Environmental Oriented Electrochemistry. Studies in Environmental Sciences, Volume 59

ENVIRONMENTAL ORIENTED ELECTROCHEMISTRY

This Page Intentionally Left Blank

Studies in Environmental Science 59

ENVIRONMENTAL ORIENTED ELECTROCHEMISTRY

Edited by

C.A.C. Sequeira lnstituto Superior Tecnico Technical University of Lisbon Lisbon, Portugal

ELSEVlER Amsterdam

-

London

-

New York

-

Tokyo 1994

ELSEVIER SCIENCE 6.V Sara Burgerhartstraat 25 P.O. Box 21 1.1000 AE Amsterdam, The Netherlands

Library of Congress Cataloging-in-Publication

Data

E n v i r o n m e n t a l o r i e n t e d e l e c t r o c h e m i s t r y / e d i t e d by C . A . C . S e q u e i r a . p. c m . -- ( S t u d i e s in e n v i r o n m e n t a l s c i e n c e ; 59) I n c l u d e s b i b l i o g r a p h i c a l r e f e r e n c e s a n d index. ISBN 0 - 4 4 4 - 8 9 4 5 6 - X 1 . P o l l u t i o n . 2. E l e c t r o c h e m i s t r y . I. S e q u e i r a , C . A . C . 11. S e r i e s . TD191.5.E54 1994 628'.01'54137--d~20 94-1 1141 CIP

ISBN: 0-444-89456-X

0 1994 Elsevier Science B.V. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means,electronic, mechanical, photocopying, recording or otherwise, withoutthe prior written permission of the publisher, Elsevier Science B.V., Copyright & Permissions Department, P.O. Box 521,1000 A M Amsterdam, The Netherlands. Special regulations for readers in the USA - This publication has been registered with the Copyright Clearance Center Inc. (CCC),Salem, Massachusetts. Information can be obtainedfrom theCCCaboutconditionsunderwhich photocopiesofpartsofthispublication may bemadeinthe USA. All other copyright questions, including photocopying outside of the USA, should be referred to the copyright owner, Elsevier Science B.V., unless otherwise specified. No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or f r o m any use or operation of any methods, products, instructions or ideas contained in the material herein. This book is printed on acid-free paper.

Studies in Environmental Science Other volumes in this series 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

Atmospheric Pollution 1978 edited by M.M. Benarie Air Pollution Reference Measurement Methods and Systems edited by T. Schneider, H.W. de Koning and L.J. Brasser Biogeochemical Cycling of Mineral-Forming Elements edited by P.A. Trudinger and D.J. Swaine Potential Industrial Carcinogens and Mutagens by L. Fishbein Industrial Waste Management by S.E.Jmrgensen Trade and Environment: ATheoretical Enquiry by H. Siebert, J. Eichberger, R. Gronych and R. Pethig Field Worker Exposure during Pesticide Application edited by W.F. Tordoir and E.A.H. van Heernstra-Lequin Atmospheric Pollution 1980 edited by M.M. Benarie Energetics and Technology of Biological Elimination of Wastes edited by G. Milazzo Bioengineering, Thermal Physiology and Comfort edited by K. Cena and J.A. Clark Atmospheric Chemistry. Fundamental Aspects by E. Meszaros Water Supply and Health edited by H. van Lelyveld and B.C.J. Zoeteman Man under Vibration. Suffering and Protection edited by G. Bianchi, K.V. Frolov and A. Oledzki Principles of Environmental Science and Technology by S.E. Jmrgensen and I.Johnsen Disposal of Radioactive Wastes by Z. Dlou hi/ Mankind and Energy edited by A. Blanc-Lapierre Quality of Groundwater edited by W. van Duijvenbooden, P.Glasbergen and H.van Lelyveld Education and Safe Handling in Pesticide Application edited by E.A.H. van Heemstra-Lequin and W.F. Tordoir Physicochemical Methods for Water and Wastewater Treatment edited by L. Pawlowski Atmospheric Pollution 1982 edited by M.M. Benarie Air Pollution by Nitrogen Oxides edited by T. Schneider and L. Grant Environmental Radioanalysis by H.A. Das, A. Faanhof and H.A. van der Sloot Chemistryfor Protection of the Environment edited by L. Pawlowski, A.J. Verdier and W.J. Lacy Determination and Assessment of Pesticide Exposure edited by M. Siewierski The Biosphere: Problems and Solutions edited by T.N. Veziroglu Chemical Events in the Atmosphere and their Impact on the Environment edited by G.B. Marini-Bettolo Fluoride Research 1985 edited by H. Tsunoda and Ming-Ho Yu Algal Biofouling edited by L.V. Evans and K.D. Hoagland Chemistry for Protection of the Environment 1985 edited by L. Pawlowski, G. Alaerts and W.J. Lacy Acidification and its Policy Implications edited by T. Schneider Teratogens: Chemicals which Cause Birth Defects edited by V. Kolb Meyers Pesticidechemistry by G. Matolcsy, M. Nadasy and Y. Andriska Principles of Environmental Science and Technology (second revised edition) by S.E. Jmrgensen and I. Johnsen

Chemistry for Protection of the Environment 1987 edited by L. Pawlowski, E. Mentasti, W.J. Lacy and C. Sarzanini 35 Atmospheric Ozone Researchand its Policy Implications edited by T. Schneider, S.D. Lee, G.J.R. Wolters and L.D. Grant 36 Valuation Methods and Policy Making in Environmental Economics edited by H. Folmer and E. van lerland 37 Asbestos i n Natural Environment by H. Schreier 38 How t o Conquer Air Pollution. A Japanese Experience edited by H. Nishimura 39 Aquatic Bioenvironmental Studies: The Hanford Experience, 1944-1984 by C.D. Becker 40 Radon i n the Environment by M. Wilkening 41 Evaluation of Environmental Data for Regulatory and Impact Assessment by S. Ramamoorthy and E. Baddaloo 42 Environmental Biotechnology edited by A. Blazej and V. Privarova 43 Applied Isotope Hydrogeology by F.J. Pearson Jr., W. Balderer, H.H. Loosli, B.E. Lehmann, A. Matter, Tj. Peters, H. Schmassmann and A. Gautschi 44 Highway Pollution edited by R.S. Hamilton and R.M. Harrison 45 Freight Transport and the Environment edited by M. Kroon, R. Smit and J. van Ham 46 Acidification Research in The Netherlands edited by G.J. Heij and T. Schneider 47 Handbook of Radioactive Contamination and Decontamination by J. Severa and J. Bar 48 Waste Materials in Construction edited by J.J.J.M. Goumans, H.A. van der Sloot andTh.G. Aalbers 49 Statistical Methods in Water Resources by D.R. Helsel and R.M. Hirsch 50 Acidification Research: Evaluation and Policy Applications edited by T. Schneider 51 Biotechniquesfor Air Pollution Abatement and Odour Control Policies edited by A.J. Dragt and J. van Ham 52 Environmental Science Theory. Concepts and Methods in a One-World, Problem-Oriented Paradigm by W.T. de Groot 53 Chemistry and Biology of Water, Air and Soil. Environmental Aspects edited by J. Tolgyessy 54 The Removal of Nitrogen Compoundsfrom Wastewater by B. Halling-S~rensenand S.E. J ~ r g e n s e n 55 Environmental Contamination edited by J.-P. Vernet 56 The Reclamation of Former Coal Mines and Steelworks by I.G. Richards, J.P. Palmer and P.A. Barratt 57 Natural Analogue Studies in the Geological Disposal of Radioactive Wastes by W. Miller, R. Alexander, N. Chapman, I. McKinley and J. Smellie 58 Water and Peace in the Middle East edited by J. Isaac and H. Shuval 34

vii

To the memory of my grandmother "a Vd Lina': and my aunts "a Tia Guiomar e a Chichgo", who believed in Honesty, Sincerety and Hardworking. To my wfe, Maria Elisa, and my children, J o q e Augusto, C h a r Josk, Rita Alexandra and Catarina,for their support and perseverance.

This Page Intentionally Left Blank

ix

PREFACE We all are aware of growing environmental concerns, and the need to provide new and improved means for maintaining a healthy environment. A broad range of interest is being sought, with special emphasis on waste minimization (e.g., process modification, material recovery), hazardous waste treatment (e.g., in surface finishing and manufacturing industries, destruction of organic solvents), and non-hazardous waste treatment (e.g., water treatment, destruction of plastics).

Electrochemistry is concerned with the way electricity produces chemical changes and in turn chemical changes result in the production of electricity. This interaction forms the basis for an enormous variety of processes ranging from heavy industry through batteries to biological phenomena. Although there are many established applications, modern research has led to a great expansion in the possibilities for using electrochemistry in exciting future developments. One of the main directions of Electrochemistry, of great importance for the future needs of mankind, is in the area of The Environment, where potentialities range from applications in monitoring various substances polluting or affecting the environment as well as in removing pollutants of any kind, and also in affectingenergy production and all kinds of transport, which are among the major polluters of the environment. This book on "Environmental Oriented Electrochemistry" concentrates on the ElectrochemistryEnvironment relationship including, among others, chapters on design and operation of electrochemical reactors and separators, process simulation, development and scale-up, optimization and control of electrochemical processes applied to environmental problems, also including economic analysis, description of unique current and future applications, in addition to basic research into developing new technologies. It is hoped that this volume will be considered interesting and extremely timely to specialists in electrochemistry and environmental sciences. Finally the editor would like to express his gratitude to the many talented contributors to this book and to his wife, Maria Elisa, for her assistance in retyping several of the chapters. Char Sequeira Lisbon,February 1994

This Page Intentionally Left Blank

xi

LIST OF CONTRIBUTORS Dr. Yakin B. Acar

Civil Engineering Department, Louisiana State University, Baton Rouge, LA 70803 U.S.A.

Dr. Steven R Alexander

Chemical Group Hoechst Celanese Corporation Corpus Christi Technical Center 1901 Clarkwood Road P.O. Box 9077 Corpus Christi, TX 78469 U.S.A.

Dr. A. AGTurkail

College of Basic Education, P M T P.O. Box 38552 Abdulla Al-Salem 12256 Kuwait

Eng. L P. S. Araujo

Laboratory of Electrochemistry, Department of Chemical Engineering Instituto Superior T6cnico Av. Rovisco Pais 1096 Lisboa Codex Portugal

Dr. A. Asiri

P.O. Box 4 191 Jeddah 2 149 1, Kingdom of Saudi Arabia

xii

Eng. I? S. D. Bra0

Laboratory of Electrochemistry, Department of Chemical Engineering, Instituto Superior Tecnico Av. Rovisco Pais, 1096 Lisboa Codex Portugal

Dr. Per Bro

Southwest Electrochemical Company, Santa FeyNM87501 U.S.A.

Dr. Christos Comninellis

Institute of Chemical Engineering, Swiss Federal Institute of Technology, CH- 1 0 1 5 Lausanne Switzerland

f i o j Thomas 2.Fahidy

Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario N2L 3G1 Canada

Dr. Joseph C Farmer

University of California, Lawrence Livennore National Lab. Livermore, CA 94550 U.S.A.

Dr. J. M.Fenton

Department of Chemical Engineering, U-139 University of Connecticut Stons, CT 06269-3139 1J.S.A.

Dr. RobertJ. Gale

Chemistry Department, Louisiana State University Baton Rouge, LA 70803 U.S.A.

xiii

Department of Chemistry University of New Orleans, New Orleans, LA 70 148 U.S.A.

Dr. S. K Haram

Chemical Physics Group Tata Institute of Fundamental Research Colaba, Bombay 400005 India

Dr. M I. Ismail

Sci & Ad Inst, P.O. Box 98029 S. Common Post Outlet 21 50 Burnhamthorpe Rd. Miss., Ontario ISL 3AO

Canada

Dr. E Kajiyamu

Pro$

!I E Bazarjnov

Dr. Yu I. Kharkats

Corrosion Science, Fundamental Technology Research Laboratory, Tokyo Gas Company, 1-16 Shibaura Tokyo 105 Japan The A.N. Frumkin Institute of Electrochemistry, Russian Academy of Sciences, Leninsky prospect 31, Moscow 117 07 1 Russia The A.N. Frumkin Institute of Electrochemistry, Russian Academy of Sciences, Leninsky prospect 3 1, Moscow 1 17 071 Russia

xiv

Prof K Kordesch

Institute of Inorganic Chemical Technology, Technical University of Graz, Stremayrgasse 16411, A-8010 G r a ~ Austria

Dr. Samuel C Levy

Sandia National Laboratories, Albuquerque, NM 87 185 U.S.A.

Dr. Heyi Li

Chemistry Department, Louisiana State University Baton Rouge, LA 70803 U.S.A.

Dr. K Micka

J. Heirovsky Institute of Physical Chemistry and Electrochemistry, Czechoslovak Academy of Sciences, 182 23 Prague 8 Cxchoslovakia

Dr. T. Nakahara

Fujisawa Research Laboratory ‘Tokuyama Soda Co., Ltd., 2051 Endo Fujisawa-city 252 Japan

Dr. D. Ohms

Deutsche Automobilgesellschafi mbH (DAUG), D-7300Esslingen, Emil-Kessler-Str. 5 . Germany

Prof A. M. G. Pacheco

Laboratory of Electrochemistry Department of Chemical Engineering, Instituto Superior T6cnico Av. Rovisco Pais 1096 Lisboa Codex Porlugal

xv

Pro$ Ya V. P h k o v

The A.N. Frumkin Institute of Electrochemistry, Russian Academy of Sciences, Leninsky prospect 3 1, 117071 Moscow Russia

Dr. G. W: Reude

Electrochemistry Group, Chemistry Department, University of Portsmouth, White Swan Road, Portsmouth PO 1 2DT U.K.

Dr. Gabriele Nocchini

Italian Electricity Board, Thermal and Nuclear Research Center, Via Rubattino 54 20 134 Milan IdY

Dr. 1. Rousar

Department of Inorganic Technology, Institute of Chemical Technology 166 28 Prague 6 Czechoslovakia

Proj K S. K Sunthanam

Chemical Physics Group Tata Institute of Fundamental Research Colaba, Bombay 400005 India

prof: Emeritus T Seiyama

Kyushu University, Tokuyama Soda Co., Ltd., Fukuoka Branch, Sanwa-

-bldg., 1-10-24 Tenjin, Chuoku, Fukuoka-city 810 Japan

xvi

Pro$ C A. C Sequeira

Laboratory of Electrochemistry, Department of Chemical Engineering, Instituto Superior Ttcnico Av. Rovisco Pais 1096 Lisboa Codex Portugal

B o j 1% Strathmann

University of 'I'wente, Faculty of Chemical Technology, P.O. Box 217 7500 M,Enschede The Netherlands

Dr. A. A. Suleiman

Department of Chemistry, Southern University, Baton Rouge, LA 70813 U.S.A.

Dr. 1: Takeuchi

Fujisawa Research Laboratory Tokuyama Soda Co., Ltd. 205 1 Endo, Fujisawa-city 252 Japan

Dr. I? Tatapudi

Department of Chemical Engineering U-139 University of Connecticut Stems, CT 06269-3 139 U.S.A.

Dr. K Taucher

Institute of Inorganic Chemical Technology, Technical University of Graz, Stremayrgasse 16mI A-80 10 G r a ~ Austria

xvii

Dr. l? Tebbutt

Chemistry Department, University of Warwich Coventry CV4 I A L U.K.

Dr. G. A. Tedoradze

The A N . Frumkin Institute of Electrochemistry, Russian Academy of Sciences, Leninsky prospect 3 1, Moscow 1 17 07 1 Russia

Dr. E C Walsh

Electrochemistry Group, Chemistry Department, University of Portsmouth, White Swan Road, Portsmouth, PO1 2DT

1J.K.

prof: K Wiesener

Institute of Physical Chemistry and Electrochemistry, Dresden University of Technology, Mommsenstr. 13, D-0-8027 Dresden, Germany

This Page Intentionally Left Blank

xix

CONTENTS Preface List of Contributors

ix xi

I. WATER CARE AND TREATMENT PROCESSES Electrochemical techniques for the treatment of dilute metal-ion solutions. by F.C. Walsh and G. W. b a d e The scope for electrochemical techniques Typical electrochemical reactions Cell design considerations Classification of electrochemical reactors Examples of reactor designs and their performance Summary References

Electrochemical methods for purification of waste waters by 1. Itousar and K. Micka Introduction Electrochemical background for the deposition of heavy metals Plate electrolyser Electrolyser with a fluidized bed Other diaphragmless electrolysers Electrolysers involving a diaphragm Other, similar electrolyser types Electrochemical destruction of toxic organic compounds References

3

5 9 15

20 25 38

40

45

45 49 56 58

62 63 69 72 74

xx

Electrochemical oxidation of organic pollutants for wastewater treatment by C. Comninellis Introduction Experimental details Definition of global parameters for the electrochemical treatment Possibilities of the electrochemical oxidation of organics for wastewater treatment Analysis of reaction products at Pt and SnO2 anodes Comparison between electrochemical and chemical oxidation Mechanism of electrochemical oxidation of phenol for wastewater treatment Pilot plant experiments for the treatment of industrial wastewater References

Electrochemical oxidant generation for wastewater treatment by P. Tatapudi and J. M. Fenton Introduction Ozone Hydrogen peroxide Potassium permanganate Chlorine and hypochlorite Chlorine dioxide References

77 77 77

83 86 93 96

98 100 102

103

103 105 110 115 116 117 121

11. BATTERIES A N D ENVIRONMENT Batteries and the environment by P. Rro and S.C. Levy

131

Introduction The environmental problem Regulatory aspects

131 138 140

xxi

The management of diswrded batteries Battery recycling Battery improvements Administrative structures Sources of general information related to the battery waste problem References

Alkaline manganese dioxide - zinc batteries. Primary and rechargeable cells with and with0u.t mercury by W. Taucher and K. Kordesch Introduction Primary alkaline manganese dioxide - zinc batteries (PAM cells) Rechargeable alkaline manganese dioxide - zinc batteries (RAM cells) Batteries and environment ‘roxicology Conclusions References

142 144 157 158 159 160

163 163 165 174 191 197 199 200

Electrochemical generators and the environment. Fuel cells and metallair batteries by P.S.D. Brito and C.A.C. Sequeira 203 Introduction Fuel cells Metal/air batteries Conclusion References

The prospective role of magnesium in the development of environment-friendly solid-state batteries. by L.P.S.AraGjo, A.M.G. Pacheco and C.A.C. Sequeira Introduction Experimental results and discussion

203 208 213 219 220

223 223 226

xxii

Conclusions References

228 229

111. SENSORS FOR POLLUTION CONTROL Overview of gas sensors for environmental use by T. Seiyama, T. Nakahara and T. Takeuchi Importance of environmental problems and growing needs for gas sensing Methods of gas sensing related with environmental problems Recent researches and developments of gas sensors for environmental use Application to pollution control Future scope References

Piezoelectric crystal detectors for environmental pollutants by A.A. Suleiman and G.G. Guilbault Introduction Experimental Conclusions References

me electrochemistry of gases of medical interest and electrochemical gas sensors by P. Tebbutt Introduction Membrane covered sensors Oxygen: sensors and electrochemistry Carbon dioxide Anaesthetic gases Solid state gas sensors References

233 233 236 242 26 1 267 269

273 273 274 299 300

305 305 307 312 315 327 336 34 1

xxiii

IV. TECHNIQUES FOR POLLUTION MONITORING Environmental pollution monitoring using DC pulse techniqu,es by M.I. Ismail, A. Asiri and A. Al-Turkait 347 Introduction Electrochemical monitoring The DC pulse system Monitoring environmental pollution Rainfall pollution Monitoring soil contamination with Kuwaiti oil field burn products Monitoring pollution of seawater with Kuwaiti oil contaminants Conclusions and recommendations References

347 347 349 35 1 355 357 360 360 362

Bacterial-environmental interactions in corrosion on buried pipes By F. Kajiyama 365 Introduction Effects of bacterial ecology on underground corrosion Experimental procedures of laboratory study on local action cell corrosion Results and discussion of laboratory study on local action cell corrosion Conclusions References

365 365 367 368 374 375

Influence of ohmic drop on the determination of electrochemicul parameters by Gabriele Rocchini 377 Introduction General considerations Monitoring by a two-electrode technique Importance of the environment Numerical methods

377 379 381 384 387

xxiv

Mathematical schematization Influence of the ohmic drop on the corrosion rate Explicit form of the function i(AE) Inversion through power series expansion Numerical determination of Rs Experimental techniques Conclusions References

389 392 394 397 399 401 410 41 1

V. PHOTOELECTROCHEMISTRYFOR A CLEANER ENVIRONMENT Semiconductor photoelectrochemistry for cleaner environment: utilization of solar energy By Yu.V. Pleskov 417 Introduction Semiconductor/electrolytesystem in dark and upon illumination PEC cells for solar-to-electrical energy conversion PEC cells for solar-to-chemical energy conversion Semiconductor dispersions for providing cleaner environments Concluding remarks References

Prospective usage of photoelectrochemistry for environmental control by S.K. Haram and K.S.V. Santhanam Introduction Photoelectrochemicalreduction of C02 Photocatalytic reduction of C02 Photoelectrochemicaldecomposition of H2S Photoelectrochemicalconversion of SO2 Photoelectrochemical reduction of 0 2 Photoelectrochemical generation of €12

417 418 422 430 436 439 44I

445 445 446 452 454 454 456 457

xxv

46 1

References

Electrochemical storage of solar energy by Yu. I. Kharkats and Yu. V. Pleskov

469

469

Introduction Solar-hydrogen plant: Principles of design and state-of-the -art Solar-hydrogen plant: Simulation Solar-hydrogen plant: Optimization Optimization of the plant "solar array + electrolyzer 4 secondary battery" References

469 475 478 484 492

VI. MISCELLANEOUS ENVIRONMENTALLY ORIENTATED ELECTROCHEMICAL PROCESSES Electrodialytic membrane procesaea and their practical application by H. Strathmann 495 Introduction Properties and structures of ion-exchange membranes Electrodialytic processes as unit operation Other electrically driven membrane processes Conclusions References

495 496 505 524 530 530

Removal of H2S through an electrochemical membrane separator by S. Alexander 535 Introduction Technical discussion Experimental Experimental run results Economic projection Conclusions References

535 536 539 543 547 563 564

xxvi

Electrochemical treatment of mixed and hazardous wastes by J.C. Farmer Introduction State-of-the-art Examples of mixed-waste feed streams Process chemistry Electrochemical generation of oxidants in batch reactor Destruction and coulombic efficiencies for ME0 process Reaction intermediates formed during M E 0 of glycol and benzene Reaction mechanisms for ME0 of ethylene glycol and benzene Theoretical model for electrochemical batch reactor Application of model to the M E 0 of glycol by silver (11) Maximum mediator concentration - no organics Other recommended reading Summary References

Magnetic field effects in environmental control involving electrolytes by T. Z.Fahidy Preamble and motivation Theoretical foundations The magnetic field effect in the treatment of waste fluids and effluents The magnetic field effect on corrosion in electrolytes The magnetic field effect on natural bodies of water Critique and concluding remarks References

Soil decontamination using electrokinetic processing by R.J. Gale, H. Li and Y.R.Acar Introduction Theoretical Results

565 565 565 568 569 57 1 574 583

586 588 590 594 597 597 598

60 1 60 1 602 605 608 610 615 616

62 1 62 1 623 632

xxvii

Summary References

Electrochemical utilization of hydrochloric m i d - waste of chlororganic production by C.A. Tedoradze and V.E. Kazarinov Hydrochloric acid electrolysis producing chlorine and hydrogen Hydrochloric acid electrolysis producing hydrogen and chlororganic compounds References

Elect roca>talysis for environmentally orientated electrochfernicalprocesses and environmental protection by K. Wiesener and D. Ohms Introduction Environmental problems in electrochemicalprocesses The special role of hydrogen and oxygen electrodes Examples for electrolytic applications of oxygen and hydrogen electrodes Conclusions References

64 8 65 1

655

657 662 675

687 688 688 689 694 707 707

Author Index

71 1

Subject Index

713

This Page Intentionally Left Blank

SECTION ONE

WATER CARE AND TREATMENT

PROCESSES

This Page Intentionally Left Blank

3

Electrochemical techniques for the treatment of dilute metalion solutions Frank C. Walsh* and Gavin W. Reade Electrochemistry Group, Chemistry Department, University of Portsmouth, White Swan Road, Portsmouth PO1 2DT, United Kingdom

* author for correspondence ABSTRACT The major sources of dilute, metal ion liquors are identified within the metals productiodprocessing and chemical industries. Problems associated with traditional methods of metal ion removal are highlighted and the developing role of electrochemica: techniques is discussed. Electrode and cell reactions are illustrated via typical examples from laboratory and industrial practice. The need to select a n appropriate cell design and to control the reaction conditions is emphasised via consideration of the problems caused by secondary reactions. Important design criteria for electrochemical reactors are summarised. Available reactors are classified according to the nature of the product which may be metal flake or powder, a metal deposited onto a disposable substrate, a metal ion concentrate or an insoluble metal compound. The applications for electrochemical techniques in environmental treatment are illustrated by examples which show features of reactor construction and their typical performance. Current trends are summarised a n d recommendations are made for further work in critical areas.

Keywords Electrochemical cells (reactors); electrochemical techniques; electrolysis; wastewater treatment; recycling of metals; electrodeposition; cementation; electrodialysis; metal ion removal; metal ion concentration.

4

NOMENCLATURE Meaning

S I Units

cathode area

m2

bulk concentration of metal ions

mol m-3

concentration change

mol m-3

initial concentration

mol m-3

concentration at time t

mol m-3

inlet concentration

mol m-3

outlet concentration

mol m-3

Faraday constant

96 485 C mot1

current

A

current density

A m-2

limiting current density

A mm2

mass transport coefficient molar mass of metal

kg mol-l

number of identical reactors in cascade

dimensionless

volumetric flow rate

m3 5-1

time

s

electrolyte volume

m3

mass of metal

kg

fractional conversion of reactant

dimensionless

number of electrons per metal atom

dimensionless

current efficiency

dimensionless

residence time

S

5

1.

THESCOpEFORELEcTRocHEMlcALTECHNIQ~

1.1

So-

ofmetal-ionliquors

Aqueous liquors which contain low levels (less than 1 g dmq3) of dissolved metal routinely arise during the industrial production, processing and surface finishing of metals. In addition, many sectors of the chemical industry also utilise metal containing cleaners, catalysts, and process reagents. A survey of these industrial sectors [ll is summarised in table 1 which indicates the diversity of such metal ion liquors with respect to: (i)

The nature of the industrial sectors, which spans most of the manufacturing and reprocessing areas of industrial endeavour.

(ii)

The solution composition, in terms of the number of metals, the pH, the ionic composition and electrolytic conductivity.

(iii)

The scale of operation which ranges from e.g. large scale hydrometallurgical processing of multimetal liquors down to laboratorybased removal of a specific metal.

Table 1 General sources of dilute metal-ion process streams Primary ore leaching Mine dump leaching or run-off water Liberator cells in electrowinning and electrorefining Electroplating including spent and contaminated baths Etching solutions and rinse waters Metal-cleaning solutions Scrap reprocessing and refining Catalyst liquors Reagants for chemical processing Photographic processing solutions such as fixers and rinse waters Manufacture and reprocessing of spent batteries Factory effluents

6

1.2

Traditional Methods of"reatment

Traditional treatment methods for these dilute metal-ion liquors usually involve combinations of the following: Dilution, in order to lower the metal concentration and other dissolved components towards local regulatory discharge limits. Mixing which, in addition to dilution, tends to moderate pH and result in some chemical precipitation. Controlled chemical precipitation; after flow-balancing and pH adjustment, a cheap reagent such as lime or caustic soda slurry is normally used. Following sedimentation, the insoluble mixed-metal hydroxide sludge may be dried by filtration. Temporary storage, followed by transportation of the waste t o a n organisation which specialises in reprocessing or disposal. Dilution is technically simple but may lead to high water costs and a variable effluent composition. The deliberate mixing of process streams may prove hazardous and very difficult t o control. Chemical precipitation is a wellestablished technology which is capable of removing a diverse range and level of dissolved metals, when the process has been suitably designed, implemented and controlled. However, sludge disposal is an increasing problem due t o restrictions on transport and ever more restricted facilities for bore-hole, sea or landfill dumping. the storage of waste liquors has attendant problems of health and safety, and space utilisation, whilst also being a strictly temporary solution. Transportation and disposal of wastes requires the services of approved and specialist organisations. These traditional approaches largely result from an 'effluent treatment' approach to waste disposal i.e. the strategy is t o utilise a single treatment plant between the factory effluent and the discharge point to the local water disposal system. Such an approach is easily criticised on the following grounds: (i)

Flexibility is limited e.g. a newly installed production process may result in species which the effluent treatment plant cannot cope with or the metals load may be excessive.

(ii)

The scope for materials recycling is very limited. Depending upon the process involved, i t may be desirable to recycle one or more of the following: a chemical reagent, the metal, a metal-ion liquor or water.

(iii)

In ecological terms, valuable resources have been wasted. In terms of the energy flow, metals are won from their ores or from scrap via a massive and concentrated energy input.

The result of sludge formation and disposal is not only a return t o the combined state i.e. metal hydroxides, oxides and other compounds but a dispersal of material and energy (figure 1).

7 (a)

(bl

FINISHED METAL ION LIQUOR

PRODUCT

I

J

4

MIXING

SURFACE

DILUTION

FINISHING

1

F A B R I C A T I O N 8. PROCESSING

RMEEFTI AN LE D

&

Ij

C O R R Ep CH T I O N

l+OH-

SLUDGE SETTLEMENT

MZ+

DEWATERED

HY DROMETAL. EXTRACTION

I Figure 1.

MINED ORE

1

SLUDGE DISPOSAL

1

The origin and treatment of dilute metal-ion liquors showing the analogy between mined ore and chemically precipitated sludge. (a) The generation of waste metal-ion, Mz+ solutions routinely occurs during the hydrometallurgical extraction, refining, processing and surface finishing of metals. (b> Conventional chemical treatment of such liquors via e.g. hydroxyl ion addition results in sludge formation, together with its attendant disposal problems; a dispersion of energy and materials also occurs (as in the case of corrosion).

8

1.3

'point of source' treatment strategies

Ideally, the need for recycling and wastewater treatment should be considered at the onset of the design of a production process. Whenever feasible, a 'point of source' treatment strategy should be adopted. Such an approach should form part of an 'integrated waste management' view. The treatment of process liquors near t o their source often has the following additional advantages. (i)

the requirements of the selected treatment method are lowered

(ii)

the choice of available techniques is widened.

(iii)

the load on the final effluent treatment plant is much reduced.

The trend toward modular point of source treatment processes is being encouraged by several factors including: (i)

increasing concern over environmental pollution

(ii)

stricter (and more stringently applied) legislation

(iii)

increasing materials costs or supply problems

(iv)

the need t o meet higher product specifications and hence to exercise control over all aspects of the production process

(v)

limited confidence in the cost, continued availability, and efficiency of certain disposal routes and organisations.

In practice, specialised devices for metal-ion removal are often adopted in a non-optimised fashion, following brief surveys of process requirements. Such 'retrofitting' normally demands compact, modular units which are robust and have low maintenance requirements. Purchase, installation and running costs must be low and the technology must be packaged in an acceptable fashion. There are now increasing economic, social, legal and environmental pressures to utilise the 'Best Available Technology' Not Entailing Excessive Cost (BATNEEC)[2]. Available techniques for the removal of metal ions include chemical precipitation, ion exchange, evaporation, solvent extraction and a variety of membrane separation processes including reverse osmosis, ultrafiltration and electrodialysis [31. Each of these methods has its own advantages but all lack the ability of certain electrochemical techniques to produce metal directly in a controlled fashion. 1.4.

The role ofe1whemica.ltechniques

At the onset, it should be realised that electrochemical techniques are unable to provide a single solution to the problem of metal-ion treatment. Adverse factors include the following:

9

The performance of electrochemical reactors is strongly dependent upon solution composition and temperature. The low conductivity of many process liquors results in an excessive cell voltage and attendant joule heating problems. Traditional electrochemical reactor designs tend t o involve a small electrode area per unit reactor volume and poor electrode/electrolyte movement. Cell components including anodes, cathodes and membranes have a restricted range of chemical (and mechanical) stability. Lack of familiarity with electrochemical reactors in comparison to other unit processes. Despite these problems electrochemical technology is being increasingly applied to a wide variety of environmental and recycling problems [l, 4-71. Continued developments in our understanding and documentation of electrochemical engineering [8-131 and electrochemical reactor design [8-151 together with increasing industrial experience of their use [16] are resulting in a more widespread acceptability of electrochemical technology. Electrochemical reactors are currently providing a powerful unit process in the armoury of available metal-ion removal techniques 141. They may be applied singly or in combination with other physical and chemical processes such as ion exchange, solvent extraction or membrane separation techniques. Extensive reviews of reactor designs for metal-ion removal are available in the literature 11, 4, 11, 17-24] but there is a predominance of laboratory cell designs which have been used to treat simple, synthetic liquors on a small scale. Unfortunately, there is a relatively small literature concerning the construction and performance of pilot- and (particularly) full scale reactors. Moreover, a case-study approach is rarely seen and the need to select the reactor design to suit the process needs is not always evident. In particular, it is important to consider the desired (and other) electrode reactions in the light of the composition of the process liquor, the reactor geometry and the electrolysis conditions. Equally, the metal must be removed from solution in a suitable form and at the desired frequency.

2.

TYPICAL ELECTROCHEMICAL REACTIONS

Electrochemical reactors necessarily involve both cathodic and anodic reactions and the cell may be divided by an ion permeable separator (figure 2).

2.1

CathodicReactions

The most generally useful method of metal-ion removal is cathodic deposition of metal which may be represented by:

Mz++ ze’

4

M

(1)

10

C U ~ ++ 2e- +

e.g.

cu

(2)

In certain cases, the metal-ion is complex rather than being a simple aquocation, as in the following cases: Ag(S20,)23-+ e-

+ Ag + 2S2032-

(3)

CU(CN),~-+ e-

+

(4)

CUC~,~+ e-

Cu + 3CN-

+ Cu + 3C1-

(5)

I O N PERMEABLE SEPARATOR - microporous p o l y m e r - ion exchange membrane

ANODE

'

--static -moving

I

1

>cations

I

I I I

Oxd + R e d

z ,e-

M'*-+M

l

I

I I I

anions

< I I I

-

t Figure 2.

1' ELECTROLYTE FLOW -zero -continuous - intermittent

I ME'TAL D E P O S I T - smooth -rough - compact powdery

-

Schematic representation of a reactor for electrolytic removal of metal-ion in the form of a metal deposit. showing the options for major cell components.

The desired (or primary) reaction may be accompanied by unwanted, secondary reactions, resulting in a loss of cathode current efficiency. Typical examples include hydrogen evolution: pHc7

2H+ + 2e- + H2

(6)

pH>7

2H20 + 2e- + H2 + 20H-

(7)

11

which may cause health and safety problems due to the possibility of aggressive electrolyte mists or explosive gas mixtures. Reduction of dissolved oxygen: pH<7

0,

+ 4H+ + 4e- + 2H,O

(8)

pH>7

0,

+ 2H20 + 4e- + 40H-

(9)

becomes more important at very low metal-ion levels, (
e.g.

Mz+ + zOH- + M(OH),

(10)

Cu2++ 20H- + Cu(OH),

(11)

Such reactions may result in a less pure metal product or even a tendency towards passivation of the cathode surface. Another common secondary reaction is the reduction of a soluble species a s in the case of ferric ions in acid solution: Fe3+ + e- + Fe2+

(12)

or thiosulphate ions in photographic fixing liquors: s2032-+ 2e- + SO,,- + ~ 2 -

( 13)

Reaction (13) is particularly problematic in photographic silver recovery as a rapid chemical reaction results in the formation of silver sulphide: S2- + 2Ag+ + Ag,S

(14)

The deposition of silver sulphide is detrimental as it contaminates the silver deposit and the solution, interferes with the kinetics of silver deposition and tends to cause chemical instability of the fixer. At extremely negative cathode potentials, cathodic decomposition of the fixer may result in emission of toxic hydrogen sulphide:

S,O,,- + 10H++ 8e- + 2H2S + 3H$

(15)

In acidic electrolytes containing antimony or arsenic, the evolution of noxious stibine or arsine is possible e.g. during the reduction of As(II1) solutions:

12

As3+ + 3H+ + 6e- + ASH,

(16)

In general, side reactions are more problematic at higher cathode current densities (i.e. at more negative cathode potentials) and they may be minimised by appropriate cell design, alteration of the solution chemistry (where possible) and control of the electrolysis conditions. 2.2.

Anodicreactiom

Commonly, the preferred anode process is oxygen evolution at an inert surface: pHc7

W,O - 4e- + 0,

+ 4H+

(17)

pH>7

40H- - 4e- + 0,

+ 2H20

(18)

Other anode processes include:

dissolution of the anode metal:

MA- ze-

+

MAz+

(19)

dissolution of the metal of interest via contact of the deposit with the anode: M-ze-

Mz+

(20)

toxic chlorine evolution in chloride liquors: 2C1- - 2e-

+ C1,

(21)

oxidation of solution species such as ferrous or thiosulphate ions: Fe2+- e-

+ Fe3+

2e-2 + 2 ~ ~ 0- ~ - S,O,%

(22) (23)

Once again, it is important to pay adequate attention to the selection of materials, features of cell design and control of the reaction conditions to avoid problems with unwanted anode reactions. In many cases, the anode will be selected to cause minimum disruption to the cathode reaction [25]. In many cases, an ion permeable separator can alleviate problems via: (a)

exclusion of aggressive species e.g. C1- from the anode compartment (figure 3a) or 0, from the cathode one (figure 3b)

(b)

prevention of redox shuttles due t o e.g. Fe3+/Fe2+reactions (figure 3c)

13

(c)

prevention of dissolution of the stray metal deposit by direct anodic contact (figure 3d)

(d)

provision of a useful anode reaction as in the regeneration of printed circuit board etchants (figure 3e).

C1- is excluded from the anode compartment to avoid anode corrosion or Cl2 evolution.

0 2 is excluded from the cathode

compartment to avoid loss of current efficiency or open-circuit corrosion due t o oxygen reduction.

Fe3+is excluded from the anode compartment to avoid loss of current efficiency due to redox shuttles involving Fe3+ and Fe2+.

Metal powder is prevented from contacting the anode where rapid redissolution would occur.

(e)

Figure 3.

The anode reaction is utilised productively as in the case of regeneration of a printed circuit board etchant, e.g Cu(II), as shown.

Examples of divided cells for metal-ion removal showing the importance of an ion permeable separator. (In all cases, the cathode process is metal deposition).

14

a3

cellreactions

The majority of reactors utilise electrolytic cells i.e. the cell reaction is driven by an external power supply; general cases have been considered in figure 3 and specific examples will be illustrated in section 5. In certain cases, the electrode materials and conditions may result in a spontaneous reaction i.e. the use of a galvanic cell, particularly in the case of the displacement (cementation) of metals. for example, cupric ions may be removed as a copper deposit on a soluble iron powder in acid solution:

-

anodic reaction

Fe 2e’ + Fe2+

cathodic reaction

CU~+

mixed electrode reaction

Fe + Cu2++ Fe2++ Cu

+ 2e- + cu

(24) (2) (25)

This technique is widely used in hydrometallurgy but is restricted in wastewater treatment due to difficulties in control and product purity. Recently, however, controlled electrolyte flow and electrode dispersion conditions have resulted in significant improvements as in the case of the Actimag process which utilises a pulsed magnetic field [261. Another example of a galvanic cell reaction is provided by open circuit corrosion of the metal deposit. Freshly deposited (and particularly finelydivided) metals are more active than their bulk, compact counter parts. Corrosion of the mixed electrode deposit may ensue if the cathode surface is left under open circuit conditions; metal dissolution is balanced via reduction of species such as dissolved oxygen, protons o r higher oxidation states of transition metal ions. Illustrative (simplified) examples of such oxidising agents include the following: (a)

corrosion of copper in acidic solutions:

anodic sites

2Cu - 4e- + 2cu2+

cathodic sites

0,

local cell process

2Cu + 4H+ + 0,

(b)

+ 4H++ 4e- + 2H20

+ 2Cu2++ 2H20

corrosion of zinc in acid solutions:

anodic sites

Zn - 2e- + Zn2+

cathodic sites

2H++ 2e- + H,

local cell process

Zn + 2H++ Zn2++ H,

(26)

(8) (27)

15

(c)

corrosion of gold in acidic chloride solutions:

anodic sites

4Au + 16C1- - 12e- + 4AuC14-

cathodic sites

30,

local cell process 4Au + 30, (d)

+ 12H++ 12e-+ 6H2O + 16C1- + 12H++ 4AuC1,- + 6H20

(30)

(8) (31)

corrosion of silver in bleach-fix solutions containing ferric-EDTA complexes:

anodic sites

Ag + 2S2032-- e- + Ag(S,0,)23-

(32)

cathodic sites

Fe(II1) + e- + Fe(I1)

(33)

local cell process Ag + 2S,032-

+ Fe(II1) + Ag(S203)23-+ Fe(I1)

(34)

So far, this paper has considered the sources of metal-ion liquors and problems with traditional methods of treatment and the importance of a 'point of source' strategy towards wastes has been emphasised. The possible role of electrochemical techniques has been outlined and illustrated by typical electrochemical reactions. In practice, the choice of reactor design and the reaction conditions are extremely important.

3.

CELL DESIGN CONSIDERATIONS

3.1

Guidelines for reactor design

There are many interrelated factors which must be considered prior to the design and implementation of electrochemical reactors [ l , 17, 203. In the present case, the following guidelines [27] apply: (i)

Moderate costs: Capital costs may be minimised by choosing a simple cell design which utilises regular components and offers a modular approach to scaleup. Running costs can be reduced by: (a>use of low cost, reliable cells components, (b) a low cell voltage which requires suitable choice of the electrode material and its shape, the reaction at the second electrode, a small interelectrode gap, a sufficiently high electrolyte conductivity, a n undivided cell, conductive electrodes and effective current feeders, (c) avoiding high power mechanical devices for electrode movement or electrolyte agitation and (d) maintaining a low pressure drop over flow-through reactors.

(ii)

Convenience and reliability in operation: The reactor must allow routine extraction of the product(s1 in a suitable form and a t a reasonable frequency. It must also facilitate maintenance as well as being safe and reliable. In many process environments, the reactor may have to operate for long periods without attention.

16

(iii)

Suitable reaction engineering: The reactor must have an adequate potential distribution and current density distribution, in order t o achieve a high selectivity together with a reasonable production rate. The need for a high and uniform rate of mass transport will often necessitate careful control of the hydrodynamics. The active electrode area per unit reactor volume should be high, particularly when a compact design is needed, or when the desired reaction is restricted t o a low current density. In certain cases, an alternating anodic and cathodic electrode region may be useful in destabilising metal complexes. These considerations are developed in the following section.

(iv)

Suitability and versatility: in use: The reactor must readily integrate into the overall process and should operate in a suitable mode with respect to both the electrolyte flow and the frequency of reactant addition and product withdrawal. The size, shape and ease of scaleup/expansion may be particularly important considerations in a pilot plant facility. following commissioning o r process changes, the performance may need to be uprated or tuned.

Choice A

Factor Mode of operation

I Number of anodes

and cathodes Electrode geometry Electrode motion Electrode connections

Choice B

Batch I

ISingle -

Continuous I

IMultiple

2-dimensional

3-dimensional

I

I Static

I Moving

I

I

[ Monopolar

1

IBipolar

Interelectrode gap

Moderate

Capillary/’zero’

Electrolyte manifolding

External

Internal

Cell division

Undivided

Divided

Sealing of cell

Open cell

Closed cell

Type of electrolyte

Liquid

Solid polymer

1

i

b

(Decision ‘B’ may result in a greater effort in design, higher costs or a more complex construction; it may, however, provide operational advantages.)

17

(v)

Simplicity: The reactor design should fulfil the process requirements with as few components as possible together with elegant engineering. Such an approach will minimise costs and will often produce hardware which is attractive to users and easy to maintain. Inevitably, the above considerations often result in conflicting requirements of a reactor design. For example in some cases, a more complex reactor design (in terms of its construction) may reward the design engineer with enhanced performance o r market advantages (table 2). Hence the electrochemical engineer must exercise considerable skill and experience in order to achieve suitable compromises in the process of cell design or selection.

32

Rates of metal-ionremoval

The rate of an electrochemical reaction is usually restricted to a certain (range) of operational current density, j, in order t o achieve a high current efficiency together with a suitable form of metal product. A t a constant current, I, the mass rate of metal-ion removal is given by Faraday's laws of electrolysis as 1281: dw -=-

$MI

dt

zF

(35)

where w is the mass of metal, t is the time, (I is the cathode current efficiency, I is the current, M is the molar mass of metal, z is the number of electrons and F is the Faraday Constant. The cathode current density is defined as:

.

J = -

I A

(36)

allowing equation (35) to be rewritten as: dw - (IjAM -_-

zF

dt

(37)

which indicates the importance of equation (37) in maintaining high values of $, j and A. The change in the concentration of metal ions due to electrochemical reaction, Ac, may be expressed as follows: (a)

For a batch electrolyte system of volume V, W

AC = -

VM

(38)

where: (39)

18 (b)

for a flow through reactor involving a volumetric throughput Q and a residence time z, W

AC = QzM

where:

Equation (37)may be integrated and rewritten in terms of a change in the metal-ion concentration. For a batch system: Ac=-

jAMt zFV

(42)

and for a continuous flow through reactor:

The last two equations show the importance of maintaining a large electrode area, a high current efficiency and a large current density.

3.3

Mass transportcontrolled depositionofmetal

The treatment of many process liquors often results in an electrochemical reaction which is under mass transport control due t o the restricted convective-diffusion of species t o (or from) the electrode (figure 4). This is particularly true in dilute liquors, where the bulk reactant concentration, cB is low. The situation may be characterised by a mass transport coefficient, k,:

where j, is the limiting current density. Combining equations (37)and (44) gives an expression for the maximum rate of metal-ion removal:

dw = -CB kL A M dt

(45)

Equation (45)clearly indicates the importance of maintaining high values of I), cB, k, and A. The combined parameter kLA is a useful figure of merit and a central driving force in reactor design is the achievement of high values. Thus high surface area (per unit reactor volume) may be achieved using various three-dimensional electrodes while high rates of mass transport may be achieved by enhanced electrode/electrolyte movement via turbulence promotion

19

I

-J

I

I

est

I

-E

0 -E

Figure 4.

Idealised current density versus cathode potential curves for copper deposition from uncomplexed acid solution. (a) Showing the cathode reactions and the effect of cathode potential on the nature of the metal deposit. The maximum rate of metal-ion removal corresponds t o operation at the limiting current density. At a given metal-ion concentration, the maximum reactor (b) duty may be enhanced by increasing the rate of mass transport by, for example, elevating the temperature or by increasing the relative electrode/electrolyte velocity.

20

4.

CLASSIFICATIONOF ELECTROCHEMICALREACTORS

The continuous and wide spectrum of cell designs and reactor processes may be conveniently subdivided according to one of several arbitrary classifications: 4.1

reactor model and mode of operation

Selected modes of operation are summarised in figure 5 [271 while table 3 provides corresponding expressions for the fractional conversion, X, assuming complete mass transport control.

YJ

6; SIMPLE

BATCH

la1

REACTOR

PLUG FLOW REACTOR (PFR1 or CONTINUOUSLY STIRRED TANK REACTOR I C S T R ) I

I

t

Figure 5.

t

I

Common modes of operation for electrochemical reactors (a) Simple batch Batch recycle around a reactor and mixer tank loop (b) Single pass through a reactor (c) (d) Cascade of n identical reactors in series electrolyte flow

21

Table 3 Expressions for fractional conversion under full mass transport control Model reactor

I Mode of

operation

I Expression for fractional conversion, X,

Simple batch reactor

Batch

1 - exp-t(+)k A

Plug flow reactor (PFR)

Batch recycle

1 - {exp-f[l-

Continuous stirred tank Batch reactor (CSTR) recycle

1

{

exp-[y)]}

exp-f[ 1 -

11

1

~-

PFR

[

pass Single

1 - exp-(

++)I

CSTR

Single pass

PFR

Cascade of n identical reactors

1 - exp-n(kLA/Q)

Cascade of n identical reactors

1-

CSTR

1 1+(kL A / Q)"

For a simple batch reactor, the fractional conversion of metal-ion is given by:

X*=l-

C(t) C(0)

while the relevant expression for a continuous flow reactor is:

(46)

22

In many practical cases, a two stage operation is attractive, where the majority of metal-ion is removed from solution (as metal) in a p r i m a r y reactor, leaving the secondary reactor t o polish the residual solution. Two variants of this approach are shown in figure 6. It is also possible for one of the stages (particularly the polishing one) to be non electrochemical i.e. chemical or physical techniques for metal-ion separation may be used in concert with an electrochemical one[4, 18, 19,l.

PRIMARY 1000 ppm

SECONDARY

100 ppm

Ib)

t PRIMARY REACTOR

process a d d i t i o n s

I

STORAGE TANK

SECONDARY REACTOR

'O ppm ( i n t e r m i t t e n t )

continuous 1

Figure 6.

1

.-j--

1000 ppm

+10 ppm

I

Examples of two-stage reactor strategies for metal-ion removal. showing the overall and individual fractional conversions. (a) Continuous series flow through a primary ('roughing') and a secondary ('polishing') reactor. (3) Batch recycle via a primary reactor with intermittent secondary treatment.

23

42

Ehtmde geometry and motion

Following Goodridge [291, Walsh has classified reactors according t o the electrode structure motion [4, 8, 171. In the present case, the cathode may be static or moving with 2- or 3-dimensional character as a subdivision (table 4) and typical cell designs suggested by this classification have been considered elsewhere [4,5,8, 17, 201. Table 4 Classification of cell design according to the type of electrode Electrode Movement

Electrode Geometry 2-dimensional Parallel Plate - plate-in-tank - filter press

Static

Concentric Cylinder - Eberson cell Stacked Discs - capillary gap cell - perforated plate trickle tower

3-dimensional Porous electrodes

- stacked mesh - reticulated - cloth Packed bed electrodes - fibre - granules - sphereslrods - flakes Particulate trickle tower

Chemelec cell (inert fluidised bed)

Swiss Roll cell

~

Dynamic

Moving sheet or wire - reciprocating plate - vibrating electrode Rotating Disc (e.g. Pump Cell)

Active fluidised beds Moving beds - moving bed electrode - slurries - tumbled bed cell

Rotating Cylinder (e.g. Eco-Cell)

4.3

Nature ofthe metal product

From the practitioner's point of view, a major selection criteria for a cell design are the nature of the metal product [23] and its frequency of removal from the reactor. This important classification is illustrated in figure 7 for selected reactors.

h,

DISCONTINUOUS METAL RECOVERY

METAL ION CONCENTRATE

r/ REUSABLE

Most 3-13cathodes - HSA Reactor - Enviro-Cell - Turbocel 1

Chemelec Cell in electroplating

I

I

CONTINUOUS METAL RECOVERY I

METAL FLAKE OR POWDER

COMPACT METAL

RCE cells CEER Cell

RCE cells - Heraeus 'Band Cell - continuous belt electrodes

1

I

DEPOSITED METAL

METAL ON A COMBUSTIBLE SUBSTRATE

Parallel Plate - Recowin Cell - Hickman Cell - Retec cell

Many 3-d Cathodes - Retec Cell - Trickle Towers - Enviro-Cell

OR POWDER RCE cells MVH cell

-

Active FBE's - Billiton FBER Rotating or Vibrating Cells - Goecomet Reactor

Concentric Cylinder - Cyclone Cell

Figure 7.

A classification of electrochemical reactors ..according to the frequency of metal product removal and the nature of the metal deposit. Examples of proprietary devices are given (see tables 5 and 6 >;further information on these reactors is provided elesewhere [1,4, 7, 16,201.

I

P

25

It is obviously important to match the reactor type and its mode of operation to individual process requirements which may involve metal: (i)

in powder or flake form

(ii)

in sheevplate form

(iii)

deposited on a combustible substrate for refining

(iv)

for reuse as a soluble anode in electroplating

(v)

a pure metal or an alloy

(vi)

insoluble metal compounds e.g. oxides, hydroxides, sulphides

(vii) a metal-ion concentrate (viii) metal ions in a regenerated (reduced, oxidised or purified form)

The metal product may have several destinations, it may be:

(a)

recycled directly in a particular production process

(b)

recycled to another process

(c)

reprocessed in house

(d)

sold for refining o r as scrap.

6.

EXAMPLES OF REACTOR DESIGNS AND THEIR PERFORMANCE

6.1

Typical designs and applications

An extremely wide range of electrochemical cell designs has been described in the academic and patent literature [l,4, 7-24]. Relatively few of these have survived the rigours of scaleup to a pilot scale and fewer still have proved successful in a competitive marketplace. In this section, emphasis will be placed upon those reactors which are (or have been) in commercial use and which have, therefore, been applied to practical problems of metal-ion removal a t a reasonable scale or in moderately large numbers. Due to the size and scope of this subfield of electrochemical engineering, only a profile of certain designs will be given. Information on suppliers and other organisations is provided in a n appendix. More detailed information is available elsewhere [l, 4, 141 including data on reactor performance [24, 271. Table 5 summarises the design concepts and means of performance enhancement for selected reactors, a subdivision being made according t o the nature of the metal product and its frequency of removal.

Table 5 Examples of electrochemical reactors for continuous metal recovery Reactor

Organisation

Type of cathode

Frequency and method of product removal

Cell normally divided?

Performance enhancement n ainly via Electrolyte movement

'Band' Cell

Heraeus

Eco-cell*

(Originally Ecological Engineer ing) now MVH bv (Originally AKZO) Billiton Research bv

FBE* reactor Goecomet reactor

Goema AG

Ac tim ag Process

Actimag (also Darcy Products Ltd in the UK) Andco Environ mental Processes Inc

Chromate Reduction Process

Outer surface of a smooth, horizontal rotating cylinder Outer surface of a rotating cylinder with powdered metal deposit Fluidised bed of metal particles i n a tube and shell type geometry Rotating tubular bed or impact rod Dissolving base Vertical plate electrodes. Anodically

I I

movement

May be continuous as a sheet of metal

May be

May be continuous via automatic powder scraping and fluidisation

May be

Continuous via withdrawal of grown particles

Yes

Continuous as settled powder via a sludge cone Mav be continuous

No

J

I No

J

via sedimentation of precipitates

High active cathode area

J

J

J

I

J

Table 6 Examples of electrochemical reactors for discontinuous metal recovery. Reactor

Bipolar Trickle Tower Reactor CEER cell

Recowin

-I

Organisation

Type of cathode

Series k 6 0 ) of bipolar elements in a column

Finishing Services Ltd

Vertical plate

Eco-tec

Vertical plates in tank, air agitatation possible Inner surface of a cylindrical foil (copper or stainless steel) Vertical mesh (or plate) in an electrolyte with fluidised glass beads Rotating cylindrical foil (usually stainless steel) or static cylindrical foil with rotating anode (larger cells)

Engineers Ltd)

Performance enhancement mainly via

Cell normally divided?

Electrode movement

Electrolyte movement

British Technology Group

Systems

Frequency and method of product removal

Discontinuous via leaching o r removal of the carbon bed Discontinuous via extraction of metal flake/powder from a sump port or cathode Discontinuous via manual stripping of metal as sheet Discontinuous brass cathode may be furnace refined in the case of gold Discontinuous by manual scraping or reuse as anodes in ~~

No

J

Yes

/

No

J

~

Discontinuous by manual scraping or flexing

I No

No

No

I

J

I J

J

High active electrode area

I

Table 6 (continued) Examples of electrochemical reactors for discontinuous recovery of metal. Cell normally

I HSA

reactor Enviro-

I SCADA

I Systems

I Deutsche

Performance enhancement mainly via

I e.g. packed bed of carbon particles within a parallel d ate and frame ex. porous carbon figre matrix Contoured Dacked bed, of carbon, Dossiblv with nonconductive particles Vertical metal (or carbon) foam electrodes in a tank

Discontinuous via anodic or open-circuit leaching

I Discontinous via anodic

I or open-circuit leaching

I Discontinous via anodic

I

or open-circuit leaching or vacuum removal of bed Discontinuous cathode may be steel (which may be chemically stripped for reuse) or carbon for precious metals cathode may be reused as an anode in plating

29

Several comments may be made regarding the information in tables 5 and 6: The most common cell geometry is the plane parallel configuration which offers many advantages including availability of materials, flexibility and uniform flow and current distributions [30]. The concentric cylindrical geometry is preferred for some small scale operations. The majority of reactors achieve an improvement performance via electrode electrolyte movement and/or a high electrode area. Porous, 3-dimensional electrodes are capable of treating particularly dilute liquors while maintaining a reasonably high fractional conversion and current efficiency. However, care must be taken to avoid potential and flow distribution problems, or 'plugging' by metal. Electrode movement may be achieved by rotation of cylindrical electrodes (the anode or the cathode), or by vibration, impaction, fluidisation or a pulsed magnetic field. Electrolyte movement may involve trickle or (more usually) flooded flow through a porous matrix o r past a solid surface, air sparging or fluidisation via inert glass beads. Some reactors may be classified into several categories, depending upon the nature of the electrode geometry which may facilitate a choice of metal removal strategy. For example, porous, 3-dimensional carbon electrodes may be leached t o produce a metal-ion concentrate or the metal depositlcarbon matrix may be removed, followed by furnace refining in which the substrate is removed as carbon dioxide. Rotating cylinder electrode reactors may operate in several different modes, depending upon the electrolysis conditions and the nature of the cathode surface. For example, possibilities include: a)

intermittent recovery of smooth, compact metal deposits

b)

intermittent recovery of roughened metal flake or powder

c)

continuous recovery of metal foil or

d)

continuous recovery of metal flake or powder.

The majority of reactors involve the direct electrolytic deposition of metal. However, other possibilities include open-circuit cementation of metal or indirect, electrolytic precipitation of metal oxides and hydroxides.

A diverse range of process liquors can be treated, in terms of their source and the concentration of metal-ion. In the following section, the construction and performance of certain reactors developed in the one of the author's laboratories will be briefly considered. The

30

examples have been chosen from a large number of possible reactor designs and serve to illustrate diversity with respect to the electrode geometry and scale of operation. In addition, the reactors have shown a creditable performance during the electrolytic treatment of a wide range of metals and electrolytes. 622

Concentriccylindricalcell

The static cylindrical geometry offers a convenient cathode design for small scale operations and has primarily been used for precious metal extraction. Mass transport may be enhanced by the use of tangential manifolds and the use of a reasonable flow rate. An example of an inner concentric cathode is provided by the 'Cyclone Cell' which is shown schematically in figure 8a; this device has been used to extract gold from electroplating rinse waters [31] and silver fom photographic processing liquors 1321. The performance of a typical cell is illustrated in figure 8b which shows data for the removal of silver from various photographic fixer solutions under constant current conditions. The rate of concentration decay is seen to depend both upon the nature of the fixer and the batch processing time. At early times, the rate of concentration decay is fixed (i.e. the reaction rate is current-limited) but the process is mass transport limited at longer times and an exponential fall in the silver level is observed 11, 32, 331. The results also show the relatively high rate of redissolution of a fresh silver deposit under open-circuit conditions, particularly in the case of the bleach-fix solutions. The current efficiency for silver removal was relatively constant during the early stages of metal removal but it decayed rapidly during the later, mass transport controlled period. In the case of colour and black and white fixers, the rapid decrease in current efficiency may be partially prevented by potentiostatic operation as described elsewhere [341. In the case of bleach fix solutions, it is particularly difficult t o maintain a high current efficiency due t o the intervention of Fe2+/Fe3+redox shuttles (see equation (34)) although the use of a divided cell shows significant improvements.

6.3

RotatingcylinderelectmdeI.eactor

Reactor designs based upon a rotating inner cylinder in turbulent flow have been well described in the literature 135-371 and a wide variety of devices has been used for metal removal from chemical process liquors [32,371 and hydrometallurgical process streams [38-401. Despite the drawbacks associated with a rotating drive assembly and associated electrical power brushes, RCE reactors have exploited the following key characteristics 1371: (i)

The rate of mass transport is high due to turbulent flow and surface enhancement of microturbulence occurs in the presence of roughened metal deposits.

(ii)

The cathode surface has an enhanced surface area due to the deposition of metal in a roughened form and the cathode current density is reasonably uniform under mass transport controlled conditions.

(iii) Continuous metal extraction is possible via the deposition of a powdery

31

cell rop~

17

x / C o t h o d e feeder

Transporent (Perspex ) Elostorneric '0' ring seol -

Tangential inlet manifold

Anode feeder

0.145 rn

I

I

I

I

0

I

I

1

1

2

I

I

1

3

4

5 t/h

Figure 8.

Concentric cylindrical cathode cell [31, 323. (a) Schematic showing the construction. Silver concentration decays with four photographic fixer (b) solutions, showing the effect of fixer type and the rate of open circuit dissolution of a fresh silver deposit. A 20 d m 3 batch of electrolyte is processed at 25OC using a cathode of 20.5 cm diameter (A = 1560 cm2).

32

deposit which may be scraped from the cathode surface. (iv)

Diverse ranges of cathode size (0.0001 m2 to 4 m2>and current rating (10-3 A to 104 A) have been utilised in the simple batch, single pass, batch recycle and cascade modes of operation; the reactor behaviour is a good approximation to that of a CSTR 1361.

The continuous extraction of cadmium powder from an acidic sulphate liquor containing adversely high concentrations of zinc has been examined on a pilot plant (500 A) scale using the reactor shown in figure 9a and the arrangement in figure 9b. As demonstrated in table 7 [401, a high rate of cadmium removal was observed if the process conditions were suitably controlled. The results refer to single pass conversion measurements and the assumption of complete mass transport control. The current efficiency values are appreciably below loo%, primarily due to the redissolution of active cadmium powder via corrosion (cf equation (29) for zinc). The cadmium deposits typically had a purity of 98.7 - 99.2%by weight, the zinc level being ~0.3%. Table 7 Summary of results for cadmium removal from a zinc sulphate liquor RCE diameter: 22.9 cm; RCE area: 1630 cm2; rotation speed: 540 rev min-1; zinc concentration: 127 g dm-3; catholyte temperature: 6OoC

I

%*

‘(IN)

‘(OUT)

/A

/v

Ippm

/ P P ~

1.9 2.4

165 180

-1.16-1.35

450570

0.420.47

60-

300

79

200 270

3.2 3.3

26 34

-0.42 -0.54

110138

69 87

0.34 0.38

6775

130 163

3.7 4.8

23 35

-0.42 -0.50

105143

75110

0.220.29

48-

80 113

PH

*

260-

cathode potential vs Hg/HgaSOdl M K2SO4

‘A

%$

k,A 1 10-3m3 s-1

64

33

Figure 9.

A pilot scale rotating cylinder electrode reactor system for continuous extraction of cadmium powder from hydrometallurgical process streams 1401. (a) Schematic plan view of the reactor to show the construction. A polypropylene cell body; B cell flange; C rotating cathode; D cation exchange membrane; E nickel mesh anode; F catholyte inlet manifold tube; G catholyte outlet manifold; H reciprocating scraper; I removable inspection cover; J reference electrode probe. (b) The process flow system. P pressure gauges; RC,RA RCE reactor catholyte, anolyte compartments; HA, HC anolyte, catholyte heat exchangers FM flow meter; S gashquid separator; C hydrocyclone; AM,CM anolyte, catholyte tank; ISE ion selective electrode; ADDS chemical additions; CW cooling water; G gas vent; D to drain; F solid/liquid separator and collection tray; PA,PC anolyte, catholyte pumps.

34

6.4

Bipolart3ickletowerI1.eactor

Trickle tower reactors containing a number (typically <60) of vertically stacked, bipolar electrode layers have been studied for metal-ion removal at the University of Southampton [41-441. Earlier studies [41-431 employed layers of hollow carbon cylinders known as Raschig rings (see figure 10a). The relatively low active cathode area per unit reactor volume together with constructional difficulties associated with these packings led to the examination of alternative materials such a s felt, particles, foam and perforated plates 1441.

A prototype commercial reactor, developed in collaboration with Wilson Process Systems Ltd, (figure lob) has been used t o remove gold from a n alkaline, cyanide-based electroplating dragout solution under the conditions stated in table 8. Operation in the batch recycle mode,at a constant reactor voltage, resulted in removal of both gold and cyanide as shown in figure lOc, although the current efficiency for gold removal was relatively low ( ~ ~ 1 0 at %) the low metal concentrations involved. The fluid contacting pattern and potential distribution in bipolar trickle tower reactors has been shown to offer a particularly effective reaction environment for the breakdown of cyanide complexes which facilitates metal-ion removal [42]. Table 8 Details of the bipolar trickle tower reactor number of electrode layers:

2 x 37 = 74

number of feeder electrodes:

1positive and 2 negative

packing material:

hexagonal sheets of 4 mm thick, carbonised perforated pegboard

cross sectional area of packing:

75 cm2

total projected cathode area:

4600 c m 2

applied reactor voltage:

86 V

voltage per layer:

2.3 V

cell current:

1.8 A - 2.8 A

volumetric flow rate:

1.0 dm3 min-1

electrolyte volume:

5 dm3

pH :

10.9 - 9.6

temperature:

15OC - 6OoC

electrolyte conductivity:

17 mS cm-1- 22 mS cm-1

35

Electrolvte inlet Porous anodeleeder

Glass or plastic tube Conductive, bipolar layers of electrodes

Porous polymer separator lectrolyteflowsasa thin film

Porouscathode feeder

+INLET

1

s p r a y j t o p feeder

t

upper packing c e n t r a l , back t o back feeder lower packing outlet

-M

1

OOULE S

(b)

Figure 10. The Bipolar Trickle Tower Reactor (a) Early Raschig ring design [1,41]. An improved reactor design [441 which incorporates a (b) modular arrangement of replaceable cartridges, perforated plate electrode layers and a revised configuration for the feeder connections.

36

Figure 10. The Bipolar Trickle Tower Reactor (c) Concentration-time behaviour, showing the removal of gold, total cyanide (via the Liebig method) and free cyanide (via an ion selective electrode).

37

In many cases of metal-ion removal, there is a need to maintain a very high fractional conversion in a continuous flow-through reactor. This situation may be accomodated via a number of reactors in series flow; both CSTRs [36, 371 and PFR's [45] have been considered. The important requirement of design simplicity often dictates the use of a static cathode geometry incorporating porous, 3-dimensional electrode materials 1461. One example is reticulated vitreous carbon which has been extensively investigated in our laboratories [45491 and elsewhere, e.g. 150,511.

A reactor comprising of eight identical RVC cathodes (figure l l a ) has been used for cupric ion removal from an acidic sulphate solution. Each of the cathode segments was controlled at a potential of -0.500 V vs SCE in order to achieve completely mass transport controlled reaction conditions. Typical steady state results are shown in figure l l b . The cupric ion level was systematically reduced from an inlet value of approximately 10 ppm to an outlet concentration of approximately 0.1 ppm in a single pass through the cascade of eight cathodes. The reactor behaviour closely approximated to a cascade of PFR's and the performance is equivalent to an overall fiactional conversion > 0.99 with an overall current efficiency of 66%.

Figure 11. Reticulated vitreous carbon cell [451. (a) Schematic showing the construction; each cathode is fabricated from 100 ppi RVC and has dimensions: 5.0 cm (long), 5.0 cm (wide), 1.2 cm (thick). a cation exchange membrane; b RVC cathodes; c lead-antimony plate anode; d sample point; e reference electrode liquid junction; f cathode feeder; g polymer mesh separator; h viton gaskets

38

out

1

L/mm ( bl

Figure 11. Reticulated vitreous carbon cell. Copper concentration versus distance along the reactor. (b) Catholyte: deaerated 0.5 M Na2S04 at pH 2 and 298 K with a mean linear velocity of 0.35 cm s-1 and at a cathode potential of -0.500 V vs SCE.The solid line shows the predicted performance based upon a cascade of eight identical PFRs, each having a kLA value of 9.67 cm s-1.

6.

SUMMARY

Numerous electrochemical reactor designs have been described in the literature for the removal and recovery of a range of dissolved metals from both synthetic and industrial process streams. Many of these have been developed into successful pilot- and full scale devices as indicated in this paper. In addition t o the choice of reactor design (which includes decisions regarding electrode geometry and electrode/electrolyte motion), the following operational factors are seen to be important: (i)

The form of the metal product.

(ii)

The mode of operation with respect to both electrolyte flow and the frequency of metal product recovery.

(iii)

The electrolyte (or electrode) velocity, which affects mass transport rates.

39

The electrode potential, which affects the current efficiency and selectivity. The electrode polarity and fluid contacting pattern, which may markedly affect the rate of electrochemical reactions involved in the breakdown of metal complexes and the anodic oxidation of toxic species. The electrolyte composition, with respect to the concentration of metals, the type and number of metals present, their speciation, pH, additive levels and dissolved oxygen concentrations, together with electrolyte temperature. The possibility of redissolution of metal product due to corrosion, which reduces the observed current efficiency. The energy consumption of an electrolytic process. increasing implementation of available hardware depends upon favourable process experience and adequate documentation. In particular, it is important to achieve progress in the following areas: (a)

Published case studies in specific industrial sectors.

b)

Evaluation of suitable anode lanolyte and membrane combinations for specific applications.

(c)

Examination of mixed metal, complexed metal and other types of (comp 1ex) process liquor.

(d)

Comparative and quantitative studies of reactor performance.

(e)

Integration of electrochemical reactor(s) with other unit processes for separation, process recycling, effluent treatment and water purification.

ACKNOWLEDGEMENTS The author is grateful to many industrial organisations (see the appendix) who have provided technical information on electrochemical devices. The tutorialheview style adopted in this paper has been encouraged by several industrial colleagues and postgraduate research students. Part of the material in this paper was presented at the BNF 8th International Conference on "Environmental Issues: Their Likely Impact on the Metals Processing and User Industries, 15-17 May 1991, Amsterdam. Financial support for the reported studies has been provided by the British Technology Group, E A Technology Ltd, Wilson Process systems Ltd and the Science and Engineering research Council. The authors are grateful to the Science Faculty of The University of Portsmouth for financial support t o one of us (G W R) via a postgraduate bursary.

40

REFERENCES

1.

Pletcher D, Walsh FC. Industrial Electrochemistry, 2nd. Edn., Chapman and Hall, London, 1990; Ch. 7 "Water Purification, Effluent Treatment and Recycling of Industrial Process Streams", pp 331-384.

2.

Hart J. Processing, Jan 1990,47-49.

3.

WR Bowen. 1.Chem.E. Symp.Ser.1987: No 105:37-43.

4.

Pletcher D, Walsh FC, Whyte I. 1.Chem.E. Symp. Ser 1990;No 116: 195210

5.

Bockris J O'M, ed. Electrochemistry of Cleaner Environments, Plenum Press, New York, 1972.

6.

Kuhn AT, ed. Industrial Electrochemical Processes, Elsevier, Amsterdam, 1971.

7.

Kammel R, Lieber HW. Galvanotechn.,l977;68 : 413;1978;69:687.

8.

As ref. 1 but Ch. 2 "Electrochemical Engineering" p. 60.

9.

Scott K. Electrochemical Reaction Engineering, Academic Press, London, 1991.

10.

Heitz E, Kreysa G. Principles of Electrochemical Engineering, VCH, Weinheim, 1986.

11.

Coeuret F, Storck A. Elements de Genie Electrochimique, Tecdoc, Paris, 1984.

12.

Rousar I, Micka K, Kimla A. Electrochemical Engineering, Vols 1 and 2,Elsevier, Amsterdam, 1986

13.

Pickett DJ. Electrochemical Reactor Design, 2nd Edn., Elsevier, Amsterdam, 1979.

14.

Ismail MI, ed. Electrochemical Reactors - Their Science and Technology, Part A, Elsevier, Amsterdam , 1990.

15.

Fahidy TZ. Principles of Electrochemical Reactor Analysis, Elsevier, Amsterdam, 1985.

16.

Genders JD, Weinberg N, eds. Electrochemical Technology for a Cleaner Environment, The Electrosynthesis Co., New York, 1992, in press.

17.

Symp. Ser. 1982;83-12:314-366. Gabe DR, Walsh FC. Electrochem. SOC.

18.

Fleet B.Coll. Czech. Chem. Commun. 1988;53:No 6:1107-1118.

41

19.

Kammel R. NATO Conf. Ser.1984; 6-10, Hydrometall. Process Fundam., 617-628.

20.

Marshall RJ, Walsh FC. Surface Technology. 1985; 24: 45-

21.

Kreysa G. Metallobedache. 1981; 35: 211-219.

22.

Weininger J. A.1.Ch.E. Symp. Ser.1983; No. 229: 79: 179-187.

23.

Roberts GM. "The Role of Electrolytic Processes for Metal Winning from Dilute Solutions", Conference a t the University of Newcastle-Upon-Tyne, UK, July 1980.

24.

Houghton RW, Kuhn AT. J. Applied. Electrochem. 1974; 4: 173-192.

25.

Couper AM, Pletcher D, Walsh FC. Chemical. Reviews. 1990; 90: 837865.

26.

Rizet L, Noual P. "A Novel Reactor for the Production of Copper Powder", paper provided by Actimag S. A.,Z. I. de Cranves-Sales, B.P.23,74380 Bonne, France.

27.

Walsh FC. Bulletin of Electrochemistry. 1990; 6: 283-295.

28.

Walsh FC. Trans. Inst. Metal. Finish. 1991; 69: 155-157, 158-162.

29.

Goodridge F. Proc. 24th Int. Cong. Pure and Appl. Chem. 1974; 5: 19-30.

30.

Hammond J, Robinson D, Walsh FC. paper presented at the Meeting on Cell Design and Optimisation, Bad Soden, 24-26 September 1990; Dechema Monograph. 1991; 123: 279-316.

31.

Walsh FC, Wilson G, Trans. Inst. Metal Finish. 1986; 64: 55-63.

32.

Walsh FC, I. Chem. E. Symp. Ser. No. 98, 1986: 139-149.

33.

Cooley AC, J. Applied Photographic Science and Engineering 1982; 8: 171-182.

34.

Walsh FC, Saunders DE, J. Photographic Science 1983; 31: 35-42.

35.

Gabe DR, Walsh FC, J. Applied Electrochem. 1983; 13: 3-16.

36.

Walsh FC, Gabe DR, Trans. I. Chem. E., B 1990; 68: 107-114.

37.

Walsh FC, The role of the rotating cylinder electrode reactor in metalion removal, ch. 5 in ref. 16.

38.

Robinson D, Walsh FC, Hydrometallurgy 1991; 26: 93-114.

39.

Robinson D, Walsh FC, Hydrometallurgy 1991; 26: 115-133.

42

40.

Gabe DR, Walsh FC, Proc. Reinhardt Schuhmann Int. Symp. on innovative technology and reactor design in extractive metallurgy, eds Gaskell DR, Hager JP, Hoffmann JE, Mackey PJ, AIME, 1987: 775-794.

41.

El Ghaoui EA, Jansson REW, Moreland C, J. Applied Electrochem. 1982; 12: 59-68.

42.

Ehdaie S, Fleischmann M, Jansson REW, J. Applied Electrochem. 1982; 12:69-74.

43.

El Ghaoui EA, Jansson REW, J. Applied Electrochem. 1982; 12: 75-84.

44.

Lesch FM, Schiffrin DJ, Walsh FC, Bulletin of Electrochemistry, submitted for publication.

45.

Pletciier D, Whyte I, Walsh FC, Millington JP, J. Applied Electrochem., accepted for publication.

46.

Pletcher D. Walsh FC, Three-dimensional electrodes, ch. 4 in ref. 16.

47.

Pletcher D, Whyte I, Walsh FC, Millington JP, J. Applied Electrochem. 1991;21: 659-666.

48.

Pletcher D, Whyte I, Walsh FC, Millington JP, J. Applied Electrochem. 1991;21: 667-671.

49.

Walsh FC, Pletcher D, Whyte I, Millington JP, J. Chem. Tech. Biotech. 1992, in the press.

50.

Konicek MG, Platek GF, New Materials and New Processes 1983; 2: 232235.

51.

Wiaux JP, Nguyen T, Galvano-Organo Traitments de Surface 1991; 60: 587-593.

APPENDM: SOME ORGANISATIONSINVOLVED IN THE DEVELOPMENT

AND SUPPLY OF ELECTROCHEMICAL REACTORS FOR METAL-ION

REMOVAL This listing refers to devices and processes in this paper. More information on hardware is available elsewhere [1,161. ACTIMAG PROCESS Actimag Z.I.de Cranves-Sales 74380 Bonne France

also DARCY PRODUCTS LTD Invicta Works East Malling, Near Maidstone Kent ME19 6BP, UK

ANDCO ENVIRONMENTAL CHROMATE AND HEAVY METAL REMOVAL SYSTEM Andco Environmental Processes Inc. PO Box 988, Buffalo New York 14240, USA

43

'BAND" CELL Heraeus Elektrochemie GmbH Werk Freigericht Industriestrasse 7-9 6463 Freigericht 2, Germany CEER CELL Finishing Services Limited Woburn Road Industrial Estate Postley Road Kempton Bedfordshire, UK CHEMELEC CELL BEWT (Water Engineers) Limited Works & Laboratories Tything Road, Arden Forest Industrial Estate Alcester, Warwickshire, UK 'CYCLONE' CELL Wilson Process Systems Limited 9 Waterworks Road Hastings East Sussex TN34 lRT, UK DECHEMA Theodor-Heuss-Allee 25 Postfach 970146 6000 Frankfurt 97, Germany DEM CELL Electrocatalytic Limited Norman Way Severn Bridge Industrial Estate Portskewett, Newport Gwent, NP6 4YN, UK E A TECHNOLOGY LTD (formerly Electricity Research and Development Centre) Capenhurs t Cheshire CH16ES, UK ELECTROCELLS Electrocell AE3 Tumstocksvagen 10 S-18366 Taby, Sweden THE ELECTROSYNTHESIS COMPANY INC PO Box 430 East Amherst New York 14051USA

44

ENVIRO-CELL Enviro-cell Umwelttechnik GmbH Gattenhofer Weg 29 D-6370 Oberusel West Germany THE ELECTRODE POREUSE PERCOLEE PULSEE ('E3P) REACTOR ENSIGC-Lab-du Genie Electrochimie Chemin de la Loge 31078 Toulouse Cedex, France HSAREACTOR SCADA Systems Inc 44 Fasken Drive, Rexdale Ontario MGW 5M8, Canada IONSEP ELECTRODIALYTIC PROCESS Ionsep Corporation Inc PO Box 258 Rockland, DE 19732, USA

MVH CELL MVH bv Prior van Milstraat 4 5402 GH Uden, The Netherlands RECOWIN CELL Eco-Tech (Europe) Limited Units 5C, D & E, Chase Park Industrial Estate Ring Road, Chase Terrace Walsall WS7 SJQ, UK RETEC CELL EES Corporation (an Eltech Systems Company) 12850 Bournewood Drive Sugar Land Texas 77478, USA SE-REAKTOR GOECOMET (GOEMA) Dr. Gotzelmann K. G. Postfach 995 Monchaldenstrasse 27A D-7000 Stuttgard 1, Germany TURBOCELREACTOR A joint development by: Sessler Galvanotechnik GmbH Industriestrasse 9 D-7538 Keltern-4, Germany

and WERNER F L U H M A " AG Postfach 630 Ringstrasse 9 CH-8600 Dubendorf, Switzerland

45

ELECTROCHEMICAL METHODS I. Roular'

FOR

PURIFICATION

OF

WASTE

WATERS

and K. Mickab

"Department of Inorganic Technology, Institute of Chemical Technology, 166 28 Prague 6, Czechoslovakia b

J. Heyrovsk2 Institute of Physical Chemistry and ElectroPrague 8, chemistry, Czechoslovak Academy of Sciences, 182 23 Czechoslovakia

INTRODUCTION The purification of waste waters must always be carried out to such a degree that the corresponding lawful norms be satisfied; only then the purified water can be discharged into the surroundings. Eventual differences between the lawful norms of various industrially developed countries are not significant. These norms indicate the permissible concentrations of contaminants besides other parameters (colour, odour, content of microorganisms, salinity, etc.); and they are currently supplemented to include further substances, whose toxicity has been determined in recent years. It should be noted that, analogously to the exposition to nuclear radiation, there is no lower limit under which the dose of the toxic agent can be considered as harmless for human organism. The noxious effects of very small doses of radiation or toxic compounds on the human organism usually cannot be separated from the influence of the "natural background", comprising all environmental factors. Similarly to low doses of nuclear radiation, to which the human body can be exposed (in some cases) repeatedly many times, low repeated doses of toxic compounds can accumulate in human body and their effect may be manifested only after several years. A typical example is the Minamata desease, caused by organic mercury compounds, or absorption of the insecticide DDT in human fat tissues and its subsequent setting free with all negative effects after losing weight. Cadmium, also a dangerous environmental pollutant, has been reported to interfere with the metabolism of some mineral sub-

46

stances, mainly calcium [l], magnesium [2], and phosphorus [3]. Interaction of cadmium with the metabolism of manganese has also been reported [ 4 ] . Moreover, cadmium can replace zinc in certain enzymes [5]. Easy replacement of Ca2+ ions in calcium-binding proteins by Cd2+ ions is made possible by their equal charges and similar ionic radii (0.097 nm for Cd2+ and 0.099 nm for Ca2+) [ 6 , 7 ] . Exposure to cadmium results in disturbances in calcium homeostasis in the body. Characteristic symptoms of cadmium poisoning, such as osteoporosis, hypercalciuria, and altered protein synthesis indeed strongly suggest that cadmium interferes with calcium-dependent processes [5,8]. Traces of heavy metals in waters are of industrial origin (waste waters from mining and treatment of ores, from metallurgical plants and rolling mills, metal finishing, photographic shops, from textile, leather, and chemical industry, and even from agriculture (e.g. cadmium from African phosphate fertilizers). An additional source are atmospheric condensations contaminated by exhalations from combustion of fossile fuels and from motor vehicles. The toxic metals are present in water in the form of ions or complexes (organic or inorganic). Lead, cadmium, and mercury are most important; they can be accumulated in sediments. Their maximum allowable concentrations in water supplies for water works in Czechoslovakia were determined by an official regulation in 1975 as 0.05, 0.0001, and 0.01 mg/l for lead, mercury, and cadmium, respectively, and in other surface waters as 0.5, 0.005, and 0.3 mg/l, respectively [9]. The three metals are especially dangerous for fish; their acute letal concentrations are indicated in Table 1. It is important to note that mercury can be accumulated in fish bodies, unlike cadmium and lead. Significant accumulation of mercury and cadmium has been observed in edible mushrooms, in contrast to lead, especially with the genera Agar caceae and Tricholomaceae. According to the World Health Organization, the acceptable daily intake (ADI) per person (weighing about 60 kg is 0.5 mg

41

Table 1 Acute letal concentrations (mg/l) of Cd, Hg and Pb for fish in surface waters. Maximum allowable concentrations are given in parentheses. Data according to ref. [9]. Cd Salmon-like

0.001-0.09 (0.0002) 0.24-3.2 (0.001)

Carp-1ike

-

Hg 0.3-1.0 (0.001-0.002) 0.2-4.0 (0.001-0.002)

Pb 1.0-10 (0.004-0.008) 10-100 (0.07)

of lead, 0 . 0 5 0.07 mg of cadmium, and 0 . 0 4 mg of mercury (in the form of methylmercury). The data referring to the desired quality of discharged waste waters according to the U . S . , Swiss, German, and Czechoslovak norms are summarized in Table 2. It can be seen that the highest permissible concentrations of heavy metal ions are in the range from 0.05 to 1.0 mg/l. The attainment of such low concentrations by means of electrochemical methods would obviously require electrodes of very large surface area, resulting in high investment and operating costs. Therefore, for heavy metal ions concentrations below 10 mg/l, other methods were proposed, namely precipitation, which is usually the cheapest one, sorption on special sorbents, or ion exchange (the metal ions are usually replaced by hydrogen or sodium ions; complex metal anions are replaced by chloride, sulphate, or other anions); and, finally, extraction of heavy metals with liquid reagents. In the latter case, special reagents are sometimes used to decrease the heavy metal concentration below 0.1 mg/l. The operating costs for purification of 1 m3 of waste water containing heavy metal concentrations of the order of 10 mg/l range from 0.50 to 3 DM according to the method used; generally the costs decrease with increasing size of the equipment or with increasing rate of flow of the waste water. The purified waste water must also satisfy the norm regarding salinity, i.e. the total content of salts. Absorption of ions in ion exchangers, which must be generated by using a four to ten-fold molar excess of salt with respect to the number of

Table 2 Maximum allowable concentrations (in mg/l) of contaminants in discharged industrial waste waters. Data taken from ref. [lo]. USA PH Suspensions NH3

P (total) CN- reacting with chlorine CN- total F A1 AS Ba Cd Cr6 + cr3 + Cr total cu Fe Pb Mn Ni Ag Zn Hg Active C1 NO; Org. solvents Phenols Petroleum and its fractions Hydrocarbons Se

v

Chlorinated hydrocarbons Fats Sulphate ions BOD COD Inorg. salts Total solutes 1

Switzerland

6 - 8.5 10 1.0 0.6

6.5 0.3

0.03 0.53 2.0 0.2 0.05 1.0 0.1 0.05 0.05 0.25 0.2 0.5 0.05 1.0 1.0 0.0 0.5

0.1

----

-

9

Czechoslovakia

--7

lo

--

-- -

0.5

lo 10 --

0.2

-2.4 -0.1 1.5 0.01 0.1 0.5

lo 1.0 0.1 2.0

--

--

0.1 --

1.0 1.0 1.0

0.1

--

2.0 0.1 2.0 0.1 0.5 3.OZ2 1.0 10 traces

-

-

--

--

lo -

--

20

-lo

10 -0.05 0.1 0.005 55 400 1000 2000 1000 3000

Higher iron content is occasionally allowed. 2 Higher values apply if the waste water is discharged into drainage. 3 Biological oxygen demand after 5 days. 4 Chemical oxygen demand.

49

moles of absorbed heavy metals, leads to a considerable increase in salinity of the treated water. If a low salinity has to be achieved, it is advisable to use, e.g. in galvanic processes, anions that can be either isolated at the anode in the form of a gas (such as chlorine) or decomposed by microorganisms in a biological waste water treatment plant. All processes for waste water treatment are expensive and cause an increase of manufacturing costs and of prices. Nevertheless, the treatment of waste waters is in the focus of interest of the whole society, hence legislative provisions and their checking are necessary. Modern approach to purification of waste waters containing heavy metal or other substances assumes that water of a certain provenience is supplied that must be freed from a particular metal or compound. The pH value can be adjusted after mixing together all charges of waste water purified separately. Such an approach permits the isolation of a metal or a sufficiently pure compound suitable to be reused in production. The classical method of purification of all waste water charges mixed together and containing many heavy metal ions by the addition of lime ensured low concentrations of these ions in the discharged water, but the side product was sludge containing heavy metal hydroxides and lime, which was deposited on dumps. (The number of such dumps in Czechoslovakia is estimated as several thousands.) In this way, the heavy metal ions were dissipated into the ground and, through the nutritional chain, they appeared in foodstuffs. Certain countries, e.g. Sweden, attempt to prevent the dissipation of toxic metals, especially cadmium, from dumps of waste metals by prohibiting the production and import of cadmium-plated metal parts. ELECTROCHEMICAL BACKGROUND FOR THE DEPOSITION OF HEAVY MgTALS The equilibrium electrode potential sition according to the equation

for cathodic metal depo-

50

MeZ+

+

ze- = Me

can be calculated from the Nernst equation E (MeZ+/Me)= Eo (MeZ+/Me)+ zF In a(MeZ*)

(2)

where Eo (MeZ+/Me) denotes standard potential of the electrode is the gas constant, T is the reaction (l), R = 8.314 J/mol K absolute temperature (in K) and F = 96 487 C/mol is the Faraday constant. The activity of metal ions, a(MeZ+) , can in approximate calculations be replaced by their concentration, c1 (in mol/l), the difference being negligible in dilute solutions. The electrode reaction proper (1) may be considered as very rapid, especially when the electrode potential is considerably more negative than the equilibrium one. Therefore, the ratecontrolling step is the transport of Me2+ ions from the bulk of the solution to the surface of the cathode. The rate of this transport depends on the thickness of the Nernst diffusion layer, which is involved in the coefficient of mass transport; the two quantities depend, in turn, on the rate of stirring of the solution near the electrode. Thus, the rate of stirring controls the limiting current density, i.e. the current density at which the concentration of the metal ions at the cathode surface approaches zero (cis+ 0 ) . The mentioned quantities are interrelated by the following equations: c -c j = -zFD1 l a 1 s

(3)

6

'1

i m

= -zFD

1

l a =

6

zFkm

a

(4)

wh j denotes the current den ity, j l i m its limiti g value (in A/m2 ) , D1 (m2/s) the diffusion coefficient of Me2+ ions, k (m/s) coefficient of mass transport, and 6 (m) thickness of the Nernst diffusion layer. The subscripts, a and s, refer to the analytical (bulk) and surface concentrations, respectively.

51

In the theory of transport in electrolysers, it is convenient to introduce a dimensionless criterion called the Sherwood number and defined as

where Lc (m) is a characteristic length of the system under study. During current passage, the electrode potential differs from its equilibrium value, E r , since the concentration of metal ions at the surface, cis, is in this situation lower than that in the bulk, cia. Since the ratio of cls/cla can be expressed from equations ( 3 ) and ( 4 ) as l-j/j,lm, we obtain for the electrode potential

I - =I

E = Er(MeZ+/Me)+ zF In 1 RT

1

(7)

where EP (Me" /Me) is given by Eq. ( 2 ) and the second term represents the so-called concentration polarization. The course of the polarization curve is shown schematically in Fig. 1. In practice, a situation is often encountered in which a higher current density than the limiting one is used. This leads to evolution of hydrogen bubbles at the cathode already at zero V (SHE) if the pH value is in the acidic region. The partial current density corresponding to the reaction 2H'

+

2e- = H2

(8)

increases with increasing negative potential; and the solution layer adjacent to the cathode is stirred by hydrogen bubbles, causing an increase of the mass transport coefficient, k l , and thus of the limiting current density for metal deposition. The partial current densities for the two processes are depicted schematically in Fig. 2 . This shows that if the limiting current density is surpassed, the electrode potential becomes more

52

0.4

I

I

I

I

I

1

I

Y -

-

0.

-

-0.4-

-0.8

-

-

-1.2 I

I

I

I

I

I

I

Figure 1. Cathodic polarization curves for a rapid electrode reaction calculated from Eq. ( 7 ) , where j / l j l i m l= y, Er= 0.34 V. (a) z = 1; (b) z = 2.

I

I

0

E

Figure 2. Partial polarization curves for (a) hydrogen evolution and (b) metal dissolution; (c) polarization curve due to superposition of the two processes. The limiting diffusion current for curve b is influenced by stirring with hydrogen. negative, hydrogen is evolved and the current yield of the electrolysis decreases. At current densities lower than the limiting one, a charge equal to zF is needed to obtain one mole of the metal from its ions; this is equal to 26.82 ampere hours. At current densities higher than the limiting one, a considerably higher charge is

53

needed, as follows from the above considerations. current yield, q I , is given as

Then, the

where the total current density, j t o t , involves the parasitic current density, j p a r ,corresponding to the evolution of hydrogen (and/or other side reactions). Thus, it can be seen that the two quantities play an important role in evaluating the economic aspects of the whole process. Consider a flow-through electrolyser with plate-shaped electrodes in a rectangular channel (Fig. 3). In this case, the Sherwood number is given as

[

: ill3

Sh = 1.85 ReSc 2 for Re < 2300, and

Sh = 0.023 R ~ " * S C ~ ' ~ for Re > 2300, where the dimensionless criterion Re

= vLc/ve

is the Reynolds number, and Sc = v e / D 1

(13)

is the Schmidt number, v (m2/s) denotes kinematic viscosity of the electrolyte solution, and C (m/s) linear velocity of the solution between the electrodes. The characteristic length, L , in Eq. (10) is equal to 2d for d << w (see Fig. 3). As an example, the limiting current for the deposition of copper is plotted in Fig. 4 as a function of the concentration of Cu2+ ions for D = 6x1O-l0 m2/s, d = 0.01 m, L = w = 1 m, v = ~ X I O m2/s. ~ Thus, if the concentration of Cu2+ ions ought to be decreased in a flow-through electrolyser with recirculation from 100 g/1 to 1 mg/l at the highest current yield of 100 %, it would be

54

N

Figure 3 . Scheme of a flow-through channel electrolyser with plate electrodes; d distance between electrodes, L electrode height, w electrode width, v linear velocity of electrolyte. Figure 4 . Dependence of the limiting current density on the concentration of copper ions; 1 Re = 400, 2 Re = 4000. Electrolyser according to Fig. 3 ; L = w = 1 m, d = 0.01 m.

necessary to monitor their concentration and to lower the current density in proportion to the decreasing concentration. This, however, would be impractical. In practice, the terminal voltage of the electrolyser is usually kept constant; if the electrolyser is sufficiently long and the electrolyte has a sufficient conductivity, the current is also constant; and the contemporary current supplies are capable to maintain a constant current automatically. However, if a constant current is maintained from the beginning of the metal deposition, corresponding to the initial limiting current density and initial concentration of 100 g Cu2+ per litre, then at the end of the electrolysis the ratio of the applied current density to the limiting one will be equal to lo5 and the current yield will drop to 0.001 %. To improve the current yield, it is necessary in practice to

55

make a compromise: firstly, the concentration is decreased in an electrolyser, usually by a factot of 10 to 100, and then an electrode with a large surface area is used to attain the final concentration. In this case, the limiting current density is attained at the end of the electrolysis. Another problem that must be taken into account with plate electrodes is the formation of metal dendrites which grow from the cathode surface toward the counter electrode, causing eventually a short circuit. If a diaphragm is placed between the electrodes, the dendrites can pass through it and they finally may destroy it. The region of Zn or Cd dendrite (or sponge) formation lies on the polarization curve below the limiting current density for deposition of zinc from zincate solutions and cadmium from CdS04 solutions. It seems that the formation of dendrites is favoured by increasing pH in the diffusion layer caused by simultaneous evolution of hydrogen; the formation of hydroxides or basic salts can increase the dispersion degree of the powder [ll]. In most cases, additional metal ions are present, mainly those with a positive standard potential, which are deposited preferentially at the cathode and have a low overpotential for hydrogen evolution. On the other hand, zinc sponge is deposited even at relatively low current densities during electrolysis of ZnSOI solutions in the presence of Cu, As or Sb ions; these are nobler than zinc and are deposited at the maximum rate (i.e. at their limiting current) in a powdery form, and the zinc deposited on them "copies" their powdery structure. Powdered metals (Fe, Cu, Ni) are usually deposited at the cathode at current densities higher than the limiting one with simultaneous evolution of hydrogen. Metal powder or sponge is also often formed during deposition of metals from their complex anions, e.g. from cyanide solutions, with simultaneous evolution of hydrogen. The formation of powdered metals in a diaphragmless electrolyser may cause a decrease in the current efficiency: the oxygen evolved at the anode passes into the solution and the sus-

56

pension of the powdered metal (Cu, Fe) can corrode: Me

+

O2

+

2H'

= Me2+

+

H20.

(14)

Moreover, in acidic medium some metal powders can react directly with hydrogen ions. Therefore, the formation of powdered metals should be avoided in purification of waste waters, except for some special cases. It is usually assumed that the metal is deposited on an electrode that must be replaced during the process, and that the anode is insoluble. The anodic reaction is oxidation of water to oxygen and/or oxidation of anions, e.g. CN-. For aqueous solutions with a low salinity, titanium anodes coated with Ti02 and Ir02 or with platinum are mostly used. It is apparent from the above considerations that the limiting current density for the given experimental arrangement and concentration of electroactive ions should be known in order to propose a suitable working regime. In the text below, we shall proceed from the simplest electrolyser types to the more complicated ones.

PLATE ELECTROLYSER

The simplest arrangement of a plate electrolyser is shown in Fig. 3, where the electrodes are placed on the channel walls. Alternatively, it is possible to place several plate electrodes in one channel (Fig. 5); then equations (10) and ( 1 1 ) will give the lower estimate of the limiting current densities. The true limiting current densities will be higher owing to the increased trurbulence at the leading edge. The arrangement of plate electrodes in the channel can also be bipolar; a typical example is shown in Fig. 6. It is apparent that the critical point is the construction of the bipolar electrodes, since the desired metal must be deposited on one side, while oxygen evolution or oxidation of anions or inorganic compounds proceeds on the other side. The manufacture of such bipolar electrodes is complicated and, of course, their

57

0-

Figure 5 . Monopolar plate electrolyser with five channels. The volume rate of flow through each of them is equal to V/5. Figure 6. Bipolar electrolyser.

arrangement of

electrodes

in

a channel

disassembling is time-consuming. For this reason, electrolysers for waste water purification using such electrodes are not available on the market. Another problem is due to parasitic (bypass) currents, which flow from the first anode, pass through the electrolyte outside the pack of bipolar electrodes, and enter the last cathode. In an extreme case, it may happen that no current passes through the innermost bipolar electrodes. This problem has extensively been analysed in the literature [12-141. Naturally, the parasitic currents vary in a broad region with varying electrolyte conductivity and total current flowing through the system. They can be suppressed in practice by inserting plates of insulating material above and below the bipolar electrodes parallel to the channel walls. The insulating plates divide the channel into several narrower channels. (Four such channels would be formed in the electrolyser shown in Fig. 6.) In this way, the resulting current efficiency is improved. Some technical applications involve a combination of plate and rod electrodes, or rod electrodes alone. Rod-shaped anodes are preferred if a higher anodic current density is needed,

58

compared to the cathodic one (e.g. in the destruction of cyanides). A possible modification of this electrolyser type is formed by a cylindrical cathode with a central rod-shaped anode placed in a cylindrical channel (Fig. 7). The Sherwood number for the cylindrical cathode can be calculated as Sh = 1.62(ReSc dc/L)1'3

(15)

for Re > 2100, where dc = d o u t -din. In the turbulent region, it is possible to use Eq. (11) for a rectangular channel. In both cases, Re is given by 'Eq. (12), where Lc= dc. The terminal voltage of the electrolyser under consideration is given by the equation U = Ea(ja)

-

do u t Ec (jc) + 0.5doutpejcln

(16)

in

where the subscripts a, c, and e refer and electrolyte.

to the anode, cathode,

ELECTROLYSER WITH A FLUIDIZED BED An increase of the limiting current density can be achieved by using a fluidized bed of inert particles (e.g. glass beads) with diameters of several tenths of a millimeter up .to about

Figure 7. Cylindrical electrolyser with a cathode placed on its wall and with a coaxial rod-shaped anode.

59

1 mm. The moving inert particles forming the fluidized bed sur-

round the electrodes and disturb the diffusion layer. If their concentration is suitably chosen, then the Sherwood number is increased even when the rate of flow of the electrolyte solution is relatively low, i.e. for relatively low values of the Reynolds number (referred to the total cross section of the interelectrode space). The pressure losses in a fluidized bed of inert conducting particles can well be estimated from the approximate equation

*P

= (l-c)(Pp- p e ) g L

(17)

where E denotes the so-called voidage of the bed (an analogue to porosity), equal to the ratio of the liquid volume to the combined volumes of the liquid and the particles, L (m) is the bed height, g = 9.81 m / s 2 is the acceleration of gravity, p P (kg/m3) is the density of the particles, and p e that of the electrolyte. For spherical particles of equal diameter in the motionless state, E 0.4. It is convenient to introduce the Reynolds number referred to the particle diameter, d : P

v

where is the "mean nominal" velocity of electrolyte (based on the total cross section of the channel) and V denotes the volume rate of flow, equal to the product of G I channel width w and wall distance s. The Reynolds number satisfies the following theoretical relationship [15]: Re

Ar

=

18

+

E

~

0.6(Ar

* E

~ ~

10

' 5'

~ ~

~

where Ar denotes the Archimedes number, defined as Ar =

d; ( P p - P e 14 2

(20)

Pe 'e

The minimum velocity at which

the particles start to move is

60

called the threshold velocity, v t h i it can be estimated from Eq. (19), where E is set equal to 0.4. Equation (19) holds good for Ar E <60; 1000> and E E <0.5; 0.95>; it can be used to calculate the voidage, c , if Re is known from Eq. (18). From theP se data, the Sherwood number can be oalculated in turn [16,17]:

.

ShP = 1.24 S C * ’ ~ ( ~ - - PE ) /~c * ~ ~ R ~ ~ * ~ ~ ( 2 1 ) This criterion is related with the limiting current density (compare Eq. (5)) on the surface of the particles: ShP = - j l m d p/cI zFDl. ~

(22)

It follows from Eq. (21) that the Sherwood number (and hence the value of jli,) has a maximum value at a certain value of The advantages of a fluidized bed of inert particles can best be judged from the dependence of the limiting current, j l l r , on

c.

the streaming velocity, v, shown in Fig. 8 (assuming glass beads with d = 0.5 or 1 nun and p = 2624 kg/m3). It can be seen P

P

40

t

0.00

i 0.04

0.08

0.12

1

0.16

Figure 8. Dependence of the limiting current on the streaming velocity in a fluidized bed electrolyser. Particle diameter dp(mm): 1 0.4; 2 0.5; 3 0.6; 4 0.7; 5 0.8; 6 0.9; 7 1.0; 8 1.1; 9 1.2. Parameters for Eqs.(l8)-(21): p = 2624 kg/m3, P p e = 1130 kg/m3, u e = 1 . 3 ~ 1 0 -m2/s, ~ g = 9.81 m / s 2 . Curve 10 corresponds to the same case but without a fluidized bed, Eqs. (10)-(12).

61

that in a certain region of linear velocities the limiting current densities are at least an order of magnitude higher than in the same channel without a fluidized bed. The electrolyser must be provided with a grid or mesh forming the bottom of the bed of inert particles, and with another such a grid at the outlet, preventing the particles from escaping into the reservoir. The whole equipment is shown schematically in Fig. 9. To maintain constant pH of the solution, a glass electrode and a device for controlled addition of an acid or a base can be placed in the reservoir. Electrolyser of this type can be used to remove metals from their dilute solutions. For example, an electrolyser of the firm CHEMELEC (England) permits the concentration of heavy metal ions to be lowered to 100 mg per litre at an acceptable current yield (tens of per cent). The metal deposited on a plate or rod-shaped cathode can be used as an anode in galvanic plating. There were some doubts about the purity of the recovered metal, which was supposed to contain microscopic glass particles, presumably from disintegrated glass beads. If this

Figure 9. Scheme of an electrolyser with a fluidized bed of inert particles. 1 Electrolyte, 2 inlet grid, 3 outlet grid, 4 flow meter, 5 valve, 6 pump, 7 reservoir.

62

were true then during exploitation of such a metal as anode in a galvanic bath the glass particles might be transported to the cathode and built-in in the deposited metal. However, the manufacturers of fluidized bed electrolysers exclude such a possibility, since the only loss of the glass beads occurs during maintenance of the electrolyser and not by disintegration during operation. OTHER DIAPHRAGMLESS BLECTROLYSERS

Many other diaphragmless electrolyser types are known, where an intense stirring takes place under more or less defined hydrodynamic conditions. One of them is the electrolyser with a rotating drum, on which the desired metal is deposited in the form of a foil that is continually removed (Fig. 10). Another electrolyser is provided with a rotating cylindrical cathode that operates at high current densities causing deposition of metals in the form of a powder, which is removed by a brush and carried away by the streaming electrolyte. The metal powder is separated from the liquid by means of a filter. An industrial type of this equipment is known under the name ECO CELL. Still another electrolyser involves a system of rod-shaped

1

2

3

Figure 10. Principal scheme of an electrolyser with a rotating drum. 1 Deposited metal layer, 2 cutting knife, 3 metal foil, 4 electrolyte.

63

cathodes arranged on the circumference of a rotating disc (Fig. 11). A cylindrical anode is placed at the inner circumference of the electrolyser, and an additional anode can be situated within the system of rotating cathodes. Such an equipment is manufactured, e.g., by Gotzelmann KG (FRG) under the trade name SE Reactor Geocomet Cell. As an alternative, the system can be equipped with a diaphragm separating the positive and negative electrodes. It is used, for example, for the deposition of gold from cyanide solutions with simultaneous oxidation of CN- ions at a graphite anode.

Figure 11. Scheme of the system of rotating cathodes.

ELECTROLYSERS INVOLVING A DIAPHRAGM The term "diaphragm" will be used to denote a separator that prevents spontaneous mixing of the anolyte and catholyte, although it is porous, so that it permits ionic current to pass through. The diaphragm should not have the properties of an ion exchanger, hence suitable materials are PVC, polyethylene, ceramics, and asbestos (although the latter has the character of an ion exchanger with a low content of bound groups). The requirements to prevent mixing of the electrolyte and to permit easy current conduction by ion transport through the pores are obviously contradictory and they require a suitable compromise to be made. In some cases, the diaphragm is used to prevent contact between a fluidized bed cathode of conducting

64

particles and the counter electrode. The advantage of such systems consists in the large surface area of the fluidized bed cathode in a relatively small electrolyser volume. The fluidized bed cathode is used, e.g. , to remove C u 2 + ions from etching baths employed in the treatment of copper objects. The acidic etching bath is regenerated by electrolysis and, at the same time, oxygen is evolved at the anode (platinum-coated titanium), the anolyte becoming acidic. The separation of the anolyte from the catholyte is important, since the oxygen from the anode would otherwise lower the current yield at the cathode. The current yield with respect to copper increases with decreasing concentration of H2S04 in the bath. An electrolyser with a fluidized bed of copper particles and with several Ti/Pt rod anodes enclosed in cylindrical diaphragms of an organic polymer is shown schematically in Fig. 12. Electrolysers with a fluidized bed cathode are constantly under investigation, since the possibility of attaining a large cathode surface area in a relatively small volume is attractive, however nothing is known about their technical exploitation. This has several reasons. During electrolysis, the particles get into contact with the auxiliary cathode for a rela-

Figure 12. Basic scheme of the electrolyser developed by Akso 1 Rod anode, 2 diaphragm, 3 one Zout Chemie Nederland B.V. of Cu particles in suspension streaming upward, Q cylindrical cathode.

65

tively short time, so that a fraction of the current flows directly through the auxiliary cathode, causing its thickness gradually to increase. To eliminate this obstacle, a current density is used that surpasses several times the limiting current density on the auxiliary plate cathode, so that copper is deposited on it in the form of a powder that eventually separates and is carried away by the streaming electrolyte and filtered off. Another obstacle is that the bed particles increase in volume during electrolysis, and therefore the rate of flow of the electrolyte must gradually be increased to keep the bed in a fluidized state. After the particles have reached a certain size, they must be separated and replaced by a new filling of small particles. The recommended parameter values for electrolysis of waste water solutions containing about 100 - 150 g of H2S04 and 5 50 g of CuS04 per litre can be estimated from the literature 0,3 A/m2 or [18,19] as follows: Cathodic current density 0.1 10 30 A per litre, anodic current density at least 0.1 A/m2, voidage of the fluidized bed c = 60 - 75 %. The consumption of 2 kWh/kg of Cu. This electrolyser electric energy attains 1 type has been developed probably to the highest perfection at AKZO ZOUT Chemie Nederland B.V. [20]. Practical tests were carried out with solutions of 0.5 - 1.0 M H2S04 containing low concentrations of Cu2+ ions. According to Fleet [21], the concentration was lowered from 2000 to 27 mg of Cu per litre at a current efficiency of 90 %. Fleischmann et al. [22] removed 99 % of copper from solutions containing 6 mg of Cu per litre in the absence of oxygen. Also other authors were successful, attaining final Cu concentrations arround 1 mg/l at a current yield decreasing from 80 - 90 % to about 30 % when the Cu concentration dropped below 100 mg/l [23-261. For very low concentrations (of the order of 1 mg/l and lower), electrolysers have been designed with a large-surface area cathode, which is, in substance, a flow-through porous electrode. This is realized in various ways, e.g. by a graphite felt

-

-

-

-

66

or by several layers of a graphite fabric, or by metallized polyurethane foam (electrolysers RETEC of the firm Eltech, Switzerland). Fixed beds of a powdered metal or graphite were also proposed. A common feature of these cathodes is that their pores become gradually clogged by the deposited metal and their functioning is thus impaired in the course of operation. The electrodes that approached the end of their service life are worked up by burning the graphite or the organic polymer skeleton. Attempts to regenerate them electrochemically showed that the metal dissolution is nonuniform and therefore requires very low current densities, hence chemical dissolution in acids is the preferable way. The service life of the porous cathodes can be somewhat affected by the direction of the electrolyte flow through the pores: if the electrolyte flows from the cathode (i.e. from its “rear” side) to the anode, the current distribution in the pores is more uniform, hence the pore ends on the anode side are less loaded by the current and their clogging sets in later than in the case where the electrolyte flows in the opposite direction [ 2 7 1 . Modular units with porous cathodes have been introduced by

43 2

\

0-

t a

/’ -@

b

Figure 1 3 . Schemes of the (a) flow-across and (b) flow-through Anode, 2 supporpatterns used by Electrocell AB, Sweden. 1 ting metal mesh, 3 diaphragm, 4 porous electrode.

67

the Swedish firm ElectroCell AB; they employ a membrane and electrolyte flow either normal to the electrode surface (socalled "flow-across pattern", Fig. 13a) or parallel to the electrode surface ("flow-through pattern" , Fig. 13b), although the parallel mode seems less advantageous from the theoretical point of view. Electrolysers with graphite felt or fiber cathodes were used in Czechoslovakia already in the seventies for purification of waste waters containing platinum and gold. The metal concentrations were thus decreased from less than 1 mg/l below 10 pg/l; the rate of flow in the "flow-across" mode was very low. Reticulated vitreous carbon (RVC) cathodes were tested for removing mercury from contaminated brine solutions [28]. Copper foam cathodes were used for electrolytic recovery of copper r.291.

Mass transfer coefficients were evaluated for flow-by porous electrodes [30] and found to be higher by factors of 2 8 95 than on a plate electrode, when the rate of flow of electrolyte 10 cm/s. in the interelectrode space was 2 Possible problems associated with the use of such electrodes (particulate beds, nets, gauzes, foam structures) may include: (a) non-uniform current and potential distributions; (b) blockage by solid products and gas entrapment; (c) difficulties in making current feeder contacts; (d) difficulties in maintaining mechanical and electrical integrity of the structure; (e) an increased pressure drop. Mass transfer coefficient ( k m ) and pressure drop per unit electrode height were studied on both flow-through and flow-by nickel foam electrodes [31]. A comparison with older literature data suggests that the values of km are higher by a factor of up to 1000 compared to a plate electrode. Although the cited work was concerned with laboratory systems, the technical use of the mentioned electrodes for metal winning from dilute solutions seems hopeful, since the nickel foam is manufactured on an industrial scale by the firm SORAPEC, France. The product is

-

-

68 distinguished by a high porosity ( 0 . 9 7 ) , low apparent electrical resistivity ( 1 . 5 ~ 1 0 -nm) ~ and a high specific surface area (up to 41 000 m2 /m3 ) . The nickel foam is delivered in 2 3 nun thick sheets. The mass transfer coefficient on a plate electrode can be increased by the factor of 4 6 if a so-called turbulence promoter is inserted into the space between the electrode and the diaphragm. This is usually a plastic mesh made of high-density polyethylene. Its thickness is 0.3 3 nun and mesh size 0.5 9 mm. The values of the mass transfer coefficient in such systems were measured by many authors [30,32-341. The use of turbulence promoters can be recommended f o r such processes where the clogging by the deposited metal particles is excluded. They are used mostly in electrosynthetic processes. One of the typical modular electrolysers designed for a wide range of electrochemical processes (including those using membranes) is the ElectroSynCell, manufactured by Electrocell AB, Sweden. Similar units, denoted as FM 01-LC through FM 21-SP, are delivered by ICI, England. Electrolysers with packed bed cathodes, so-called Envirocells (Deutsche Carbon AG/ENVIROCELL Urnwelttechnik GmbH, Germany) with graphite spheres in the cathodic compartment and an inert or ion-exchange membrane (Fig. 14), are used in indust-

-

-

-

-

Figure 14. Scheme of a packed bed electrolyser. 1 Anode compartment, 2 cathode compartment filled with graphite spheres and separated by a diaphragm.

69 rial applications for removing mercury and copper ions from waste water. A list of various applications is given in Table 3 [35]. The advantages of packed bed electrolysers are as follows: ( a ) Favourable operational costs compared to classical waste water treatment systems, even (and, practically, only) for low metal concentrations down to some ppm; (b) limited space required allows installation of the system even beside existing installations; (c) apart from cathodic metal separation, anodic oxidation enables double benefits to be obtained, e.g. when destruction of cyanides or chlorinated hydrocarbons, etc., is required. Destruction of organic compounds will be dealt with in a subsequent section. Table 3 Some applications of ref. ( 3 5 1 )

packed bed

electrolysers (according to ~

Origin of waste water

Metal extr. ~

Chlor-alkali electrolysis Dye stuff production Dye stuff production Cell. acetate production Oxychlorination process (under construction) Prod. of mater. for plastics (under constr.)

Flow rate m3/h ~

~

~

Concentr. in out mg/l mg/l

En. consum. kWh/m3

Anode size sq. m

~~~~

Hg

3.5

1.0

0.05

0.16

20

Hg

2.0

4.0

0.05

2.5

15

cu

6.0

400

2.0

4.0

90

cu

20

20

1.9

0.08

40

Cu

22

6.0

0.5

0.15

40

Cu

3.0

50

0.5

1.8

26

OTHER, SIMILAR ELECTROLYSER TYPES

Attempts to increase the active surface area of the cathode and the mass transfer coefficient and to prevent the formation

70

of dendrites or metal powder or the cementing of the bed particles together led to many variants of the electrolyser with a non-stationary bed of conducting particles. There are two principles that are utilized nowadays; and these will be discussed below. The first one, so-called rolling layer cell, is realized by a rotating drum, partially filled with small metal balls of 4 6 mm in diameter, occupying 10 15 % of the inner volume. The balls are in contact both mutually and with an auxiliary cathode (Fig. 15) and they are thus polarized to a negative potential. About one half of the available volume is filled with the electrolysed solution; a stainless steel or titanium grid coated with platinum, I r O z or RuOz and TiOz serves as an insoluble anode, permitting a rapid removal of the evolved oxygen. The axis of the drum is slightly inclined; small balls are fed into the drum at its higher-positioned end, they travel by gravity while growing in size, and larger balls (8 12 mm in diameter) are removed from the drum at the other end. The electrolyte flows through the drum in the axial direction and the

-

-

1

7

5

Figure 15. Basic scheme of the electrolyser with a moving particulate cathode, developed by Bergakademie Freiberg, FRG. 1 Cylindrical steel mantle, 2 layer of polypropylene or hard electrolysed solution, 4 perforated lead anode, 5 rubber, 3 cathodic current lead, 6 bed of metal spheres, 7 diaphragm, 8 supporting wheels.

71

impoverished solution is reused in the main process. Electrolysers of this type (manufactured by Bergakademie Freiberg, FRG) are recommended for metal winning from rinse water containing 1 - 40 g of the metal per litre. The length of the drum is up to 2 m, its diameter up to 1 m, number of revolutions per minute 0.1 - 5, current load up to 2000 A/m2 (referred to the surface area of the bed), current yield 90 % or even higher, and current load for copper deposition from solutions containing 8 40 g/1 Cu is up to 2000 A. The electrolyser was tested in the deposition of copper [36], silver from photographic fixing baths [36], and nickel [37]. Another system, manufactured by Martineau Comp. (31120 Portet -sur-Garonne, France) utilizes a periodically pulsating bed of small metal balls [38] and its scheme is shown in Fig. 16. It was realized and tested on a pilot scale in the electrowinning of silver from photographic fixing baths. The rate of flow of the solution was 0 . 2 m3/h, the current yield was 4 g Ag per 1 Ah and the input power was 600 W. The bed consisted of about 4 6 kg of silver balls. The same electrolyser was used in removing cadmium from rinse

-

-

Figure 16. Scheme of the electrolyser with a pulsating bed, developed by Martineau Comp., France [38]. 1 Driving motor, 2 piston, 3 supporting mesh, 4 cathodic current lead contacting the bed of silver balls, 5 metal mesh anode.

12

water, but the results do not seem satisfactory. The concentration of Cd was lowered from the original 100 to 50 pg/l after electrolysis. Maintenance of chromium plating baths causes ecological problems that can be solved by electrochemical methods, especially by using membrane electrolysers (i.e. electrodialysis) [39-411. In these processes, chromium(II1) is oxidized at the anode, usually made of lead, to chromium(VI), whereby chromic acid is regenerated. The chromic acid consumed in the plating process is then replaced by adding Cr03 in the solid form. The electrolyser for regeneration of chromium plating baths has two compartments, a cathodic and an anodic one, separated by a cation-exchange membrane. The solution to be regenerated flows through the anodic compartment, while impurities such as Fe3+, Mg2+ and other cations are transported into the cathodic compartment containing a weak acid solution. The applicability of the electrodialysis method was verified on a pilot scale [39]. The operating costs were found to be substantially lower than with the common process using ionexchange and detoxification units. Considerable savings are due to the fact that almost no chemicals are consumed, and the expenses f o r maintenance and salaries are much lower than with the common process. ELECTROCHEMICAL DESTRUCTION OF TOXIC ORGANIC COMPOUNDS Contamination of both industrial and household waste waters with organic chemicals represents an ecological problem of increasing importance. Halogenated organic compounds, especially halogenated hydrocarbons, are particularly problematical since they are produced in very large quantities on an industrial scale, they are very poisonous and chemically stable. Various methods have been considered for their destruction and some of them are used in water works, e.g. biodegradation and oxidation with ozone. However, chlorinated hydrocarbons are resistant against such treatment. Therefore, electrochemical reduction as a method for their destruction has been studied by many aut-

13

hors, who showed that this method is feasible and leads to practically complete detoxification [ 4 2 ] , Its further development seems therefore desirable. Advantages of the cathodic dehalogenation are: (i) treatment at ambient temperature, (ii) no additional chemicals, and (iii) selective removal of chlorine while the organic skeleton remains to be digested by the biological route. This is still the cheapest way. Thus, various chlorinated aliphatic and aromatic compounds were dechlorinated in a flow-through electrochemical cell with a graphite fibre cathode, a Nafion (cation-permeable) membrane and a Pt gauze anode. The concentration of pentachlorophenol decreased from 50 to about 1 mg per litre after 2 0 min of electrolysis at a current efficiency of about 1 %, and the product was phenol. Similar results were obtained with other chloroderivatives. The expected total costs of the process are of the order of 10 DM per 1 m3 of waste water, which is comparable with the cost of adsorption on active carbon [ 4 2 ] . Chlorinated hydrocarbons can also be decomposed by cathodic reduction at a copper electrode in aqueous solutions at pH = 4 to 7 . The reaction was carried out in a fixed-bed, flow-through electrolyser filled with copper balls, 0 . 2 - 0.6 mm in diameter, supported on a platinum gauze [ 4 3 ] . The best results were obtained with hexachlorocyclohexane, which was completely and rapidly dechlorinated; less reactive compounds were tetrachloroethylene, trichloroethane, and chlorobenzene; unsatisfactory results were obtained with a polychlorinated biphenyl. The dechlorination can also proceed on a suspension of powdered iron that has been partially coated with copper by cementation [ 4 3 ] . In this case no external current is needed, since the reaction proceeds similarly to corrosion, where the free iron surface acts as the local anode and dissolves to give iron(I1) ions, whereas the copper-coated surface represents the local cathode, at which the given chloroderivative is reduced to C1- ions. The reduction rate is lower in this case, since the potential of the local cathodes is less negative than in

14

the case of a cathodic reduction with imposed d.c. voltage. Destruction of toxic halogenated hydrocarbons and other similar contaminants can also be effected by photocatalytic oxidation [44] using suspensions of semiconductor particles under illumination. If the radiation energy is higher than the band gap energy of the semiconductor, absorption of a light quantum causes excitation of an electron from the valence band into the conductivity band. Thus, an electron/hole pair is formed, of which the hole acts as an extremely strong oxidizing agent, causing destruction of any organic molecule present in the semiconductor/electrolyte interface. The electron is then captured by some reaction intermediate, which is thus reduced. Thus, there is a distinct analogy to electrochemical corrosion reactions. An aqueous suspension of Ti02 can be used as a photocatalyst inducing complete mineralization of toxic halogenated hydrocarbons in waste water effluents under illumination with sun light or a UV lamp, yielding C 0 2 and non-toxic mineral acids as the only products. The authors [44] studied the destruction of chloroform in detail. A possible technical realization is based on a cylindrical UV lamp, surrounded by a cylindrical glass mantle containing a glass mesh coated with Ti02; the contaminated water flows through the mantle in the axial direction and the organic contaminants are oxidized in contact with the irradiated Ti02 layer [45]. Another equipment uses sun light, concentrated by a cylindrical mirror [46] on a glass tube reactor.

RBFBRBNCBS

1 2 3 4 5 6

Verbost PM, Flik GI Pang PKT, Lock RAC, Bonga SEW. J Biol Chem 1989; 264: 5613-5615. Lundberg U, Milanes CL, Pernalete N, Weisinger JR, et al. Am J Physiol 1987; 253: F401-F407. Wesenberg GBR. Scand J Dent Res 1982; 90: 95-101. Gruden N. Environ Res 1987; 43: 19-23. Vallee BL, Ulmer DD. Annu Rev Biochem 1972; 41: 91-129. Flick G I van de Winkel JGJ, Part P, Bonga SEW, Lock RAC. Arch Toxicol 1987; 59: 353-359.

75

7 8 9 10

11 12 13 14 15

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

Sotiroudis TG. Biochem Int 1986; 13: 59-64. Yoshiki SYT, Kimura MO, Suzuki SMT. Arch Environ Health 1975; 30: 559-564. Cibulka J et al. Pohyb olova, kadmia a rtuti v biosfkPe (Circulation of lead, cadmium, and mercury in the biosphere!: Academia, Prague, 1991; Chap. 4. Ptacek MI Erlebach J, Lischke P I Matdjka 2. tigtgni odpadnich vod z galvanotechniky a chemickk povrchovd lipravy kovh. (Purification of waste waters from metal plating and finishing. ) Prague, Stitni nakl. tech. liter., 1981; 299. Ibl N. In: Tobias CW, ed. Advances in Electrochemistry and Electrochemical Engineering. New York: Interscience, 1962; 121. Rougar I, Cezner V. J Electrochem SOC 1974; 121: 648-651. Rougar I, Micka K, Kimla A. Electrochemical Engineering, I. Elsevier: Amsterdam, 1986; Chap. 8. White RE, Walton CW, Burney HS, Beaver RN. J Electrochem SOC 1986; 133: 485-492. Aerov ME, Todes OM. Gidravlicheskie i teplovye osnovy raboty apparatov so statsionarnym i kipyaschim zernistym sloiem. (Hydraulic and thermic operation principles of devices with stationary and fluidized beds of particles.) Khimiya: Leningrad, 1968; 176. Carbin DC, Gabe DR. Electrochim Acta 1974; 19: 645-652. Coueret F, Le Goff P. Electrochim Acta 1976; 21: 195-202. Goodridge F, Vance CJ. Electrochim Acta 1979; 24: 1237-1242. van der Heiden G, Raats CM, Booh HF. Chem Ind 1978; 13:465. Anonym. Chem Engineering 1976; 18: 72. Fleet DS. Chem Ind 1971; 300. Fleischmann MF, Oldfield JW, Tennakoon LJ. J Appl Electrochem 1971; 1: 103-112. Tsong-Yang W, Sneshaw T, Jing-Chuhoh. J Chin Ch E 1987; 18: 83. Haines KP, Wilkinson JAE. Trans Inst Min Metal1 1972; 81: 157. Haines KP, Wilkinson JAE. Ind Recovery 1971; 17: 26. Wilkinson JAE. Trans Inst Metal Finish 1971; 50: 16. Rouiar I, Micka K, Kimla A. Electrochemical Engineering, 11. Elsevier: Amsterdam, 1986; Chap. 36. Matlosz MI Newman J. J Electrochem SOC 1986; 133:1850-1859. Tentorio A, Casolo-Ginelli U. J Appl Electrochem 1978; 8: 195-205. Hammond JK, Robinson D, Brown CJ, Pletcher D, Walsh FC. Dechema Monographs, 1990; 123: 299-316. Langlois S, Coueret F. J Appl Electrochem 1989; 19: 51-60. Carlsson L, Sandergren B, Simonsson D, Rihovsk9 M. J Electrochem SOC 1983; 130: 342-346. Hammond JK, Robinson D, Walsh FC. Dechema Monographs, 1990; 123: 279-288. Muller V, Rougar I. Dechema Monographs, 1990; 123: 331-344. Muller KJ. Dechema Monographs, 1990; 123: 199-204. Schab D, Hein K. Metallwiss Technik 1990; 44: 362.

76 37 38 39 40 41 42 43 44 45 46

Hein K, Kirchhof A, Schab D. Metallwiss Technik 1987; 41: 148. Lacoste G. Dechema Monographs, 1990; 123: 411-426. Bormann G I Mobius A. Galvanotechnik 1992; 83(1): 2-5. Hartinger L. Handbuch der Abwasser- und Recycling Technik. Munchen: Hanser, 1991. Giilbas M. Metalloberflache 1988; 42: 4. Schmal D, van Duin PJ, de Jong AMCP. Dechema Monographs, 1991; 124: 241-247. Bachmann T I Vermes I, Heitz E. Dechema Monographs, 1991; 124: 221-239. Bahnemann D, Bockelmann D, Goslich R. Dechema Monographs, 1991; 124: 261-281. Matthews RW. Solar Energy 1987; 38: 405-413. Matthews RW. Water Res 1986; 20: 569-578.

Electrochemical oxidation of organic pollutants for wastewater treatment Christos Comninellis Institute of Chemical Engineering, Swiss Federal Institute of Technology CH-1015 Lausanne, Switzerland

1. INTRODUCTION

Besides inorganic materials, (heavy metals. acids, bases, ....) industrial wastewater also contains organic pollutants which have t o be treated before the water can be discharged. Biological treatment is the most economic process and is usually used for treatment of "readily degradable" organic pollutants present in the wastewater. The situation is completely different when the wastewater contains refractory (resistant to biological treatment) organic pollutants or if their concentration is high or/and very variable. In this case, an other type of treatment must be used. Many treatment technologies are in use or have been proposed for recovery or destruction of these pollutant. These include activated carbon adsorption and solvent extraction for recovery chemical o r electrochemical oxidation for destruction. Several applications of chemical oxidation have been reported. Oxidants employed include hydrogen peroxide (in the presence of Fe+2), ozone and chlorine. The electrochemical method for the treatment of wastewater containing organic pollutants has attracted a great deal of attention recently (1). Mainly because of the ease of control and the increased efficiencies provided by the use of new electrode material (2,3,4,5) and compact bipolar electrochemical reactor (6).

In the present investigation the aim was t o examine the anodic oxidation of some benzene derivatives (model organic pollutants) at platinum and DSA type anodes to elucidate the possibilities of the electrochemical method and to elucidate the mechanism of the reaction.

2. EXPERIMENTALDETAILS

ZL Determination of the curmnt eflbiciency During the electrochemical treatment of wastewater containing organic pollutants a side reaction of oxygen evolution always occurs resulting in a decrease in current efficiency of oxidation.

Two methods have been used for the determination of the current efficiency during the electrochemical oxidation of the organic pollutant (OP). The oxygen flow rate method and the COD (Chemical Oxygen Demand) method. In the oxygen flow rate method the oxygen flow rate was measured during electrolysis and the Instantaneous Current Efficiency (ICE) for oxidation of the OP a t given experimental conditions was calculated using the relation

where

9,

= Oxygen flow rate in absence of the organic pollutant (OP) in the

electrolyte (cm3min-1) ( V t I o r ~ Oxygen flow rate a t a given time t in the presence of OP (cm3min-1) In the COD method the Chemical Oxygen Demand was measured (COD, Hach Dr/2000) during electrolysis and the ICE was calculated using the relation

where (COD),

I F V

: Chemical oxygen demand at time t ( g [O] dm-3) : Chemical oxygen demand a t time t c A t (g [Ol dm-3) : Current (A) : Faraday constant (96487 C eq-1) : Volume of electrolyte (dm3)

The choice of the method to measure ICE depends on the solubility of the electrolysis product. Thus the COD method is used only if the electrolysis products are soluble in the electrolyte contrary t o the oxygen flow rate method where the electrolysis product may be soluble or insoluble. Measurements of ICE by both techniques gives information about the formation of polymeric products a t the anode during treatment. The ICE decreases with time during electrolysis t o finally reach a value of about zero. A typical example is given in Fig.1 where the evolution of ICE ( from the oxygen flow rate method) during the electrochemical oxidation of phenol (model pollutant) is given.This figure shows that for the experiment conditions used ICE decreases from 0,28(initial value) to 0,015 after 3,5 h of electrolysis.

79

w

0.30

0

0.20

0.10

I1

-m-0

>

0.

-. \

0 '

0,

I

\O \ \

0 \

o\ \ \

0

0.00

I

I

'

I

-

'

Figure 1. Evolution of the ICE (from the oxygen flow rate method) during the electrochemical oxidation of phenol. i = 57 mA cm-2, T = 70 "C, pH 12,5 (constant). Initial phenol concentration: 23 mmol dm -3

Electrochemical cell for the determinationof ICE For the determination of ICE using the oxygen flow rate method, a two compartment cell of 150 ml capacity was used; the anode was a cylinder of 35 cm2 surface area and the cathode was a platinum spiral (4 cm2) enclosed in a 10 ml porous porcelain pot; stirring was provided by a magnetic bar. The pH of the anolyte was maintained constant (pH = 12,5 or 3) during electrolysis by continuous introduction of a solution of NaOH or H2S04 to the anolyte (150 g dm-3Na2S04) containing the OP. The oxygen formed in the anolyte during electrolysis in the presence and in the absence of OP was measured using a gas burette and analyzed by gas chromatography. For the determination of ICE using the COD method and to study the influence of hydrodynamic conditions on ICE a parallel plate cell with external circulation of the electrolyte was used. The parallel plate cell is constructed in two halves each with two electrodes Anode-Cathode (100 cm2 surface each) inlaid in polypropylene blocks. On one of the halves inlet and outlets were provided for circulation of the electrolyte. The interelectrode gap was 4 mm (thickness of the viton gasket). Circulation of the electrolyte through the parallel plate cell was provided by a magnetic pump (Iwaki MD 50 R) and the electrolyte flow rate was measured with a magneto hydrodynamic flow meter (Deltaflux). Details of the hydraulic equipment are given in Fig. 2.

80

P 0 Electrolytic cell Q Flow meter 0 Recirculation pump @ Heat exchanger 0 4 litre reactor

Figure 2. Equipment used for the determination of the Instantaneous Current Efficiency (ICE) during the electrochemical treatment of wastewater. Preparationof the DSA type anodes Different techniques have been used for the preparation of the so called Dimensionally Stable Anodes (DSA) which consist of a thin layer of active coating (1- lop) on an inert substrate usually titanium. PreDaration bv the thermal decomDosition techniaue (Ti/Ru02, Ti/IrO2) I

.

The Ti/RuOz and TYIrO2 anodes are prepared by the thermal decomposition technique in which the following steps are involved: Dissolution in isopropanol of the coating component (RuC13 for the Ti/RuO2 and H2IrCls for the Ti/IrOa)and application on the pretreated titanium substrate by brush, drying at 8OoC, thermal decomposition a t high temperature (" 500°C) cooling and repeating the above operation until the desired amount of coating is reached (" 35 g/m2), finally post-heat treatment a t 550°C for 1 h. Fig. 3 shows a microphotograph of a TYIrOz electrode prepared by this technique. Much more details concerning electrode preparation and characterization are given elsewhere (2,3,4).

81

Figure 3. Microphotograph of a Ti/IrOz electrode prepared by the thermal decomposition technique.

Figure 4. Microphotograph of a Ti/SnOa electrode prepared by the spray pyrolysis technique.

82

PreDaration bv the smav Dvrolvsis techniaue (TUSnO2) The SnO2 coating doped with Sb2O5 (Ti/SnOz) are prepared on titanium substrate using the spray pyrolysis technique. In this technique alcoholic solution containing the coating components (SnCl4 and SbC13) was sprayed on a heated (at 550°C) titanium substrate for about 30 minutes. Fig. 4 shows a microphotograph of the Ti/SnO2 electrode. Details for the preparation and characterization of this coating are given elsewhere (7). PreDaration bv anodic deDosition (TiPbOz) The Ti/Pb02 anodes are prepared by anodic deposition from a plating bath containing 350 g dm-3 Pb(N03)2 and 30 g dm-3 Cu(N03)2. The applied anodic current density was 2,5 A dm-2 and the working temperature was 65°C. The pH of the bath was maintained constant (at pH 3) by periodic addition of CuCO3 2.2 Electrochemical reactor used for wastewater treatment A new undivided "bipolar electrochemical reactor" of channel type with flat parallel electrodes has been developed in our laboratory. Its design is very simple, it consists of a bipolar stack of 5 - 30 metallic sheets (electrodes) separated from each other by insulating plastic spacers at a fixed distance.

The electrode stack is introduced in the reactor body and the wastewater is treated discontinuously o r continuously. Fig. 5 shows a small pilot scale bipolar electrochemical reactor (with 5 elements) having 2,3 litres volume and 370 cm2 (5 x 74 xm2) total anodic area. The treatment of wastewater is effected at constant current density and temperature.

83 residual g a z e s

0 pump 8 heat exchanger 0 bipolar electrode stack

Figure 5. Electrochemical reactor used for wastewater treatment.

3. DEFINITION OF GLOBAL PARAMETERS FOR THE EJXCTROCHEMICAZ,

TREATRlENT Wastewater contains in general a variety of organic pollutants. Analysis of these pollutants and their oxidation products during the electrochemical treatment is not only a complex matter but does not give direct informations about the current efficiency and the oxidation state of the organic carbon during treatment. We have defined the following global parameters for the electrochemical treatment of wastewater containing organic pollutants (OP).

84

. .

1

(EOI) s n t Efficiency (ICE) - time curve (Fig. 1) an average value can be calculated and defined as the Electrochemical oxidation Index”(E01) . I

ICEdt 0

EOI =-

(3)

z

where z = time of electrolysis at which ICE is almost zero. EOI expresses the average current efficiency and is a measure of the facility of the electrochemical oxidation of the OP at given experimental conditions. From the technological point of view the EOI value gives information about the specific energy consumption (kWh/kg COD) for the electrochemical treatment (eq. 4) E,, = 355

VC

(4)

Considering a linear relation between cell potential (VJ and current density (i)

V, = Vd + p . 1 . i

(5)

eq. 4 becomes

where ESP= Specific energy consumption Vc = Applied potential vd = Decomposition potential p = Resistivity of the wastewater 1 = Interelectrode distance i = Current density

(kWhkg COD) (Volt) (Volt)

(Rcm) (cm) (A/cm*)

Fig. 6 represents graphically eq. 4 for different applied cell potential, this figure shows that for EOI c 0,4 the specific energy increases rapidly.

85

400 c1

300

0 0

2 E Y

200

100 0

0.0

0.5

1 .o

EOI

Figure 6. Specific energy consumption for two cell potentials as a function of EOI values

1Oxwen Demand(EOD) It expresses the amount of "electrochemically" formed oxygen used for oxidation of the organic pollutants. The Electrochemical Oxygen Demand (EOD) can be calculated using the relation

EOD=

8EOIxQ F

(7)

where Q = specific electrical charge passed [Ah (g OP)-1] m e e of- 0 The degree of oxidation (XI expresses the ratio between the amount of "electrochemically" formed oxygen used for oxidation of the organic pollutants (EOD), to the amount needed for complete oxidation (COD*) EOD X=-COD* where COD* is the initial COD value expressed in g 0 2 (g 0P)-1. Using relation (7),X can be given by

(8)

86

Values of X close to 1 indicate complete oxidation and values of X close t o zero indicate that the organic pollutants present i n the waste water are not electrochemically oxidized.

4. POSSIBILITIES OF THE ELECTROCHEMICAL OXIDATION OF ORGANICS FOR WASTEWATER TREATMENT

The electrochemical method for the treatment of wastewater containing organic pollutants is economically viable only if high EOI and EOD values can be achieved. To evaluate the possibilities of this technique we first studied the influence of benzene substituted on EOI values using a platinum anode. In a second step the influence of the organic concentration, pH, current density and anode material on EOI and EOD values was studied (using phenol as model organic pollutant) , and finally the influence of the nature of anions present in the wastewater and the type of the electrochemical cell used on EOI values was elucidated. i) Muence of the nature ofbenzene substituanton EOI at Pt anode The nature of organic pollutant influences strongly the EOI values. In this paper the influence of benzene derivatives on the EOI value has been studied using a P t anode. Effort has been made to treat the experimental EOI data by correlation analysis [the term "Quantitative Structure - Electrochemical Activity Relationship" (QSEAR) was used for this approach] using the equation

log EOI = p Q

+

c

.....

(10)

where 0 is the substituent constant (which depends on the nature and position of the substituent) and p is the electrochemical reaction constant (which depends on the anode material). The electrochemical reaction constant (p) reflects the sensitivity of the reaction t o the effect of various substituents. The positive and negative signs indicate nucleophilic and electrophilic reactions, respectively. For monosubstituted benzene (R = - OH, - NH2, - COO-,- SO;, - NO21 the logarithm of EOI is plotted against the substituent constant Q (Fig Z).

87

I

-3!

"

1

'

"

"

- 0 . 2 0.2

1

0.6

'

sigma

Figure 7. Correlation of EOI with the substituent constant o of monosubstituted benzene derivatives. Initial organic concentration 23 mmol dm-3 The regression equation obtained (n = 5 ) is logEOI = -20 -1,3

(11)

with correlation coefficient R = 0,99.

For ortho-substituted anilines, the log EOI - Q relation is given in Fiv 8

-0.4

i1

-Os2

u

-0.6

2

-0.8

x

01""' i

-1 .o

-1.2 - 1

0

1

2

Figure 8. Correlation of EOI with the substituent constant o of ortho-substituted anilines. Conditions as in Fig. 7

88

The regression equation obtained (n = 4) is

Equations (11)and (12) are one-parameter equations of the form of equation (10). The reaction constant p is invariably negative, which indicates an electrophilic attack on the reaction centre. This fact suggests that the limiting step of electrochemical oxidation of aromatic compounds is a n electrophilic substitution. In such cases, the reaction is slowed down on electron withdrawal from the reaction centre by substituents possessing high positive values of the substituent constant ( 0 ) . This is why the reaction is slowed down by nitrogroups and why, conversely it is promoted by substituents possessing high negative values of the substituent constant (electron repulsing groups).

ii) Influence of the organic concentration The influence of organic concentration on EOI and EOD values is complex; 9 shows the influence of phenol concentration on EOI and EOD.

fig..

- 1.5 0.3

- 1.0

n W

0.2

M

W

8 w

6 - 0.5

0.1

CI

0

w

1

0.o

M

Y

-Ll-T-L 0

40

0.0 80 120

c (mmol

dm-3)

Figure 9. Influence of the initial phenol concentration on the EOI and EOD values. Conditions as in Fig. 1.

89

The dependence of EOI on phenol concentration (C)can be described by the equation

where (EOI)max is the maximum average current efficiency a t the given operation conditions and 1 K, the phenol concentration at which EOI = S(EOI),~~. At very low phenol concentrations (C<< K,) EOI varies linearly with the phenol concentration and at high phenol concentration (C >>K,. EOI is independent of the phenol concentration. The value of K, depends on the condition of oxidation (pH, T, hydrodynamic conditions) and on the anode material used. and K, values, a Since it is very difficult t o determine from FiP 9 the (EOI),, linear representation is proposed by taking reciprocals of both sides of eq. 13.

1 1 By plotting EOI against (Fip 10) yields a straight line from which the values of (EOI)maxand K, can be measured directly from this graph.

0 0.00

0.01

0.02

0.03

i/c

Figure 10. Linear representation of eq. 13 for the determination of (E0Umax and K,

90

This dependence of EOI on phenol concentration (eq. 13) can be explained by the formation of a polyoxyphenylene film at the anode surface which modifies it's electrochemical properties, this film formation is favoured a t high phenol concentration (C >>KJ and high temperature (250' C). iii) Influence of pH The pH of the wastewater during the electrochemical treatment influences strongly the EOI and EOD values. This is due to the fact that the values of the substituent constant (6)are significantly different for the ionic and non-ionic forms of the substituent (table 1). The fact that the values of CJ are smaller in alkaline medium than in acidic medium (table 1) indicates that higher EOI and EOD values are generally obtained in alkaline medium for the electrochemical oxidation of substituted benzene derivatives.

Table 1 Values of the substituent constant 6 for various benzene substitutents in acidic and alkaline medium. Benzene Substituent

forme

Carboxylic acid

- COOH

Sulfonic acid

-so$

Amine

-N

dimethylamine

Alkaline medium

Acidic medium

H~

- N(CH&

substituent constant u -0

-m

-p

-

0,36

0,44

0,55

-

forme

substituent con ;ant d -0

-m

-0,69

-0,06

-

cooso;

-

0,17

0,86

0,60

NH2

0,OO

-0,13

0,93

0,90

- N(CH3)z

0,30

-0,12

iv) Muence of c m n t density The influence of current density on EOI and EOD values depends on the nature of the organic pollutant and on the anode material used. FiF 11 gives the influence of current density on the EOI and EOD for the electrochemical oxidation of phenol at platinum anode.

91

0.3

n

0.2

I

W

8 W

0.1

0.0 0

20 40 60

i (mA cm-2) Figure 11. Influence of current density on EOI and EOD values; Anode: Pt, i = 57 d c m 2 , T = 70° C, pH = 12,5.Initial phenol concentration: 10,6mM The fact that the product i x EOI (which represents the effective current density used for the oxidation of phenol) varies linearly with the current density & U )indicates that the process is not limited by mass transfer at the anode and the oxidation occurs probably by active oxygen ( O H ) formed at the anode by the electrochemical oxidation of water.

lo 86-

420

I

0

20

.

I

40

60

i(appl.) mNcm2

Figure 12. Variation of the effective current density with the applied current density for the electrochemical treatment of phenol. Conditions as in Fig. 7

92

V) Intluence of anode material The influence of anode material on EOI values for the electrochemical oxidation of phenol is given in Fiy 13

* O * 0.8 '

-0

1

0.6

-

0.4

-

W

-

0.2 0.0

Pt

Ti/Ru02 Ti/lr02 Ti/Pb02 Ti/Sn02

ANODE Figure 13. Influence of the anode material on EOI values for the electrochemical treatment of phenol for a given degree of oxidation (x = 0,71 ) i = 50 mA/cm2, T = 70" C, pH = 12,5.Initial phenol concentration: 10,6mM This figure shows that for the given experimental conditions all "traditional" anodes give almost the same EOI value (0,15f 0,021,only the Ti/SnO2 (doped with Sb205) anode gives high EOI values (- 0,581. vi) Influence of anions present in the wastewater

The influence of halide ions (Cl-, Br-) on EOI values for the electrochemical oxidation of phenol on IrO2 and SnO2 anodes are given in Table 2 which shows that in the presence of NaCl EOI values for IrOp increase to values very close to those obtained with SnO2 anodes. The presence of NaCl does not influence the EOI value for SnO2 and the presence of NaBr decreases strongly the EOI value. The increase of EOI values for IrO2 electrodes in the presence of NaCl is due to the participation of electrochemically generated active chlorine on the oxidation of phenol.

93

Table 2 Influence of the halide ions (present in the electrolyte) on EOI values for the electrochemical treatment of phenol; i = 50 mA/cm2; T = 50" C; pH = 12,5. Initial phenol concentration: 10,6 mM Anode

TiArO2

I

Ti/Sno2

NaCl gll 0 1 5 10 0 0 5

NaBr gll

0

5 0

0

EOI

0.17 0.56 0.56 0.56 0.06 0.58 0,56

vii) Influence of the electrochemical cell used (dividedor undivided)

The type of the electrochemical cell (divided or undivided) can influence the EOI values especially for the treatment of benzene derivatives containing a -NO2 substituent. A typical example is the electrochemical treatment of p-Nitro Toluene Sulfonic acid (p-NTS); low EOI values (- O , l > were obtained in the divided cell contrary to the undivided cell where high EOI values (0,5)were obtained. The increase of EOI values in the undivided cell is due to the cathodic reduction of -NO2 group t o -NH2 group, this transformation promotes the electrochemical oxidation as the substituent constant ( 0 )for -NH2 has negative value (favouring the electrophilic attack on the benzene ring) contrary to the NO2 substituent which has positive value (see 8 4 i).

5. ANALYSIS OF REACTION PRODUCTS AT Pt AND SnO2 ANODES The pH of electrolysis strongly influences the nature of intermediates formed; thus at electrolysis in alkaline media (pH > 12) hydroquinone and 1-4 benzoquinone are not detected (benzoquinone itself is unstable in alkaline solution) and the conditions of film formation a t the anode are favorable. The situation is quite different for electrolysis in acidic media (pH < 4) where hydroquinone and benzoquinone are the main oxidation intermediates and the formation of the anodic films is inhibited. It must also be noted that, in acid media, the carbon dioxide formed during the oxidation of phenol (or its intermediate product)

94

escapes from the electrolyte; contrary t o alkaline media where it reacts with alkali forming CO; or/and HCO;. Figure 14 compares the rate of phenol and TOC removal i n acidic medium obtained with SnO2 and platinum anode; the rate of phenol elimination is almost the same for both anodes (complete elimination of phenol after 25 - 30 Ah dm-3) but the rate of TOC removal is much higher for SnOn electrode. Thus using SnOa anode 90 % TOC removal can be achieved after the passage of 50 Ah dm-3 in comparison to 38 % TOC removal obtained with Pt anode. h

I500

3

20 E a

0

1000

0 C

10

Q,

500

c

a

FG

W

0 0

40

,O 80 120 A h dm-3

Figure 14. Rate of phenol and TOC removal 0 Platinum anode (TOC) Q SnO2 anode (TOC) 0 Platinum and SnO2 anodes (Phenol) i = 57 mA cm-2, T = 70OC, pH = 2 (const.); Initial phenol concentration 21 mmol dm-3 The evolution of the oxidation products during electrolysis of phenol at platinum and Sn02 anodes are given in Fig. 15.

95

Pt

A h dm-3 h

?

E

?

-

-u

-3

0

E

0

E

E

E

Figure 15. Evolution of oxidation products during electrolysis of phenol (Conditions as in Fig. 14)

@

Platinum anode 1 Hydroquinone 2 Catechol 3 Benzoquinone 4 Maleic acid

@

SnOzanode

5 Formic acid 6 Oxalic acid

7 Other intermediate (A)

96

The concentration of the other non identified intermediates (Curve 7 in fig. 15) has been calculated from the relation Nonidentified] intermediates

-

[Initial] Phenol

-

[

Mass balance

3

identified intermediates

The mass balance of the identified reaction intermediates (or product) has been calculated using the relation

[c02]-2

Fumaric Ma'eic

1

Comparison of the oxidation products obtained with platinum and SnO2 anodes shows two main differences:

- A t the SnO2 anode there is only a very small amount of aromatic intermediates (hydroquinone, catechol, benzoquinone), these intermediates are formed in large amount on the platinum anode.

-

Aliphatic acids (fumaric, maleic, oxalic) are rapidly oxidized a t the SnO2 anode and are practically electrochemically inactive at the Pt anode.

6. COMPARISON BETWEEN ELECTROCHEMICAL AND CHEMICAL OXIDATION

For better understanding of the mechanisms of the electrochemical oxidation of phenol a comparative study was undertaken between electrochemical (with P t and SnO2 anodes) and chemical oxidation of phenol with H202 in the presence of Fe+2 catalyst (Fenton's reagent) which is well known to occur by electrophilic attack of hydroxyl radicals on the organic compound (8). There are two techniques for the oxidation with Fenton's reagent. i) Room temperature oxidation, in which the oxidation is effected at 20 - 30°C by mixing phenol with an excess H202 (8).

ii) High temperature oxidation, in which oxidation is effected at high temperature (140OC)and pressure ( 5 bar) by continuous introduction of H202 to the reaction medium (9).

97

In the case of room temperature oxidation, large amounts of aromatic intermediates (hydroquinone, catechol, benzoquinone) are initially formed and further oxidized to aliphatic acids (maleic, fumaric, oxalic). These acids are stable towards further oxidation. [Phenol]

Aromatic

+

Intermediates

1

[

Aliphatic Acids

1

This reaction scheme is similar to the one observed with the platinum anode (Fig. 15A). The main difference between room temperature Fenton's reagent and electrochemical oxidation with platinum anode is that the level of TOC elimination is higher for the electrochemical oxidation with platinum (60 % of TOC elimination) than the chemical oxidation a t room temperature with Fenton's reagent (30% of TOC elimination) (10). Concerning the oxidation at high temperature and pressure (14OoC,5 bar) with Fenton's reagent only a small amount of aromatic intermediates is formed. The principal intermediates are aliphatic acids which are further oxidized to carbon dioxide. [Phenol]

+

r

Intermediates

7

1Aromatic + Aliphatic 1

4

[C021

This reaction scheme is very similar to the one observed with SnO2 anodes (Fig 15B).For both oxidations the level of TOC elimination is higher than 90 %. It is well known that Fenton's reagent generates hydroxyl radicals in solution (8). Fe+2+ H202

+

Fe+3+ OH- + O H

which oxidizes the organic compound (phenol o r its oxidation intermediates) by electrophilic attack. Aromatic

[PhOHl

qOH' > [ Intermediates]

OH*

Aliphatic

[

Acids

]

?>

The rate constant kl for phenol hydroxylation is very high k l 4010 dm3 mol-1s-1 (ll)] and this reaction can be considered as instantaneous (molecular diffusion will determine the rate of the reaction). Rate constants kz (ring opening) and k3 (combustion) are relatively low a t room temperature [kz = 1O*dm3mol-1s-l,k3 = 105dm3mol-~s-~ (1113. But as these reactions have high activation energies (9)the rate constants increase considerably with temperature and probably at 140°C (high temperature, Fenton's reagent) all three reactions are instantaneous.

98

A similar model can be proposed for the electrochemical oxidation of phenol at platinum and S n 0 2 anodes. Thus, for the electrochemical oxidation a t the platinum anode the rate constants k2 and k3 are relatively low contrary to the ,31102 anode where all three reactions are instantaneous.

7. MECHANISM OF ELECTROCHEMICAL OXIDATION OF PHENOL FOR WASTEWATERTREATMENT

The mechanism of electrochemical oxidation of phenol has been studied by various workers as a means of producing hydroquinone and benzoquinone. However the reaction sequence transforming a dilute solution of phenol t o maleic acid, oxalic acid and finally carbon dioxide is not well understood. The experimental results obtained in this paper have shown that traditional anode materials (Pt, Ti/Ir02, Ti/RuO2, TiEbO2) give relatively low current efficiencies (low EOI) for the anodic oxidation of phenol. Contrary to the WSnO2 anode which not only gives high EOI values but allows quasi complete Total Organic Carbon (TOC) elimination. Comparison between electrochemical and chemical oxidation (using H202/Fe+2)has shown that the reaction scheme at platinum and SnO2 anodes is similar t o the one obtained with H202/Fe+2 at room temperature (20 - 30 "C) and high temperature (14OoC,5 bar) respectively. To explain these unexpected results we propose the reaction sequence given in Fig. 16 for the electrochemical oxidation of phenol. In this scheme, hydroxyl radicals are initially produced by the electrochemical oxidation of water. The hydroxyl radicals thus formed are adsorbed at the electrode surface and react with phenol in a three step process [hydroxylation (kl), ring opening (k2) and combustion (ks)] o r can subsequently react giving 0 2 [side reaction (k)]. The reaction rate constants k, k l , kz and k3 depend strongly on the anode material (table 3). It is interesting t o note that the combustion rate constant (k3) is very low at a platinum anode, contrary to Ti/SnO2 where k3 is high. Further investigation is necessary to get more specific information about the mechanism of electrochemical oxidation a t Ti/SnOz electrodes.

99

Oxygen evolution (Side reaction)

Oxidation of phenol (Desired reaction)

ki

Hydroxyla tion

7 (Aromatic intermediates)

Ring opening

k

(Aliphatic Acids) Combustion

Figure 16. Mechanism of the electrochemical oxidation,

100

Table 3 Influence of the anode material on the reaction rates k: Oxygen evolution k2: Ring opening kl: Hydroxylation k3: Combustion Electrode

k

kl

k2

k3

Pt

medium

high

medium

very low

Ti/SnO2

low

high

high

high

8. PILOT PLANT EXPERIMENTS FOR THE TREATMENT OF INDUSTRIAL WASTEWATER

A pilot plant "bipolar electrochemical reactor" of 1,6 m2 of total anodic surface area was constructed and tested for the treatment of industrial wastewater produced during t h e production of aminothiazol derivatives (the characterization of this wastewater is given in table 4).

Fig. 17 shows the installation used. The evolution of TOC during treatment is given in Fig. 18. This figure shows that after the passage of 150 Ah-dm-390 % of TOC are eliminated. Table 4 Characterization of the industrial wastewater TOC : 7'800 mgA NaCl : 50 g/l

DCO : 25'800 mg/l

I

DBO5 : 3'000 mgA

Electr. conductivity : 50 mskm

101

Figure 17. Pilot plant installation used for the electrochemical treatment of industrial wastewater.

102

10000

8000

? E U

6000

F

4000

Y

0

e

2000 0 -

0

50

100 150 200

Ah/l

Figure 18. Evolution of TOC during the electrochemical treatment of the industrial wastewater (see table 4).

REFERENCES 1. 2.

3. 4.

5. 6. 7. 8. 9. 10. 11.

Ch. Comninellis, and E. Plattner, 1988, Chimia 42: Nr. 7/8: 250-252. G.P. Vercesi, J. Rolewicz, and Ch. Comninellis, 1991, Thermochim. Acta 176: 31. Ch. Comninellis, and G.P. Vercesi, 1991, J. Appl. Electrochem. 21: 335345. G.P. Vercesi, J.Y. Salamin, and Ch. Comninellis, 1991, Electrochem. Acta 36: 991. Ch. Comninellis and G.P. Vercesi, J. Appl. Electrochem. 21 (1991): 136142 Ch. Comninellis, E. Plattner and P. Bolomey, J. of Appl. Electrochem. 21: (1991)415-418 R.KBtz, S.Stucki, and B. Carcer, 1991, J. Appl. Electrochem. 21: 14-20. N. Al-Hayek, and M. Dore, 1985, Environmental Technology Letters 6: 3750. H. Debellefontaine, P. Striolo, F. Haddoud, J.N. , and J. Besombes VailhB, 1990, Informations Chimie 223: 192-196. Ch. Comninellis, and C. Pulgarin, 1991, J. Appl. Electrochem. 21: 703708. J. Hoign6, 1988, in "Process Technologies for Water Treatment", S. Stucki (edt.), Plenum New York, p. 121-143.

103

Electrochemical Oxidant Generation for Wastewater Treatment Pallav Tatapudi and James M. Fenton Department of Chemical Engineering, U-139 University of Connecticut Stows, CT 06269-3139 USA

Toxic and hazardous organic compounds are often present in water supplies, groundwaters and industrial wastewaters at low concentrations which can make their removal arduous and expensive by conventional treatment processes. Chemical oxidation is being considered as a complete technology for the treatment of organic compounds in order to meet the regulations for toxicity as well as COD reduction [1,21.

Chemical oxidation, as a treatment process, can be more beneficial than other available treatment technologies which: a) shift the problems (hauling and landfilling), b) concentrate the problems (carbon adsorption, reverse osmosis), c) transfer the problem to another medium (air stripping), or d) have an extremely narrow range of operating conditions (biological treatment). Chemical oxidation is one of the few processes which can destroy hazardous and toxic organics compounds on site. The degree to which these compounds decompose can be described as (31:

Priniary Degradotion:

A structural change in the parent compound occurs,

104

which allows it to be eliminated more easily by other processes (eg., biological treatment, adsorption, etc). Acceptable Degradation:

Decomposition of the parent compound to the extent that its toxicity is reduced.

Ultimate Degradation:

Mineralization of the parent organic compound, i.e., its conversion to inorganic substances such as CO, and H,O.

An environmentally acceptable oxidant should possess the following features: 1.

Reactive with compounds to be treated.

2.

Neither produce nor leave undesirable by-products during the course of the reaction.

3.

Readily available.

4.

Reasonably inexpensive to purchase.

A number of chemical oxidants are currently being used either as disinfectants for potable or swimming pool waters, or for the destruction of organic and inorganic chemical species found in wastewaters. The most commonly used oxidants for the above purposes are ozone, hydrogen peroxide, chlorine, chlorine dioxide, sodium and calcium hypochlorite, and potassium permanganate. This chapter discusses the electrochemical synthesis of these oxidants in relation to their use in treating organic wastewaters. The benefits of generating

105

oxidants electrochemically may include:

1.

Close regulation of product yield and purity through control of applied potential or current.

2.

Elimination or minimization of chemical by-product generation during oxidant synthesis.

3.

Transport and storage of toxic and hazardous oxidants can be eliminated by producing them on site at an amount proportional to the waste concentration.

4.

Electrochemical processes could also have an economic advantage over traditional routes for oxidant generation, especially in small scale uses.

Those electrochemical processes which are either on an industrial scale or those which produce oxidants continuously are reviewed. More importance will be placed on the synthesis of ozone and hydrogen peroxide as these two oxidants (with

or without use of UV radiation) have been receiving tremendous attention lately in treating wastewaters containing toxic and hazardous organics [4].

Ozone-0,: Ozone, a triangular shaped molecule is an unstable gas possessing a characteristic sweet odor. It is detectable in air (often near copier machines) by most people at a concentration as low as 0.01 ppm [5]. Ozone is the second most powerful oxidant molecule, exceeded in it's oxidizing power only by fluorine. Since ozone is a non-polluting oxidant (ozone is

106

reduced to oxygen during organic oxidation), its use far exceeds that of fluorine. The largest application of ozone is for the treatment of drinking water which was developed in Europe in the early 1900's. Ozone's second largest application is in treating odors and colors from industrial and municipal wastewater treatment plants. It is also used as a bleaching agent and as a precursor for the manufacture of pharmaceutical intermediates. Currently, ozone is commercially produced by the corona discharge process. This involves the application of an alternating voltage of a very high magnitude (20,000 V) between two electrodes with dry air or oxygen passing in between. The

oxygen atoms that form combine with an 0, molecule to form ozone. This technique produces ozone concentrations of about 2-6 weight % in the gas phase. In the mid 1970s, ozone received a considerable amount of attention when it was shown that hydroxyl radicals are formed in ozonated water in the presence of either ultra violet (UV) light or hydrogen peroxide. These hydroxyl radicals are far more powerful than ozone as an oxidizing agent [4,6]. However, the low ozone concentrations available from the corona discharge process restrains ozone from being used for wastewater treatment purposes.

Further, low ozone transfer

efficiencies from the discharge gas to the water and thus the need for large volumes of carrier gas have also restricted the use of an otherwise useful and powerful oxidant.

In order to circumvent these difficulties, various electrochemical processes for ozone evolution were and are being investigated. Advantages in capital costs,

107

ozone

concentration, and operating characteristics show the electrochemical

generation of ozone to be economic for small-scale uses. Foller and Goodwin [7] found that electrochemicalmethods have economic advantages for ozone productions rates less than 6.8 kg/day. Ozone was first discovered accidentally by electrolysis of sulfuric acid in 1840'. Since then electrolytic generation of ozone has developed rather slowly, owing to the relatively low current efficiencies that have been observed at practical

operating temperatures. Ozone is formed by the electrolytic decomposition of water, which can be represented by the following half cell reaction: 3H20

> 0,t6Htt6e-

Eo = 1.51 V vs NHE

(1)

Oxidation of evolved oxygen could also possibly produce ozone: > 0 , t 2 H t +2e-

H,O t 0,

Eo = 2.07 V vs NHE

(2)

The evolution of oxygen will occur preferentially over ozone evolution as it

is a lower potential process: 2H,O

> 0,t4Htt4e*

Eo = 1.23 V vs NHE

(3)

In order to obtain ozone at reasonable concentrations, and for the reaction to proceed at significant current efficiencies, the evolution of oxygen must be inhibited. This can be achieved by using high oxygen overvoltage (poor oxygen evolution kinetics) anodes. Other requirements for practical geneiation of ozone include: electrolytes whose anions and cations engage in no competitive oxidation

'Schobein, C.F. Comptis Rendis Hebd. Seances Acad Sci 10, 706 (1840).

108

or reduction; anodes present in their highest oxidation state or kinetically resistant to further oxidation and; anodes that are stable in highly acidic environments produced by the anodic decomposition of water. The majority of laboratory investigations on the anodic evolution of ozone used Pt and PbO, electrode materials and H,SO,, HCLO,, or H,PO, electrolytes [8- 131.

Based on these early studies, two main paths have evolved for generating ozone electrolytically. Figure 1. shows a schematic of a cell to synthesize ozone using glassy carbon anodes and a specialized electrolyte, tetrafluoboric acid (HBF,) [ 141. The electrodes remained unattacked at current densities of 400 mA/cm2 evolving ozone at 35% volume concentrations from 48 wt. % HBF, at 10°C. The cell potentials for this system were on the order of 3.2-3.4 V vs NHE. The ozone gas formed within the cell was immediately diluted with air to lower the ozone concentration to 15 wt. % which is below explosion limits. OxyTech (UK) Ltd. is currently selling ozonators based on this technology using 20 cm long, 2.5 cm diameter tubular glassy carbon anodes. The cathodic reaction for this system is the reduction of oxygen by using an air cathode. Inherent advantages of the above system are the relatively high current efficiencies (35 %), lower cell voltage, and elimination of hydrogen management through use of the air cathode. However, the ozone that is liberated has to be used

in the gaseous form for treating wastewaters. An obvious disadvantage of this process is that the treatment process will be limited by the rate of mass transfer of

109

ozone from the gas phase (where it is formed) to the liquid phase (where it is needed for reaction with organics). Figure 2. shows the schematic of a cell to synthesize ozone in deionized water from the backside of a three dimensional porous inert anode (lead dioxide) in contact with a solid polymer electrolyte at room temperature [15,16]. High concentrations of dissolved ozone (20 mg/L) are possible at a current density of 1 A/cm2. Current efficiencies, however, are low (14%). Asea Brown Boveri of Switzerland has commercialized the MEMBREL process for the electrolytic generation of ozone. The synthesis takes place in a cell whose anodic section is made of titanium and the cathodic section of stainless steel. A NafionR membrane (Du Pont) acts not only as the electrolyte for the system, but

also as the separator. This membrane is sandwiched between the anode (lead dioxide) and the cathode (platinum). Electrolysis takes place at a total cell voltage of 3-5 V with a current density of 0.5

- 2.0 A/cm2.

The corresponding cathodic

reaction is hydrogen evolution. This method has a distinct advantage in that it produces ozone in pure water (electrolyte free), making this system more desirable for wastewater treatment. Higher dissolved ozone concentrations can be achieved by pressurizing the system [ 161 which will eliminate the need for low efficiency gas-liquid contact spargers.

However, some of the disadvantages of this system are:

1.

Low current efficiencies (< 15%).

110

2.

The process is highly dependent on the type of lead dioxide used (alpha or beta forms), its morphology, its method of preparation, the substrate used for the lead dioxide, and the change in electrode morphology with time.

Tatapudi and Fenton [17] explored the synthesis of ozone in a proton exchange membrane (PEM) electrochemical flow reactor as a part of an overall scheme to study the paired synthesis of ozone and hydrogen peroxide in the same

PEM reactor. A mixture of commercially available lead dioxide powder and TeflonR, deposited on a NafionR 117 membrane was used as the anode. Current efficiencies ranged from 2.5% at an applied potential of 3.0 V to 5.5% at 4.0 V.

The

corresponding cathodic reaction was hydrogen evolution.

Hydrogen Peroxide-H,O,: Hydrogen Peroxide is a weakly acidic, clear colorless liquid which is miscible with water in all proportions [MI.Its use has increased especially in bleaching of paper pulp and textiles and in wastewater treatment [ 19,201 since it is a stable, easy to handle, and a non polluting oxidant. It is rapidly replacing chlorine, and hypochlorite as a bleaching agent for paper pulp as they lead to the formation of toxic chemicals such as dioxins and other chlorinated hydrocarbons [21]. The use of hydrogen peroxide as an oxidant in conjunction with UV light or ozone for treating wastewaters contaminated with organics has shown promise since it leads to the formation of hydroxyl radicals which are more powerful in reacting with organics than ozone or peroxide alone [4,22]. These processes, commonly

111

known as Advanced Oxidation Processes (A0Ps)are currently receiving considerable attention in the wastewater treatment area [2,4,6].

It is estimated that the

environmental market for hydrogen peroxide could be challenging the paper industry usage by the year 2000 [20]. On an industrial scale, the predominant manufacturing process for hydrogen peroxide is by the auto-oxidation of anthraquinone. The North American hydrogen peroxide production was estimated to be 445-520 million Ibs in 1989 [20]. Electrochemical processes can compete with this process for small scale operations. At hydrogen peroxide production rates below 3-4 tons/day, electrochemical processes have a cost advantage [23]. Hydrogen peroxide can be electrochemically produced by reducing oxygen at the cathode. The formal half-cell reaction in acidic media is given by:

0, t 2Ht t 2e-

___-

> H,O,

Eo = 0.68 V vs NHE

(4)

This shows hydrogen peroxide to be the product of a two electron transfer to oxygen. However, if oxygen is reduced by a four electron process, water is formed as the end product: > H,O

0, t 4Ht t 4e-

Eo = 1.23 V vs NHE

(5)

Hydrogen peroxide can reduce further to water: > H,O Eo = 1.776 V vs NHE

H,O, t 2Ht t 2e-

(6)

In alkaline media, the reaction stoichiometry is given by: 0,

+

HO-,

H,O

+ 2e'

+ H,O

t 2e' _

> OH- t HO, _

>3O~ H

Eo = -0.076 V vs NHE

(7)

Eo = 0.88 V vs NHE

(8)

112

The perhydroxyl ion, HO,is formed by hydrogen peroxide dissociation in base: H202

> Ht + HO;

(9)

The reverse case of Reaction (7) and Reaction (8) could result in a loss of product or a lowering of observed current efficiency. In addition, peroxide solutions can be catalytically decomposed by trace metal ions [24]: 2H202

> 2H,O

+ 0,

(10)

Reaction (10) may occur homogeneously(in bulk solution)or heterogeneously (at the electrode surface). Although reaction (8) is thermodynamically favored over reaction (7), kinetically, it is sluggish enough that cathodic loss of peroxide maybe significant [24]. Which course oxygen reduction takes depends upon the electrolyte media, its composition, and the electrode used. Kinetic and mechanistic studies on the electroreduction of oxygen have been performed in both acidic [25-311,and alkaline media

[31-381using different metals as well as various types of carbon electrodes. Studies on the production of hydrogen peroxide in significant amounts by the electroreduction of oxygen have been done using carbon cathodes in alkaline electrolytes (i.e. reaction (7)). The trickle bed cell (Figure 3) was investigated by Davison et. aL [39] using graphite chips and Reticulated Vitreous Carbon (RVC)cathodes in 2M NaOH electrolyte. At potentials more positive than -0.W vs. SCE, current efficiencies for low catholyte flowrates were loo%, but at more negative potentials, the current

113

efficiencies dropped steadily. Under identical conditions, the RVC electrodes exhibited similar behavior up to -0.7 V. At more negative potentials, however, they maintained a higher current efficiency and generated higher concentrations of peroxide than graphite electrodes. It was also observed that at slower electrolyte flow rates, the performance of the RVC cathode was found to be significantlybetter than that of graphite chips. For example, at a flow rate of 7 ml/min and a current density of 800 A/m2, the RVC cathode produced 2.32 wt. % peroxide as opposed to only 0.64 wt. % peroxide using the graphite chips. A method for both producing caustic hydrogen peroxide and concentrating it by electrodialysis into an acidic product, using a dual membrane batch cell (Figure 4) consisting of three chambers, was investigated by Kuehn et al. [40]. The cathodic

chamber contained a carbon black porous gas diffusion electrode in 0.5M KOH to which oxygen was bubbled. The anodic chamber contained sulfuric acid electrolyte where water was oxidized to oxygen. The central chamber which was separated by an anion exchange membrane on the cathode side and a cation exchange membrane on the anode side contained 0.1M sulfuric acid. Perhydroxyl and hydroxide ions generated at the cathode migrated into the central chamber, where they were protonated by hydrogen ions entering the chamber form the anode. With such a configuration, acidic hydrogen peroxide accumulated in the central chamber for over

six hours, resulting in a final concentration of 3.3M (11 wt. %) in a 200 ml volume. A modification of the trickle bed cell (Figure 5.) was recently developed by Dow Chemical [24]. The cell contained

two chambers separated by a liquid

114

permeable membrane. Caustic electrolyte seeps into the packed bed cathode (made of graphite chips coated with a TeflonR/carbon black matrix) from the anode chamber through the membrane, wetting the chips but keeping the bed essentially dry. Oxygen gas was pumped downward through the cathode and product was drawn off at the cell bottom. Two percent hydrogen peroxide in 1M NaOH was generated

in a single pass operating at 2 V with 67% current efficiency. Oloman and his co-workershave been investigating the synthesis of hydrogen peroxide in electrochemical flow reactors since the early 1970's [41-441.Their most recent work, Kalu and Oloman [44],involves the simultaneous synthesis of alkaline hydrogen peroxide and sodium chlorate in a bench scale 'flow-by' single cell electrochemical reactor. The peroxide was obtained by the electro-reduction of oxygen in NaOH (0.5-2.0

M) on a fiied bed of graphite felt cathode while the

chlorate was obtained by the reaction of anodically generated hypochlorite and hypochlorous acid using a dimensionally stable anode (DSA). The anodic an cathodic reactions can be represented as follows:

> ClO;

C1- t 6 0 H

30,

+ 3H,O

t 6e-

+ 3H,O

t

6e-

> 3HO; t 3 0 R

(11)

(12)

A RaiporeR 1035 anion membrane, covered with an asbestos diaphragm on

the anode side was used as the separator between the two chambers. The peroxide current efficiency varied from 2046% (0.069-0.80 M), while the sodium chlorate current efficiency vaned from 5141%. Recently, the oxidation of phenols by electrochemically generated hydrogen

115

peroxide was studied (451. The peroxide was generated at the cathode using ultra "F purity grade graohite. During a preliminary experiment, 140 pprn of peroxide was generated at pH 3 in 24 hours. The concentration of phenol decreased form an initial 50 ppm to 25 ppm in 24 hours. Tatapudi and Fenton [46] investigated the continuous production of hydrogen peroxide by oxygen reduction in a proton exchange membrane (PEM) electrochemical flow reactor using commercially available gold, activated carbon and graphite powders. This study was undertaken as a part of an overall scheme to study the simultaneous synthesis of ozone and hydrogen peroxide in the same PEM reactor. Using the graphite powder at a loading of 10 mg/cm2 with 20% TeflonR binder yielded a current efficiency of 10% at 2.5 V.

Potassium Permangunate-KMnO,: Potassium permanganate has long been used for the removal of taste and odors in drinking waters. Its utility in the removal of Trihalomethane (THM)precursors is also known [47].

However, very little

information exists in its use in the treatment of toxic and refractory organic compounds [48]. There is some evidence that it can be used to treat a variety of phenolics, and may yield a decrease in

toxicity resulting in improvement of

biodegradability (491. It has also been shown to oxidize heptachlor with 8040% efficiency [50]. World production of potassium permanganate is about 40,000 tons per year. Electrochemically, potassium permanganate is produced by the anodic

116

oxidation of potassium manganate:

Mn0;-

t

Ht + e t

H,O

> MnO,' t O H

+ Ht + e-

> 1/2 H,

(Anode) (Cathode)

The process is carried out using an electrolyte containing potassium hydroxide and potassium manganate. The anode is made of either nickel or monel and the cathode is made of iron or steel. The oxidation occurs at 60' C at a current density of 5-15 mA cm-2. The current efficienciesare generally between 60-90%. A more detailed review on the synthesis of potassium permanganate can be found in Reference [5 11.

Chlorine and Hypochlorite-C1, & OCk The use of chlorine and hypochlorites is well

established in the sterilization of potable and swimming pool waters, sewage treatment, and paper and wood pulp bleaching. World chlorine production was estimated to be 40.4 million short tons in 1989 [21]. The paper and pulp industry accounts for 14% of the chlorine market in the United States while the disinfection of drinking water represents 5%. Both these numbers are expected to decline in the future as these industries are shifting to the use of less toxic oxidants (ozone, hydrogen peroxide and chlorine dioxide) for bleaching and disinfection purposes. The use of chlorine in the oxidation of toxic and refractory organics is also extremely limited due to their ability to form chlorinated organics. The electrochemical synthesis of chlorine takes place through the electrolysis of aqueous sodium chloride. The electrode reactions are:

117

> C1, t 2e'

2c1-

> H, t O H

2H,O t 2e-

(anode) (cathode)

Low tonnage electrolyzers, which are typically needed for wastewater treatment purposes use the membrane cell technology for the manufacture of chlorine [fill. Unlike mercury or diaphragm cells, the membrane technology is more suited for intermittent production of chlorine. The reactions involved in the synthesis of hypochlorite are essentially the same

as that for chlorine evolution. A major difference in cell design is that a separator is not used since the anodically formed chlorine is hydrolyzed by the cathodically formed hydroxide: C1, t 20H-

~~

> H,O t OC1- t C1-

The cathode material is usually stainless steel, a nickel alloy, or titanium. Graphite, lead dioxide, platinized titanium, and dimensionallystable anode are some of the materials that have been used as an anode. The cells are operated

between a current density of 0.1-0.5 A cm'2. The majority of the cells used to produce hypochlorite use the parallel plate type geometry [Sl].

Chlorine Dioxide-Cl0,: Chlorine dioxide, like chlorine and hypochlorite, has found

wide use as a disinfectant in water treatment/purification and as a bleaching agent in the paper and pulp industry. Unlike chlorine and hypochlorite, however, it is not known to form residual chlorinated organic species when used as a chemical oxidant in treating organic wastewaters [2] and has been shown to react with polyaromatic

118

hydrocarbons, alkenes, and phenolic compounds [48,52-551. Chlorine dioxide must be manufactured where it is going to be used because of its explosive character. Large scale production uses the chlorate reduction technology which takes place in an acidic media. Small scale generation (which is what will be needed for waste water treatment) currently use a chemical reaction between a chlorite salt (sodium chlorite), and an acid and or/oxidizing agent (hydrochloric acid and chlorine), preferably in combination. A number of methods of producing chlorine dioxide electrochemically using perm selective membranes are found in the patent literature. Harke et. aL produced chlorine dioxide in a three compartment electrolyticcell [56]. A buffer compartment located between the anode and cathode compartments was separated from the cathode side by a cation exchange membrane and from the anode side by an anion exchange membrane.

Hydrogen chloride was fed to the anode compartment,

aqueous alkali metal chlorate and chloride to the buffer compartment, and water to the cathode.

Chlorine and chlorine dioxide are taken off from the anode

compartment while hydrogen and alkali metal hydroxide are removed from the cathode compartment. Sweeney et. ul. used a similar three compartment cell, but included a bipolar electrode in the middle compartment between the anion and cation exchange membranes in addition to a cathode and an anode in their respective compartments [57]. When the cell was filled with brine and a D.C. current applied to it, hydrogen

was evolved at the cathode and a mixture of chlorine and chlorine dioxide at the

119

anode. Lipsztajn ef. d ,[58] produced chlorine dioxide in the cathode compartment

of an electrolytic cell using a three dimensional high surface area cathode. The cathodic section was separated from the anodic section by a cation exchange membrane. The hydrogen ions generated at the anode migrate to the cathode where they react with sodium chlorate along with chloride ions to form chlorine dioxide. The chlorine gas co-produced with the chlorine dioxide is reduced at the cathode to provide chloride ions. More recently, Kaczur, ef. d (591 patented a process for electrolytically producing an aqueous solution of chlorine dioxide in a cell possessing anode and cathode compartments and at least one cation exchange resin compartment between the anode and cathode compartments. An aqueous solution of an alkali metal

chloride (sodium chlorite) is fed to the cation exchange resin compartment. Hydrogen ions are liberated due to water electrolysis at the anode using a nonoxidizable acid (sulfuric) as the anolyte and an oxygen generating anode. The hydrogen ions pass into the ion exchange compartment through a cation exchange membrane. As a hydrogen ion enters the stream, a sodium ion passes through a different cation exchange membrane located adjacent to the cathode to maintain electrical neutrality. The exchange of ions and the formation of chlorine dioxide is represented by the following reactions: 4Ht t 4NaC10, 4HC10,

> 4HC10, t 4Nat > 2C10,

+

HClO,

+

H,O

120

The production takes place at 50°-7$ C and at a current density of 5 X 10”

-

A/cm2 1 A/cm2.

121

References 1.

Fxkenfelder, W.W., Jr. Industnizl Water Pollution Control, 2nd edition. McGraw-Hill Inc., New York, 300-311(1989).

2.

Bowers, A.R., Toxicity Reduction in Industrial Effluents, (P.W. Lankford and W.W. Eckenfelder, Jr., ed.), Van Nostrand, Reinhold, New York, 247-272 (1990). Scow, K.M., Handbook of Chemical Property Estimation Methods, (W.J. Lyman, W.F. Reehl, D. H. Rosenblatt, ed.), McGraw-Hill,Inc., New York, 9-3 (1982). Glaze, W.H, J. Kang, and D.H. Chapin., Ozone Science and Engineering, 9,335 (1987). Kirk, R.E and D.F. Othmer (Editors), Encyclopedia Of Chemical Technology, V. 16, 3d Ed, John Wiley & Sons, New York 683 (1981). Nebel, C., Seventh Annual Semiconductor Pure Water Conference, 279 (1988).

7.

Foller, P.C., and M. L Goodwin, Chem. Engr. Progress, 81:3, 49 (1985).

8.

Fischer, F and K. Massenez, Anoq. Chem, 52, 202 (1907).

9.

Seader, J.D and C.W. Tobias, I d Eng. Chem, 44,9 (1952).

10.

Briner, E, R. Haefeli, and H.Paillard, Helv. Chim. Acta, 120, 1510 (1937).

11.

Semchenko, D.P., E.T. Lyubushkina, and V . Lyubushkin, Electrochimyiya, 9 1744 (1973).

12.

Semchenko, D.P., E.T.Lyubushkina, and V.Lyubushkin, IN. Sev-Kuk Nmtchn. Tsentra Vyssh Shk Ser. Tekn. Nauk, 3, 98 (1975).

13.

Foller, P.C., and C.W. Tobias, J. Electrochem. Soc., 129, 506 (1982).

14.

Foller, P.C., and M.L Goodwin, Ozone Science and Engineering, 6,29 (1984).

15.

Stucki, S., G. Theis, R.Kotz, H.Devantay, and HJ. Christen, J Electrochem .Sot., 132, 367 (1985).

16.

Stucki, S.,H. Baumann, H.J. Christen, and R. Kotz, J AppL Electrochem., 17

122

773 (1987). 17.

Tatapudi, P.,and J. M. Fenton, Proceedings of the Symposia on Electrochemical Engineering and Small Scale Electrolytic Processing. (C.W. Walton, R.D. Varjian, and J.W. Van Zee, Ed.), The Electrochemical Society Inc., 90-10,275 (1990).

18.

Kirk, R.E and D.F. Othmer (Editors), Encyclopedia OfChemical Technology, V. 13, 3d Ed, John Wiley & Sons, New York (1981).

19.

Chemical Engineering, June 20, 1988, p. 32.

20.

Chemical & Engineering News, September 11, 1989, p. 15.

21.

Talbot, J.B., and A.D. Fritts, Report of the Electrolytic Industries for the Year 1991, The Electrochemical Society, May 1992.

22.

Bellamy, W.D., G.T. Hickman, P.A. Mueller, and N. Ziemba, Res. J. Water Pollut. Control Fed., 63, 120 (1991).

23.

Schmidt, A.H.H., Chem. Ing. Tech, 37, 832 (1965).

24.

Mclntyre, J.A., and R.F. Phillips, Proceedings of the symposium on Electrochemical Process and Plant Design, (R.C Alkire, T.R. Beck, and R.D. Varjian, ed.), The Electrochemical Society Inc., 83-6 79 (1983).

25.

Hoare, J.P.,,Electrochim. Acta, 11, 311 (1966).

26.

Gensahw, M A , A.Damjanovic and J.O'M Bockris, J. Electroanal. Chem, 15, 163 (1967).

27.

Sabirov, F.Z., M.R. Tarasevich, and R.U. Burshtein, Electrokhimiya, 6, 1130 (1970).

28.

Sabirov, F.Z., M.R. Tarasevich, and R.U. Burshtein, Electrokhimiya, 7, 404 (1971).

29.

Bianchi, G., F. Mazza, and T. Mussini, Electrochim Acta, 2, 1509 (1966).

30.

Otsuka, K., and I. Yamanaka., Electrochim. Acfa, 35,319 (1990).

31.

Hoare, J.P., The Electrochemistry of Oxygen, John Wiley & Sons, Inc., New York, (1968).

123

32.

Damjanovic, A, M.A. Gensahw, J.O'M Bockris, 1 E l e c t r o d Chem, 15, 173 (1967).

33.

Wroblowa, H.S., Y.C. Pan, and G. Razumney, J.Electrochem Soc., 69, 195 (1976).

34.

Berl, W., T i m . Electrochem. Soc., 83, 253 (1943).

35.

Yeager, E., P. Krouse, and K.V. Rao, Electrochim Acta, 9, 1057 (1964).

36.

Taylor, R.J., and A.A. Humffray, J. Electrochem. Soc.,64, 63 (1975).

37.

hvrecek, B., M. Batinic, and J. Caja, Electrochim Acta, 28, 685 (1983).

38.

Appleby, A.I., and M. Say, J. Electrochem. SOC., 51, 15 (1978).

39.

Davison, J.B., J. M. Kacsir, P. J. Peerce-Landers, and R. Jasinski, J. Electrochem. SOC., 130, 1497 (1983).

40.

Kuehn, C., F. Leder, R. Jasinski, and K. Gunt, Electrochem. Soc., 130,1117 (1983).

41.

Oloman, C., and k P . Watkinson, Can. Pulp. Pap. I d . , 54, 312 (1976).

42.

Oloman, C., and A. P. Watkinson, J. Appl. Electrochem., 9, 117 (1979).

43.

Oloman, C., Can. Pulp. Pap. Ind., 63 (July, 1980).

44.

Kalu, E.E., and C.Oloman, J. Appl. Electrochem., 20, 932 (1990).

45.

Huang, C.P., and C. Sheng, Extended Abstract 28K American Institute of Chemical Engineers, Summer National Meeting, Aug. 18-21, 1991.

46.

Tatapudi, P. and J. M. Fenton, "Synthesis of Hydrogen Peroxide in a Proton Exchange Membrane Electrochemical Reactor", submitted for publication in Journal of the Electrochemical Society, July 1992.

47.

Kirk, R.E and D.F. Othmer (Editors), Encyclopedia OfChemical Technology, V. 24, 3d Ed,John Wiley & Sons, New York 404 (1981).

48.

Benschote, J.E., W. Lin, and W.R. Knocke, Environ. Sci. TechnoL, 26-7, 1327 (1992).

49.

Throop, W.M.,J. Hazard Muter., 1,312 (1977).

124

50.

Kirk, R.E and D.F.Othmer (Editors), Encyclopedia OfChemical Technology, V . 24,3& Ed, John Wiley & Sons, New York 310 (1981).

51.

Pletcher, D., and F.C. Walsh, Industrial Elecrrochemirrry (2"d Ed.), Chapter 3,5, 7, Chapman and Hall, New York, 1990.

52.

Amor, H.B., J. DeLaat and D. Moore, Water Research, 18, 1545 (1984).

53.

Rav-Acha, Ch. and R. Blits, Water Research, 19, 1273 (1985).

54.

Rav-Acha, Ch. and E. Chosen, Water Research, 21, 1069 (1987).

55.

Masschelein, W.J., J. Amer. Water Works Assoc., 76, 70 (1984).

56.

U.S.Pat. 3,904,496 (Sept. 9, 1975), Harke, C.J., and co-workers (to Hooker Chemicals and Plastics).

57.

U.S. Pat. 4,308,117 (Dec. 29, 1981), Sweeney, C.T.

58.

U.S. Pat. 4,853,096 (Aug. 1, 1989), Lipsztajn, M., and co-workers (to Tenneco Canada, Inc.).

59.

U.S. Pat. 5,106,465 (Apr. 21, 1992), Kaczur, et. al., (to Olin Chemical Corporation).

125

i

Diluent Gas

Product Gas

v

+

/

Porous Air Cathode

02+ 4H** 4e'= 2H20

02current efficiency)

\

Anode

3 H 2 0 * O,+ 6H'+ 6e' and 2H20 = 02+4H'* 4 e *

Figure 1.

Schematic of the cell used in the synthesis of ozone using the HBFJGlassy Carbon route [51]. (By permission from Chapman and Hall)

H,0+H2

H,O+O,+O,

Membrane Electrode Assembly

Figure 2.

Schematic of the cell used in the synthesis of ozone using the Lead Dioxide/Solid Polymer Electrolyte route [IS]. (By permission from The Electrochemical Society)

126

Hollow Stainless Steel Anode

Naflon Membrane Gas/Electrolyte Inlet

3mm Glass Beads

Hollow Stainless Steel Cathode Feeder Plate

Gas/Electrolyte Outlet Cathode Chamber Packed With Graphite Particles

Figure 3.

Schematic of the two compartment trickle bed cell used for hydrogen peroxide synthesis [39]. (By permission from The Electrochemical Society)

127

H20

+

I

Cation Membrane

Anion Membrane

-

+

H$

J

L

Hd*

!H'+

2

OH-

H2s04

H202

Figure 4.

in H+04

Schematic representation of the operations involved in the production of acidic hydrogen peroxide [40]. (By permission from The Electrochemical Society)

128

Anode Gas out

Conductive Screen

Anolyte

Figure 5.

Schematic of a diaphragm flow controlled trickle bed cell used by Dow Chemical Co. for the synthesis of hydrogen peroxide [24]. (By permission from The Electrochemical Society)

SECTION TWO

BATTERIES AND ENVIRONMENT

This Page Intentionally Left Blank

131

BATTERIES AND THE ENVIRONMENT Per Bro Southwest Electrochemical Company Santa Fe, NM 87501 USA. and Samuel C. Levy Sandia National Laboratories Albuquerque, NM 87185 U.SA

1. INTRODUCTION The awareness that batteries may pose environmental and health risks has been recognized since the toxicity of various battery materials became known. Lead is an example. It has been known for some time that the inhalation or ingestion of lead fumes and lead compounds from paint and automobile exhaust may cause damage to the central nervous system, and that exposures to lead-containing materials from any source, including batteries, should be limited. Although battery manufacturers began to recycle lead for economic reasons long before its harmful effects were fully recognized, the general awareness of its toxicity had led them to limit worker exposures and to limit the amount of lead discharged from their lants. The case of mercury is similar in some respects. Its toxicity was well known Pons before mercury batteries were developed. When mercu batteries first appeared, they caused little public concern, primarily because t eir distribution and use were quite limited. The mercury s ill at Minamata Bay in Japan, the growing use of mercurials as fungicides, and t e finding that mercury had become widely distributed in various ecosystems and had entered the food chain in various ways led to an increased concern regarding the possible spread of mercury from any source, including batteries. The case of cadmium is more recent even thou h cadmium has been used in Ni/Cd batteries ever since their invention in 1899. "ke reason for the belated awareness of cadmium as a toxic material is connected with the rowth of Ni/Cd batteries as a general consumer item. Indeed, the growth of pub ic concern about the possible toxic effects of batteries is closely connected with the rapid growth and ever widening distribution of batteries for consumer electronics applications and their subsequent appearance in municipal landfills in increasing amounts.

P

g

B

It is not so much that the disposal of batteries in uncontrolled landfills has an immediate and serious effect on the environment. Rather, it is the recognition that their accumulation, and eventual entry into the environment, may cause harm. Regardless of when that may occur, it is thought advisable to take the necessary steps at the earliest possible time to prevent such an eventuality from occurring, preferably by recycling the batteries, even if there are no compelling short term economic incentives to do so.

132

In the design of reprocessing facilities for high volume operations, conventional metallurgical techniques are employed to the fullest extent ossible. Both the simple thermal methods and the more complicated hydrometa lurgical techniques are used. As more restrictive limits are imposed on the allowable airborne, waterborne, and solid dischar es from plants, however, it seems inevitable that the more elaborate, the more efficient, and the more expensive chemical processing techniques will come to dominate the recycling field. The ideal objective is to operate fully closed loop facilities, apart from the materials to be rocessed, but that is a practical im ossibility. Instead, re clin o erations should e designed to satisfy two primary o jectives in regard to p ant isc arges:

P

7 (BR

\

!

1. The concentration of toxic materials in all the waste streams must be below prescribed limits. 2. The total amounts of toxic materials discharged from a plant in specified time periods must be below prescribed limits. The rates of discharge are also important. In this chapter we introduce batteries in general, with few technical details. We consider the possible environmental impact of batteries relative to that of other sources of the same toxic materials, and we review briefly some re latory aspects of battery manufacture, recycling and disposal. Also included is a Ecussion of the possibility of developing new battery systems and modifications aimed at the design of environmentally friendly batteries. Various recycling schemes are discussed from an elementa point of view; expert readers are referred to the literature cited at the end oft e chapter for more detailed technical information.

x

1.1 The Classification of Batteries

Many different kinds of batteries have been developed to serve a variety of needs in various markets. Based on their mode of operation, we distinguish between the following types of batteries: Primarv batteries are distinguished from others bv the characteristic feature of essentid non-rechargeability." They are discarded when their electrical energy has been consumed. Examples: Zinc Carbon, Alkaline Manganese, Silver/Zinc, Lithium/Manganese, Lithium/Sulfur ioxide, Lithium/Thionyl Chloride.

d

- R Reserve batteries are a special type of primary battery distinguished by the feature that they need to be activated by mechanical or pyrotechnic means 'ust prior to use, whereby electrolyte is injected into the electrode chamber o f tlle batteries. They are discarded after a one-time use. Examples: Silver/Zinc, Water activated Magnesium/Lead Chloride, Ammonia activated Magnesium/Dinitrobenzene, Spin activated Lead/Fluoroboric Acid. Thermal Batteries Thermal batteries represent another e of primary reserve battery distinpished by the feature that they are activated y heating to a high temperature by internal

??

133

pyrotechnic means. They operate only at high temperatures and are discarded after a single dischar e. Examples: CalciumfTungstic Oxide, Calcium/ Calcium Chromate, Lithium/Iron Sulfide. Fuel Cells Fuel cells represent another variant of primary batteries distinguished by the interesting feature that the oxidizin agent, the cathodic material, is oxygen taken from the surroundin air. These atteries must remain open to the air during operation. The meta /air fuel cells are discarded when their negative electrodes have been discharged. Examples: Aluminum/Air, Zinc/Air, Methanol/Air, Hydrogen/Oxygen.

s

%

Secondary Batteries Secondary batteries are rechargeable batteries that can be re-used many times, several hundred times for consumer batteries and several thousand times for s ecially designed secondary batteries. Lamples: Nickel/Cadmium, Lead-Acid, Silver/Zinc, Nickel/Metal Hydride. Advanced B a t t h Other types of batteries have been developed, but, with some notable exceptions, they have not yet reached commercial importance. Among these, we may mention flow batteries and hi h tem erature solid state or fused salt batteries. Some of these batteries are rec argea le and some are suited for high capacity applications only. Examples: Zinc/Bromine, Sodium/Sulfur, Solid Electrolyte, and Polymer Systems.

a !

The distribution of the different types of batteries among various ap lications is an important consideration from the point of view of the control o their possible entry into the environment. Reserve and thermal batteries are used almost exclusively by military organizations and their disposal can, in principle, be controlled to a much higher degree than can the disposal of primary and secondary batteries used by the general ublic. A further classification may be considered for the rechargeable batteries. &e use and disposal of traction and automotive type batteries, as well as any kind of battery used by fleet operators, can, by the very nature of their users, be well controlled. In fact, a very large fraction of batteries entering into these cate ories of usage are already being recycled, rather than discarded in landfills. d e principal environmental problems are associated with the smaller size batteries, both primary and secondary, that are widely used by the general public in a variety of applications, and that are discarded in an uncontrolled manner.

P

An interesting observation (1) is that the types of batteries used by the general

public in different countries may vary considerably. For example, the low cost zinc/carbon batteries are more prevalent in underdeveloped countries than are the more expensive alkaline manganese batteries. The latter are more important in highly developed countries. Battery usage appears to be closely coupled with the usage of consumer electronic devices. The product mix of batteries used in any particular country needs to be considered when battery recycling and disposal processes are designed for that country.

134

1.2 Battery Life Cycles

The typical life cycles of consumer batteries depend to some extent on their applications. Primary batteries employed in portable devices are most often discarded casually with domestic trash and end up in landfills. Exceptions exist in communities where deliberate efforts are made to collect spent batteries for revcling or proper disposal and in communities where battery vendors partici ate in incentive programs to return spent batteries to the manufacturers or co lect them for recycling or proper disposal. The magnitude of such programs is still uite limited and has contributed little to alleviate the environmental problems %at may be created by the disposal of spent, primary consumer batteries.

P

Small nickel/cadmium and lead-acid batteries generally experience the same fate. With some exceptions, the larger rechargeable batteries, automotive batteries in particular, are returned to the vendors to a large extent for subsequent recyclin by the manufacturers, or for processing by scrap metal operators for recovery o f t eir intrinsic metal values.

fl

During normal battery usage, there is no reason to expect the batte contents to leave the batteries and enter the environment in an uncontrol ed manner. Accidents do occur, however, due to inadvertent or deliberate abuse, or due to battery damage caused by other events such as collisions or fires. Occurrences of this kind may generate excessive local concentrations of toxic or harmful substances or other hazards, such as acid or alkali burns, toxic vapors, or explosions, with undesirable effects on nearby persons or on the local environment. Apart from the local and temporary effects, events of this kind are rare and do not significantlycontribute to environmental contamination in general.

7

1.3 Battery Materials

Most of the batteries available today are of a high quality, robust, and able to withstand a reasonable amount of abuse without rupturing or spilling their contents. Our concern is, therefore, less with what may happen durin normal more importantly, what is likely to happen when batteries are &carded whether in landfills or during recycling, and their contents enter the For the design of the proper handling, disposal, and recycling methods it is necessary to know what materials are contained in the batteries. We present a brief summary below of the major chemical species present in the most frequently used batteries. Information on the contents of less commonly used batteries may be obtained from their manufacturers or by chemical analyses of the batteries themselves. Zinc/Carbon Batteries (Regular and Heaw Duty Iron case, (tin and copper ma be present in sma1)amounts). Zinc anode, (may contain cadYmium and lead). Manganese dioxide cathode with some metal impurities. Carbon, acetylene black, and graphite. Ammonium chloride electrolyte, zinc chloride,(chromate may be present). Starch, paper, asphaltics, polypropylene. Alkaline Maneanese Batteries Nickel plated steel can (tin and copper may be present in small amounts).

135

Zinc anode with unspecified corrosion inhibitors (very little mercury). Manganese dioxide cathode (high purity). Carbon, conductive component. Potassium hydroxide electrolyte with dissolved zinc oxide. Cellophane, nylon, carboxymethylcellulose. Silver Oxide Cells Essentially the same constituents as alkaline manganese cells, except that silver oxide has replaced manganese dioxide. Mercurv Cells Essentially the same constituents as alkaline manganese cells, except that mercuric oxide has replaced man anese dioxide. Silver may be present in the cathode. Mercury cells are being p ased out wherever feasible.

fl

Zinc/Air Cells Essentially the same constituents as alkaline manganese cells. excem that man anesedioxide is replaced with a catalytic electrode that may contain carbon, smal amounts of silver, manganese dioxide, and some heavy metals as catalysts in addition to teflon type binders.

f

Lithium ManPanese Batteries Nickel plated steel case. Lithium anode with or without a metal screen. Manganese dioxide cathode (high urity). Carbon conductive component, te on binder. Polypropylene separator and insulators. Organic solvent electrolyte. (The solvents actually used may differ in cells from different manufacturers. Typically, they include propylene carbonate, dimethoxyethane, or similar solvents). Electrolytic salts such as lithium perchlorate, lithium hexafluoroarsenate, and lithium tetrafluoroborate.

R

Thionyl Chloride Batteries Stainless steel or nickel plated steel case. Thionyl chloride active cathode material. Teflon bonded carbon current collector with nickel grid. Lithium aluminum chloride electrolytic salt. Fiber glass separator. Lead-Acid Batteries Pol ropylene or hard rubber case. Lea anode with antimony or calcium alloying constituents (Low concentrations of other metals may be present). Lead dioxide cathode. Sulfuric acid electrolyte. Polyvinyl chloride, polyolefins, phenolic bonded paper, epoxy sealants.

ys

Nickel-Cadmium Batteries Nylon, polypropylene, or stainless steel case. Nickel plated copper terminals. Cadmium anodes.

136

Nickel oxyhydroxy cathodes. Potassium hydroxide electrolyte. Nylon or pol propylene separators (wetting agents may be present in low concentrations!. Nickel-Metal Hydride Batteries Essentially the same constituents as Ni/Cd batteries, except that cadmium and cadmium hydroxide have been replaced with a metal hydride formed from hydrogen and a mixture of metals that may contain nickel, cobalt, lanthanum, and various mischmetal components (rare earth metals). The approximate metal contents of various batteries in terms of weight percentages are as follows (2):

Pb-Acid Ni/Cd 1.4 Toxicity

The preceding list indicates the variety of materials that are employed in the many different types of batteries that are in use today and which are available to the eneral consumer. Some of these materials may cause harm to users if the atteries are grossly abused, i.e., under conditions not likely to occur unless deliberate attempts are made to abuse the batteries. Althou h mishaps ma occur due to inadvertent abuse, with few exce tions, batteries have een found to e very safe from the time they leave the manu acturers until their disposal.

5

P

%

i:

A different situation exists during battery manufacturing and during batte disposal/recycling operations where workers may come into direct contact wit toxic or harmful battery components. In addition, members of a community may be unknowingly exposed to toxic battery materials due to the improper disposal of battery wastes and the injection of toxic battery materials into their ecosystems. The latter may affect both man, animal, and plant life adverse1 . In general, the toxic effect of a particular material depends upon the nature ofythe exposure and its concentration; whether it be by contact, ingestion, inhalation, or a combination of these.

2

Although the regulatory climate varies from one country to another, the manufacturing environments have been subject to more or less stringent state control for some time now in many of the countries where batteries are made. The regulations take into consideration the nature of the materials used, as far as is known, and they provide limits applicable to short term and lon term exposures as well as procedures and/or rules governing the handling andp discharge of toxic waste materials. Since battery manufacturing operations are reasonably well controlled in most countries where batteries are made and since the manufacturers are well aware of the hazards associated with their operations, the battery

137

manufacturing aspects of the environmental problems will not be discussed here. It is only recently that attention has been drawn to the potential community hazards associated with the dis osal of batteries in general. We focus attention on the disposal/recycling aspects trlat may be of interest to the general public. The battery materials of foremost environmental concern at the present time are mercury, lead, and cadmium, but recent efforts have contributed significantly to the reduction of their dispersal in the environment. For example, the redesign of alkaline manganese batteries that contained an appreciable amount of mercury in the past has led to a new eneration of alkaline manganese batteries that are ractically free of mercury. %he recycling of lead-acid batteries has been practiced !or some time, and it has led to a significant reduction in the amount of lead that might otherwise have entered the environment. In the case of cadmium, much remains to be done, but the pending introduction of nickel/metal hydride batteries as a replacement for nickel/cadmium batteries may do much to alleviate this problem. For our purposes, there is no need to discuss the detailed medical and public health aspects of the toxic substances present in batteries. It is sufficient to recognize that their effects are many and varied, and that they may range from short term to long term effects, as well as long term-delayed effects due to the cumulative retention of the toxic substances or their derivatives in the bod . Information on the effects of many battery materials on human health is availab e in the comprehensive listing prepared by Sax and Lewis (3) and the references cited by them.

1

r

The relative toxicit of a material may be expressed by the maximum allowed exposure in terms o the time weighted avera e exposure (TWA) or the threshold limit value (TLV) for short term exposures. f i e r e 1s a possibility that the currently acce ted values may be lowered as more information becomes available, especially on t e effects of long term, low level exposures. Some values selected from Sax and Lewis are:

R

Lead Cadmium Mercury

TLV: 0.05 mg/ms

These values apply to workers exposed to battery processing operations, whether manufacturing, recycling, or disposal operations. The values ap licable to drinking water may be of greater interest to the general public. From t e Code of Federal Regulations in the United States we obtain the following values (4):

R

Lead Cadmium Mercury

0.050 mg/l 0.010 mg/l 0.002 mg/l

The introduction of advanced batteries based on lithium or on metal hydride technologies has not yet reached the point where they have created an environmental concern, but when their usage becomes more widespread they will introduce additional problems. In particular, lithium batteries containin thionyl chloride (TLV = 1 ppm) or sulfur dioxide (TLV = 2 ppm) and acetonitri e (TWA = 40 ppm) may create disposal problems not only because of their toxicity, but also

k

138

because of the fire hazard associated with lithium metal. In addition to various flammable organic solvents (for example methyl formate, tetrahydrofuran, and others), lithium batteries contain salts (lithium tetrafluoroborate, lithium hexafluoroarsenate, lithium perchlorate, and others) and various heavy metals, some of which may create environmental problems, not so much at the user level as at the disposal/recyclin level. The metal hydrides that replace cadmium in Ni/Cd batteries a r e like y to contain heavy metals such as vanadium (V), lanthanum (La), cobalt (Co), and various misch metal components (cerium group metals with atomic numbers ranging from 58 through 71), some of which may be toxic.

P

An important aspect of the battery problem is the question of the pathways of the

toxic battery materials in various ecosystems, whether they enter by leaching from landfills or via the atmosphere from incinerators. We do not discuss this difficult problem here. It is an area about which little is known, although work in progress is providin useful information of direct interest to both regulatory organizations and the pu lic in general. The reader is referred to the literature cited at the end of the chapter for information on this subject.

%

2.

THE ENVIRONMENTALPROBLEM

The problem created by the uncontrolled disposal of batteries may be stated simply: some batteries contain toxic materials whose injection into ecosystems may cause harm, When added to landfills, the toxic materials may enter the local groundwater system and propagate through the food chain in various ways. If processed in improperly designed or operated incinerators, the toxic materials may enter the ecosystem via the airborne route. Although the organic solvents and other materials in lithium batteries do not present a significant problem at this time, they may also cause harm if improperly processed after being discarded. The many factors that affect the athways of toxic materials in ecosystems are not well known. Therefore, it is fifficult to predict exactly how and when such materials may enter the foodchain and to quantify the magnitude of their harmful effects. Faced by the actual and potential threat to public health, the most reasonable policy is to prevent the potential entry of the toxic materials into ecosystems at the source. Apart from the long known toxicity of lead and the measures taken to decrease the hazards associated with the improper handling and disposal of lead-acid batteries, only recently has attention been focussed on the need to control the environmental release of other toxic batte materials. The most critical of these materials are cadmium and mercury. Ef orts underway in many countries today suggest that significant reductions will be achieved in the amounts of mercury and cadmium likely to be dischar ed into various ecosystems during the coming years. However, the uncontrolled (f!isposal of alkaline manganese and nickel-cadmium batteries during the ast years has added an unknown amount of mercury and cadmium to older land ills. Their distribution in various ecosystems represent a potential threat that is likely to persist for some time, even if no further additions of mercury or cadmium are made to existing landfills. These problems are more likely to be encountered in urban rather than rural communities and in communities near landfills. It is an open question whether the water purification methods practiced

r

P

139

in vulnerable communities are effective in removing mercury, lead, cadmium, and other toxic materials originating from batteries and other wastes. We do not discuss the important problems associated with the secure containment of toxic materials in landfills nor the possibility of detoxifying landfills ( 5 ) . The serious nature of the health problems caused by the ingestion of mercury, cadmium, and lead is beyond uestion. However, sources other than batteries also contribute to the magnitude o the potential and actual health problems caused by these materials. Barnett and Wolsky (6) have collected information on the fractions of the total amounts of lead, cadmium, and mercury produced in the United States that enter into battery manufacturing. Their estimates for 1989 are as follows:

9

Material

Total Consumption (Metric tons)

% of Total in Batteries

Pb Cd Hg

1.28~10s 4.1 xloS 1.2 xi03

78.9 27.0 20.6

Comparable estimates for Western Europe are (2): Pb Cd Hg

1.58~106 6.1 x103 2.0

44 23 3

r

It ma be noted that a substantial fraction of the lead in spent lead-acid batteries is recyc ed and not discarded. In the United States, about 95% of the lead used in lead-acid batteries is recycled (1989) and similar figures are likely to apply to other industrialized countries, if not now, then soon. In the case of Ni/Cd batteries, a smaller fraction is recycled, but that situation is likely to improve as more recycling facilities come on stream. The battery industry has made a concerted and relatively successful effort during the past decade to eliminate mercury from their products, and, in combination with mercury recycling, these efforts may be expected to significantly reduce the addition of mercury from batteries to communal waste streams. It becomes clear that the battery contributions to the overall problem is a matter of concern. The solution of the public health problem needs to be approached as a whole by extending the required treatments to cover all the sources of these public pollutants. Since toxic materials from improperly disposed batteries and other sources may enter ecosystems that extend beyond national boundaries, the solution of the environmental roblem requires international cooperation as well as technical cooperation. d e efforts of the European Community of Nations (7,8) is a good example of what can be accomplished within an international framework, although much more needs to be done.

140

3.

REGULATORY ASPECTS

Many countries worldwide are implementing legislation to re late various types of batteries, from manufacturing through disposal and/or recyc ing. It is the intent of this chapter to review the trends that are occurring worldwide in the area of regulations affecting batteries, not to get into the specifics of the re ulations. Legislation is still evolving and varies, not onl from country to country, ut within countries, i.e., between individual states and&r municipalities. A good source of information concerning the legislation passed or under consideration in various countries or states are the Proceedings of the International Seminars on Battery Waste Management, sponsored by Ansum Enterprises and BDT, Inc.

P

6

The trend in the United States is towards increasing concern at the local overnment levels. In many instances, state and munici a1 regulations are %ecomingmore stringent than federal regulations. Recent egislation in several states prohibit the disposal of used batteries in munici a1 solid waste (MSW) and require the manufacturers of sealed lead/acid and NiyCd batteries to rovide for proper collection, transportation and processing of waste batteries. n addition, products containing these batteries must be labeled and the batteries must be easily removable by the consumer. The issue of household batteries is also being addressed. The trend is to prohibit these batteries from disposal in MSW.

f

f

In an attempt to standardize state regulations, the Battery Council International has drafted model le islation for the states. Some states have adopted this language, while others i?ave not. At the federal level, disposal of waste batteries is regulated by several statutes (9). The Resource Conservation and Recovery Act (RCRA), enacted in 1976 as an amendment to the Solid Waste Disposal Act is intended to regulate hazardous waste from its initial generation to its final dis osal. This act defines hazardous solid waste in two general categories; listed or c aracteristic. If a solid waste is not listed on one of the Environmental Protection Agency’s (EPA) lists of hazardous wastes, the generator must determine if the waste exhibits a characteristic of a hazardous waste. Four general characteristics of hazardous waste have been identified by the EPA. These are: 1) ignitability, 2) corrosivity, 3) reactivi? and 4) extraction procedure toxicity. Once the generator determines that the so id waste is hazardous, he must meet specific regulatory requirements that govern how to manage that waste.

hp

All waste batteries are considered RCRA solid waste except those that are returned to the manufacturer for regeneration, reused as an ingredient without reclamation, reused as a substitute, or returned as raw material. The largest ercentage of disposed batteries are part of household waste and therefore exempt From regulation. The remaining waste batteries must be determined by the enerator whether they ualify as a characteristic hazardous waste. Penalties can %eassessed if not correct y determined.

9

The Comprehensive Environmental Response Compensation and Liability Act (CERCLA) of 1980, known as the “Superfund,” as amended by the Superfund Amendments and Reauthorization Act (SARA) in 1986, regulates past and present releases of hazardous substances into the environment. This is the first major environmental law designed primarily to correct past disposal practices. It seeks to

141

impose liability without regard to fault upon owners, operators, generators and transporters of hazardous substances at sites where there has been a release into the environment. This legislation enables the government to either spend funds to clean up and then seek reimbursement from the owners, operators, generators and transporters, or sue to compel liable parties to clean up a site. Other pertinent legislation includes the Clean Air Act, the Clean Water Act, and the Toxic Substances Control Act (TSCA). Within the European Community, 25 directives, regulations, and decisions have been enacted relatin to waste management (8). They are designed to encourage the prevention andfor reduction of waste by either the marketin of products designed to reduce the amount of hazardous waste durin their fulIg life cycle, or the recovery of waste by re cling, reuse or reclamation. ?$Ie cost is to be borne by the holder of the waste andror the previous holder or producer. Specific legislation relating to batteries includes the Toxic and Dangerous Waste Directive which gives priority consideration to mercury, cadmium and lead. This directive is being replaced by a Directive on Hazardous Waste which specifically includes batteries and other electrical cells. Additional metals to be covered include nickel, cobalt, silver, zinc and lithium. Thus, all commonly used batteries will be covered by this directive. The Directive on Batteries and Accumulators is focused on lead, cadmium and mercury. Its goal is to approximate the laws of the member states on the recovery and dis osal of spent batteries and accumulators containing: >25 m Hg per cell, > .O259! Cd by weight, and > .4% Pb by weight. In addition, Ni/C dg and Pb/acid batteries may only be incorporated into appliances where they can be readily removed by the consumer when spent. The batteries and appliances, where appropriate, must be labeled to indicate: separate collection, recycling (where appropriate), and heavy metal content. Other re ulations related to the safe handling and disposal of waste batteries include: birective 67/548 on the classification of dangerous substances, the Base1 convention which addresses the movement of waste, the Directive on Landfill Waste, and the Directive on Packaging and Packaging Waste. Individual countries within the European Community handle the battery waste problem differently. For example, in Switzerland all used consumer batteries are considered hazardous waste and must be collected separately from ordinary household waste. Batteries must be recycled or stored in warehouses, not landfilled. A tax is collected on all new battery purchases to help defray the cost of recycling. In Italy, spent dry batteries are considered as hazardous waste and must be collected separately. In Sweden (lo), the environmental issues relating to waste batteries are addressed in the Control of Chemicals Bill and in the Decree on Environmentally Hazardous Batteries. All used batteries containing cadmium or mercury are collected separate1 under government control. The cadmium is then recycled. Regulations are in p ace for the manufacture of nickel/cadmium cells, limiting the exposure of workers and the emission of toxic materials.

r

In Japan, a number of laws and ordinances have been in effect since the late 1960's to regulate the disposal of batteries containing hazardous materials (11). Included

142

are: the Environmental Pollution Control Law, the Air Pollution Control Law, the Water Pollution Control Law, and the Waste Disposal and Public Cleansin8 Law. Strict controls are placed on cadmium and certain users of Ni/Cd batteries are responsible for their proper disposal.

In Canada, several articles of legislation relate to the handling and disposal of used batteries (12). These include: the Canadian Environmental Protection Act of 1980, the Transportation of Dangerous Goods Act, and the National Waste Reduction Plan. In addition, Canada is a signatory to the Basel convention. An Environmental Choice Program is also in effect in which environmentally friendly products are so labeled. Lead/acid batteries can have the Eco-Logo if they contain >50% recycled lead and have instructions for safe disposal. To date, this has been successfully opposed by industry groups. The rovince of British Columbia has im osed a $5 surcharge or "green tax" on the purc ase of Pb/acid batteries to be use for a "sustainable Environmental Fund." A monitoring and tracking system has been initiated and incentives from the fund have increased the return of used batteries, especially from remote areas.

B

f!l

New South Wales Australia has adopted the United States' system with some modifications. 4.

THE MANAGEMENTOF DISCARDED BATTERIES

The problems associated with discarded batteries from the highly distributed domestic sources are complicated. They involve questions of social structures, economics, technology, and re latory climates. The first problem is that of the efficient retrieval of discarded atteries. The next question that must be answered is what kinds of batteries may be allowed to enter municipal waste streams without treatment? Another question is how will the consumers know which is which? This is related to the question of the amounts of pollutants contributed by particular batteries relative to other sources of the same pollutants. It is not reasonable to consider batteries alone as a problem from the point of view of municipal solid waste (MSW) treatments when other sources contribute the same ollutants as well. Nickel/cadmium batteries are an example. Sources other than gatteries contribute significant amounts of cadmium to MSW. The elimination of nickel/cadmium batteries from MSW or the elimination of cadmium from Ni/Cd batteries will not solve problems attributed to cadmium in MSW.

i

The batteries of principal concern at this time are lead-acid batteries, nickelcadmium batteries, and mercury batteries. Even though they may contribute smaller amounts of toxic pollutants to MSW than other sources, concerted efforts are and should be made to prevent pollutants from these batteries from entering the foodchain and from becoming a health hazard. The first step in reventing battery derived pollutants from entering the food chain is the creation o an effective collection and processing system. Since none of the toxic materials (Pb, Cd, H ) can be detoxified, in contrast to organic solvents, they must be extracted from t e discarded batteries and recycled or disposed of in secure landfills (potential future hazards?) or rendered innocuous in some other way. The recycling of battery materials will be considered in Section 5.

P

fl

143

Fleet operators and large scale industrial users of batteries are in a better position than the general consumer communities to collect discarded batteries. Because of the organizational structure of the former group, the establishment of collection systems within their existing organizations is relatively easy. The more difficult problem lies in the domain of domestic users, particularly the problem of the retrieval of batteries other than lead-acid batteries. The retrieval of spent automotive lead-acid batteries from domestic users, by far the most significant fraction of lead acid batteries, is relatively efficient because these batteries are sold on a trade-in basis in most cases. However, most consumer batteries are not sold in this manner. Various schemes have been tried for the collection of regular consumer batteries, They include curbside collections, public drop boxes, and various monetary incentives, but the collection efficiencies have been disappointingly low, less than 20% in Germany and Japan (13) and less than 35% in Switzerland (14). It is even more difficult to induce domestic users to presort discarded batteries. It may be argued that the collection efficiency of spent consumer batteries via public or vendor channels can be increased by suitable incentives, by penalties for failure to comply with appropriate regulations, and by sustained educational efforts. Recent efforts along these lines in Sweden have shown that considerable progress can be made (15). It must be recognized, however, that an appreciable fraction of consumer batteries is likely to end up in the general MSW regardless of any such efforts. It is necessary to supplement public collections with other methods, the most important of which are: 1. Processing of municipal solid wastes to remove the toxic materials contributed by batteries and other categories of waste.

2. Reduction in the amounts of toxic materials in batteries at the source, i.e., at the manufacturing level. The possibility of separating batteries from MSW needs to be considered within the context of MSW processing as a whole. Some data from the United States may be of interest. In 1988 the average rate of accumulation of municipal solid waste amounted to 2.8 kg per person per day (16), and it was expected to increase rather than decrease during subsequent years. (The amounts of domestic waste enerated in some other countries are (17): Japan: 1.08 Kg/person,day; France: 8.99 Kg/person,day; Germany: 0.66 Kg/person,day.) The concentration of batteries in this amount of waste is vanishingly small. Various methods have been found to be somewhat successful in extracting batteries from MSW on a limited scale, for exam le magnetic separation (18), but the large scale feasibility and the economic viabi ity of this and other methods remain to be demonstrated. Wiaux and Nguyen have discussed the sorting problem in some detail (19).

P

Interestingly, in 1988 in the United States, 14% of the MSW was incinerated. The increasing rate of generation of MSW and its accumulation, combined with the decreasing availability of facilities or areas for storing such wastes in an acce table manner in most countries, suggest that incineration of MSW will be practicegto an increasing extent during the coming years. It represents a relatively effective means of reducing the volume of wastes to be stored in landfills. The exportation of solid and toxic wastes to less developed countries is an indefensible practice.

144

Insofar as mercury and cadmium are concerned, and lead to a lesser extent, no matter how the incinerators are operated, a significant fraction of these materials will be volatilized during incineration and enter the ecosystem via the airborne path, unless recovered from the flues by fl ash preci itation and va or condensation, methods of questionable merit for arge scale &W operations. h e remainder of the cadmium and lead will end up in the incinerator ash and in the incinerator residues, but all the mercury may be expected to be volatilized. This means that unless the reduction of the toxic materials at the source can be racticed, the incinerator residues and flues will need to be processed to remove ead and cadmium for recycling or for safe disposal in some other manner. The most effective and also the most economical way to prevent mercury from entering the environment from batteries is to phase out the use of mercury in batteries to the fullest extent possible, an effort already instituted by the battery manufacturers, and to maintain an effective collection system for the mercury batteries still in use.

Y

P

5.

BATTERY RECYCLING

There is general agreement that toxic materials from batteries must not be allowed to enter the environment and cause harm. The question is not whether or not to render spent batteries and battery wastes innocuous, but how best to proceed to achieve that objective in the least costly manner. Three approaches may be suggested: 1. The development of non-toxic batteries. 2. The recycling of toxic materials that cannot be rendered harmless by chemical processing.

3. The storage of toxic battery residues in hazardous waste disposal sites until an acceptable recycling process can be developed. Among these alternatives, the first is the most desirable from an environmental point of view, but it may entail a significant reduction in the performance capability of batteries. Some progress has been made already by battery companies in their development of environmentally safe batteries, notably by the reduction of the mercury content of batteries and the development of a technology that may make it possible to replace the cadmium in Ni/Cd batteries with metal hydrides. It is unrealistic to think, however, that all batteries with toxic materials will disappear entirely in the foreseeable future. The batteries most likely to remain in circulation the longest are the lead-acid batteries. Interestin ly, these are the batteries that are recycled to a far greater extent than any ot er battery. Next come the nickel/cadmium batteries. Although they may begin to be replaced to some extent with the more environmentally benign nickel/metal hydride batteries during the coming years, it is doubtful that they will disappear soon, primarily because of their excellent performance capabilities. Batteries that contain mercury are already being phased out wherever feasible, and they are not expected to pose an environmental threat in the future. Probably, it may not be necessa to establish dedicated facilities for the recycling of mercury batteries. The xird alternative of storing toxic battery wastes in secure landfills or in some other secure storage facilities appears to be acceptable to many at the present time. But this is

a

145

not a solution; it is a means of postponing a solution. Eventually it will be necessary to reprocess the materials stored in hazardous landfills. It is an acceptable alternative only as an interim method of handling toxic wastes. The second alternative is the preferred solution to the potential environmental problems that may be caused by batteries that cannot be rendered non-toxic, and it is the solution most generally accepted as long as batteries are made that contain toxic materials. The operational incentives for recycling of most consumer batteries are regulatory rather than economic. Wallis and Wolsky (20) cite estimates of US $0.80/lb for the recycling cost of batteries, exclusive of any credits for recovered materials, and US $O.lO/lb for their dis osal in hazardous waste landfills. Clearly the economics favor disposal, not recyc ing. The recycling of batteries will be driven, therefore, primarily by regulatory pressures.

f

The state of the art of recycling varies for different battery systems and battery components. The processing of metallic wastes, Pb, Cd and Hg, when available as such, is straightforward and is based on well established metallurgical technologies. To the fullest extent possible, recycling should preserve metallic wastes as such to reduce the energy and other costs of reprocessing, i.e., efforts should be made to segregate the metallic wastes from materials that need to be processed chemically. In most cases, this may not be possible. Batteries comprise an intimate combination of metallic and non-metallic materials, both organic and inor anic, and they may be mixed with other trash. Given this situation and the dif erent chemical compositions of the various battery systems, the nature of the chemical processing steps required to recycle a particular waste will depend on the battery type under consideration. The recycler cannot expect to have a uniform feedstock, and, for this reason, an process under consideration should be robust, i.e., able to handle highly variable eedstocks with a consistent processing efficiency.

B

r

Considerable emphasis has been and will continue to be placed on methods of obtaining battery wastes free of other trash. In the case of lead-acid batteries this is less of a problem since the majority of lead-acid batteries are traded via automotive outlets on a trade-in basis that facilitates recycling of batteries with no admixtures. The waste admixture problem is associated primarily with general consumer batteries. To the extent that the collection of batteries as presorted waste at their points of origin can be practiced, the preferred approach, can the admixture problem be alleviated. But, however desirable it may be to separate eneral household trash from batteries, it is unrealistic to think that batteries will e available with no such admixtures, Various methods have been pro osed for the separation of batteries from other wastes at processing sites that inc ude both manual and magnetic separations and methods based on bar coding and X-ray attern recognition. However, no satisfactory method has been developed et. {he waste separation problem is likely to receive continued attention. +he unsatisfactory state of the required separation methods notwithstanding, our discussion of recyclin methods will be based on the assumption that feedstocks are available with little, i any, waste admixture.

%

P

P

Regardless of the type of batteries under consideration, attention needs to be given to their safe storage and transportation from the time they are collected until they enter actual recycling o erations. Most batteries contain aggressive electrolytes that may cause persona[ equipment, or environmental damage if permitted to

146

escape from battery wastes. The containment and control of potential spills require that adequate holding facilities be provided during storage, handling and transport; casual storage under open air conditions is not a satisfactory method. The chemical process technology for battery recycling is still in a state of flu. Established smelter operations need to be improved to meet more stringent emission standards, and some of the recycling processes for different kinds of batteries have not emerged from their developmental stages yet. The construction of some full scale plants have recently reached, or will soon reach, completion, and as operating data become available it should be possible to make more realistic assessments of the various processes based on actual data rather than on projections based on pilot plant data. The information of particular interest relate to the quality and quantity of all plant effluents on a sustained basis, plant maintenance and o erating problems lant economics, ener efficiency, as well as information on t e feedstocks. ij)nfortunately, very Byittle of that kind of information is available for full scale plants today. In the following sections, we present simplified descriptions of representative processes either employed at the resent time or proposed for the recycling of spent batteries. In the case of the fatter, many of the processing steps have passed through the pilot stage with results sufficiently encouraging to warrant advancement to the design of full scale plants.

K

5.1 Lead-Acid Batteries

Only a minute fraction, about 0.1%, of the total lead consumed by the battery industry enters into the manufacture of small consumer type lead-acid batteries, and they are likely to be discarded as part of general household waste. The recycling of batteries in that category ma be handled by processes described in Section 5.3 below. Almost all the lea consumed by the battery industry is employed in the manufacture of large prismatic automotive and industrial type batteries.

B

Discarded lead-acid batteries may be recycled by processing in conventional lead smelter operations, although the resent trend is towards recycling battery wastes in dedicated facilities operated y the battery manufacturers themselves or by independent reprocessors.

1

Thermal Processinp One of the simplest Drocesses for recovering lead from lead-acid batteries is the thermal process’developed and operated by varta (2). The main processing steps are shown schematically in Figure 1. Spent batteries are drained and fed to a blast furnace operated under reducing conditions to convert lead compounds to metallic lead and to convert sulfate to iron sulfide. The bottoms removed from the furnace is a mixture of molten lead, iron sulfide, and slag, all of which can be separated mechanically. The crude lead with its content of antimony is sent to a lead smelter for refining. The iron sulfide may be disposed of as such or reprocessed to recover iron and sulfuric acid. The off-gasses from the blast furnace contain organic vapors and some lead chloride formed by reaction between the polyvinyl chloride degradation products and lead. The off-gases are subjected to a partial quench and filtration to recover the lead chloride. The lead chloride is converted to lead carbonate and returned to the blast furnace to recover the lead. Although some of the slag may be returned to the furnace, most of it must be removed from the process to prevent an excessive accumulation of inert materials in the process

147

Ha0 Coke Ume Iron

Furno ce

Ouench

+

.c

Filter

After Burner

Fuel

I

w J 'r' PbCI.

NazCOJ

I

Reactor

I

I

Filter

r

HzO Ash

PbCOj Filter

Melt Seporator

Monual Seporator

Woste Woter

z

c FeS

c

Figure 1. Varta Process. Simplified Diagram.

Slag

Pb(Sb)

148

streams. Depending on the a plicable regulations and the quality of the operations, further reprocessing o some of the waste streams may be necessary to reduce the discharge of lead and other materials to acceptable levels.

F

Hvdrometallurgical Processin5 A-more compkated process has been developed by Engitec (21). It exploits chemical processing techniques to recycle almost all the materials in spent leadof the extensive use of acid batteries, and it produces minimal internal recyclin of process streams. in Figure 2. It egins with electrolyte and screening, flotation and hydrodynamc is smelted and refined to produce a re-usable lead grid alloy. The lead compounds in the slurry from the screening operation are digested chemically to produce a soluble lead salt that is electrolyzed to give a high purity lead. The sodium sulfate solution generated in the slurry digestor is electrolyzed to produce battery grade sulfuric acid and a sodium hydroxide solution that is recycled to the slurry digestor.

%

Although the hydrometallurgical process is more complicated than the thermal process, its principal virtue resides in the essential recovery of all of the materials in spent lead-acid batteries, including the plastic materials, and in the minimal generation of waste streams. As actual data become available on the operational characteristics and the economics of full scale plants, it will be possible to make a meaningful assessment of this process and its merits relative to other processes for the recycling of lead-acid batteries. Other processing methods and variations on the above processes have been roposed that may merit consideration. Interested readers are referred to the roceedings of the International Seminar on Battery Waste Management and the relevant journals cited at the end of the chapter for more detailed information on these and other processes.

F

5.2 Nickel-Cadmium Batteries

The recycling of Ni/Cd batteries is of more recent origin than that of lead-acid batteries. It is motivated by concern for the harm that cadmium and its compounds may do to the environment and to man and to some extent by the economic incentive to recover nickel. Although some income may be generated by the sale of recovered nickel and cadmium, it should be realized that cadmium is available in some abundance as a byproduct of zinc refining, and such income can do little more than offset the cost of recycling. The recycling of Ni/Cd batteries can be carried out by various methods, and we select a thermal and a hydrometallurgical process for discussion. As in the case of lead-acid batteries, small Ni/Cd batteries are generally discarded as part of household waste and have to be processed as discussed in Section 5.3 below. The two processes we consider rely on spent batteries as feedstock with no admixture of other wastes. Thermal Processing The recycling process developed and operated by NIFE in Sweden (22) is a pyrometallurgical process that accepts both large and small Ni/Cd batteries. A simplified diagram of the process is shown in Figure 3. Lar e batteries are dismantled manually and the separated components are fed to dif erent processing

B

Batteries

-

Shredder

HISO, Pb

Solids HBF&O) H202 Solids NaOH(H,O)

I

Filter

HzO

I

NaSO. Solution

I

A'Evopomtor

NoOH

Electrolysis

Htso4

9

Organics

I t

wc

151

sections. The plastic parts are rinsed, dried, and disposed of as plastic waste or sold to plastics fabricators for reworking. The nickel electrodes and metal cases are also rinsed and collected as iron-nickel metal scrap and sold as such. The cadmium electrodes are fed to a furnace operated at about 900°C under reducing conditions to produce a cadmium condensate and off-eases that are vented to the atmosphere after filtration to collect sus ended solids. The furnace residues comprise metallic iron and nickel mixed wit slag.

K

Small Ni/Cd batteries are processed in a relatively simple manner in the NIFE rocess, the initial steps being different from those used in the case of the large gatteries. The small batteries are heated to about 400°C in a retorting furnace to rupture the cells and to pyrolyze their organic constituents. The solid residues are fed to the high tem erature furnace for processing with the cadmium feed from the lar e batteries. &e organic vapors are oxidized in an afterburner and filtered be ore being vented to the atmosphere as mostly carbon dioxide and water vapor.

B

The solid residues from the filter and from various other plant sources contain some cadmium as well as iron, nickel, and cobalt. Since they may not be discarded as waste, they are rocessed further by a sequence of selective dissolution steps, electrolysis, and ifferential precipitation to recover metallic cadmium and se arate solutions of cadmium sulfate, nickel sulfate, and cobalt sulfate that are so d to metal refiners for recycling. Depending on their compositions, the rinse and neutralizing liquors may be discarded as waste or reprocessed to recover their metal values and to reduce the plant emissions to acceptable levels.

8

P

The NIFE process has been in operation for several years, and it has been finetuned over the years to produce very low emissions and products that can be reused by the battery industry or sold to chemical a n d / o r metallurgical manufacturers. Hvdrometallureical Processing A process developed by the Department of Environmental Technology (TNO) in the Netherlands represents an interesting departure from the pyrometallurgical methods (23). It requires a feed with no admixtures of other wastes. A simplified diagram of the process is shown in Figure 4. The discarded batteries are first shredded and separated into a coarse and a fine fraction. Most of the plastic components and many of the metal parts are retained in the coarse fraction. The coarse fraction is subdivided into two fractions by magnetic means, and both subfractions are leached with hydrochloric acid to dissolve residual cadmium before being discarded as waste. The acidic cadmium extract is further enriched by usin it to wash the fines from the shredder. After filtration of the fines and the leac ing solution, the residual solids, mostly iron and nickel, are discarded. The filtrate contains dissolved cadmium, iron, and nickel. It is extracted with tributylphosphate (TBP)to remove the dissolved cadmium, and the cadmium salt is stripped from the extract by acid extraction. The TBP is recycled to the solvent extractor. The acidity of the cadmium chloride extract is adjusted to precipitate residual iron as ferric hydroxide that is collected by sedimentation and/or filtration, Metallic cadmium is recovered by eletrolysis and the stripped solution is discarded. The aqueous phase from the solvent extractor contains dissolved iron and nickel chlorides, and some cobalt chloride as well. It is oxidized at a controlled acidity t o precipitate ferric hydroxide that is removed by

fl

152

Batteries HCI Salty Waste

Plastics

Fe, Ni Waste

NaOH

TBP Solvent

NaOCl

Cd

Ni

Figure 4. TNO Process. Simplified Diagram.

153

sedimentation/filtration, and nickel with some cobalt is stri ped from the filtrate by electrolysis. The waste liquors are discarded as relatively armless waste.

R

The TNO process produces no exhaust gases, except for some oxy en from the electrolysers, but the various solid and liquid wastes may contain sma 1 amounts of residual cadmium. When full scale TNO plants become operational, it is likely that modifications may have to be introduced to bring all the wastes into compliance with the relevant emission regulations. It is expected that sufficient operational data will become available on a full scale TNO plant in the near future so that the relative technical and economic merits of this process can be assessed.

f

5.3 General Household Batteries The recycling of batteries from domestic sources presents two serious roblems. The most pressing one is that of the efficient collection and separation o batteries from other trash. An additional complication is that many batteries from both household and other sources may be permanently affixed in various devices and not easily retrievable for disposal. The cost of sorting wastes has been estimated to be about U.S. $0.50/kg (24), a significant fraction of the cost of a battery. The other serious problem is that the compositions of the feedstocks to battery recycling facilities are likely to change from time to time and from place to place. This may present a real process design problem, particularly for the less robust hydrometallurgical processes that need to be well controlled to operate consistently. The thermal processes tend to be more robust and forgiving of feedstock variations.

P

Several processes have been pro osed for recycling domestic batte wastes, and we select two for discussion. ey have been chosen because of t eir otential usefulness and since they illustrate TE a variety of techniques. They are both ased on feedstocks of unsorted batteries with no admixture of other wastes. It is likely, however, that both may be able to operate reasonably well even if some such wastes are present in the feedstock.

xg

Thermal Processincr The process devegped by Sumitomo Hea Industries, Ltd., in Japan is a good example of a thermal process well suited or the recycling of spent household batteries (1). Although it was designed to handle only zinc/carbon, alkaline man anese, and zinc/mercuric oxide batteries, it can probably be adapted to han le nickel/cadmium, lead-acid, silver/zinc, zinc/air, and lithium batteries as well. A simplified diagram of the process is shown in Figure 5. Some of the operations and internal processing loops have not been shown in this diagram or reasons of simplicity.

'r

8 seconda;y

Spent household batteries with no admixture of other trash are fed to a shaft furnace operated at temperatures above 500.C where the organic battery components are decomposed and volatilized together with metallic mercury and electrolyte vapors. Practically all the mercury in the waste is volatilized in the shaft furnace. The gaseous effluents and vapors are passed through an afterburner to oxidize the organics and then passed through a scrubber and condensor to remove mercury and waste water. Part of the waste water is recycled to the scrubber and the remainder is discharged as waste after removal of sludge in a thickener. After the mercury removal from the gaseous stream, the exhaust contains mostly water

154

155

vapor and carbon dioxide. This stream is filtered to remove suspended solids, passed through an adsorber to remove residual mercury, and vented to the atmosphere. The solid residue from the shaft furnace is fed to a smelter operated under reducing conditions at about 1400°C where manganese and zinc oxides are converted to metals. The zinc is volatilized and separated from the flue gases by condensation and the zinc-free flues are recycled to the shaft furnace after scrubbing and reheating. The scrubber effluent is separated into sludge and waste water before disposal. The smelter bottoms contain an alloy of iron and manganese and slag. The two are separated and collected for sale and disposal, respectively. Although the information available on this process is somewhat limited, it appears to operate well, but it may be necessary to introduce additional rocessing steps to improve the uality of some of the waste streams and to render t e process capable of handling a 1 types of batteries likely to be present in general household wastes. No doubt such modifications and improvements will be made once sufficient experience has been gained in the operation of full scale plants.

i

4

Hvdrometallureical Processinp An interesting-recycling process has been developed by the Recytec company in

Switzerland in cooperation with ETH in Zurich (25). It combines an initial thermal treatment wth a subsequent electrochemical process to recover separated metal values from spent household batteries. A virtue of the rocess is that it accepts unsorted mixtures of just about all types of batteries like y to be found in household wastes, and it can probably tolerate admixtures of some non-battery wastes as well. Although it is not designed to handle lead-acid batteries, it may well be capable of doing so after some process modifications.

f

A simplified diagram of the process is shown in Figure 6. Unsorted batteries are fed directly to a pyrolyzing furnace without prior shredding. The organic materials are decomposed in the furnace, the electrolyte solvents evaporated, and the mercuric compounds decomposed to yield mercury vapor. The volatilized materials are passed through a condenser to recover mercury and condensable liquids, mostly water and some organics. The condensate is fractionated by centrifugation into mercury, waste water, and an organic phase. Depending on the feedstock, the discharged mercury may also contain zinc and cadmium. The water collected from the centrifuge is passed through an aluminum cementor to extract residual mercury as an aluminum-mercury alloy, and some of the effluent from the cementor is recycled to the scrubber for the noncondensible gases from the furnace. The remainder of the effluent is sent to an evaporator to generate process steam/water and a disposable salt mixture in order to prevent an excessive salt buildup in the process. The liquid scrubber waste is recycled through the cementor. After scrubbin , the residual ases and organic vapors are oxidized in an afterburner and vente to the atmosp ere as water vapor and carbon dioxide after an adsorptive step to remove final traces of mercury.

d

a

The solids from the furnace are shredded and leached prior to magnetic separation of iron and nickel scrap from nonmagnetic solids. The leaching solution contains suspended oxides of manganese, carbon, insoluble zinc salts, and some dissolved salts. The suspended solids, mostly manganese oxides, are collected by

Batteries Fuel

:H

= Furnace

Solids

Condensate

I

Scrubber

Fuel

4 1 -

Condensate

After Burner

m

-

1

I

Condenser

Filter

#

t

NH9) Carbon

4

HZ0

Adsorber

Adsarber

4

Figure 6. Recytec Process. Simplified Diagram.

Fe, Ni Scrap

COz

157

decantation, dissolved in a reducing ste under acidic conditions, and electrolyzed to produce manganese dioxide anodical y and zinc and cadmium cathodicall . The bottom solids from the decanter are fed to the magnetic separator for urther processing. The nonmagnetic fraction from the magnetic separator is fractionated electrochemically by means of a unique dissolution-deposition batch process in which the relatively ure metals are collected sequentially in the order: zinc, co per, silver, nicker The residual solids, mostly carbon and graphite, are co lected by filtration.

P

r

f

The Recytec process is quite complicated and may need to be modified before its acceptance as a viable process for the plant scale re cling of spent batteries. When plant scale data become available it will be possi le to make the necessary assessments in terms of both plant economics and recycling efficiencies.

'41

6.

BATTERY IMPROVEMENTS

The battery manufacturing community is responding to the increasing environmental concerns and regulations with a number of changes in the manufacturing processes and ingredients in consumer batteries. Foremost is the reduction of mercury in alkaline cells to <.025%, with a continuin effort to completely eliminate mercury from this system. In the area of Ni/C batteries, much work is underway to develop a Ni/metal hydride system, to eliminate cadmium. However, it is unlikely that the Ni/Cd system will be replaced completely. In a number of applications, e.g., very hieh rate or very fast recharge, it is not clear that the Ni/metal hydride system will be able to duplicate the performance of Ni/Cd.

f

Studies have indicated that the alloys used in the Ni/metal hydride batteries pose less of an environmental risk versus the Ni/Cd or Pb/acid systems (26). However, some le islation, e+, the EC Directive on Hazardous Waste, specifically includes some o the materials used in these batteries (Ni and Co). In addition, several other heavy metals, e.g., rare earths, are used in some of the alloys. It remains to be seen how, or if, these metals will be regulated.

P

Much research is underway on rechargeable lithium batteries. These systems hold out the promise of an environmentally benign system, especially the newer "lithium ion" systems, depending on the cathode and electrolyte selected. If any of these efforts are successful, the rechargeable lithium ion battery, which utilizes a carbonaceous material as the anode, could eventually replace the Ni/Cd and Ni/metal hydride systems. Many batte manufacturers have programs in place to reduce or eliminate toxic discharges uring the manufacturin process. The elimination of halogenated solvents for degreasing and improve manufacturing processes, e.g., soldering and welding, to reduce the amount of scrap, are examples.

7

d

The lead/acid battery will remain the primary system for starting, lighting and ignition (SLI) a plications. This does not pose a serious threat to the environment, since >85% o the batteries are presently being recycled. However, the sealed Pb/acid cells used in many consumer applications will probably go the way of

P

158

Ni/Cd, and be replaced by Ni/metal hydride in the near future and possibly by lithium ion in the long term. 7.

ADMINISTRATIVE STRUCTURES To succeed in minimizing hazardous waste from batteries, a detailed plan must be develo ed and put into place, both at the manufacturing and disposal/recyclin ends o the battery life cycle. A considerable effort is already taking place in bot these areas.

a

P

7.1 Manufacturing The aim of this effort is source reduction and waste minimization in factory o erations. A combination of steps are required to achieve the desired results ($7). Th ese include: 1) Environmental audits to identify and rioritize waste areas and to develop solutions to these problems. Also, ealth and safety audits need to be performed.

E

2) Source reduction at vendors who have hazardous materials in products and materials purchased. Work closely with these vendors and inspect their plants. 3) Employee involvement through training and incentives for good housekeeping and improved quality. 4) Emergency response team creation for pollution control and prevention.

5 ) Reuse and recycling of materials and packa ing. Keep materials clean and well defined so they may be reintroduced into t e production process. Specify the configuration and construction of containers that components are received in. They can then be reused, which will reduce the used container disposal volume.

fl

6) Recycle paper, cardboard, plastic and metals offsite. 7) Involvement of all departments in the plant in the source reduction and waste minimization process. 7.2 Disposal/recycling A number of steps are necessary to achieve a successful battery disposal and/or

recycling pro ram. There are several ways in which a community can approach this problem b8). 1) Separation at centralized facilities after the batteries have been discarded into the waste stream.

2) Curbside collection. 3) Voluntary centralized collection, i.e., drop-off centers.

I59 4) One day collection events.

5) Original Equipment Manufacturers (OEM) reverse distribution. 6) Focused retail programs. Since batteries make up a small fraction of the total waste stream, their collection and separation from the household waste stream is expensive. Therefore, a number of steps need to be put into place for this type of program to succeed. These include: a) Development of battery collection and reclamation programs based on the ideas listed above. b) Distribution of information to the public via newspaper, TV,radio, billboards, etc. c) Expansion of representation in legislative and environmental forums. d) Providing opportunities for diverse groups to exchange information on various options available for the safe handling, disposal and/or recycling of batteries. There are many good reasons for recycling spent batteries, however certain basic conditions must be met for a successful program (29). A sufficient quantity of spent batteries must be available, containing an adequate high concentration of recyclable substances, to become economically feasible. A simple technique must be available with justifiable energy requirements and low emissions and residues. Finally, the recovered materials must have a market. It must also be proven that battery recycling has significant ecological advantage over disposal.

8.

SOURCES OF GENERAL INFORMATION RELATED TO THE BATTERY WASTE PROBLEM The International Seminars on Battery Waste Management held annually at Deerfield Beach in Florida provide an excellent source of information on battery recyclin and related matters. The papers presented at the seminars are published in the roceedings that may be obtained from Ansum Enterprises, Inc., 1900 Coconut Road, Boca Raton, Florida 33432, U.S.A.

1

Since no journals are devoted exclusively to the disposal and recycling of batteries, readers in search of technical information not available in these Proceedings may find it useful to consult the general environmental literature and journals devoted to metsllurgical and chemical engineering. The following is a limited selection of journals that contain information on environmental technologies and roblems in general, including information on ecosystem dynamics that may be o interest to battery processors and recyclers. Many other journals are also available, especially civil engineering journals, that contain technical articles of interest to waste management engineers.

P

"Environmental Science and Technology," The American Chemical Society, Washington, DC.

160

"Journal of Environmental Systems," Baywood Publishing Co., Amityville, NY. "Journal of Environmental Engineering," American Society of Civil Engineers, New York, NY. "Journal of the Society of Environmental Engineers," Society of Environmental Engineers, London. "Pollution Abstracts," Cambridge Scientific Abstracts, Bethesda, MD. 9.

REFERENCES

Note:

ISBWM is an abbreviation for International Seminar on Battery Waste Management.

1. M. Toshio, "Sumitomo Used Dry Battery Recycling Process. Process Concept and Pilot Plant Results." Proc. 2nd ISBWM, Florida, November 1990.

2. Eurobat Bulletin. "Realized and Projected Recycling Processes for Used Batteries," June 1991. 3. N. I. Sax and R. J. Lewis, Jr., "Dangerous Properties of Industrial Materials." Van Nostrand Reinhold, New York, 1989.

4. United States Code of Federal Regulations. "Primary Drinking Water Regulations." 40 CFR 141.11

5. M. D. Royer, A. Selvakumar and R. Gaire, "Control Technologies for Defunct Lead Battery Recycling Sites." Proc. 3rd ISBWM, Florida, November 1991. 6. B. M. Barnett and S. P. Wolsky, "The Battery Waste Problem and Alternatives to Small Sealed Rechargeable Lead Acid and NiCd Batteries." Proc. 2nd ISBWM, Florida, November 1990. 7. E. Bennett, "European Community Waste Management and Its Application to Used Batteries." Proc. 2nd ISBWM, Florida, November 1990.

8. R. Eloy, "Battery Waste Management Legislation i n t h e European Communities." Proc. 3rd ISBWM, Florida, November 1991. 9. C. H. Tisdale, Jr., "Legal Issues Associated With Battery Disposal." Proc. 1st ISBWM, Florida, November 1989. 10. T. Kertesz, "Battery Collection in Swedish Municipals." Proc. 3rd ISBWM, Florida, November 1991.

11. S. Oda, "The Disposal of Ni-Cd Batteries in Landfills and the Affect of Cadmium on the Human Systems." Proc. 1st ISBWM, Florida, November 1989.

161

12. J. Grach and B. G. Terry, "Battery Recycling in Canada." Proc. 3rd ISBWM, Florida, November 1991. 13. G . B. Baum, "Model Legislation for Nickel Cadmium and Small Lead Battery Recycling." Proc. 2nd ISBWM, Florida, November 1990. 14. J. Fiala-Goldiger, M. A. Rollor and J. Hanulik, "The Status of Battery Recycling in Switzerland." Proc. 2nd ISBWM, Florida, November 1990.

15. T. Kertesz, "Battery Collection in Swedish Municipalities." Proc. 3rd ISBWM, Florida, November 1991. 16. H. Pillsbu , "Battery Recycling and Disposal in the United States." Proc. 2nd ISBWM, Frorida, November 1990. 17. J.-P. Ribes, F. Mendel, J. Zeboulon and C. Denis in L'Express Nr. 2119, February 21,1992. 18. D. E. Freeman, "Batteries in Garbage. Treatment by Mineral Dressing Techniques." Proc. 2nd ISBWM, Florida, November 1990.

19. J.-P. Wiaux and T. Nguyen, "The Sorting-Out of Spent Batteries: From Pilot Scale to Industrial Application." Proc. 3rd ISBWM, Florida, November 1991. 20. G . Wallis and S. P. Wolsky, "Options for Household Battery Waste Management." Proc. 2nd ISBWM, Florida, November 1990. 21. R. M. Reynolds, E. K. Hudson and M. Olper, "The Engitec CX Lead-Acid Battery Recovery Technology." 1st ISBWM, Florida, November 1989. 22.T. Anulf, "Recycling of NiCd Batteries - An Economic Solution to the Cadmium Dilemma." Proc. 1st ISBWM, Florida, November 1989. 23. J. van Erkel, J. J. D. van der Steen, G. van der Veen and C. L. van Deelen, "Recovery of Metals from Spent Nickel-Cadmium Batteries." Proc. 2nd ISBWM, Florida, November 1990. 24. J. David, "Cadmium Nickel Battery Treatment - An Economic Point of View."

Proc. 1st ISBWM, Florida, November 1989.

25. J. Fiala-Goldiger, J. Hanulik, Thinh Nguyen and G.C. Kin , "The RecytecTM Process for Spent Dry Cell Batteries." Proc. 1st ISBWM, lorida, November 1989.

f?

26. C. R. Knoll, S . M. Tuominen, J. R. Peterson, L. M. Metz and T. R. McQueary, "Environmental Impact Status of Select Battery Alloys in 1991." Proc. 3rd ISBWM, Florida, 1991. 27. R. L. Balfour and T. J. Anderson, "Source Reduction and Waste Minimization at Rayovac Corporation." Proc. 3rd ISBWM, Florida, 1991.

162

28. N. England, "Portable Rechargeable Battery Association's Response to Environmental Concerns." Proc. 3rd ISBWM, Florida, November 1991. 29. J. Fricke, "Recycling of Batteries - The View of the European Battery Industry." Proc. 3rd ISBWM, Florida, November 1991.

APPENDIX

Principal Recyclers of Consumer Batteries (1991)

(This list is incomplete. Please contact Dr. S. C. Levy if you wish your company to be included in future listings.) Ni/Cd

Societe Nouvelle d'Affinage des Metaux (SNAM), l a Verpilliere, France

Ni/Cd

Societe Aveyronnaise de Valorisation des Metaux (SAVAM). Viviez. France

Ni/Cd

SAB NIFE, Oskarshamn, Sweden

Ni/Cd

T N O , D e p a r t m e n t of E n v i r o n m e n t a l Technology, Netherlands Organization for Applied Scientific Research

Ni/Cd

Toho Zinc Co., Ltd., Iwaki, Japan

Ni/Cd

International Metals Reclamation Co., Ellwood City, PA, USA Mercury Refining Compan Albany, NY, USA, Viviez, prance

Ni/Cd Zn/C Alkaline cells

Recymet, SA Aclens, Switzerland

Zn/C Alkaline cells

Sumitomo Heavy Industries, Ltd. Nihama, Japan

Lithium batteries

BDT, Inc., Clarence, NY, USA

Acknowledgements

We are indebted to Dr. S. P. Wolsky for ermission to quote freely from his Proceedings of the "International Seminars on attery Waste Management" and to Dr. Brian Barnett for providing information on consumer battery recyclers.

163

-

ALKALINE MANGANESE DIOXIDE ZINC BATTERIES PRIMARY AND RECHARGEABLE CELLS WITH AND WITHOUT MERCURY W. Taucher and K. Kordesch Institute of Inorganic Chemical Technology, Technical University of Graz, Stremayrgasse 16/III, A - 80 10 Graz, Austria 1. INTRODUCTION 1.1. Overview

Battery recycling has become a contraversial topic and it is not expected that there will soon be a consens among environmentally concerned lawmakers in different countries. There is no question that batteries which contain poisonous or dangerous materials should be prevented from coming into the hands of consumers who may simply discard them after use. Presently, primary batteries are mostly disposed together with the household refuse, up to now, various collecting schemes were not too successful. For that reason it may be necessary to produce mercury-free manganese dioxide-zinc primary batteries. Worldwide, more than 12 billions of such batteries are manufactured - and thrown away. It can be done, industry has produced mercury-free manganese dioxide batteries. The performance has suffered in some instances power output and shelf life characteristics are somewhat inferior, differing between brands, mainly depending on the fact that corrosion inhibitors, which guarantee a longer shelf life, reduce the load capability and battery companies adopted their specific compromises. Spent mercury-free manganese dioxide-zinc batteries do not contain any other definitely poisonous materials. The container is a steel can, the reaction end products like zinc carbonates and low manganese oxides are existing in nature in various forms, the plastic parts and components can be made from materials which can be burned or dumped without any environmental objections. In this case "recycling" would be a costly and not very productive enterprise. This argument becomes even more plausible if these "throw away" batteries can now be produced in a rechargeable form which allows a re-utilization of 50 to 500 times, depending on the mode of operations. The manufacturing costs of the rechargeable version are not much higher than those of the primary cell. The batteries are completely exchangeable, the shelf life is very high, even at extreme summer temperatures (unchanged from the primary cells) the capacity of the first cycle is essentially the Same and, most important, the cells can be made on the same mass production machinery with the only changes in the formulations of cathodes and anodes and the separator to make the cells safely and reliably rechargeable. The best of all: the production costs are nearly the same. In order to make all these aspects clear, this chapter is extensively describing the regular primary alkaline manganese dioxide-zinc system and also the new rechargeable system. In respect to the question "Are mercury-free cells possible?" the comparisons are made in very specific applications. The reader will certainly come to the conclusion that efforts, which must be done to eliminate the mercury without reducing the performance of the batteries are certainly worthwhile. In contrast to the manganese dioxide-zinc system, all other rechargeable batteries contain poisonous components and definitely should be collected and processed or recycled.

164

1.2. History

The history of the alkaline manganese dioxide cell dates back to 1882 [I]. The first commercial alkaline Mn02-Zn cell was produced in the mid 1950s and used a manganese dioxide-graphite cathode in form of a pellet, KOH or NaOH as electrolyte soaked into an absorbent separator and an increased surface area zinc anode [2]. It was called the "crown"cel1, because of its bottle-cap-shaped closures. This cell was mainly used for low-current applications and was commercially produced as a 9 V battery for radios. During the period of 1960-1970the high-drain alkaline manganese dioxide cell was developed. At that time a large number of innovations and improvementswas established [3]. An essential improvement could be achieved through the construction change from the "bobbin" type (inside cathode) as used in Leclanche cells (carbon zinc), to the sleeve type cathode (outside cathode) [4]. The latter construction assures a relatively thin Mn02-electrode with improved diffusion properties, and the zinc powder anode provides the large area needed to overcome the passivation tendency of the zinc. In the early 1950s the rechargeability of the Mn02-Zn system was recognized and efforts commenced to make the "throw-away" primary alkaline cell rechargeable [5,6]. The development work revealed that primary alkaline cells, with some design modifications, could be rendered rechargeable for applications requiring a relatively inexpensive cell delivering a moderate number of cycles. Recognizing the economical and environmental benefits of the rechargeable system the first commercial cells appeared in the 1960s [3,7]. The sizes available for secondary cells were limited to "D" cell and larger sizes and were not sold as individual cells. Cells were typically assembled into a metal battery case which provided physical containment and acted as a barrier against leakage and gas build up when the power pack reached the end of its useful life. The first commercial cells were plagued with high internal resistance, low nominal capacity (nominal "D" cell capacity 2Ah) and required electronic means to determine the end of discharge [8]. The inability of those cells to handle any kind of consumer abuse resulted in the withdrawal from the market in the early 1970s. Research work was intensified since the late 1970s at the Technical University of Graz (TU Graz),Austria. New cell designs appeared in the early 1980s [9,10]. Further development of the rechargeable Mn02-Zn system was done in a specific project funded by the "Austrian Funds for the Support of Scientific Research'' within the framework of a larger program including also alkaline fuel cells. In 1987 the technology was transferred to the Battery Technologies Inc. (BTI) laboratories in Mississauga. In 1991 Environmental Battery Systems Inc. (EBS) was founded. A pilot production line of "AA" cells was established in 1988 at BTI progressing towards commercial application. In the meantime several important changes have been introduced making the Mn02-Zn system competitivewith other rechargeablebatteries. Limiting the amount of metallic zinc in the anode formulation prevents deep discharge of the manganese dioxide cathode which was one of the most important problems with earlier constructions. A physical restriction of the cathode and the use of modified manganese dioxide containing lattice stabilizers contributed to the improvement of cycle life [ 11,12,13]. The accumulation of corrosion hydrogen in the cell can be prevented through catalysts in the cathode mix oxidizing the hydrogen to water [ 141. Other catalysts suppress the formation of 6-valent manganates during overcharge by evolving oxygen stoichiometricallyequivalent to the applied current [ 151. These overcharge capabilities via an oxygen cycle are important for the application of small cells in serial arrays. The mercury content of the cells has been lowered from 0.6 wt % at the beginning to 0.1 wt % in series connected cells and to 0.025 % in single cells and has become zero in 1992.

165

The use of rechargeable batteries instead of primary cells diminishes the amount of waste and the absence of mercury makes them less hazardous from the environmental point of view. Separators are now available which allow far beyond 200 cycles under partial discharge conditions. The earlier problem of zinc dendrite formation is minimized. Continuous testing verified that the self-discharge at elevated temperatures is considerably lower than that of nickel-cadmium and nickel-metal hydride systems [ 161. The manufacturing lines for the production of rechargeable cylindrical cells are nearly identical with those of the low-cost primary cell version. This is an important cost-determining factor [ 171. New chargers, some with resistance-free voltage sensing, have been developed [ 181.

--

2. PRIMARY ALKALINE MANGANESE DIOXIDE - ZINC BATTERIES (PAM CELLS) 2.1. General Characteristics of Alkaline MnO2-Zn and LeclanchC Cells

The alkaline MnO2-Zn primary cell represents a major advantage in portable power sources over the conventional Leclanche cell, which is often called carbon-zinc cell because of the carbon-rod collector in the center of the Mn02-"bobbin". It has a higher capacity compared with Leclanche cells, especially at high discharge and continuous discharge rates [ 191. Figure 1 illustrates the discharge characteristics of the two different Mn02 systems. The degree of the slope depends on the discharge mechanism and on the current-carrying ability of the battery. Alkaline Mn02-Zn cells can sustain heavy loads far better than Leclanche cells. In Figure 2 the

-

>

Y

W

1.5

12.5

1.3

," 10.0 a

L3

a I-I 0

L_r

1.1

-

>

-I

7.5

U

2 2

0.9

A W

" 0.7

5.0 2.5

0 0

500 TIME

1000

1500

STARTING DRAIN

CmAl

C MINI

Figure 1. Comparison of discharge characteristics of alkaline MnO2-Zn and Leclanche D-size cells for 500mA starting, discharged continuously at 2 1"C.

Figure 2. Capacities of alkaline MnO2-Zn and Leclanche cells at different load conditions to 0.8 V.

166

capacities at different load conditions are shown. It becomes obvious that alkaline Mn02-Zn cells excel at higher load levels. In Table 1 the comparison data of both systems are shown [20,21,22]. Although the theoretical energy content is nearly the same, the actual energy content differs strongly. At an roughly estimate the actual energy density is one fourth of the theoretical value for low current densities. Further advantages of the alkaline Mn02-Zn system are the low-temperature performance and the far lower resistance due to the use of KOH as electrolyte instead of the NH4ClxZnC12 electrolyte (half conductivity) used in Leclanche cells. Another feature of the alkaline cell is its leakproohess, derived fiom the hermetically sealed construction. In general the alkaline Mn02-Zn primary cell as well as the Leclanche cell is available in different cell sizes. In Table 2 the characteristics of the most important commercial sizes of alkaline Mn02-Zn cells produced before 1989 are summarized. These cells contain 0.3 Yo Hg. Table 1 Comparison data of alkaline MnO2-Zn and Leclanche cells ~

~

system

Alkaline Mn02-Zn

LeclanchC cell

Thermodynamic E" (V) Terminal OCV Average CCV Theoretical energy content (Wh/kg) Actual energy content of cells* ( w g ) (Wdm3)

1.44 1.55 1.25 290

1.78 1.58 1.20 280

77 215

66 120

~

~~

~

-

~

~

~

~

~

~

* Current drain, cl10 rate. Data reflect variations of cell constructions. Table 2 Characteristics of 1.5 V alkaline Mn02 cells* IEC+

Cell size

Capacity (Ah)

Diameter

Height

(-1

Weight (8)

Volume (cm3)

11.5 22 65 130

3.6 7.5 25 50

Cylindrical cells LR-03 LR-6 LR- 14

AAA AA C

LR-20

D

0.75-0.8 1.5-1.7 4.5-5.0 8-10

10 14 26 33

44 50 50 60

* The values presented are average figures from different manufactures. IEC: International Electrochemical Commission. The L denotes alkaline MnOq-Zn. In the Japanese code the same sizes are: AM3, AM4,AM2, AMl. The IEC code is also the German DIN code.

+

167 2.2. Chemistry of the Alkaline Manganese Dioxide Zinc Cell

In alkaline Mn02-Zn systems the same electrochemically active materials, manganese dioxide and zinc, are used as in Leclanche cells. The only difference to the latter system is the cathode sleeve construction and the highly conductive KOH electrolyte (7 M: 0.55 (Qcm)-1) as already mentioned. The anode is formed of zinc powder instead of a zinc can. Electrolytic manganese dioxide is used for the cathode material. Due to its low polarization and high voltage characteristic it performs far better than chemically produced manganese dioxide or natural ore. The simplified reaction mechanisms, based on a one-valence-step reduction for the manganese dioxide are: Anode:

Zn + 2 OH-

Zn(OH)2 + 2e-

->

Zn(OH)2 + 2 OH-

->

Cathode:

2Mn02 +H20+2e-

Overall reaction:

Zn + 2 Mn02

-> ->

[Zn(OH)4I2Mn2O3+2OHZnO + Mn2O3.

The Gibbs free-energy is calculated between -276 and -284 KJ/mol, the cell potential is 1.4 to 1.5 V, depending on the type of manganese dioxide. Mn02 is reduced in a homogeneous phase reaction and its potential (OCV) decreases gradually as the discharge proceeds. The electrode potential is determined by the homogeneous distribution of Mn(II1) and Mn(IV), whereby the ratio of the two species determines the state of reduction and also the shape of the discharge curve. However, the internal resistance of an Mn02 cathode steadily increases and a maximum is reached if the value of x in MnO, is 1.5 to 1.3. This is mainly due to lattice expansion. The expansion of the Mn02 lattice is small between Mn02 and Mn01,6,.This behaviour is shown in Figure 3 [23]. Zinc is thermodynamically unstable in contact with alkaline solutions. The electrode potential is above the potential of hydrogen. Thus zinc reacts with water and hydroxyl-ions to evolve hydrogen and forms a complex ion [24] : Zn + 2 H 2 0 +2 OH-

->

[Zn(OH)q12- + H2

Therefore the amalgamation of zinc particles or the use of alternative corrosion inhibitors is necessary. But also harmful effects of dissolved metal impurities can be reduced. In commercial batteries the mercury content has been gradually reduced from 0.5 % of total cell weight in 1985 to 0.025 YOin 1990 [25]. The zinc passivation, which will be increased with high current densities and low temperatures, is very important for practical applications. The zinc-covering layers may be dense or porous and more or less soluble depending on accompanying ZnO precipitations from over-saturated zincate solutions. Large-surface powdered zinc does not easily passivate, while smooth surfaces show this tendency. As a result, powdered-zinc anodes operate satisfactory with small amounts of electrolyte, while zinc can anodes of Leclanche cells do not. The advantage of KOH solutions as electrolyte for alkaline Mn02-Zn cells is due to its excellent conductivity and the solubility of Zn(OH)2. A maximum conductivity (0.5 S/cm)

168 occurs with a 30 wt % KOH with a freezing point of -66°C(eutectic). In alkaline Mn02-Zn cells the KOH concentration may range from 30 wt % (7 M) to 50 wt % (14 M). The immobilization of the electrolyte is achieved by the addition of 4 to 7 % of a gelling agent (e.g. , sodium carboxymethyl cellulose) or by soaking-up of a porous separator system or by the capillary action of the powdered-zinc anode (compressed type). Zinc oxide addition has an effect of corrosion inhibition, paste-forming and overcharge protection. During the operation of the cell the conductivity of the KOH solution changes by dissolution of Zn(OH)2. The bulk conductivity of a powdered-zinc gel anode is determined by the metallic structure. In Figure 4 the relationship between weight percent of zinc and the resistivity of an electrolyte/metal-powder paste is shown [26]. Good electronic conductivity is achieved when the mixture contains 35-70 % zinc. The usable cell capacity is exhausted as soon as the amount of metallic zinc in the anode gel decreases to about 30 % of weight [27,28]. The admixture of materials with good surface conductivity improves the anode quality with respect to efficiency and capacity [29,30].

>

0 1.9

1.7

1.5

1.3

1.1

X in Mn02

Figure 3. Discharge of a Mn02-electrode at 5 mA/cm2 in 7 M KOH, showing potential and resistence changes.

10

30

50

70

90

ZINC I N PASTE C u t % l

Figure 4. Resistivity of electrolyte/metalpowder paste. <- Point at which metal phase conductivityexceeds that of electrolyte. Zinc contains 3.0 % Hg.

2.3. Cell Components and Materials 2.3.1. Manganese Dioxide Cathode Formulation Different types of Mn02 show very different electrochemical behavior in batteries. For alkaline MnO2-Zn cells EMD is exclusively used. The purity level of battery-grade EMD, as shown in Table 3, is equivalent to that of reagent grade material [3]. The level of impurities is very critical in view of the expected battery shelf life. The composition of a typical cathode of alkaline Mn02-Zn cells falls in the following range: 80-85 % Mn02 (EMD), 8-15 % graphite flake and other carbon material, 10-15 % KOH solution (7-9 M), 0.3-4% binder andor other additives. The blended mix is molded into a steel can under high pressure to form a sleeve. The increase of capacity of alkaline Mn02-Zn cells

169

can be partly attributed to the greater amount of Mn02 (30-50 % more) packed in the same size cell. But the main reason is the use of graphite instead of bulky acetylene black. Table 3 Typical range of chemical analysis of battery - grade EMD samples Substance

Percent

Substance

Percent

MnO2 Total Mn H20 (-I* Fe Pb cu Ni co Cr

92.0 2 0.3

Sb As

0.0001 0.0003 0.0002 0.24 0.001 0.02

60.0+0.12 1.52 & 0.20 0.008 0.0007

0.0003 0.0005 0.0008 0.0008

MO

Na

NH? SiO? Insoluble to HCI ~042-

0.62

0.79

* H 2 0 (-) is adsorbed moisture that can be removed by heating at 110" - 120°C. Balance is structual water, which is released by heating above 120°C. 2.3.2. Zinc Powder Anode Formulation

Zinc powder used in alkaline Mn02-Zn cells must have a high quality (99.85-99,90%). It is produced commercially by electroplating or by distilling. Sometimes 0.04 % to 0.06 % Pb is added for corrosion inhibition. Powdered zinc is produced by discharging a thin stream of molten zinc into an air jet and is then atomized. A size distribution of about 8 to 800 pm is required for a typical battery grade zinc powder. The surface area will be 0.2 to 0.4 m2/g. In commercial primary alkaline MnO2-Zn cells two types of powder-zinc anodes are normally used: gelled anodes and porous zinc anodes. Gelled anodes contain powdered zinc mixed with an gelled electrolyte. A few percent of sodium carboxymethyl cellulose are often used as gelling agent. The anode mass has a high viscosity and can be extruded into the cell. Porous zinc anodes are produced by cold-pressing of zinc powder wetted with mercury, that welds the particles together [31]. In another process the porosity is controlled by filler materials, e.g. W C l , which can be easily removed later. Plastic binders are also used to fabricate porous zinc electrodes. 2.3.3. Separators

Alkaline Mn02-Zn primary cells in general use macroporous separators made from woven, bonded or felted materials. The choice of the separator material is not so sensitive as for rechargeable alkaline MnO2-Zn cells, where microporous membrane-type materials have to be used additionally to prevent zinc dendrites from causing cell shorts. Pore diameters range from 2.5 to 10 nm (membranes) to several hundred nanometer for porous-sheet separators [8,32].

170

2.4. Mercury Reduction of PAM Cells

Several battery manufacturers have tried to reduce the mercury content of cylindrical alkaline Mn02-Zn primary cells. Up to 1985 the batteries contained about 0.5 % Hg based on the total cell weight. Since 1988 the mercury content has been reduced to 0.025 YOHg and the final target will be a complete elimination of mercury from these cells [25]. The major problems dealing with the use of nonamalgamated zinc are an increased hydrogen gassing, a reduced load capability and an increased sensitivity to rapping. The high hydrogen overvoltage of mercury is one of the reasons for the amalgamation of zinc electrodes. In Figure 5 the effect of the amalgamation level on the hydrogen evolution is demonstrated [33]. Figure 6 shows, that the gas volume in a LR14 cell after storage for 1 month at 7OoC strongly depends on the mercury content of the cell. The hydrogen pressure build-up within the battery should not exceed 30 YOof the opening pressue of the safety vent (e.g. about 20 bar for LR14 cells). A gassing rate of 1.5 pVgh at 6OoC may be acceptable as a maximum value, taking a pressure equilibrium under steady state conditions into account. With nonamalgamated zinc powders of high quality even gassing rates below this value can be obtained. But nevertheless, the hydrogen evolution at the zinc electrode can never be totally stopped. To keep hydrogen gassing rates low, a high purity of manganese dioxide and zinc will be absolutely necessary. Additional factors of influence are battery components like current collector, separator, electrolyte, gelling agent, as well as impurities resulting from the manufacturing process

120 r

t2

2

n

loo/-

-

0

0

~ -

5

40-

-I

0

> vl

-z

U U

ME R C U R Y

Lpg / c m31

Figure 5. Hydrogen evolution current as a hnction of amalgamation level.

'\ 1

w

-

0

8

0

500

, 1000 1500 2000 2500 3000

MERCURY L p p m l

Figure 6 . Hydrogen evolution in a LR14 cell as a function of mercury content after 1 month storage at 70°C.

Many efforts has been made to replace toxic mercury with environment-friendly substances. This replacement can be approached either with zinc alloys or with additives dissolved in the electrolyte. Typically used zinc alloys are In, Ga, Al, Pb, Bi and combinations thereof [34]. Very promissing additives will be organic inhibitors, e.g. polyglycols and their derivates. The influence of mercury on the load capability of a zinc powder electrode (0.2 mm) is illustrated in Figure 7. It is obvious, that amalgamated zinc exhibits an initial peak current of

171

about 24 mA/mg Zn while the peak current of mercury-free zinc is more than 20 times smaller. Mercury also influences the contact resistance between adjacent zinc particles in the zinc gel electrode and its current collector when cells under load are exposed to a mechanical shock, e.g. dropping on a steel plate from a definite height. If amalgamated zinc is used, the disturbance of the voltage (on load) is much smaller compared to cells with unamalgamated zinc, which exhibit an additional voltage drop. The different behaviour is due to the wetting properties of mercury: the zinc particles stick together, have a good contact to the current collector and reestablish the contact quickly. Generally the rapping behaviour of a mercury-free zinc electrode can be improved by increasing the contact area. Mercury-free cylindrical alkaline Mn02-Zn primary cells are produced in a pilot production [25]. In Figure 8 a discharge comparison between cells containing 250 ppm Hg and 0 ppm Hg is shown. In all important modes of discharge the curves are very similar and the cells exhibit good gassing properties and load capabilities.

0,16% Hg

'"1 '\

- 0 70

IC

m

>

u

Hg

W

13 Q I-

_1

0

>

u ---Jw o

I

I

l

m

2

4

6

a

o

TIME

8

0

Isecl

Figure 7. Effect of mercury on load capability of 0.2 mm Zn powder electrode. Potentiostatic pulse, charge of potential versus H m g O from 1.380Vto 1.363V(lOMKOWO.l MZnO).

20

TIME

40

60

60

CHRS 1

Figure 8. Comparison of 250 ppm Hg versus 0 ppm Hg. Transistor radio discharge of a LR6 cell: 39 R, 4Wd, 7dw.

2.5. PAM Cells Characteristics containing 0.3 % Hg 2.5.1. Voltage and Discharge Performance

The open-circuit voltage of the alkaline Mn02-Zn cell is about 1.52 V and the nominal operating voltage is 1.25 V. The actual voltage depends on the discharge load and the state of charge of the cell. Usually the cell performance will be cut-off at 0.9 V, but also lower end voltages, particularly at high current drains are realistic. In Figure 9 and 10 the discharge curves of the AA size alkaline Mn02-Zn cells on a continuous discharge at various loads at 2OoC are shown. The current drains are related to the nominal operating voltage of 1.25 V. The sloping shape of the discharge curve is typical for

172

alkaline cell performance, as the reduction of Mn02 is characterized by a homogenous phase reaction. For Leclanche cells the slope of the discharge curve will be more pronounced.

0

3

6

9

15

12

0

30

SERVICE C HRS 1

60

90

120

15

SERVICE L H R S 1

Figure 9 and 10. Typical discharge curves of Mn02-Zn cells (AA size) at 2OOC. 2.5.2. Effect of Temperature

The alkaline MnO2-Zn cell performs well even at low temperatures and far exceeds the Leclanche cell. Figure 11 illustrates the room temperature service of a C size cell and its performance at lower temperatureson a 2.25 R continuous discharge test. The cell can operate at temperatures as low as - 4OOC. Figure 12 shows the capacity of typical C and D size alkaline 100 ri U

% 80 hl U

uO

60

g

40

& U

a a 20

a.

U

0

100

200

300

400

500 600

TIME [ M I N I

Figure 11. Discharge performance (2.25 R) of a C size cell at various temperatures. These cells contain 0.3 % Hg. Without Hg the performance drops considerably(30 YO below OOC).

-40

-30

-20

-10

0

10

20

T E M P E R A T U R E C"C1

Figure 12. Capacity of typical Mn02-Zn cells (C and D size cells) at various temperaturesunder high and low discharge load. Without Hg the high-load curves (300 and 500 mA) will drop about 50 % at -10°C.

173

Mn02-Zn cells under high and low discharge conditions. The advantage of the low disharge load ((3200) is evident. At an operating temperature of -20°C the capacity of a D size cell is still 52 % and for the C size cell 43 % of the room temperature capacity related to the corresponding decreased values of 13 % and 20 % at the high load (C20). 2.5.3. Internal Resistance Table 4 shows the internal resistance of fresh, undischarged alkaline MnO2-Zn cells. Datas are given for AA, C and D size cells at various temperatures [35,36]. Without mercury the internal resistance at - 1O°C is expected to double.

Table 4 Internal resistance of AA, C and D size cells versus temperature ~

~

20 0 -10 -20 -30 -40

~

~~

Temperature ("C)

~

~~

Internal resistance (Q) of cells containing 0.3 % Hg

AA

C

D

0.21 0.25 0.32 0.49 0.72 1.06

0.15 0.21

0.11 0.17 0.22 0.29 0.37 0.50

0.28

0.43 0.64 0.96

2.5.4. Service and Shelf Life with and without Mercury

In Figure 13 the service life of alkaline Mn02-Zn cells at various discharge loads and temperatures is shown. The curves are normalized for unit weight (amperes per kilogram) and the discharge performance is cut-off at 0.9 V. These curves are based on an energy density of 75 Ah/kg. The service life of a given cell under particular discharge conditions can be approximated and the size or weight of a cell can be estimated to meet a particular service requirement. The shelf life of alkaline MnO2-Zn cells is excellent. After one year storage at room temperature the batteries can be expected to supply 90-95 % of their rated capacity. Figure 14 demonstrates the capacity loss as a finction of storage temperature. Points A and A' represent 20 and 10 % loss after 3 years at 20°C, points B and B' represent 20 and 10 % loss after 3 months at 45OC and point C represents a 20 % capacity loss in 20 days. As a rule of thumb, 1 week of storage at 70°C is equivalent to 3 months storage at 45°C.

174 I-?

t.

a

0

s Y

t.

t U Q

a

a U U

0 v) v)

0 1

5

10

20

50

SERVICE

100 200

90

500

70

C HRS I

50

TEMPERATUR

Figure 13. Service life of the MnO2-Zn cell at various temperatures and discharge rates to a cut-off voltage to 0.9 Volt. Cells without Hg will have a considerably reduced service life.

30

20

C"C1

Figure 14. Capacity retention after storage at various temperatures. It will not suffer due to the absence of Hg if proper inhibitors are used.

-

3. RECHARGEABLE ALKALINE MANGANESE DIOXIDE ZINC BATTERIES

(RAMCELLS) 3.1. General Characteristics of RAM Cells with 0.025 YOand Zero YOMercury

The major design changes made to primary alkaline cells were the use of improved cathode and anode formulations, the limitation of the anode capacity to approximately 1/3 of the cathode capacity to prevent overdischarge of the cathode, the application of improved separators and the integration of means to enable moderate cell abuse. Cell components (cans and closures) and raw materials (EMD, graphite, zinc) used are identical to the ones used in primary alkaline cells. The electrode capacity balance accounts for the reduced capacity of RAM cells when compared to primary alkaline cells of similar size. By 1986 C and D size cells were developed to the point that 50 deep discharge cycles (100 % DOD) to 50 % of the original capacity and 250 shallow discharge cycles (10 % DOD) were reached. Figure 15 shows the construction of the C size and D size cell respectively [10,37]. The cumulative capacity of the C cells at that time ranged, depending on the depth of discharge @OD), fion 75 to 150 Ah. Since 1987 the development emphasis was shifted to the AA size cell which represents over 50 % of the consumer market. Figure 16 shows that the 1992 cell design has become very similar to the one of primary alkaline cells. Most noticeable is a simplification of the cell construction. Both mechanical cathode confinement and the complicated current collector design were abandoned [38].

175 I

I

Negative Terminal Plastic Seal Top

1

Brass Contact Nail Cu-Anode Cage Anode

2 3

Brass Contact Nail

4 5

Separator S t e e l Cage

4 MnO

Negative Cap Plastic Seal Top

Anode

6

Rings form t& sleeve -

Separator

17

Positive Terminal

Cathode

I

Can 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0

I

Figure 15. 1984 "C" and "D" cell design

Figure 16. 1992 "AA"cell design.

3.2. Manganese Dioxide Cathode

Investigations of the International Mn02 Common Samples revealed that electrolytic manganese dioxide materials (EMD) exhibit an appreciable cycle life if the discharge is limited to a fraction of the first electron capacity [9].With the depth of discharge limited to one third of the Mn02 first electron capacity premature failure of electrodes could be attributed to the mechanical disintegration caused by expansion (discharge) and contraction (charge) of the EMD electrode. Other failure mechanisms involve irreversible capacity loss due to overdischarge beyond the Mn02-MnOOH homogeneous phase discharge and due to haeterolite formation (ZnOxMn203). Confinements of the Mn02 electrode by mechanical means [10,39] used to achieve the required cycle life were ultimately replaced by an unconstrained RAM electrode design [38]. More recently various additives have been identified which have a beneficial influence on the cycle life and are the subject of various patent applications [40]. The cathode formulations hrther employ metallic catalysts to promote the recombination of hydrogen gas generated within the zinc anode at very low (0.025 %) or zero mercury content [41,42,43]. The hydrogen scavenger can form a discrete auxiliary electrode in electronic contact with the cathode or can be introduced simply as a cathode additive in a ratio of 30170 in the cathode preparation process. A very reliable catalytic material even at elevated temperatures (65°C) is Ag20. The auxiliary electrode may be an annulus or a disk of similar

176

diameter to the Mn02-cathode and located at one end of the anode. Gas diffusion electrodes can be employed if higher recombinationcurrent densities are desired [44]. The hydrogen pressures may range from 5 to 15 psig up to the relief pressure of the cell. In Figure 17 the pressure build-up of hydrogen with time is shown. The pressure builds up faster in the conventionell cell (curve A) than in the cell employing As20 (curve B) as catalytically active cathode ring. These results were obtained fiom tests made with discharged C size half cells submersing a Ni-wire into the electrolyte. During the galvanostaticallycharging process with 50 mA hydrogen gas evolves at a rate of 20 ml per hour on the Ni-surface. Even better results of hydrogen recombination are achievable with gas difision electrodes, containing a mixture of Pd/Rh on a PTFE-bonded carbon foil (Figure 18). The cell with the catalytically active ring recombines hydrogen to an extent, that a cell pressure of approximately 6 psig can be maintained for over four hours (curve D).

n

r

30

si

a CL 4 3

20:

v)

LL

10:

0

7 /

00-

0'

'0 0/ O R

w

0

0:

3 lA Ln ..

o/o'

-fig.,

W

LL

a I

I

I

0 TIME C M I N j

Figure 17. Pressure build-up of hydrogen with time. Curve A: Conventionell cathode, Curve B: Cell with Ag20 - ring.

50

100

150

200

250

T I M E [MINI

Figure 18. Pressure build-up of hydrogen with time. Curve C: Conventionell cathode, Curve D: Cell with gas diffusion electrode.

3.3. Zinc Anode - Reduction of Mercury

RAM cells are zinc limited. M e r complete discharge of the cell the zinc anode capacity is exhausted while a substantial amount of the cathode capacity remains. Zinc anodes of RAM cells do not differ in principle fiom primary cell anodes. A gelled anode is prepared by immersing the zinc powder in an immobilized KOH electrolyte. Levels of amalgamation are generally kept low to minimize shape change of the secondary zinc electrode. To reduce the gassing on the shelf, of intermittent discharge and in cycled cells to a minimum a mercury content of 0.6 % per total cell weight was maintained in the early 1980s. The overall mercury content has been reduced to 0.025 % by employing a combination of inorganic and organic corrosion inhibitors. The mercury content of primary alkaline cells of 0.025 % based on cell weight is currently the industry standard in North America and Europe. Mercury-free alkaline

177

primary cells have been introduced recently by several manufacturers. Extensive progress has been made in the field of mercury free-RAM cells. 3.4. Separator

In general secondary cells have more stringent requirements for separator materials than primary cells. RAM cells typically apply two components: a non woven absorbent and a barrier material. The absorbent material used is formulated from polyvinyl alcohol and rayon fibers, acts as mechanical spacer between anode and cathode and provides as an electrolyte reservoir. The barrier is unglycerinated cellulose and prevents zinc dendrites from causing cell shorts. Prior to insertion into the cell the two materials are wound into a tube. Separator materials are selected which are not subject to oxidation in the alkaline electrolyte even at elevated temperatures and which combine chemical and mechanical stability with long life expectancy. 3.5.

Basic Designs for Cells with 0.025 YOand Zero YOMercury

3.5.1. Cylindrical Cells using Sleeve Electrode Construction The best developed types of RAM cells are cylindrical cells using sleeve electrode construction, which are simply discribed as cylindrical cells. They are produced in A4 C and D size types with a typical trend to small sizes. In the Figures 15 and 16 of chapter 3.1. the design changes of the different cell sizes are shown. The principal construction of the cylindrical RAM cell is identical with the equivalent PAM cell design consisting of the same cell components. Differencies between these systems exist mainly in the capacity limitation of the anode to approximatly 113 of the cathode capacity, the application of improved separators and better anode and cathode formulations as already mentioned in the above citied chapter. A lot of experimental data exist for low-mercury (0.025 wt YOHg) as well as for mercury-free (0 % Hg ) RAM cells. They are summarized in Chapter 3.6.. 3.5.2. Cylindrical Cells using Spirally Rolled Construction RAM cells can also be designed as spirally rolled cells, which may be placed into the same container as the cylindrical cells. The advantages of this type of cell, which is called "jelly-roll" cell, too, are multiple. Figure 19 shows the principle construction of a "jelly-roll"cell [45]. The electrodes are manufactured in form of thin layer sheets providing good electrical conductivity by the incorporation of a metal screen. The improvement of the utilization of the active materials increases the rechargeability of the electrodes. ionic and electronic transport processes within the electrode masses are accelerated. RAM cells of spirally rolled construction have the advantage of better Mn02 utilization of thin plates compared with cylindrical cathodes. Figure 20 shows a diagram in which the utilization of Mn02 is plotted as a hnction of the layer thickness [46]. As an example, the cathode thickness in a cell of cylindrical construction is 3.2 mm and the layer thickness in a spirally rolled cell is only 0.6 mm. The utilization of the Mn02 at a load of 50 mA/g of Mn02 increases from 30 YOto 65 % and at 100 mA/g it increases from 10 % to 50 % . The diagram also illustrates that high power cells are only possible with thin electrodes. Assuming that a rolled D size cell contains 50 g

178

Mn02 than it could provide 1.5 hours of service at a load of 5 A or at an extreme it could support a load of 12.5 A for about 20 minutes. These values are competitive to rolled lead-acid or Ni-Cd batteries

CATHODE CURRENT COLLECTOR

% -

a

Mn07- UTlLlSATlO N 100 L

[%I

Ah I g H n 0 2

80

60 4.0 20 BOTTOM END

0 mA I g MnO2

Figure 20. Utilization of MnO2 as a finction of layer thickness.

Figure 19. Principle construction of a "jelly-roll" cell.

PERATURE: 22°C

z

0

:

a

20 '

0

-30 -20 -10

0 10 20 30 40 50 TEMPERATURE [ O C I

Figure 2 1. Temperature behaviour of thin layer alkaline Mn02-Zn cells on heavy and light load (0.03 % Hg).

OO

CHARGE : 1.72 V CONST. 15hrs.

! I

la

10

10

i,

60

CYCLES

Figure 22. Cycle performance of spirally rolled RAM cells (0.03 % Hg).

179

Spirally rolled cells provide a many times increased interface between cathode and anode, therefore reducing the current density at a given load current by the same factor. The electrode interface in a cylindrical D size cell is about 30 cm2 and in a rolled D size cell it would be approximately 300 cm2. The reduction of current density by a factor of 10 has a profound influence on the recharging behaviour and on the cycle capacity and cycle life of the cell. The low current density is also improving the low temperature performance. In Figure 21 the temperature behaviour of alkaline Mn02-Zn cells on heavy and light load is shown. At -20°C the capacity of the rolled cell will be still 40 YOof the nominal capacity instead of the 10 % experienced with cylindrical D size cells (heavy drain, cutoff 0.9 V). The cycle life performance related to the utilization factor expressed in percent of the initial capacity is not as steep as that of cylindrical cells and completely flattens out after 50 cycles. Figure 22 shows a typical cycle performance of spirally rolled cells [47]. Principally two versions of spirally rolled RAM cells can be produced: fi~llycharged and nearly discharged cells. The Mn02 cathode can either be partly or completely prereduced in a batch process [48] and the zinc anode may consist mainly of zinc oxide. This special version is easier to manufacture and the shelf life in the nearly discharged state is probably unlimited. But initial charging is required for a 100 % discharge capacity of the first cycle. The rolled cell design has also been studied for the application in Mn02-H2 cells [49]. In that system the influence of zinc on the cycling behaviour is not present and the cycle life is improved. However, it should be noted that rolled Mn02-H2 cells are capacity-wise far below the 4 AA or 7 AAA RAM cell arrangement unless electrodes without supporting metal screen are developed. 3.5.3. Bipolar Flat-Plate Design

Flat plate cells are usually made with electrodes spread on screens or grids like lead-acid batteries or industrial Ni-Cd batteries. In these types of batteries the electrodes itself are good metallic conductors. To build manganese dioxide batteries as plate cells good conductive screens would be needed. A well known example for a bipolar Mn02-Zn cell in commercial production is the 6 Volt Polaroid camera film battery, a primary battery with a weakly acidic electrolyte. The advantages of the flat plate design are principally the same as cited for the cyclindrical cells using spirally rolled construction: better utilization of the active materials, excellent Mn02 thin-plate utilization (Figure 20, 0.6 mm only), increased cycle life performance, accelerated ionic and electronic transport processes within the electrodes and the possibility of high power cells. In Figure 23 the correlation between cathode utilization and sheet thickness in the range from 0.6 mm to 3.2 mm under different current loads is illustrated. It can be seen, that the slope of these hnctions demonstrates the performance limits of different plate designs [50].The thickness of the cathodic layer should not exceed about 0.6 mm for high current applications. In bipolar design no external wiring for cells connected in series is necessary thereby saving total battery weight. Between the layers of bipolar electrodes the current load is evenly distributed and recharging will be uniform. This is not the case with cylindrical cells or with grid containing plate or spirally rolled cells which have locally different specific electrode loading and varying resistances. "Bipolar" means that cathode and anode are forming positive and negative layers of a sandwich type assembly which contains a conductive liquidimpermeable barrier, e.g. a plastic bonded carbon foil, between them. A porous separator

180

between the bipolar plates establishes a stack arrangement. Single electrodes of MnO2 and Zn are used as current collector and as positive and negative terminal. Figure 24 shows the principle construction of a bipolar 3 Volt alkaline Mn02-Zn battery [Sl]. For manufacturing of thin plate and layered bipolar cells two production schemes are possible: fully charged and nearly discharged cells. The latter version is easier to handle, the shelf life is probably unlimited, but initial charging is often required.

70

PLASTIC

~

0 150mAlg A 200mAlg

50mA/g 0 100mAlg III

t

60-

BIPOLAR

= 5oc

0 I-

4 404

i r

7 30: w

0

c

r

20100.6

I

I

1

I

I

.

I

I

I

,

.

,

,

1.4 1.8 2.2 2.6 3 THICKNESS OF SHEET [mml

I

M n 9 Zn +c

SEPARATOR

Figure 23. Correlation between MnO2 utilization and thickness of the cathodic layer.

Mt2

1

Zn

(-1

ENOPLATE

SEPARATOR

Figure 24. Principle construction of a bipolar 3 Volt alkaline Mn02 - Zn battery.

Assuming that a cathodic layer contains 10 g Mn02 than it could provide 1.5 hours of service at a load of 1 A or at an extreme it could support a load of 5 A for about 20 min. These current values are competiting with the datas of thin-plate lead-acid cells or spirally rolled Ni-Cd cells. In addition, the MnO2-Zn flat-plate battery surpasses both systems in Ah capacity per weight and volume by a factor of at least two [47]. The geometric size of a flat plate is the interface area between anode and cathode. As an example a 6 cm x 6 cm bipolar plate delivers about 1 A at a current density of 30 mNcm2. In comparison a cylindrical D size cell has about the same interfacial area and a comparable output. But the thickness of one assembly of a bipolar plate plus separator is only about 1.5 mm and the modular volume is therefore 5.4 cm3. A battery with 4 bipolar cells including endplates will have a volume of 25 cm3 and can deliver 1Aat6V. Principally the construction of a bipolar battery is not size limited. The plates can be as large as needed with the possibility of any shape desired. Like in fuel cell systems battery plates can be built in m2 dimensions. This is an important aspect, because regular plate batteries are size limited by the current collector grids like the lead-acid automobil starter battery, which get unproportionally heavy at larger sizes and heavy loads. Generally we can expect energy

181

densities in the range of 100 Whikg with bipolar alkaline Mn02-Zn cells of the primary type and about 80 Wh/kg with the rechargeable version. No other design will approach these values. For that reason the bipolar RAM cell would be an ultimate package for storage of electricity, for vehicles and stand-by purposes alike. 3.5.4. Bundle Cell Approach

In the former chapters we have seen that the thickness of the Mn02 electrodes of flat plate cells and of spirally rolled cells is decisive for their performance. This knowledge is the basic concept for the bundle cell approach. Cylindrical D and C size cells can be replaced by a package of AA and AAA cells increasing the efficiency of the Mn02 utilization, or single cells can be arranged as an array leading to heavy-duty power packs. In general a bundle cell is built from multiple paralleVseries connected cylindrical RAM cells. The cathode wall thickness in commercial primary cells is about 6 mm in a D, 4 mm in a C, 2 mm in a AA and 1 mm in a AAA cell [52]. If one would build a D or C cell with a thinner wall thickness, the Ah capacity would be decreased too much. The obvious solution is therefore to produce corresponding cell bundles of comparable total weight. In Table 5 the weights of various cylindrical cell sizes, their capacities under different constant currents to a cut-off voltage of 0.9 V and the short circuit currents are shown. The theoretical capacity of 1 g Mn02 is 0.3 Ah. Table 6 shows the corresponding values for some parallel cell arrays. Generally there are two parameters determining the use of a certain cell type in a batterypowered device: the volume and the weight. Weight-wise the 6 AA axray is equal to the single D cell and volume-wise the 4 AAA combination fits the C size can. If volume is the limiting dimension, one D size cell can be replaced by only 4 AA cells fitting into the same size container [50]. Even in this disadvantageous case the package performs better after 10 cycles (Figure 25). The situation is even more promising, if weight is the critical parameter. Then one D size cell may be replaced by a parallel arrangement containing 6 AA cells having about the same total weight. Besides the improvement of cycle life the parallel assemblies show also high load capability (Figure 26). Table 5 Weight, capacity (cut-off: 0.9 V, const. currents) and short circuit current (SCC) of single cells Sizes

D

C

AA

AAA

Cell weight (8) MnO2 weight (g) 0.30 A 1.25 A 2.50 A SCC (A)

130 80 12 Ah 3Ah 1Ah 17

62 40 6Ah 1 Ah

22 10 2Ah

11 6 0.8 Ah

16

15

13

182

Table 6 Weight, capacity (cut-off: 0.9 V, const. currents) and short circuit current (SCC) for packages b Y S

4AA

6AA

4AAA

7AAA

Cell weight (g) MnO2 weight (g) 0.30 A 1.25 A 2.50 A SCC (A)

88 40 8Ah 6Ah 4Ah 60

132 60 12 Ah 10 Ah 7Ah 90

44 24 4Ah 3.5 Ah 3Ah 50

77 42 7Ah 6Ah 5Ah 90

6 5

4

3 2 1

0

20

40

60

CYCLES

Figure 25. Discharge capacity of 1 D size cell and arrays of 4 AA and 7 AAA size cells (4 i2 load).

I

20 0.0

0.4

0.8 CONTINUOUS CURRENT [A]

Figure 26. Comparison of the discharge of 1 D size cell and a parallel array of 6 AA cells.

The energy densities of the parallel assemblies are very high, inspite of the high load carrying capability, e.g. for the 4 AAA array 110-120 Whkg at a 1.25 A load. The larger cathodelanode interface cuts the current density down by factors of 2 to 6 and simultaneously the internal resistance is divided by the number of cells connected in parallel. This will be especially important for cells containing low or even no mercury, because they passivate easily under heavy load conditions. Figure 27 illustrates the effect of lowering the internal resistance and the decrease of the polarisation of the Mi102 by bundling 6 AA cells instead of using 1 D size cell. In Figure 28 the corresponding relationship of capacity versus current is shown. The increase of capacity in an 6 AA cell array is obvious compared with a single D size cell.

183 V 1. 2 .4 1.O .3 0 . 8 .2 0 .6 .1 0 . 2 28 HRS, 0 0

HI

V

.5

1.2 1.0 0.8 0.6 0.2 0

0

4

8

1

12

16

2

20

3

24 4

4

8 1

12

2

16

3

20 24 4

28 HRS,

Figure 27.Comparison of the polarisation and the internal resistance of 1 D and 6 AA cells. The better efficiency and the more evenly utilization of the Mn02 cathode in bundle cells lowers the loading per cell and leads to an improved cycle life. But if required, the cell can also supply very high currents for special applications, e.g. hand tools, video cameras, portable TV's and wheelchairs. Figure 29 shows how the change from 1 D sue cell to a multi-cell

100

> t 3

2 4 a

U

2 2 a

U

20I

0

u 1

30

1000 2000 CONTINUOUS CURRENT [ m A l

100

300

Figure 28. Relationship of capacity versus current for 1 D cell (1) and an array off 6 AA cells (2).

0 CYCLES

Figure 29. Cycling (100 % DOD) of 1 D cell compared with a 4 AA and a 7 AAA cell array. Every 6 th cycle is shown (1. - 60. cycle).

arrangement in the same size container will change the cycling characteristics under similar load conditions. Further more the anodic capacity limitation realized by an accurately balanced formulation of the gelled powder zinc electrode is easier to achieve. And separator resistance becomes less important, therefore a higher quality membrane-type barrier can be used, restricting the growth of zinc dendrites on charge. Another advantage of bundle cell

184

restricting the growth of zinc dendrites on charge. Another advantage of bundle cell arrangement is the better reversibility of the swelling of RAM cathodes on discharge and the volume contraction on recharge. In 1990 a bundle cell was designed which powered an electric motor-bicycle partizipating in a Solar Mountain Race in Austria [53]. The race included a crossing of the Alps via a 2000 m high pass. A hlly charged battery was projected to bring the vehicle up to the highest point. During the subsequent run down it should be recharged by regenerative braking. The battery consisted of 480 AA RAM cells arranged in 20 layers, each of them containing 24 cells in parallel. The layers were then connected in series to give an open circuit voltage of 30 V. Nominally this was a 25 V, 24 Ah or a 0.6 kWh battery. The battery weight was 12 kg and yielded an energy density of 50 W g . It should be noted, that the average discharge cycle Ah-capacity of a single AA RAM cell was rated as about 1 Ah. For comparison the primary AA cell may be rated as about 2 Ah (100 Wh/kg) under average load conditions. The vehicle passed the competion and the inventor - who was driving it - won a prize. 3.6. Performance Data of Cylindrical Cells with 0.025 YOand Zero YO Mercury 3.6.1. Electrical Performance At the moment there are no accepted standards available for RAM cells. They have to be subjected to standard tests for primary alkaline manganese dioxide zinc (PAM)cells. Table 7 Table 7 Electrical performance of Mercury-Free and Low-Mercury AA RAM cells (service hours)

0 % Hg

Discharge Test

Load Condition

IEC Minimum Average

RAM

0.025 % RAM

Premium Primary 0.025 % Hg

Transistor Radio Walkman Motor & Toys Pulse Test (Photoflash) Performance Comparison (%)

75n 4 m

100

97.4

105

150

13.1 5.8 40 1

15.0 5.8 387

16.0 6.6 516

m 11.0 l0Q I 3.923 IH/D 5.8 1.823 15/45Sec 320 100

(100

-

127)

Note: The capacity of the first cycle of a RAM cell is approximately 75-85 % of the capacity of a PAM cell. shows the electrical performance of Mercury-Free (h4F) and Low-Mercury (LM) AA RAM cells compared to the International Electrotechnical Commission (IEC) proposed new standards for single use alkaline cells [54]. These standards are designed in a way that 80 % of the primary alkaline cell manufacturers on a global basis meet the requirements. It is apparent that

185

LM RAM cells surpass these requirements by over 25 %. The table also indicates that LM RAM cells are capable of providing approximately 80 % of the service life of North American premium primary cells. In summary, one can say that RAM cells provide a similar service life on their initial use cycle compared to the performance of PAM cells. RAM cells of the 1991 total service life performance are typically replacing between 7 and 20 PAM cells depending on various use conditions. When operated to cell exhaustion (deep discharge) on each use cycle, the operating time gradually decreases and afler about 20 to 30 cycles the operating time is reduced to the one of a typical zinc carbon cell. If the RAM cell is recharged prior to exhaustion (shallow discharge) the cycle life increases dramatically. On shallow discharge cycling the performance of AA RAM cells exceeds 350 cycles removing 350 mAh per day (7 hours a day, 24 R) and in excess of 100 cycles 500 mAh per day (4 hours a day, 10 Q). Figure 30 and 3 1 [55] show the total service comparisonbetween P A M and RAM

LLW

ou

0

5

10 15 CYCLES

20

Figure 30. Total service hours comparison between PAM and RAM AA cells on 10 R continuous deep discharge to 0.9 V.

25

0

20

40

60

80

100

CYCLES

Figure 3 1. Total service hours comparison between PAM and RAM AA cells on 10 R shallow discharge (4 hours a day).

AA cells on 10 R for continuous discharge to cell exhaustion and on shallow discharge (4 hours a day). The benefits RAM cells represent as an alternative to throw away dry cells are obviously significant. In Table 8 performance data of LM and MF AA RAM cells on various continuous drain rates are summarized.Even at high drain rates of 1 R continuous discharge to 0.75 V MF RAM cells provide similar service as LM cells. The development of mercury-free RAM cells to an excellent performance level is pursued for the near future. Presently the MF RAM cells have advanced to that point that both the electrical and the high temperature storage performance have reached acceptable levels.

186

Table 8 Performance data of Mercury-Free and Low-Mercury AA RAM cells on various continous drain rates Test

0 Yo Hg

0.025 % Hg

Cumulative Capacity (Ah)

Cumulative Capacity (Ah)

Cycles la 2.2 a 3.9 R 10 a

1 0.776 1.253 1.466 1.598

OCV(mV) SCC (A)

1529 6.51

10 5.32 7.17 8.30 9.93

25 10.1 12.66 13.90 17.12

1 0.988 1.287 1.518 1.645

10 5.78 7.60 8.44 9.66

25 9.91 12.40 13.64 15.92

1566 11.34

3.6.2. Performance at Different Temperatures

The LM and MF RAM cells improve their performance on the 3.9 R and 10 R load test when the operating temperature is increased to as high as 65 "C without leakage or bulging. They show excellent discharge performance compared with commercial PAM cells. In 3

1

RAM (O%Ho)

100

200

300

400

500

CONSTANT CURRENT CmAl

Figure 32. Capacity of RAM cells (0 % Hg) and commercial cells (0 % Hg) at different temperatures.

Figure 33. Comparison of PAM and

RAM cells containing 0.025 % and 0 % mercury per cell weight regarding charge retention after high temperature storage (3.9 Q to 0.75 V).

Figure 32 the capacity curves of RAM cells and of commercial cells, both with 0 % Hg, are illustrated at different temperatures. The behaviour of the h4F RAM cells far exceeds the

187

discharge capacity of the commercial cells, e.g. at 22OC. This effect can be attributed to the use of catalysts for the hydrogen recombination (Ag02) in MF RAM cathodes. Commercial cells with 0 % Hg contain organic inhibitors for the suppression of hydrogen evolution in the anode. 3.6.3. Charge Retention

The charge retention of a new RAM cell is quite similar to the one of a PAM cell as demonstrated in Figure 33. It has been shown in accelerated tests that RAM cells retain approximately 80 % of the charge for a period of five years from the date of manufacture. Table 9 shows a comparison between LM and MF RAM cells regarding charge retention in an accelerated test. It can be seen that MF RAM cells are capable of excellent charge retention. Table 9 Electrical performance comparison after high temperature storage between Mercury-Free and Low-Mercury AA RAM cells Test (3.9 a)

Time Temp. Cycle 1 (weeks) ( " C ) (Ah) 2 1

2 3 4

0.025 Yo Hg

0 % Hg

55 65 65 65 65

1.466 1.407 1.355 1.303 1.113 1.074

%Loss

10 Cycles Cum.(Ah)

Cycle 1 (Ah)

%Loss

10 Cycles Cum. (Ah)

-

8.30 7.84

1.518 1.351 1.175 1.026 0.902 0.738

-

8.44 8.25

4.0 7.6 11.1 24.1 32.8

11.0 22.6 32.4 40.6 51.4

3.6.4. Abuse Tests

There are a variety of abuse tests for RAM cells. They have to sustain the following procedures: Permanent shorting for 24 hours, heating of cells to 95"C,cooling to -30°C and heating to +65"C for 20 temperature cycles. The arrangement of 1 discharged cell in series with 4 fully charged cells is considered successfully, if no decrimping of cells occurs; safe venting is allowed. Abuse tests for mercury-free cells are far more critical than for mercurycontaining cells. 3.7. Comparison between RAM Cells and other Battery Systems

Table 10 shows a comparison of RAM cell performance with other commercially available dry cells [56]. RAM cells have an excellent shelf life comparable with PAM cells and a good cycle life depending on depth of discharge (50-500 cycles), nearly in the range of Ni-Cd cells. The average discharge capacity even exceeds the values for Ni-Cd cells, only PAM cells are better. But one of the most important advantage of RAM cells is the low toxicity (0.025 YOHg

188

Table 10 Performance specifications of RAM cells and commercially available dry cells

PAM

Ni - Cd

RAM

2-3

4

0.1 - 0.2

3-4

1

1

200 - 800

50 - 500

AA C D Voltage (V) Manufacturing Cost Operating Cost Toxicity

200 500 2500 5500 1.5 Low High Low

800 2500 7000 14000 1.5 Medium High Higher (Hg)

175 600 1200 1200 1.2 High Low High (Cd)

Quantity of discard

Great

Great

Less

350 900 2500 4500 1.5 Medium Low Low (Hg-free, rechargeable) Less

Heavy Duty Zinc Carbon

75 % Charge Retention at 20 "C (years ) Life Cycle Depending on the Depth of Discharge *Average Discharge Capacity per cycle (mAh) AAA

* A battery with a capacity of 1000 mAh will operate a wallanan (requiring 250 mA for 4 hours). or Hg-free) and the reuse of materials due to its rechargeability leading to a lesser quantity being discarded. A comparison between energy densities by weight and volume for some existing and advanced battery systems is demonstrated in Table 1 1 [47]. Research studies have shown that Mi102 electrodes are as powerhl as lead-acid or nickel-oxide electrodes when used in correspondingly thin layers. Energy density (60 Whkg), power density (450 Wkg) and charge retention (2-3 years, 20°C) of flat-plate RAM cells are excellent and many times better than for spirally rolled Ni-Cd and flat-plate Pb-acid batteries. In Figure 34 the cumulative capacity of PAM and RAM cells in the AA configuration is shown [57]. Depending on depth of discharge (shallow or deep) RAM cells are able to replace a multiple of PAM cells. Figure 35 shows a comparison of PAM, RAM and Ni-Cd AA cells [55]. At room temperature and above the discharge capacity of RAM cells is higher than for Ni-Cd cells. The improved performance of "spirally rolled" Ni-Cd cells becomes obvious at low temperatures compared to the cylindrical construction of PAM and RAM cells. The charge retention of Ni-Cd, Ni-metalhydride and RAM AA cells after storage at 65OC obtained on a 3.9 SZ discharge to 0.75 V is illustrated in Figure 36. At this storage conditions Ni-metalhydride cells fail and Ni-Cd-cells can only be used meaningfilly aAer 1 week storage [56]. RAM cells are nevertheless able to supply 500 mAh after 3 weeks at 65OC. Tests comparing the temperature stabilty of Mn02-Zn (RAM), Mn02-H2 [49] and Ni-rnetalhydride cells resulted in the data of Figure 37 [16]. It is confirmed, that the shelf life of nickel oxide is

189

far worse than that of manganese dioxide at elevated temperatures. The presence of a hydrogen atmosphere accelerates the reduction of the NiO, even more. Table 11 Comparison of energy densities of advanced batteries ~~

Property Energy Density (Wh/kg) Energy Density (Wh/L) Power Density (Wkg) Power Density (WL) Cycle Life at 20 % DOD Cycle Life at 100 % DOD Operating Voltage (V) Charge Retention at 20°C Charge Retention at 65°C Memory Effect Disposal Hazard

~~~

Ni - Cd (spiral)

RAM (thin layer)

Pb - acid (flat-plate)

27 66 72 170 500 100 - 200 1.2 3 - 6 month 1 day Yes Toxic

60 110 450 900 300 - 400 50 1.2 2 - 3 years 6 month No Non - toxic*

35 80 150 340 300 - 400 20 2 3 - 6 month about 2 wks No Toxic

* Mercury - Free Version

n L

a 150,

1

U

ed DEEP

"

U

I

RAM

PAM

Figure 34. Comparison of the cumulative capacities of PAM and RAM cells in AA configuration.

40°C

209:

-20°C

Figure 35. Comparison of PAM, Ni-Cd and RAM AA cells on 3 10 mA constant current discharge to 0.75 V obtained at various temperatures.

190

- 1.2- n T E M P E R A T U R E 65°C c

d

c

100

-$ 80

n OWEEKS

1.0-

t 3 1 WEEK

2 WEEKS 0 3WEEKS

20.8-

60

> k 240

5 0.6W

z

a 0.4-

I

2 0.20 0-

RAM

Ni-MeH

Figure 36. Comparison of PAM, Ni-Cd, Ni-MeH and RAM AA cells regarding charge retention (discharge: 3.9 Q to 0.75 V).

20 0

\

0 250( Mn02-Zn AND Mn02-H2 (EQUAL TO 1WEEK AT 55°C

',

Ni-MeH (OVONIC 1988)

\

25°C

\

55°C CAPACITY LOSSES AT 25°C

.

.

10

20 TIME (DAYS)

AND 5 5 9 30

Figure 37. Shelf life and temperature stability of Ni-MeH, Mn02-H2 and Mn02-Zn cells at 25°C and 55°C.

3.8. Charging Procedures

RAM cells have to be charged with voltage limited taper chargers (VLTC). A typical VLTC-charger for the major cylindrical RAM cell sizes (D, C, AA, AAA) employs a maximim current of 500 mA and a maximum voltage of 1.72 V. In Figure 38 the current and voltage response of a multicell charger is shown [57]. Due to the constant voltage charge regime, consumer chargers for RAM cells will be designed in a parallel arrangement. The effect of charging voltage on RAM cell performance indicate, that an acceptable performance is achieved between 1.65 and 1.69 V at room temperature. Performing long term taper charge experiments revealed that RAM cells have been maintained at 1.65 V for up to one month without bulging or leakage. The typical charger circuit design is similar to the usual transformer-rectifier power supply used in Ni-Cd chargers except for the voltage regulator. Due to the sleeve electrode construction the time to recharge a RAM cell to 100 % state of charge (SOC) takes longer than it does to recharge a spirally-rolled Ni-Cd cell. On the other hand up to 75 % SOC for RAM cells can be achieved at a faster charging time than would be required for Ni-Cd cells. A general purpose charger for AA cells is capable to completely recharge the RAM cell within a period of 8 to 12 hours. Several of these circuitries provide an optimal state of charge indication for RAM cells and include temperature compensation. If properly designed the taper charger minimizes cell heating and internal gassing. The recent development of special pulsechargers is able to decrease the charging time for RAM cells at about 50 % [51]. In Figure 39 pulse charging of a single RAM cell and of 4 AA RAM cells in parallel arrangement is shown. Up to 1 hour of charging time a typical charge current peak exists. After 2-3 hours cells have obtained a sufficient charge capacity.

191

" -2

3.0 2.5

U

Y

k-

> 2.0

z k

1.5

1.00 II

cia

0.75

2 0.5 0

CURRENT 5

10 15 TIME [ H R S 1

100

=

u

4

u

1.0

0.5 20

Figure 38 .Current and voltage response of a multicell charger.

0 0

40

80

120

TIME

[MINI

160

200

Figure 39. Pulse charging of 1 AA and 4 AA RAM cells.

4. BATTERIES AND ENVIRONMENT 4.1. Environmental and Waste Management Issues

Waste management and waste reduction have become important issues for governments and concerned citizens. The disposition of toxic waste has been a concern of private agencies for many years. However, growing public awareness and active pollution lobbying by environmental organizations has caused many governments to initiate or propose legislation for the sale and disposal of many products deemed to be hazardous to the environment. Cells containing mercury in any amount are of particular concern and environ-mental agencies are petitioning governments to enact legislation to ensure the cells are reclaimed and reprocessed or disposed of in controlled disposal sites. The reduction of the amount of mercury used to reduce the gassing and increase the shelf life and performance of zinc anodes is now a major priority for all primary cell manufacturers. The industry must demonstrate its ability and intention to meet minimum mercury content in response to the growing public concern. Recycling efforts have been implemented for rectangular lead-acid, nickel cadmium automotive and industrial batteries. Several efforts are directed at the recycling of cyclindrical consumer batteries. Small format battery collection boxes have been set up in several municipalities in an attempt to prevent batteries containing hazardous materials from entering garbage incinerators, composters and landfills. A strong demand exists for low-cost environmentally benign rechargeable batteries with long shelf life which are completely dischargeable and conveniently rechargeable. Low cost and safety are also issues of increasing importance. RAM cells are a viable alternative to single use "throw away" batteries resulting in waste reduction as well as reuse of valuable raw materials and resources. The battery waste generated by PAM, RAM and Ni-Cd cells in a popular video game application (Nintendo "Gameboy" utilizing 4 AA cells) was modelled for 1000 units operating for 7 hours a day over a period of a year [ 5 5 ] . In the model 1000 life cycles were assumed for Ni-Cd cells. Figure 40 and 41 demonstrates the reduction of disposed cells and of total waste volume due to the use of RAM cells. In Figure 42 and 43 the total waste weight

192

and the toxic waste weight is shown. Depending on the frequency of the recharge the waste generated by spent RAM cells in this application represents a waste reduction between 86 % and 97 % when replacing PAM cells. Ni-Cd cells also reduce the total waste very successfblly, but they contain toxic waste (4 kg Cd). The toxic waste of RAM cells (< 0.1 kg Hg) is extremely small compared with P A M cells (1.2 kg Hg).

2 300 0 0

BATTERY

BATTERY

WASTE

WASTE

7

U

a 200 W

Vl 0

a

:

100

-I i l W U

0 PAM

RAM

Ni-Cd

Figure 40. Cells disposed created by PAM, RAM (containing 0.025 % mercury per cell weight) and Ni-Cd cells in a Nintendo Gameboy application.

0 I-

PAM

RAM

Ni- Cd

Figure 41. Total volume of battery waste created by PAM, RAM (containing0.025 % mercury per cell weight) and Ni-Cd cells in a Nintendo Gameboy application.

n 5 z 0 I-4 Vl

Y

? 3 Vl a 3 2 -I a I-

0 PAM

RAM

Ni-Cd

Figure 42. Total waste created by PAM, RAM (containing 0.025% mercury per cell weight) and Ni-Cd cells in a Nintendo Gameboy application.

PAM

RAM

Ni-Cd

Figure 43. Toxic waste created by PAM,

RAM (containing0.025 % mercury per cell weight) and Ni-Cd cells in a Nintendo Gameboy application.

193 4.2. Battery Recycling

A few battery systems are containing dangerous substances like mercury, cadmium and lead and can cause significant health problems if allowed to conterminate our environment through misuse or improper disposal. Therefore battery recycling issues are of relevance to our present environmental situation. The EC (European Community) waste management directive and regulation proposals regarding hazardous waste, movement, landfill and incineration of waste already include: nickel, zinc, lithium, silver and cobalt in addition to mercury, cadmium and lead. This will influence indirectly the disposal of spent batteries and accumulators. Table 12 illustrates the classification of some substances contained in spent batteries [581. In Western Europe about 120,000 todyear of portable batteries (99 % Mn02-Zn) are sold that would represent about 0.1 % of the European household waste. Every Western European citizen needs seven to eight portable batteries per year. For lead-acid and nickel-cadmium accumulators and for some buttom cells established and advanced recycling technologies exist. Table 13 shows the status of battery recycling technology for different battery systems [59]. In Switzerland a pilot plant for unsorted "household" batteries with 500 - 2,000 tons/year started its operation successhlly in 1991. No technology on a large commercial scale is available for carbon-zinc and alkaline manganese dioxide-zinc batteries. For portable lithium batteries there is no special recycling technology available. Table 12 Classification of some substances contained in spent batteries Substance

Classification

Mercury Cadmium compounds Lead compounds Lithium Zinc chloride Potassium dihydroxide Sulphuric acid

Toxic Harmful Harmful Highly Flammable Corrosive Corrosive Corrosive

Nickel Nickel dihydroxide Cobalt

Harmful + Carc Cat 3 Harmful + Carc Cat 3 Irritant

Adapt

12 th (3.91)

15 th (9.91)

EC directive 67.548 A significant recycling potential of metals is granted for high concentration of reuseable metals in spent batteries. This situation is given in the case for recycling of lead batteries in which lead is present in the order of magnitude of 65 % of the total battery weight. Today lead is the most recycled of all metals with about 54 YOof Western World lead production coming from secondary sources. In USA recycling rates for SLI and industrial batteries for 1987, 1988 and 1989 were pronounced with 89 %, 91 % and 95 % respectively. But also Ni-Cd

194

Table 13 Status of the battery recycling technology System

Technology

Components

Recycling*

Economy

Lead-Acid Nickel-Cadmium Zinc-HgO Zinc-Ag2O Zinc-02 Zinc-Carbon Zinc-MnO2 Lithium Mixtures

existent existent existent existent none pilot pilot none pilot

Pb Ni, Cd, Fe Hg (zn, Fe) Ag (Hg, Zn) Zn, Fe, Hg Zn, Mn, Fe Zn, Mn, Fe Li, Fe, Mn Ag, Cd, Fe, Hg, Li, Mn, Ni, Pb, Zn

>> 95% ap. 60% >> 60% ap. 60%

yes I no yes I no

no Yes no no no no no

*) for the former West Germany

Table 14 Rounded percentages for metallic substances used in battery systems Pb

Ni

Cd

Zn

20

20

-

Mn

Ag

Hg

Li

Fe

20 30

-

0-0.01

25

0-0.025 <1 30
35 40

~~~~

~~

Pb-PbO2 Ni-Cd Zn-Carbon Zn-Mn02 Zn-Ag2O Zn-HgO Zn-02 Li-MnO2

65

20 15 10 10 30

-

25

30

35

2

40 45 50

accumulators with a proportion of 20 % for both active masses is a good canditate for battery recycling, as the materials can be separated easily. The percentages for metallic substances used in battery systems are listed in Table 14 [59]. The profitability of the recycling of lead accumulators and Ni-Cd accumulators depends on the metal prices. Due to fluctuations it swings between losses and profits. The recycling process for mercury button cells and for Mn02-Zn primary cells require financial subsidues. In October 1989 the Canadian Council of Ministries of Environment adopted as its goal the reduction of wastes in Canada by 50 % by the year 2000 [60].Therefore the government has spelled out the "four Rs" of waste management in order of preference: reduce, reuse, recycle and recover. With respect to reduction, rechargeable batteries are promoted as preferable to primary batteries, even though the relative environmental risk is still debated by experts. In March 1991 the European Community has adopted the following directive on batteries and accumulators containing dangerous substances: alkaline manganese dioxide batteries

195

should not contain more than 0.025 % mercury and mercuric oxide, nickel-cadmium and leadacid batteries must be collected separately from household waste for recycling or special disposal [59]. The development of new products with less or without dangerous substances must be a future goal of priority to avoid hazards to the environment. Good examples are lithium primary and secondary cells and the mercury-free rechargeable alkaline manganese dioxide cell, which is produced at a pilot plant in Canada [57]. But also the nickel-metalhydride cell will be a favourite to supplement and partly substitute nickel-cadmium cells. 4.3. The Status of Battery Recycling in Switzerland In 1991 the world’s first commercial plant for battery recycling for unsorted household batteries, which uses the Recytec Battery Recycling Process, started operation in Aclens, Switzerland. The technology applies a thermo-mechanical and chemical-electrochemical separation process for recovering valuable materials from unsorted household battery waste. It efficiently refines waste components to raw materials that can be recycled into the industry. The plant started in a first phase with a 500 tondyear capacity. In a 3-year period it will be extended to the projected 2,000 tondyear [61]. The facility in Aclens is additionally designed for neodfluorescent tubes and electronic scrap, that is estimated to be on the order of ten to thirty times larger compared to battery waste. Two million fluorescent light lamps and metal vapor tubes will be recycled per year. The urgent need for treatment and recycling of household batteries produced and consumed in Switzerland can be attributed to particular circumstances. Total national battery production has climbed in 1990 to between 3,700 - 5,000 tondyear. The distribution of battery sales for 1987 and 1988 is provided in Table 15 [62]. Despite this rather significant level of battery

Table 15 Distribution of commercial battery consumption in Switzerland for 1987 and 1988 Battery Type

Zinc-Carbon Alkaline Manganese Nickel-Cadmium Hg-Button Cells Silver Oxide Zinc-Air Lithium

Total Consumption (“A) 1987

1988

55 39 3.5 0.28 0.08 0.0028 0.11

49 46

4.4 0.25 0.18 0.006 0.15

Trends

Decreasing Increasing Increasing Constant Increasing Increasing Increasing

Reference: Import and production statistics in the year 1987 and 1988 of the Swiss Federal Office of the Enviroment of Bern.

196

consumption, there are no hazardous waste landfills located in Switzerland. And by Swiss law, batteries are defined as hazardous wastes. In addition, burning batteries and other hazardous wastes in incinerators is prohibited and since 1991, there are no export permits for spent batteries by the government. Up to that point in Switzerland, like in other Western European countries, most collected batteries were transported to the landfill of Schoneberg in the former East Germany. In 1989 an amount of 1,590 tons of household batteries were collected and exported. The Swiss current collection rate of spent batteries is about 35 %; an intermediate storage of battery waste has been established at the recycling facility in Aclens. The Recytec Process can accept any type of household battery: zinc-carbon, alkaline manganese dioxide-zinc, nickel-cadmium, lithium or any other battery. Previous sorting of the battery mix is not required, only visual inspection for non-battery wastes is done. One of the great advantages of this recycling process compared to other technologies, e.g. the Japanese Sumitomo process, is its excellent recycling efficiency of 95 %, producing a total output of 13 different products of high purity (> 99 %) and only 5 % of secondary waste. In Figure 44 an overall mass balance based on production-scale illustrates the flow of the materials recovered in the process [61]. Important steps of the overall Recytec Process for battery recycling are as follows: pyrolysis and gas treatment (condensation), shredding and washing, magnetic-inductiveseparation and electrolysis of the non-ferrous scrap and magnetic separation followed by the electrolysis of manganese dioxide and zinc. The first process step is thermal treatment (Pyrolysisto 55OoC) under reducing atmosphere. The mercury content of the battery material is reduced down to approx. 2 ppm and a reduction of battery weight of roughly 18 % is achieved. Effluent gases are passed through a condensation step for further purification. After thermal treatment the remaining solids are shredded and water washed to mobilize the manganese dioxide into suspension and to dissolve different salts. Zinc and graphite used in a finely distributed form are associated with the manganese oxides and wash out into the suspension. Manganese dioxide and zinc are 1,000k g UNSORTED BATTERIES 130 kg w a t e r 20 kg oils 1 kq mercury +zinc/cadmium 30 kg pyrolyss gas (buried in-situ)

L ~SHRE,ODER] b

MAGNET'C

r(ELECTROLYSlg230 1

(MAGNETIC/ INDUCTIVE SEPARATION] I

IELE c TRoLY s IsJ

30 k.a o salts-

100 kg manganese dioxide kg zinc + 5 kg cadmium 190 kg manganese oxides+ 50 kg graphite

'

150 kg iron

+ 20 k g graphite t

35 k g zinc + cadmium 30 kg copper 9 kg nickel + silverlgold

Figure 44. Material balance and flowsheet diagramm for the Recytec Recycling Process

197

solubilized in an acidic leaching procedure and separated in a simultaneous electrodeposition process. Magnetic separation is used to remove the ferro-magnetic materials. The nonmagnetic and non-inductive fraction is predominantly graphite (96 %). The remaining solids of the magnetic-inductive separation step enter the electrochemical system and solution processing as anode material. A selective anodic dissolution leads to good separation of the metals and purities higher than 99 % for the recovered metals. Due to its closed liquid loops for water and electrolytesonly about 50 m3 salt brine per year based on the 500 tondyear battery capacity has to be transferred to a waste water facility outside. The dust and activated carbon filters keep the process emissions to acceptable legal and environmentallimits. The costs of the recycling facility are established by the Swiss Model for financing the recycling of household batteries. A surcharge on every battery sold ($.05to $.30) and an prospective collection efficiency of 80 % are the two basic objectives. The advantage of this Swiss system is that only the actual battery consumers pay the bill and the administrative overhead maintained at a relatively low level. Perhaps the only disadvantage is that, since there is no direct financial gain for the consumer, "public education" represents the only means to stimulate consumers to improve battery collection rates. 5. TOXICOLOGY

Most of the commercial battery systems, e.g. zinc-carbon, manganese dioxide-zinc, nickelcadmium, lead-acid and mercury button cells contain toxic substances. Strong efforts have been made to recycle these batteries, to lower the concentration of their toxic substances or to replace them with alternative systems. Nevertheless, battery production processes as well as disposal or recycling activities of spent batteries are responsable for the infiltration of a few toxic substances in our environment. The following chapters describe the toxicology of mercury, cadmium and lead, which are the most toxic components found in different battery systems. 5.1. Mercury

Mercury is known as a toxic substance. Because of its high vapor pressure, metallic mercury disperses relatively quickly into the atmosphere. Its toxicity depends mainly on the state of aggregation and the degree of dispersion. In comparison to the liquid metal fine particulate dust containing mercury and mercury vapors are very toxic. Compounds containing bivalent mercury are generally more poisonous than monovalent ones. With increasing solubility the toxicity of inorganic mercury compounds increases, but organic lipid soluble mercury compounds are in most cases more toxic. Thus metallic mercury and all mercury compounds are toxic, with the exception of the practically insoluble red mercury sulfide. In human beings mercury acts as a general cell and protoplasmic poison, that may bonding to the sulfhydryl groups of proteins, denaturating proteins, damaging membranes and reducing the RNA content of cells. This leads to blocking of many enzyme systems. The kidneys and nervous system are especially vulnerable, but also toxic dermatitis may occur. Mercury is mutagenic, teratogenic and embryotoxic, especially as methylmercury compound [63, 641. Environmental contamination with mercury leads to a critical concentration effect in animals that occupy higher positions in the food chain. Concentrations of 10 ppm Hg and more have been found in certain fish. Although the tolerated concentration of mercury in foods, e.g.

198

in USA is only 0.05 ppm, fish originatingin Japan's Miniamata Bay contained up to 9.6 mg Hg per kilogram ("Minamata disease"). Acute poisoning occurs when mercury ion concentrations reach 0.2 mgJ100 ml of blood. Chronic inhalant intoxication can be expected with mercury vapor concentrations of 0.1-1 mgJm3. A daily 5 hour exposure to inhaled mercury vapor concentrations of 0.1 mdm3 leads to severe chronic mercury poisoning [65]. 5.2. Cadmium

About 40 years ago it was already established that long-term inhalation of cadmium oxide dust could cause a syndrome characterized by damage of the pulmonary and renal systems [66]. And it is a well known fact, that during the last decades in Japan a large population suffered from a chronic cadmium poisoning with severe bone damage ("Itai-Itai desease"), caused by the intake of highly cadmium contaminated rice. Waste disposal, especially by incineration, as well as phosphate fertilizers and acid rain are pronounced risks for cadmium exposure. The average normal daily intake via food is 10-20 pg in most countries. The highest cadmium concentrations are found in some basic foodstuffs such as wheat, rice, liver, kidney and certain seafoods. Drinking water and atmospheric exposure (1-10 ndm3) are generally of minor importance. Smoking twenty cigarettes a day has been estimated to result in an inhalation of about 3 pg [67]. The toxicology of inhaled cadmium compounds depends on particle size and solubility. Finely divided cadmium oxides, especially fumes, deposit in the lower respiratory tract. About 30 % of the inhaled amount of this relatively high soluble substance is absorbed. Orally ingested cadmium is normally absorbed to a few percent, but nutrititional deficiences, e.g. in iron and calcium, will cause absorption levels up to 20 % [68]. Absorbed cadmium from the lungs or the gut is mainly stored in the liver, that has a high capacity to synthesize metallothionein, a cysteine-rich protein, which strongly binds cadmium, but also zinc and copper [69]. The kidney can also synthesize metallothionein to ensure cadmium trapping in an inert form. Normally excretion of cadmium from the kidney is very small. Long retention times in liver and kidney, as well as in other tissue, lead to an accumulation of cadmium in the kidneys from birth to middle age. The gross biological half-life of cadmium in the body has been estimated to be about twenty years [70]. Therefore past exposure is often more important than the present one. Average cadmium concentrations in the liver of adults are about 1 mgkg wet weight and in the renal cortex about 20 m a g in most European countries and in North America. In the blood of non-smokers the cadmium concentration is < 1 pg/l and for smokers up to 5 pg/l [67]. But also in the urin smokers excrete about twice as much as non-smokers. Generally the urinary excretion is a usehl aid for monitoring the total body burden of cadmium. Acute cadmium poisoning by inhalation has mainly been caused by accidental exposure to cadmium fumes. A short-term exposure limit for cadmium oxide fumes and respirable dust of 0.25 m@m3 has been recommended to prevent acute lung reactions; for longer times exposure should not exceed 20 pg/m3 [71]. Chronic effects of cadmium after long-term inhalation are mainly seen in the lungs and kidneys. The critical effect of cadmium is tubular dishnction. An increased excretion of low molecular mass proteins is regarded as the first sign of renal tubular effects.As the tubular damage is irreversible, prevention is therefore absolutely necessary.

199

5.3. Lead

Today lead is found in the environment and measureable amounts of it exist in all adult body tissues and fluids, although lead is not an essential element of the body. Ingestion of food is generally the major source of lead intake. Lead in the diet can come from the uptake of lead into plants, deposition on vegetables, soil or water and from lead solder in canned food. Drinking water can be a significant contributor to the lead intake resulting from natural sources, lead-lined tanks, lead pipes and lead solder-joints. Concentrations in drinking water exceeding 300 pgA have been reported [72,73]. Very soft water, low pH and high temperature favour high lead contents. By far the greatest contribution to the atmospheric lead level comes from automotive lead emissions. In most countries strong efforts have been made to reduce or replace the amuont of lead in the gasoline, where it is added as tetraethyl lead. Another important source of atmospheric lead emission, estimated to be about 10 % [74], is contributed from industry by smelters and recovery operations. Significant values have been found in the proximity of primary and secondary smelters, e.g. peak levels in excess of 200 pg/m3 are reported in the former Yugoslavia [75], whereas the standards are in the range of 1.5 - 0.5 pg/m3 (EPA). Concentration of lead in street dust of heavy traffic areas and surface soil can be as high as the soil of lead smelter industry. Dust analysis of 77 US cities showed a lead content of 1600 ppm in residual areas and 2400 ppm in industrial ones [76]. Lead-bearing dust can be especially dangerous to very young children, because of a high level of ingestion through their play and eating habits. Lead is absorbed from the gastrointestinal tract. The amount of absorbed lead through the digestive system is greatly influenced by stress, fasting and the solubility of the lead compound. As an example, lead acetate is 25 times more soluble than lead sulfide. The fine particles (0.1-lpm) of lead fume can readily enter the respiratory system and be absorbed in the blood stream. Lead can almost inhibit all steps in the biosynthesis of hemoglobin and globulin. From the blood stream it deports to soft tissue and gradually accumulates in the bone. Particles of larger size (> 1-2 pm) may be deposited in the nasopharangeal area, trachea and bronchi and leave by swallowing or expectoration. Organic lead compounds (e.g. tetraethyl lead) are readily absorbed through the skin. Prolonged lead exposure can cause neurological effects and muscle weakness. The excretion of lead may produce renal dysfunction, which is reversible, if slight lead toxicity exists. 6. CONCLUSIONS

It has been shown that the primary manganese dioxide-zinc system can be recharged to such an extent that a commercial version of a rechargeable battery can be obtained. This new rechargeable system can well compete with the nickel-cadmium and lead-acid battery system in about 70 percent of all consumer applications. Especially in all electronic devices which use at the present time the primary cell, the rechargeable version can be interchangeably inserted. Due to the low manufacturing cost a saving to the consumer in the order of 20 to 100 times over the use of throw-away primary cells is achieved. Special types which combine high power output (200 W/kg) with high capacity (100 Whkg) are under development in Canada at Environmental Battery Systems, Inc.. These future versions are the "Bundle Batteries", "Spirally Rolled Cells" and the "Bipolar RAM-Batteries". These types may even be used for electric

200

vehicle propulsion purposes. The rechargeable manganese dioxide-zinc system can now be produced without any mercury in the anode. Patented processes allow the corrosion hydrogen to recombine with the cathode and therefore no large amounts of inhibitors, which usually decrease the performance of batteries, need to be used. The new batteries contain no environmentally objectionable poisonous substances. A recycling is therefore unnecessary and it would also not produce any valuable substances on reprocessing. However, should any country enforce the principal recycling of all batteries, then all components of the spent rechargeable manganese dioxide-zinc batteries can easily be separated from the very poisonous components of other rechargeable cells. 7. ACKNOWLEDGEMENTS

The authors express their sincere thanks to the "Funds for the Support of Scientific Research in Austria" for financial help within the Main Program S-27 (Subtitle "Electrochemical Energy Storage and Conversion") and continuing projects. Thanks are also due to Dr. Klaus Tomantschger from Battery Technologies, Inc., 2480 Dunwin Dr., Mississauga, Ontario, and to Dr. Josef Daniel-Ivad from Environmental Battery Systems, 30 Pollard Str., Richmond Hill, Ont., Canada, L4B 1C3, for supplying recent technical test results. 8. REFERENCES

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Leuchs G. German Patent 1882; 24552. Herbert WS. J Electrochem Soc1952; 99: 190C. Kordesch K. Batteries, Vol. 1, Manganese Dioxide. New York: Marcel Dekker, 1974. Marsal PA, Kordesch K, Uny LF. US Patent 1960; 2,960, 558. Kang HY, Liang CC. Electrochimica Acta 1968; 13: 277-281. Uny LF. US Patent 1962; 3,024,297. Eveready Battery Applications Engineering Data, UCC. New York, 1971. Falk SU, Salkind AJ. Alkaline Storage Batteries. New York: John Wiley & Sons, 1969. Kordesch K, Gsellmann J, Chemelli R, Pen M, Tomantschger K. Electrochimica Acta 1981; 26: 1495-1504. Kordesch K, Gsellmann J. U s Patent 1983; 4, 384,029. Kordesch K, Harer W, Taucher W, Tomantschger W. 22nd IECEC. Philadelphia, Aug. 10-14, 1987; 1102-1107. Kordesch K, Gsellmann J, Harer W, Taucher W, Tomantschger K. Progress in Batteries & Solar Cells 1988; 7: 194-204. Taucher W. Docteral Dissertation. Technical University Graz, 1988. Kordesch K, Tomantschger K. US Patent 1990; 4, 925, 747. Kordesch K, Binder L. Patent application filed, 1991, Kordesch K, Binder L, Gsellmann J, Kahraman E, Winkler G. Int Power Sources Symp. BournemouthRJK. Power Sources 13. 1991; 273-284. HIBAR Systems Ltd. Richmond Hill, Ontario, Canada, L4B 1A8. Kordesch K, Ujj R. US Patent 1990; 4, 977, 364. Crompton TR. Small Batteries, Primary Cells. New York: John Wiley & Sons, 1983; 2. Latimer W. The Oxidation States of the Elements and their Potential in Aqueous

20 1

21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56

Solution. 2nd ed. Prentice-Hall, Englewood Cliffs. New York, 1952. Pourbaix M. Atlas of Electrochemical Equilibria in Aqueous Solution. New York: Pergamon, 1966. Euler J, Fleischer A. Primary and Secondary Cells - History and State-of-the-ArtSummary. 29th Meeting of AGARD, Propulsion and Energy Panel. Liege, Belgium, June 1967. Boden D, Venuto CJ, Wisler D, Wylie RB.J Electrochem SOC1967; 114: 415. DirkseTP, DeHaan F. J Electrochem SOC1958; 105: 311. von Borstel D, Spahrbier DH. Progress in Batteries & Battery Materials 1991; 10: 38-49. Daley JLS. Annual Power Sources Cod. 1961; 15: 96. Cheesman PG, Stock MG. In: Pearce LJ, ed. Power Sources 10. Int Power Sources Symp Comm 1985; 217. Haynes A, Binder L, Kordesch K. Progress in Batteries & Solar Cells 1989; 8: 39. Kordesch K, Binder L, Taucher W. Patent application filed, 1991. Taucher W, Binder L, Kordesch K. J Appl Electrochem 1992; 22: 95-98. Jammet JF,US Patent 1971; 3,556,861 Permion Battery Separators. RAI Research Corporation. Company Brochure 1990. Baugh LM, Tye FL, White NC. In: Thompson J, ed. Power Sources, 9. LondodNew York: Academic Press, 1983. Glaeser W. In: Kelly T, Baxter B, eds. Power Sources, 12. Int Power Sources Commitee, 1989. Eveready Battery Engineering Data. Union Carbide Corp. New York, 1982,II. Kordesch K. In: Linden D, ed. Handbook of Batteries and Fuel Cells. Chapt 7. New York: McGraw-Hill, 1984; 1- 1 7. Kordesch K, Gsellmann J, Tomantschger K. US Patent 1990; 4, 957, 827. Tomantschger K, Michalowski C. US Patent 1992; 5, 108, 852. Kordesch K, Gsellmann J, Tomantschger K. US Patent 1990; 4,957,827. Kordesch K, Taucher W, Daniel-had J. Patent application filed, 1991. Kordesch K, Tomantschger K. US Patent 1990; 4,925,747. Tomantschger K, Kordesch K, Oran E. US Patent 1991; 5,043,234. Tomantschger K, Kordesch K, Findlay RD. US Patent 1991; 5, 069, 988. Tomantschger K, Kordesch K. US Patent 1988; 234,933. Daniel-Ivad J. Doctoral Dissertation. Technical University Graz, 1991 Kordesch K, Kozawa A. US Patent 1976; 3, 945, 847. Daniel-Ivad J. Internal EBS-Report. Canada, 1992. Kordesch K, Gsellmann J, Tomantschger K. US Patent 1991; 5, 01 1, 752. Winkler G. Doctoral Dissertation. Technical University Graz, 1991. Binder L, Daniel-Ivad J, Kordesch K. ICHEME Symposium. Loughborough, England. April 7-9 1992; 127: 259-268. Mussnig RJ. Doctoral Dissertation. Technical University Graz, 1992. Kordesch K, Daniel-Ivad J, Kahraman E, Mussnig RJ,Toriser W. IECEC-Meeting. Boston, 4-8 Aug. 1991; 3: 463-468. Toriser W. GroRglockner-Austria Solar Race. May 1990 International Electrotechnical Commision, EC.1987; 86-1. Tomantschger K, Book J, Findlay RD, Ginther S, Goodman P. Internal BTI-Report. Canada, Dec. 1991. BTI-Brochure. The RAM Battery System. Canada, January 1992.

202 57 58 59 60 61 62 63 64

65 66 67 68 69 70 71 72 73 74 75 76

Tomantschger K. The Third International Rechargeable Battery Seminar. Deerfield Beach, Florida, March 5-7, 1990. Weinberg DB. Third International Seminar on Battery Waste Management. Deerfield Beach, Florida, November 4 - 6, 1991. Fricke J. Third International Seminar on Battery Waste Management. Deerfield Beach, Florida, November 4 - 6, 1991. Grach J, Terry BG. Third International Seminar on Battery Waste Management. Deerfield Beach, Florida, November 4 - 6, 1991. Stammbach MR, Rollor MA. Third International Seminar on Battery Waste Management. Deerfield Beach, Florida, November 4 - 6, 1991. Fiala-Goldiger J, Rollor MA, Hanulik J. Second International Seminar on Battery Waste Management. Deerfield Beach, Florida, November 5 - 7, 1990. Nriagu JO, ed. The Biochemistry of Mercury in the Environment. Amsterdam-New York-Oxford: Elsevier-North Holland Biomedical Press, 1979. Greenwood MR. Queckilber. In: Merian E, ed. Metalle in der Umwelt. Weinheim: Chemie, 1984; 5 1 1-539. Daunderer M, Simon M. In: Elvers B, et al, ed. Ullmann's Encyclopedia of Industrial Chemistry, 5. Completely Revised Edition. Weinheim: VCH, 1990. Friberg L. Acta Med Scand Suppl 1950; 138: 240. Piscator M. In: Gerhartz W, ed. Ullmann's Encyclopedia of Industrial Chemistry, 5. Completely Revised Edition. Weinheim: VCH, 1985. Flanagan RR,McLellan JS, Haist J, Cherian G, et al. Gastroenterology 1978; 71: 841-846. Kagi Nordberg M, eds. Metallothionein. Basel: Birkhauser, 1979. Friberg L, Piscator M, Nordberg GF, Kjellstrom T. Cadmium in the Environment, 2nd Ed. Chemical Rubber Co. Cleveland, 1974. World Health Organisation. Recommended Health-based Limits in Occupational Exposure to Heavy Metals. Techn Rep Ser 1980; 647. Worth D, et al. Environmental Lead. New York: Academic Press, 1981. Lacey RF, Moore MR, Richards WN. Sci Total Environ 1985; 41: 235-257. Teindl H. In: Elvers B, et al. Ullmann's Encyclopedia of Industrial Chemistry, 5. Completely Revised Edition. Weinheim: VCH, 1990. Internat Symp on Environ Lead Research. Dubrovnik, Yugoslavia. Arch Ind Hyg Tox 1975: 26. Supplement. Lead, Airborne Lead in Perspective. National Acedemy of Sciences, National Research Council. Washington DC, 1972.

203

Electrochemical generators and the environment. Fuel cells and metauair batteries. P. S. D. Brito and C. A. C. Sequeira Instituto Superior TBcnico Av. Rovisco Pais, 1096 Lisboa codex Portugal

INTRODUCTION The increase of atmospheric pollution may provoke irreparable damage to the Earth, particularly the emissions of gases that are related to the greenhouse effect and oxides of nitrogen and sulphur. The production of energy to sustain the human activities, is one of the greatest responsible for this fact. Ninety per cent of world energy is provided by burning fossil fuels and the consequent result is to add more carbon dioxide to the already burdened atmosphere and so increase global warming with direct impact on the climate. To base on the global climatic models, we can estimate that Earth will suffer a rise in the average surface temperature between 1.5"C and 4.5% if the concentration of greenhouse gases duplicate; moreover, if the actual tendency does not change, this temperature increase will occur before the year 2050. These alterations would have a bad impact on human activity [l]. The combustion of fossil fuels for electricity generation emits also oxides of nitrogen and sulphur which give rise to the acid rain, contributing to the dying forests, sterile lakes and corroded buildings 121. The necessity of reducing the gaseous products of the combustion lead to the need for alternative technology, concerning the conservation of energy resources and more efficient burning of petroleum and coal, the use of alternative fuels with lower or no emissions of greenhouse gases, the use of satisfactory renewable energy sources, and so on. Fuel cells and metallair batteries are a set of energy generators which appear to hold considerable potential for future efficient, and cleaner, generation of electricity. These systems are presently being developed for stationary power plants and for transport applications. The main advantage of fuel cells is to produce electricity and heat with negligible emissions of nitrogen oxides, carbon monoxide and dioxide, and others airborne pollutants. A study of the World Fuel Cell Council realized by Arthur D Little in 1991 indicates that fuel cells could reduce energy use and associated carbon dioxide emissions by 40-60% in those sectors that use over 90% of energy, and the emissions of harmful oxides of nitrogen would be reduced by 50-90% compared even to conventional technologies that employ pollution control equipment in a wide range of both stationary and transportation applications. A further advantage of fuel cells and metal/air batteries over all other methods of energy conversion, is that chemical energy is converted directly into electrical energy. The conventional methods of energy conversion, apart

204

from hydroelectric, involve the intermediate conversion of heat to mechanical energy, the efficiency of which is Carnot limited, thus the efficiencies of fuel cells are higher than those of other energy conversion methods (table 1) [3]. Table 1 Efficiencies of energy conversion devices. Energy Generator Thermoelectric Thermionic Photovoltaic Magnetohydrodynamics Fuel cells Gas turbines Internal combustion engines

Efficiency

(%I

10 22 18

60

el

30 30

Other advantages of fuel cells are the modular construction. Cells can be made in any shape or size, which allows them to be used for a wide range of applications, adaptability to operate over any desired temperature range and simple operation compared to other energy conversion devices. In this chapter we present an overview of the state of the art and short-term perspectives of fuel cells and metal/air batteries and analyse their environmental benefits and drawbacks. Greenhouse&ect The climatic conditions of the Earth are determined, in large scale, by the atmosphere compositions. The Earth's atmosphere acts as a filter that controls the exchange of energy between the sun, the Earth and space. Certain gases in the atmosphere, like carbon dioxide, methane, nitrous oxide, tropospheric ozone and chlorofluorocarbons (CFC), can act as greenhouse gases, by absorbing infrared radiation and retaining heat, thus contributing to the calculated global mean surface warming. The main greenhouse gas is carbon dioxide, which is responsible for 49% of the effect. The emissions result chiefly of burning of fossil fuels in the electricity generation, transports and industry (5 Gton of C per year), and deforestation (0.5 to 2 Gton of C per year) [ll.Figure 1 shows the percentage contribution to carbon dioxide emissions from human activities between 198085 [41. Between 1960 and 1985, the atmosphere concentration of carbon dioxide increase from 315 ppm to 345 ppm. Measurements of carbon dioxide in bubbles of air trapped in the Antarctic ice have shown that for 10 000 years the concentration remained constant at 270 ppm and only started rising 100 years ago. Actually the rate of increase of the concentration is 1.5 ppm per year [ll.

205

Figure 2 gives an evaluation of the carbon dioxide emissions in Portugal up to the year 2010, with basis on a median energy demand scenery [5].

Desflorestation Electricity Transport Industry Others

Figure 1. Percentage contributions to CO, emissions from human activities (1980-85).

1990

2OOo Year

2010

Figure 2. Carbon dioxide emission evaluations in Portugal (1990-2010).

206

Methane arises from agricultural animals, paddy fields and oil and gas wells. Recent observation shows clearly that methane emissions have been rising since the beginning of the last century, with a more rapid rise in the last few decades. This is partly because changing agricultural practices, waste disposal and mining have produced significant amounts of methane. In 1983 the concentration of methane in the atmosphere was of 1.68 ppm and is increasing by 1.1%per year since 1951 [ll. The Chlorofluorocarbons are chemical products with application in air refrigeration, refrigerators, solvents, packings and expulsants in aerosols. In 1985 the atmosphere concentration of CFC-11 and CFC-12 in four different places in the planet was respectively of 0.21 to 0.32 ppb and of 0.37 to 0.39 ppb. Although its concentration is lower than that of carbon dioxide, CFC is responsible for 14% of the actual greenhouse effect because its retention efficiency is 10 000 times greater. The total elimination of CFC can be expected by the year 2000 by application of restriction to the production, reciclage and destruction of CFC in the existing products [l]. The emissions of nitrous oxide are resulting chiefly from burning of fossil fuels and biofuels. In 1985 the concentration in the atmosphere was of 300 ppb. Actually the rate of increase of the concentration of nitrous oxide is 0.25% per year 111. Figure 3 shows greenhouse gases contributions to global warming of the Earth 141.

14

Carbon dioxide E 4 Methane 0 CFC-11& CFC-12 0 Nitrous oxide El Others

Figure 3. Greenhouse gas contributions to global warming.

Oxides of nitrogen and sulphur

The emissions of nitrogen oxides, NO,, arise from two sources, the oxidation of nitrogen in the combustion air and the oxidation of the nitrogen contained in the fuel. Sulphur oxide, SO,, arises from oxidation of the sulphur in the fuel

207

and all the sulphur is normally converted to the oxide. Sulphur occurs naturally in coal and oil. The principals transmitters are refineries, chemical product factories, termoelectric centrals and automobiles (solely NO,). As a n example, a 1000 MW oil termoelectric central emit 7 ton of SOz, 1 ton of NO,, and 0.1 ton of particles per hour, and a n automobile circulating a t 90 Km/h emit, per Km, approximately 4 g of CO, 3 g of NO, and 0.7 g of hydrocarbons 141. Table 2 presents the sectorial emissions of SOz and NO, in Portugal, during 1987 [51. Table 2 Sectorial emissions of SO, and NO, in Portugal (1987).

NO,

SO2

Sector Combustion: Electricity production Oil refinery Industry combustion Production processes Transports TOTAL

ton

%

ton

%

83 877

33 7

15834 2343 13534 13402 70 748 115861

14 2 12 11 61 100

14318 91 823 19815 8 489 218322

42 9 4 100

The industry combustion sector, that includes the combustion installations of the chemical industry, paper paste, cork, siderurgy and various units of small and median dimensions, and the energy production sector, set almost exclusively by termoelectric centrals, are the greatest responsible for the total emissions of sulphur oxides; their contributions are respectively 42% and 38%. The main contributor to the total emission of nitrogen oxides is the transport sector, with 61%. There are strong indications that sulphur dioxide (SO,) emissions increased by about a factor of two between 1950 and 1970 in most of Europe and the same general trend also holds for nitrogen oxides (NO,). The emissions of sulphur dioxide have generally levelled off since 1970, while the nitrogen oxides increase with the number of motor vehicles. Figure 4 gives an evaluation of the oxides of nitrogen and sulphur emissions in Portugal, until the year 2000 [6]. As we can see, the tendency is to increase the global emissions. This is due, by one side, to the prevision that energy search and industrial production will rise, and by other side, t o the enlargement of the automobile park, in spite of the larger restrictions to the vehicle emissions. The reduction observed for the years 1986-1989 was due mainly to the utilization of coal with less sulphur contents, in addition to the reduced use of fuel oil in the thermoelectric centrals (responsible of 30%).

208

320

280 240

s2O0 160

w

m 80

40 0

1985

1988

1991 1994 Year

1997

m

Figure 4. Emissions of nitrogen and sulphur oxides in Portugal (1985-2000).

FUELCELZS Fuel cells (FCs) are electrochemical devices that directly convert fuel energy into electricity without the need for a thermal cycle. They are essentially galvanic cells in which the electrodes only collect and convey electrical charges, but (unlike in the Volta pile and all other electric cells and batteries) they do not participate in the electrochemical reaction, since they are chemically and electrochemically inert conductors (amorphous carbon, sintered nickel oxide, etc.). In a FC, such electrodes are porous t o allow fluid reactants t o be continuously adducted t o the reaction zones: the contact surface between the anode and the electrolyte for the fuel, and that between the cathode and the electrolyte for the oxidant. In principle, all those exothermic reactions between fluid reactants that can be subdivided into two electrochemical semireactions could be used in a FC. In practice, only the oxidation reactions of Hzto form HzO and of CO to form COz are considered, because the reaction products a t the FC working temperature are non-aggressive and gaseous: they, therefore, involve no problems in the choice of structural materials, and are easily eliminated. If we take the H,oxidation as an example, the theoretical cell voltage is:

209

where AG is the Gibbs free energy of the reaction, expressed in joules, n is the number of electrons participating in the reaction, per molecule, and F is the Faraday number. In practice, polarisation effects reduce the voltage available at the electrodes to nearly half of its theoretical value: V,=0.71 V/cell is the open-circuit voltage, V=0.65 VJcell the payload voltage with current density J=200 d c m 2 . To realise useful voltages it is therefore necessary to pile up several cells (as in any electrical battery): this is the fuel-cell stack. Electrode surfaces of about one square foot, with current intensities in the range of 200 A are very common when using the present technology, and stacks of some hundreds of cells are also commonly used. However, hydrogen and carbon monoxide are not freely available in nature, i. e. they are not "primary fuels". A pre-fuel reformer must therefore convert any commercial fuel (gasoline, LPG, methane, naphtha, methanol, etc.) into a hydrogen-rich and/or carbon-monoxide-rich gaseous mixture to feed the fuelcell stack. Moreover, since fuel cells are electrochemical devices, they produce direct current: an invertor must therefore be set after the stack, when a.c. current is needed. The resulting block-scheme of a complete FC generator is reported in Figure 5. The invertor subsystem is obviously not necessary for transportation applications.

c

r

P

C

CP

c

cc

i

ca

cr

Figure 5 . Block scheme of a FC generator set: c, commercial fuel in; r, reformer; H2, hydrogen rich reforming gas; H20, water vapour; p, fuel-cell stack; cc, direct current output; i, invertor; ca, alternating current output; cp, process heat; cr, recoverable heat.

Advautagesanddrawbacks The main advantages of FC generators are: high electrical-conversion efficiency, modularity and very low environmental impact. Among the devices that produce electricity using fossil fuels, FCs are the only ones that produce no NO, since this mixture only forms well above 1000°C, as in burner flames, but not in FCs, which all operate below this temperature. FCs obviously produce the same stoichiometric amount of COP as conventional generators do per amount of fuel used. However, since their

210

conversion efficiencyis higher, the rate of C02 production per kwh produced is lower. FC exhausts do not contain SOz, since FCs do not accept sulphurated fuels. Finally, FCs produce negligible noise since they have neither moving parts nor intermittent irreversible reactions (like combustion or explosion). They can have high power-to-weight ratios (up t o 0.7 kW/Kg) and power density (up to 1kWAitre). So simple is the principle and so many are the advantages, that one might wonder why FCs are not yet commonly adopted. Indeed, they were conceived by Sir Humphrey Davy in 1801, only one year after the invention of the Volta pile, and first realised by William Grove in 1839, as a laboratory prototype. From then on, a century-and-a-half long development took place to bring that prototype to the threshold of industrialisation. The reason is that the kinetics of the oxidation and reduction semireactions is unfavourable at room temperature, and this hindrance can be overcome either by increasing the operation temperature (high-temperature FCs, 600-950°C) or by adding suitable catalysts to the reaction zone (low-temperatureFCs, 60-300°C). In the first case materials stability and in the second case catalyst lifetime, and in both cases current density, have raised century-lasting problems that could be solved only recently, with very mature technologies. One further consideration may be added. FCs are typical high-added-value, low-material-cost devices (in spite of precious catalysts, which only cover 0.5 per cent of the cost of the present commercial prototypes, and 5 per cent of the final expected asymptotic value for mass-produced FCs). So high is the added value, that only large-scale mass-production would make FC generators competitive, at least until environmental impact was not considered as a bottle-neck, and economy and performances were the only comparison grounds for powering devices. Three facts have recently reversed this situation: first, improved technologies raises FC expected lifetimes to 40 000 hours and in some cases, even to 60 000 hours, which is more than competitive for many applications: second, recent results with new proton-conducting membranes [61 allow very high current densities (up t o 4 A/cmZ) to be reached: third , i. c. engine pollution seems t o raise hardly solvable problems for downtown transportation. Taking into account that the automobile industry is a typical massproduction industry, it is deemed worthwhile to draw the attention of automobile designers to the FC technology and performances.

Alkaline FCs Alkaline FCs (AFCs) use KOH as electrolyte and work at 70-90°C: they are fully developed and very reliable (they powered on-board instrumentation of the Apollo spacecrafts and they power on-board instrumentation of the space Shuttles). Electrodes are mostly sintered nickel (anode) and sintered, lithiated nickel-oxide (cathode). Unfortunately, KOH electrolyte does not bear traces of carbonaceous gases (CO and CO,) in the reactants even at percentages below 0.1 per cent, which rules out not only commercial-fuel-reforming-gasas a stack-fuel, but also air as an oxidant. Higher CO or CO, percentages bring about formation of

21 1

insoluble, viscous and less-conducting carbonates or bicarbonates, which would obstruct the pores of the electrodes, thus determining their rapid death PI. For these reasons, AFCs are excellent for spacecraft, where weight limitations anyway impose pure H, and pure 0, as reactants, but they can be envisaged for ground transportation only in those countries where public opinion would accept pure hydrogen and oxygen on board terrestrial vehicles. As a matter of fact, Elenco, in Belgium, has been successfully testing an AFCpowered VW van for several years, and AFCs have also been proposed to substitute i. c. engines in the German project for two thousand hydrogenfuelled buses. It is also known that a large international project is being developed within CEC for transatlantic transportation (in methylcyclohexane form) of hydrogen produced electrolytically in Canada (exploiting the immense reservoir of unused hydroelectric power of that country), to be subsequently pipelined in several European Countries. Such a vigorous implementation of the hydrogen economy would certainly be beneficial to AF'Cs for standing generators and possibly even for transportation, although some countries, Italy as a n example, might get ready to accept hydrogen instead of methane for fixed installations well before hydrogen and oxygen on board terrestrial vehicles.

Phosphoric acid FCs Phosphoric acid FCs (PAFCs) are the most extensively tested ones, at least for fixed installations: about fifty 40-kW PAFC units, working at about 200"C, have powered a variety of users, including an automatic laundry, restaurants, swimming pools, a telephone-switching central office, kindergartens, etc., in the United States, Canada and Japan for about three years; two 5-MW demonstrators have been realised, namely one in downtown Manhattam, one in Tokyo, and a 11-MW plant is now being implemented for the Tokyo Electric Power Company; a 1-MW plant of US technology is under construction at AEM (the Milan Municipal Electricity and Gas Utility) and another one, of Japanese technology, has been in operation for years now in Sakaiko a t the Kansai Electric Power Company. A 2.5-KW demonstration and training unit has been in use at CISE, in Milan, for the last seven years 181. PAFCs can accept air at the cathode and reforming-gas of any commercial fuel at the anode: in particular, methanol reformers, working at about 240"C, are simple, compact, suitable for transportation. PAFCs can be water-cooled, air cooled or organic-liquid-cooled. All three concepts have been considered for ground transportation. In the early 1980s the Engelhard Corporation designed a fork-lift powered by an organic-liquid-cooled PAFC. This programme was discontinued, and all rights were transferred t o Fuji, who also independently developed a watercooled PAF'C and finally realised a PAF'C-powered fork-lift truck. More recently, the US Department of Energy and Department of Transport launched a programme for PAFC-powered minibuses, and tenders were submitted for water-cooled and air-cooled PAFC powering. A preliminary study of a PAFC-powered minibus was presented at ISM-8, Tokyo 1988.

212

Solid polymer electrolyte FCs Solid polymer electrolyte FCs (SPEFCs) were developed last, but, paradoxically, they were the first to have a "commercial" application: they powered the on-board instrumentation of the Gemini spacecraft and still power that of the biosatellite. Their acid electrolyte is micro-encapsulated or laminated onto a proton-exchange membrane. The working temperature is a little above r. t. (60-90°C)171. The development of new membranes by Dupont and Dew Chemicals recently granted a very sharp increase in current density, that is, from the "conventional" 200 mA/cm2 to 4 A/cmz. This success obviously reflects onto power density and power-to-weight ratio, and has recently allowed a twoperson submarine to be equipped, which can be visited in operation in Florida. With this power source, the range of the submarine was increased by a factor 11. It is evident that such performances call for immediate attention in the field of ground transportation too. In Italy, to mention but one, a collaboration has been established between De Nora (which has a prototype of SPEFC and plans to develop a 10-kW stack) and IVECO (the industrial vehicle concern of the FIAT Group), under the sponsorship of ENEA (the Italian analogous of DOE) within the "Volta project" for the development of FC technology. One major difficulty has yet to be solved before SPEFCs can gain interest for transportation. Although SPEFCs accept minor percentages of CO, (i. e. they can use air as oxidant), their limitations about CO content in the reactants are very severe: to purify reforming gas from CO at the required level, reformers are now so big that they cannot be installed on board commercial vehicles. To give an idea of the importance of this problem a 3 M$ programme, to develop a more compact methanol reformer for SPEFC, has been established in Canada.

Solidoxide FCs Solid oxide fuel cells, operating at 800-1000"C, offer an attractive, albeit longer term prospect of cells capable of converting hydrocarbon fuels to hydrogen in situ, avoiding the expense and complexity of reformers. If the high temperature reject heat can also be used to generate steam for a "bottoming cycle" turbine, overall efficiencies of up to 60% are anticipated. These cells use stabilised zirconia as the electrolyte, relying on solid state diffusion of oxygen for conductivity, thus imposing a lower limit to the temperature of operation. The largest unit constructed to date (3 kW) has been built by Westinghouse in the USA, using a tubular format, and this has been evaluated by the Tokyo Gas Company. Dornier, and Brown Boveri (Switzerland) have also developed tubular structures of calcia o r yttria stabilised zirconia electrolyte, the latter proposing to work on fuel cells as a development of their high temperature "Membrel" electrolyser technology. A recent development is the concept, employing similar materials, fabricated into a lightweight honeycomb structure of small cells, t o give a high surface area (and hence increased energy density) per unit volume. One version of this, from Argonne National Laboratories, is built up from flat and corrugated plates. A similar concept is being evolved at Imperial College, London, as part of an EEC sponsored programme 191.

213

Moltencarbonate Fcs Work on molten carbonate electrolyte fuel cells in Europe is mainly concentrated in Italy and The Netherlands. These programmes have been set up with the aid of existing molten carbonate fuel cell developers in the USA and Japan, to make maximum use of available technology. The Dutch programme has led to the construction of a 1kW unit during 1989, with plans for 10 kW plants to be operational during 1993. Much of the research work has been carried out at ECN, Petten, with contributions from several Dutch universities and a manufacturing company, Hoogovens. Technology has been licensed from the Institute of Gas Technology in the USA to establish a sound basis for further development. The EEC has sponsored a programme to draw together the expertise in materials throughout Europe, with the intention of overcoming problems such as corrosion and electrolyte migration. The Italian effort was mainly coordinated by ENEA, with a number of universities, CISE, CNR-TAE and Ansaldo participating to develop 10 kW units with their associated hydrogen generator plant by 1993 [lo].

MetaVair cells are a semi-fuel cell, a hybrid between a battery and a fuel cell. The system combine in situ a metal anode, which dissolve in an alkaline o r neutral electrolyte, and a porous diffusion cathode, in which air is continuously admitted. These systems can be electrical o r mechanical recharged. Electrical recharging can be performed either in the battery itself or in a special recharging o r electrolysis cell. Mechanical recharging consists of the replacement of the metal anode plate. If the battery is mechanical recharged, it will be necessary to remove the reaction products by recirculation of the electrolyte or by replacement periodically of the electrolyte. The concept of the metauair battery is shown schematically in figure 6 together with the overall cell reaction. The first metallair cell mentioned in the literature was a zinc/air cell constructed by Smee in 1840 [ll].In 1878, Maich [12], modified a Leclanchb cell by replacing the conventional manganese dioxide cathode by a mixture of platinum and carbon powder. Ever since the work of Maich, various metals have been studied for use in this type of system in different applications. Among others, zinc, iron and aluminium have been considered t o electric vehicles applications. It is, however, in the field of electric vehicles that most of the research has been conducted.

Advantagesanddrawbacks The main characteristics of electrochemical power sources that must be considered in electric vehicles applications are specific energy density, specific power and cost. The main advantage of metal/air batteries is primarily their high specific energy in comparison with other types of batteries. This is achieved through not requiring to incorporate positive active components within the cell (table 3).

214

n

- Me Z

+ nOH- +

n If 0, + ij H,O 4

Me(OH), + ne-

+ ne- + nOH

Figure 6.MetaYair cell.

Table 3 Metal/air systems. Metal Li A1 Mg Zn Fe

Electrochemical equivalent (Ahlg)

Theoretical cell voltage

3.86 2.98 2.20 0.82 0.96

3.4 2.7 3.1 1.6 1.3

(V)

Theoretical specific energy (Wh/g)

13.1 8.1 6.8 1.3 1.2

Operating voltage

(V) 2.4 1.6 1.4 1.2 1.o

The operational costs of metal/air batteries are low. Air is costless and inexhaustible. The metals that have been used, such as aluminium, zinc, iron, are cheap and abundants. The major limitation of metaYair batteries, that have prevented their more wide-spread applications, are their low specific power. This limitation arises from the air performance electrode, which suffers from considerable polarization losses on discharge, largely due to mass transport limitations. However, advances have been made in reducing polarization by improving the physical electrode properties.

215

The metal/air systems produce electricity without emissions of airborne pollutants. Thus, the use of large population of metallair electric vehicles may greatly reduce the air pollution and associated health impacts in urban areas. However, even though metal/air electric vehicles do not themselves emit exhaust gases, their running will incur emission increases from metal industries. The main pollutants of aluminium industry are fluoride, sulphur dioxide, carbon dioxide and carbon monoxide, any of them can be reduced with control technologies. In the zinc industry the pollutants are chiefly sulphide dioxide, carbon dioxide, carbon monoxide and zinc vapour (in pyrometallurgy). However, with electrolytic zinc production, with modern roasting equipment, fume emissions are minimized and a rich sulphur dioxide stream is produced, suitable for acid. More than three-quarters of the World’s primary zinc is currently being produced by this process. On other hand, the air emissions for the metal production plants are principally due to fossil-fuel generated electricity, this emissions can be reduced by using alternatives electricity generators, such as, fuel-cells. Consequently, it is important to the implementation of metal/air electric vehicles, that metal production plants develop their technology to reduce global impact on the environment.

owm-

High performance and low cost oxygen (air) cathodes are of vital importance in metallair batteries concerning systems for automotive applications. Energy and power-losses, mainly due t o the high cathodic polarization, expensive catalysts are the main problems of oxygen cathode for metallair batteries. Removal of carbon dioxide from the air is an important prerequisite for prolonged operation of all alkaline metallair cells, because in alkaline solutions the carbon dioxide precipitate the carbonate which could block the air electrode. Development of operational air electrodes encounters two main problems: design of a structure of the gas-diffusion electrode and selection of an effective and stable catalyst for the specific operational conditions. The electrode reaction can only occur in the region where solid, liquid and gaseous phases come together. Thus the construction of oxygen electrodes for practical cells must maximise the interfaces between them. This may be achieved using support material with high porosity, such as carbon, and dispersing on its surface the active catalyst. Most air cathodes for metal/air batteries are of semi-hydrophobic type. The semi-hydrophobic electrode has two layers: a hidrophilic electrolyte-layer, which is wetted by the electrolyte, and a strongly hydrophobic gas-side layer, which is not wetted by the electrolyte at all. The hydrophobic gas-side layer is impregnated with a waterrepellent coating, such as, polytetrafluorethilene (PTFE). Carbon has been used as a support material for cathodes because of its large surface area, high electric conductivity, large availability and low price. Besides, carbon materials have also catalytic activity for oxygen reduction. Active carbon materials, which have high content of mineral traces and large surface area, and different types of carbon blacks (acetylene blacks, furnace blacks) are often applied.

216

Noble metals applied as electrocatalysts for the oxygen reduction have been largely utilized because of their high electrocatalytic activity and stability. Investigations are concentrated on platinum, palladium, silver and gold. The application of noble metal catalysts is limited by two fundamental disadvantages: high cost and low availability. Thus, it is important t o construct cathodes with small amounts of the noble metal which are obtained, for example, by dispersed platinum on an appropriate support. Metal chelates, sulphides and thiospinels have been examined for their usefulness as oxygen electrodes in metauair cells because they show sufficient stability and activity in alkaline medium. The three macrocyclic complexes, tetraazaannulene (MeTAA), tetraphenylporphyrin (MeTPhP) a n d phthalocyanin (MePc), have showed relatively high activities for oxygen reduction. Their activity is influenced by the central metal ion, Fez+and Co2+ are the metal ions which give the highest catalytic activity 1131. Other group of oxygen catalytic materials that has shown good electrocatalytic properties and stability are some inorganic semiconductors. In alkaline media a number of spinels (CozNi04, Co304), perovskites (lanthancobaltites, e. g. Lao.*Sro,zCoO,) doped with metals of the second main group (esp. strontium), have shown good results [131. A number of research groups have been engaged in the research and development of alkaline air (oxygen) electrodes. Among these are ELTECH Systems Corporation, Westinghouse Electric Corporation, International Nickel Corporation, Gould Inc., Union Carbide Corporation, Giner Inc.,Prototech Corporation and Case-Western University in the United States; Siemens A. G., Accumulatorenwerk Hoppecke and University of Bonn in Germany; Compagnie GBn6rale d'Electricit6 in France; Matsushita in Japan; Polish Academy of Science; Bulgaria Academy of Science; and University of Beograd in Yugoslavia.

Zindaircell Zinc is the most electropositive element that can be electrochemically deposited in aqueous solution. Consequently, the zindair cell has the highest voltage among electrically rechargeable metal/air batteries. It is the combination of this fact with zinc low equivalent weight, large terrestrial abundance and low cost, that has stimulated significant research and development activity on zindair batteries for electric vehicle applications [14]. The problems of this system are the low specific power and efficiency, related with oxygen electrode, and short life, due t o the change of zinc morphology during the cell charge [15]. Gulf General Atomic Company was one of the first parties to develop a zindair and zindoxygen system for an electric vehicle. A 20 kW zindoxygen battery was tested in a mini-moke jeep. The system employed electrolyte circulation and continuous removal of Zn(OH), reaction product. Zinc electrodes were regenerated from spent alkaline zincate solution in external recharging cells 1161. In the 1970s, a mechanically rechargeable 35 kWh zinc/air battery was demonstrated by General Motors in a 1350 Kg test bed vehicle, but the development was not continued to commercialisation 1171.

217

Sanyo systems make use of forced circulation of air and electrolyte. A 124 V, 560 Ah traction battery was tested in large vehicles 1181. The most promising system has resulted from the work of Compagnie GBn6rale d'Electricit6 in France and shown in figure 7. Here a slurry containing zinc particles flows through tubular cells whose outer surface comprises the air electrode and for which the negative current collector is a n axial brass rod. Zinc particle regeneration occurs in a special component external to the discharge cell modules [191.

electrolysis cell (zinc regeneration)

1-'

reservoir2

1

Figure 7. Zindair system with a slurry zinc anode.

hdaircell The irodair cell is especially attractive as it can utilize resources that are virtually inexhaustible. Only electrically rechargeable batteries have been developed owing to the good reversibility of the iron electrode and cycling behaviour . A shortcoming of the system is the high ratio between the charge voltage and the discharge voltage resulting in a low energy efficiency. The charge voltage is about 1.75 V and the discharge voltage is 1.0 V. In order to minimize the hydrogen evolution corrosion reaction that occurs at the open circuit potential and on the charge, which lowers the energy and current efficiency, electrolyte additives are used in order to increase the hydrogen overpotential. In particular, sulphide ions have favourable effects.

218

The cell design employs iron electrode which is located in the centre of the cell faced with two air electrodes. The air electrodes are bifunctional, serving both for oxygen reduction and evolution [151. The development of irodair batteries was initiated at the Swedish National Development Company in 1969. Since then other groups and companies have been developing irodair systems. Among these are the Indian Institute of Sience, Matsushita in Japan, Westinghouse Electric Corporation in United States and Siemens A. G. in Germany.

Aluminidair cell The high specific energy and voltage, and the feature of mechanical rechargeability of aluminiudair cell, as well as high abundance and low cost of aluminium, lead to the viability and attractiveness of the a l u m i n i d a i r cell for electric vehicles application. Regardless of these advantages, aluminium presents some difficulties which arise from the formation of a passivating and resistive oxide layer when it comes into contact with water containing electrolytes. This results in a much less negative rest potential compared with the thermodynamic one (-1.9 V us. M E ) . When the oxide is removed, by the action of some aggressive ions such as chloride and hydroxide, a corrosion reaction takes place and hydrogen evolution occurs. This effect is more pronounced in alkaline hydroxide solutions because the rest potential of aluminium is made more negative. The inhibition of hydrogen evolution may be achieved by modifying the composition of the anode material (instead of pure aluminium using aluminium alloys with gallium, indium and thallium), or by adding specific inhibitors to the anolyte [201, Two types of aluminiudair batteries are being developed oriented toward different applications: alkaline batteries for electric vehicles, due to favourable specific energy, and saline batteries for special applications, such as, emergency lighting, reserve power, marine objects and so on. The use of aluminium metal/air batteries was attempted by S. Zaromb in 1962 [23]. He proposed a battery similar to a fuel-cell. The Norwegian Defense Research Establishment was the first party to develop an aluminiudair battery. In 1980 they built a mechanical rechargeable 10 cell battery that delivered 40 W at 12 V. The battery was based on an electrolyte mixture containing KCl and KF with citrate and stanate as additives. The unit consists of a cell stack, an electrolyte container and an electrolyte pump. In addition, it has five hydrocyclones t o continually remove the finely dispersed reaction product and a heat exchanger. The aluminiudair battery project of the Norwegian Defense Research Establishment continues with the collaboration of industry, of the Norwegian Research Council and the Foundation of Scientific and Industrial Research at the Norwegian Institute of Technology [211. In Germany, Hoppecke has developed and built several prototypes of a l u m i n i d a i r batteries with alkaline electrolytes and cylindrical aluminium anodes. The rods of aluminium are continually fed through gaskets at a rate which is identical to the rate of their consumption 1151.

219

The Lawrence Livermore National Laboratories, as the prime contractor for the United States Department of Energy, has been working in aluminiudair battery programs with the collaboration of sub-contractors, such as, Electromedia Corporation, Case Western Reserve University for the air cathode development; Alcan and University of Akron for the aluminiudair batteryhystem engineering; SRI International and Oklahoma State University for the aluminium anode development; and ELTECH Systems Corporation for performance of aluminium alloys, air cathodes and all the cited previous areas of development. Various cells have been constructed with different approaches: flat or wedge aluminium anodes; anodes and cathodes movables with flat plate configuration [221. In figure 8 it is shown a scheme of a wedge approach of an aluminiudair cell. Neutral aluminiudair cells have been developed at the Serbian Academy of Science and Arts in the Yugoslavia and Bulgaria Academy of Science in Bulgaria. A 24 W, 10-cell saline or sea water was constructed with galliumtin-magnesium aluminium alloy anode [231.

electrolyte out

electrolyte out air in

air in

air out

air out

t i

electrolyte in Figure 8. Schematic representation of an a l u m i n i d a i r wedge type cell.

CONCLUSION Industrial processes saving primary energy, and clean and efficient energy conversion systems will become essential in the near future. Electrochemical processes can theoretically be conducted with no energy losses, and therefore

220

with no heat pollution, and also with little greenhouse gas evolution in comparison with processes involving combustion reactions. Furthermore, the electrical energy involved in such processes, as upstream energy for electrolysis processes, or as stored or produced energy for electrochemical batteries and fuel cells, can be readily transported and efficiently converted to mechanical energy, for example in the case of electric vehicles. Fuel cells and metauair batteries offer an attractive and viable alternative source of electric energy, both for large electric energy production and for transportation applications. The systems are not aggressive to the environment and are also, very flexible, quiet and have high efficiency. Significant advances have been made in many parties and companies around the World to develop more efficient and low cost metallair systems, but much work is necessary before a large population of electric vehicles of this type of batteries takes place. The development of a metal industry less pollutant is also a requisite to obtain a more efficient global process and a clean source of energy for electric vehicles.

ACKNOWLJ3DGEMENT The authors wish to express their thanks to JNICT, Portugal, for financial support (BD/899/90-RM).

1 CE Document, COM(88)656/2, "The Problem of Greenhouse Effect and the

Community", 1989. 2 I. Fells, "Energy and the Environment", Ed. J. Dunderdale, Royal Society of Chemistry, G. B., 1990; 1-12. 3 B. V. Tilak, R. S. Yeo, and S. Srinivassan, "Comprehensive Treatise of Electrochemistry: Vol. 3 - Electrochemical Energy Conversion and Storage", Ed. J. O'M. Bockris, Brian E. Conway, Ernest Yeager and Ralph E. White , Plenum Press, New York, 1981; 39-122. 4 UNEP Industry and Environment, April-May-June 1990; 2-8. 5 A. P. Carneiro and G. C. Neves, "Inventfirio das Emissijes de Poluentes Atmosfdricos: Uma Metodologia", Direc@o Geral da Qualidade do Ambiente, Portugal, 1990. 6 M.A. 0.V. G. ColaGo, P. M. A. S. Miguel, P. A. L. R. Santos and C. A. C. Sequeira, "Chemistry and Energy I", Ed. C.A.C. Sequeira, Elsevier, Amsterdam, 1991; 69-84. 7 A. J. Appleby and F. R. Foulkes "Fuel Cell Handbook", Ed. Van Nostrand Reinhold, New York, 1989; 261. 8 H. Van den Broeck, Fuel Cell Seminar Abstracts, National Fuel Cell Coordinating Group, Long Beach, California, 23-26/10/1988;375. 9 B. C. H.Steele, presentation at "Fuel Cell Technology and Applications: International Seminar", The Netherlands, 1987. 10 R. Vellone, presentation a t "Eureka Industrial Fuel Cells, Forum", London, 1987.

22 1

11A. Smee, Phil. Mag. 111, 16, 1840; 315. 12 L. Maiche, French Pat. 127,069, 1878. 13 K. Wiesener and D. Ohms, "Electrochemical Hydrogen Technologies: Cap. 3 - Electrocatalysis of the Cathodic Oxygen Reduction", Ed. H. Wendt, Elsevier, Amsterdam, 1990; 63-103. 14 E. L. Littauer and J. F. Cooper, "Handbook of Batteries and Fuel Cells: Cap. 30 - Metal/air Batteries", Ed. D. Linden, McGraw-Hill, New York, 1984; 30.1-30.21. 15 F. G. Will, "Studies in Electrical and Electronic Engineering: Vol. 11 Power Sources for Electric Vehicles", Ed. B. D. McNicol and D. A. J. Rand, Elsevier, Amsterdam,1984; 573-619. 16 G. Caprioglio and A. Weinberg, Proc. 6th Intersoc. Energy Conv. Eng. Conf., Boston, 1971; 140. 17 R. Witherspoon, E. Zeitner and H. Schulte, Proc. 6th Intersoc. Energy Conv. Eng. Conf., Boston, 1971; 96. 18 H. Ikeda, N. Furukawa and M. Ide, paper presented at J . Congr. Am. Jap., Honolulu, 1979. Chem. SOC. 19 F. Bonino and M. Lazzari, "Modern Batteries: Cap. 8 - Secondary Hybrid Cells", Ed. C. A. Vicent, Edward Arnold, London, 1984; 227-239. a0 D. D. MacDonald and C. English, Journal of Applied Electrochemistry, 20, 1990;405-417. 2l 0. Hasvold, Chemistry and Industry, 1February 1888; 85-88. 22 "Aluminiudair Battery Development, Final Report", Lawrence Livermore National Laboratory, California, November 1987, UCRL-21018. 23 A. Despic and K. Krsmanovic, Journal of Applied Electrochemistry, 19, 1989;323-330.

This Page Intentionally Left Blank

223

THE PROSPECTIVE ROLE OF MAGNESIUM IN THE DEVELOPMENT OF ENVIRONMENT-FRIENDLY SOLID-STATE BATTERIES L. P. S. Aracjo, A. M. G. Pacheco and C. A. C. Sequeira Dept. Engenharia Quimica, lnstituto Superior Tecnico Av. Rovisco Pais, 1096 Lisboa Codex, Portugal ABSTRACT A bird's-eye view is given on the estimated burdens and foreseeable trends of some major atmospheric pollutants after an entire era of fossil fuels consumption. The ever rising importance of renewable power sources and the progress in solidstate batteries are briefly outlined. Several experiments with PEGnMg(C104)21PN (30% wt. IPN; n=8, 12, 16, 24, 30) have been carried out, since these polymeric electrolytes are tentatively entertained to offer an alternative to lithium systems. Preliminary results of AC conductivity measurements (25 - 135 C) performed on PEG1sMg(C104)21PN by means of electrochemical impedance spectroscopy (EIS; 65.5 kHz - 10.4 Hz) are reported and discussed. INTRODUCTION

The environmental impact of current needs and foreseen demands for energy must not be overlooked and hardly need to be emphasized, even from the standpoint of a sustainable development. In fact, and in spite of the general concern about issues such as acid rain and global warming, some 90% of the world energy budget still comes from burning fossil fuels [l],thus adding more and more greenhouse gases - namely, C02 - and acidic species - e. g., SOx and NOx to the already burdened atmosphere. Albeit other contributors to the greenhouse effect (methane, nitrous oxide, tropospheric ozone, various CFCs) have been considered, the major contribution comes from carbon dioxide, which has been held responsible for ca. 49% of the effect. Scavenging of C02 at ground level is accomplished mainly through plant metabolism, bacteriological decay, and solution in the sea; the uppermost nontropospheric sink appears to be its photochemical decomposition above the mesopause [2]. Deep concern has been expressed over several decades about the abundance and fate of carbon dioxide in the atmosphere. The historical background of about 270 ppm started rising with the Industrial Revolution one hundred years ago, yet solely between 1960 and 1985 was there found an increase from 315 to 345 ppm: by the turning of the former decade (1969-70), estimates of C02 concentrations near ground level and in the vicinity of the stratopause were around 330 and 320 ppm respectively [2,3], Suggesting an increase at a rate of 0.7 to 1 ppm per year. At such a rate, and on the basis of global climatic models developed thereupon, the Earth's radiation balance would end severely affected, with the average surface temperature shifting upwards some 1.5-4.5 C in the near future (before the year 2050). However, time Series are not that conclusive, and fortunately (or unfortunately..., as already pointed out by Cadle and Allen [2]) the predicted warming seems to have been offset or even reversed to

224

a certain extent by a cooling effect due to the accumulation of particulate matter in the atmosphere. Still, such an issue deserves continuing attention inasmuch as the very fate and escape routes of carbon dioxide in the lower atmosphere are not well understood as yet: for instance, despite an extensive oceanographical research in general and an intensive effort to study the surface waters in detail, no one seems to know for sure the scavenging power of the ocean and/or the kinetics of the CO2 transfer across it. Regarding this major tropospheric sink, the situation could be pretty much the same that Maclntyre [4] depicted some 20 years ago, that is questions like how quickly the excess gas will be absorbed by the top microlayer of the ocean, what are its main routes into the deep water (and how fast and pervasive they are), and just how much of it is left to influence the Earth's climate, such questions remain unanswered. Following a similar pattern, the sulphur content of glacial snow and ice has been steadily increasing for the past several decades, having previously remained constant over the centuries [5-lo]. Although a definite, quantitative link between airborne concentrations and the glacial record may be difficult to establish, inasmuch as gross seasonal variations in those concentrations are believed to occur [l 11 and a strong background of non-anthropogenic sulphur compounds is known to exist [12], such a trend seems hardly unaccountable for the mid-latitude, man-made emission from Northern Hemisphere's urban-industrial sources. Besides, there is a relative agreement upon man's contribution to the sulphur budget rising by a factor of two just in the period 1950 through 1975 [13,14]; nevertheless, some dark prognosis of that share getting to and overtaking Nature's by or even before the end of the present century are likely to become outdated, due to an ever growing pressure for industrialized countries to limit pollutant fluxes into the atmosphere. As a consequence of either stringent national policies such as the US Clean Air Act of 1970, or acknowledging the recommendations of international bodies like OECD, the anthropogenic emissions of sulphur dioxide appear to have generally levelled off in the early seventies, and followed a downward trend thenceforth (see, for instance, [15-181);it is anticipated that this trend will continue, as regulations concerning the conservation of energy, the use of cleaner fuels, and the installation of FGD systems spread all over the world. An important contribution on this subject came with the 1988 EC directive committing Member States to a 15year plan against acid-rain-maker pollution from power plants and other heavy industrial facilities: emissions of sulphur dioxide are to be cut by 60 percent from their 1980 levels in three stages through 2003 (25% by 1993, 40% by 1998, and 60% by 2003) [19]. As for NOx, a precursor of the nitrate ion found ubiquitously in precipitation [20] and in airborne dusts [21,22], a similar trend applies: in 1977, NOx accounted for 12% (23 Mt) of the total air pollutant fluxes in the United States, whereas SOx accounted for 14% (27.4 Mt) [23], and its emissions were expected to grow relative to sulphur dioxide's at least through the end of this century [24]. Unlike SOx, which is primarily emitted from stationary combustion sources, NOx concentrations are strongly dependent on transportation-related sources, and thus, to a certain extent, much more difficult to control: it may be worth mentioning that, as early as 1950-51, in an area with a heavy traffic density such as Los Angeles' (CA, USA), nitrogen oxides were added to the atmosphere at a rate of 200 to 300 tonne per day [25]. NOx emissions increased slightly in the late ' ~ O S , began to decrease in the early ' ~ O S ,and became relatively stable for the next few years, again as inferred from US

225

figures [17]. It is hoped that one can rely on engine performance (combustion efficiency) and on exhaust technology (catalysts) to counteract an ever rising number of motor vehicles, in order to regain some negative slope in NOx time series charts; besides, despite both NOx and SO, emissions having apparently peaked in the seventies, their levels still remain too high [26] (in the EC, nitrogen oxides are to be reduced by 30 percent from their 1980 levels in two stages through 1998 [19]). Roughly pictured, this is the situation thus far, i e., after an entire era of fossil fuels consumption and only a few decades of environmental concern: things could really be worse, yet they should actually be a lot better... Electrochemical energy storage has become more and more important with increasing complexity of power distribution systems, increasing environmental protection, and decreasing conventional energy resources. Rechargeable batteries appear as prime candidates for managing renewable energies (e. g., solar or wind), for load levelling, and for delivering some rather clean power to electric vehicles and remote areas. Unfortunately, the energy density of commercial secondary batteries ( 4 0 Wh.kg-1, typically) stands well below that of current fuels such as gas, oil or coal, and their by-products (up to 12 kWh.kg-1); moreover, and in general terms, the energy density of electrochemical power sources is much lower than what could be expected from the weight and the thermodynamic features of current battery-active materials (beyond 1 kWh.kg-1). Hopefully, such a large discrepance between theoretical expectations and practical achievements will be diminished to an appreciable extent by future work. The last few years have witnessed a high level of activity pertaining to the research and development of all-solid, thin-film polymer electrolyte batteries: most of these use lithium as the active anode material, polymer-based matrices as solid electrolytes, and insertion compounds as active cathode materials. Highperformance prototypes of such batteries stand currently under research, whose trends are expected to include the development of amorphous polymers with very low glass-transition temperatures, mixed polymer electrolytes, and fast-ion conductors in which the cationic transport number approaches unity. Mainly on account of its lightness, extreme electropositivity, theoretical energy density, and electrochemical reversibility in polymer electrolytes, lithium appeared as an obvious, if not the best eligible anode for solid-state batteries: therefore, such batteries should be capable of storing more energy (wt. by wt.) than the other commercially available alternatives (e. g., Ni-Cd or lead-acid cells). However, several drawbacks are known to be associated at large with lithium technology, namely the toxicity and cost of the material itself, plus an eventual liability to economic and/or political speculation, owing to the extent and/or location of the worldwide reserves. Besides, anodic efficiency is often impaired by lithium passivation: this phenomenon takes place in polymer networks as well as in liquid organic electrolytes, and seems to depend in an yet unpredictable, apparently erratic way on both the type of polymer matrix and the temperature of operation. On the other hand, polymer electrolytes must have high ionic conductivity, and good dielectric properties, electrochemical compatibility, and mechanical strength, in order to act as suitable cell separators. A large number of polymeric fast-ion conductors have been proposed for the last few years: in terms of specific ionic conductivity at low temperature, they now stand two to four orders of magnitude (ca. 10-5S.cm-1) above those in the original reports by Wright [27];however, for room-

226

temperature applications toward a cleaner environment, further improvements should be sought. Magnesium polymeric electrolytes are likely to offer an interesting, environment-friendly alternative to lithium systems. First, magnesium poses no special cost and/or supply problem(s). Second, it does not bear the noxious effects associated with lithium: actually, the alkaline-earth metal has been closely related to basic human needs such as medical care, the protection (corrosion prevention) of drinking-water distribution networks, and manned flight, to quote only a few. Thirdly, magnesium features an impressive anodic suitability: it exhibits a highly negatke electrode potential and a theoretical charge density that, apart from lithium, stands second only to aluminium's; in practical terms, the de-electronation liability of magnesium has been long recognized [28-311, and is currently put to work, 8. g., in high-performance cathodic protection systems [32,33]. Lastly, still regarding polymer-Li salt complexes and accounting for the similarity in ionic radii, Mg-based analogues would lead to an increase in the anion-to-cation ratio (a twofo Id increment).

EXPERIMENTAL

RESULTS

AND

DISCUSSION

Attempts have been made in our laboratory to extend the former work by Chiang et a/. [34], in order to produce the interpenetrating polymer network (IPN) based on Mg(C104)2, polyethylene glycol (PEG), and an epoxy resin; no special precautions were taken to dry matrices, salts, or solvents before use. Although a full characterization of the system is still ongoing, it appears that such a doped IPN may turn into an attractive material for energy-storage purposes. Bulk PEGnMg(C104)2IPN (30% weight IPN) was prepared/obtained from admixing 4,4'- isopropylidenediphenol epichlorohydrin resin, diethylenetriamine (DETA), PEG 400, and Mg(C104)2, in acetonitrile (5% wt. solutions), all chemicals being of reagent grade quality; forthcoming solid electrolytes were designed to bear n=8, 12, 16, 24, 30, where n stands for the molar ratio [PEG 400 unit]:[Mg salt]. In either case, acetonitrile solutions were poured into beakers (150 ml) and magnetically stirred for about one hour at room temperature; after this, they were transferred into desiccator-enclosed PTFE moulds and furnace-dried at 80-100 C; a slow evaporation was allowed to occur by keeping the desiccator slightly open. Solvent-free cast films were then removed from the PTFE underlayers and stored in another desiccator until further required. The introduction of resin and DETA enables the mechanical properties of the polymer electrolyte to be upgraded, insofar as such additives intermingle to form a solid matrix which PEG is held upon in an amorphous conformation at room temperature, as well as above it: due allowance made, all happens as if one had a wet sponge, the water corresponding to the liquid PEGnMg(C104)2 complex and the sponge itself to the aforementioned mixed network. AC impedance measurements were carried out on 300 p m thick polymer electrolyte samples, with n=16 and an exposed area of 0.20 cm2, using a Pt/PEGl~Mg(Cl04)2/Ptcell configuration; results for the other molar ratios will be reported later. Regarding the experimental set-up, major equipment included a Schlumberger* 1255 HF Frequency Response Analyser, electrochemically interfaced to a Western Systems* 486 PC via a PAR* 273A

227

Potentiostat/Galvanostat, and a Buchi* GKR-51 glass oven (*registered trade names). Table 1 lists the bulk resistance and ionic conductivity of PEG16Mg(CIO&lPN (30% wt. IPN) as a function of temperature; an Arrhenius plot of the conductivity data is shown in Figure 1. Table 1 PEG16Mg(C104)21PN bulk resistance and ionic conductivity as a function of temperature (25 - 135 C), from AC impedance measurements (65.5 kHz - 10.4 Hz).

0-6 1.434~1 1.129x1 0-5 2.1 83x1O‘5 2.820~1 0-5 5.21 4x1 0-5 1.023~1 0-4

104600 13290 6872 5319 2877 1467

25 63 80 90 112 135

-3.0

-4.0 h

r

E r!

?

.Q

-5.0

0)

-0

-6.0

-7.0 2.4

2.6

2.8

3.0

3.2

3.4

lOOO/T (K-’l

Figure 1. Arrhenius plot for the conductivity data in Table 1.

228

As it can be inferred from Figure 1, there exists an obvious linear relationship when polymer conductivity and absolute temperature are plotted in Arrhenius mode. The correlation is highly significant: its Pearson coefficient reads 0.996 when rounded off to the nearest thousandth, which makes it significant at the .001 level (allowing for just four degrees of freedom), and the corresponding regression line explains some 99.3 percent of the total (corrected) variance in the results, that is only a few residual tenths are unaccounted for by the regression. In these terms, a classical, unmodified Arrhenius model may be seen to hold for the temperature range under consideration: apart from essentially excluding any phase change, such a model yields an apparent activation energy (Ea) for the conduction process of about 40 kJ.mole-1 (38.8 kJ.mole-l), which seems fairly agreeable with single ion transport through the abovepictured sponge analogue. Moreover, other than confirming the thermal stability of the PEGl6Mg(C104)2 -soaked epoxy phase over this temperature window, compliance with an Arrhenius-type behaviour to give such an energy barrier suggests that the PEG-alkaline earth complex stays very largely, if not completely amorphous for the present molar ratio (n=16) [35]. At and above room temperature, this polymeric electrolyte shows a greater conductivity than pure PEO: the magnitude order transition 10-6-10-5S.cm-1 occurs at around 60 C, which seems consistent with former results by Patrick et a/. [36], and turns this compound into an attractive solid-state electrolyte for battery applications [37].On the other hand, PEGnMg(C104)2lPN films can be easily preparedlhandled and they show good mechanical and electrochemical properties: these are essential requirements to be met by any polymer candidate to such applications. Also, it should be noted that the historical drawback of magnesium cells, the so-called "delayed action", i.e., the time elapsed before the protective anodic layer is disrupted and full operating voltage is achieved (see, for instance, Robinson [38] and references therein) cannot be envisaged as a potentially serious handicap of the present systems in service conditions, insofar as the perchlorate ion most likely is a strong pitting agent [39]. CONCLUSIONS The casting of interpenetrating polymer network (IPN) films based on Mg(C104)2, polyethylene glycol (PEG 400),and an epoxy resin results in an easy laboratory task, at least down to a PEG-to-salt molar ratio of 8; it is anticipated that their production in a larger scale would put no particular problems. Conductivity data from AC impedance measurements on PEG16Mg(C104)21PN (30% wt. IPN) samples over the range 25 to 135 C fits in an Arrhenius model to yield an apparent activation energy of about 40 kJ.mole-1, thus suggesting a single-ion hopping mechanism for conduction; such a model accounts for all but 0.7 percent of the total variance in the results, which makes it significant at the .001 level. Since the conductivity of polymer electrolytes generally follows the VogeI-Tamman-Fulcher (VTF) equation, further investigations are required to elucidate whether true Arrhenius behaviour or free volume and configurational entropy models should be considered. Albeit these are preliminary, illustrative results, and system optimization is still ongoing, PEGnMg(C104)21PN are likely to become an alternative to Li-polymer electrolytes for low-temperature devices, not only in terms of specific cell features,

229

but also on ecological grounds, inasmuch as the magnesium friendship vis-A-vis the environment lies far beyond that of lithium. Among other advantages, such an alternative holds out prospects of technological risk abatement and better waste disposal management. REFERENCES

1

2 3 4 5 6 7 8 9

I. Fells (1990):in Energy and the Environment (J. Dunderdale, Ed.), Royal Society of Chemistry, London, p. 1. R. D. Cadle, E. R. Allen (1970):Science, 167,243. T. G. Scholz, L. E. Heidt, E. A. Martell, D. H. Ehalt (1969):Trans. Am. Geophys. Union, 50, 176. F. Maclntyre (1 974):Scientific American, 230(5),62. M. Koide, E. D. Goldberg (1971):J. geophys. Res., 76,6589. H. V. Weiss, K. Bertine, M. Koide, E. D. Goldberg (1975):Geochim. cosmochim. Acta, 39,1. A. Nyberg (1977):Quart. J. Roy. Meteor. SOC.,103,607. R. Delmas (1 979):in Proc. lnfernational Symposium on "Sulphur Emissions and the Environment", Society of Chemical Industry, London, p. 72. A. Neftel, J. Beer, H. Oeschger, F. Zurcher, R. C. Finkel (1985):Nature, 314,

611. 10 P. A. Mayewski, W. B. Lyons, M. J. Spencer, M. Twickler, W. Dansgaard, B. Koci, C. I. Davidson, R. E. Honrath (1986):Science, 232,975. 1 1 C. I. Davidson, J. R. Harrington, M. J. Stephenson, M. J. Small, F. P. Boscoe, R. E. Gandley (1989):Atmospheric Environment, 23,2483. 12 A. M. G. Pacheco (1988):in Microbial Corrosion 1 (C. A. C. Sequeira, A. K. Tiller, Eds.), Elsevier Science Publishers Ltd., London, p. 361. 13 C. F. Cullis, M. M. Hirschler (1980):Atmospheric Environment, 14,1263. 14 D. Mdller (1 984):Atmospheric Environment, 18,19. 15 M.-L. Weatherley (1 977):Clean Air U. K., 7(27),4. 16 M.-L. Weatherley (1979):Clean Air U. K., 9(1),8. 17 Environmental Protection Agency (1 985): National Air Quality and Emission Trends Report 1983, EPA, Research Triangle Park (NC, USA).

18 Staff Feature (1 991):Materials Performance, 30(8),10. 19 Staff Feature (1991):Materials Performance, 30(1 l), 12. 20 T. E. Graedel (1988):in Degradation of Metals in the Atmosphere ASTM STP 965,ASTM, Philadelphia (PA, USA), p. 327. 21 H. W. Hermance, C. A. Russell, E. J. Bauer, T. F. Egan, H. V. Wadlow (1971): Environ. Sci. Techno/., 5,781. 22 J. D. Sinclair, L. A. Psota-Kelty, C. J. Weschler, H. C. Shields (1990):J. Elecfrochem. SOC.,137,1200. 23 Environmental Protection Agency (1978):National Air Quality, Monitoring and 24 25 26 27 28

Emission Trends Report 1977, EPA, Research Triangle Park (NC, USA). J. Wisniewski (1982):Water, Air, and Soil Pollut., 17,361. A. J. Haagen-Srnit (1952):Ind. Eng. Chem., 44, 1342. R. Baboian (1 993):Materials Performance, 32(5),46. P. V. Wright (1975):Br. PolymerJ., 7,319. J. Bussey (1 830):J. Chim. Med., 6,141.

230

29 30 31 32 33 34 35 36 37 38 39

L. Whitby (1933): Trans. Faraday Soc., 29, 844. L. Whitby (1933): Trans. Faraday Soc., 29, 853. G. D. Bengough, L. Whitby (1933): Trans. lnst. Chem. Eng., 11, 176. J. N. Britton (1991): Materials Performance, 30(8), 23. K. C. Lunden, T. M. Stastny (1991): Materials Performance, 30(11), 24. C. K. Chiang, B. J. Bauer, R. M. Briber, A. T. Davies (1987): Polymer Comm., 28, 124. J. E. Weston, B. C. H. Steele (1981): Solid State lonics, 2, 347. A. Patrick, M. Glasse, R. Latham, R. Linford (1986): Solid State lonics, 18/19, 1063. M. B. Armand, J. M. Chabagno, M. J. Duclot (1979): in Fast /on Transport in Solids (P. Vashishta, J. N. Mundy, G. K. Shenoy, Eds.), Elsevier North Holland Inc., Amsterdam, p. 131. J. L. Robinson (1976): in The Primary Battery (N. C. Cahoon, G. W. Heise, Eds.), John Wiley and Sons Inc., New York, p. 149. E. Gulbrandsen (1992): Electrochim. Acra, 37, 1403.

SECTION THREE

SENSORS FOR POLLUTION CONTROL

This Page Intentionally Left Blank

233

OVERVIEW OF GAS SENSORS FOR ENVIRONMENTALUSE Tetsuro Seiyamaa, Takeshi Nakahara b and Takashi Takeuchib aProfessor Emeritus, Kyushu University (Mailing Adress: Tokuyama Soda Co., Ltd., Fukuoka Branch, Sanwa-bldg., 1-10-24 Tenjin, Chuoku, Fukuoka-city 810, Japan) bFujisawa Research Laboratory, Tokuyama Soda Co., Ltd., 2051 Endo, Fujisawa-city 252, Japan 1. IMPORTANCE OF ENVIRONMENTAL PROBLEMS AND GROWING NEEDS FOR GAS SENSING

Recently, air pollution became a serious problem again as global environmental problems such as greenhouse effect, acid rain and ozone layer destruction. Of these, the greenhouse effect is the most important and difficult problem. Many scientists warned global warming due t o greenhouse effect originated from the increase of CO, concentration in the air. The increase of the atmospheric temperature for past twenty years was 0.3"C whereas it is said to be another 0.3"C for next ten years and 3°C at the end of the twenty-first century, which will bring the sea level rising of 65cm. The second global environmental problem is the acid rain. Some components, especially NO, and SO,, of the exhaust gas from combustion engines, boilers and furnaces make rain acidity (acid rain) to destroy trees and forests. The damage by acid rain has become serious especially in Europe and the east coast of US. The third problem is ozone layer destruction. Chloro-fluorocarbon is decomposed into C1 atom by ultraviolet light and induces chain reactions t o destroy ozone molecules in the stratosphere, which results the increase, of ultraviolet light fallen to the ground and causes the increase of skin cancer. Under the background of growing interest to the global environmental problems, countermeasures and regulations for not only problems directly related to the global environment but also more local environmental problems are now being actively reinforced. In Japan, NO, pollution especially due to Diesel engine vehicles is considered to be one of the most important and urgent problems. The rapid reduction of NO, emission from Diesel engines, however, is considered to be difficult so that besides the control of emission from individual Diesel engines, the control of vehicle type to promote replacing high emission Diesel engines by lower emission engines is being planned to be introduced in the large cities in Japan. In US also, the clean air act was revised in 1990 and the regulations and standards for NO,, CO, HC, SO, emissions were widely altered to diminish urban air pollution, acid rain and ozone layer destruction. Typical air pollution problems are summarized in Table 1, in which the related gases (both harmful and beneficial gases) and the harmful emission sources are listed. To solve these pollution problems, various countermeasures, which are also given in the table, are trying to be taken. The monitoring and measurement of gases related t o the air pollution are the most fundamental to performing many of these countermeasures successfully, so that handy and

234

Table 1 Air pollution problems and related gases Air pollution problem

Global problem Greenhouse effect

Related gas

Main cause or source Countermeasure

C02, CFC, CH4, N20,

Acid rain

SO2, NO,, NH3, HCI

Combustion of fuel Decrease of forest Eruption of volcano Use and loss of CFC Substitution and collection of CFC Combustion of fuel

Ozone layer destruction

CFC, BCFC, BFC, CH3CC13 CCI4, O3

Decomposition of O3 with C1 in W-light

sO2, HCI, CO, Sulfuric acid mist NO,,SO,, Non-methane HC

Industrial waste gas Combustion of coal Combustion of fuel

03

Local problem Smog Photochemical smog

Saving energy Improving efficiency Changing fuel & energy Forest conservation

Desulfurization & denitration from fuel & exhaust gas Improvement of combustion Substitution & collection of CFC

Smoke treatment Same as acid rain

CFC: Chlorofluorocarbon BCFC: Bromochlorofluorocarbon BFC: Bromofluorocarbon

reliable gas sensors are strongly desired to be developed. The needs of gas sensing for controlling air pollution can be classified in three cases. The first one is for grasping the actual circumstances of the air pollution, for example, understanding the details of the generation and consumption of CO, and CH, on global scale. The second is the routine watching of pollutant gases in the atmosphere and in combustion exhaust gases. For this purpose, conventional analytical instruments are used at present but more simple and easy methods are desired t o increase the number of monitoring stations and emission sources. The third need of gas sensing is for closed-loop control of combustion engines, boilers and furnaces. The application of oxygen sensors to the feedback control of the air-fuel ratio of automobile engines is the only one example in which gas sensors are widely used at present for the use of closed-loop control. Table 2 shows the concentration range and Japanese standard for several pollution gases existing in the air and exhaust gases, which are attracting special interest recently. Considering from the standpoint of sensor development, pollution gas sensors can be divided into two groups with different required specification. The one is for sensing in the atmosphere. For

Table 2 Concentration range and Japanese standard for gases related to environmental problems

co Concentration range to be detected

Air pollution

Air pollution (0.001-2ppm) Coal or oil burner (1-1OOOppm) Incinerator (10-500ppm)

Air pollution (0.01-1Oppm) Gas poisoning (5-1OOOppm) Tunnel of mine (1-1OOppm) Boiler, burner (10-1OOOppm) Automotive exhaust (0.1-10%)

Air pollution on the ground (0.01-lppm) Stratosphere (0.1-lOppm or 109-10* molecules/cm3)

NO2
<0.04ppm (dayly average)

oxidant <0.06ppm (hr. average)

60-8OOp~m(*~) 0.48gflrm (passenger car) 6.8-7.8&Wh (heavy d i e d ) 5ppm (NO21 25ppm (NO)

0.1-5m3/hr(*3)

(0.01-0.5ppm) Boiler,combuster (10-2OOOppm) Automotive exhaust (10-5OOOppm) Heater, water heater (105OOppm)

CFC(*l)

03 Air pollution (200-4OOppm) Boiler, burner (1-10%) Automotive exhaust (5-20%)

Air pollution (lppt-lppb) Leak from refrigerator blppm)

Japanese Standard

Air pollution

Exhaust gas Fixed facility Automobile

General allowable limit

2PPm

2.7gh (passenger car) 9.2gkWh (heavy diesel) 5Oppm O.lppm

5000Ppm

(*1) CFC: Chlorofluorocarbon (*2) depend on kind and scale of facility

(*3) depend on height of chimney and location of facility

N W

wl

236

this purpose, the required concentration range is very low, in general, under 1 ppm so that extremely high sensitivity is required for sensors. The response time and the resistance against the environment are not so severe. The other group of sensors is for sensing in exhaust gases from emission sources. In this case, fast response and strong resistance against high temperature exhaust gases are required for the sensors. The concentration range is generally above 1 ppm, which is rather high in comparison with the sensing in the air. The accuracy and selectivity are equally required for both sensing in the air and in exhaust gases. 2. METHODS OF GAS SENSING RELATED WITH ENVIRONMENTAL PROBLEMS 2.1. Conventional AnalyticalMethods Conventional analytical methods for environmental use have been established for the routine watching of pollutant gases in the atmosphere and in combustion exhaust gases. Equipments based on various analytical methods have been put into practice and are widely used in the field. Table 3 presents conventional analytical methods for measuring gases related to environmental pollution. The objective gases can be classified into two groups, i.e., gases at relatively high concentration in the emission and those a t extremely low concentration diffused in the atmosphere. In the table, therefore, measurable gases of conventional analytical methods are categorized by concentration ranges. Classifying various kinds of these methods, there are three main types, that is, optical type which utilize the change of absorption spectrum or luminescence spectrum of gas, chemical analysis based on the chemical reaction of an absorbed gas with a specific reagent, and gas chromatography which use a variety of detectors. Of these, chemical analysis, which is a wet method, has a few disadvantages such as low reaction efficiency between the absorbed gas and the specific reagent, and difficulty for the continuous measuring of the gas. For these reasons, chemical analysis is being displaced by dry methods i.e. optical type or gas chromatography. Among the dry methods, infrared spectroscopy, an optical type, has some features such as variety of measuring gases and simplicity of the measuring principle. Therefore, the applied fields of this method must be further widespread if high sensitive infrared detector and infrared laser which enable both to enhance the sensitivity and to miniaturize is developed. Analytical conventional methods presented in the table are basically used t o measure gases in one spot. It is, therefore, difficult to monitor the degree of the environmental pollution in terrestrial scale. Recently interesting new approaches have been tried in order to extend the monitoring area. For example, perpendicular distribution evaluation of ozone concentration in the stratosphere by a laser radar, total ozone monitoring in extended area by a total ozone mapping spectrometer equipped in an artificial satellite, multi-gas measurement such as CO,, CH, and N,O by an atmospheric periphery infrared spectrometer, high sensitive and accurate analysis of various pollutant gases by a semiconductor laser infrared spectrometer have been carried out lately. Gas monitoring in terrestrial scale must become more important in the future but further details of the new methods is omitted in this paper.

237

Table 3 Conventional analytical methods for measuring gases related to environmental Dollution Analyticalmethod

NOx

SOX

Infrared spectroscopy

8

Q

Ultraviolet spectroscopy Absorptiometry

8

8

. 3 4

Conductometric analysis Thermal conductivity detector Flame ionization detector Flame photometric detector a:Chlorofluorocarbon

Q

0 3

CFC'

HCb

CO

Q Q c m

Q .

8.

. . a

(asoxl ant)

0

Chemiluminescence analysis Coulometry

C02

0

(as oxidant)

@a

Q

8

Q Q c m

rn b:Hydrocarbon

8:Available for monitoring gases in the emission. 0 :Available for monitoring diluted gases in the atmosphere. 2.2. various

of Gas sensors Equipments which utilize conventional analytical methods exhibit excellent accuracy and reproducibility but generally are large, heavy and expensive. Moreover, they are complicated in operation and require much labor in maintenance. Gas sensors which are remarkably smaller, cheeper, simpler and easier than conventional equipments are, therefore, urgently desired to get more data a t more points. Various types of gas sensors which applied a variety of principles and materials have been proposed. Table 4 shows types of gas sensors which is specified by detecting principles and their detectable environmental pollutant gases. Most gas sensors listed in the table have not been put to practical use yet but are on the way to development, so that the sensors in the table were chosen by the feasibility judgment of their detectable gases and concentration ranges. All kinds of environmental pollutant gases except for SO,, chlorofluorocarbon and hydrocarbon in the atmosphere can be detected by the gas sensors as shown in the table. It is, therefore, expected that they will greatly contribute to solve the environmental problems in near future. Principles and features of gas sensors are briefly given as under.

2.2.1. Semiconductor Qpe Semiconductor gas sensor is based on electrical conduction changes, caused

238

Table 4 Gas sensors for measuring gases related to environmental pollution Type of gas sensor

NOx

Semiconductor

@a

Electrochemical (liquid electrolyte)

Q

Electrochemical (solid electrolyte) Catalytic combustion

8

Optical

Q

Oscillator

G3.Q

SOX

C02

0

8

Q

0 3

CFCa

O

Q

O

Q

HCb

GO

Q

Q

Q

Q

Q

8 Q

a:Chlorofluorocarbon b:Hydrocarbon :Available for monitoring gases in the emission. 0 :Available for monitoring diluted gases in the atmosphere.

by contact with gases, of metal oxide semiconductors such as SnO, or ZnO and organic semiconductors such as phthalocyanine or polypyrrole. A number of studies have been done on semiconductor gas sensors since they were developed independently by Seiyama et al.[ll and Taguchi[21. Gas sensors consisting of SnO,, in particular, have been first put to use t o detect leakage of combustibles by Figaro Engineering Inc., Japan, in 1968. Since then the applications of SnO, gas sensors have been widespread in various fields for detecting leakage of toxic gases, checking for alcohol, monitoring for cigarette smoke and sensing for odor substances. The relationship between sensor resistance of semiconductor type and gas concentration can be expressed by the exponential function In R = k + aoln C where R is the resistance of the sensor, k and a are the constants and C is the concentration of gas. This equation means that the semiconductor gas sensor is able t o detect wide concentrations of gases, with high accuracy at low concentrations. Furthermore it can detect a variety of gases, electron donative and attractive gases, which undergo the electronic reaction with semiconductor surfaces. Fig. 1 shows the sensitivity characteristics for several gases of the commercialized SnO, gas sensor(TGS822, Figaro Engineering Inc.). The resistance of the sensor decreases as a logarithmic function of the concentration of gases. The sensor can detect various electron donative gases, e.g. ethanol, H,, CO and so on. On the other hand, the lack of selectivity become a problem in case of detecting a specific gas in mixed gases. In order to enhance the selectivity t o a specific gas, optimization of the operating temperature, addition of sensitizers or dopants or control of the morphology are

239

Figure 1.Resistivity variations of comercialized Sn02 semiconductor sensor for various gases.

Gas Conc. / ppm widely investigated. Semiconductor gas sensors are especially expected to be applied for detecting environmental pollutant gases such as NO,, CO, ozone, chlorofluorocarbon and hydrocarbon. 2.2.2. Solid Electrolyte Qp

Electrochemical gas sensors detect gases based on the electromotive force(EMF) o r the current of an electrochemical cell due to the electrochemical reaction of a particular gas. Solid electrolyte which a specific ion can selectively permeate is used as a diaphragm. Potentiometric type gas sensors have been most widely adopted. Among them potentiometric oxygen sensors composed of partial stabilized zirconia have already had practical application and been extensively used for the feedback control of the air-fuel ratio of automobile engines. The oxygen sensor elements are composed of the following electrochemical cell.

EMF is represented as E = (RT/4F)ln(Po2JJ/PO,I ) where Po, I and Po, II are the partial pressure of oxygen in reference gas and the one in sample gas respectively, and RT/F has usual meaning. As far as is kept constant, PO2n can be obtained by monitoring the value of the

%id.

Numbers of detectable gases have been increased since Gauthier et a1.[31 first proposed the feasibility of solid electrolyte gas sensors for SO,, NO,, CO, and so on. Moreover the couple of a metal oxide solid electrolyte and an auxiliary electrode, investigated recently, makes it possible for stable operation and disuse of a reference gas. Electrochemical gas sensors has a strong point of detecting gases selectively because only one kind of ion can permeate through

240

the electrolyte. In the present stage, however, there are unsettled problems such as the discrepancy of EMF from the Nernst's equation at a low temperature and the insufficiency of sensitivity t o gases at low concentrations. To overcome these problems, further researches on materials of solid electrolytes and electrodes should be needed. Solid electrolyte gas sensors are anticipated to be developed for detecting SO,, NO, and CO,.

2A3.Liquid ElectrolyteType Liquid electrolyte gas sensors, which employ electrolytic liquid solution, detect gases by taking out the current as sensor signals, which generates when these gases are electrochemically oxidized o r reduced a t an electrolytic liquid/electrode interface, This type has earliest been put into practice among gas sensors. There are two main types, constant potential electrolysis type and galvanic cell type. The former is more excellent because the sensor of this type is available for detecting various gases related to environmental pollution such as NO,, SO,, CO and ozone. Furthermore, because of diversity of the electrolytic potential for each gases, constant potential electrolysis type is capable of detecting a gas selectively by applying proper voltage. Short life caused by the evaporation and leakage of electrolytic liquid have been pointed out, these disadvantages was fairly solved by modifying electrolytic liquid, electrodes and gas permeable membranes. 22.4. Catalytic CombustionType

Catalytic combustion gas sensor is based on the resistance change of an element composed of a platinum coil covered with a highly oxidative catalyst. When inflammable gases contact with the element, the resistance change is caused from the temperature rise due to a combustion reaction and is converted into the electrical signal. From the practical point of view, a bridge circuit consisting of a sensing element and a compensating element is adopted in order to get rid of the influence of the ambient temperature change. The sensitivity of catalytic combustion gas sensors t o a inflammable gas is determined by the molecular heat of combustion of the gas. Therefore, it is easy for this type to measure single gas but difficult to detect a specific gas among mixed gases. In recent years, many investigations on stabilizing the catalyst have enabled the long term stability of the sensor. Catalytic combustion gas sensors are applicable for monitoring relatively high concentration of CO, chlorofluorocarbon and hydrocarbon in exhaust gases. 22.6. Optical'Qpe

Optical gas sensors utilize the optical change due to the interaction between a gas and a sensing material. The development of this type of sensor is being forwarded through the improvement of fiber optic communication and its circumferential techniques. Optical fiber, therefore, is often used for this type of sensor. Fiber optic sensors can be classified into two types. That is, unfunctional type in which an optical fiber only works as an optical propagation path, and functional type in which the whole or partial path of the fiber plays a role for gas sensing. In the former, the material in charge of the interaction with a gas is usually applied on the edge surface of an optical fiber, and the change of the intensity or phase of light is propagated through the fiber. In the latter, the method which utilizes evanescent waves is most widely used.

24 1

Gas sensing materials are doped in the cladding of an optical fiber. Refractive indices of core and cladding are adjusted as t o satisfy the condition for perfect reflection. In this system, the objective gases are detected from the change of evanescent waves caused by the interaction between a gas and a sensing material. The sensor can be sensitized by repeating perfect reflections many times. Optical gas sensors generally have some characteristics such as a high insulating capacity, independence of electrical noises and operation in safety. Moreover it is possible for them t o detect a variety of gases by choosing a gas sensing material. For environmental use of optical sensors, applications for NO,, SO,, CO, and CO are expected. 2.2.6. Oscillator

Oscillation type gas sensors consist of a material capable of selectively adsorbing and desorbing objective gases and the slight change in the weight of the material due to adsorption or desorption is converted into electrical signals by piezo-electric actuators. These sensors can be classified into two types, i.e., the one utilizes bulk acoustic waves of piezo-electric oscillators and the other makes use of SAW(surface acoustic wave) devices. The interaction between a gas and an adsorbent occurs at the surface of a sensor, so that the SAW type sensor responds t o the surface state more sensitively than the bulk acoustic wave type sensor. SAW sensors generally consist of a dual delay line of which two interdigital electrodes were fabricated onto a piezo-electric actuator as shown in Fig. 2[4]. The sensor layer, an adsorbent for a objective gas, was formed on one delay line. The other line is used t o compensate the temperature change. Oscillator type gas sensors are being investigated for detecting NO,, SO,, hydrocarbon, NH, and H,S. 2.2.7. Other Types Besides the sensors mentioned before, FET(fie1d effect transistor) type and

RF amplifier Sensor layer

Low pass filter

u+ Mixier

. Acoustic absorber RF amplifier Figure 2. Surface acoustic wave gas sensor consisting of dual delay line oscillator[4].

242

capacitor type are interesting regarding the detection of gases related t o an environmental pollution. The former, first proposed by Lundstrom[51, utilizes the gate potential change of a metal-oxide-semiconductorfield effect transistor due to contact with a gas . NO,, NH, and H,S sensors were fabricated by selecting the proper gate material. The latter detects gases by the capacitance change of a metal oxide capacitor or a MOS(meta1-oxide-semiconductor) capacitor and so far CO, and NH, sensors have been investigated. 3. RECENT RESEARCHES AND DEVELOPMENTS OF GAS SENSORS FOR ENVIRONMENTAL USE

3.1. General lkends Recently gas sensors for environmental use have been focused on the atmospheric monitoring and the emission control as increasing demands in the world. In order to digest trends of researches and developments on gas sensors for environmental use, the sensors are picked out among reports which were presented in latest five years at two main international conferences on sensors, "International Meeting on Chemical Sensors" and "International Conference on Solid-state Sensors and Actuators". The total 132 gas sensors concerning environmental problems, reported a t the conferences, were classified by their types and detectable gases and the results are shown in Fig. 3.As for the type of sensor, semiconductor type, solid electrolyte type and liquid electrolyte type are predominant and occupy above 70% of total numbers of the reports. Optical type and SAW type based on new detection principles also hold 20% and the number of the reports is increasing year by year. On the other hand, speaking of gases to be detected, five kinds of gas i.e., 0,, NO,, CO, hydrocarbon and CO, occupy above 80% in the whole reports. Among those gases, 0, as for the emission control of automobiles, CO and hydrocarbon as for the detection of the leakage of combustible or toxic gases have intensively been studied in the past. A number of reports on these gases indicate the movement of research from security use to environmental use. Furthermore, the fact that the number of reports on NO, and CO, have rapidly increased in several years reflects the concern about environmental problems. On the other hand, there are only a few reports concerning SO,, ozone and chlorofluorocarbon. This is due to the lack of enough sensitivity of the present gas sensors t o detect these gases diffused in the atmosphere at extremely low concentrations. Now, gas sensors are required t o have several characteristics as listed below in order to detect gases related to environmental pollution. (1)sensitivity: excellent sensitivity is necessary to detect gases especially those of very low concentrations in the atmosphere. (2)selectivity: gas sensors are required high selectivity so as to detect a specific gas in the exhaust where various kinds of inteference gases coexist and their concentrations widely change. Of cource, selectivity is also necessary in case of monitoring a gas in the atmosphere because the concentration of coexisting gases are generally much higher than that of the gas to be detected. (3)response time: fast response is needed especially for the monitoring in the exhaust. In order to control the emission of gases, shorter response time than the time constant of actuator systems is demanded. (4)reliability:because of the lack of long-term stability or rather large

243

Fig.3 Numbers of gas sensors for environmental use presented a the International Meeting on Chemical Sensors (1986,1990)and the International Conference on Solid-state Sensors and Actuators (1987,1989,1991).

dependency on temperature and humidity, practical applications are limited only in a few fields though many types of gas sensors have been proposed. Reliability of the sensor must be decisive for the practical application in environmental use. In order to satisfy these characteristics, many attempts are being made for various types of sensors. As for semiconductor type, the improvement in sensitivity is proceeded by thinning a sensor layer using new manufacturing techniques such as sputtering, evaporating and langmuir-blodgett method. Regarding the enhancement of selectivity, not only new sensing materials such as metal phthalocyanines, TiO, and In,O,, but also processing of sensor signals obtained from a multi-sensor array enable to detect a specific gas selectively. In solid electrolyte type, the measuring range of oxygen has been enlarged by limiting current type oxygen sensors and those sensors are being applied for controlling the emission from lean burn engines. NO,, SO, and CO, sensors, using the combination of chemically and thermally stable metal oxide electrolytes and auxiliary electrodes composed of metal salts which enable the stable operation, are also been widely studied. Also in liquid electrolyte type, interesting researches such as chlorofluorocarbon sensor with the combination of a pyrolyzer and a sensor, and NO, and SO, sensors using a new electrolyte have been reported. Furthermore a variety of sensors based on new principles e.g. optical type, oscillator type, pattern recognition method and so on have been investigated. Further studies on sensor materials, sensing principles and signal processing are desired t o put gas sensors in practical use in this field.

244

3.2. Individual Environmental Gas Sensors 3.2.1. N0,SenSor Acid rain is mainly caused by NO, and SO, emitted from combustion facilities such as automobiles and boilers. Of these gases, NO is generated in the combustion flame by the reaction of N, and 0, and gradually converts to oxides of higher order, which are mainly NO,, in the atmosphere as the time elapsed. There are, therefore, two applications of NO, sensor, the one for detecting NO which is a predominant component of NO, in the combustion emission and the other for monitoring NO, in the atmosphere. NO, is, in general, chemically active and gives rise t o electronic, optical and thermal reactions with a variety of materials due t o the adsorption on their surface. By utilizing those characteristics, various types of NO, sensors have been proposed. Present status of NO, sensing summarized by Takeuchi[6] is shown in Table 5, where

Table 5 Present status of NOx sensing [6]

-

Concentration range of NOx 1PPb IPPm 1% 0 1 . . . . . I . I

Requirment for: air pollution monitor combustion gas H Analytical mefhod: chemiluminescence infrared absorption H Handy sensors: Semiconductor type MePc a H PolYPyrrole H SnO2 H TiOz Wo3 H Inz03-Sn02 H hO3-v206 H CnO3-Nb206 H Electrochemical type phosphoric acid Na-P/P”-AhOdNaN03 H NASICON/NaN03-Ba(NOs)z H RbAg416 H Nafion H Optical type metal porphine H MePca H SAW type MePc a Others Sn02(FETC) H MePca (FE‘f ,seebeck effect) a:Metal phthalocyanine c:Field effect transistor

b:Surfase acoustic wave

Accuracy

(%I 0.1

1

10

I

I

em ma l5a mp

5

5

eaa mp

rma rrm mp

rn

rm Ra

rn rza prm

Em

ma ma

m

czz?a

245

Figure 4. Molecular structure of metal phthalocyanine. the concentration range and accuracy of various kinds of NO, sensors developed are listed together with conventional analytical methods and with requirements for monitoring of air pollution and of combustion gas. As for air pollution monitoring, semiconductor sensors [7-151 and SAW sensors[4,16-181 using metal phthalocyanines and solid electrolyte sensor utilizing RbAgI5[191 are expected because of their high sensitivity. Especially, phthalocyanines are chemically stable and are, therefore, most feasible materials for NO, monitoring in air. For detecting NO, in combustion exhaust, metal oxide semiconductor sensors[20-291, electrochemical sensors[30-331, optical sensors[34-361 and FET sensors[37-39] have been developed. Among them, sensors using ceramics as NO, sensitive materials have the ability of in-situ monitoring due to their thermal, chemical and mechanical stability. Judging from the performance of NO, sensors, several significant sensors are described in some detail in the following. Metal phthalocyanines, macrocyclic compounds in which various metals cdinate at the center as shown in Fig. 4, have been extensively studied on NO, sensors since Sadaoka et al. reported the p-type semiconductance of metal phthalocyanines being changed by the adsorption of NO, on their surfaces[40,411. Detailed studies on coordinated metals done by Jones et a1.[42,431 revealed that some metal phthalocyanines, like Pb phthalocyanine, prepared by vacuum sublimation could detect NO, in the concentration range of 1 to lOOppb and were promising materials for monitoring NO, in the atmosphere. There were, however, some disadvantages such as large influence of humidity, slow response especially in recovery and the lack of reproducibility. Many attempts was carried out to overcome the disadvantages, and the following several noticeable results were obtained. Sadaoka et a1.[10,11] of Ehime University, Japan, found that the response t o NO, is much improved by annealing in a proper condition after preparing metal phthalocyanine films by vacuum sublimation. For example, Pb phthalocyanine annealed in air or N2 at 603K showed the excellent response characteristic i.e., 20 seconds in response and a few minutes in recovery even in operating at room temperature. The fast response, which allow the operation in room temperature, owes to the morphology of Pb phthalocyanine films consisting of fine triclinic crystals with mean diameters of the order of 0.2pm. The influence of preparation conditions of Pb phthalocyanine on NO, sensing characteristics was also studied with spectroscopic and electrical techniques by Mockert et all71 of University of Tubingen, F.R.G. Fig. 5 shows 0 1s XPS spectra of Pb phthalocyanine films after different treatments, that is, (a) as preparation under UHV(u1tra-high-vacuum) conditions, (b) after exposure of UHV-prepared films to air and (c) after heating the air-exposed films in

lo' gatzysisorbeci I 102- chemisorbed IOH-

10'- lead oxide

a I

1

540

1

536

1

I

I

1

524

528

532

Binding Energy I eV

Figure 5 . 0 Is XPS spectra of Pb phthalocyanine thin films after different treatments. a:as preparation under UHV(ultra-high-vacuum) conditions, b:after exposure of UHV-prepared films to air, c:after the air-exposed filmes in 300mbar 0 2 a t 423KL71.

300mbar 0, at 423K. Literature data of the binding-energy values of O,(gas), O,(physisorbed), Oi(chemisorbed), OH- and 02-(as in PbO) are also indicated. During first exposure of the UHV-prepared thin films to air, small amounts of surface OH groups are formed, which are a prerequisite for subsequent reversible conductivity changes in sensor application. By this preparation procedure, the formation of lead oxide at the surface, which is the cause of the deterioration in sensor performance, can be avoided. Pb phthalocyanine sensor in this study can detect NO, at ppb orders of concentrations reproducibly with fast response. To improve sensor performances, a new sensor structure was proposed by Ruihua et a1.[131 of Semiconductor Institute of The Chinese Academy of Science, China, and by &in et all141 of University of Tongji, China, taking into account of the fact that the electronic interaction between metal phthalocyanines and NO, occurs at the surface of metal phthalocyanines. The configuration of the coplanar sensor and the current paths through the sensor layer are illustrated in Fig. 6. A pair of interdigital electrodes are formed on the one face of the layer whereas a control electrode is deposited on the other face, and they were then connected as shown in the figure. When the same voltage is applied across the interdigital electrodes and across the phthalocyanine layer, most of the electric current through the bulk phthalocyanine flows into the control electrode and the rest of the electric current through the surface region flows into the interdigital electrode. Therefore, the current flowing between two interdigital electrodes is sensitive to the change in the surface state. This is the

I

I

I

f' Pb(ok Cu)phthalotyanine

I

1

'

I

Figure 6. Configuration of the coplanar sensor and the current Daths through Pb(or Cu) phthalocyaniie film[13,1&

247

reason why the sensor with the coplanar structure shows high sensitivity and fast response. To enhance the heat-resistance of metal phthalocyanines, the polymerization of silicon and germanium phthalocyanines was investigated by Jeffery et a1.181 of Admiralty Research Establishment, U.K. Metal phthalocyanine chlorides were autoclaved to form hydrates, followed by condensated t o obtain thin polymer films. 130"Chbar PcMCl&M:Si,Ge)+PcM(OH), 2hr

400°C d(PcMO),(n250) lhr

Both polymers respond t o ppb levels of NO, at a sensitivity comparable t o that of Pb phthalocyanine. The polymers exhibit good heat-resisting characteristics and are stable even in operating in the 200-250°C temperature region. Moreover the silicon polymer exhibits a negligible response to water vapor. Another important NO, sensors using metal phthalocyanines are SAW gas sensors. SAW devices are attractive for gas sensor applications because of their high sensitivity and reliability, so that NO, sensors using SAW devices with metal phthalocyanines have been extensively studied by many researchers. Nieuwenhuizen et al. of Prins Maurits Laboratory TNO, Netherlands, have actively studied on the SAW gas sensors and as an example of the researches they reported the response of SAW gas sensors for NO,, in which different metal phthalocyanines were deposited on one delay-line of a dual delay-line oscillator[l61. Fig. 7 indicates the response of the sensors as a function of NO, concentration at 150°C. The response is defined as frequency differences divided by the thickness of a phthalocyanine layer. As shown in the Figure. Co phthalocyanine is most sensitive t o NO,, followed by Cu phthalocyanine and H,

2 m

.

1

-

1

'

1

.

1

-

H2Pc -k MgPc COPC r(

x

CUPC

A FePc,

. Figure 7. Response variations of metal phthlocyanine SAW sensors with NO2 concentration[l61.

248

phthalocyanine. Co phthalocyanine is also preferred for selectivity and Cu phthalocyanine is preferred for response time. It is concluded that the transduction mechanisms of SAW sensors are a combination of changes in mass and in conductivity caused by several chemical and physical processes. The Langmuir-Blodgett film deposition technique, a new preparation technique for metal phthalocyanine films, was applied on the fabrication of SAW sensors. Tetra-4-tert butyl Si phthalocyanine chloride films was formed on a SAW oscillator by using LB technique and the NO, sensing behavior was studied by Holcroft et a1[41 of University of Oxford, U.K. The room temperature response of the sensor coated with monolayer showed response and recovery times comparable with those reported for other phthalocyanine-based sensors operated at much higher temperature. The detection limit of the LB film sensors was found to be 40 ppb of NO,. Various kinds of metal oxide semiconductors have been investigated since SnO, was found to be able t o detect NO, sensitively by Chang[44,451. At the beginning of the researches and developments, SnO, was focused as an active material, but recently other metal oxides have been studied intensively rather than SnO, because of the problems of SnO, such as a lack of sensitivity and a drift of sensor signals. For the use of combustion monitoring, a TiO, semiconductor NO, sensor has recently been developed by Satake et al.[251 of Tokuyama Soda, Japan, and is 0.4 I

-0.8

1 0.0

I

I

1

0.8

1.2

Ta

0.4

Figure 8. Sensitivity of Ti02 semiconductor sensor doped with various elements[251. sensitivity : Sg = log (Rb/Ra) S NO a:Oppm b:200ppm(5%,02) SC3H6 a:Oppm b:500ppm(500ppm,NO + 5%,02)

249

Figure 9. Sensitivity variations of In-doped T i 0 2 semiconductor sensor with NOx concentration[251. 0.0 1

10

loo

loo0

loo00

NOx Conc. / ppm going to be used practically. The TiO, sensor has great advantages of the ability of in-situ monitoring, being stable, easy handling and maintenance free. Thesensing characteristics of the TiO, sensor was found to be further improved by doping with three valent elements such as Al, Sc, Ga and In. Fig. 8 shows the relationship between the sensitivity t o C,H, under co-existing NO and that t o NO. In the figure, the sensitivity S is defined as logarithm of the ratio of the resistance Rb in the atmosphere composed of the gas species g of the concentration b t o the resistance Ra in the atmosphere consisting of the same gas species of the concentration a. Sg = log(Rb/Ra) The sensitivities t o NO of TiO, are enhanced by adding the elements both in case of three valent elements and five valent elements. The sensitivities to C,H,, in turn, become higher by doping with five valent elements, lower with three valent elements. NO and NO, concentration dependencies of the sensitivity of TiO, doped with In are shown in Fig. 9. The sensitivity to NO of the sensor is much higher than that to NO,, and increases linearly in a wide concentration range of NO. The sensor also exhibited the fast response to NO with 90% response time of shorter than 30 seconds. These results indicate that TiO, doped with three valent elements are the promising material for in-situ detection of NO, in combustion gases. For other metal oxide sensors, WO, thick filmsr241, IT0 thin films(90%In2O,+10%Sn0,) prepared by magnetron sputtering technique[26,27], mixed oxides such as Al,0,-V,06[281 and Cr,0,-Nb,O6[291 with proper compositions and are also investigated t o utilize as NO, sensors. As for electrochemical NO, sensor, liquid electrolyte type has been developed for a long time, but these days solid type, which solid electrolyte is adopted instead of electrolytic liquid, is widely studied. Ba(NO,), was used as a solid electrolyte in the beginning of investigations[3], but there were some disadvantages like EMF drifts and necessity of reference gas. A new sensor composed of P/P"-Al,O, as a solid electrolyte and NaNO, as an auxiliary

250

Septum Argon gas

Platinum mesh

NaX (+electro. cond.)

Na-reference electrode Glass tube Nd-solid electrolyte

Figure 10. NOx sensor using Na-P/P'-A1203as solid electrolyte[311.

electrode was proposed by Hotzel et a1.[31]. of Max-Planck-Institute, F.R.G.The structure of the galvanic cell is illustrated in Fig. 10 and the cell arrangement is shown below. Pt/Na/JVa-P/P"- Al,O,/NaNO,/Pt ,NO, ,O, And the overall cell reaction is expressed in the following. Na + NO, +1/20, = NaNO, Also in the sensor, reference gas is unneeded by using Na as the solid reference electrode. The EMF of the cell t o the NO, partial pressure is in excellent agreement with the values calculated from Nernst's equation. A t 430K NO, can be measured down t o less than 0.1 ppm and the response time varies from a few seconds up to some minutes for high and low concentrations of NO,, respectively. There are, however, some disadvantages in the sensor operation. That is, the limitation in operating temperature due to the melting point of NaNO3(307"C)auxiliary electrode and large dependence on water resistance. Shimizu et a1.1321 of Kyushu University, Japan, investigated materials for an auxiliary electrode and indicated that NO, sensing performance is further improved by using Ba(NO,),-NaNO, or Ba(NO,), as an auxiliary electrode. These nitrates enable the operation a t higher temperature, the decrease of humidity dependence and fast response because of their high melting points and the water resistances. The use of the binary nitrate electrode of Ba(NO,),NaNO,, for example, made it possible to operate at temperature up to 500°C without deteriorating the sensor performance. The sensor is expected for insitu monitoring of NO, in exhaust gases from the viewpoints of stability and selectivity. The overall development of NO, sensors described above is in advance of other global environmental gas sensors. Present major problems for developing simple and handy sensors are in their durability and reliability, so that detailed studies to improve their durability and reliability should be urgently carried out.

25 1

3.2.2. SO,Sensor SO,, one of main causes of acid rain, is exhausted by the combustion of sulfur compounds in fuels. Therefore, SO, sensors are required for monitoring both in the atmosphere and in the emission from combustion facilities. Table 6 shows the present status of the research and development of SO, sensor in comparison with conventional analytical equipments and the requirement of environmental SO, monitoring[61, in the same manner as Table 6 for NO, sensors. As is seen in the table, highly sensitive SO, sensors suitable for air pollution monitoring are hardly obtained yet even in the form of prototype. Simple and handy sensors, in the present stage, are for monitoring SO, more than 1 ppm corresponding to the concentration in exhaust gases. Among them, solid electrolyte gas sensors are actively proposed and examined. In the beginning of the researches and developments, alkali-metal sulfates such as &SO,, Na,SO, and Li,SO, have been evaluated as solid electrolyte materials. Their cell potentials, however, unstable because microcracks were generated in the electrolyte due t o the phase change during thermal cycling. Difficulties with gas reference electrodes such as maintaining a constant reference gas composition were also pointed out. To overcome these disadvantages, multiphase sulfate-electrolytes[46-501 with a solid reference electrode and metal Table 6 Present status of SOXsensing 161 Concentration range of sox 1PPb 1PPm 1%

e

m

.

e

.

(

.

Requirment for: air pollution monitor 1 - 1 combustion gas H Analytical method: solution conductivity flame photometry infrared absorption H ultraviolet absorption H Handy sesors: Electrochemical type phosphoric acid K2So4 Li~S04-AgzSO4 P-Ak?03/NazS04 NASICON/NazS04 CaF2-CaS04 NazS04-Y2(S04)3 N~ZSO~-L~ZSO~-Y~(SO~)C~SiOfliS04-NiO Optical type Cu(PBza )thiophenolate H pyrenelperylene H ____

____

~____

a:Tribenzylphosphine

Accuracy (%)

0.1

1 I

10 I

ma FZza

Em Ez4 md Ea

F a

ap Dzi

mzr

eua ma

em

Dzi

nm rn

252

Two phase

-[g

e'ectrolytF

Platinum mesh

Silica tubes

/

I

Silver embeded in two phase electrolyte

Themocouple

Gas in Gold leads

Catalyst

Figure 11.'ho-phase sulfates electrolyte SOX sensorC481. oxides solid electrolytes[51,52] or a metal fluoride solid electrolyte[531 together with sulfates like Na,SO, as an auxiliary electrode have been recently proposed. Fig, 11illustrates the schematic diagram of a successful example of the twophase sulfate-electrolyte sensors developed by Worrell et al.[48] of University of Pennsylvania, USA. The cell composition can be expressed as,

Ag/Li,S0,-23mol%AgzS04/S0,,0,,S0, As shown at the top of Fig. 11,silver metal is embedded in the right-hand side of the two-phase electrolyte t o form the solid reference electrode which enable the disuse of the reference gas. The left-hand side of the two-phase electrolyte is exposed to the gas mixture containing SO,. The Ag,SO, chemical potential is the same in both sides of the two phase electrolyte, resulting no chemical interaction between the reference electrode and the two-phase electrolyte. The cell potentials for the two-phase sulfate-electrolyte sensor are in excellent agreement with the values calculated from Nernst's equation for the range of 3 to 10000 ppm of SO, concentration. The lower detection limit for SO, is determined by the thermodynamic decomposition of Ag,SO,, which is around ppm SO, in air at 803K. Pt and V,O,, oxidation catalysts, are used t o minimize the equilibration time and improve the response time of the sensor because the response time of the sensor depends upon the equilibrium reaction of S0,-0,-SO, gas mixture. Results for long-term tests showed excellent stability that the measured cell potentials(EMF) were maintained within f 3 mV of the calculated values for 100 days. Solid reference electrode, another key component of the solid electrolyte SO, sensors, was also widely investigated to disuse the reference gas and to simplify the sensor structure. S o far metal-sulfate(e.g., Ag-Ag,SO,) and oxidesulfate(e.g., MgO-MgSO,) were proposed and among them most successful solid reference electrodes are Ag-Ag,SO, mixture and NiO-NiSO, mixture. For example, the EMF of the cell using NiO-NiSO, mixture as the solid reference electrode shows good agreement with the calculated values according t o Nernst's equation for the range of 30 ppm t o 1% of SO, at 923m49,50].

253

There are some optical SO, sensors which utilize a change of photo transmittance caused by the interaction of SO, with photoactive material[54] and a quenching of fluorescence from fluorescent materials like pyreneperylene charge transfer couple due to SO, adsorption[551. These sensors, however, have rather low sensitivity to SO, and are considered not to be suitable for environmental monitoring.

3.2.3.C0, Sensor Carbon dioxide is chemically inactive so that some kind of sensors such as the semiconductor sensor is not adequate to detect CO, because the semiconductor sensor operates on the principle based on chemical reaction of gases a t the surface. Most of simple and handy sensors are based on the effect of adsorption, desorption and electrochemical reaction. In Table 7 , current status of CO, sensors are summarized together with the requirement of environmental monitoring and conventional analytical equipments[6]. Recently, electrochemical sensors, especially solid electrolyte CO, sensors have been actively studied and developed[56-631. A metal oxide solid electrolyte with an auxiliary electrode, investigated recently, enable the stable operation and the disuse of a reference gas. Table 7 Present status of CO2 sensing [6] Concentration range Accuracy of c02 (%I 1PPb 1PPm 1% 100%0.1 1 10 * . . . . ] . I I I Requirment for: air pollution monitor combustion gas Analytical mefhod: infrared absorption thermal conductivity Handy sensors: Electrochemical type P-f%03/Na2C03 NASICON/NazC03 YSZ a INASICONMazCO3 NASICONDJazCOs-Bacos Li+ conductor/LizCOs h ydroxya patitdCaClz K2COdPEG Optical type fluoresceine/F’EG LiTaOs SAW type polyethyleneimine Others mixed oxides (capacitor) zeolite (thermal conductivity) a:Y203-Zr02 b:polyethyleneglycol

n

H

H

H

H H

H H

H H H

Fza

ma ma zzan Qza

nm mp

Em

am

eza nm

Em

rm P a

Rza

rn

254

Maruyama et a1.[57] of Tokyo Institute of Technology, Japan, reported CO, sensing characteristics of galvanic cells composed of Na+ conductor like NASICON or P-A1,0,, and Na,CO, as an auxiliary electrode. The cell composition is expressed as follows. Au,C0,,0,/Na2C03/NASICON or P-Al,O,(Na,O)/O,,Au The EMF of the cell depends on the partial pressure of CO, and is good agreement with the values calculated from Nernst's equation because the activity of Na,O is constant in NASICON o r P-Al,O,. Furthermore, the simplification and the miniaturization were accomplished by using a solid reference electrode. One of the problem of the Na,CO, auxiliary electrode is the degradation of EMF response caused by a water vapor interference[59,60]. The interference occurs because Na,CO, tends t o change chemically or electrochemically t o other compounds such as NaOH, NaHCO, and Na,CO,.xH,O. New auxiliary electrodes employing binary carbonates composed of Na,CO, and alkaline earth metal carbonates was found to remarkably diminish the water vapor interference by Miura et a1.[59,601 of Kyushu University, Japan. Fig. 12 illustrates the structure of the sensor element using Na&O,-BaCO, as an auxiliary electrode. Binary carbonate was fixed to NASICON disk tightly by melting and crystallizing method. The EMF is perfectly linear to the logarithm of CO, concentration in the whole range tested(4-400000ppm) at 823K as shown in Fig. 13. It is again noted that the presence of water vapor(2.7kPa) scarcely

Inorganic adhesive

Na'conductor

Figure 12. C02 sensor using BaC03-Na2C03 as auxiliary electrode and NASICON as solid electrolyte[59,60].

255

-150

I

-250 -350

0

-650

I

I

//!

-

-550 -450

I

dry C02

\

w

I

A wet C 0 2 (2.7kPa-H2O)

Figure 13.EMF variations of the COZ sensor using BaCOs-Na2C03 as auxiliary electrode and NASICON as solid electrolyte for dry and wet C02[59,601.

/ I

I

I

1

I

affect the EMF values as seen in the figure. The solubility of BaCO, in water is extremely smaller than that of Na,CO, so that BaCO, enhances t o the water vapor resistance of the binary electrode, and in the result the sensor exhibits the excellent stability against water vapor. The rate of response and the longterm stability are also improved by using the binary electrodes. Judging the CO, sensing characteristics of solid electrolyte sensors, the sensor using the binary electrode is considered t o be closest to practical applications. A new type of CO, sensor, a capacitive type sensor consisting of BaTiO, mixed with a metal oxide, was recently proposed by Ishihara et a1.[64,65] of Oita University, Japan. The CO, sensing characteristics of the mixed capacitor were examined and the results is summarized in Table 8 . Both sensitivity and optimum operating temperature strongly depend on the kind of oxides mixed with BaTiO,. In particular, CuO mixed with BaTiO, exhibits the high sensitivity a t the CO, concentration up t o 2% so that it is suitable for air pollution monitoring. On the other hand, NiO-BaTi03 is able to detect CO, at the higher concentration up t o 20%, showing to be applicable for combustion monitoring. The optimum operating temperature for CO, sensing gives a good correlation with the decomposition temperature of carbonate corresponding t o the metal oxide mixed with BaTiO,. Capacitance change is, therefore, considered to be caused by carbonation of the metal oxide so that a high selectivity is expected. Other examples of CO, gas sensors are, polyethylene grycol-inorganic salt sensor based on the conductivity changer661, hydroxyapatite-CaC1, sensor based on the ion conductivity change[67], Ca2+exchanged zeolite sensor utilizing the thermal conductivity changeI681, fluoresceine dispersed fiber optic sensor[671 and polyethyleneimine coated SAW sensor[701. They have their own features but there are many problems to be solved for accomplishing the sensors. All of the sensors described above are chemical sensors utilizing the chemical interaction between CO, and a sensing material. In point of conventional analytical equipments, a extremely miniaturized IR sensor for CO, was recently developed by Shibata et.a1[71,721 of Sanyo Electric, Japan. Miniaturization was achieved by using a new type of pyroelectric IR detector composed of LiTaO,, a solid-state chopper and anoptical filter in a same

256

Table 8 CO2 sensing characteristics of mixed oxide capacitor [641 Oxide mixeda

Upper limit o@ detection (%I

Dielectricb constant

~~~~~~e (K)

Sensitivityd (CcwCair)

La203 Nd203 Y203

19.64 22.21 11.28 11.27 10.21

>1173g ll40 1039 823 1032

0.891 0.329 0.451 0.641 0.794

8 10 8 6 10

CeO2 PbO

22.09 66.94

934 774

0.410 0.711

8 6

NiO CUO

47.81 109.61

828 7%

0.441 2.892

20 6

zro2

17.45 171.93

915 801

0.740 0.362

10 6

Fez03

27.17

614

0.678

2

Biz03

27.83

718

0.824

2

v206

19.86 13.42

1,000 1.000

0 0

with BaTiOs CaO

MgO

c0304f

Si02

a:BaTiO3:oxide=l:l(molar ratio). b:Dielectric constant of element a t 573K in air. c:Optimum temperature for C02 detection. d:Sensitivity to 2%C02 e:A linear relationship between sensitivity and COZconcentration exists below this concentration. f:BaTiOa:C~04=3:l(molarratio). g:Higher than 1173K

package. The volume of the gas sensor is 1/4 smaller than that of the conventional IR equipment. The accuracy of measurement is within + 5 % for the concentration range of 0 to 20% CO,. It is expected the application for environmental monitoring because of high sensitivity and accuracy. 36.4. Ozone Sensor

Table 9 shows the summary of the present status of ozone sensing[61. From the view point of global environment, a major need for ozone sensing lies in the measurement of ozone concentration in high altitude of atmosphere. Especially, small and light sensors which can be mounted on a sonde or a rocket are desired. Liquid electrolyte sensors are being used now for such

251

Table 9 Present status of 0 3 sensing [61 Concentration range of 0 3

Requirment for: air pollution monitor Analytical mefhod: chemiluminescence ultraviolet absorption Handy sesors: Semiconductor type CeOz.-Inz03/SiOz Pd-SnOda-Ab03

Accuracy (S)

H

H H

c0304

Electrochemical type galvanic cell Nafionherchlobc acid

H H

purpose but are restricted to use up t o 35 km height because they utilize electrolytic liquid. Solid-state semiconductor sensors, such as Iq0,[731, SnO,r74] and Co,O,[75] sensors are expected to overcome the restriction. Among the sensors, an In,O, thin film sensor developed by Takada[73] of New Cosmos Electric, Japan, has highest sensitivity and superior response speed. The structure of the sensor is illustrated in Fig. 14. In,O, thin film is formed by physical vapor deposition. In order t o enhance the sensitivity and selectivity, CeO, is impregnated a s a form of (NH,),Ce(NO,),, followed by hexamethyldisiloxane(HMDS), which would convert t o SO,, is deposited by chemical vapor deposition. It is reported that the sensor can detect ozone below 1 ppb and is good in humidity dependence of the sensitivity and in long-term stability. HMDS-CVD thin film impregnated with (m4)2Ce(N03)6

1-03

Pt film electrodes A1203 substrate P t film heater

Figure 14. Schematic diagram of

111203 thin

film ozone sensor[73].

258

Table 10 Present status of CFC sensing [61 Concentration range of CFC lppt 1PPb 1PPm

Accuracy

(a)

1%

1..1..1...1

Requirment for: air pollution monitor 1 - 1 leak detection Analytical mefhod: gas chromatograph 1 - 1 infrared absorption Handy sesors: Halogen detector Semiconductor type S-Sn02 V-Mo-AhOdZnO LazOa-LaFs pyrolyzer+SnOz Electrochemicaltype pyrolyzer+galvaniccell

H H H

I--------I H H

H H

1 I

10 I

100

I

m

mp

m eza

Qza ma

em

E?a lGz4

3.2.6. Chlomfluorocarbon Sensor There are two major needs for chlorofluorocarbon sensing. The one is the detection of leaked chlorofluorocarbon from refrigerators, air conditioners, etc. The other is the monitoring chlorofluorocarbon concentration diffused in the atmosphere. Chlorofluorocarbon sensors being developed now are available only for the detection of the leakage because the level of the atmospheric chlorofluorocarbon concentration is too low t o be detected by the present sensors, as is seen in Table 10, which shows the present status of chlorofluorocarbon sensing[6]. Halogen leak detectors have been used widely to check the leakage of chlorofluorocarbon from refrigerators. Recently, semiconductor type chlorofluorocarbon sensors have been developed. Semiconductor materials being used are Sn0,[761,Zn0[771 and LqO,-LaF,[781. As chlorofluorocarbon is rather stable, the former two sensors are assisted by a catalyst or catalytic dopant to enhance the selectivity to chlorofluorocarbon. A variety of dopants were added to SnO, and sulfur was found t o be effective for sensitivity by Nomura et a1.[76] of Figaro Engineering, Japan. Fig. 15 shows the sensitivity t o various gases of the sensor doped with 1 mol% of sulfur together with that of the sensor without sulfur. As is seen in the figure, sulfur enhances the sensitivity to chlorofluorocarbons whereas it decreases the sensitivity to alcohols which are interference gases. It is considered that the sensitizing effect of sulfur is due t o the acceleration of SO:- species on the SnO, surface on the cleavage of C-C1 bond in chlorofluorocarbon molecule. The sensor has already been put into practice and is utilized in the detection of leaked chlorofluorocarbon and in the chlorofluorocarbon recovery system which will be described in section 4.2. Chlorofluorocarbon can be measured with usual electrochemical halogen sensors after decomposing chlorofluorocarbon with a platinum heater. Fig. 16

259 10

-

t

J

9

j

0 Without sulfur

8

With sulfur

‘ 7 8 h c,

‘5 c,

6 5 4

.H

TI 3 m 8 2 1

R-11

R-12

R-113

R-22 Ethanol n-Octane i-Propanol

Figure 15. Sensitivity of SnOz sensor doped with and without sulfur for chlorofluorocarbonsand interference gasesl761. illustrates the sensor system reported by Komiya et a1.[791 of Bionics Instrument, Japan. Sample gas containing chlorofluorocarbon is introduced into a pyrolyzer and decomposed by a heated Pt filament. The decomposed gases are detected by the electrochemical sensor. All kinds of chlorofluorobabon could be detected by the combination of the pyrolyzer and the electrochemical HF sensor. 3.2.6.0ther Sensors From the view point of former environmental hydrocarbon sensing, measurements have been made mainly on non-methane hydrocarbon because

sensor Flow meter

Power source

Pump

Figure 16. Chlorofluorocarbon sensor system composed of a pylolyzer and an electrochemical sensor[791.

260

unsaturated hydrocarbon is one of the source material of photochemical smog whereas methane is hardly related to phtochemical reaction. In the field of security, rather high concentration (several thousands of ppm) of leaked methane and propane has been needed to be detected. On the other hand, relatively low concentration (several ppm) of methane in the atmosphere becomes a problem in relation to global environment because methane makes the second largest contribution to the greenhouse effect. Unfortunately, we do not have any sensors which are simple, handy and are able to detect ppm level of hydrocarbon including methane selectively, even i n the research and development stage. Semiconductor sensors have the highest potential to detect such a low concentration of hydrocarbon among various hydrocarbon sensors being investigated at present. To improve the sensitivity and selectivity to hydrocarbon of semiconductor sensors, some proposals have been made by several researchers, for example, SnO, sensors assisted by various sensitizers[80,81] or a catalyst[82], multi-layered SnO, sensorC831 and 'y-Fe,O, thick film sensor[84]. Also for CO sensing, the present sensors are available only for the field of security not for environmental use because of the insufficient sensitivity and selectivity to monitor CO in the atmosphere. Examples of CO sensor which have been improved their sensitivity and selectivity are, for example, SnO, semiconductor sensors operated under periodic temperature cycle[85-871, a electrochemical sensor using nafion membranel881, a catalytic combustion sensor composed of catalysts and hydrophobic polymer[891, a SnO, diode sensor doped with Pd[901 and an optical fiber catalytic sensor with Au/Co,O, a s combustion catalyst[9 11. Oxygen sensors have played an important roll to control the emission of NO,, CO, and hydrocarbons from combustion facilities. Especially, electrochemical sensors using partial stabilized zirconia and TiO, semiconductor sensors have been extensively studied and had practical applications as h-sensors for automobiles. A new type of oxygen sensors, limiting current type oxygen sensors which utilize the oxygen concentration dependence of the limiting current characteristics of zirconia electrolyte cells, with wider measuring concentration range of oxygen have been developed[92-971. Practical limiting current type oxygen sensors are divided into two groups according t o gas diffusion methods, that is, pinhole type sensors[92-941 and porous layer type sensors[95-971. In the former sensors, a cover with a pinhole attached to the outside of the cathode of a zirconia cell acts as an oxygen diffusion ratedetermining device. The limiting current of the sensor depends only on the dimension of the pinhole and the ambient oxygen concentration. In the latter sensors, the porous layer attached to the outside of the cathode of the cell play the same role as the cover with the pinhole of the pinhole type sensor. The primary advantage of the porous layer type sensor is quick response because the inner space on the cathode side is very small in comparison with the pinhole type sensor. Because of many features of the limiting current type oxygen sensor such as small size, simple structure, low cost besides wide measuring concentration range, the sensor is promised t o be applied for environmental monitoring. H,S and NH, are not only poisonous but also offensive-odored so that the sensors for environmental use are required enough sensitivity comparable to the human olfaction. Many types of sensor have been proposed by many

26 1

researchers. SnO, semiconductor sensor doped with sensitizers[98] is preferred for H,S sensing because the sensor detect H,S selectively down to 10 ppb. Electrochemical sensor[99] and FET sensor[100] are preferred for NH, sensing because of their high sensitivity. 4. APPLICATIONTO POLLUTIONCONTROL 4.1. Automobilea The first practical application of oxygen sensors t o automobiles was made in 1978 for three-way catalytic converter systems by several automobile makers in Europe, Japan and US to meet Japanese or California emission standards. In these systems, the aidfuel (A/F)ratio of engine exhaust gas is detected by an oxygen sensor and controlled at an optimum value near stoichiometric A/F ratio, where NO,, CO and HC are simultaneously reduced by a three-way catalyst[lOl]. The realization of the three-way catalytic converter system gave not only a real solution for Japanese and US regulations but also the following three impacts t o automobiles. First, it showed that low emission can be compatible with low fuel consumption and good drivability, whereas the other low emission systems were difficult to. Second, it was the first introduction of microprocessors to automobiles and later developed into various computercontrolled systems i n automobiles. Third, understanding of ceramics was greatly improved because key components of the system, the catalyst and the oxygen sensor were composed of ceramics and found to be durable even in the severe condition of the automobile exhaust gas. In Japan, the three-way catalyst system is now used in most passenger cars because of such excellent performance. Most of oxygen sensors being used in this system are zirconia concentrationcell type oxygen sensors, whereas a part of oxygen sensors are titania semiconductor type oxygen sensors. The basic researches and developments of these oxygen sensors were almost completed and further researches and developments are being tried in relation to the improvements on detailed characteristics and durability. Since the second oil crisis in 1978, saving energy became the most important subject in the world and low fuel consumption vehicles were actively developed. As one of these trials, a closed loop control lean burn engine system shown in Fig.17 was developed and put into practical use by Toyota in 1984(Kimbara et a1.[102] Matsushita et a1.[103]). The system was the first real lean burn system which satisfies low fuel consumption without deteriorating drivability. In the case of Toyota 1600 cc vehicle, fuel economy was improved about 18% from 14.4kmA to 17kmA in Japanese ten mode test by introducing the lean A/F control[l02]. In this system, a zirconia limiting current type oxygen sensor developed also by Toyota (Saji et a1.[95], Kamo et a1.[1041) was installed. The sensor can detect A/F continuously t o control the A/F of exhaust gas to predetermined values in lean A/F. The limiting current type oxygen sensor has many varieties and was studied by many workers since then. They are summarized in the reference [105]. After the conquest of the second oil crisis, the trend of the automotive market turned and headed from low fuel consumption vehicles t o high-grade and highperformance vehicles. As a result, the lean burn system with the limiting

262 Sequential Injection

Swirl Control Valve

/

Distributor

Lean A/F Ratio Sensor

F I

P I

Pressuren

I

I@ Ignition,Coil

Figure 17. Automotive lean combustion system with lean A/F(oxygen) sensor[l02]. current type oxygen sensor used in only part of vehicles. Recently, as the interest in global environmental problems increased, lean combustion engine systems with lean A/F sensors began t o be actively developed again by several Japanese makers t o reduce CO, emission. One of the most important problems for the lean combustion systems t o be used widely is that the systems are applicable only for small vehicles at present because the heavier the vehicles, the more deficient in power and also the larger in the amount of harmful emission. Therefore, for the lean burn system to spread more widely, a more efficient lean burn engine has to be developed. One of approaches solving the problem is t o control the engine t o its optimum A/F ranging all part of A/F according t o the driving conditions. For this purpose, wide range A/F sensors are being developed also. Applications of oxygen sensors to systems other than AD' control are also being tried. An example is a closed loop control of the EGR rate of an engine to reduce NO, generation, where EGR (Exhaust Gas Recirculation) rate is the ratio of the mass flow rate of exhaust gas recirculated into the intake manifold to the air mass flow rate in the intake manifold. To investigate the performance of the closed-loop EGR control system, Nishida et a1 al. of Mitsubishi group used a wide range limiting current type oxygen sensor, whose output varies nearly linearly with oxygen concentration and does not depend appreciably on pressure. They showed the effectiveness of the precise control of EGR ratio to reduce NO, emission[l061. 42. Industry A major need for industrial applications of simple and handy gas sensors exists in detecting gases in the emission from industrial facilities. Gas sensors

263 Sensing element

Inlet I

Housing

Lead I

Connector

Figure 18. T i 0 2 semiconductor NOx sensor probe designed for in-situ monitoring of exhaust gases[l071. for this use are classified into two groups. The one is for sensing NO,, SO,, CO, CO,, hydrocarbons and oxygen emitted from combustion facilities like boilers o r turbines for monitoring combustion conditions and for controlling the combustion o r exhaust treatments. The other is for detecting chlorofluorocarbon, hydrocarbon, ozone, H,S and NH, generated from oil refining plants, semiconductor factories and food manufacturing factories to detect the leakage of gases and to control exhaust treatment. In both cases, fast response and selectivity to a specific gas are required for the sensors. Moreover in the former case, strong resistance against high temperature exhaust gases is peculiarly required for the sensors. Because the sensors in researches and developments hardly satisfy these requirements, irrespective of a lot of needs described above, only a few sensors have been put in a practical application, of which some examples will be given in the following from the practical standpoint. Handy NO, analyzer for in-situ exhaust monitoring has been first developed byTokuyama soda applying TiO, NO, sensor described in section 3.2.1.[251. The structure of a sensor probe designed for in-situ monitoring is shown in Fig. 18[107]. The sensor element composed of TiO, with a dopant is incorporated with a heater and a thermistor and is mounted in stainless housing so as to protect the element from particulates in exhaust gases. The thermistor is used to maintain the operating temperature of the sensor element. That is, applied voltage on the heater is controlled so as to maintain the thermistor output constantly when the temperature or the flow rate of the exhaust gas changes. Fig. 19 shows the relationship between the sensitivity of TiO, sensor and NO, concentration simultaneously measured by a chemiluminescence type NO, analyzer for the exhaust gases of various combustion facilities. In the figure, results obtained by model gas experiments(N0+5%,0 +lO%,H,O+N,) are also drawn as the solid line. As seen in the figure, the sensigvity to exhaust gases of NO, sensor is in good agreement with that t o model gas, indicating NO, can accurately detected by the sensor irrespective of the kind of combustion apparatuses and fuels. Furthermore, results of long-term operating test in the exhaust gas of a kerosene heater showed that the sensor was stable over 2000 hours. Commercialized handy NO, analyzer is shown in Fig. 20. The analyzer consists of a sensor probe and a controller. By directly inserting the sensor

264 1.0

I

'

0.8

,

0.6

-

0.4

-

(/3

Modelgas 0 Stove(Kerosene) A WaterheateraPG) 0 Boiler (Fuel oil A)

.

0.0

I

10

100

Figure 19. Correlation between sensitivities of Ti& semiconductor NOx sensor to NOx in exhaust gases of various combustion facilities and the sensitivities to NOx in model gas[lO71.

L

loo0

NOx Conc. / ppm

Figure 20. Handy NOx analyzer using Ti02 semiconductor NOx sensor.

probe into a exhaust gas, NO, concentration for the range from 10 to 3000 ppm in the exhaust gas is displayed at the controller. The analyzer is much smaller and lighter than conventional analytical equipments and needs no maintenance. A new type of liquid electrolyte gas sensors capable of simultaneously monitoring SO, and NO in flue gases at elevated temperature was constructed by Bergman et al.[108] of Health and Safety Executive, U.K. A mixture of pyroand meta-phosphoric acids as liquid electrolyte and a metallized-membrane electrode were employed to cope with the corrosive substances in flue gases. The upper operating temperature, 280°C, of the sensor is determined by the heat-resisting temperature of the gas-permeable PTFE(po1ytetrafluoroethylene) membrane. Two gold electrodes, separated by a narrow strip, were sputtered on t o the PTFE membrane. A pair of potentials of 0.70 and 0.78 V were selected to maximize the differential response to SO, and NO, while to avoid any significant interference from coexisting CO. Fig. 21 shows the response curves of the sensor t o stepwise changes in the concentration of SO, and NO. The system shows temporary cross-talk peaks, but no long-term interferences remain. A metal-sinter filter and a thermistor were also adopted

265 1200

.!l loo0

r

400ppm so2

130ppm 850ppm NO SO2

3%pm

NO

P .

Figure 21. Response variations of the liquid electrolyte sensor with step change in concentration of SO2 and NO[ 1081.

so as to protect the device from abrasive flue dusts and compensate the temperature dependence of cell responses, respectively. The accuracy for SO, and NO is zk 10 ppm or 5 % of the reading between 60 and 280"C, which cover the range of 0 to 50000 ppm with 90 % response time shorter than 1minute. A chlorofluorocarbon recovery system incorporated with SnO, sensor was proposed by Setoguchi et a1.[109] of Figaro Engineering, Japan. Fig. 22 illustrates the configuration of recovery system for R- 113(CC1F,-CC1F2),which is one of main chlorofluorocarbons used as a cleaning solution and a solvent. The recovery of R-113 is achieved by adsorbing R-113 on activated charcoal and separating it from air. In this method, however, the recovery efficiencyof R-113 decreases as the capacity of activated charcoal deteriorates. Gas sensors in the system are utilized not only for checking the capacity of activated charcoal by monitoring R-113 at the outlet of exhaust gases, but also for selecting the flow path of gases by measuring R-113 both at the inlet of gases to be treated and the outlet of separated R-113. The sensor, which was enhanced the sensitivity and selectivity to R-113 by doping with sulfur, is further improved in the long-term stability and in corrosion resistance considering the condition which the sensor contacts constantly with concentrated R-113. 43. Other Applications

In consumer's use, applications of gas sensors related to environmental pollutant gases are carried out mainly for security and amenity. Examples of the applications are, for instance, in the protection from a incomplete burning of combustion apparatus for home use and in the ventilation or condition of indoor atmosphere. A combustion monitoring sensor using SnO, was developed by Tanaka et a1.[1101 of Figaro Engineering, Japan. The gas water-heater

266 Air, R-113

Figure 22. Chlorofluorocarbon recovery system using SnO2 semiconductor sensors[l09]. R-113

configuration fitted with a SnO, sensor is shown in Fig. 23. The sensor is placed in exhaust gases and is operated by an intermittent heating cycle between 400°C for 20 seconds and 150°C for 40 seconds in order t o sensitize t o reducing gases in the emission. The resistance of the sensor is measured 39 seconds after heating to 150°C. At 150"C, the sensor resistance changed considerably with an increasing amount of reducing molecules generated by the incomplete burning. Actual tests for trial sensors in the exhaust gases of the water-heater confirmed that the sensing characteristics are quite stable for a long period of time. Applications of simple and handy sensors under research and development t o atmospheric air monitoring are considered t o be restricted because of the lack of sufficient sensitivity. Only an electrochemical sensor used for the monitoring of ozone in high altitude atmosphere and an 1%03semiconductor sensor for monitoring ozone in open air have had practical application. Although conventional analytical equipments are being utilized i n this field, requirements for the application of simple and handy sensors are getting increased. The sensors feasible for atmospheric air monitoring, therefore, are chosen among a variety of sensors in the research and development stage. As for NO, sensing in the atmosphere, semiconductor sensors and SAW sensors using metal phthalocyanines are most feasible owing t o their excellent sensitivities t o NO,. Solid electrolyte sensors and capacitor type sensors are most preferable for detecting CO, in air because of their sufficient sensitivity. The most important key point to put them t o practical use is considered to be in reliability and durability.

267

aT+

Exhaust gas

Water

+-

FT1 Air

1:Sensor 2: Exhaust gas assembler 5: Heat absorber

Gas

Figure 23. Water-heater equipped with SnO semiconductor sensor[ll0].

5. FUTURESCOPE

In the above sections, the state of the developments of gas sensors related t o environmental problems have been discussed. The final subject which should be discussed will be that how the gas sensors will play a role in satisfying the growing needs of gas sensing related to environmental problems in future. As for the first need of gas sensing, that is, grasping the actual circumstances of the air pollution, highly precise and selective measurements such as those of the decomposed or reacted products of chlorofluorocarbon in the stratosphere will progress by means of the improvement of analytical instruments rather than sensors as before. Here, we distinguish 'sensor' from 'analytical instrument' by that the 'sensor' means handy and small scale sensing device although it is difficult to draw a definite boundary line between them. On the other hand, sensors will play an important role on multiple and wide area measurements such as NO, and CO, distribution in a specified region. The second need is the routine watching of pollutant gases in the atmosphere or in combustion exhaust gases. In the present, analytical instruments are mainly used for this purpose. Handy and low-cost sensors, on the other hand, are being developed actively and will probably be used widely in place of conventional analytical instruments in future. It is likely that the handy sensors will expand their use and extend into a place where the gas detection has never been made before. For the third need of gas sensing, that is, the closed-loop control of exhaust gas, realization of handy and tough gas sensors are especially expected. Real closed-loop control systems installed on boilers or combustion engines will never be realized without sensors. Generally speaking, however, it will take several years until the realization of the practical application of gas sensors for this use with the exception of a few kind of sensors because the durability of the sensor in exhaust gas is difficult to attain.

268

The next subject is what types of sensors will be promising in future to satisfy the various needs mentioned above. Unfortunately we do not know the definitive type of sensors yet, which will surely satisfy the above needs and will be superior to any other type of sensors in future. From the standpoint of 'practical application within two t o five years', semiconductor type and electrochemical type (both solid and liquid) will still maintain the leading position among various types of gas sensors. As for the research and development of gas sensors, new types of sensors such as FET, junction, optical, SAW and capacitive type gas sensors will be studied more and more actively. Of these new types of gas sensors, FET, junction, SAW and capacitive type sensors operate based on a mechanism common with the semiconductor type gas sensor. That is; the adsorption and desorption of gas molecules occur on the surface of sensing materials at the first step of sensing processes and finally the electrical or electro-acoustical properties of the sensing materials changes with gas concentration. Therefore, these types of sensors have common problems such as rather poor selectivity and necessity of heating, which essentially come from adsorption and desorption process. These new types of sensors, however, have their own advantages. For example, both the junction and FET type sensors are very sensitive and the SAW type sensor has frequency output, which is adequate for digital signal processing. Future success of these new types of sensors will depend on how their advantages can surpass their weak points. The optical type gas sensor is in different situation from other new type of sensors described above. There are two types in the optical type gas sensors. One is the sensor based on an interaction between optical material and objective gases, where adsorption and desorption process play an important role too. The key point of the development of this type of sensor is in finding and utilization of new materials. The other type of optical gas sensor is based on the principle which measures the characteristics of objective gases directly with optical means. This type of optical sensor has the following special features which many of other type gas sensors do not have. The first one is that it is possible to operate without contacting with objective gases directly, which means i t does not necessarily have the problems originated from adsorption and desorption. The second is its high selectivity, which comes from the ability of utilizing spectrum information. The third is that an optical sensor which operates without any aid of electricity is possible to be fabricated if necessary. Thus the latter type optical sensors seem to be very attractive and in fact optical methods are used successfully in conventional analytical instruments although they are too large i n size and complicated t o apply to sensors a t present. Trials to miniaturize the infrared spectrometer have been made actively as described in previous sections. These efforts are bringing the image of analytical instruments closer to that of sensors. Further drastic simplification, in future, will realize handy optical sensors, which are completely different from the concept of the analytical instrument. As for the utilizing technique of sensors, multi-sensor systems composed of several gas sensors and multi-signal processing circuits can greatly enhance the selectivity even if the individual sensor has low selectivity. Furthermore, the instability of sensors is also possible to be compensated or reduced by the systems. Many efforts have been made for investigating multi-sensor systems. At the present level of the research and development, the identification of a gas species in simple gases is possible but that in mixed gases is difficult in

269

laboratory experiments. The practical application of these trials will be realized at first in a use where the objective gases exist to be easily distinguishable. The combination of the technique of the multi-sensor system with thin or thick film- and micromachining- technology makes it possible to produce array sensors which enable us to get a sensing system being drastically miniaturized and having the same functions as the multi-sensor systems. The array sensors are ideal sensors in a sense. There are, however, the following several problems which have to be solved to realize the practical application. (1) In principle, the array sensors are promising to attain the high selectivity and stability, but it is not sure what extent it can. (2) Generally speaking, less durability is a weak point of the sensors fabricated with thin-film and micromachining technique. (3) The cost of the array sensors are expected to be very low when they are mass-produced. It, however, is really difficult to produce low cost sensors at present because of being in the dilemma of low yield, huge investment and not large quantity. To solve the problem, great progress and maturity will be necessary not only in the research and development of the array sensor itself but also in the manufacturing technique of IC and the related fields. The goal seems to lie far away, but the image of the sensor will completely be changed if the above problems are solved and the practical array sensors are realized.

1 Seiyama T, Kato A, Fujiishi K, Nagatani M. Anal. Chem. 1962; 34: 1502-1503. 2 Taguchi N. Japan Patent 45-38200. 1977; 1 2 4 1579-1583. 3 Gauthier M, Chamberland A. J Electrochem. SOC. 4 Holcroft B, Roberts GG. Thin Solid Films 1988; 160: 445-452. 5 Lundstrtim I. Sensors and Actuators 1981; 1:403-426. 6 Takeuchi T. Proc. Fukuoka Int. Symp. '90 on Global Environment and Energy Issues, Fukuoka, Japan, 1990; 327-333 7 Mockert H, Schmeisser D, Gopel W. Sensors and Actuators 1989; 19:159-176. 8 Jeffery PD, Burr PM. Sensors and Actuators 1989; 17: 475-480. 9 Temofonte TA, Schoch KF. J Appl.Phys. 1989; 65: 1350-1355. 10 Sadaoka Y,Jones TA, Revel1 GS, Gopel W. J Mater. Sci. 1990; 25: 5257-5268. 11 Sadaoka Y,Jones TA, Gopel W. Sensors and Actuators 1990; B1: 148-153. 12 Hamann C, Kampfrath G, Mueller M. Sensors and Actuators 1990; B1: 142- 147. 13 Ruihua w, Jones TA. Sensors and Actuators 1990; B,12: 33-42. 14 &in SJ,Bott B. Sensors and Actuators 1991; B,3: 255-260. 15 Cranny AWJ, Atkinson JK, Burr PM, Mack D. Sensors and Actuators 1991; B,4: 169-174. 16 Nieuwenhuizen MS, Nederlof AJ, Berendsz AW. Anal. Chem. 1988; 60: 230-235. 17 Nieuwenhuizen MS, Nederlof AJ, Vellekoop MJ, Venema A. Sensors and Actuators 1989; 19: 385-392. 18 Rapp M, Binz D, Schickfus MV, Hunklinger S, Fuchs H, Schrepp W, Fleischmann B. Sensors and Actuators 1991; B,4: 103-108. 19 Nagashima K, Meguro K, Suzuki S, Hobo T. Bunseki Kagaku 1988; 37:

270

400-404. 20 Hanawa T, Kuwabata S, Yoneyama H. J Chem. SOC.,Faraday Trans. 11988; 84: 1587-1592. 21 Gopel W, Schierbaum D, Schmeisser D, wiemhofer. Sensors and Actuators 1989;17: 377-384. 22 Sberveglieri G, Groppelli S, Nelli P, Lantto V, Torvela H, Romppainen ,€' Leppavuori S. Sensors and Actuators 1990; B1: 79-82. 23 Sberveglieri G, Groppelli S, Nelli P, Sensors and Actuators 1991; B,4: 457-461 24 Akiyama M, Tamaki J , Miura N, Yamazoe N. Digest 13th Chemical Sensor Symp., Nagoya, Japan, 1991; 133-136. 25 Satake K, Kobayashi A, Inoue T, Nakahara T, Takeuchi T. Proc. 3rd Int. Meeting Chemical Sensors, Cleveland, USA, 1990; 334-337. 26 Sberveglieri G, Groppelli S, Coccoli G . Sensors and Actuators 1988; 15: 235-242 27 Sberveglieri G, Benussi P, Coccoli G, Groppelli S, Nelli P, Thin Solid Films 1990; 186: 349-360. 28 Ishihara T, Shiokawa K, Eguchi K, Arai H. Sensors and Actuators 1989; 19: 259-265. 29 Ishihara T, Shiokawa K, Eguchi K, Arai H. Chemistry Letters 1988; 997-1000. 30 Chang SC, Stetter JR.Electroanalysis 1990; 2: 359-365. 31 Hotzel G, Weppner W. Sensors and Actuators 1987; 12: 449-453. 32 Shimizu Y, Okamoto Y, Yao S, Miura N, Yamazoe N. Denki Kagaku 1991; 59: 465472. 33 Buttner WJ, Maclay GJ, Stetter JR.Sensors and Actuators 1990; B1:303-307. 34 Eguchi K, Hashiguchi T, Sumiyoshi K, Arai H. Sensors and Actuators 1990; B1: 154-157. 35 Schoch KF, Temofonte TA. Thin Solid Films 1988; 165: 83-89, 36 Zhu DG, Petty MC, Harris M. Sensors and Actuators 1990; B,2: 265-269. 37 Peschke M, Hansch W, Lechner J , Lorenz H, Riess H, Eisele I. Sensors and Actuators 1991; B,4: 157-160. 38 Kolesar ES Jr, Wiseman M. Anal. Chem. 1989; 61: 2355-2361. 39 Willis MR, Markland KJ, Fahy MR. Synthetic Metals 1989; 28: C781-C786. 40 Sadaoka Y, Yamazoe N, Seiyama T. Denki Kagaku 1978; 46: 597-602. 41 Sadaoka Y, Sakai Y, Aso I, Yamazoe N, Seiyama T. Denki Kagaku 1980; 48: 486-490. 42 Bott B, Jones TA. Sensors and Actuators 1984; 5: 43-53. 43 Jones TA, Botto B. Sensors and Actuators 1986; 9: 27-37. 44 Chang SC. B E E Trans. Electron Devices 1979; ED-26: 1875-1880. 45 Chang SC. SAE paper 1980; 800537. 46 Lui QG, Worell WL. Phys. Chem. of Extractive Metallurgy, eds. Kudryk V, Rao Y. (The Metal. SOC., AIME< Warrendale, PA) 1985; 387. 47 Lui QG, Worell WL. Solid State Ionics 1988; 28-30: 1668-1672. 48 Mali CM, Beghi M, Pizzini S, Faltemier J. Sensors and Actuators 1990; B,2: 51-55. 49 Imanaka N, Yamaguchi Y, Adachi G, Shiokawa J. J Electrochem. SOC. 1987;134: 725-728. 50 Imanaka N, Kawai K,Shiokawa J, Adachi G. Chemistry Letters 1988; 11351136. 51 Akila R, Jacob KT. J Appl.Electrochem. 1988; 18: 245-251.

27 I

52 Akila R, Jacob KT. Sensors and Actuators 1989; 16: 311-323. 53 Jacob KT, Iwase M, Waseda Y. Solid State Ionics 1987; 23: 245-252. 54 Cook RL, Macduff RC, Sammells AF.Analytica Chimica Acta 1989; 226: 153-158. 55 Sharma A, Wolfbies 0s. SPIE 1988; 990: 116-120. 56 Ogata T, Fujitsu S, Miyayama M, Koumoto K, Yanagida H. J Mater. Sci. Lett. 1986; 5: 285-286. 57 Maruyama T, Sasaki S, Saito Y. Solid State Ionics 1987; 23: 107-112. 58 Liu J, Weppner W. Solid State Communications 1990; 76: 311-313. 59 Yao S, Shimizu Y, Miura N, Yamazoe N. Chemistry Letters 1990; 2033-2036. 60 Miura N, Yao S, Shimizu Y, Yamazoe N. Proc. 6th Int. Conf. Solid-State Sensors and Acutuators, Sanfrancisco, USA, 1991; 558-561. 61 Watabe K, Sasaki T, Ono T, Maruyama T. Proc. 6th Int. Cod. Solid-state Sensors and Acutuators, Sanfrancisco, USA, 1991; 1002-1005. 62 Imanaka N, Kawasato T, Adachi G. Chemistry letters 1990; 497-500. 63 Imanaka N, Kawasato T, Adachi G. Chemistry letters 1991; 13-16. 64 Ishihara T, Kometani K, Hashida M, Takita Y.J Electrochem. SOC. 1991; 138: 173-176. 65 Ishihara T, Kometani K, Mizuhara Y, Takita Y. Sensors and Actuators 1991;B,5: 97-102. 66 Shimizu Y, Komori K, Egashira M. J Electrochem SOC. 1989; 136: 2256-2260. 67 Nagai M, Nishino T, Saeki T. Sensors and actuators 1988; 15: 145-151. 68 Ishiguro Y, Nagawa Y, Futata H. Proc. 2nd Int. Meeting Chemical Sensors, Bordeaux, France, 1986; 719-722. 69 Kawabata Y, Kamichika T, Imasaka T, Ishibashi N. Analytica Chimica A ~ t a1989; 219: 223-229. 70 Nieuwenhuizen MS, Nederlof AJ. Sensors and Actuators 1990; B,2: 97-101. 71 Yokoo T, Shibata K, Takeuchi K, Tanaka T, Kamino M, Nishikawa S, Nakano S,Kuwano Y. Proc. 4th Int. Cod. Solid-state Sensors and Acutuators, Tokyo, Japan, 565-568. 72 Shibata K, Takeuchi K, Tanaka T, Kamino M, Nishikawa S, Kuroki K, Nakano S, Kuwano Y. Ferroelectrics 1989; 95: 117-120. 73 Takada T, Komatu K. Proc. 4th Int. Cod. Solid-state Sensors and Acutuators, Tokyo, Japan, 1987; 693-696. 74 Takashima Y, Ogino K, Futata H. Digest 6th Chem. Sensor Symp., Tokyo, Japan, 1987; 3-4. 75 KogaO, Hori Y, Suzuki S. Nippon Kagaku Kaishi 1987; 147-151. 76 Nomura T, Matsuura Y, Takahata K, Matsuura S. Digest 6th Chem. Sensor Symp., Chiba, Japan, 1990; 13-16. 77 Shiratori, Matsura M, Tsuchiya T. Proc. 1st Int. Meeting Chemical Sensors, Fukuoka, Japan, 1983; 119-124. 78 Cao 2, Stetter JR,Proc. 3rd Int. Meeting Chemical Sensors, Cleveland, USA, 1990; 196-900. 79 Komiya H, Kimura S. Sensors and Acutuators 1990; B1: 68-72. 80 Komori N, Sakai S, Komatsu K. Proc. 4th Int. Cod. Solid-state Sensors and Acutuators, Tokyo, Japan, 1987; 591-595. 81 Coles GSV, Williams G, Smith B. Sensors and Actuators 1991; B,3: 7-14. 82 Portnoff MA, Grace R, Guzman AM, Runco PD,Yannopoulos LN. Proc. 3rd Int. Meeting Chemical Sensors, Cleveland, USA, 1990; P96-P97. 83 Shimizu Y, Morimoto T, Egashira M. Proc. 3rd Int. Meeting Chemical

272

Sensors, Cleveland, USA, 1990; P120-P121. 84 Lee DD, Choi DH. Sensors and Actuators. 1990; B1: 231-235. 85 Murakami N, Takahata K, Seiyama T. Proc. 4th Int. Conf. Solid-state Sensors and Acutuators, Tokyo, Japan, 1987; 618-621. 86 OyabuT, Matsuura Y, Murai R. Sensors and Actuators. 1990; B1: 218-221. 87 Watson J , Davies G. Sensors and Actuators 1990; B,2: 219-222. 88 Otagawa TI Madou M, Wing S,Rich-Aiexander J , Kusanagi S, Fujioka TI Yasuda A. Sensors and Actuators, 1990; B1: 319-325. 89 Marcinkowska K, Macgauley MP, Symons EA. Proc. 3rd Int. Meeting Chemical Sensors, Cleveland, USA, 1990; 184-187. 90 Matsushima S, Miura N, Yamazoe N. Chemistry Letters 1987; 2001-2004. 91 Kobayashi T, Haruta M, Tsubota S, Sano H, Delmon B. Sensors and Actuators 1990; B1: 222-225. 92 Usui T, Nuri K, Nakazawa M, Osanai H. Jpn. J Appl. Phys. 1987; 136: L2061-L2064. 93 Usui TI Asada A, Nakazawa M, Osanai H. J Electrochem. SOC.1989; 136: 534-542. 94 Asada A, Yamamoto H, Nakazawa M, Osanai H. Sensors and Actuators 1990; B1: 312-318. 95 Saji k, Takahashi H, Kondo H, Takeuchi T, Igarashi I. Proc. 4th Sensor Symp.,Tsukuba, Japan, 1984; 147-151. 96 Takahashi H, Saji k, Kondo HI Takeuchi T, Igarashi I. Proc. 5th Sensor Symp.,Tsukuba, Japan, 1985; 133-137. 97 Kondo H, Takahashi HI Saji k, Takeuchi T, Igarashi I. Proc. 6th Sensor Symp.,Tsukuba, Japan, 1986; 251-256. 98 Nakahara T, Takahata K, Matsuura S. Proc. Symp. Chemical Sensors, Electrochem. SOC.,Hawaii, USA, 1987; 87-9: 55-64. 99 HCifele E, Kaltenmaier K, Schonauer U. Sensors and Actuators 1991; B,4: 529-531. 100 Lundstrom I, Spetz A, Winquist F, Ackelid U, Sundgren H. Sensors and Actuators 1990;B1: 15-20. 101 Zechnall R, Baumann G, Eisele H. SAE paper 1973; 730566. 102 Kimbara Y, Shinoda K, Koide HI Kobayashi N. SAE paper 1985; 851210. 103 Matsushita S, Inoue T, Nakanishi K, Kato K, Kobayashi N. SAE paper 1985; 850044.

104 Kamo T, Chujo Y, Akatsuka T, Nakano J , Suzuki M. SAE paper 1985; 850380. 105 Takeuchi T. Sensors and Actuators 1988; 14: 109-124. 106 Nishida M, Suzuki HI Inoue N, Kumagai S. SAE paper 1988; 880133. 107 Inoue TI Satake K, Kobayashi A, Ohkoshi HI Nakahara T, Takeuchi T. Digest 12th Chemical Sensor Symp., Noda, Japan, 1991; 129-132. 108 Bergman I, West KO. Proc. 2nd Int. Meeting Chemical Sensors, Bordaeux, France, 1986; 735-738. 109 Setoguchi Y, Nomura TI Amamoto T,Tanaka K, Mtsuura S. Digest 11th Chemical Sensor Symp., chiba, Japan, 1990; 17-20. 110 Tanaka K, Morimoto S, Sonoda S, Matsuura S, Moriya K, Egashira M. Sensors and Actuators 1991; B,3: 247-253.

273

PIEZOELECTRIC CRYSTAL DETECTORS FOR ENVIRONMENTAL POLLUTANTS A. A. Suleimana and G. G. Guilbaultb 'Department of Chemistry, Southern University, Baton Rouge, LA 70813, United States bDepartment of Chemistry, University of New Orleans, New Orleans, LA 70148, United States INTRODUCTION

The phenomenon of piezoelectricity was first discovered by the Curie brothers in 1880 (1). They proposed that when compressed, some crystals show positive and negative changes proportional to the pressure on certain portions of their surfaces. In the following year, it was asserted on the principles of conservation of electricity that a converse effect should exist. The Curies verified the same conclusion by showing that a quartz crystal is strained when a voltage is applied to it. Piezoelectricity remained a scientific curiosity until the war in 1914, when the Ilecho method" utilizing piezoelectric quartz plates was developed to explore the ocean bottom. The development of other devices including microphones, loud speakers, phonograph records and pickups using crystals as control elements followed. In later days, the art of crystals developed to become one of the cornerstones of the network of communication systems. Alpha quartz (low-quartz) is most often used for piezoelectric crystal detectors. It crystallizes at temperatures below 573OC, and transformes to beta-quartz (high-quartz) between 573'C and 870OC. Only AT and BT cut quartz plates are useful. The term IIcutlt designates the direction of the normal to the major faces. A standard AT cut, describes a plate with thickness in the y direction rotated 35' 15' counterclockwise about the X-axis Fig.1 (2).

2 B

OT --

CT,DT

0T

X Figure 1. Angular orientation of some special cuts.

274

The use of piezoelectric (PZ) devices as potential analytical chemistry sensors was not realized until Sauerbrey (3) described the relationship between the resonant frequency of an oscillating piezoelectric crystal and the mass deposited on the crystal surface. The relationship between the weight of an evenly distributed film of a metal and the resonant frequency of an AT cut crystal vibrating in the thickness shear mode can be expressed by: AW AF

= -2.3

X

lo6 F2 A

where: AF = Frequency change due to metal coating in Hz. F = Frequency of quartz crystal in MHZ. AF = The weight of deposited film in grams. A = Area of quartz plate in cm2 The above equation predicts that a commercially available 9 MHz crystal with two electrodes 5 mm in diameter would have a sensitivity of 475 Hzlvg, and a 15 MHz crystal a sensitivity of 1318 Hz/pg. The detection limit was estimated to be 1O-I2g ( 4 ) . King (5) reported that the first crystal sorption detector was the IIMartini Detector", a fun project he started in the basement of his house. He coated the crystals in his ham radio transmitter with the insect repellant (di-n-butylphthalate), and found that the frequency shifted more when the coating was exposed to alcoholic breath than to normal breath. Out of that fun project came the pioneering work of King in the development of the piezoelectric sorption detector. The high sensitivity, simplicity, and ruggedness of the piezoelectric sorption detector led to a variety of applications in air pollution research. EXPERIMENTAL A block diagram of the simplest typical experimental setup is shown in Figure 2. The piezoelectric crystal is housed in the detector cell and driven by a low frequency transistor oscillator. The frequency of the crystal is monitored by a frequency counter modified by a digital-to-analog converter. The change in the frequency of the vibrating crystal can be monitored directly from the frequency counter or can be translated to a permanent record. Subsequent improvements in the design include incorporating two crystals in the detector cell, a reference and a sampling crystal. Although, a number of piezoelectric crystal based devices for different applications are commercially available, Universal Sensors

275 P -

8

-

7

P

offers the only device (Figure 3) for pollution monitoring, coating materials for several common air pollutants.

Figure 3. Universal Sensors PZ 105 gas phase detector.

and

276 Coating Methods

Several techniques have been used for the application of the substrate to the piezoelectric crystal. These include dropping, dipping, and spraying methods. In the dropping method, the substrate is dissolved in a volatile solvent, and a known amount of solution is placed at the center of the crystal electrodes using a microsyringe, then spread evenly. The solvent is then allowed to evaporate leaving a dry thin film of coating on the electrodes. In the dipping method, the crystal is usually dipped into the solution of the substrate then allowed to dry. In the spraying method, the crystal surface is sprayed with the substrate solution then allowed to dry. The most critical factor in the coating process seems to be the ability to reproduce the operation, where the same amount of coating is applied on the crystal in the same manner and covering approximately the same area. Most recently, the use of LangmuirBlodgett (LB) technique to coat crystals have been reported, and showed good promise. Dilution Methods

In earlier studies, most dilutions were made using the syringe dilution method. A known amount of pure gas is usually withdrawn in a gas tight syringe from a lecture bottle or purified compressed gas, and then diluted with air. The mixture is allowed to equilibrate by diffusion, then the process is repeated to obtain the desired concentration. Although this method is simple and fast, the inaccuracy in the procedure is very obvious, compounded with contamination, and adsorption on the syringe walls between dilutions. Dilution can also be accomplished using a dilution flask or a bag. The container is first evacuated, then filled with a carrier gas, and a known volume of the pure gas is injected. The mixture is stirred sufficiently to obtain a homogeneous diluted mixture. The process may be repeated if very dilute mixtures are required. Barring any adsorption on container walls, this method is more reproducible and more accurate than the syringe dilution method. Alternatively, the use of permeation devices has been adopted in later studies for the generation of more accurate gas mixtures. In addition, sample introduction via injection valves has replaced the syringe injection method. Aerosol Measurement

The detection of airborne particles using piezocrystal devices was first reported by Chuan (6,7) who used an adhesive coated 10 MHz crystals to detect aluminum oxide particles and mass concentration as low as 1 g/m3 were detected. Particles were impacted by A aerodynamics acceleration with reported sensitivity of 10-"g. portable continuous, direct reading instrument for measuring

277

particulate mass from 5-5,000 pg/m3 with a precision of 15 pg/m3 was subsequently developed (8). Olin et a1 (9) conducted several preliminary experiments to verify the feasibility of quartz crystal microbalance for mass concentration measurement of suspended particles. It was concluded that assuming perfect bonding of the particles to the crystal surface, the experimental and theoretical mass sensitivities (for example, 5.5 Hz pg-l) were identical. Daley and Lundgren (10) evaluated the performance of various instruments for aerosol measurement. The effects of several parameters including temperature, humidity, particle collection characteristics, linearity and mass sensitivity were studied. The observed linear response limits ranged from 0.2 to 6 pg/mm*, and no relationship between linearity limit and point of complete saturation was apparent. The mass sensitivity was found to be dependent upon deposit size and location and decreased for particles size beginning at 2 pm in diameter reaching essentially zero at 20 pm: However, viscous coating appeared to improve sensitivity in this range. The characteristics of several QCM instruments for aerosol measurement have been reviewed (11). Particles are collected by impaction, electrostatic precipitation or both. The mass sensitivity is reported to be affected by the location of deposited particles on the crystal, the size of the particles, and the type of coating. In addition, the sensitivity changes as the crystal becomes loaded. Despite some limitations, most of the studies indicated that QCMs can be successfully used for aerosol measurement with good correlation coefficient with the reference filtration method. Applications included measurement of aerosol in ambient air, particulate emission from automobiles and diesel engines, smoke plume from a coal-fired power plant, solid fueled rocket plume, and particulate matter in the effluents in combustion sources.

-

Qas Detection systems Acetoin

Acetoin, 3-hydroxy-2-butanoneI is present in many foods and drinks. It is a fermentation product of bacteria, degrading 2,3-butanediolI certain fungi degrading sugar can juice, and also formed by the action of yeast in diacetyl. An acetoin detector could be used in law enforcement area, for the detection of contraband at points of entry from abroad. A PZ detector for acetoin based on the coating tetrabutylphosphonium chloride was developed by Suleiman et a1 (12). The response of the detector was linear over the concentration range 8-120 ppb of acetoin. Complete response with maximum sensitivity, and at shorter sampling time was obtained at a flow rate of 270 ml/min. About 95% reversibility was obtained in 10 min at this high flow, and in 20 min at flow rate of 130 ml/min. However, significant response was obtained after a sampling time of 3 0 seconds, and at a flow rate of 130 mllmin with complete reversibility accomplished in less than 5 min. The coating did not respond to various potential organic interferences with only

218

response to acetone and ethanol, however at very high concentrations. In another study (13), a semicarbazide coating was prepared by dissolving 100 mg of semicarbazide hydrochloride in 10 ml of 1:l acetonejethanol, followed by the addition of 5 ml of 6 M ammonia solution. The clear solution was allowed to stand to ensure neutralization of the hydrochloric acid before application to the crystal electrodes. The coated 9 MHz crystal was driven at its third overtone of 27 MHz. The detector had a sensitivity of 12.4 Hz p1-l and a linear response in the concentration range 5.50 pg 1-l. the highest response was obtained in a dry nitrogen stream at a flow rate of 90 ml min-l, and decreased to about 33% in a wet air stream (58% relative humidity) at a flow rate of 200 ml min-l. The relative sensitivities to the potential intererences studied (chloroform, acetaldehyde, ethanol, acetone, n-octanol, ethylbutyrate and n-hexylacetate) ranged from 0.000000 to 0.000029 compared to a relative response of one to acetoin. The response decreased to about 88% at 40% humidity, then decreased by about 22% between 90-95% humidity. A t 97% humidity level, the detector was overloaded by moisture and became useless.

Amines The use of various metal salts such as FeC13, ZnC12, HgBr2, CoC12 and ZnI2 as coatings for the detection of mono-, di- and trimethylamines was investigated (14). The coated crystal was attached to a vacuum line, and its frequency was measured after the system was evacuated to torr. The gas trapped in a 1-liter glass storage bottle at 200 torr was then allowed to react with the coating, and the reaction was evaluated at room temperature with respect to time and sensitivity. It was found that FeC13 adsorbs amines 2-4 times more than the other salts, and adsorption was greatest for trimethylamine. The reported sensitivity is 369 Hz pg-' for 13.23 - 15.45 pg of coating.

Ammonia Ammonia and nitrogen dioxide were determined using a coated crystal with Ucon 75-H-90,000 and Ucon-LB-300-X after conditioning the coating in a NO2 containing atmosphere (15). The response to ammonia was linear in the concentration range 1 ppb to 1 ppm. Ascorbic acid and the extract of Capiscum annuum pods were also used as coatings (16). The alcohol extract was the most sensitive, with a frequency response of 120 Hz for 10 ppm to 50 Hz for 0.001 ppm of ammonia. The chloroform extract gave 45-15 Hz frequency response for the same concentration, while no response was obtained with the toluene extract. The addition of silver nitrate to the coating greatly increased sensitivity, suggesting that ascorbic acid is probably the active component in the extract. Hence, ascorbic acid was tested as a coating, and greater response was obtained. Similarly, the addition of silver nitrate to the coating increased the response. A coated crystal with silver nitrate alone was also

219

very sensitive to ammonia; however, the results obtained were erratic. The infrared study suggested that the reaction of ammonia with ascorbic acid was a simple acid-base reaction and occurred at the solid surface. L-glutamic acid-HC1 and pyridoxine-HC1 were reported to have greater sensitivity and better selectivity than the ascorbic acid silver nitrate coating (17). According to XR spectra, a reaction between the carboxyl group and ammonia could have occurred yielding the ammonium salt of the carboxylic acid. The response time for both coating was less than 1 min and complete reversibility occurred in 5 min. No significant interferences were reported, and moisture was eliminated by a precolumn of silica gel. The pyridoxine HC1 coating exhibited sensitivity down to parts per trillion range. A plot of the logarithm of change in frequency vs. the logarithm of concentration was linear over the concentration range of 0.01 ppb to 1 ppm. since the pyridoxine - HC1 is very sensitive to ammonia, it saturates if exposed to higher concentration of ammonia, and poor reproducibility is obtained. The role of a nonylphenoxypolyethoxylate (Antarox CO-880) as a support matrix for pyridoxine - HC1 coating has been investigated (18-20). The addition of the polymer prolonged the useful lifetime of the coating to 10-53 d. The interference profiles were generally similar to those for pyridoxine-HC1 alone, except for hydrogen chloride gas which gave a frequency response of 1496 Hz for a 109 mg dm-3. The increase in sensitivity to HC1 was attributed to hydrogen bonding between HC1 and the ethoxylate oxygen of the Antarox CO-880. The use of polyvinylpyrrolidine (PVP) as a sensitive coating for the determination of ammonia in the range 6 ppb to 10 ppm was also reported (21). The PVP coated crystal was part of a multi sensor sysem containing a silver chloride coated crystal for water, and a temperature sensor. The optimum flow-rate was found to be 50 ml min-', with equilibrium established after 15 min.

-

Carbon Dioxide

The use of piezoelectric crystals for measuring dissolved C02 was investigated (22). Crystals were coated with didodecylamine (DDDA) or dioctadecylamine (DODA) and placed in a chamber separated from the sample solution by a Teflon membrane, a Polyvinyl chloride membrane, or a Fluoropore filter with polyethylene web backing. The DDDA was 33 times more sensitive to H20 vapor than to C02. The membranes were not sufficient for a selective measurement of C02, and a differential mode of measurement was used to circumvent H20 vapor interference. A carbon dioxide sensor for monitoring levels in a closed exhalation anesthesia system was developed by Jordan (23). 7,IOdioxa-3,4-diaza-1,5,12,16-hexadecatetrol was prepared by mixing monoethanolamine, often used as a C02 scrubber, with ethylene gylcol diglycidyl ether in a 2:l ratio and was used as a coating. The respone was 391 Hz for 10% C02, exposure time less than 30 sec, and complete recovery in less than 60 sec. No interferences were reported from nitrous oxide, halothane, or oxygen tested at normal anesthesia concentrations.

280

A piezoelectric crystal coated with tetrakis(hydroxyethy1)ethylenediamine was proposed as a monitor for C02 in fermentation processes (24). The response curve was linear over the concentration range 1.8 - 16% (v/v) COz. Complete reversibility was obtained in 2-5 minutes at sampling times of 3-30 sec for a 10% (v/v) C02 sample, and varied from 2-5 min in the linear portion of the calibration curve at a sampling time of 5 sec. Potential interferences were eliminated by passing the sampling stream through a column of molecular sieve 3A. The proposed sensor was shown to have real time monitoring capability, and could be used for more than 15 days without significant loss in sensitivity. Carbon Monoxide A piezoelectric crystal detector for carbon monoxide in ambient air was developed by Ho et a1 (25). Carbon monoxide was first reacted with mercuric oxide at 210'c to produce mercury vapor, which can be amalgamated by the gold electrodes of the crystal, increasing its mass and thereby decreasing the frequency proportional to CO content of the air. A reference stream through a silver oxide column was used to eliminate the background due to thermal decomposition of mercuric oxide. After each measurement, mercury was desorbed from the gold by heating the crystal up to 170'C in a hot purified air or nitrogen atmosphere for 10 min. organic interferences could be removed effectively using activated charcoal or molecular sieve 5A. Sulfur dioxide, the only interfernce from inorganic gases can be also easily removed by a molecular sieve precolumn, and moisture was removed by calcium chloride or phosphorus pentoxide. The response was linear over the concentration range of 1-50 ppm at a sampling time of 2 see. The detector is capable of monitoring CO in the parts-per-million and parts-per-billion levels by varying the sample size. At parts-perbillion levels, higher flow rates and longer sampling times are required, and the detector can monitor as low as 5 ppb of carbon monoxide.

Formaldehyde

In the first published paper on the use of enzymes as coatings for the assay of substrates in the gas phase, Guilbault (26) reported that formaldehyde was selectively detected using a protein coating. The piezocrystal was coated with a mixture of formaldehyde dehydrogenase and the cofactors reduced glutathione and nicotinamide adenine dinucleotide. Exposure to formaldehyde vapor using the syringe dilution method with lab air ( - 50% relative humidity) was utilized. A linear response from 10 ppb to 10 ppm was obtained, with frequency changes of about 200 Hz per decade of concentration. The reaction occuring on the crystal electrode was believed to be: E

+

Cofactors

+

Formaldehyde

@

[complex] -, products

281

It was speculated that if the reaction time is kept short (less than 1 min) , only the complex is formed reversibly, and not products. The coating was good for 3 days or 100 analyses. The coating is fairly specific for formaldehyde, with little responses for other aldehydes and alcohols ( > 100O:l selectivity ratio at 1 and 10 ppm formaldehyde). In another study (27), formaldehyde was determined using a coating prepared by mixing saturated solutions of 7,10-dioxa-3,4diaza-1,5,12,16-hexadecatetrol and chromotropic acid (in chloroform), in 2:l v/v ratio. Water vapor and other potential interferences were eliminated by passing the sampling stream through a column of anhydrous magnesium perchlorate. Although about 24% of the CH20 was also absorbed by the column, a calibration curve with adequate sensitivity was constructed. The response curves were linear in the concentration ranges 0.4 - 4.5 and 0.4 - 3.6 ppm v/v CH2O with and without the scrubber column, respectively. Complete reversibility was obtained in 4 min at a flow rate of 100 mllmin with sampling time of 2 min for 3.4 ppm CH20 sample, and varied from 3-6 min over the concentration range 0.4 - 4.5 ppm. Hydroaarbone

Sorption detectors were first proposed by King (28) using gas chromatographic substrates for the detection of hydrocarbons. A solvent leak detector consisting of a 9 MHz crystal coated with rubber showed sensitivities towards several organic and inorganic compounds (29). The observed sensitivities were very low for low molecular weight materials but very high for high molecular weight materials. It was suggested that the sensor may have future use in the medical field because it was sensitive to some of the anesthetic vapors tested. A device based on a crystal coated with nugol-transchlorocarbonyl bis.(triphenylphosphine)irridium(l) was reported to detect aromatic hydrocarbons preferentially over aliphatic ones and ordinary olefines (30). Xylenes, benzaldehyde, 1,3,5-trimethylbenzene, anisole, n-butyl benzene, could be detected at low concentrations, but hexane, heptane, octane, and cyclohexane were detected at high concentrations. The coating was less sensitive to benzene and toluene, and did not respond to moisture. The detector was used to monitor auto exhaust for aromatic hydrocarbons. The behavioral pattern of several coatings for monitoring a number of organic gaseous pollutants was investigated (31). The performance of model systems set up arbitrary for traces of chloroform and toluene was in agreement with expectations. The evaluated coatings included carbowax 20 M, dinonyphthalate, Apiezon-M, squalane, rubber, tricresyl phosphate and pluronic L 64 for the detection of seven organic analytes. The use of liquid crystals as coatings was reported for the detection of aromatic hydrocarbons and their derivaties, as well as of organophosphorous compounds (32-33). Liquid crystalline substances such as 4-pentyl-4t-cyano-diphenyl, 4-pentyl-4,-propylazobenzene, 4-propyl-4~-methylazoxy-benzene and cholesteryl oleycarbonate can be used as coatings, and some showed different

282 sensitivities for isomeric compounds (e.g. 4-cyan0-4~-pentyldiphenyl coating) showed sensitivities of 19.23 x lo6, 16.15 x lo6 and 23.85 x lo6 Hz.rn1.g-' for 0-, m-, and p-diethylbenznee, respectively. The cyclodextrins have been shown to behave as molecular receptors toward a wide variety of guest species, especially aliphatic and aromatic hydrocarbons. Three chemically modified cyclodextrins have been prepared, characterized by N M R spectroscopy, and studied for their possible role as coatings for benzene (34). 2,6-per-o-(t-butyldimethylsilyl)-a-cyclodextr~n (DSaCD) was recommended for its better storage Yuality and overall performance The coating was overloaded in the range of 0.08 to 4 0 0 mg dmwith benzene at 396 mg dm-3, and reversibility was achieved by placing the coated crystal in hot air oven at 8OoC for 5 min after each calibration set. There were no significant interferences from low relative molecular mass alkane gases. Although significant response was observed for toluene, the sensitivity for benzene is so great that the response for toluene is < 20% of that for a corrsponding amount of benzene. Toluene in ambient air was detected using a crystal coated with carbowax 550 (35). The detector has a linear range of 30-300 ppm toluene with a relative standard deviation better than 4% and a response time of 30 sec and reversibility of less than 40 sec. No interferences were observed at 5% (v/v) level from benzene and other alkylbenzenes. Interfernce from water was suggested to be eliminated using nafion tubing. A lightweight portable instrument was developed and reported to be capable of monitoring toluene in 1-200 ppm concentration range, using pluronic F-68 coating. The device was tested at a printing plant located in Copenhagen, Denmark, and the results obtained were in agreement with the HNU photoionization detector and the Drager tube (36).

.

Hydrogen Chloride Hydrogen chloride gas is a noxious byproduct of many industrial chloride processes, and a major product from the combustion of most solid-rocket boosters. A quartz crystal microbalance using triphenylamine as coating with glfrequency was reported (37). The detector was capable of detecting lO-'g of or 2.5 x molecules of HC1. The performance of the detector was demonstrated during the Titan-Centaur launch in 1975. Triphenylamine and trimethylamine HC1 were also used as coaing by Hlavay et a1 (38). Both coatings were found very sensitive, whereas 1 ppb HC1 could be detected with a reported response time of a few seconds, and complete reversibility obtained in less than 1 minute. Triphenylamine exhibited a lifetime of about 20 days, while the lifetime of trimethylamine was very limited due to strong interaction with moisture. In another study (39), thiosulfate ion was reacted with HC1 to yield sulfur dioxide, which reduces mercurous ion to free mercury. The reaction sequence is as follows: S 2 O 3 * -+ 2HC1

-

2C1-

+

So

+

SO2

+

H20

283

H9z2+ + 2S02 + 2H20 * Hg(S03)z2-

+ 4Ht + Hgo

The generated mercury forms a gold amalgam, causing a frequency change proportional to HC1 concentration. The detector has a linear response in the 0 . 5 - 92 ppm range, with a sensitivity of 8.3 Hz/ppm and an estimated limit of detection of approximately 20 ppb HC1. The process was made specific by the use of several scrubber reagents placed before and after the magnesium thiosulfate tube. These included mercuric nitrate to trap H2S from the gas stream, silver nitrate to trap unconverted HC1, untraped H2S and NH3 and KMn04 and BaC12 to absorb any residual SO2 in the sample gas stream. A scheme for the detection of HC1 utilizing a Zn-coated crystal has ben described recently ( 4 ) . The high sensitivity of the sensor is based on the formation of the hygroscopic salt ZnC12, and its high affinity for water. Numerical differentiation was used to compensate for ubiquitous low frequency background drifts. Although, a non-linear response was not obtained, it was suggested that the proposed scheme could be still used for chemical alarm application where triggering is based on exceeding a specified threshold rate of mass accumulation. Hydrogen Cyanide A group of metal salts, the transition metal p-diketonates known to react with hydrogen cyanide were examined as potential coatings (41). A crystal coated with bis(pentane-2,4-dionato)nickel was used to detect hydrogen cyanide over the range 13-93 ppm. The IR study supported the overall reaction.

Ni(acac)

+

2HCN

=

Ni(CN)

+

2H (acac)

The displacement of the heavy ligand by the lighter molecule HCN yields an increase in the sensitivity, a concept identified as Itmassamplificationgg, and was not applied to piezoelectric detectors before. As mass is lost from the coated crystal when exposed to HCN, the frequency increases whereas a decrease in frequency is expected if moisture is absorbed by the coating. The effect of humidity was studied over the range of 40-92% RH, and it was shown that humidity plays a part in the reacation. Since the reaction is irreversible and the nickel complex is prone to hydrolysis, the coated crystal could be used once only. Substitution of other ligands in nickel complexes, including dibenzoylmethane, dipivaloylmethane, formate and acetate, did not improve the selectivity towards HCN. Hydrogen Bulfide

Piezocrystal detectors for hydrogen sulfide were first suggested by King(28), using lead acetate, metallic silver, metallic copper, and anthraquinonedisulfonic acid.

284

Silver acetate dissolved in n-butyl amine was proposed as a coating (42). Although the reaction is irreversible, a linear calibration curve was obtained over the concentration range 10 ppb 10 ppm, and no statistical treatment of data was reported. Webber et al. ( 4 3 ) investigated the use of the acetone extracts of various soots. Soots were obtained by burning several substances in air, collecting the black residue in acetone, then applying a thin film of soot on the crystal. The extract of chlorobenzoic acid soot provided the best coating, and was sensitive to H2S in the concentration rage 1-60 ppm. Attempts were made to characterize the reaction by gas chromatographic measurement of major components, elemental analysis and IR. However, there was not sufficient information either to determine the reaction mechanism or characterize the soot extract. In a recent review, it was reported that H2S was determined with crystals coated using the Langmuir-Blodgett film deposition method (44). Linear responses over the concentration range 0-100 ppm using oTA and ttbPcSiClz coatings. A simpler and improved detector using a mixture of cadmium iodide/urea/glycerol (6:2:1 v/v) was found to be the most sensitive Urea and glycerol appeared to modify the and reporducible ( 4 5 ) . irreversible nature of the cadmium sulfide product formed in the adsorption process, yielding a completely reversible reaction. The proposed detector has a linear response over the concentration range 2-25 ppm with sensitivities of 1.3, 4.2 and 7 Hz/ppm at sampling times of 30, 60 and 120 sec, respectively. Only ammonia caused a significant frequency change (7% of that caused by the same concentration of hydrogen sulfide). The elimination of moisture was also necessary for continuous monitoring. Mercury

The first use of piezoelectric crystal for the detection of mercury in soil samples was reported by Bristow(46). Detection limits of 5 ng/l for mercury vapor in a continuous flow system, and 0.7 ng for a total mass of vapor in a drawn soil gas sample were observed. The use of gold coated piezoelectric crystals as sensors for mercury in air was investigated by Scheide et al. (47). The crystal was incorporated into a variable oscillator circuit, and changes in the frequency due to the increase in mass resulting from the ability of gold to amalgamate mercury facilitated the measurement of mercury in air at sub-part-per billion level. Calibration curves were obtained from part-per-million to subpart-per-billion concentration of mercury. Reversibility was achieved by heating the crystal in an oven to 150'C. In a similar study, 14 MHz gold crystal were used for the measurement of mercury in air (48). Linear relationship were established between the frequency change and the action period of mercury vapor. The calculated detection limit based on 3 0 criterion is 3 x 10-6g/m3with a measurement period of 10 min.

285

Nitro-aromatic Compounds Several substances were evaluated as coatings for the detection of mononitrotoluene (MNT), which can serve as a reliable indicator for the presence of trinitrotoluene (TNT) (49). A crystal coated with carbowax 1000, was found to be the most sensitive at optimum conditions of 30 ml/min flow rate, 5OoC exposure chamber temperature, and 30 pg of coating applied to the crystal. The calibration curve was linear over the concentration range 3 ppb 7.5 ppm, with a reported detection limit of 1 ppb. No interferences were observed from CO, SO2, NH3 and NO2 at 100 ppm levels. Most of the organic vapors tested caused no interference, except for some perfumes and high concentrations of organic solvents. It was also suggested that interference by humidity can be compensated for by using air at constant humidity level as the carrier gas. The coating showed a reasonably long lifetime, where a coated crystal gave almost the same sensitivity after one month of use. A series of coatings were evaluated for their sensitivity to nitrobenzene (50). The characteristics of the most sensitive coatings are listed in Table 1. Table 1 Performance of Coatings sensitive to Nitrobensene

Charcoal PEG-400 PEG-750

Quadro1 Tetrabase

Amount/ KHz

Linear range

Slope, Hz ppm-l

0.57 8.8 8.9 9.9 8.9

0.73-7.6 2.1 -9.7 2.2 -9.7 2.2-10.7 2.2 -7.0

-13.1 -7.3 -11.4 -6.4 -4.1

c +c 2

t

0.3 0.0 0.5 0.4 0.0

The greatest sensitivity was obtained when very small amount of the finest charcoal particles is used as coating. Unfortunately, these coatings are not selective and have long response times. Other potential coatings such as diphenylamine, dicyclohexylamine and trioctylamine were examined but were rejected because they suffered from high bleed rates. A recent attempt was made to develop a piezoelectric crystal sensor for nitroaromatic compounds using an (aminopropy1)triethoxysilane (APTES) coating (51). The selectivity was achieved by converting the silanol structure of the silanized sensor to a siloxane structure by the interaction with water. The characteristics of Al-plated and APTES-coated sensors for several compounds are shown in Table 2. It was proposed that the specific adsorption on the APTES coated crystals occurred through H-bonding, dipole-dipole and n-So d orbital overlap. The sensor can be successfully used for the determination of nitrotoluene with a limit of detection in the vicinity of 20 pg/l at 20°C.

Table 2 Performanae of Al-plated and APTEB Coated Bensors Al-plated sensor Conc. a Compound Benzene Nitrotoluene Toluene o-Nitrotoluene m-Nitrotoluene p-Nitrotoluene Phenol Benza ldehyde Aniline

4.1 0.013 1.2 0.009 0.009 0.051 0.19 0.06 0.035

aConcentration, mmol/dm';

-----------------

APTES Coated Sensor

-----------------

AF (HZ)

Sens.

AF HZ

Sens.

2 29 4 32 30 28 2 10 13

0.5 2230 3.0 3350 3330 549 10.5 167 370

122 112 105 112

29 8600 87 12400 10600 1647 590 17100 4900

95 84 112 1024 172

'sensitivity, (Hz dm'mol-I)

Nitrogen Dioxide Two polyalkylene glycols, Ucon-75-H-9000 and Ucon-LB-300-X were used as coatings for NO2 detection (15). The two coatings became sensitive to NOz only after the coated crystals were exposed to 1:3 mixture of NO2 and nitrogen in a small chamber for 5 min. New compounds were formed upon exposure to the mixture as confirmed by IR study. The response of both coatings was linear over the concentration range 1 ppb - 1 ppm NO2. The response was fast and reversible at low concentrations, however at concentrations above 1 ppm, a long recovery time (about 20-30 min) was reported. The coatings are highly hygroscopic, and the elimination of water vapor is necessary. In addition, the coating lost its sensitivity if it was exposed to air for one hr., but it was activated upon re-exposure to N02-N2 mixture. The use of a manganese dioxide coated crystal for the detection of NO2 was also described (52). The coating procedure involved melting manganese nitrate tetrahydrate onto the surface of the The temperature was then raised to crystal electrode at 36'C. 177'C, and the crystal was baked at this temperature to produce manganese dioxide. The best results were reported at humidity around 1500 wl 1-', and it was concluded that this detector would have a limited practical use. Organophosphorus Compounds Many pesticides and chemical warfare agents are organophosphorus based compounds which are potent cholinesterase inhibitors. They are structurally related, and undergo similar reactions. Diisopropylmethyl phosphonate (DIMP) and dimethylmethyl phosphonate (DMMP) have been used as simulants because of their relatively low

287

toxicity. Scheide and Guilbault evaluated a number of inorganic salts as coatings for the detection of DIMP (53). The studied compounds showed an increasing response in the order of CdClz
Table 3 Response of Coatings to organophosphorus Compounds

Response (Hz) Coating DIMP a

MalathionD

L-Histidine hydrochloride 30 DL-Histidine hydrochloride Succinyl choline chloride 55 succinyl choline 41 290 2 -PAD 403 3 -PAD Sample concentration; a

= 15

ppm, b

= 1

ppm, c

Parathion'

1414 474 519 490

126

64

24

= 1.5

23 76 44

ppm

The optimum coating mixture was found to be 31% 3-PAD, 13% NaOH, and 56% Triton X-100. The calibration curve for DIMP was

288

linear over the concentration range 1 ppb to 15 ppm. Similarly, malathion was determined in the range 1 . ppb 10 ppm using L-histidine hydrochloride coating. No serlous interference from inorganic gases were observed with the ternary mixture coating except from SO2 due to its reactivity with NaOH. Copper complexes including Cu(butyrate)2.ethylenediamine,

-

Cu(butyrate)2.diethylenediamineI

Cu(butyrate)2.2-ethy1enediaminel

XAD-4-Cu2+ diamine, XAD-4-Cu2+ triamine and XAD-4-Cu2+ tetramine complexes were evaluated as coatings (57-59). The overall reaction occurs in two steps: first, the copper complex binds the phosphorus ester reversibly; second, the adduct-product is irreversibly broken down by hydrolysis. The bidentate complexes proved to be the best coatings, with the X A D - ~ - C U ~diamine + giving a greater affinity for DIMP. The calibration curve of the XAD-4-Cu2+ diamine coating was linear from 1-20 ppb. To prolong the lifetime of the coating, the crystal was first coated with poly(hexadecylmethacrylate), then finely ground XAD-4-Cu2+ diamine was sprayed onto the crystal, and the excess was wiped off with a soft paper. The adsorption of DIMP on the polymer is only 0.05 Hz/ypb DIMP/pg coating, approximately 25 times less than on the m D - 4 - c ~+-diamine coating. In addition, two other copper(I1) chelates were investigated (59-61). Tetramethylethylenediamine copper(I1) chloride(TMEDA) was connected to polyvinyl pyrrolidone (PVP), and to polyvinyl benzylchloride(PVBC) polymer, resulting in PVP-TMEDA and PVBC-TMEDA) polymer bonded chelates. A linear response in the 0-20 ppb range was obtained with the PVBC-TMEDA chelate, and no saturation observed above 20 ppb. The PVBC-TMEDA response was linear in the range 0-30 ppb, however it was saturated above 30 ppb. The response characteristics of the PVP-TMEDA coating are cited in Table 4. Table 4 Response characteristics of the PVP-TMEDA complex

Concentration ( PPb1 5.1 8.9 14.0 20.9 25.9 30.0 36.5 41.1 47.0

Response (Hz1 13 36 62 100 122 138 155 172 180

Average deviation Response (Hz) Time (min) 1 0.25 0.5 0.25 1 1.7 0.75 3.0 4.7

1 1 1.5 1.5 2 2 3 5 5

Recovery Time (min) 1 2 2 3 3.5 4 4.5 6 7

No serious interferences were observed except for NH3 and water with the PVP-TMEDA, and for chloroform, with the PVBC-TMEDA. For practical applications, the PVBC-TMEDA was recommended since up to 8 0 % of relative humidity can be tolerated. Uncoated crystals with different metal electrodes were examined for the detection of DIMP (59,62). Linear calibration curves were

289

obtained in the pgfl range, with increasing sensitivity and decreasing selectivity in the order: gold, silver, and nickel electrodes. The nickel-plated crystal showed the highest sensitivity, however with poor reproducibility, and long response and recovery times. Better reproducibility was obtained with silver crystal but recovery times were still high, and NO2 was adsorbed on the electrode irreversibly. The gold crystal was recommended, where better reproducibility and selectivity were both obtainable. The use of proteins as coatings for the detection of organophosphorus compounds was first reported by Van Sant (63). A crystal coated with acetylcholinesterase was used for the detection of DIMP and Malathion, with reported detection limits of 0 . 4 ppm DIMP and 5 ppb malathion. In subsequent studies (64,65), various cholinesterases were tested as coatings, and were immobilized using different agents: glutaraldehyde, bovine serum albumin, diazo coupling, and carbodiimide. Eel Cholinesterase, immobilized with glutaraldehyde exhibited the highest sensitivity €or DIMP. Anti-parathion antibodies were used as a coating material for the determination of parathion in the gas phase (66). The results demonstrated the possibility of using antibodies in gas phase assays of the corresponding antigens at ppb levels. All responses were ascribed to specific antigenjantibody interactions after corrections for nonspecific adsorptions. For each measurement, active antibody, and reference BSA-coated cyrstals were used. A linear response over the concentration range 2 to 35 ppb parathion was reported, with a relative standard deviation of 6%. Most of the pesticides showed some response to the parathion antibody coating, but at substantially higher concentrations than parathion. The average lifetime of the antibody coating was about a week, and a completely reversible response was observed within 6% experimental reproducibility. In a follow-up study (67), the performance of acetylcholinesterase, GD-1 anti-soman, anti-DMMP antibody, and bovine serum albumin (BSA) coatings was evaluated, and different immobilzation methods were also tested. The responses obtained with the different proteins, immobilized via crosslinking with glutaraldehyde were acceptable, provided that the reference crystal was coated with dextran. The proposed coatings showed good stability and resonable lifetimes that ranged from three weeks in the case of the antibody coatings to several months in the case of BSA. Although moisture, gasoline, and sulfur are potential interfernces, their effects were eliminated by using a sodium sulfate scrubber. The question of antigenjantibody interactions on the crystal in the gas phase was addressed by Thompson et al. (68,69). Both film free and protein coated sensors showed high sensitivity towards pesticides containing the thiophosphonate functional groupI with lower limit of detection of about 0.01 pgjl in nitrogen carrier gas. Responses were reversible, and were attributed to chemisorption or physisorption processes, but no signal corresponding to selective immunochemical binding was observed.

290

oaone A piezoelectric crystal coated with lI4-polybutadiene was used for the detection of ozone in the ambient air of working environments (70). The detection principle is based on the nonreversible reaction between ozone and the unsaturated hydrocarbon polymer coating, probably with the formation of ozonides. The fresh coated crystals were kept over activated carbon in order to eliminate traces of the solvent toluene released from the coating, and the ozone from the surrounding air. Ozone measurement was performed during a TIG welding operation, and the results obtained using the crystal device were in reasonable agreement with those obtained using an AID ozone monitor (chemiluninescence) and a Drsger detector tube. A homogenity of variance in the concentration range 70 to 425 ppb O 3 was obtained, with an estimated standard deviation of 2.31, and a detection limit of less than 10 ppb. Although the coating is weakly hygroscopoic, no series effects were observed at constant relative humidity levels of 28% and 75%. Difficulties were observed only where electrical transients occurred during the welding process or, when the measurement was performed very close to the breathing zone inside the helmet, where abrupt change of humidity takes place. Phosgene

A crystal coated with methyltrioctylphosphonium dimethylphosphate was found to be a good detector for phosgene in the pg/l range (71). The flow rate in this study showed a marked effect on the sensitivity of the detector, with maximum sensitivity obtained between 5-10 ml/min. It was speculated that the detector response was probably due to a rather weak adsorption of phosgene. For rapid desorption, the flow rate was increased to 70 ml/min, resulting in complete reversiblity. At very high concentration (320) mg/l, H2S, HC1 and NH3 gave a relatively high response. The respone to NH3 was irreversible, and it was observed that exposure of the coating to large doses of NH3 resulted in an irreversible and a higher response to phosgene, however the lifetime of the coating was shorter, and the crystal ceased to vibrate after several exposures. The response curve was linear in the concentration range 5-140 pg/l, with a useful lifetime of 6 weeks limited only by exposure to moisture and high concentrations of ammonia. Propylene Glycol Dinitrate

Several substances were evaluated as coatings for the detection of propylene glycol dinitrate (PGDN), a highly toxic compound used as a propellant in torpedoes (72). The highest sensitivity was reported for dicyanoallylsilicone (OV-275), producing a response of 146 Hzlppm. The frequency response vs concentration of PGDN plot indicated that the response was nonlinear with concentration. However, it was suggested that the response curve can still be adequately described by a straight line at high concentrations (up

29 1

to 10 ppm). The response time at low concentration was less than 30 6, with full recovery of less than 1 min. It was noted that at high coating levels, there was an increase in the sensitivity to other contaminants, e. g., 2-pronanol, but not to PGDN. Therefore, the crystal coatings were deliberately kept below 5 0 KHz to get maximum sensitivity. To improve selectivity, two identically coated crystal were incorporated in the oscillator design, one preceded by a trap of cellulose acetate isobutyrate which specifically absorbs PGDN. To improve the sensitivity, an autozero fnction was also incorporatA ed, and the detection limit was extended to below 50 ppb. prototype instrument with a built-in microprocessor was described, and it was run in the laboratory continuously for up to 3 months with no obvious reduction in sensitivity, selectivity, or stability. sulfur Dioxide Hartigan (73) investigated several amine coatings for the determination of sulfur dioxide, and suggested that within a series of amines the response increased as the basicity decreased. The highest response was obtained with didodecylamine with an average slope of 100.3 Hz nmol-l. The use of the well-known West-Gaeke reaction was utilized for the determination of SO2, by coating the crystal with sodium tetrachloromercuriate(Na2HgC14),(74). Several gas chromatographic phases including: apiezon N, Silicone SE-30, Silicone QF-1, Carbowax 20 M and Versamide 900 were less sensitive than Na2HgC14. The coating absorbed 0.18 - 4.01 vg SO2 in a 15 min period, with a sensitivity of 369 HZ p 9 - l . In a subsequent study (75); Carbowax 20 M was found to be a suitable coating for SO2 determination in 1-100 ppm range, with a response time of 5 sec, and a recovery time of 1 min. No serious interferences from inorganic gases were reported, but NO and NO2 seemed to dissolve or attack the coating. Frechette et al. (76) studied the response of 14 substances to S O 2 , Tridodecylamine and tripropylamine were found most sensitive, but they bleeded easily and showed partial retention of S O 2 producing serious detector fatigue. The best coating studied was 1:l co-polymer (SDM the styrene-dimethyl-aminopropylmaleimide polymer) with a detection limit of 5 ppm. A new cell design, where the column effluent is split into two equal streams which directly and simultaneously fall on opposite faces of the coated crystal was developed (77). Various amine compounds were tested, and good resonses were obtained with armeen Quadrol and 25! amine 220, triethanolamine, and quadrol. triethanolamine were the most sensitive, and showed a linear response in 10 ppb to 30 ppm range. Response times were in the order of a few seconds, and complete recovery in 5 min. The use of a GC column, or a drier was suggested to separate SO2 from the potential interferences NO2 and moisture. In a later study, the use of a hydrophobic membrane filter was proposed to eliminate the effect of moisture on the quadrol coating (78). The response to lab air was reduced from 300 Hz to 70 Hz by the filter. The limit of detection decreased to 0.1 ppm, compared to a 1 ppb in nitrogen

292

stream in the previous study. The apparent loss in sensitivity was attributed to the reaction of SO2 with moisture to form H2S03. In addition, the linear range of the response was 0.1-40 ppm using a 5 ml sample and 10 ppb 10 ppm using a 10 ml sample. Cheney et a1 (79,80) followed a systematic approach by determining isobaric and isothermal adsorption rates in order to evaluate the effects of temperature on the sensitivity of triethanolamine coating. The response time and sensitivity were found to be affected by SO2 concentration and temperature, however, the adsorption rate was not greatly affected by temperature. Higher sensitivity was obtained at lower temperature and higher amount of coating. In addition, an instrument was designed, utilizing an uncoated reference crystal, a coated crystal, and a diode mixer system to measure the frequency difference of the two crystals. Attempts were also made to improve the stability of the TEA coating. The use of triisopropanolamine as a stabilizer, although decreasing the volatility of the coating resulted in decreased sensitivity and longer response time. A coating modified by applying a Teflon film over the TEA coating showed higher sensitivity, and the same rate of SO2 adsorption. Although the film reduced coating loss, undesirable loss was still occuring at higher temperature which was explained by lack of complete coverage of the coating with the film. The attempts to remedy coating loss led to a new SO2 sensitive coating, ethylenedinitrilotetraethanol (81). The new EDT coating was found to be more suitable than compounds used previously, and showed improved stability and detector response. Moisture caused no interferences except when condensation occurred on the crytal, however NO2 was found to react with the coating and resulted in its r emova1. A portable instrument operated on a car battery based on quadrol coating for SO2 monitoring in auto exhausts and refinery stack gases was developed by Karmarker et a1 (82). To eliminate the effect of moisture, layers of a hydrophobic membrane filter were placed in the flow system. The response of moisture was completely eliminated with 4 filter layers, however the response for so2 also decreased. The S02/H20 response ratio was optimum when using 2 filter layers of small pore size ( . 2 vm). The study of amines, especially TEA, quadrol and attempts to improve their characteristic responses continued including parameters as varying detector cell temperature, the use of gas permeable membranes to minimize coating loss, and bonding the coating to divinylbenzene-styrene copolymer (83-85). A mercury displacement method based on enhancing the disproportionation reaction of mercurous ion solution by So2 was utilized for the detection of SOz (86).

-

2S02 + 2H20 + Hg2”

0

(Hg(S03)2’+ + H g o

+

4Ht

The mercury vapor produced by bubbling the sulfur dioxide through a mercurous nitrate solution was detected by the gold-coated crystal due to the formation of a mercury amalgam. The optimum conditions were: temperature = 27OC; flow rate = 55 ml/min; [ H g Z 2 + ]= 4 x lo-’ M; purge time = 10 min. The sensitivity of the detector depended on the sample size, and linear calibration curves in the concentration

293 ranges 2.5 - 450 ppb and 2.5 - 85 ppm were produced using 2.5 and 1 ml samples, respectively. The only major interferences were the sulfide gases tested at their threshold limit values set by the occupational safety and hygiene association (OSHA). However, the interfernce effects of the sulfide gases were eliminated by passing the sample through a silver nitrate solution scrubber. Most recently, trioctylmethylammonium dichromate which irreversibly bonds SO2 in an oxidation-reduction type reaction was used as a coating for SOz (87). The detector signal is based on integration of the decrease of the frequency over at least 10 min time interval. It was reported that nonlinearity can be compensated by mathematical correction which can increase the lifetime of the coating by several orders of magnitude.

Toluene Diisocyanate Dithiozone and tri-n-octylphosphonium oxide (TOPO) were the most sensitive coatings among several substances tested for the detection of toluene diisocyanate (TDI) (88). The sensitivity to TDI was found to be 12 i 4.4 Hz/ppm over a range of 0.2 - 14 ppm. Although dithiozone was about 1000 times more sensitive to TDI than to water, the reversibility to TDI was only about 50%, but complete to water. Tri-n-octylphosphine oxide coating showed a sensitivity of about 75 Hz/ppm. However, irreversibility was still a nuisance. It was postulated that the actual absorber is a product of TDI with TOPO and water. A microcomputer-controlled prototype instrument was described, and evaluated for its ability to correct for changes in relative humidity levels (89). The instrument incorporated crystal pairs coated with polyethylene glycol 400 or 1540 or tri-n-octylphosphine oxide and cobalt(I1) chloride. It was shown that TDI can be detected over the range 0.1 - 15 ppm with relative humidities ranging from 30 to 60%. Morrison and Guilbault (90) evaluated silicon materials as coatings for TDI, Table 5.

Table 5 Response Characteristics of Bilicon Coatings Coating FS-1265 DC High Vac Grease LS-420

Response Hz 98 64 57

Response time min 2.83 2.70 5.04

Recovery time min 3.74 4.01 7.58

FS-1265 showed a linear response in the concentration range 10 ppb 10 ppm, while both DC-high vacuum grease and silastic LS-420 showed Only minor linear response in the 10 ppb - 1 ppm range. interferences were noted from ketones and alkylbenzene, however, the

294

selectivity for TDI was > 1OOO:l at 100 ppb TDI. The high specifity toward TDI was reported to be due to an irreversible chemical modification of the silicone surface by the isocyanate. Although water vapor did not interfere directly, it was observed that the amount of absorbed water on the crystal face (coated or not) was a function of the relative humidity, and not of the coating material. The response to TDI was not altered over the range 10-85% relative humidity. A coating of tetrakis(hydroxyethy1)-ethylenediamine v/v in (THEED)/dithiozone(diphenylthiocarbazone) solution (1:3 acetone) was reported for the determination of TDI in the concentration range 3-24 ppb (90). Interferences were noted only from moisture and S02, however a scrubber column of chromosorb W NAW showed some promise in eliminating S O 2 . Humidity levels up to 15% could be tolerated, and the crystal ceased to vibrate at humidity levels of 35%. A coated crystal was used for more than 2 months (about 500 determinations) without significant loss in sensitivity. Vinyl Chloride

Amine 220 was used as a coating substrate for the detection of vinyl chloride in the linear range of 0.6 75 PFln (63). Vinyl chloride was passed through chromate oxidation tubes (Drlger), where the reaction products were detected with the coated crystal. It was suspected that the amine 220 responds to the HC1 reaction product but other decomposition materials caused an irreversibility problem. The chromate oxidation tube had a lifetime of 75 injections of 2.5 cc of 33 ppm vinyl chloride, however, after 17-20 injections the response decreased drastically and the coating was overloaded with coating-reaction products complex adsorbed on the crystal. Chlorobenzene and tetrachloroethylene were found to respond significantly, and presented interference problems. The sensitivity of the detector was found to be affected by sample volume and exposure time. Although three linear curves were obtained for 2.5, 5.0 and 10.0 cc sample volumes, the best sensitivity was for the 10 cc sample, however the linearity decreased to a range of 0 . 6 6 ppm. Coated crystals were exposed continuously to 2-6 ppm samples for exposure times ranging from 0.5 5 min. The linearity of the obtained responses was not as good as for the injection mode responses.

-

-

-

Uae of PZ Deteatora i n Water earnplea

Piezoelectric detectors were used for the detection of dissolved gases in aqueous solutions. Ammonia and hydrogen sulfide were determined utilizing gas-permeable membranes to isolate the crystal from the solution (91). The crystal coatings were the nickel dimethylglyoxime complex for NH3, and the acetone extract of chlorobenzoic acid soot for H 2 S . The concentration vs. AF curves were linear to about 0.45 and 9 x M for ammonia and hydrogen sulfide, respectively. Dissolved C02 was also determined using didodecylamine and dioctadecylamine coatings, and was mentioned earlier (22).

295

The early applications of the piezoelectric crystal detectors were limited to the measurement in the gas phase, because of the common impression that stable oscillation cannot be obtained in the liquid phase. However, recent advances in PZ research have shown that quartz crystals can oscillate in contact with solution, and several studies have been reported addressing the theoretical aspects of the oscillating frequency of piezoelectric crystals in solution. Nomura and Okuhara (92) demonstrated that the frequency change of a crystal immersed in an organic solvent depends on the density and viscosity of the solvent, and was not affected by the dielectric constant: AF

=

+

bq112 -

C

where d = density, q = viscosity, and a, b, and c are constants that depend on the properties of the crystal. Bruckenstein and Shay (93) conducted an in situ measurement of the resonant frequency of an oscillating quartz crystal having one electrode face in contact with an electrolytic solution. It was experimentally shown that the volume and the height of the liquid above the crystal has no significant effect on the resonant frequency, and the resonant frequency varies as the square root of the solution viscosity. In addition, changes in temperature caused significant frequency changes as a result of concomitant alternations in the viscosity and density of most liquids. The oscillation frequency of a quartz resonator in contact with The liquid was also investigated by Kanazawa et a1 ( 9 4 , 9 5 ) . relationship which expresses the change in the oscillation frequency of the quartz crystal in terms of fluid and quartz parameters can be described by: AF = - F o 3 1 2

(qLPL/mpQPQ) ' I 2

where F, = oscillation frequency of the free (dry) crystal, q L and p L = absolute viscosity and density of the liquid, and p Q l p q = elastic modulus and density of the quartz. The theory was based on a simple physical model which treats the quartz as a lossless elastic solid, and the liquid as purely viscous fluid. The frequency shifts arise from coupling the oscillation of the crystal, a standing shear wave with a damped propagating shear wave in the liquid. The accuracy of this model was demonstrated using aqueous solutions of glucose and ethanol at various concentrations. A systematic comparative study was conducted by Yao and Zhou (96) to investigate the oscillation behavior of a piezoelectric crystal immersed in various liquid systems including organic liquids, waterlorganic mixed solvents, and aqueous solutions of various electrolytes. It was found that, for a crystal to oscillate in the liquid phase, it is necessary to maintain the temperature at a level higher than a critical temperature at which the crystal ceases to oscillate. Expressions for the frequency shifts in different liquids were derived, and the general equation proposed

296 is : AF = C1 d l ”

+

C2q1’2

-

-

C ~ E C4K

+

C,

where d, q, E and K are the density, viscosity, dielectric constant and specific conductance of the liqid, and C,, C1, C 2 , C3 and C 4 are constants depending on the crystal oscillator and other experimental conditions. For organic liquids, the influence of dielectric constant, the specific conductance is negligible. For organic/water mixed solvent, the influence of specific conductance can be neglected. For an aqueous electrolyte solution, in the range where the density and viscosity do not vary significantly with electrolyte concentration, the influence of the dielectric constant can be neglected. In a recent study (97), it was concluded that the frequency change depended on the circuit used. One side of the crystal was coated with a silicon sealant, and the coating was used for measuring the oscillation of crystals in solutions for a wide range of density and viscosity values, and in electrolyte solutions. The experimental data showed that the frequency change in pure water is a linear function of p q except for salt and high polymer solutions. Finally, a continuous flow cell system was designed, and a comprehensive analysis of the interaction between an 11-MHz AT-Cut quartz crystal and viscous liquids was studied (98). The derived equation was basically identical with that reported earlier (94,95) assuming that the solution height is sufficiently large.

Bromide The catalytic effect of bromide on the oxidation of iodine to iodate by permanganate in acidic medium was utilized for the determination of bromide with a piezoelectric crystal detector (99). 10 I -

+

+

2 Mn04-

+

16 H + > -

2 ~ n 0 4 -+ 4 H+

Br -

--+

5 I2

+

2 Mn2+

+

0

H20

2 1 0 ~ +- 2 M n 2 + + 21-1~0

After extraction of iodine in toluene, the resulting frequency change caused by iodine adsorption on the crystal electrode is 5 x proportional to bromide concentration over the range 0.5 M. Only silver(l), mercury(l1) and large concentrations of chloride interfered significantly.

-

copper A crystal coated with poly(4-vinylpyridine) was used for the determination of copper (11) in solution (100). Metal ions adhering to the resin were removed by passing 1 mM EDTA solution through the cell. The frequency change resulting from the sorption of copper was linearly proportional to the flow rate of copper solution up to 8 ml min-l, and to the volume of the solution that passed over the crystal over the range 1-7 min. The calibration graph was linear

297

over the range 5-35 pM, in a maleate buffer at pH 6.6 flowing over the crystal for 5 min. Cadmium and equal molar concentrations of iron(II1) interfered giving positive errors, while ten fold amounts of cyanide, thiosulfate and thiocyanate gave negative errors, and sulfide gave negative error. Cyanide

Namura and co-workers have utilized piezoelectric crystals for the determination of cyanide in solution (101,102). A drop (5 pl) of the sample solution adjusted to pH 10.4 was placed on the electrode of the horizontally oriented crystal in an air path, and the cyanide reacted with the gold causing an increase in the crystal frequency. The reaction time of the cyanide with the gold was about 2-3 min, and was independent of cyanide concentrations. The response curve was linear over the range 1 x 4.0 x M. Only Silver(1) and Mercury(I1) interfered, while other cations which form stable cyanide complexes or hydroxides (Cd2+,C02+,Cu2+,Fe3+, P b 2 + , Ni2+, Zn2+)were completely masked by the addition of EDTA.

-

Endotoxin

Endotoxin, a fever inducing pyrogen produced by microbes is widely found in water samples. The gelation process of Limulus amebocyte lysate (LAL) by endotoxin was utilized to measure endotoxin via viscosity monitoring with a piezoelectric quartz crystal (103). The gelation time was computed by using an approximation to a polynomial equation from the resistance or the frequency change measurements. The maximum rate of change of the resistance (R) was confirmed as a good index of endotoxin concentration with a detection limit of 1 pg/ml. The changes in resistance and resonance frequency were also considered as indices with detection limits of 10 and 100 pg/ml, respectively, and were recommended for determination of high concentrations of endotoxin. Iodide

Iodide in solution was determined by means of a piezoelectric crystal with silver electrodes connected to a normal TTL oscillator (104). In a follow-up- study (105), iodide was determined in the range 0.5 - 7 pM by electrodeposition on the silver electrodes of a crystal. Thiosulfate, cyanide, sulfide, Fe(II1) , Hg(I1) , and Ag interfered, but procedures to eliminate their interferences were described. In another study (106), iodide was oxidized to iodate, and after the destruction of excess oxidant, iodine was formed on addition of excess iodide, which was extracted in toluene, and detected by adsorption on a silver-plated crystal. The calibration graph was linear over the concentration range 0.005 - 1 x MI and only bromide interfered significantly. Iron The effect of metal ions on a piezoelectric crystal in aqueous

298

solution and the adsorptive determination of Fe(II1) as phosphate was studied by Nomura and Maruyama (107). One side of the crystal was covered with a thin glass plate to prevent the electrodeposition of metals. The phosphate of some metal ions precipitated and adsorbed on the crystal from acidic solutions especially lead (11), Fe(II1) and Al(II1). The adsorption of FeP04 was greatest at pH 2.7-2.8, and facilitated the determination of Fe(II1) in the concentration range 1 x 10'' 1 x M. Ten-fold molar amounts of lead and aluminium, and equivalent amounts of bismuth, sulfide, and thiosulfate caused low recoveries resulting from precipitation of the cations by phosphate and reduction of Fe(II1) by the anions.

-

Lead Nomura et a1 (108) determined lead reproducibly in the range 1 x -3x M by adsorption of the extracted lead 8-quinolinolate on the electrodes of a piezoelectric crystal. Fe3+, Nil Co*+, Zn, Cd and Agl+ interfered, but they were masked by L-ascorbic acid and cyanide. In a follow-up study, lead was determined using a crystal coated with copper oleate (109). Copper was removed from the coating by passing 1 mM EDTA until the frequency became constant, followed by 0.01 M acetate buffer (pH 5.6). The remaining coating reacted with aluminium, copper (11), iron (111) and lead in a flowing acidic solution. Lead exhibited the largest and most reproducible frequency change, and was determined in the 3-40 pM range at pH 5.5 - 5.8. The mass of an underpotentially deposited monolayer of lead on gold was measured in situ by using a quartz crystal (110). Crystals were operated at their third harmonic, and a frequency change of 46 Hz corresponding to a mass gain of 2.8 x lo-' g/cm2 was noted. Mercury The use of a gold coated piezoelectric crystal, coupled with a direct reduction technique was described for the determination of mercury in water (111). The mercury in an aqueous sample was reduced to elemental mercury by tin(I1) chloride, the mercury vapor produced was volatilized from the solution and detected with the gold crystal. The reported linear range was 5-100 ng with a sensitivity of 1.78 Hzlng. Reversibility was achieved by thermal desorption at 170'C. In a different approach (112), mercury and silver were electrodeposited on the electrodes when a potential of 0.1 V was applied. However, Hg(I1) can be deposited with no voltage applied without interference from silver, but with less reproducibility. Reversibility can be achieved by immersing the crystal in 1:1 mixture of 0.01 M ammonium peroxydisulfide and 0.01 M HNOB for 5 min. Microbial Detection A piezoelectric biosensor system was applied for the detection of E . c o l i using an antibody coating (113). The frequency shifts correlated with an E. coli concentration in the range of l o 6 - lo* ~ells.cm-~ for a reaction time of 30 min. The frequency shift was

299

further increased by further treatment to bind anti-E. coli antibody modified polystyrene particles to the E . coli allowing determination limit of 10’ cells.~m-~. In another study (114), three different immobilization methods were evaluated for the detection of E. coli using an antibody coated crystal. The best results were obtained with a protein A coated method, and a response for lo6 - lo9 cells 1 ml of E. coli K12 and other antigens of the Enterobacteriaceae family was observed. Repeated usage of the coated crystal by removing the bound antigen with urea or glycine HC1 buffer (pH 2.6) was not very successful. B ilver

Silver was determined in solution using a three electrode deposition system, comprising of the platinum-plated gold electrode of the crystal as cathode, a coiled platinum wire anode, and a silver-silver chloride reference electrode (\15). Linear calibration curves over the concentration ranges 10- - 5 x lo-’ M and lo-’ lo-’ M were obtained after electrodeposition for 5 min, and by recycling 20 ml solution for 3 h, respectively. In another design (116), a zinc rod was used as a counter electrode, and silver in the range l o F 5 - 10 - 6 M was determined by internal electrodeposition on the platinum electrode. A simpler method for electrodeposition of silver was described, where no equipment was used for applying voltage (117). The test solution contained EDTA to mask other interferences. The response to silver was linear over the concentration range - 3 x 1 0 - ~after electrodeposition for 10 min, and 2 x - 1x M for 1 hr. A special transistorized circuit was also constructed (118) to minimize spontaneous electrodeposition. Siver adsorption occurred when tertrate, citrate, EDTA or their mixtures were present, and the response to silver was linear in the range 2 x lo-’ - 1 x l o w 5 M after 10 min from a 1 mM EDTA/3mM tartrate solution. No significant interfernces were noted, including mercury(I1) and copper(I1). Finally, a method for the determination of total salt content of natural waters, compleximetric and precipitation titrations monitoring were reported by Yao et a1 (119,120). Analytes at M were determined. Several metal ions concentrations down to were titrated with EDTA in the absence of buffer and competing ligands, in buffered media, in acidic solutions and in presence of competing ligands. A method for the determination of total organic chlorine content of water was described (121), based on the combustion of organic compounds to produce HC1 which was subsequently detected with a crystal coated with amine 220. The reported detection limit was in the order of 100 pg of HC1, and 3 ppb of chloroform were detectable in a 100 p1 aqueous sample.

-

CONCLUBIONB

The fundamental theory and the principles of operation of the crystal detectors are deceptively simple. However, the simplicity of the technique might be its main drawback, and may have

PZ

300 led to several erroneous conclusions at the early stages. Although most of the published literature continues to describe systems limited to laboratory environment, the future of such detectors is still bright. The use of pattern recognition techniques in conjunction with sensor arrays constitutes a promising approach for multicomponent analysis and for improved selectivity (122,123). In addition, the development of new coating materials which possess improved characteristics such as stability, selectivity, sensitivity, fast response and short recovery time is still desirable. Further advances in sensors design and in electronics will generally play a crucial role in expanding the versatility of the sensors, and will facilitate overcoming some of the limitations and drawbacks. Finally, more emphasis should be directed toward thorough undestanding of the nature of interactions between the coatings and the analytes. REFERENCES 1. Curie J, Curie P. Comp Rend 1880; 91: 294-297. 2. Lu C. In: Lu C., Czanderna AW, eds. Applications

of Piezoelectric Quartz Crystal Microbalances. New York: Elsevier,

1984; 21-22. 3. Sauerbrey G. 2. Phys 1959; 155: 206-212. 4. Warner AW, Stockbridge CD. In: Behrndt

Microbalance

Techniques.

New

York:

Plenum

KH, ed. Vacuum Press, 1963; 3:

55-73. 5. King WH. Res/Development 1969; 20(4): 28-34. 6. Chuan RL. Aerosol Science 1970; 1: 111-118. 7. Chuan RL. In: Steven RK, Herget Wf. , eds. Anal Methods Appl. Air Pollut Meas. Ann Arbor: Ann Arbor Science, 1974; 163-189. 8. Chuan RL. Instrum. Pulp Pap Ind; 1975, 16(20), 1-6. 9. Olin JG, Sem GJ. Atmos Environ 1971; 5: 653-668. 10. Daley PSI Lundgren DA. Am Ind Hyg Assoc J 1975; 35, 518-532. In: Lu C, Czanderna AW, eds. Applications of 11. HO MH.

Piezoelectric Quartz Crystal Microbalances. New York: Elsevier, 1984; 351-388. 12. Suleiman A&, Hahn EC, Guilbault GG, Cavanaugh JR. Anal Lett 1984; 17(A19) : 2205-2211. 13. Hahn EC, Suleiman AA, Guilbault GG, Cavanaugh JR. Anal Chim Acta 1987; 197: 195-202. 14. Guilbault GG, Lopez-Roman A, Billedeau SM. Anal Chim Acta 1972; 58: 421-427. 15. Karmarker KH, Guilbault GG. Anal Chim Acta 1975; 75: 111-117.. 16. Webber LM, Guilbault GG. Anal Chem 1976; 48: 2244-2247. 17. Hlavay J, Guilbault GG. Anal Chem 1978; 50: 1044-1046. 18. Colins SI, Moody GJ, Thomas JDR. Anal Proc 1985, 22: 10-14. 19. Moody GJ, Thomas JDR, Yarmo MA. Anal Chim Acta 1983; 155: 225-229. 20. Lai CSI, Moody GJ, Thomas JDR. Analyst 1986; 111: 511-515. 21. Fraser SM, Edmonds TE, West TS. Analyst 1986; 111: 1183-1188. 22. Fogelman WW, Shuman MS. Anal Letters 1976; 9(8): 751-765. 23. Jordan JM. MSc. Thesis 1985; University of New Orleans.

30 1 24. Fatibello-Filho 0, de Andrade JF, Suleiman AA, Guilbault GG. Anal Chem 1989; 61: 746-748. 25. Ho MH, Guilbault GG, Scheide EP. Anal Chem 1982; 54: 1998-2002. 26. Guilbault GG. Anal Chem 1983; 55: 1682-1684. 27. Fatibello-Filho 0, Suleiman AA, Guilbault GG. Talanta 1991; 38(5) : 541-545. 28. King WH. Anal Chem 1964; 36: 1735-1739. 29. King WH. Bull NY Acad Med 1972; 48(2): 459-467. 30. Karmarkar KH, Guilbault GG. Environmental Letters 1975; 10 (3): 237-246. 31. Edmonds TE, West TS. Anal Chim Acta 1980; 117: 147-157. 32. Mierzwinski A, Witkiewicz Z. Talanta 1987; 34: 865-871. 33. Micrzwinski A, Witkiewicz Z. Chem Anal (Warso) 1985; 30: 429-441. 34. Lai CSI, Moody GJ, Thomas JDR, Mulligan DC, Stoddart JF, Zarzycki RJ. Chem SOC Perkin Trans I1 1988; 3: 319-324. 35. HO MH, Guilbault GG, Rietz B. Anal Chem 1980; 52: 1489-1492. 36. Ho MH, Guilbault GG, Rietz B. Anal chem 1983; 55: 1830-1832. 37. NASA Tech Brief, MFS-23357, 1976. 38. Hlavay J, Guilbault GG. Anal chem 1978; 50: 965-967. 39. Hahn EC, Suleiman AA, Guilbault GG. Anal Lett 1989; 22(1): 213-224. 40. Neuburger GG. Anal Chem 1989; 61: 1559-1563. 41. Alder JF, Bentley AE, Drew PKP. Anal Chim Acta 1987; 182: 123-131. 42. Hlavay J, Guilbault GG. Proceeding, 4th Joint Conference on

Sensing of Environmental Pollutants, American Chemical Society (1978); 775-777. 43. Webber LM, Karmarkar KH, Guilbault GG. Anal Chim Acta 1978; 97: 29-35. 44. McCallum JJ. Analyst 1989; 114: 1173-1189. 45. De Andrade JF, Suleiman AA, Guilbault GG. Anal Chim Acta 1989; 217: 187-912. 46. Bristow Q. J Geochem Explor 1972; 1: 55-76. 47. Scheide EP, Taylor JK. Environ Sci Tech 1974; 8: 1097-1099. 48. Magilevskian, Mayorov AD, Stroganova NS, Stavrovski DB, Galkina

IP, Spassov L, Mihailou D, Zaharieva R. Sensors and Actuators 1991; 28: 35--39. 49. Tomita Y, Ho MH, Guilbault GG. Anal Chem 1979; 51: 1475-1478. 50. Sanchez-Perdreno JAO, Drew PKP, Alder JF. Anal Chim Acta 1986; 182: 285-291. 51. RajakoviC LV. J Serb Chem SOC 1991; 56: 521-534. 52. Edmonds TE, Hepher MJ, West TS. Anal Chim Acta 1988; 207: 67-75. 53. Scheide EP, Guilbault GG. Anal Chem 1972; 44: 1764-1768. 54. Shackelford WM, Guilbault GG. Anal Chim Acta 1974; 73: 383-389. 55. Tomita Y, Guilbault GG. Anal Chem 1980; 52: 1484-1489. GG, Tomita Y, Kolesar ES Jr. USAF Report 56. Guilbault (SAM-TR-80-21) , 1980. 57. Guilbault GG, Affolter J, Tomita Y, Kolesar ES Jr. Anal chem 1981; 53: 2057-2060. 58. Kolesar ES Jr., Guilbault GG, Affolter J, Tomita Y, USAF Report (SAM-TR-80-6) 1982.

60.

Kolesar ES Jr., Guilbault GG, Affolter J, Kristoff J, Tomita Y, Scheide EP, Warner RBJ. USAF Report (SAM-TR-83-28) 1983. Guilbault GG, Kristoff J, Owen D. Anal Chem 1985; 57:

61.

Guilbault

59.

1754-1756.

GG,

(SAM-TR-82-14) 62. 63. 64. 65. 66.

Kristoff

J,

Kolesar

ES

Jr.

USAF

Report

1982.

Kristoff J, Guilbault GG. Anal Chim Acta 1983; 149: 337-341. Van Sant MJ. Ph.D. Thesis 1985; University of New Orleans. Guilbault GG, Lubrano GJ, Ngeh-Ngwainbi J, Jordan JM, Foley PH. SMCCR-RSB 1986. Guilbault GG, Ngeh-Ngwainbi J, Foley PHI Jordan JM. In: Smyth MR, Vos JG, eds. Electrochemistry. Sensors and Analysis. Amsterdam: Elsevier, 1986; 25: 335-341. Ngeh-Ngwainbi J, Foley PHI Kuan SS, Guilbault GG. Amer Chem SOC 1986;

108:

5444-5447.

68.

Suleiman AA, Pender MI Ngeh-Ngwainbi J, Lubrano GJ, Guilbault GG. CRDEC-CR-075 1990. Rajakovid L, Ghaemmaghami V, Thompson M. Anal Chim Acta 1989;

69. 70. 71. 72. 73. 74. 75. 76.

Rajakovid L, Thompson M. J Serb Chem SOC 1991; 56: 103-109. Fog HM, Rietz B. Anal Chem 1985; 57: 2634-2638. Suleiman A, Guilbault GG. Anal Chim Acta 1984; 162: 97-102. Turnham BD, Yee LK, Luoma GA. Anal Chem 1985: 57: 2120-2124. Hartigan MJ Jr. PhD Thesis 1970; University of Rhode Island. Guilbault GG, Lopez-Roman A. Environ Lett 1971; 2: 35-45. Lopez-roman A, Guilbault GG. Anal Lett 1972; 5: 225-235. Frechette MW, Fasching JL, Rosie DM. Anal Chem 1973; 45:

77. 78.

Karmarkar KH, Guilbault GG. Anal Chim Acta 1974; 71: 419-424. Karmarkar KH, Webber LM, Guilbault GG. Environ Lett 1975;

67.

217:

111-121.

1765-1766. 8:

345-352. 79. 80. 81. 82.

Cheney JL, Homolya JB. Anal Lett 1975; 8: 175-193. Cheney JL, Momolya JB. Sci Total Environ 1976; 5: 69-77. Cheney J, Norwood T, Homolya J. anal Lett 1976; 9: 361-377. Karmarkar KH, Webber LM, Guuilbault GG. Anal Chim Acta 1976;

83. 84. 85.

Hepher MJ. PhD Thesis 1984; University of Aberdeen. Cooke S. PhD Thesis 1979; University of Aberdeen. Edmonds TE, Hepher MJ, West TS. Anal Chim Acta

81:

265-271. 1986;

187:

293-299. 86. 87. 88.

Suleirnan AA, Guilbault GG. Anal chem 1984; 56: 2964-2966. Pribil R, Bilkovd E. Talanta 1992; 39: 361-366. McCallum JJ, Fielden PR, Volkan MI Alder JF. Anal Chim Acta

89.

Fielden PR, McCallum JJ, Stanios TI Alder JF. Anal Chim Acta

90.

De Andrade JF, Fatibello-Filho 0, Suleiman AA, Guilbault AA. Anal Lett 1989; 22: 2601-2611. Webber LM, Guilbault GG. Anal Chim Acta 1977; 93: 145-151. Nomura T I Okuhara M. Anal Chim Acta 1982; 142: 281-284. Bruckenstein S, Shay M. Electrochim Acta 1985; 30: 1295-1300. Kanazawa KK, Gordon JG. Anal Chem 1985; 57: 1771-1772. Kanazawa KK, Gordon JG. Anal Chim Acta 1985; 175: 99-105. Yao SZ, Zhou TA. Anal Chim Acta 1988; 212: 61-72.

1984; 1984; 91. 92. 93. 94. 95. 96.

162: 162:

75-83. 85-96.

97. Kurosawa S , Tawara E, Kamo N, Kobatake Y. Anal Chim Acta 1990; 230: 41-49. 98. Shana ZA, Radtke DE, Kelkar U R , Josse F, Haworth DT. Anal Chim Acta 1990; 231: 317-320. 99. Yao SZ, Mo ZH. Anal Chim Acta 1989; 217: 327-334. 100. Nomura T, Sakai M. Anal Chim Acta 1986; 183: 301-305. 101. Nomura T, Hattori 0. Anal Chim Acta 1980; 115: 323-326. 102. Nomura T, Anal Chim Acta 1981; 124: 81-84. 103. Muramatsu H, Tamiya E, Suzuki M, Karube I. Anal Chim Acta 1988; 215: 91-98. 104. Namura T, Mimatsu T. Anal. Chim Acta 1982; 143: 237-243. 105. Nomura T, Watanabe M, West TS. Anal Chin Acta 1985; 175: 107-116. 106. Yao S Z , Mo ZH, Nie LH. Anal Chim Acta 1988; 215: 79-89. 107. Nomura T. Maruyama M. Anal Chim Acta 1983; 147: 365-369. 108. Nomura T, Yamashita Y, West TS. Anal Chim Acta 1982; 143: 243-247. 109. Nomura T, Okuhara T, Hasegawa T. Anal Chim Acta 1986; 182: 261-265. 110. Melroy 0, Kanazawa K, Gordon JG, Buttry D. Langmuir 1986; 2: 697-700. 111. Ho MH, Guilbault GG, Scheide EP. Anal Chim Acta 1981; 130: 141-147. 112. Nomura T, Fujisawa M. Anal Chim Acta 1986; 182: 267-270. 113. Muramatsu H, Watanabe Y , Hikuma M, Ataka T, Kubo I, Tamiya E, Karube I. Anal Lett 1989; 22: 2155-2166. 114. Plomber M, Guilbault, GG, Hock B. Enzyme Microb Techno1 1992; 4: 230-235. 115. Nomura T, Iijima M. Anal Chim Acta 1981; 131: 97-102. 116. Nomura T, Nagamune T. Anal Chim Acta 1983; 155: 231-234. 117. Nomura T, Tsuge K. Anal Chin Acta 1985; 169: 257-262. 118. Yao SZ, Dan SL, Nie LH. Anal Chim Acta 1988; 209: 213-221. 119. Yao SZ, Mo ZH. Anal Chim Acta 1987; 193: 97-105. 120. Yao SZ, Mo ZH, Nie LH. Anal Chim Acta 1990; 230: 51-58. 121. Gebhardt JA. US Environmental Protection Agency, Report 60014-84-003, 1983. 122. Carey WP, Beebe KR, Kowalski BR, Illman DL, Hirschfeld T. Anal Chem 1986; 58: 149-153. 123. Olness DU, Kishiyama KK, Lane JE, Steinhaus RJ, Hirshfeld TB. CRDC-SP-86007, 1986.

This Page Intentionally Left Blank

305

The Electrochemistry of Gases of Medical Interest and Electrochemical Gas Sensors Peter Tebbutt Chemistry Department, University of Warwick, Coventry CV4 7AL, UK 1. Introduction

Chemical sensors operate by converting a physical or chemical change into a signal which can be interpreted in terms of the concentration of the analyte of interest. In most cases a transducer converts this change into an electrical signal which is output as a digital or analogue scale. In this respect electrochemical cells are ideal as sensors. The reaction of a substance at an electrode results in changes of an electrical nature, that is, changes in current, voltage or resistance may be generated. These electrical changes are easily measured using quite simple electronic cicuitry. Electrochemical devices are also relatively cheap to manufacture or to buy. Clinical gas sensors may cost several hundred to thousands of pounds whereas clinical mass spectrometers cost tens of thousands of pounds. The basic design is also flexible so that they can be tailored for specific purposes. They may be reduced in size and be made flexible so that they can be used intraveinously or made larger and sturdier for use in harsh manufacturing environments. They can be made light and portable for environmental monitoring or be incorporated in larger, non-portable machinery. One criticism of electrochemical methods is the possible lack of specificity. There is, however, some specificity inherent in all electrochemical measurements. Every chemical has its own specific potential at which it will react at an electrode (the standard electrode potential). In mixtures the observed potential difference generated by the sensor will be a function of all the components present. When using amperometric measurements, the application of a given potential to the working electrode may cause all components with standard electrode potentials lower

306

than the applied potential to react at the electrode. Thus the observed current will, again, be a function of all components present. Measurements made using conductivity may be even less selective. One way to impart specificity is to ensure that only the component of interest is allowed into the electrochemical cell. This may be achieved by the use of a membrane which will separate the internal components of the cell (the electrolyte and electrodes) from its working environment. Another method, which has been used in the development of solid state sensors, is to incorporate a chemical reactor into the sensor [ll. In the development of amperometric methods, the use of short pulses to different potentials may separate signals arising from the presence of different components [ 2 - 4 1 . The apparent lack of specificity is therefore seen by electrochemists as an interesting, but not insurmountable, problem. One approach used in solving such problems is the detailed study of the electrochemistry of the system of interest. Kinetic studies may shed light on the chemistry and electrochemistry involved and this information may be used to provide a suitable protocol for the analysis of the mixture in question. In this review I hope to emphasise the importance of such studies in the development of working sensors. It is possible to construct sensors in an empirical fashion ie. by trial and error methodology, selecting different electrode materials, membranes, electrolytes etc. This approach has been described as haphazard and inadequate [ 5 1 . In particular it does not give information about possible routes to improving stability, response range and response time of the device. As will be shown this kind of investigation may lead to analytical methods that would not be likely to arise from empirical studies.

307

This review is written from the point of view of a chemist who is interested in sensors rather than from that of one whose main interest is the analysis of gases as a discipline in its own right. Therefore although the current literature on solid state gas sensors greatly outnumbers that on membrane type sensors this review concentrates on the latter. The reader interested in solid state devices is referred to a number of excellent reviews and conference proceedings which have appeared recently [6-91. 2 . Membrane Covered Sensors.

In this section the basic design and operation of these types of sensors are described. Examples of such sensors in common use are the Clark oxygen electrode [ l o 1 and the Severinghaus carbon dioxide electrode [lll. The basic design of these and related sensors is shown schematically in Figure 1. The sensor is an electrochemical cell which consists of two electrodes (strictly half cells), e and f, immersed in an electrolyte solution, d. The electrolyte and

Figure 1. The electrochemical cell as a sensor. See text for description.

e

f

d

a c-t

3

a

a

b

308

electrodes are separated from the external environment by a polymer membrane, c. The response to analyte, a, which reacts at the working or sensing electrode e,is measured by the external circuitry, g . Electrode, f, is the reference electrode which is often a silver wire coated with silver chloride which constitutes the stable half cell reaction;

The electrochemical potential of such an electrode is, ideally, independent of the amount of current flowing through it. Therefore any changes in potential difference imposed on or measured across the cell can be attributed to changes at the other electrode. The working electrode is usually some noble metal such as platinum, silver or gold. Commonly the electrolyte is an aqueous solution of an inert salt such as potassium chloride or a buffer solution such a s potassium hydrogen carbonate. In some of the studies described here the electrolyte has been replaced by a non-aqueous solvent containing an inert solute such as tetrabutylammonium tetrafluoroborate. The advantage of using a nonaqueous electrolyte is that it is possible to reach potentials that are unobtainable in water due to solvent electrolysis. Thus it is possible to observe the electrochemistry of many more potential analytes. Also the product of many electrochemical reactions in dry solvents is a stable radical which would not exist in aqueous media (due to immediate protonation). These radicals may undergo interesting and useful chemical reactions. For example in aprotic solvents oxygen is reduced to the superoxide radical anion (02’-). In water this would be protonated but in dimethyl sulphoxide it may undergo other reactions with species such as carbon dioxide. This reaction has been exploited ( v i d e infra) to produce a sensor for the simultaneous determination of oxygen and carbon dioxide. Another example is the reaction between the electrochemically generated fluoranthene radical anion and anaesthetic gases. This

309

has been used by Compton and Northing in the amperometric measurement of these gases ( v i d e i n f r a ) Difficulties do arise, however, in the long term operation of sensors based on such systems. It is extremely difficult to keep non-aqueous solvents dry. The presence of even small amounts of water changes the observed electrochemistry dramatically (due to the protonation of radical species). Also, most of these solvents are toxic nd therefore it would be difficult to justify their use in clinical environments. The membrane has two purposes. Firstly it separates the internal components of the sensor from the external working environment. This is useful in that the electrolyte composition may be maintained and that fouling of the electrode by components of the analyte mixture may be prevented. Secondly, the membrane forms a well defined diffusion barrier for the analyte to pass through. The steady state current observed under potentiostatic control is a function of the kinetics of electron transfer at the electrode and of mass transport to the electrode surface. At high potentials when electron transfer is fast the current is solely a function of mass transfer. This may be controlled by changing the thickness of the membrane and changing the membrane material. The sensitivity and selectivity of the sensor may therefore be controlled to some extent by judicious choice of the membrane material.

It is sometimes desirable to make a sensor whose response is independent of the membranes properties. This would improve the reproducibility of measurements which may differ due to the intrinsic differences in individual membranes. Such differences may arise during the manufacturing process. This problem may be solved by considering the response of a Clark type sensor to the application of a potential pulse.When such a pulse is applied to the working electrode it takes a finite amount of time for the diffusion gradient to develop across the electrolyte layer and the membrane. At short times the diffusion gradient does not

reach the membrane and therefore the observed current should be independent of the membrane properties. It is therefore possible, in principle, to make the sensor response independent of the membrane, by using transient rather than steady state methods. This review will focus, mainly, on amperometric sensors. These operate by applying a potential to the working electrode and measuring the current thus generated. A plot of steady state current versus applied potential (Figure 2 . ) shows an initial increase in current as the rate of the electron transfer increases and a mass transport limited plateau where the current is independent of the potential. In this region the current is proportional to the concentration of analyte in the test solution or gas mixture. A non-linear proportionality to the concentration is usually indicative of some complex reaction mechanism involving kinetically slow steps rather than the simple electron transfer which gives rise to linear current - concentration plots. Mechanistic investigations usually shed light on the reason for non-linearity and lead to useful predictions of the sensor response.

'L

1

I

E 11%

potential

E/ V

Figure 2 . Typical current-potential plot for membrane type sensor. Ell2 is characteristic of a particular species and is the potential at which the current is half the diffusion limited value .

31 1

Potentiometric methods involve the measurement of the potential difference between the sensing electrode and the reference electrode. This has certain advantages over amperometric methods. One is that because the measurement is made at equilibrium (between the redox states of the analyte) none of the analyte is consumed. Amperometric methods do consume some analyte though this may be minimised by using smaller electrodes and by using pulse methods. Another is that the circuitry needed for control is much simpler. In principle all that is needed is a high impedance voltmeter. In amperometric measurement quite sophisticated electronics are needed to control the potential of the working electrode whilst simultaneously measuring the current. Although this circuitry is readily available, it must add to the cost of commercial devices. The advantages of amperometric measurement do, however, outweigh the disadvantages. The output of the potentiometric measurement is logarithmic, following the Nernst equation,

where [ox1 and [redl are the concentrations of the oxidised and reduced species which constitute the electrochemical equilibrium ox + ne-

* red

(3)

and E, EoI F, R and T are the observed potential, the standard electrochemical potential of the cell, the Faraday constant, the gas constant and the temperature respectively. n is the number of electrons transferred in the reaction. Because of the logarithmic nature of the response, a large change in concentration produces a relatively small change in potential. Sensitivemeasuring devices are therefore necessary. Amperometric

312

measurements are potentially much more sensitive. It is also significant that the measured potential relies on the potential of the reference electrode and the temperature. The stability of the reference electrode is therefore critical to the measurement being made as is knowledge of the temperature response and the ambient temperature. This necessitates constant recalibration of devices which may be time consuming. By comparison the measurement made amperometrically does not depend so critically on the reference electrode since it is made in a potential independent region of the current potential plot. In the following sections some advances in the development of electrochemical sensors for gases of medical interest are discussed. These include oxygen, carbon dioxide and anaesthetic agents such as nitrous oxide, halothane and isoflurane. 3 . Oxygen: Sensors and Electrochemistry.

The measurement of oxygen concentration or partial pressure is important in many areas. These include aquatic, atmospheric, clinical and laboratory environments. In clinical measurements the monitoring of arterial and veinous oxygen tensions is important (in conjunction with carbon dioxide measurements) in diagnosing and treatment of cardiopulmonary disorders. The lungs are in effect gas exchange devices, allowing the blood to absorb oxygen from the atmosphere and to release carbon dioxide. At the body cell, oxygen is released by the blood and carbon dioxide is absorbed. The efficiency of these processes can be monitored by making blood gas measurements and these can be used to aid diagnosis and, in the operating theatre or intensive therapy unit, to monitor the patients well being. Oxygen sensors are also used as safety devices in anaesthetic delivery equipment. Nitrous oxide is still wideley used as an anaesthetic but is lethal in concentrations above about 80% vlv. Therefore oxygen sensors are used to ensure that oxygen concentration in gases delivered to the patient does not fall below about 20% vlv. A typical use of oxygen sensors in the laboratory is in the

313

assaying of enzymes which catalyse reactions which consume oxygen.

e

Figure 3. The Clark Type Sensor. The b working and reference electrodes, e & d, are part of the main body, c, which is screwed into the external body C a. It is held tightly against the membrane g which is held in place by O-ring f. A thin layer of the electrolyte, b, is held between electrode and membrane.

The measurement of oxygen concentration or partial pressure is usually effected using Clark type sensors (Figure 3 ) . Most innovations in oxygen measurement have been in the engineering of sensors rather than in the electrochemistry. They nearly all rely on amperometric detection following the application of a suitable reducing potential. Innovations in design include miniaturisation for insertion into blood vessels [12,13], the inclusion of heaters for the transcutaneous measurement of blood gas [14,151 and shaping for mounting on the eye for measurement via the palpebral conjunctiva [16,171. It is useful, at this point, to remind the reader of the considered mechanism of the oxygen reduction reaction. Although this not a new development, it is relevant to work which will be described on anaesthetic gas sensors. It is thought that the simplest mechanism involves two parrallel pathways. One is the four electron reduction to water,

314 Oa +

4e- + 4H'

-. 2H2O

(4 1

The other is a two by two electron reduction via hydrogen peroxide , ie. 0, + 2e- + 2H'

-. 2 4 0 ,

(5a)

That the observed current is a function of these two mechanisms is complex enough, but the amount of oxygen being reduced via either route may change with potential. This proposed mechanism is also simplistic because it only considers the electrochemistry and ignores the chemical reactions such as protonation, the cleavage of the dioxygen bond, the chemical disproportionation of hydrogen peroxide without electron transfer to the electrode and the adsorption and desorption of intermediates. It is possible to investigate the mechanism using the rotating ring disc electrode (RRDE) technique. This has been described in detail by Albery and Hitchman [181. Briefly the method is as follows. Oxygen is reduced at the rotating disc electrode and the potential of the concentric ring is set such that the oxidation of hydrogen peroxide is under mass transfer control. Plots are then made of the inverse of the observed collection efficiency versus the inverse of the square route of the rotation speed (in accordance with normal RDE theory). Linear plots are usually observed. Extrapolating to infinite rotation speed allows the prediction of the maximum possible collection efficiency. This is the maximum amount of hydrogen peroxide produced. In terms of the two possible reduction routes (equations 4 and 5 above) this gives the parameter x which is defined as,

315

x =

f l u x of O2 f l u x o f O2

-. -.

H,O

Ha02

(6)

The slopes of these plots give the rate constant for the reduction of hydrogen peroxide. There are many studies on the oxygen reaction and some of these are discused in reference 18. More recent examples are given in references 71 - 7 5 . In terms of the measurement of oxygen in solution it is fortunate that hydrogen peroxide is often minor product and the reduction can be considered solely as a four electron reduction. The measurement of dissolved oxygen has been discussed at length by Hitchman C191. 4 . Carbon Dioxide

The measurement of carbon dioxide is of great importance. Being a gas involved in respiratory processes it can be used to monitor patient wellbeing in the clinical environment and can be used as a tool for investigation by physiologists measuring respiratory functions. It is also important in ecological studies and, in view of the present concern about the greenhouse effect and global warming, in atmospheric monitoring.

4 . 1 Electrochemical Studies.

There is a great wealth of literature concerned with the electrochemistry of carbon dioxide. However it is fair to say that most of this is not concerned with analysis. It is concerned either with the use of carbon dioxide as a synthetic unit or with the recycling of carbon dioxide in the atmosphere to counter the build up of the gas and the global greenhouse effect. It is useful to discuss some of this work as it may be possible to utilise the electrochemistry in working sensors.

316

The electrocatalytic reduction of carbon dioxide has recently been reviewed 1201. Work in this area has concentrated on transition meatal complexes such as metal polypyridyl complexes [21], metal macrocycle complexes [22,231 and the use of different metals and alloys [241. Most of these studies are either mechanistic or in terms of the current efficiency, energy efficiency and selectivity towards products. Products are usually formic acid, methane, methanol or carbon dioxide. The complex of nickel(1) with the ligand cyclam (cyclam 2 1,4,8,11-tetraazacyclotetradecane) has been found to be particularly effective and selective for the reduction to CO at mercury and gold amalgam electrodes [223. Recently Balasz and Anson [231 published the results of an extensive study of the adsorption of nickel cyclarn on mercury. These show that the catalytic species is probably adsorbed Ni(cyc1am)' and that one of several possible configurational isomers is important. It is this metastable configuration that is thought to be the source of the complexes unusually high reactivity. Preliminary results obtained in this laboratory suggest that a linear response to C 0 2 is obtained across a wide concentration range (Figure 4). These results were produced in bulk solution, it has yet to be seen if the effect can be translated to a membrane sensor. 0'Toole and co-workers have investigated the reduction of C02 by polymers of complexes containing rhenium and ruthenium. The main polymer of interest was poly-[(~inylbipyridine)Re(CO)~C11.This was compared to a copolymer with the complex [(bipyridine)ZRu(vinylpyridine)2].Although the investigation was essentially a mechanistic one, it was interesting to note the variation of current observed with concentration of Cot. For the copolymer a linear response was observed for the whole concentration range. For the homopolymer curvature was observed at high C 0 2 concentation. This could be explained mechanistically

317

80

60

I 20

60

40

KO,] /

O/O

80

100

VlV

F i g u r e 4 . Plot of peak current for the reduction of nickel cyclam on gold amalgam vs. C 0 2 concentration (Tebbutt & Wilmott, unpublished results)

and therefore would not be a problem in exploiting this polymer. Nagoaka and coworkers [251 have investigated the reductive addition of C02 to a number of quinones in acetonitrile. For quinones with reduction half wave potentials more negative than 0.9V (vs Ag/Ag' ) large currents were observed due to the addition of COz. This effect has also been reported by Mizen and Wrighton [261 for the reduction of 9,lO-phenanthraquinone. Mechanistic details are given but no actual structures for the adducts are suggested. Nevertheless the coupling is chemically reversible and it may therefore be possible to exploit this reaction as a sensor without loss of the mediator. From the details of the electrochemical response to changes in C 0 2 concentration (Fig 5) it is unclear how such a sensor might operate. It is unlikely that continuous mode operation would be approproate but transient measurements may be possible. A number of studies on the use of different metals for the reduction of C 0 2 have been published. It has been shown that the

318

*

O

O

S

Figure 5. Dependence of voltammetric peaks of 5 mM 1,4napthoquinone on carbon dioxide concentration: [C021 = OM --- , 7.5mM - - - ,411nM.. + . . Sweep rate 50mV/s. Reproduced from Nagaoka et d l [251.

-

..

reduction product depends largely on the metal used as the electrode material. Formic acid (HCOOH) is produced with high faradaic efficiency on metal such as mercury [27,28l,lead [29,301, indium and zinc [30,311 whilst carbon monoxide is the major product on silver and gold [291. Copper has been shown to particularly efficient at producing hydrocarbons such as methane and ethene [32-341. A recent study by Kyriacou and Anagnostopolous C351 has shown that preparation of the electrode surface is also important. The current efficiency of smooth copper electrodes was found to diminish rapidly. This was probably due to adsorption of products and intermediates. Surface roughening (by anodisation) or thermal pretreatment of the electrode lead to changes in adsorption behaviour and extended the life of the electrode significantly. It is thought that these changes are a result of

319

new crystal planes appearing or the formation of catalytic sites as a result of the treatments involved. These results are similar to other studies involving the reduction of organic compounds on platinum [361. Watanabe and coworkers have found that alloy electrodes are often more efficient than electrodes made of pure metals. They have studies the reduction of C 0 2 on copper alloys with nickel, tin, lead, zinc cadmium and mercury [ 2 4 1 . The alloys were all produced by electroplating from mixtures of the metal ions in solution. In the cases of nickel, tin, lead and zinc alloys, current densities and faradaic efficiencies were all found to be greater than those of the pure components. A l s o product selectivity was found to be a function of the metal alloyed with copper. There is one major problem that must be addressed if any of the above studies are to be adapted for COl analysis. This is the reduction of oxygen. Oxygen is likely to interfere with C 0 2 reduction or will be reduced at lower potentials. Therefore oxygen must either be removed from the cell or a separate measurement of oxygen concentration must be made. These might possibly be achieved in one step by incorporating another electrode which will reduce oxygen prior to the C 0 2 measurement. This problem has been addressed in studies concerning the development of sensors for C 0 2 which are discussed below. 4 . 2 Sensors for Carbon Dioxide.

At the present time the most common method of determining CO2 concentration is by potentiometric measurement. The Severinghaus C 0 2 electrode is a pH electrode covered by a gas permeable memrane. The measurement depends on the dissolution of gaseous COl to produce carbonic acid;

The carbonic acid is present as its normal dissociation products, bicarbonate ( HCOJ-) , carbonate ( Cog2-) and protons. If the solution contains sodium bicarbonate at a concentration in excess of 10-3M it can be shown that the partial pressure of C 0 2 , pC02 is given by;

where

ax is the activity of species x a is the solubility coeffiecient of C 0 2 K1 is the first dissociation constant of carbonic acid

The activity if hydrogen ions is measured by the potential at the glass pH electrode. Taking logs of equation ( 8 ) it can be shown that

A 1gpC02

=

-A pH

(9)

Thus a well behaved C02 electrode will register a change of one pH unit, or 59mV, for a ten fold change in C 0 2 concentration. Developments of Cot electrodes, like those of the Clark electrode, have mostly been in terms of changing shape and size. However the equilibrium between hydroquinone and benzoquinone ('quinhydrone')has been exploited as a pH sensitive equilibrium. These developments have been reviewed by Hahn C371. The problem of interference from other gases occurs in all measurements of COz. In potentiometric measurements this may be due to other acidic gases, such as sulfur dioxide and nitrogen oxides. Dissolution of these gases forms acidic solutions which, like the dissolution of C 0 2 causes changes in solution pH. In

32 1

clinical environments these gases are not usually present and so this is not an inconvenience. The problem of interference from acidic gases is also be a problem when conductometric measurements are made. These rely on the change in conductivity of a sample when dissolution of a gas involves hydrolysis as shown in equation (7). The change in conductivity depends on both the nature of the gas and the solution in which it dissolves. If the gas dissolves to form ions the conductivity would be expected to increase but if an acidic gas dissolves in a basic solution the the conductivity may decrease as ions are effectively removed by neutralisation. In a sensor developed by Bruckenstein and Symanski C381 the conductance of a sample of pure water increases as CO1 dissolves. The sensor is shown schematically in Figure (6).

Figure 6 . Schematic representation of the continuous conductometric sensor. Reprinted with permission from S Bruckenstein & J S Symanski. Anal Chem 58 ( 1986), 1766. Copyright 1986 American Chemical Society.

C02 diffuses through the PTFE membrane (F) and dissolves in the pure water layer (D). At the rear of the water layer is a water porous spacer screen (C) which supports a mixed bed ionexchanger ( A ) . The ion-exchanger removes ions produced by the continuous dissolution of C02 and a steady state concentration gradient of C02 is established as a result of these two processes. The sensor is mounted in a Pyrex tube ( B ) and the

322

conductance is measured using platinised electrodes which are deposited on the surface of the membrane in contact with the water. It is found that the conductance is proportional to the concentration gradient and is proortional to ( pCOl)l/’. The sensor was only calibrated for partial pressures up to 1% (v/v). The response was rapid for step changes up to 0.05% but much slower for larger partial pressures. The response for decreases in partial pressure was as long as 180 seconds. In its favour the method would supply a cheap and convenient way of analysing C02 but there are problems with sensor drift and interference from other gases. When attempting the amperometric measurement of COz, interference from other gases is still a serious problem. Any gas with a standard electrode potential below or near that of C02 may react at the electrode. Although most of these gases will not be present in clinical environments, or may only be present in small amounts, oxygen is nearly always present. This problem has been approached in several ways. If separate reduction waves are observed for O2 and CO1 and there is no interference between the two it may be possible to measure the two gases using a double pulse method. This is analogous to the determination of oxygen and nitrous oxide on silver ( v i d e infra). If the pulses are short enough this might be considered to be a simultaneous determination. In aqueous solution the reduction of C 0 2 usually occurs at very high potentials and cannot be distinguished from the hydrogen evolution reaction. This approach is therefore not appropriate. In aprotic media, although the reduction of C02 can be observed a more interesting case arises as carbon dioxide interferes with the oxygen reduction reaction. This is shown in Figure 7 . In the absence of C 0 2 the reversible one electron reduction of oxygen to superoxide ion is observed. When C02 is present the anodic peak disappears and the cathodic peak is enhanced. This is typical of

323

an electrochemical system where the product of the electron transfer reacts with another species in solution and the starting material is regenerated. Thus no reverse peak is observed since the initial product is removed by chemical reaction and the forward peak is enhanced due to regeneration of the starting material (which undergoes further electron transfer). The reaction between C02 and superoxide ion has been investigated by Sawyer and coworkers [391 and by colleagues in my own laboratory [ 4 0 ] . The reaction has been exploited by the latter in the development of a working sensor.

0

Figure 7 . CV for the reduction of oxygen (0.9mM) in DMF/ te tr a et h yl ammo ni urn perchlorate, in the absence (solid line) and presence of carbon dioxide. Reprinted with JL permission from Roberts, TS Calderwood & DT Sawyer, J Am Chem Soc, 1 0 6 ( 1 7 ) , 1 9 8 4 , 4 6 6 8 . Copyright 1 9 8 4 American Chemical Society

-0-5 -1-0 -1.5 E/V vs. SCE

The following reaction scheme has been proposed to explain the observed current response of a system where both oxygen and carbon dioxide are present 1 4 0 1 .

324 0, + e- -. 0;-

q - + co,

-

Cod- + C02

c,og-

+

q-

( electrode)

(solution)

(20;-

-+

-

(solution)

C20e-

c20:-

+ 0,

(solution)

After controlled potential electrolysis a deposit appears on the electrode surface. The laser raman spectum of this is consistent with the percarbonate species produced in the last step of this scheme [ 4 0 1 . A sensor based on this reaction scheme has been described by Hahn [ 3 7 1 . This is a Clark type sensor which operates using a double pulse method. The first pulse is to a potential at which the reduction of oxygen occurs at its diffusion limited rate. The integrated current is used as a measure of oxygen concentration. The second pulse is to a potential where the superoxide ion is oxidised back to oxygen. In the absence of C 0 2 the integrated current should match that of the forward pulse. When C 0 2 is present the reverse pulse is diminished. The difference between the two pulses is therefore a measure of the amount of carbon dioxide present. The interested reader is referred to the review by Hahn for further details of this system. Although this sensor is interesting it has a number of disadvantages. One is that in long term use build up of reaction products (such as percarbonate salts) may adversely affect the sensor performance. Another is that the solvent used, dimethyl sulphoxide rapidly absorbs water. Protonation of the superoxide ion would compete with the C 0 2 reaction and affect the response. Maintaining a dry solvent would therefore be a major problem in

325

most sensor applications. This has been highlighted in a recent study by Hamnett's group [411. Using in s i t u Fourier Transform Infra Red spectroscopy they showed that the product of carbon dioxide reduction was in fact carbonate ion which could only have formed if there was water adsorbed on the surface of the electrode. All sensors based on non-aqueous solvents will suffer from this problem. A sensor based on aqueous electrochemistry would therefore be preferable. Nevertheless workers interested in the non-aqueous electrochemistry of C02 will no doubt find the publication of solubilities of the gas in various non-aqueous solvents a useful aid [681. Evans and coworkers have developed a sensor which relies on the change in pH of a solution when CO1 dissolves [42,431.

OIO

coz

Figure 8. (left) Calibration curve for the sensor in Figure 9. Figure 9. (right) Schematic representation of the carbon dioxide sensor designed by Evans, Pletcher et al. This is a three electrode design with working, a, and counter,h, electrodes consisting of porous PTFE discs,b, coated with platinum or carbon,f. The reference, d, is a silver wire which is separated from the other electrodes by filter pads c. The electrolyte is in a reservoir, 8, and is transferred to the electrodes by a wick, g. Both figures reproduced with permission of the authors, Anal Chem 61, 1989, 577. Copyright 1989 The American Chemical Society.

326

The solution used is copper ( 1 1 ) bis-1,3-propanediamine in potassium chloride. This complex dissociates reversibly in the presence of hydrogen ions,

The copper ion produced by this dissociation is then reduced at the electrode to copper (I) in the presence of excess chloride ion (which stabilises the univalent copper),

This reduction occurs at 0.15V (vs Ag/Ag'), well removed from the oxygen reduction reaction and has been shown to be unaffected by the presence of oxygen. Obviously the position of the equilibrium will be affected by the pH of the solution which changes according to the dissolution of CO1. The response to C02 (Fig 8 . ) is dependendent on many variables; temperature, copper ( 1 1 ) bis 1,3-propanediamine concentration, potential, electrolyte and electrode material. The working sensor (Fig 9 ) has a porous carbon or platinum working electrode and contains a solution of 2.5mM copper complex and 1M potassium chloride. The response is rapid and though non-linear was found to be reproducible. The response is also largely independent of oxygen concentration although this depends on a suitable choice of potential. The only serious interferents are, as expected, other acidic gases such as sulfur dioxide and nitrogen dioxide. Finally, a note on the calibration of sensors for carbon dioxide. This is traditionally achieved using commercially available gas mixtures or producing mixtures using flow meters. In some laboratories sensors may be calibrated against a mass spectrometer although the expense of such instruments limits their use considerably. In a recent communication Calabrese and Maillet describe a photochemical method of producing reproducible

321

levels of C02 and O2 in solution [441. The method is based on the photooxidation of oxalate:

where Ox = CIO, Irradiation of ferrioxalate solutions with light between 250 and 500nm rapidly leads to constant C02 concentration in solution which depends on the initial concentration of the ferrioxalate solution. The overall reaction is somewhat more comlicated than that suggested by equation (13) and involves the consumption of oxygen. Thus the level of oxygen may also be controlled. Complete experimental details are given in reference 44. 5. Anaesthetic Gases The measurement of concentrations of anaesthetic gases in inspired and expired air and in the blood is not routinely carried out. It is of secondary importance compared to the monitoring of the respiratory gases. On the other hand it is of interest to physiologists who may wish to monitor the uptake of such gases by the body. There is also interest in monitoring the amount of anaesthetic gas which escapes into the atmosphere of the operating theatre [451. This could be used to monitor the long term exposure of theatre staff to low levels of anaesthetic gases. Sensors for anaesthetics could also be incorporated in feedback mechanisms in automatic anaesthetic delivery equipment. Gases of interest are nitrous oxide and the inhalation vapours (or volatile halogenated anaesthetics) such as halothane ( CF3CHBrC1) , isoflurane ( CHF20CHC1CF3) and enf lurane (CHC1FCF20CHF2). In this section electrochemical studies on nitrous oxide, halothane and isoflurane will be discussed.

328

5.1 Nitrous Oxide The electrochemistry of nitrous oxide (N20) has been of interest since it was found that the presence of the gas can interfere with the routine monitoring of oxygen with Clark electrodes [461. This is also the case for the interest in halothane electrochemistry (vide infra). This interference has been traced to the plating of silver on the working electrode from the silver - silver chloride reference electrode used in commercial Clark electrodes [ 4 7 1 . It has been demonstrated that N20 can be reduced on silver 1481. A recent study by Parsons and coworkers has shown that the gas can also be reduced on platinum [491. Using single crystal electrodes they showed that the reduction is closely associated with the hydrogen evolution reaction. This my also be a source of error in oxygen measurement if too high a reduction potential is applied to the working electrode. On silver the reduction of N20 occurs as a single two electron step

N,O

+ H,O

+

2e

-

N,

+

20K

Prolonged electrolysis results in the build up of gaseous nitrogen behind the membrane and therefore transient measurements must be made. The half wave potential for N 2 0 reduction is -1.15V (vs SCE) compared to -0.425V for oxygen [571. There appears to be no chemical interference beteen the two and it is therefore possible to measure the concentrations of these gases in mixtures by sampling the current at the diffusion limited plateaus of the two reduction waves. The first gives the concentration of O1 and the second the sum of the concentrations of both. The concentration of N20 is therefore given by the difference. A

329

working in viva sensor for both gases using this principle has been demonstrated [2,4,501. Recently Compton and Northing have described a method for the measurement of N20 in acetonitrile [511. They have shown that N 2 0 reacts with the radical anion of the organic compound fluoranthene (Figure 10). Fluoranthene is reduced on mercury plated copper electrodes at potentials more negative than 1.5V ( v s SCE). The reaction in the presence of N20 is a classical EC’ or catalytic mechanism, ie.

The sensor designed by this group is also of interest. Instead of the Clark type sensor they have modified a channel electrode employing a rectangular flow cell. This is shown schematically in Figure 11. The cell is a rectangular channel with inlets and outlets for solution. The flow is usually controlled by a gravity feed. The gas permeable membrane is placed upstream of the rectangular working electrode. In thee studies this was a copper strip plated with mercury. The gas diffuses through the membrane and is carried by the flow over the working electrode. The flow of solution in the channel can be mathematically modelled and the observed limiting current is seen to follow a Levich equation (analogous to the RDE),

330

where U is the mean solution velocity in the cell h, w and x, are defined in Figure 11. [S]* is the bulk concentration of solution species eg. fluoranthene. F and D are the Faraday constant and diffusion coefficient respectively. The theory of channel electrodes have been fully reported elsewhere [ 5 2 1 .

x='o

membrane

detector electrode

x&:;mx=x5

x=xc

Figure 10. (left) The structure of fluoranthene, used in the mediated reduction of anaesthetic gases by Compton and coworkers. Figure 11. (right)Schematic representation of the flow cell used by-Compton and- co-workers. Reproduced from Compton and Northing [511 In this system the catalytic current is shown to vary with NtO as expected for the above mechanism. The concentration of N20 can also be measured in the presence of oxygen. This is achieved by exploiting the reduction of oxygen by the anthraquinone dianion;

where AQ is anthraquinone. This is included in the solution and the anion is generated by an electrode placed upstream of the membrane.

33 1 5 . 2 Halothane

Halothane is a widely used inhalation anaesthetic which gives rapid induction and recovery from general anaesthesia.It was originally introduced as a replacement for ether. Being nonflammable it was considered to be much safer. Its chemical instability, however, means that it is potentially toxic in long term or repeated use and possible liver damage to surgical peraonnel from chronic, subclinical environmental exposure has been suggested [531. That halothane can be electrochemically reducedwas demonetrated by Severinghaus in1971 [ 5 4 1 and Stossek C 5 5 1 and Batsa in 1975 [561. It is reduced at potentials close to the oxygen reduction reaction on silver (hence the source of interference), ie, w i t h a half wave potential of -0.565V (vs S C E ) . On gold it is reduced at more negative potentials, with a half wave potential of -0.94V (vs SCE) 1 5 7 1 .

-

-

Hall and coworkers [571 suggest that it is possible to deconvolute oxygen and halothane currents on gold using a single pulse. The current transien!: is integrated at short times (up to 20me) and at long times (after 70ms). Manipulation of the data is complicated but is controlled entirely by computer. The result is a halothane measurement independent of oxygen concentration. The response to halothane is, however, non-linear but reproUucible.The authors also consider the effect of age on the sensor response. Thia aging effect is considered to be largely due to silver deposition. The interested reader is referred to the original paper for full details of the discussion. My own approach to this problem hag been to observe the effect of halothane on the oxygen reduction reaction [ 5 8 1 . As is shown in Figure 12. continued potential cycling of a gold electrode in

solution saturated with halothane- nitrogen mixtures results in some passivation of the surface. The oxygen reduction reaction wa8 then atudied on this surface and compared to that on the clean gold electrode, T h i s was done using the method described above. a

332

Figure 12. Partial passivation of gold by repeated potential cycling in the presence of halothane. Reproduced from Tebbutt & Hahn C581 12a

a

10-

K

8-

3

9

60 0

A

4-

2-

I

012

d.4

0.;

I

I

0.8

1.0

(wmz)""

Figure 13. Plots of inverse collection efficiency vs. inverse rotation speed for the reduction of oxygen on clean gold (filled circles) and in the presence of increasing amounts of halothane (all other points). Reproduced from Tebbutt & Hahn l583 Figure 13 shows plots of the inverse of the observed collection efficiency versus the inverse square r o o t of the rotation speed. These are on clean gold and in the presence of increasing amounts of halothane in solution.

333

€1

El

Figure 14. Defining the shape of the two wave polarogram as the ratio of to il.

5-

4-

5 320-

1-

1

-

t

-

-0

I

I

I

1

1

20

40

60

80

100

I

1021 v/v %

4-

[halothane] / % v/v

Figure 15. Plots of iz/il vs. oxygen concentration, left, at increasing concentrations of halothane, right. Reproduced from Tebbutt & Hahn 1 5 8 ) .

The implication of this result is clear. Since these p l o t s all collapse onto one line (within experimental error) in the presence of halothane the mechanism of the reaction must be unnaffected by the presence of increasing amounts of halothane in solution. The mechanism is, however, different to that observed on the clean metal surface.

3 34

The result is that the shape of the voltammogram for oxygen reduction is independent of the halothane concentration. The observed voltammogram is the sum of voltammograms for oxygen and halothane reduction. The shape of the voltammogram maybe defined mathematically as the ratio of the height of the second wave to that of the first wave (Fig 14). This, as can be seen in Figure 15, is independent of the oxygen concentration but is proportional to the halothane concentration. Also the response is linear over the whole of the concentration range which is commonly used. Transfer of this system to membrane bound sensors has unfortunately not been straight forward. The method works well at high oxygen concentration but curvature is observes at lower concentrations. Further work is obviously necessary to solve this problem, 5 . 3 Isoflurane

Isoflurane is a halogenated ether and potent anaesthetic. It is stable chemically (and therefore less toxic than halothane), photochemically and electrochemically. This obviously presents problems when trying to develop methods of measuring its concentration using chemical or electrochemical means. Compton and coworkers have investigated the mediated reduction of isoflurane on mercury by the radical anion of fluoranthene [591. Figure 16 shows RDE voltammograms for the reduction of fluoranthene in the presence of increasing amounts of isoflurane. As can be seen the current increases with isoflurane concentration. The mechanism of this reaction has been shown to be one of the many variations of the catalytic mechanism. Compton terms it the EC' (pre-eqm) since it involves a pre-equilibrium prior to the irreversible product forming step;

335

3-

2Q

E

\

c

C

e 1L

3

V

0-

SCE Figure 16. RDE voltammograms for the reduction of fluoranthene in the presence of (a) 0.0, (b) 1.7, (c) 3.3, ( d ) 5.0mM ispflurane. Reproduced from Compton et al [ 5 9 1

F + e - + F

F-

+

isoflurane

intermediate

-

-

F + intermediate

products

The mechanism was deduced from rotating disc data. This involves producing a mathematical model of the expected responses for the various nuances of the mechanism and comparing the experimental result to theoretical working curves. The method has been fully described by Compton and coworkers [ 6 0 1 . The above mechanism can be deduced since a smaller catalytic current than would be expected for a simple CE' process is observed. This occurs because when more of the reactive species, F , is added the equilibrium is shifted to the left. Thus less Fis recycled.

In this report Compton et a 1 describe the synthesis and use of poly-(vinylfluoranthene) and its use in mediating the reduction of isoflurane. The polymer was deposited on a 2platinum electrode. This film exhibits reversible electrochemistry in acetonitrile and catalytic behaviour when isoflurane is present. Unfortunately the films were unstable and could not be used for long term monitoring of the gas. Finally the flow cell described above has been used to monitor isoflurane in the presence and absence of oxygen (using the AQ2scavenger method) [51]. The response time to changes in gas concentration is less than ten seconds and it is suggested that this could be easily improved. Furthermore the mediated reductions of nitrous oxide and isoflurane are considered to be independent of each other. Thus it is also suggested that the sensor could be used to monitor both gases provided there was an independent measurement of one of them available. This however, remains to be seen. 6. Solid State Gas Sensors

These types of sensors are represented by semiconductor devices and devices incorporating solid electrolytes. These have certain advantages over sensors utilising liquid electrolytes. These are the inherent, robust nature of such sensors, the lack of problems due to evaporation of solvent or corrosion due to spillage, the possibilities of miniturisation and the possibility of mass production. Sensors based on solid electrolytes usually operate in the potentiometric or amperometric mode but the semiconductor devices usually operate by measuring changes in conductivity.

6.1 Semiconductor Devices

These rely on the changes in electrical resistance brought about by the adsorption of gases on the surface of the semiconductor

331

material. They may be thin film, thick film or bulk pellet devices. Most use tin oxide and represent the commercially most successful sensors for the control of fuel and oxygen in combustion engines [ 6 1 . They have also been used in devices for measuring leakages of town gas and other toxic gases 1 6 1 1 and for the determination of low level vapours such as odourous components [621. The observed change in conductivity and the mechanism by which this occurs has been the subject of much study [ 6 3 1 . It is, however, apparent that changes 3re due to changes in surface states and charge carrier concentration brought about by the adsorption of oxygen. If combustible gases are includedthese may react with the adsorbed oxygen and thus affect the conductivity of the uderlying semiconductor [ 6 4 1 . Selectivity may be improved by addition of a metal or metal oxide to the surface. Two mechanisms for this improvement in selectivity are proposed. One is the spillover effect. This is the adsorption of molecules onto the metal particle which may then overflow onto the semiconductor. Platinum is thought to operate by this mechanism and has been shown to improve sensitivity to carbon monoxide [ 6 1 . The other is when the metal or metal oxide can be considered to be a sensitizer. In this case the adsorption causes a change in oxidation state of the sensitizer and this, in turn, affects the surface states of the underlying semiconductor. Metal oxides such as silver oxide and palladium oxide are considered to be sensitizers. Examples of improved selectivity are sensors for hydrogen sulfide using zinc oxide 1 6 5 1 as a sensitizer and a sensor for ethanol using lanthanum oxide (LaZOj)[ 6 4 1 . Semiconductor devices have been reviewed in more detail by Yamazoe [ 6 4 ] and many more examples (such as the use of other materials such as pthalocyanines etc) appear in the reviews and conference proceedings mentioned in the introduction.

6.2 Devices Based on Solid Electrolytes.

These devices use fast ionic conducting solids as the electrolyte which is in intimate contact with solid electrodes. The mechanism of conduction of such solids have been described by Alcock [661 using the mobility of ions through the crystal lattice although more complex band models have also been described C671. One of the most common materials used in these devices is doped zirconia (ZrOZ). The addition of lower valent metal oxides such as calcium oxide stabilize the cubic crystal structure which is normally only stable at much higher temperatures. The structure thus produced has a number of oxygen sites missing and on heating to 600 " C 02- ions are able to move into the vacant sites. Because of this temperature dependence sensors based on this material operate only at high temperatures. This is suitable in the monitoring of combustion effiency in boiler flues or in engines but is obviously not suitable for clinical applications. Another material commonly used is l3-alumina, (originally named as a form of A1203but actually contains other metal ions such as Nat which stabilise the structure). Conductivity arises from the mobility of loosely bound sodium ions in layers between the more rigid spinel type layers of the alumina. Mobility is such that significant conductivity is observed even at room temperature. Other fast ion conductors are a-AgI, stable above 146°C in which the small silver i o n s can move over the surface of the rigidly placed iodide ions, avoiding the points of contact. A related silver ion conductor is RbAg415which is stable and conductive at room temperature. Antimonic acid (HSb03.nH20)is a fast protonic conductor. Most sensors are potentiometric and are based on concentration cells. A typical arangement is shown schematically in Figure 17. The solid electrolyte, b, is sandwiched between two electrodes, c. One is exposed to the test gas and the other either to a known concentration of gas or some reference half cell. The potential

339

Test Gas

17. Schematic representation of a typical concentration cell based on a solid electrolyte. Figure

a

Reference

Gas

difference is measured by the external circuitry d and the whole system is enclosed on a tube a. The potential difference arises because of the potential dependence on concentration or partial pressure and is therefore expected to be Nernstian. This has been shown to be the case for a hydrogen gas sensor based on antimonic acid, although at low temperatures ( < 3 0 0 K )kinetic factors become important [ 6 9 1 . In the case of sodium-R-alumina the Na/Na’ couple has been used using sodium as the electrode material. Wienhoffer and Gopel have defined a number of modes of operation [67]. One is the direct participation of the mobile ion in the electrochemical equilibrium involving the gas of interest. Thus the mobile oxygen dianion in zirconia is in equilibrium with gaseous oxygen;

and the potential at the electrode will be defined by the Nernst equation. Related to this (and given the same classification) is

the use of some sensitive layer on the surface of the electrolyte. This usually consists of a component in common with the ionic conductor and a component in common with the gas. An examples is the coating of silver iodide with silver chloride as a chlorine sensor [ 6 7 1 in which the electrochemistry of the mobile silver is coupled to the electrochemistry of chlorine by the common chloride in the sensititizing layer. Others are the use of sodium sulfate (Na2S04)and sodium carbonate (Na2C03)as sensitive layers on sodium conductors for sulfur dioxide [671 and carbon dioxide C701 respectively. The second type is where the electrolyte bahaves as a classical electrolyte an takes no part in the electrochemistry. Rather it acts only as a solvent. Although these sensors have the potential for operation at lower temperatures they do rely on high current densities and high concentration of the gas in the solid phase. A more in-depth review of the theoretical aspects of fast ion conductors as sensors and more examples are given in reference. As before, the interested reader is referred to the conference proceedings listed in the introduction. Concluding Remarks.

In this review I have attempted to highlight some of the recently published work on the electrochemistry of gases which is related to the development of sensors for these gases. Although work on solid state devices is somewhat under-represented I am sure that in the near future these will be more promising as rapid sensors for gas detection and measurement at room temperature. However, as a solution electrochemist I am also sure that there is much interesting work to be done on the solution electrochemistry of pure gases and gases in mixtures.

34 1

REFERENCES 1. Megede DZ. S e n s o r s & A c u a t o r s B,1991, B4( 3-4), 373-8, 2. Hahn CEW, Brooks WN, Albery WJ & Rolf P. A n a e s t h e s i a , 1979; 34, 263-6 3. Brooks WN. Electroanalytical Techniques For Anaesthetics; 1980; D P h i l T h e s i s , University of Oxford. 4. Brooks WN, Hahn CEW, Foex P, Maynard P & Albery WJ. B r J A n a e s t h 1980; 52, 715-22. 5. Bartlett PN, Tebbutt P & Whitaker RG. P r o g R e a c t i o n Kinetics 1991; 16(2), 55-156. 6. Janata J & Bezegh A. A n a l Chem 1988; 6 0 , 62R-74R. 7. Wohitjen H. A n a l Chem 1984; 56(1), 87A-103A. 8. Liu C-C & KO WH, eds. Proceedings of Third International Meeting on Chemical Sensors, Cleveland OH, 1990. S e n s o r s & A c t u a t o r s B,1991, B5(1-4). 9. Hardtl KH, ed. Proceedings of Eurosensors IV, 1991. S e n s o r s & A c t u a t o r s B 1991; B4(1-4). 10. Clark LC. T r a n s Am SOC A r t i f Intern O r g a n s 1956; 2, 41-8. 11. Severinghaus JW & Bradley AF. J A p p l P h y s i o l 1 9 5 8 ; 13, 515-20. 12. Mindt W. Proc J a h r e s t a g u n g D Ges Biomed Tech Erlangen 1973; 29-30. 13. Parker D & Soutter L. In. Payne JP & Hill DW, eds. Oxygen Measurements in Medicine & Biology. London: Butterworths, 1975; Ch. 18, 269-283. 14. Eberhart P , Hammacher K & Mindt W. Biomed T e c h n i k 1973; 18, 212-221. 15. Huch R , Lubbers DW & Huch A. In: Payne JP & Hill DW, eds. Oxygen Measurement in Medicine & Biology, London: Butterworth, 1975; Ch.9, 121-138. 16. Fink S, Wayne WC, McCartney S et al. C r i t C a r e Med 1984; 12, 943-8. 17. Abraham E, Smith M & Silver L. Ann Emerg Med 1984; 13, 1021-6. 18. Albery WJ& Hitchman ML. Ring Disc Electrodes. Oxford: Clarendon Press, 1971. 19. Hitchman ML. The Measurement of Dissolved Oxygen. New York: John Wiley, 1978. 20. Sullivan BP, Bruce MR, O'Toole Tr, et al. In: Ayres WM,ed. Catalytic Activation of Carbon Dioxide; ACS S y m p o s i a S e r i e s No. 366. Washington DC: ACS, 1988, 52-90. 21. O'Toole TR, Sullivazn BP, Bruce MR-M et al. J E l e c t r o a n a l Chem 1989; 259, 217-239. 22. Fujihara M, Hirata Y & Suga K. J Electroanal C h e m 1990; 292,199-215. 23. Balazs GB & Anson FC. J E l e c t r o a n a l C h e m 1992; 322, 32545. 24. Watanabe M, Shibata M , Katoh A, et al. J Electroanal Chem 1991; 305, 319-28. 25. Nagoaka T , Nishii N, Fujii K & Ogura K. J E l e c t r o a n a l Chem 1992; 322, 383-9. 26. Mizen MB & Wrighton MS. J Electrochem Soc.1989; 136, 941. 21. Udupa KS, Subramanian GS & Udupa HVK. Electrochim Acta 1971; 16, 1593.

342

28. Russel PG, Kovac N, Srinivasan S & Steinberg M. J E l e c t r o c h e m SOC 1977; 124, 1329. 29, Hori Y, Kikuchi K & Suzuki S, Chem L e t t 1985; 1695. 30. Ikeda S, Takaga T & Ito K. B u l l C h e m SOC J a p a n 1987; 60, 2517. 31. Kapusta S & Hackerman N. J E l e c t r o c h e m SOC 1983; 130, 607. 32. Hori Y, Kikuchi K, Murata A & Suzuki S. Chem L e t t 1986; 897. 33. Cook RL, MacDuff RC & Sammells AF. J E l e c t r o c h e m SOC 1987; 134, 2375. 34. Cook RL, MacDuff & Sammells AF. J E l e c t r o c h e m SOC 1988; 135, 1320. 3 5 Kyriacou G & Anagnostopoulos A. J E l e c t r o a n a l Chem 1992; 322, 233. 36. Luna AC, Giordanao MC & Arvia AJ. J E l e c t r o a n a l Chem 1989; 259, 173. 37. Hahn CEW. C l i n i c a l P h y s i c s , P h y s Meas 1987; 8, 3-38. 38. Bruckenstein S & Symanski JS. A n a l Chem 1986; 58, 1766. 39. Roberts JL, Calderwood TS & Sawyer DT. J Am Chem SOC 1984; 106, 4667. 40. Tebbutt P, Clark D, Robinson G I e t al. In Smyth MR & Vos JG, eds. E l s e v i e r A n a l y t i c a l S y m p o s i a S e r i e s Vol 25; Electrochemistry, Sensors & Analysis, 315-322. 41. Christensen PA, Hamnett A, Muir AVC & Freeman NA. J E l e c t r o a n a l Chem 1990; 288, 197-215. 42. Evans J, Pletcher D, Warburton RG & Gibbs TK. Anal Chem 1989; 61, 577-580. 43. Evans J, Pletcher D, Warburton PRG & Gibbs TK. J E l e c t r o a n a l Chem 1989; 262 , 119-29. 44. Calabrese GS & Christian-Maillet L. A n a l Chem 1992, 64, 120-3. 45. Eiceman GA, Shoff DB, Harden CS, e t a l . A n a l Chem 1989; 61, 1093-99. 46. Evans MC, Cameron IR. L a n c e t 1978; 11, 1371. 47. Gibson SP. Honours T h e s i s , Universuty of Oxford, UK, 1976. 48. Albery WJ, Brooks WN, Gibson SP & Hahn CEW. J A p p l P h y s i o l 1978; 45, 637-43. 49. Ebert H, Parsons R , Ritzoulis G & VanderNoot T. J E l e c t r o a n a l Chem 1989; 264, 181-93. 50. Brooks WN, Hahn CEW, Foex P I Maynard P. B r J A n a e s t h 1978; 50, 1082-3. 51. Compton RG & Northing RJ. J Chem SOC F a r a d a y T r a n s , 1990; 86(7), 1077-81. 52. Unwin PR & Compton RG. In: Compton RG & Hamnett A, eds. 1

C o m p r e h e n s i v e C h e m i c a l Kinetics. Amsterdam, Elsevier.

1989; 29, 173-296. 53. Spence AA. B r J A n a e s t h 1987; 59/96. 54. Severinghaus JW, Weiskopf RBI Nishimura M & Bradley AF. J A p p l P h y s i o l 1971; 31, 640-2. 55. Stosseck K. A n a e s t h e t i s t 1977; 26, 453-6. 56. Bates ML, Feingold A & Gold MI. Am J C l i n P a t h o l 1975; 64, 448-51. 57. Hall EAH, Conhill EJ & Hahn CEW. J B i o m e d Eng 1988; 10, 319-25.

343

58. Tebbutt P & Hahn CEW. J E l e c t r o a n a l Chem 1989; 261, 20516. 59. Compton RG, Northing RJ, Waller AM, e t a l . J E l e c t r o a n a l Chem 1988; 244, 203-19. 60. Compton RG, Day MJ, Laing ME, e t a l . J Chem S O C . F a r a d a y T r a n s . 1. 1988; 84(6), 2013-25. 61. Kanefusa S , Nitta M & Haradome M. J Chem SOC J a p a n 1980; 1591-5. 62. Shimizu Y, Takao Y & Egashira M. J E l e c t r o c h e m SOC 1988; 135, 2539-40 63. McAleer JF, Mosely PT, Norris JOW & Williams DE. J Chem S O C . F a r a d a y T r a n s 1, 1987; 83, 1323-1346. 64. Yamazoe N. S e n s o r s & A c t u a t o r s B5, 1991, 7-11. 65. Nakahara T, Takahata K & Matsuura S. P r o c S y m p C h e m i c a l S e n s o r s , Honolulu H I USA, 1987; 55-64. 66. Alcock NW. Bonding & Structure, Structural Principles in Inorganic & Organic Chemistry. London, Ellis Horwood; 1990. 67. Weinhoffer H-D & Gopel W. S e n s o r s & A c t u a t o r s B , 1991; B4( 3-4), 365-72. 68. Gennaro A , Isse AA & Vianello E. J E l e c t r o a n a l Chem 1990; 289, 203-215. 69. Forano C & Besse JP. M a t e r i a l s Chem & P h y s , 1988; 19, 56777. 70. Liu J & Weppner W. E u r J Solid S t a t e I n o r g Che m , 1991; 28,1151-60. 71. Strbac S, Anastasijevic NA & Adzic RR. J E l e c t r o a n a l Chem 1992; 323, 119-195. 72. Wroblowa HS & Qaderi SB. J E l e c t r o a n a l C h e m 1990; 295, 153-161. 73. Biloul A , Coowar F, Contamin 0 e t a l . J E l e c t r o a n a l Chem 1990; 289, 189-201. 74. Su YO, Kuwana T & Chen S-M. J E l e c t r o a n a l Chem 1990; 288, 177-195. 75. Sawaguchi T, Itabashi T, Matsue T & Uchida I. J E l e c t r a n a l Chem 1990; 279, 219-230.

This Page Intentionally Left Blank

SECTION FOUR

TECHNIQUES FOR POLLUTION MONITORING

This Page Intentionally Left Blank

347

Environmental Pollution Monitoring using DC pulse technique M.I. Ismaila. A. Asirih, and

A.

C

Al- Turkait

aSci & Ad Inst, P.O. Box 98029 S . Common Post Outlet 2150, Burnhamthorpe Rd.Miss..Ontario, Canada L5L 3A0.

bP.O.Box 4191 Jeddah. 21491 Kingdom of Sauidi Arabia C College of Basic Education, PAAET, P.O. Box 38552 Abdulla A1-Salem.72256 Kuwait

INTRODUCTION The DC pulse response method is used as nondestructive techhnique (NDT) for measurement of variations in the composition and characteristics a s a result of environmental pollution. The response to short DC pulse (e.g. 9V for 0.5 s ) in a wire cell with elecrodes in Contact with the polluted Media 1 1 1 proved to Correlate to the Contaminant level and type 12-41. The proposed electrochemical technique is of prime interest for in-situ monitoring of environmental pollution various phases (Media). Details about Contaminated Soil , Water , Plants and Atmosphere a s aresult of oil field burns in Kuwait in 1991 are presented in this chapter .

ELECTROCHEMICAL MONITORING Electrochemical techniques are known for their use a s evaluation test methods. They are fast analytical techniques. Several applications are used in various fields such as quality control tests. The response to DC pulse is a typical example in this respect 151. I t is sensitive to the physical, chemical. and electrochemical characteristics of the tested Media. The DC pulse technique The response of short duration DC pulse imposed in wire electrodes correlates to the composion and property of the media. The following are more details about the DC pulse proposed for monitoring environmental pollution of any media

348

whether single or multiphase system(s). The DC pulse device The DC power supply imposes the applied potential in the wire cell [11. The open cell voltage (OCV) and load voltage could be in the range 3 to 9 V DC for a short period. The time of application of the load voltage should be less than that neederd for evoloution of hydrogen or other ion discharge which lead to gases evolution at the electrodes of the utilised wire cell. Half second is considered fair enough time for the DC pulse application in the copper wire electrodes. The commercial device [ 6 3 has its timer and controller to keep the duration of the DC pulse constant in all the experiments. However , ordinary 9 V DC cells could be utilised using conventional on-off switch for the application of the DC pulse in the wire cell.

The measurement of the response to DC pulse

High

impedance

millivoltmeter

is

used to measure the response to short duration (for 0.5 s > DC Pulse applied to the copper wire pair (the probe). The monitored mV signal level i s function of the separation distance between the electrodes in the utilsed copper wire cell. and duration of the applied DC potential has the signal level. Pulse duration should be during the experimental set. The variation of

The amplitude its effect on kept constant the DC pules

response correlates with the media composition and other characteristics. The time - OCV curves, known a s the decay curves, are used for the comparison between the various experimental conditions in the monitoring program. Aspracticized by all the standart testing devices, a refernce signal is utilized. Standard samples are used for this purpose to generate reference signal.

349

The signal measurement steps The basic steps of the experimental operation of the DC pulse response technique proposed for monitoring the environmental pollution level is as simple as follows 1. An

in

electrolyte is used to make the wire cell electrodes

physical contact with the test media. A simple filterpaper. pre-wetted with water, will act as electrolyte a

holder. The possibility of theoccurance of

shortage between

electrodes is minimized when the wire electrodes contact conducting metallic surface if this situation is applicable in the monitorng process 2 . Apply 9 V to the copper wire electrodes contacting the test media through the wetted paper separator. 3 . Put the power "on" for 0.5 s in the copper wire electrodes and measure the decay in the open cell voltage after the is "off" using high impedance power(app1ied potential) millivoltmeter. 4. Record the decay curves for every sample in the experimental set used. 5 . use freshly cut end copper wires (the wire cell) in every experiment to keep exactly the same experimental condition to minimize the experimental error.

THE DC PULSE SYSTEM The DC pulse response measurements are useful for fast monitoring of contaminant level in the tested media. Figure 1 shows a schematic experimental set-up of the test. The generated electrical charge, as a result of the DC pulse, dissipates in the monitored media at a rate proportional to its impedance and resistance. The rate at which the OCV decays is an indication of the media contamination with the monitored pollutant. The transient time from the application

350

2

Figure 1 The experimental set-up of the DC pulse response (schematic) 1, DC pulse power supply or DC battery, 2, Copper electrodes wire cellIll; 3, High impedance millivoltmeter; 4 , the test sample (solid, liquid, or gas); 5 , Water pre-wetted separator contacting the wire cellIll and the test media.

35 1 of the 0.5 s DC pulse specific OCV. for

the

relative

electrodes to any

could be considered as an

resistivity

the signal, mV, after s

in the wire cell

e.g. 100 mV

of

the

medium

tested.

index Also,

time, e.g., after 5 or

specific

10

from the application of the DC pulse is to be considered

as an index to the pollutant level in the test samples. The higher

the

rate

of

the

OCV

decay,

the

higher

the

resistance/impedance of the tested material. Figure 2 shows a typical response of the various materials e.g., relatively highly conductors or highly insulators . There is a significant difference in the monitored signal.

,

mV. levels. MONITORING ENVIRONMENTAL POLLUTION

The of

pulse response monitoring technique i s interest in monitoring the contaminant level

proposed

potential

DC

of the tesed media. This NDT technique is an economic test method a s it saves time and effort done in-situ. The sensing probe pairs

the

(

should

be

wire

fixed

1s

cell,[l)).

so

the

. A simplified test i s

a simple pure copper write

The

contaminant

experimental

conditions

concentration

will

be

the only test parameter. The electrode separation is fixed.The the cross section of the wire electrodes i s also constant. Fresh metal

surface

is used

in every experiment by simply

cutting the end of the wire pairs by a sharpe shear.

in the State of Kuwait is affected by the o i l field mass burns in Feb. 1991 and its extended period Almost

till

everything

Nov.1991. Huge

ammount

of

pollutants

were

generated

and diffused in everywhere depending on the various geographic and other conditions. Air , water and soil were severely polluted with the emmited pollutants. Almost

all

living creatures were affected

by atmospheric

pollution by the oil burn products. Also, the oil slick and contaminated

seawater

in the Gulph affected

large areas in

352

+

-0

I/' / TIME, seconds, s Figure 2 Typical DC pulse response monitored signals ( Decay curves) after positive ,(a), or negative. (b) DC Pulses. 1, Relatively insulating materials; 2, Relatively conducting materials.

353 Kuwait and the Kingdom of Saudi Arabia. Several evidents existed for the environmental pollution transferthousands of killometers and were carried by the wind. Such pollutants have reached the Far East countries as was proven by the colered precipitation of grey snow in the Himalaya area. This action was attributed to the dark (smoke) clouds from the Kuwaiti burning oil fileds in 1991. The color of the leaves of various trees had changed their conventional green to darker blue black color due t o precipitation and absorption of the Kuwaiti oil products in the atmosphere.

field burn

The rain falling in kuwait during the oil field burn period (Feb.-Nov. 1991). On Oct.1991 it rained grey color water drops. After the rain fall the soil was loaded with oil layer of a thickness of about 2 cm at about 3 cm below the soil surface. The soil- oil layer is also wetted with water (composite oil-water-sand with little air voids).

Monitoring of plant contamination

The environmental disaster in Kuwait which resulted from the oil field burn showed clearly the relative survival of plants and trees in the polluted atmosphere of kuwait. The number of dead trees is related to the type of the tree and also to the soil condition and irrigation system which was almost nil1 during the Kuwait invasion period. Even the dead dry plant

leaves which are usually brittle were soft a s a result o f the absorbed oil contaminants due to atmospheric pol 1 ut ion. Figure 3 shows the change in the DC pulse response of plant leaves exposed to oil field burn products. The exposure time was 3 months in comparison with the non exposed control samples.

354

ocu, PiU

OCU,

l11U

n

2 \

\

200

\oo

0 01

,

L

s

1G

5

10

15

c 3

1 :PIEq 1’ ,? t’ciuia JEJ Before 2 minutes Treatment (8)

1-1 PIE After

, c

4

c 3

Figure 3 Monitoring contaminated plant leaves (palm-tree) exposed to Kuwaiti oil field burns during the period Feb.Nov. 1991. 1, samples exposed for 3 months (Feb.-May 1991); 2, control samples (from Jeddah area in the Kingdom of Saudi Arabia)

I t is clear that the plant leaves exposed to the oil field

burns are loaded with contaminants compared to the control (unexposed plant) leaf. This was comfirmed by a simple test using a commercial microwave device 181 exposure time was 2 minutes. The dry palm

-

tree leaves exposed to environmental

pollution by oil field burns contaminant showed severe weight loss compared to the conventional control (dry palm tree unexposed leaves). Figure 4 shows details. More details are ~

available elsewhere IS). RAINFALL POLLUTION ~ c i drain is known to result from the presence of acidic gases in air . The falling rain absorbs the acidic gases toform dilute acid solutions. This dilute solutions change to more concentrated acids on evaporation of their water content causing the degradation and corrosion of almost all materials in contact with such corrosive acid ( e . g . , sulfuric acid from sulfer dioxide gases). Toxicants still present a major danger to all living creatures. When it rained in Kuwait in Oct.l3th, 1991, the rain was loaded with dark grey colored oil contaminants from the polluted atmosphere due to oil field burns in Kuwait in 1991. The contaminant contents were as high a s 10,000 ppm. Even higher contaminant content samples of falling rain was about 2 . 5 % wt in some locations depending on the intensity of the atmospheric pollutants and the dust and fine sands in the air. The precipitated contaminants were mainly oil-water emulsion adherant to the fine dust particles present in the atmosphere. Figure 5 shows the variation of DC pulse response data for various rain samples during the period of Kuwaiti oil field burns (Feb. -Nov.1991). More details about polluted rain fall in Kuwait is available[lOl.

There exist a clear difference in the monitored signal and the contaminant level of the falling water during and

356

0.1 5

wt

LOSS

9

-

t r q 'mq

I

my'ctn

9

2

0.1

1 0.05

0

I

1

2

Figure 4 Effect of microwave treatment weight loss of contaminanted palm-tree to environmentally polluted atmosphere mass burn in 1991) in Kuwait 1. Kuwaiti contaminated palm-tree leave; 2, control samples

2 minutes) on t h e dry leaves exposed (due t o oil field

(2

351

after the Kuwaiti oil burn as shown in Figures 4 and 5.

MONITORING

SOIL

CONTAMINATION

WITH

KUWAITI

OIL

FIELD

BURN

PRODUCTS

The Kuwait soil was contaminated with oil in various ways 1. The

oil flooding from Lhe oil tanks during the invation

period due to the Iraq attack o f Kuwait. 2.

The

Iraqi strategic plans to

isolate the UN

forces by

oil buning area facing the UN forces. 3. The

oil leak from sorage tanks attacked by the various military actions for Kuwait loiberation.

4.

The

oil

flow

in

the

Gulph

just

before

the

Kuwaiti

liberation in Feb. 1991. 5. The atmospheric precipitation during the Kuwaiti oil burn

per i od . 6. The

rainfall since the Kuwait o i l

field burn disaste in

Feb. 1991.

This resulted in a major soil contamination with oil and oil

related

products.

The

chemical

and

electrochemical

characteristics of the soil is depending on the contaminant types and level. The response to DC pulse shows details for various

soil samples tested.

Figure 6 shows details about the monitor ng of contaminated soil with oil-related products

.

There exist a significant difference in the monitored signal

in mV (the response to DC pulse).

358

200

100

10

0

20

TIME, s Figure 5 Monitoring pollution levels of rain samples during the period of Kuwait oil field burns (Feb.-NoV.l991) 1, Dark oil contaminated rain in Oct.13, 1991; 2, Rain sample after the oil field burn period ( in Jan. 13, 1992); 3. Rain samples in March 17, 1992; 4 , Potable water samples ( a s a control)

359

oct

mU

300

200

100

0 0

10

20

TIME, s Figure 6 Monitoring contaminated s o i l by DC pulse technique 1 , Soil samples before the oil field burn (control); 2, Soil samples exposed t o atmosphere for 7 months during the oil field burns ( i n Oct.1991); 3, Kuwaiti Soil samples exposed t o rain in Oct.,1991.

360 MONITORING CONTAMINANTS

The

POLLUTION

Gulfwaters were

OF

SEAWATER

contaminated

WITH

with

KUWAITI

the

oil

when

OIL

the

Kwait invador flooded the seawater with oil from the storage tanks in Kuwait. The direct contamination of seawater with oil

from

the

Kuwaiti

oil

wells

was

the

main

source

of

contamination. Moreover, the continued precipitation during the Kuwaiti oil fields was a major source of Gulfwaters contaminantion. The discipated oil-bnased contaminants are expected to affect the chemical and electrochemical characteristics of the seawater. The marine life was endangered. Several sea birds were rescued. However, thousands of birds were killed mainly a s a result of the oil slick. Most of the fish species suffered from the floating , dispersed and sinking oil contaminants. Figure 7 shows the significant difference in the level of the response to DC pulse.

CONCLUSIONS AND RECOMMENDATIONS

The proposed electrochemical technique for mon toring the change in the chemical composition due to contam nation with oil related contaminants is proving its validity. This fast and economic electrochemical test method which involves the use of short DC pulses imposed into copper wire pair cellIl1. The applied DCV is about 9V for 0 . 5 s and the response is measured a s the decay in the open cell voltage (OCV) for about 15 s . The signal level(OCV, mV). The device proposed is a simple commercial high impedance millivoltmeter and simple recording system which is available almost everywhere at reasonable cost.

36 1

0 0 mU

30C

200

100

0 0

10

20

TIME, s Figure 7 Monitoring contaminated seawater level with the use of DC pulse technique 1 , Gulf seawater sample before the Kuwaiti oil f ield burn disaster; 2 , G u l f seawater contaminated with after- oil slick

362 Standard

reference materials should

be

used

to

generate

reference signal for comparison with experimental condition. The proposed DC pulse technique is simple , fast and present an economic method for monitoring contaminated media in any form. However, the identification of the individual polluting materials is still to be done by other chemical techniques such as the gas chromotography and other systems. The polluted media show significantly different mV signals (OCV after the DC pulse) from the non or less polluted media which meets the objectives of most of the environmental monitoring programs. The devices are simple and could be used in-situ with no previous training experience.

REFERENCES 1

Ismail MI, ed. Electrochemical

Reactors: Their Science

and Technology, Part A, El-Sevier Publ., Amesterdam.

1989,

504,(ISBN-0-444-87139-X).

2

Ismail

Kirchner

MI,

K,

Environmental

monitoring

by

transient in-situ technique, The 40th Int. SOC. Electrchem meet . , Kyoto, 1989. 43. 811. Ismail MI, Ahmed NA. Motairy H.. Al-Zafiry S . Org. Ind. Counc., 1989, 37, 112-127. 3

4

Ismail MI,

A

New

in-situ system

for monitoring

J.Gulf

phase

and property change in chemical engineering system. 3rd world Congress Chem Eng, Tokyo, Sep. 1986. Vol lV, 119-203. 446449.

Ref 1 P.344-388. 6 Ismail M I , ed, Simplified techniques; Applied research, graduate study and technology transfer, CRM Publ., Toronto 1989. ( ISBN-0-921478-18-6). 5

363 Ismail MI, Cooperative Education by Research and Referred Publications. CRm Publ, Canada, 1991, 20-24 (ISBN-0-9214787

39-91.

8

Sharp Carousel Convection Microwave Model.

Asiri A , Ismail MI, Role of Geographic factors i n the environmental pollution of Kuwait during the oil field burns,

9

CRM Publ.,( in Press) 10 Ismail MI. Young researchers and environmental programs, ibid (in press). 11 Ismail M I , techniques for

Kuwaiti

volunteers

Farguhar G J , Prasad T, Non-destructive monitoring soil/groundwater immiscible

contaminants, The Electrochem. SOC. Meeting

, Toronto, May

12-17, 1985, 8 5 - 1 , 510-511.

Ismail MI, Farquhar GI, Environmental control systyem 12 monitoring techniques f o r immiscible liquids spilled onto soil and contaminated groundwater, 34th Can. SOC. Chem. Eng. Conf.. Quebec, Sep. 30 - Oct. 3rd. 1984, 9 4 - 9 7 . 13.

Ismail

MI,

A

Novel

technique

for

immiscible contaminants in Soil and Electrochem. SOC. Meeting, Toronto, May 514.

monitoring groundwater. 12-17.

1985,

water The 513-

This Page Intentionally Left Blank

365

Bacterial-environmental interactions in corrosion on buried pipes F. Kajiyama Corrosion Science, Fundamental Technology Research Laboratory, Tokyo Gas Company, 1-16 Shibaura, Tokyo 105, Japan

INTRODUCTION Bacterial-environmental interactions, with reference to the cycles of sulfur and other elements, in corrosion on buried pipes are presented based on the recent field and laboratory studies by the author. Attention is focussed on the action of iron bacteria (IB), iron-oxidizing bacteria (IOB), sulfur-oxidizing bacteria (SOB), sulfate-reducing bacteria (SRB) and methane-producing bacteria (MPB). Correlation between the extent of microbiologically influenced corrosion damage and the activities of above mentioned bacteria is also discussed from electrochemical aspects.

EFFECTS OF BACTERIAL ECOLOGY ON UNDERGROUND CORROSION When we think of the bacterial-environmental interactions in corrosion on buried steel and cast iron pipes, we have in mind the different reductions and oxidations of inorganic sulfur compounds performed by living bacteria of the sulfur cycle. Depending on the bacteria and soil conditions which can be described appropriately as an extremely heterogeneous system, these transformations may be assimilatory or dissimilatory metabolic functions. Based on the recent field surveys and laboratory studies, the bacterial-environmental interactions, with reference to the cycles of sulfur and other elements, in corrosion on buried pipes are shown as Figure 1. This Figure demonstrates that microbiologically influenced corrosion (MIC) results from the activities of a microbial community.

Corrosion under anaerobic conditions Since the abnormally high corrosion rates in some natural anaerobic environments were traced to the presence of dissimilatory sulfate-reducing bacteria (SRB) by von Wolzogen Kiihr and van der Vlugt[ll in 1934, most studies of MIC have been concentrated on SRB as principal, or only, important microorganisms in MIC process. Therefore, corrosion of buried ferrous metals, influenced by the activities of SRB, is one of the best known of the economic activities. It is only recently that correlation between the extent of MIC damage and SRB levels (or their activities) has being suspected, and microorganisms other than SRB have become interest. In the complete anaerobic degradation of organic matter to COZ and methane, methane-producing bacteria (MPB) as well as SRB play a major role. This consideration shows that MPB should be considered to influence the activities of SRB.

Corrosion under aerobic conditions Less well known is the phenomenon of corrosion of metals by thiobacilli such as sulfur-oxidizing bacteria (SOB) and iron-oxidizing bacteria (IOB) which can proliferate in symbiosis with SRB and MPB particularly in acid sulfate soils such as Neogene strata and reclaimed lands with very low pH values. Essentially, the process requires an inor-

366

Asmblc

I

Ansmroblc

Figure 1. Bacterial-environmental interactions, with reference to the cycles of sulfur and other elements, in corrosion on buried pipes. (-Bacterial routes) SRB ; Sulfate-reducing bacteria MPB : Methane-producing bacteria APB : Acid-producing bacteria SOB : Sulfur-oxidizing bacteria IOB : Iron-oxidizing bacteria IB : Iron bacteria ganic energy source of sulfur or sulfide from pyrites or from sulfides such as FeS and H2S. SOB oxidize elemental sulfur or sulfur compounds (S2-, S20z-, SO$- etc.). IOB oxidize ferrous ions or inorganic sulfur compounds. The S source supports the autotrophic growth of these two bacteria with concomitant formation of sulfuric acid and it is this acid that is the highly corrosive agent, causing leakage and corrosion of the pipe metals. Recent extensive field and laboratory studies by Kasahara and Kajiyama [2], by means of quantification theory analysis and electrochemical techniques, have revealed that sulfuric acid produced by SOB and IOB in acid sulfate soils is mostly responsible for accelerated corrosion. Under aerobic conditions with neutral pH, the formation of iron tubercles due to the activities of the iron bacteria (IB), is often found. IB are obligatory aerobic, heterotrophic organisms. They are able to oxidize ferrous ions to ferric ions with the consequent precipitation and accumulation of ferric hydroxides on the surface of the pipe metal. Colonization by IB always encourages corrosion due to the formation of differential aeration cell. The small area beneath the tubercles becomes the anaerobic site (anode site) which sustains severe localized corrosion. Under aerobic environments containing active IB and a significant amount of bicarbonate ions, localized corrosion on buried ductile cast iron pipes was investigated by Kasahara, Kajiyama and Okamura [3] placing the main emphasis on the morphology of corrosion attack and the action

367

Figure 2. Cross section of a 155 mm-inside diameter ductile cast iron pipe buried for 567 Ms. of microorganisms. Figure 2 shows the cross section of a 155 mm-inside diameter ductile cast iron pipe buried in sandy soil for 567 Ms. The extensive iron tubercles were formed on the surface of the pipe. The form of corrosion was classified as graphitic corrosion. Microbial enumeration and electrochemical observations strongly suggested a possibility that a symbiotic proliferation of IB and IOB (SOB) was primarily responsible for accelerated localized selective leaching as well as tubercle formation. Figure 3 shows the proposed mechanism for the graphitic corrosion of buried ductlie cast iron pipes for this case [31.

EXPERIMENTAL PROCEDURES OF LABORATORY STUDY ON LOCAL ACTION CELL CORROSION Effects of microbiological activities on the corrosion of ductile cast iron in the soil were evaluated in the laboratory by burying a probe in a 1 dm3 of Pyrex glass soil cell. The probe used in the study was a triangular type in which a ductile cast iron plate 30 mm long x 20 mm wide x 8 mm thick and a platinum plate 30 mm long x 20 mm wide were used as working and auxiliary electrodes, respectively, and a Cu/CuS04 electrode was used as a reference electrode. To simplify the corrosion system, an exposed surface of 6 cm' was obtained for working and auxiliary electrodes by applying waterproof paint leaving two surfaces opposite to each other. The exposed surface on a ductile cast iron plate was slightly abraded with 210 emery paper followed by degreasing with acetone. Soils used in the study were collected from Neogene strata and reclaimed lands in and around Tokyo in Japan. Each cell was held at 303 K in an incubator for a period of 7,78 Ms. During testing period, corrosion potential (E,,,,), AC impedance and polarization measurements were taken at a regular interval. AC impedance measurements were taken potentiostatically by a combined use of a potentiostat and a frequency response analyzer. The applied potential was controlled at 10 mV rms. AC impedance was measured from 10 kHz down to 1 mHz, and impedance data were plotted on a Cole-Cole diagram, wherefrom a charge transfer resistance, RCt, could be read off. Potentiostatic polarization measurements from Ecorr to Ecorr +500 mV were taken by changing the

368

Figure 3. Proposed mechanism for the graphitic corrosion of buried cast iron pipes under aerobic environments containing active IB and a significant amount of bicarbonate ions (IB: Iron bacteria, SOB: Sulfur-oxidizing bacteria) [31. applied potential alternately towards anodic and cathodic directions at a step of 2 mV by compensating IR drop. Soils used in the study were analyzed with respect to Eh, pH, FeS, SRB, MPB, IOB, SOB and IB before and after the 7,78 Ms of burial testing. Samples for such analyses were collected from the immediate vicinity of respective working electrodes. Upon conclusion of testing, every probe was removed from each soil box, followed by cleaning and weight loss determination. The weight loss was then converted into corrosion rate by applying Faraday’s law.

RESULTS AND DISCUSSION OF LABORATORY STUDY ON LOCAL ACTION CELL CORROSION Corrosion rate Figure 4 shows the results of MIC testing reproduced in the laboratory, wherein corrosion rate estimated from AC impedance measurements by applying Eq. (1) [41 was correlated to the corrosion rate converted from actual weight loss determination using Faraday’s law. d(m/Ms) = 4,12 . 1 0 - 7 / ( m ) .................. . . . (1)

369

B I B and I O B ( S 0 B )

/ I

10

'

actiw

0 I O B active 0 S R B active x IOB sterile I

1

I

10

10

10

Corrosion rate from weight loss, m/Ms

Figure 4. Relationship between corrosion rate predicted from AC impedance measurements and that determined from weight loss measurements.

is a charge transfer resistance averaged over extended periods (ohm. In Eq. (11, m'). From Figure 4 , the following observations may be drawn: 1) Particularly noteworthy is a marked deceleration in the corrosion rate in a sterilized soil as 1,87x l o p 5 m/Ms of corrosion rate was reduced to 1x m/Ms, which clearly indicating the contribution of IOB and/or SOB to accelerated corrosion. 2) Under aerobic conditions containing active IB and a significant amount of bicabonate ions, the formation of iron tubercles was found. The small area beneath the tubercles sustained severe localized corrosion. 3) A group of corrosion rate data for high SRB activity remained at levels more than one order of magnitude lower than that for high IOB activity, rejecting the relation of SRB to accelerated corrosion in soils. 4) Whole data fell closely on a one-to-one line irrespective of the differences in species and activities of bacteria, indicating the validity of the AC impedance technique in predicting rates of MIC in soils.

Electrochemical Aspects of the Action of SRB Table 1 shows the result of environments analysis for the testing cell wherein the activities of SRB remained at a high level in a clayey soil throughout the testing period as evidenced by an increase in SRB count, FeS and pH. Figures 5 to 7 show the results of electrochemical measurements. Impedance data indicated that charge transfer controlled process, characterized by capacitive semicircles, was predominant throughout the testing period. Increasing diameter of semi-

370

Table 1 Result of environments analysis under active SRB condition Time' Ms

Location

0

Bulk

77

E1ectrode/soil 78 interface

EhIVvs. FeS/ NHE pH wt'ppm

Number of bacteria/kg-l

SRB

MPB

IOB

SOB

IB

0,591 5,91

116

2x106 1x106

0

0

0

0,254

253

1 ~ 1 0 1' ~ 1 0 ' ~0

0

0

7,14

circles and decreasing soil resistivity with time indicated that low resistivity of FeS layers, that was formed by active SRB thrived in the ductile cast irodsoil interface, play a significant role in reducing corrosion rate to a level of 3,2X m/Ms, that is, more than one order of magnitude lower than that in high IOB activities. As reported by Kiihr et al. [ll, cathodic depolarization was observed also in the present study. However, as evident in Figure 7, anodic polarization was concurrently activated, presumably as a result of formation of stable FeS layers. Such concurrent anodic polarization and cathodic depolarization must be the reason why corrosion rates remained at a low levels in those environments where SRB were highly active. Therefore, in anaerobic environments with active SRB, micro-cell corrosion rates are very low on condition that the continuous FeS layers remain stable.

d

131 2

. d

14

0.505

VI

69Ms

179Ms

Surlsc. area 0cm' 1.0-

0,60?

f

rl

b .-m ri" 0.70

20

0 MZ

a .-C

,,----*+-

__-_---

a ---_____.__. .-______----____ g

,/'

0.80

0 I 0

0.6

I

I

1.0

1.5

Re[Z]/kQ

Figure 5. Time course of l/Rct and E,,,, under active SRB condition.

Figure 6. Sequential complex plane impedance plots under active SRB condition.

37 1

-Initial

----i.ims

-0.4

0

Current density, I I A - m - 2

Figure 7. Polarization curves under active SRB condition.

Electrochemical Aspects of the Action of IOB Table 2 shows the result of environments analysis for the testing cell wherein the m/Ms was reproduced in the laboratory severest general corrosion rate as 1,87x in a sulfate enriched silty soil of 1400 wt.ppm of SO:-. Appreciably low pH and decrease of FeS during test period would lend a strong support to an interpretation that sulfuric acid generated as a result of proliferation of IOB and SOB was responsible for such an observed accelerated corrosion rate. Table 2 Result of environments analysis under active IOB condition

FeSI

EhIV vs.

Time’ Ms

Location

0

Bulk

71

78 interface

NHE 0.655

Number of bacteria/kg-’

2,82

1110

0,409 3,36

549

IOB

SOB

IB

7x104 lx104 8x10’ 2x10’

0

4x104 lx104 3x10’

0

pH wt*ppm SRB

MPB

0

3 72

Figures 8 to10 show changes with time of 1/Rct and Ecorr, AC impedance and polarization curves were obtained for the same testing cell as Table 2. Although corrosion ~ from the start of testing rates remained at levels as high as 2,16 - 1 , 5 5 ~ 1 0 -m/Ms till 2,59 Ms, they began to decrease between 2,59 and 3,46, eventually attaining to levels around 3 , 2 ~ 1 0 -m/Ms, ~ which are still about one order of magnitude higher than the average general corrosion rate in the soils in the Tokyo area. In response to such decrease of corrosion rate, an inductive semicircles on sequential complex plane impedance diagram, characteristics of the adsorption and desorption of sulfate ion onto the electrode surface, was replaced by two semicircles after day 3,46 Ms, indicative of the overlap of two kinds of electrical double layers. Such AC impedance behavior together with polarization behavior was thought to indicate a change in the corrosion process, that is, accelerated dissolution in IOB and SOB active environment became being disturbed by anodic and cathodic polarization that could be accompanied by the stifle cathodic reaction due to the action of IOB and SOB.

Electrochemical Aspects of the action of IB Table 3 shows t h e result of environments analysis for the testing cell wherein the average general corrosion rate as 1,08x m/Ms was obtained in the laboratory in sandy soil. Appreciably low pH at the electrode/soil interface after test could be attributed to the proliferation of SOB in symbiosis with IB. Figures 11 to 13 show changes with time of 1/Rct and Ecorr, AC impedance and polarization curves obtained for the same testing cell as Table 3. The continuous decrease of 1/Rct, together with continuous shift of corrosion potential towards noble direction, was thought to indicate that the electrode surface in sandy soil was covered with protective oxide layers. Appearance of the specimen showing extensive tuberculation lends a strong support to this electrochemical behavior. In this case, only capacitive semicircles were obtained. Significantly depressed semicircles were obtained with time which were thought to be due to the repartition of the time constants of the processes. The repartition of the time

8or-----

60

,

Figure 8. Time course of l/Rct and E,,,, under active IOB condition.

I

I

I

Figure 9. Sequential complex plane impedance plots under active IOB condition.

373

Initial

7.7BMs

-1.2

h ' 102

10

2

10

1

10'

10

Current density. / / A - r n

Figure 10. Polarization curves under active IOB condition. constant is generally explained either by the surface roughness or by non-uniform of the current distribution on the surface leading to localized type of corrosion. The decrease of resistivity at electrode/soil interface with time was responsible to the action of SOB producing sulfuric acid. Polarization behavior was thought to be agreement with that observed in the AC impedance analysis. Table 3 Result of environments analysis under active IB condition Location

0

Bulk

0,519

8,22

Electrodelsoil

o,559

4,06

79

NHE

78 interface

Number of bacteria / kg-

FeS/

EhIV vs.

Time/ Ms

SRB

MPB

IOB

SOB

5

2x10'

1x10'

0

4x10'

+E

8

7 ~ 1 02x107 ~

0

5x10'

-+H

pH wt'ppm

IB

314 80,

I

I

2

4

I

I

6

0

t

"'"I

1

I

I

60

,

,

Duration of burial testing. t / M s

Figure 11. Time course of l/Rct and E,,,, under active IB condition.

-0.2

c

- 1

I

I

Re[Z]/kO

Figure 12. Sequential complex plane impedance plots under active IB condition.

,

Current density, 1 I A . m ~ '

Figure 13. Polarization curves under active IB condition.

CONCLUSIONS Bacterial-environmental interactions, with reference to the cycles of sulfur and other elements, in corrosion on buried pipes were studied. The results obtained are summarized as follows: 1) Severe corrosion on ductile cast iron was observed both in the field and laboratory being attributable to acid sulfate soils containing appreciable concentration of sulfuric

375

acid formed as a result of IOB and SOB proliferation. 2) Under aerobic environments containing active IB and a significant amount of bicarbonate ions, localized corrosion on buried ductile cast iron was observed. 3) Cathodic depolarization was observed in high SRB activity; which did not always result in an accelerated corrosion because of concurrent activation of anodic polarization. 4) AC impedance technique was indicated to provide useful information in MIC testing concerning the corrosion processes as well as the rates of corrosion.

REFERENCES 1 C.A.H. von Wolzogen Kiihr en L.S. van der Vlugt, Water 1934; 18: 147-165. 2 Kasahara K. and Kajiyama F., Microbially influenced corrosion and biodeterioration NACE 1991; 2-33 - 2-39. 3 Kasahara K., Kajiyama F. and Okamura K., Zairyo-to-kankyo 1991; 40: 806-813 (in Japanese). 4 Kasahara K. and Kajiyama F., Corrosion 1983; 39: 475-480.

This Page Intentionally Left Blank

377

Influence of ohmic drop on the determination of electrochemical parameters Gabriele Rocchini Italian Electricity Board - Thermal and Nuclear Research Center, Via Rubattino 54, 20134 Milan, Italy

INTRODUCTION When operating in aggressive environments, if appropriate active or passive protection techniques have not been used, metallic materials may be subject to progressive decay of their mechanical properties and to a reduction of structural functionality as a result of uniform, localized or selective, corrosion attacks. The thermodynamic instability of technological materials, whose entity depends on the chemical and physical characteristics of the operating environment, stimulates corrosionists, designers and users to get a deeper knowledge of the corrosive behaviour. The main aim of these efforts concerns the adoption of the most effective remedies ensuring integrity and a high degree of reliability. A characteristic aspect of this problem is the circumstance that the loss of the performance of materials, which have the greatest interest for practical applications, is due to corrosion processes that take place through an electrochemical path. This feature makes it possible to evaluate the behaviour of a metal both in the laboratory [I], simulating the environment, and in the field [2], by means of appropriate electrochemical techniques. The difficulties inherent in a faithful reproduction of the actual operating conditions and the impossibility, in some fields, of obtaining objective laboratory data have led researchers to develop some specific themes aiming at the monitoring of the corrosive state [3] of plant components in real time [4]. The use of electrochenlical techniques for studying the mechanisms of a given process, evaluating the mass loss of a metal quantitatively or determining the degree of aggressiveness of the environment poses the problem of the significance and reliability (21 of the information obtained. But the existence of some limitations of the linear polarization technique [5] in predicting the correct trend of the corrosion rate is not a novelty, for some discrepancies between direct and electrochemical mass loss determinations had been pointed out by Indig and Groot [6]. In order to give a satisfactory answer to these questions the first step to be taken is to eliminate the interferences due to the operating environment, which has also the important task of supporting the current flow necessary to polarize the electrochemical system. The role of the environment is twofold; on the one hand it helps the kinetics of a corrosion process by providing the electroactive species, on the other hand, it attenuates its intensity by hindering the motion of the ions that transport the current necessary for the overall process between the anodic and cathodic microcells. Thus the electric resistance of the medium [7] can control the kinetics of the anodic and cathodic reactions directly. This concept is clearly set out in reference [8] that

378

examines the behaviour of the aluminium alloy 3004 in a solution having a resistivity of 2.1 x lo4 ncm, a value typical of many industrial and waste waters [9]. A very wide field, where the electric resistivity of the medium plays a decisive role on the possibility of studying the behaviour of a system accurately by means of electrochemical techniques, is that of organic environments. To examine the behaviour of ARMCO iron and some low alloy steels in methanol solutions Mazza et al. [lo, 111 were compelled to add 0.1 M of lithium perchlorate to them. This device, which can be adopted because the adsorption of LiC104 on the surface of the specimen may be disregarded [12], enables one to avoid the use of too high anodic potentials, which might cause oxidation of the methanol [13]. A further example, which confirms the necessity of evaluating the resistivity of the medium very carefully, concerns the corrosion of rebars in reinforced concrete. In this case the intensity of the current flowing between the anodic and cathodic zones of a macrocell depends on the resistivity of the concrete and the extent of the region involved. To determine the concrete resistivity various methods have been developed, which can be applied in the laboratory [14] as well as in the field [15]. It should be noted, however, that in the latter case most researchers have pursued the approach suggested by Wenner [16] for the evaluation of the resistivity of soils. The contribution of the ohmic drop to the electrode overvoltage cannot be neglected when the values of the corrosion rate of the rebars are appreciable, even if the current intensity is small within a given polarization potential interval, because under such conditions the interpretation of experimental results could be completely distorted. This observation concerns particularly polarization resistance which, according to Stern and Geary's theory [5], is inversely proportional to the corrosion rate. In fact, the contribution of the environment resistance might increase the apparent value of the polarization resistance considerably and mislead the operator as to the evaluation of the true corrosion state of rebars. In one of his studies [17] the present author has shown that, with reference to the geometry of the concrete specimen immersed in artificial sea-water, the contribution of the medium resistance to the polarization resistance, though slightly dependent on the position of the rebars, was of the order of 5 . 2 ~ 1 0Rcm2 ~ and remained practically constant in the course of the test. One year later, the value of polarization resistance of the rebar that exhibited the least satisfactory behaviour was 7x lo4 R cm2, which experimentally confirms what has been stated above. Lastly, it may be useful to recall that the resistivity has an important role in the field of cathodic protection by impressed current because, especially in soils [MI, it helps to determine the real polarization potential of the structure considered. In all of these cases the attainment of the optimum degree of protection depends on the a priori knowledge of the medium resistance between the reference electrodes and the structure to be protected. Of course, these observations apply also in the case of media having a high electric conductivity and characterized by an aggressiveness that may impair the integrity of the metallic materials that are in contact with them for a long time. Under such circumstances, in order to attain a good level of protection it is necessary to supply highintensity currents so that the contribution of the ohmic drop to the electrode potential will not be negligible. In the case of highly resistive media, when cathodic protection is based on the use

379

of sacrificial anodes, their positioning relative to the structure must be correctly defined so that they can fulfil their function effectively, supplying the current that is needed to prevent corrosive attacks. It has been observed, in fact, that, because of the modest amplitude of the available electromotive force, sacrificial anodes are generally unable to supply high-intensity currents.

GENERAL CONSIDERATIONS In order to polarize an electrochemical system by an external signal three electrodes must be used and the electric contact of the tip of the reference electrode, which cannot be made too close the interphase of the working electrode, takes place in the solution bulk. Thus the presence of the ohmic drop is a consequence of that and of the technique adopted in assembling the working electrode and the electric connections outside the electrolytic cell. Figure 1 shows, schematically, the electrolytic cell that is used in the author's present laboratory for measurements up to 65 "C and the assembly of the working electrode. The reference electrode is immersed directly in the electrolytic solution and its porous plug is quite far from the surface of the working electrode. This arrangement was adopted because the electrochemical cell is used also for alternating current measurements over the frequency interval [0.01, 2x104] Hz. The adoption of the Luggin capillary causes undesired phase-shifts and an attenuation in the amplitude of the external signal [19] in the high-frequency region. The assembling technique of the working electrode, which is screwed on to a threaded stainless steel stem, clearly shows the existence of the problem of contact resistances and electric connections. The use of the four-wire technique permits complete elimination of all ohmic contributions from the conductors outside the electrolytic cell but does not interfere with the contact resistance related to the assembly of the specimen, which depends also on the cleaning degree of the two surfaces. This is quite important because, if the two surfaces that are in contact are oxidized, the value of the resistance may be greater than 1 0. Under normal conditions, however, the values of the contact resistance between the working electrode and the stainless steel stem are negligible, so that it is correct to consider only the resistance of the electrolytic solution. Generally, the ohmic drop contribution is exclusively determined by the geometry of the cell and the conductivity of the solution under consideration. Yet, our experience concerning the performing of polarization curves in acid solutions has shown that the resistance may be influenced by mass transport and that the extent of the perturbation becomes significant when the intensity of the external current is sufficiently high. This observation can be easily explained by examining the general expression of the current density. If, for convenience, we consider the ideal case of a planar cell and assume the ion H 3 0 + to have a concentration c, the value of the current density J in the electric field of intensity X is given by dc

J = epX - eDdx

,

380 n

w)

WmkmQ Ebclrode R ) Relerence Electrode C ) Counter Eleclrode 1) Thwmmwler

2) Gas hlei 3) Thermostatlc Liquia lnlel 4) ThermOstaliC Ltqwa Outlei

5) Gas Oullel 6 ) lnoi Eleclric Conlacl

a) Tefbn b) Viton Gasket c) lnoa Confact Screw d) Test Soecmen

Figure 1. Schematic diagram of ENEL-CRTN mod. EC02 double-wad cell with a total capacity of 1.5 1. where e, p and D are the charge, the mobility and the diffusion coefficient of the species considered, respectively. The relation (l),which in the case under discussion yields the predominant contribution to charge transport, shows that current density increases when the diffusion process is activated. Hence the increase in current density causes a decrease in the value of the apparent resistance of the acid solution. Therefore, the existence of an electric resistance having a value different from zero between the working and the reference electrodes implies that the experimental currentvoltage characteristic of an electrochemical system may sensibly differ from its ideal behaviour, which is determined in accordance with the theory of the transition state [20]. This is because the absolute value of the actual polarization governing the electrode process is lower than that of the polarization imposed from the outside.

38 I

This problem is particularly important when the current-voltage characteristic i ( A E ) of the electrochemical system is described by the ideal law

i = I,[exp(aAE)

-

exp(-pAE)]

,

(2)

where I, is the corrosion current density, while A E is the potential difference referred to the mixed potential Em.The quantities a and ,B are related to the anodic and cathodic Tafel slopes B, and B, by the relations 1

= - In 10,

1

B = -In 10 .

(3) Ba B, The law (2) represents the behaviour of a system containing only two elementary processes controlled by the activation energy [21]. Its importance, however, is connected with the experimental observation that it satisfactorily describes the kinetics of a very large class of electrochemical systems. This is particularly true of metals and alloys of technological interest that are subject to uniform corrosion in a great number of aggressive environments. The specific interest in the law (2) is due to the possibility of dealing analytically with the problem of the influence that the solution resistance exerts on the kinetics of a corrosion process. Experience has shown, in fact, that the direct determination of the mass loss of a metal in a given environment differs from the one obtained by electrochemical measurements. It should be noted, however, that the origin of this discrepancy is of a more general nature and is not only ascribable to the ohmic drop. On the other hand, the expression of the law ( 2 ) , where the corrosion current density I, appears as a factor in the right-hand side member, allows the extent of the anodic process to be correctly evaluated. Generally, the difference in form between ideal and real responses, which is exclusively due to the experimental procedures adopted for performing the polarization curve, affects the determination of the values of the corrosion current density and the Tafel slopes. The practical impossibility of a well-defined determination of these quantities, which, with reference to the law (a), characterize the behaviour of the system under examination, may result in an unsatisfactory formulation of the reaction mechanisms and an incorrect evaluation of the corrosion rate. When the corrosion current density values are high or the exposed surface area of the working electrode is large the effect of the electric resistance of the electrolytic solution is not negligible. Yet it is not easy to find in the literature specific studies dealing with the problem of the reliability of the information contained in experimental results and of the indetermination of electrochemical parameters as a consequence of the practical impossibility of evaluating the ohmic drop correctly. (Y

MONITORING BY A TWO-ELECTRODE TECHNIQUE An interesting way of illustrating the need to eliminate the contribution of the ohmic drop to electrode potential is offered by the discussion of corrosion rate monitoring by the two-electrode technique introduced by Marsh [22].

382 The method is schematically illustrated in figure 2. The operational amplifier A, working in the follower configuration, is used to apply, between the points D and T, the potential difference V present at the terminals of the generator G, which is assumed to be positive. This hypothesis simplifies the analytic expression of the potential difference V , because its polarity determines the behaviour of the two electrodes. The differentialinput voltmeter Q determines the intensity of the current that, flowing through the electrochemical cell C from D to T, polarizes the electrodes Wl and W2. These electrodes, made of the same material, are identical. Their surface areas are equal to S and, conventionally, they are polarized anodically and cathodically.

Figure 2. Schematic diagram of a two-electrode monitoring system. Legend: G voltage generator, A follower, Q differential voltmeter, and C electrochemical cell. Based on experimental conditions, assuming the electrodes Wl and W2 to have the same free corrosion potential and denoting by R8 the solution resistance between them, we can set the relation

+

+

V = AEa(i) AEc(-i) R,iS

,

(4)

where the functions AE,(i) and AEc(-Z), which take only positive values, give the anodic and cathodic overvoltages, respectively, while i is the current density. Following Marsh’s indications [22], the value of V is generally assumed to be equal to 20 mV, so that the approximation of the linear response can be considered valid. Under such a hypothesis and assuming the electrochemical behaviour of the two electrodes to be identical, the expression (4)becomes = [ I d a2+ P )

+ BSS] i .

(5)

Setting R = VJi,from the equation ( 5 ) the corrosion current density is given by the relation n

383 which clearly shows that omission of the term due to the ohmic drop in the relation (4) results in an underestimate of the corrosion rate. It is important to note that this term can be neglected only when the inequality R >> R,S holds. In fact, setting Q = R,S/R and using the series expansion of the function 1/(1 - Q ) the expression (6) takes the form

which confirms the validity of the foregoing remark. The purpose of this simplified discussion was to define, through an analytic representation, the dependence of I, on the quantity R, and, consequently, on the environment. The general problem, however, is more complex because the free corrosion potentials of the electrodes Wl and Wz may be different. In support of this assertion mention can be made of some results concerning the behaviour of the SA 213 grade T9 low alloy steel in a solution containing 100 g/l of EDTA at pH 9 and 100 "C [23]. In the course of the dynamic test using the three solution flow rates 0.5 ms-', 1.0 ms-' and 1.5 ms-' the absolute values of the free corrosion potential of the samples examined, referred to electrodes of the same material, were generally less than 5 mV. However, shifts of 46 mV, 29 mV and 14 mV were observed. This observation is more general than it might appear at a first glance, because a similar situation was noticed in the case of the SA 106 grade B carbon steel and the SA 213 grade T11 low alloy steel under similar test conditions. In these two cases, however, the shift of the free corrosion potential from zero was more moderate (about 5 mV). Finally, in the course of some experiments related to the monitoring of the corrosion state of aluminium brass tubes in a steam condenser [24] cooled with sea-water conditioned with ferrous chloride, the potential difference between two similar electrodes of commercial probes was found, in some cases, to be greater than 100 mV. Yet, the absolute value of the potential difference between two aluminium brass electrodes having a larger surface area and used on our experimental probes never exceeded 15 mV. The formula (4) can be generalized by schematizing the real electrochemical cell into an ideal cell, in which the electrodes have the same free corrosion potential, and a voltage generator G that takes into account their different behaviour, as shown in figure 3.

Figure 3. Schematization of a real electrochemical cell. Legend: G voltage generator, and C electrochemical cell. Designating by

Vi the

potential difference at the terminals of G, the formula (4)

384

assumes the more general expression

V = AE,(Z)

+ A&(-Z) + R,Si + VI .

(4')

This relation is of considerable practical importance because it justifies the fact that the response of the electrochemical system varies as the polarity of the applied voltage changes. Admitting that the two electrodes present different corrosion rates, 1,1 and Icz, and applying the linear approximation of the formula (4'), we obtain the expression

where the quantity Re and the apparent corrosion density I,, are equal to ( V - &)/i and IclIc~(Icl Icz)-', respectively. Examination of this formula shows that the determination of two distinct values for Ica does not mean that the corrosion process is of a localized type [25]. The two determinations of I,, can simply be explained by examining the behaviour of the voltage generator G, which in0uences the current 00w between the points D and T.

+

IMPORTANCE OF THE ENVIRONMENT In most cases, the corrosion rate of a metal in contact with an aggressive environment is monitored using direct current and specific equipments [26, 27, 281 whose operation is based on the polarization resistance method [5]. Yet, although this method is widely used in the laboratory as well as for applications of industrial interest, it is impossible to find in specialized literature an exhaustive study on the validity and reliability of its application as a function of the geometry of the three-electrode probe and the characteristics of the medium. This concept is of primary importance when on-line corrosion rate monitoring is applied to materials used for the manufacture of components of industrial plants. In this case the ultimate objective is to obtain data that will ensure the optimum operation of the equipment, reducing unavailability due to corrosion. In fact, the aim of monitoring the evolution of a corrosion process is to attain an objective evaluation of the extent of the phenomenon. On the other hand, the use of an electrochemical technique has clear advantages: it provides information in real time and, where suitable probes are available, it can be employed easily in field. The most serious problem related to the practical application of the polarization resistance technique concerns the correct use of commercial equipments, which can be conventional-type or computerized. Conventional-type equipment is calibrated by the manufacturer who, in most cases, makes use of literature data and bases his choice on the premise that electrochemical measurements can give absolute indications. The goodness of the reading depends on the accuracy of the calibration of the instrument which, in principle, should be based on tests performed under controlled conditions. In actual fact, as can be seen also from Mansfeld's monograph [26], the manufacturer sets the instrument for some specific

385 applications, taking no account of the characteristics of the experimental technique that has been adopted for performing electrochemical measurements. Apart from the validity of the theory of the linear response, the main purpose of calibration should be to reduce the influence of the medium, the specific test conditions and, possibly, the ohmic drop on the determination of the mass loss. This is possible because the calibrating operation consists in establishing an empirical relationship between the mean corrosion rate values and the electrochemical quantity i / A E . This concept is clearly illustrated by the results of figure 4,concerning the behaviour of the two SA 106 grade B and SA 213 grade T11 carbon steels in a 5% by weight HCl solution, inhibited with 2 g/kg of the commercial product Rodine 213, at 75 "C [as]. The mean corrosion rate is expressed in mgcm-2h-1, while the indications of the Ate1 mod. AP 102 corrosion meter were converted into the same physical unit, assuming the density of the two steels to be equal to 7.8 g ~ r n - ~The . mean value of the corrosion rate, determined by electrochemical measurements, was obtained using the trapezia method for calculating the integral.

corrosion rate [ m g

05

SA 106 Gr

h-l]

1

B

SA 213 Gr T l l

04

0.3 0.2 0.1

00

0.5

1.0

15

0.5

1.0

15

Solution flow rate Crns-11 Corrosion meter ( C M )

69 Mean weight loss (MWL) of 0 Mean weight loss (WWL) of

5 electrodes working electrode

Figure 4. Comparison of different corrosion rate determinations of the two carbon steels SA 106 grade B and SA 213 grade T11 in an HC1 5% inhibited solution at 75 "C. Figure 4, which compares three corrosion rate determinations related to the electrochemical measurement (CM), the mass losses of the 5 electrodes of the probe (MWL) and the mass loss of the working electrode (WWL) as a function of the flow rate of the solution, testifies to the goodness of the new calibration of the corrosion meter. After this operation, the instrument can be used for monitoring the corrosion of carbon steels in inhibited HC1 solutions quantitatively. Of course, in principle, the calibration of the corrosion meter is valid only for the

386

electrochemical cell depicted in figure 5 because its geometric configuration determines both the hydrodynamic characteristics of the solution and the value of the resistance between the working and reference electrodes. In fact, it has been found experimentally [30] that the corrosion rate may depend on the geometry of the electrodes and on their position relative to the direction of the flow. The electrochemical cell in figure 5 contains 5 cylindrical electrodes, made of the same material and having an internal surface area of 60 cm’. This makes it possible to evaluate the scattering of the data and check the reliability of the extrapolation of an electrochemical measurement.

Figure 5. View of the electrochemical probe. Legend: W working electrode, C counter electrode, R reference electrode, A alternative electrode, and R1 true reference electrode. It should be noted, however, that the calibration of the Ate1 mod. AP 102 corrosion meter is of a global type because the instantaneous corrosion rate is generally dependent on time. The importance of the influence of the environment on the calibration of a corrosion meter has been demonstrated by a study of the behaviour of the two steels mentioned above in a mixture containing formic acid and glycolic acid, both at a concentration of 1.5% by weight, and inhibited with 2 g/kg of the commercial product Rodine 31A. In this case the two steels exhibited distinct electrochemical behaviours, so that the calibrating operation was rather complex and revealed some perplexities. The use of computerized systems for corrosion rate monitoring is more flexible because the proportionality constant [31] can be changed by software and is selected by the

387 operator on the basis of his own experience. In this way it is possible to reproduce the global trend of the corrosion kinetics quite faithfully, without the need for preliminary tests but using the results obtained during the monitoring operation. This enables one to optimize the use of the electrochemical technique and obtain a reliable representation of the evolution of the real process.

NUMERICAL METHODS

A chapter of electrochemical corrosion that has been developed only recently concerns the use of computers for measurements and data processing by means of mathematical models. The numerical methods [32, 33, 34, 35, 36, 371 developed for the analysis of experimental polarization curves described by the current-voltage characteristic (2) with the aim of determining the electrochemical parameters a , p and I,, are of considerable importance in the field of basic research as well as in corrosion rate monitoring. They permit, in fact, a more objective evaluation of these quantities with reference to a given potential difference interval A E , removing the degree of subjectivity that is inherent in the graphic determination of the Tafel slopes. The main characteristic of numerical methods is that, in principle, they do not require that a system be studied over such a wide A E interval that, if there are no specific problems with the conditioning of its geometric shape, its response will exhibit the Tafel behaviour. Among the factors that may determine the geometric shape of an experimental polarization curve and the region where the validity of the Tafel law begins to be evident, are the quantities a and p. The influence these parameters exert on the graphic shape of a polarization curve is discussed in some studies [37, 381. From an experimental point of view, particularly notable is the possibility of determining the electrochemical parameters without exciting the system with signals of such amplitude that its free corrosion state will be irreversibly perturbed. The usefulness of this procedure is apparent if one considers the problem of comparing different corrosion rate determinations. In this connection, reference might be made to a study [39] dealing with the monitoring of corrosion of ARMCO iron in an H z S 0 4 1 N solution through the use of potentiodynamic curves and the determination of the concentration of Fe2+ ions by the spectrophotometric technique. Experience shows, however, that the results of the processing of experimental data generally depend on the width of the interval AE, which determines the content of the information. This concept can be immediately explained on an intuitive basis because in an electrochemical system some secondary reactions may occur whose contribution in the vicinity of the mixed potential is not negligible. Another element not to be overlooked are the oscillations of the mixed potential, which cause variations in the current intensity and are, therefore, of great importance for small values of AE. The existence of processes of negligible importance results in a departure from the ideal behaviour around the mixed potential and, consequently, in a distortion of the

388 experimental curve, whose degree depends on the quantity AE and tends towards zero at a sufficient distance from the mixed potential of the system. It is evident from the foregoing considerations that there exists an influence of the ohmic drop, which will be dealt with further on, on the determination of the electrochemical parameters and the correct application of the methods of numerical analysis. Moreover, experience has shown that the success of numerical analysis depends also on the way the contribution of the ohmic drop to electrode overvoltage is reduced. In this respect, it may be mentioned, for example, that in the case of iron and carbon steels serious difficulties are met with the analysis of polarization curves performed in uninhibited HC1 solutions at temperatures above 65 "C [40] because the corrosion current density assumes very high values. Our experience with regard to HC1 solutions and electrodes having an exposed surface area of about 10 cm2 has shown that the demand for high-intensity currents for polarizing the electrode to the desired values implies the presence of an appreciable ohmic drop contribution, even if the solution resistance is rather low. In the case of acid solutions, the reproducibility of the experimental points has been found to depend also on the exposed surface area of the specimen. As a consequence, the validity of the numerical method may fall short, i.e. the evaluation of the corrosion current density may give a value that is quite different from the true one. This assertion is supported by the data in Table 1 concerning the behaviour of ARMCO iron in 1 m HC1 solutions at various temperatures [40]. The direct evaluation of the corrosion current density, Id, was obtained from the determination of the concentration of ferrous ions, entered the solution, by the spectrophotometric technique. The galvanostatic pulse polarization curves were performed using the CORRCONTROL system [41] and the GALIMP program [42]. The corrosion current density, Ic, was computed using the NOLI method [34]. Table 1 Comparison of the two corrosion rate evaluations, Id and I ,

25 35 45 55 65

0.171 0.732 1.780 5.590 17.700

0.177 0.761 1.830 5.750 18.400

The values of the corrosion current density determinations listed in Table 1 clearly show that their agreement tends to deteriorate as the dissolution rate of ARMCO iron increases. This observation is of a general nature because the same behaviour was observed in the case of the behaviour of ARMCO iron in 0.5 m HzS04 solutions [40]. Finally, the presence of a contribution of the ohmic drop to electrode overvoltage becomes more important when one tries to evaluate I, using the three different versions of the three-point method [43, 44, 451. In its simplest form, the theory of this method,

389

suggested by Barnartt [43], imposes the choice of the three pairs of points (AE, i(AE) = il), (2AE, i(2AE) = i2) and (-2AE, 2(-2AE) = 23). When this choice criterion is adopted, the calculation of the anodic and cathodic Tafel slopes is reduced to the solution of the equation of the second degree (z= exp(crAE)) 5’

- (iz/k1)2

+ li2/i311’2 = 0 ,

(8)

provided sets of three points meeting the above requirements are available. For this purpose, reference [45] suggests the use of the best-fitting of experimental data, corrected to account for the contribution of the ohmic drop, by means of the polynomial of the fourth degree. i = AAE4 BAE3 CAE2 D A E . (9)

+

+

+

The use of this polynomial, which reproduces the polarization curves for A E values within the interval [-50,501 mV quite faithfully, permits the generation of the three pairs of points that satisfy the choice criterion.

MATHEMATICAL SCHEMATIZATION The influence of the solution resistance on the geometric shape of a polarization curve can be satisfactorily illustrated by means of a simplified equivalent electric representation of the double layer [46]shown in figure 6. This scheme is particularly valid in the domain of high frequencies, where all the faradaic components of the current do not respond to external excitation [47].

Figure 6. Equivalent electric representation of the metd-solution interphase. The components R,, Rt and C of the dipolar element in figure 6 represent the resistance of the solution between the working and reference electrodes, the transfer resistance [48]and the capacitance of the double layer [49],respectively. The quantities & and C,

390 which schematize the behaviour of the metal-solution interphase, depend on the potential difference A E and are of a differential type. The impedance 2, schematizes the behaviour of the reference electrode and does not require any splitting up into simpler elements because, not being involved in the external current flow, it takes no part in the subsequent developments. When an experimental polarization curve is performed, the overvoltage is applied from the outside between the points A and D of the dipole. Consequently, the true polarization, which coincides with the potential difference between B and D and determines the kinetics of the electrode process, generally differs from the measured value. It is evident from these considerations that in the ideal case, characterized by the condition R,=O 0, the current-voltage characteristic of an electrochemical system whose kinetics are controlled by the activation energy is described by the expression (2). In the real case, on the other hand, between the true potential difference AE' and the quantity A E , corresponding to the potential difference between A and D, there exists a relation AE' = A E - R,iS, (10) where S is the value of the exposed surface area of the working electrode. This expression shows that the absolute value of the ratio AE'IAE is always less than one. Thus equation (10) shows that the main effect of the ohmic drop is a reduction in the external polarization value. The amount of this attenuation is very important because it determines not only the variation of the geometric shape of the polarization curve but also the alteration of the values of the electrochemical parameters. An interesting aspect of the expression (10) concerns the case of metals and passive alloys because the real polarization potential exhibits a discontinuity around the zone of transition from active to passive state. In fact, if Ipdenotes the passivity current density, the value of the discontinuity is of the same order of magnitude as R,I,S because during this transition the current intensity falls very rapidly. The discontinuity may be very pronounced because the values of Ip,which depend on the type of metal, the environment and temperature, may be very high. In the case of the AISI 321H titanium-stabilized, austenitic stainless steel in 1 M HCIOl 0.3 M NaCl solutions at 25 "C, the value of Ip depends on the thermal history of the specimen [50]. In many instances it was found to be about 10 mAcm-'. The influence of the ohmic drop on the geometric shape of a polarization curve concerning an electrochemical system that contains only two predominant processes, controlled by the activation energy, is evident, if one examines the analytic expression of the experimental curve i = i(AE), which, with reference to the relation (lo), is represented by the implicit function

+

i - I,{exp[a(AE

- R,Si)] - exp[-P(AE - R,Si)J} = 0 .

(11)

Figure 7, concerning the behaviour of the SA 213 grade T22 low alloy steel in a solution containing 100 g/l of EDTA at pH 6 and a temperature of 100 "C, gives a very effective representation of this mathematical concept because it clearly shows the difference in shape between the experimental polarization curve (R,=O 0) and the curve closest to the ideal evolution of the corrosion process (R,=l.l 0). In the case under examination

39 1

4

t

1

i [mA ~ r n - ~

I

-300 -200

I

I

I

I

I

-100

0

100

200

300

AE

)

[mVl

Figure 7. Comparison of the geometric shapes of an experimental curve (R,=O 0) and its corrected form (R,=l.l 0). the exposed surface area of the working electrode was 60.5 cm2, while the polarization curve was obtained using the galvanostatic pulse technique. The most probable value for the solution resistance, R,, was 1.1 R. This evaluation was based on some determinations of the impedance Z ( w ) over the frequency interval [6, 101 kHz, the limit of its modulus being evaluated for values of w tending towards infinity [51]. Examination of figure 7 shows that the trend of the experimental polarization curve markedly differs from that of the curve corrected using the expression (10). It also reveals that the exposed surface area of the specimen plays a determinant role when, to attain a given overvoltage, use must be made of a high-intensity external current. In the present case, the difference in width (about 400 mV) between the two A E and AE’ intervals shows that, in some situations, the relevant contribution of the ohmic drop makes it difficult to perform significant measurements. This point is very important in the case of computerized systems for both the performing of polarization curves and the processing of experimental data without an operator. The success of the application of some numerical analysis techniques [34, 371 depends on the absence of problems concerning the convergence of numerical sequences used by the method adopted. Such problems may arise when the interval width of the potential difference A E is so small that the available experimental data do not contain the information required for a correct use of the numerical technique. In this case, the evaluation of the electrochemical parameters I,, LY and ,B by other methods not subject to convergence criteria is, in principle, physically unacceptable because in the region examined the law (2) cannot be deemed valid. This particular problem has been dealt with by the

392

present author and is discussed in detail in the work [52]. In this study he proposed a numerical technique, based on the best-fitting of the computed electrochemical parameters, that removes any troubles related to the interval width and yields the values of these quantities for the ideal case described by the law (2).

INFLUENCE OF THE OHMIC DROP ON THE CORROSION RATE Most corrosionists agree on the fact that corrosion current density is a very important parameter for the evaluation of the kinetics of a corrosion process and the proper choice of a metal to be used in a given environment with no prejudice to its integrity and performance. Hence it is very interesting to examine analytically the influence of the ohmic drop on the determination of the corrosion rate. In fact, this analysis makes it possible to detect a priori situations that may cause the behaviour of an electrochemical system to diverge from its ideal trend and render the use of equation (10) mandatory for a more reliable evaluation of the kinetics of the corrosion process. From a mathematical point of view, this problem can be readily solved by introducing the concept of polarization resistance, Rp, [5] associated with the implicit function defined by the equation (11). For this purpose, it may be useful to remember that, after Bonhoeffer and Jena’s definition [53], the differential polarization resistance is related to the first derivative of the function i(AE), calculated at the point AE=O, by the relation 1

- = i’(0) = Ic(a+ p ) .

(12)

RP

Denoting by F ( i , AE) the left-hand side member of equation ( l l ) , by Dini’s theorem on implicit functions [54] the apparent polarization resistance, Rp,, is defined by the relation

so that

the values of the two partial derivatives (dF/dAE)ooand (aF/di)oobeing given by

(dF/dAE)oo= -Ic(n

+ p) ,

+

+

(dF/dk)oo= 1 IcRsS(a p) .

(15)

It is evident from the formula (14) that the quantity Rp represents the actual polarization resistance. Although the mathematical development used for obtaining the previous formula differs from the one suggested by Mansfeld [26], the final results coincide and are consistent with the physical intuition and the approximate electric representation of the metalsolution interphase given in figure 6. Under steady-state or slowly variable conditions, the current does not flow through the capacitor C but is entirely sustained by the charge involved in the electrochemical process.

393 Experimentally, these conditions are satisfied when static or dynamic polarization curves are performed at small scanning rates of the voltage. Another indication of the formula (14) is that of the necessity of taking into account the solution resistance whenever the true corrosion rate of the specimen is sufficiently high, for in this case the quantity [ I c ( , @)I-' may be found to be of the same order of magnitude as the term R,S. This usually happens when a metal is in contact with an acid solution that has not been properly inhibited. In fact, when the temperature of the electrochemical system is higher than room temperature, the polarization resistance values are comparatively small. Such a situation was observed in the case of ARMCO iron specimens immersed in 0.5 m H2S04 solutions at 65 "C. In the case under discussion, for instance, the geometry of the electrolytic cell and the relative position of the reference and working electrodes, the latter with an exposed surface area of 10.7 cm2, gave rise to an electrolyte resistance of 0.36 0.On the other hand, the true polarization resistance value after 150 minutes was 1.08 R cm2, Hence the value of the term R,S was higher than the polarization resistance value. The corrosion current density value computed using the NOLI method was 24.1 rnAcm-', while the Tafel slopes were B,=152 mV and B,=102 mV. From the relation (15) we find that the corrosion current density is given by the expression

+

+

where the term [l R,S(&,, - R,S)-'] = A represents the correction factor. This factor takes into account the fact that the polarization resistance, R,,, is related to the experimental current-voltage characteristic. Apart from its speculative value, this observation represents experimental reality correctly because, in principle, one can only determine the approximate value of R, or eliminate its influence on the shape of the polarization curve by analog techniques. It is evident from the foregoing that every experimental polarization curve allows an implicit relationship to be set between current density and the electrode overvoltage due to an external signal, even though in many cases this link does not appreciably differ from the function that can be established on a theoretical basis. Examination of the expression (16) shows that the value of I, comes close to the real one only when the inequality

RsS <1, -R 3

Rpa

is satisfied. Taking into account the analytic expression of the relation (14), this inequality implies a well-defined determination of the polarization resistance because the two possible representations tend to coincide. An immediate consequence of (17) is that the influence of the ohmic drop is negligible when the corrosion process exhibits high inertia at any perturbation of its steady state. Thus the inequality (17) confirms the physical intuition that the contribution of the electric resistance of the solution to electrode overvoltage is negligible when the corrosion current density values are very small. This situation can be observed, for example, in the case of aluminium brass and of the 9O:lO Cu:Ni and 70:30 Cu:Ni copper nickel alloys interacting with artificial sea water

3 94

[55] because the formation of a protective film from corrosion products improves their resistance to corrosion. The polarization resistance values, which were calculated inside the AE interval [-30, 01 mV using the first derivative approach [53],were greater than lo4 R cm2, while the determined value of R, was of the order of some ohms. Finally, it is important to note that, since the quantity ( a p) refers to the ideal polarization curve, the computed value of I , may differ sensibly from the real one if no correction factor is used in analysing an experimental polarization curve. The validity of the formula (16) can be verified considering the hypothetical case described by the three parameters I,=O.l mAcm-', B,=30 mV and 8,=120 mV and by R,=15 R. Using the above formulas we have A=1.145 and &,=118.38 Rcm2, from which we obtain Ic=0.1008 mAcmP2. The slight difference between the exact value of I, and the value obtained using the expression (16) largely depends on the goodness of the determination of the polarization resistance, 4,. This was not achieved using the expression (14) but analyzing the curve that had been generated numerically by means of the relation (11) over the AE interval [-20,201 mV. This procedure was adopted because direct determination is impossible when one considers an experimental polarization curve, its three characteristic parameters being unknown. This observation is very significant, because it shows the validity of the mathematical formulation, and is corroborated by the self-consistencyof the I , evaluation with its exact value, which is not limited to the case under examination.

+

EXPLICIT FORM OF THE FUNCTION i(AE) The explicit relation i = i(AE) can be determined numerically starting from the implicit function (ll),which represents the central problem of the foregoing discussion, and using the Newton method [56], which, as everyone knows, is based on the iterative formula

and has a convergence of the quadratic type. In the expression (18) the quantity AE must be regarded as a parameter of fixed value and not as an independent variable. By the iterative process (18) the function i = i(AE)is represented in a tabular form, which permits the graphic visualization of its trend and the study of the dependence on R, of the values of the electrochemical parameters that characterize the behaviour of a given system. This last point is very important because it gives the opportunity to establish a criterion which permits to decide when the influence of the ohmic drop on the shape of the polarization curve must be taken into account, unless only general indications are desired. Therefore, it may be useful to examine in detail two situations that might be encountered in practical applications and help elucidate the concepts expressed above. Figure 8 shows the geometric shapes of some polarization curves pertaining to a hypothetical electrochemical system whose parameters are I,=O.l mAcm-2, B,=30 mV and 8,=120 mV, for values of AE falling within the interval [-20, 201 mV. This interval

395

1

I

[mA cm-*]

0.4 -

0.3 0.2

-

0.1

-

0.0

-

-0.1

-

'

-0.2) -25 -20

'

I

I

15

10 - 5

I

I

I

I

0

5

10

15

1

'

)

2 0 25

AE [mV]

Figure 8. Influence of the solution resistance R, on the geometric shape of the polarization curve. was chosen only to reduce calculation time because the mathematical treatment adopted does not require determination of the anodic and cathodic Tafel slopes. Curve 1 relates to the ideal case, while curves 2, 3 and 4 give the trends for R, values of 10 R, 50 R and 100 R, respectively. These values were selected to emphasize the difference in shape of the curves in the anodic zone. The first result that can be immediately inferred from the graphic observation of the polarization curves concerns the asymmetric behaviour in the anodic and cathodic zones, the latter being less affected by the contribution of the solution resistance, R,, to the electrode potential. This asymmetry can be explained by the fact that, at equal true lAEl values, the cathodic current density is lower than the anodic current density. The decrease in value of the first derivative i'(AE) at the point AE=O, as R, increases, is in line with the indications of equation (14) and clearly shows that, as can be seen from figure 6, the effect of the ohmic drop apparently results in an increase in the inertia of the electrochemical system. The assertion can be immediately verified by comparing, for example, curves 1 and 4. To study the influence of the resistance R, on the determination of the values of the electrochemical parameters a , and I, some i = i(AE) functions, in a tabular form, were analyzed using the NOLI method [34] over the interval [-20, 201 mV. This interval was chosen because it was indispensable that the iterative process should converge. The necessity of choosing a AE interval of moderate width over which to analyze numerical curves depends also on the significance of the analysis. For a numerical method to be correctly applied, the experimental curve must not differ excessively from

396 the ideal behaviour. In the case under discussion it was found that when the values of R, were high, analysis inside the interval [-50, 501 mV might yield meaningless results or give rise t o sequences that did not converge. Some results of these calculations are displayed in Table 2, where the quantities B, and B, are given in columns 2 and 3, while I, is shown in column 4. Table 2 Values of the electrochemical parameters computed using the NOLI method and the function (11)

R, [ill 0 1

2 3 4 5 7 9 11 13 15

B, [mVl 30.00 31.20 32.39 33.59 34.79 36.01 38.47 40.99 43.59 46.28 49.08

B, [mV] 120.0 130.3 141.2 153.0 165.8 179.2 211.8 211.8 301.7 368.8 463.3

I, [mAcm-'1 0.100 0.104 0.108 0.112 0.116 0.120 0.128 0.136 0.145 0.154 0.163

Examination of Table 2 shows that the variation of B,, as R, increases, is more moderate than the variation of B,, while the corrosion current density I, exhibits no appreciable variation when the value of the solution resistance is less than 4 R. The excessive increase in the value of B, is in good agreement with the flattening of the polarization curves in the cathodic zone, as shown in figure 8, and accounts also for the fact that the differentiation is not as marked as in the anodic region. The analysis of polarization curves by the NOLI method [34] is helpful also in determining the variations of the electrochemical parameters when R, is held constant and the corrosion current density is increased. This situation is of great practical importance because, for a large class of electrochemical systems, the corrosion rate varies considerably, while the value of the resistance R, remains practically constant. This is particularly true of acid solutions containing an inhibitor at variable concentration, so that the corrosion rate of the metal examined may vary from the maximum value, when the solution is not inhibited, to the minimum value, which can be observed in correspondence of the optimum concentration of the product. Our experience has shown that, in general, with a cell of the same geometry and analogous electrode positions, adding an inhibitor to an electrolytic solution does not greatly modify the electric resistance of the environment. The results obtained in the hypothetical case, characterized by B,=30 mV, 8,=120 mV and R,=2 R, are given in Table 3. The corrosion current density values shown in the f i s t column are very realistic and can be obtained, for example, by studying the behaviour of ARMCO iron in HZS04 solutions at variable concentration and at a temperature greater than room temperature.

397

Table 3 Values of the electrochemical parameters computed using the NOLI method and the function (11) with R,=2 0,B,,=30 mV, B,=120 mV and I, variable

I, [mAcm-’] 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 0.900 1.000

B, [mV] 32.39 34.79 37.23 39.72 42.28 44.92 47.67 50.54 53.54 56.71

B, Imvl 141.2 165.8 195.0 230.4 274.8 332.6 411.6 527.1 713.8 1071.0

I,, [ m A ~ m - ~ ] 0.108 0.232 0.372 0.528 0.702 0.894 1.106 1.342 1.603 1.895

Table 3 shows that a high corrosion rate of the specimen can noticeably influence the determination of the quantities B, and Ica.The latter quantity designates the apparent corrosion current density derived from the experimental curve without eliminating the contribution of the solution resistance to overvoltage. Thus, with reference to the interval examined, an important result is achieved: the difference between the two corrosion current density determinations is found to be a monotonic increasing function. Association of the formula (11) with the NOLI method serves the twofold purpose of providing a theoretical justification to the experimental intuition and stressing the necessity of eliminating or reducing the negative influence of the resistance R,, when the true corrosion rate of a metal in a given environment is quite high.

INVERSION THROUGH POWER SERIES EXPANSION Concurrently with the above mathematical treatment there exists another approach to the problem of the determination of the explicit form of the function i ( A E ) starting from equation (ll),which is based on its MacLaurin’s series expansion [57]

+

~3 i ( A E ) = i’(0)AE ~ ( z ) ( o ) A E ~i ( 3 ) ( 0 ) ~+... 2! 3! +

The values of the derivatives of the function i ( A E ) , calculated at the point AE=O, which appear in the expression (19), can be determined considering the differential equation and taking the total derivative of the left-hand side member with respect t o the indepenis defined by dent variable AE. Thus the expression of the second derivative i(2)(AE) the equation

398

which can be derived from (20) by simply applying the rule of the derivation of a function of function. With regard to equation (19), it should be noted that all the second partial derivatives of the function F ( A E , i ) that appear in equation (21) are known and their expressions at the point (0, 0) are given by

= - I , ( R , S ) ~ (~ ~p2)

.

00

By combining the relations (20), (21) and (22) it is possible to determine the only unknown, i@)(O).Thus we have a recurrence procedure which, in principle, makes it possible to determine the value of the derivative i(.)(AE) at AE=O and to approximate the value of the function i = i ( A E ) by means of the polynomial of degree n. Generally, the goodness of the approximate representation of the function i = i ( A E ) by a polynomial of degree n depends on the value of the resistance R,. Moreover, considering a fixed A E interval, it can be numerically ascertained that the difference between the present methodology and the one based on equation (11) becomes more accentuated, as R, increases. Despite the critical aspect of the foregoing consideration, which depends on the validity of a given representation over a generic A E interval, it may be useful to study the influence of the resistance R, on the response of a real system by means of a polynomial of the same form as equation (19) because, from the point of view of numerical calculus, this allows a considerable saving in calculation time. A numerical study of the influence of the ohmic drop on the evaluation of electrochemical quantities has been conducted, for example, over the A E interval [-20, 201 mV by means of the IRCOM program, which makes use of a polynomial of the sixth degree, considering some experimental polarization curves and taking the values of the electrochemical parameters obtained by the NOLI method. The examples examined have shown that the representation of experimental data by a polynomial of the sixth degree is very good and that the evaluation of the correct order of magnitude of the corrosion current density, in the presence of an ohmic contribution to the electrode potential, requires that the actual values of the Tafel slopes be known. The importance of this problem is underlined by the data in Table 4 concerning the behaviour of ARMCO iron in 1 N HC1 solutions inhibited with the commercial product Borg P16 at a temperature of 75 "C.Borg P16 is a specific commercial inhibitor of the corrosion of iron and carbon steels in hydrochloric acid solutions and is a derivative of the commercial product Rodine 213. This inhibitor is used both for the acid cleaning of steam generators and for pickling baths.

399

The determinations of R,, which was made using high-frequency alternating current, showed that its value was approximately 0.3 R. The polarization curves were performed using the galvanostatic pulse technique and their geometric shapes were analyzed by the NOLI method [34]. The polarization resistance values in column 2 relate to curves corrected for eliminating the contribution of the ohmic drop. The corrosion current density values I,, and I,, in columns 4 and 5 were computed using the approximation of the linear response and the same values of a and p. I,, denotes the corrosion current density obtained from the corrected polarization curves, while I,, refers to the values obtained directly from the experimental curves. Table 4 Comparison of two I, determinations Inhibitor [cm3/1]

I&[Rcm2]

0.001 0.01 0.05 0.1 0.5 1 .o 3.0

1.67 78.3 358.0 493.0 783.0 812.0 779.0

I,, [mAcm-’1 49.958 0.347 0.099 0.071 0.050 0.048 0.049

I, [mAcm-’] 20.358 0.333 0.098 0.070 0.050 0.048 0.049

Examination of the two I, determinations confirms the result, already known, that there is no difference in the information contents between the experimental and the ideal curves when the polarization resistance value is much greater than the R,S value. Yet the data for a concentration of 0.001 cm3/l demonstrate the practical impossibility of using the values of the Tafel slopes without considering the experimental conditions. This observation should warn against the arbitrary employment of literature data when no check has been made on the way they have been obtained and the correctness of their use. Our experience has shown that the application of the polarization resistance method for a reliable evaluation of the corrosion rate requires that the Tafel slopes should refer exclusively to the polarization curve under examination.

NUMERICAL DETERMINATION OF R, An example of numerical correction for the ohmic drop contribution is discussed in reference [58], even though the subject is not dealt with in its generality. The best way to treat this theme is to start from equation (11) and assume that the two arguments aJAE- R,Sil and p(AE- R,Sil of the exponential function are much greater than one. If, for example, we examine the polarization curve in the cathodic zone, the formula (11) takes the approximate form In J i J= In I ,

-

pAE

+ ,BR,SZ.

(23)

400

Examination of this relation shows that the difference between the actual trend of the polarization curve and the one conforming to the Tafel law increases with increasing the ohmic drop contribution. It also shows that, in the case of an experimental curve, the parameters to be determined are generally I,, p and R,. The importance of the relation (23) lies in the fact that it can be used for the evaluation of the solution resistance R,, adopting the numerical technique of successive approximations and assuming that the electrochemical behaviour of the system under examination is described by the Tafel law. For the application of this method the program IRCOR has been developed. This code can be used with IBM compatible personal computers and is written in basic language. The present version investigates only the cathodic region, permitting also the examination of some hypothetical cases. It should be noted, however, that this method does not find current application in our laboratories. The example that will be discussed refers to an ideal curve described by 1,=0.8 mAcm-', B,=40 mV and B,=120 mV. The value of R, was assumed to be equal to 1 0.The current density values inside the AE interval [-270, -1301 mV were computed using the Newton method. Table 5 lists some initial data of the problem which were used for determining I,, B, and R,, the values of AE' and those of the ohmic drop contribution. Table 5 Initial data for the application of the relation (23)

AE [mV] AE' [mV] i [rnA~m-~]RSSi[mV] -9.69 -8.27 -130 -121.7 -14.23 -11.42 -150 -138.6 -170 -154.5 -20.88 -15.51 -190 -169.4 -30.65 -20.63 -37.13 -23.61 -200 -176.4 -54.50 -30.41 -220 -189.6 -42.71 -207.3 -96.92 -250 -142.26 -52.23 -270 -217.8 To determine the quantities I,, B, and R, 15 iterations were used and the values obtained were the following: I,=0.799 mAcm-', BC=119.96mV and R,=1.46 R. It is evident from this example that, though the evaluation of the electrochemical parameters I, and B, is very good, this method for the numerical determination of the solution resistance introduces an unacceptable error in the calculation of R,. But this indetermination does not depend on the numerical method that has been adopted because the values of I, and B, are reproduced very faithfully. This aspect is very important because it shows that the accuracy of a technique should be assessed by examining some cases in which all the parameters are known. The validity of this observation is more general and our studies have demonstrated that the evaluation of R, depends on the value of the corrosion current density. With a corrosion current density equal to 0.1 mAcm-2 the value of R, is 1.09 R.

40 I

The most plausible explanation for this behaviour is that the omission of the contribution of the anodic process requires more careful evaluations, especially when the values of the ohmic drop are not negligible. The data in Table 5 also show that the effect of the environment on the geometric shape of a polarization curve is twofold: while the characteristics of the metal and the aggressiveness of the medium are determinant for the corrosion current density, the nature of the environment has a strong influence on the value of R,. Hence the contribution due to the ohmic drop is determined by a combination of these effects.

EXPERIMENTAL TECHNIQUES The points discussed so far should have shed some light on the problems associated with the processing of experimental data and shown how mistaken it may be to neglect the presence of the ohmic drop. But the assessment of the validity of a given reaction scheme for explaining the kinetic behaviour of an electrochemical system depends, above all, on the reliability of the experimental data. From this point of view, it is very important that a corrosionist be familiar with the experimental techniques that are adopted for performing electrochemical measurements and with their mode of application. An important chapter in the study of experimental techniques concerns the methods that can be used for evaluating the quantity R, or reducing the effect of its presence. Experience has shown, in fact, that absolute determinations are practically impossible, also because they depend on the distribution of current density in the bulk of the electrolytic solution. A distinction must be made between evaluation and attenuation of the presence of R, because a reduction in the effect of R, on the polarization potential does not necessarily imply that the value of R, is known. As has already been discussed, the presence of an ohmic drop is not exclusively imputable to the separation of the working electrode from the reference electrode in the solution bulk but it also depends on the electric connections between the electrolytic cell and the potentiostat and the way the working electrode has been assembled. The electric connections are very important when the current intensities absorbed by the working electrode are high and the measurements are made over long distances. In fact, electrode potential measurements may be distorted on account of the contact resistances and the resistance of the wire connecting the potentiostat with the working electrode. A significant case is that of potentiostatic pulse measurements performed by inserting a mercury wetted relay between the working electrode and its input to the potentiostat. The advantage of these measurements, which can be carried out using, for example, an Amel mod. 551 potentiostat, is that they do not modify the state of free corrosion of the specimen by introducing undesired polarizations when no pulse is applied. In addition, this technique makes it possible to ascertain whether the perturbation caused by a polarization potential being too far from the free corrosion potential is such that the system will be allowed to return to its initial state. From this standpoint, the potentiostatic pulse technique does not differ from the galvanostatic technique. An example of application of the galvanostatic technique is given in figure 9 concerning

402

the CORRCONTROL computerized system [41]controlled by the GALIMP program (421.

Figure 9. Block diagram of the computerized CORRCONTROL system in the galvane static mode. All the problems related to measurement distortions can be overcome by adopting the four-wire technique. This technique is indispensable when using alternating current or scanning several cells. In the one case, it is very important that the inductive effect of wirings in the region of high frequencies be eliminated; in the other case, the purely ohmic contribution due to the relays may be overwhelming. At this point, it is important to note that some commercial potentiostats, used in the potentiostatic mode, do not adequately respond to a sinusoidal signal in the high frequency region because they introduce an undue phase shift or attenuate the amplitude of the input signal considerably. Thus, under such conditions, the use of a Luggin capillary [59] may not be sufficient to guarantee accuracy in the measurements because of the presence of the contact resistance and of the resistance of the wire connecting the working electrode to the potentiostat. Several experimental methods of reducing or evaluating the electric resistance of the solution between the working and reference electrodes have been reported in the literature

403 [59, 60). The commonest techniques are the following:

a) the Luggin capillary; b) ohmic drop compensation by positive feed-back; c) the current interruption technique using the galvanostatic mode; d) infinite frequency extrapolation of the modulus of surface impedance.

The Luggin capillary The functionality of the Luggin capillary in reducing the ohmic drop contribution to the electrode potential depends also on the choice of the experimental technique. The use of four wires permits the elimination of any type of resistance outside the electrolytic cell. The use of three wires, on the other hand, is satisfactory only if the resistances outside the cell can be neglected, regardless of the intensity of the current flowing in the system. Moreover, a Luggin capillary should not be used when its presence may perturb the hydrodynamic conditions of the electrolytic solution. In the solution bulk the value of R, is reduced by positioning the tip of the capillary very close to the surface of the working electrode. The principle on which the application is based is that the value of R, tends towards zero when the tip of the capillary touches the surface of the working electrode. The necessity of positioning the capillary tip near the surface of the electrode might result in a perturbation of the distribution of the current density over its surface while polarization curves are being performed. This problem is generally solved by positioning the tip of the capillary, with respect to the working electrode, in such a way that the current lines will not be cut [61]. All the above difficulties can, however, be overcome by adopting the extrapolation technique suggested by Bockris [62]. A critical survey on this subject has been made by Sundheim [63].

Compensation by positive feed-back The method of positive feed-back has been suggested by Lauer and Osteryoung [64] and is commonly utilized in most commercial equipments. This method can be employed using either the four-wire or the three-wire technique, as it permits compensation for any type of resistive load between the working and reference electrodes. A schematic diagram is given in figure 10. The mode of application of this method need not be described in this article, since it is thoroughly illustrated in many user’s manuals of commercial potentiostats. It should be recalled, however, that when this technique is adopted use must be made of an oscilloscope and a function generator because for an optimum choice of the feed-back amplitude it is necessary that the wave shape of the current peak be strictly controlled. It may be added that the use of the Solartron mod. 1286 electrochemical interface makes it possible also to choose the amount of the positive feed-back numerically. However, ohmic drop compensation by positive feed-back has some disadvantages: it can only be used in the potentiostatic mode and it requires the use of a fixed current range. This is because, in order to reduce the effect of the presence of R, on the polarization potential of the working electrode, a voltage signal (Ej) coming from the output of the

404

Figure 10. Schematic diagram of ohmic drop compensation by the positive feed-back technique in potentiostatic mode. zero-resistance operational amplifier (C) is sent to the input of the summing amplifier (A) of the potentiostat and is added to the initial polarization potential (Ei) and the modulation potential (EM). Thus compensation for the resistance R, is achieved by increasing the absolute value of the actual polarization potential. As a result, the choice of the current range cannot be optimized because any change, at a fixed current intensity, after the amplitude of the positive feed-back has been optimized involves a variation of the potential Ef. From the point of view of a corrosionist who wishes to have an accurate and reliable picture of the corrosion process, there are two principal limitations to a generalized use of this method. First, when the current intensity variations cover a very wide interval, it is practically impossible to examine the kinetics of the electrochemical system accurately. In fact, the current range must be selected in view of the maximum value of the current intensity that will be obtained during the performing of the polarization curve. Thus, in principle and in the course of each measurement, the study of the behaviour of a system near its free corrosion potential is practically incompatible with the examination of the currentvoltage characteristic in the anodic and cathodic irreversibility zones. Second, if the actual value of R, should vary in the course of a measurement, the

405

geometric shape of the polarization curve might undergo an irreversible alteration.

The current interruption technique The current interruption technique [59] can be simply and accurately described with reference to the circuit in figure 11 and using the Laplace transform [65] to examine its behaviour over the time interval [0, t].

3

Figure 11. Circuit scheme illustrating the current interruption technique for the ohmic drop evaluation. With reference to figure 11 and denoting by I ( p ) , % ( p ) and & ( p ) the Laplace transT) forms of the current intensity I ( t ) = l o [ l - H ( t - T ) ] (H=O for t < T , H=l for t and the potential differences K ( t ) and & ( t ) we have the set of equations

>

Using the inverse Laplace transform of the function K ( p ) and recalling that I ( p ) is give by the function l o [ l - exp(-~p)]/p we obtain the two functions

+

V I ( ~= ) IoR lo&{ 1 - exp[-t/(GR,)]}

( t < 7)

(26)

and (t 2 7 ) . (27) T)/(C&)I - exp[--t/(C&)lI Comparing the two expressions (26) and (27) and assuming, for sake of simplicity, T to the greater than the time constant CR, (exp[-~/(C&)] M 0) we find that the instant

K(t) = lo&(exp[-(t

-

the external current is interrupted there is a sudden change in the potential difference K ( t ) equal to IoR. Currently, the current interruption technique can be employed following two distinct procedures. One is based on the high-speed sampling of the voltage signal [66] and

406

the other on the conventional approach of its storage by means of an oscilloscope with direct-current coupling. With most of the latest commercial equipments the high-speed sampling method can be used for evaluating the solution resistance. This procedure is attractive and may be of considerable assistance but it has some drawbacks: high-speed sampling imposes some resolution limits, the data obtained may be spurious, the signal must be processed so as to determine the value of its discontinuity when the current is interrupted. The application mode of this technique is described in some detail in the user’s manual for the EG&G mod. 273 potentiostat. The conventional approach demands greater attention of the corrosionist and enables him to check the validity of the various operations directly. It also permits the current intensity to be selected so that the value of the discontinuity R,IO can be determined with great accuracy, even in the case of accidental or systematic errors. For a correct use of this method, which can be applied using either the three-wire or the four-wire technique, the electric characteristics of the equipment must be examined very carefully in order to define the experimental procedures. In principle, no problems are encountered if use is made of potentiostats like EG&G’s mods. 173 and 273 or of the Solartron mod. 1286 electrochemical interface. The electrochemical system, however, must be polarized by means of a current square wave of such duration as to permit the polarization potential to reach a steady-state value. It is important to note that, when the equipment is comparable to the Amel mod. 551 potentiostat, in which, in the galvanostatic configuration, the working electrode is not grounded, use must be made of a differential input or floating oscilloscope. However, the Amel mod. 551 potentiostat gives the opportunity to perform polarization curves of the galvanostatic pulse type by driving a mercury wetted relay with an external square wave. A very interesting application of the classical current interruption technique has been reported by Lorenz and Eichkorn [67], who showed that by adopting the galvanostatic configuration it is possible to evaluate the importance of the ohmic drop realistically, point by point, and obtain polarization curves with a trend very close to the ideal one, In fact, it can be experimentally demonstrated that the value of R, is not constant but is influenced by the mass transfer when the current flowing in the electrolytic cell is sufficiently high. In other words, it cannot be excluded a priori that the quantity R, depends on the electrode overvoltage. The application of the current interruption method is very simple in all the cases in which the values of the solution resistance are sufficiently high. A drawback of this technique is that, in many situations it is necessary to use a high intensity current in order to determine the ohmic drop contribution to the polarization potential and this may perturb the steady state of the system significantly. Problems of this kind may be encountered, for example, when studying corrosion inhibitors for acid environments characterized by high electric conductivity. This is at once apparent in the case of a good corrosion inhibitor having an effectiveness greater than 90% [68]. In this case the values of the current densities that are needed to obtain a significant shift of the electrode potential are rather small. Hence to solve the contribution of the term R,I use must be made of such current intensities that

401

will cause shifts of hundreds of millivolts in the electrode potential with respect to its free corrosion potential.

High-frequency alternating current This experimental approach is based on the use of alternating current (AC) in the high frequency range compatibly with the reliability of the response of the potentiostat employed. This concept is illustrated in figure 12, which shows the trends of the function Zr=Zr(w) over the frequency interval [l, 201 kHz.

t

Z, [ ~ c r n * ]

300

0 Aluminium brass

A 90 10 0 70 30

!

:

0

2

Cu NI Cu NI

2140 180 140

Y

I

0

I

I

I

40000

80000

w

Y

I )

120000

[s-'1

Figure 12. Shapes of the real part Z r ( w ) of the electrode impedance over the frequency interval [I, 201 kHz. 2 x lo4] The alternating current measurements over the frequency interval [8 x Hz were performed using a Solartron mod. 1286 electrochemical interface and a Solartron mod. 1250 analyzer, controlled by the SOFTCOR-AC-PIC program [69], adopting the potentiostatic configuration and applying a sinusoidal voltage of 5 mV RMS amplitude. The curves relate to the behaviour of aluminium brass and of the 9O:lO and 70:30 copper nickel alloys in artificial sea water at pH 8.2 and a temperature of 40 "C. The surface area of the electrodes was about 33.7 cmz, while the sea water flow rate was 1.0 ms-'. The geometry of the three electrolytic cells was the same. Examination of the shape of the curves clearly shows that the electrochemical interface introduces an unwanted inductive contribution in the high frequency range. The validity of the previous remark is evident because in the three cases considered the R, determinations should practically coincide. Actually, three distinct values were found: 4.8 R (aluminium brass), 3.7 0 (90:lO Cu:Ni) and 4.4 R (70:30 Cu:Ni). Our experimental results seem to indicate that the response of the electrochemical interface depends on the nature of the system examined.

408

I INTERFACE COMPUTER APPLE I I e

.!i

GP

-

IB

mod. 2 7 6

POTENTIOSTA EG&G mod. 1 7 3

-

i l i I

E

I

FREQUENCY RESPONSE

1

PRINTER

'7MODULATION '

Figure 13. Schematic view of the computerized system employed for performing AC measurements in galvanostatic mode. Figure 13 shows a schematic sketch of the system which is adopted in our laboratory for performing alternating current measurements. Our laboratory experience, however, has shown that, when measurements are performed adopting the galvanostatic mode, the use of alternating current is a very satisfactory means of determining the value of the resistance R, accurately. In fact, unlike the direct current interruption technique, the AC technique can be utilized for a very large class of electrochemical systems. This method falls short only when the capacitance values of the metal-solution interphase are so small [70] (of the order of some picofarads) that the limit of the impedance modulus as 1/w 4 0 does not provide a reliable evaluation when the maximum frequency value is about 60 kHz. This situation can be observed especially in the case of organic coatings where, for practical purposes, the presence of microscopic defects is negligible. From a theoretical point of view, the application of this method is based on the equivalent electric representation of the metal-solution interphase (figure 6).

409

Since the impedance Z ( w ) of the dipole in figure 6 is given by

where the quantity Q(w) is equal to Q(w) = 1

+ ( w C R ~,) ~

while j is the imaginary unit, we have

R, = lim IZ(w)l. W'CC

I+------IZl

[QI

1.2 1

0 = S A 213 T22+EDTA 10% T=100T

:SA 213 T 9 +EDTA 10% Tz100"C

A = S A 213 T22+HCI 5% l g / l BORG

pH.6 pH17

P16 T-76"C

Figure 14. Illustration of the numerical extrapolation technique for computing the solution resistance R,. Figure 14 gives axi example of the evaluation of R, by the extrapolation method, which is usually used in our laboratory. The evaluation relates to the three electrochemical systems SA 213 grade T22+EDTA 100 g/1 at pH=6 and 100 "C, SA 213 grade TSSEDTA 100 g/l at pH=7and 100 "C and SA 213 grade T22+HC150 g/kg 1 g/kg Borg P16 at

+

76 "C. With reference t o figure 14, it can be observed that the value of the resistance R, was determined considering a first-degree polynomial, (Z(l/w)l=R,+A/w, of the independent

410

variable l / w and that the behaviour of the experimental points is represented very well by a straight line. Figure 14 does not depict a successful application but illustrates a behaviour that is common to all the electrochemical systems that have been studied in our laboratory, for it is physically correct to believe that at high frequencies the current flow through the metal-solution interphase is assured by the displacement current. Based on this consideration and on the certainty that this technique is more effective than the current interruption method, the codes SOFTCOR-AC-GS [71] and SOFTCORAC-GE [72] have been developed for determining the value of the resistance R, through surface impedance measurements over the interval [7, 131 kHz when the capacitance values are rather small. The two codes, written in basic language for an Apple IIe computer, use the galvanostatic mode and drive, respectively. the Solartron mod. 1286 electrochemical interface and the EG&G mod. 173 potentiostat, which is equipped with a mod. 276 interface. In both cases use is made of a Solartron mod. 1250 frequency response analyzer. The principal application of the two codes, a version of which for IBM compatible computers is under development, concerns especially the study of the behaviour of ferrous materials in acid environments with or without corrosion inhibitors. In particular, the use of the frequency interval [0.08, 20 x lo3] Hz in the SOFTCOR-AC-GS code permits a more accurate characterization to be made of the properties of corrosion inhibitors by evaluating some electrochemical parameters under the assumption that the experimental curve 2; = f(&) is satisfactorily represented by a circle equation.

CONCLUSIONS The subject of the present work has been dealt with in all its theoretical and experimental aspects with a view to providing useful data and hints for further study, also because the success of numerical methods is strictly connected with the value of the solution resistance. The change in the geometric shape of the polarization curve depends, in fact, not only on the value of R, but also on the corrosion current density, which determines the intensity of the current flowing in the electrochemical systems. From this point of view, the function of the environment is twofold because, while, on the one hand, it tends to introduce a resistive control, on the other hand, its intrinsic aggressiveness has a strong influence on the form of the response to external perturbations. The ohmic drop exerts a sensible influence on the evaluation of the electrochemical parameters as well as on the definition of the reaction scheme that is most suitable for describing the behaviour of a metal in a given environment. It also determines the success of many operations, such as cathodic protection by means of sacrificial anodes or impressed current and corrosion rate monitoring. It is, therefore, very important for a corrosionist t o be able to adopt experimental and theoretical techniques that will enable him to evaluate the discrepancies between actual response and ideal trend and assist him in the study of particular processes. The presence of a ohmic drop may, in fact, result in the introduction of systematic errors in

41 1

the evaluation of the aggressiveness of an operating environment or the characterization of corrosion inhibitors. It should be remembered, however, that there exist many electrochemical systems where the contribution of the ohmic drop to the electrode potential may be ignored with no risk of incurring appreciable errors. The necessity of determining the value of R, with a good degree of accuracy becomes more pressing when computerized systems are employed or when experimental data must be processed in real time and with no operator, using non-linear numerical methods, because in this case the reduction in the ohmic drop contribution to the electrode potential might influence the convergence of the numerical methods very unfavourably.

REFERENCES 1 Praiik M. and Bartofi K., Corros. Sci. 1967; 7: 159. Feitler H., Mater. Perf. and Prot. 1970; 9 - No. 10: 37. Pra&ik M., Werkst. Korros 1974; 25: 104.

2 3 4 5 6

Wilde B.E., Corrosion 1967; 23: 379. Stern M., Geary A.L., J. Electrochem. SOC. 1957; 104: 56. Indig M.E. and Groot C., Corrosion 1969; 25: 455. 7 Mansfeld F., Corrosion 1976; 32: 143. 8 Jones D.A., Corros. Sci. 1968; 8: 19. 9 Lohr E.W. et al. Geological Survey Circular 203, 232, 288. Washington, D.C. 1952. 10 Mazza F., Torchio S. and Ghislandi N., Proc. 9th Int. Cong. on Met. Corrosion. Toronto 1984; 2: 102. 11 Mazza F., Bellucci F., Faita G., et al., Corros. Sci. 1988; 28: 371. 12 Farina C.A., Faita G. and Olivani F., Corros. Sci. 1978; 18: 465. 13 Szklarska-Smialowska Z. and Mankowski J., Corros. Sci. 1982; 22: 1105. 14 Whittington H.W., McCarter 3. and Forde M.C., Magazine of Concrete Research 1981; 33 - NO. 114: 48. 15 Stratfull R.F., Corrosion 1957; 13: 173t. 16 Wenner F., Bulletin of National Bureau of Standards 1915; 12: 469. 17 Rocchini G., CORROSION/91. Cincinnati 1991; paper No. 126. 18 Wall& B. and Linder B., Brit. Corros. J. 1973; 8: 7. 19 Perboni G . , Private Comm., April 1984 20 Vetter K.J., Electrochemical Kinetics Theoretical Aspects. Academic Press, New York 1967: Ch. 2. 21 Andersen T.N. and Eyring H., Principles of Electrode Kinetics. In: Physical Chemistry an Advanced Treatise. Academic Press, New York 1970: IXA; Ch. 11. 22 Marsh G.A., Proc. 2nd Int. Cong. on Met. Corrosion. New York 1963: 936. 23 Rocchini G., CORROSION/91. Cincinnati 1991; paper No. 227. Corros. Reviews (in press.). 24 Rocchini G. and Colombo A., CORROSION/92. Nashville 1992; paper No. 419.

412

25 Bird D.W., Corros. Sci. 1973; 13:913. 26 Mansfeld F., The Polarization Resistance Technique for Measuring Corrosion Currents. In: Advances in Corrosion Science and Technology, eds. Fontana and Staehle. Plenum Press, New York 1976; 6: 163-262. 27 Kilpatric J.M., Oil Gas J. 1964; 62:155 28 Townsend C.R., CORROSION/74. Chicago 1974; paper No. 64. 29 Perboni G. and Rocchini G., in: Corrosion Prevention in the Process Industries, Proceedings of the First NACE International Symposium, edr. Parkins. NACE, Houston 1990: 337. 30 Rocchini G., CORROSION/86. Houston 1986; paper No. 353. 31 Hausler R.H., CORROSION/86. Houston 1986; paper No. 275. 32 Mansfeld F., Corrosion 1973; 29: 397. 33 Greene N.D. and Gandhi R.H., Mater. Perf. 1982; 21: 34. 34 Rocchini G., ENEL-DSR/CRTN Internal Report No. G6-1; 1979. 35 Rocchini G., Corrosion 1987; 43: 326. 36 Rocchini G., CORROSION/88. St. Louis 1988; paper No. 103. 37 Rocchini G., CORROSION/91. Cincinnati 1991: paper No. 159. 38 Rocchini G. and Perboni G., Proceedings of Corrosion Week. Budapest 1988; 1: 394. 40 Perboni G. and Rocchini G., Proc. 10th Int. Cong. on Met. Corrosion. Madras 1987; 1: 193. 41 Colombo A. and Rocchini G., Proceedings of EUROCORR’82. Budapest 1982; 2: 13. 42 Perboni G. and Rocchini G., in: Dechema Monograph, eds. Heitz, Rowlands and Mansfeld. Frankfurt a m Main 1986: 327. 43 Barnartt S., Electrochim. Acta 1970; 15: 1313. 44 Le Roy R.L., J. Electrochem. SOC.1977; 124: 1060. 45 Rocchini G., Corros. Sci. 1990; 30: 9. 46 Macdonald D.D., J. Electrochem. SOC.1978; 125: 1443. 47 Rocchini G., Corrosion Prevention & Control (in press). 48 Epelboin I., Gabrielli C., Keddam M. and Takenouti H., in: Electrochemical Corrosion Testing. ASTM Special Technical Publication 727, 1981: 150. 49 Hackerman N. and McCaEerty E., Proc. 5th Int. Cong. on Met. Corrosion. NACE, Houston 1974; 542. 50 Zucchi F., Private Comm., May 1982. 51 Rocchini G., Corrosion Reviews (in press). 52 Rocchini G., Proceedings of UK Corrosion 92. Manchester 1992 (in press). 53 Bonhoeffer K. and Jena W., Z. Elektrochem. 1951; 59: 151. 54 Scorza Dragoni G., Elementi di Analisi Matematica. Cedam Casa Editrice Dott. Antonio Milani, Padua 1963; 2: Ch 18, Sec. 3. 55 Rocchini G., Proceedings of International Conference on Monitoring and Predictive Maintenance of Plants and Structures. Florence 1992; session 7: 595. 56 Demidovitch B. and Maron I., Elements de Calcul Numdrique. Mir Publishers, Moscow 1979. Ch. V, Sec. 4.

413

57 Smirnov V., Cours de MathCmatique SupCrieures. Mir Publishers, Moscow 1970; 1: Ch 4, Sec. 2-3. 58 Kajimoto Z.P., Wolynec S. and Chagas H.C., Corros. Sci. 1985; 25: 35. 59 Yeager E. and Kuta J., Techniques for the Study of Electrode Processes. In: Physical Chemistry an Advanced Treatise. Academic Press, New York 1970; IXA: Ch. IV. 60 Macdonald D.D., Transient Techniques in Electrochemistry. Plenum Press, New York 1977; Ch. 2, Sec. 2-4-2. 61 Piontelli R., Bianchi G. and Aletti R., Z. Elektrochem. 1950; 56: 86. 62 Bockris J.O’M., Electrode Kinetics. In: Modern Aspects of Electrochemistry. Academic Press, New York 1954: Ch. 11, Sec. 3. 63 Sundheim B.R., J. Electrochem. SOC.1968; 115: 158. 64 Lauer G. and Osteryoung R.A., Anal. Chem. 1966a; 38: 1106. 65 Ghizzetti A., Calcolo Simbolico. Nicola Zanichelli Editore, Bologna 1943. 66 Kuhn M., Schiitae K.G., Kreysa G. and Heitz E., in: Dechema Monograph, eds. Heitz, Rowlands and Mansfeld. Frankfurt a m Main 1986: 265. 67 Lorenz W.J. and Eichkorn G, Ber. Bunsengens. Phys. Chem. 1966; 70: 99. 68 Rocchini G. and Perboni G . , Corrosion Reviews 1988; 8, Nos. 1&2: 35. 69 Rocchini G., work in progress. 70 Colombo A., Rocchini G. and Spinelli P., Corrosion Reviews (in press). 71 Rocchini G., work in progress. 72 Rocchini G., work in progress.

This Page Intentionally Left Blank

SECTION FIVE

PHOTOELECTROCHEMISTRY FOR A

CLEANER ENVIRONMENT

This Page Intentionally Left Blank

41 7

SEMICONDUCTOR PHOTOELECTROCHEMISTRY FOR CLEANER ENVIRONMENT UTILIZATION OF SOLAR ENERGY

Yu. V. Pleskov A. N. Frumkin Institute of Electrochemistry, The Russian Academy of Sciences, 117071 Moscow. Russia.

SUMMARY Generation of photocurrent at the semiconductor/electrolyteinterface upon its illumination makes it possible to carry out photoelectrochemical reactions which can be used either for chemical fuel production, or purification of waters. Principles of operation of electrochemical cells with semiconductor electrodes for solar energy conversion to electrical and chemical energy are formulated. Most efficient cells for electricity and hydrogen production are surveyed. Certain processes for photo-destruction of pollutants, recovery of metals, etc. with making use of semiconductor dispersions are briefly discussed. INTRODUCTION The utilization of solar energy can basically solve two joint problems, energetic and ecological. As such, the conversion process is ecologically clean. Indeed, no wasteproduct emerges; carbon dioxide does not evolved (which is believed to give rise to the so-called greenhouse effect); finally, the thermal balance of Earth is not disturbed since the process under discussion is just "built-in", as an intermediate stage, into the continuously proceeding natural process of solar energy transformation into heat Therefore, by bringing solar energy into use the mankind makes it possible to scale down the consumption of fossil fuels with their inherent ecological problems. On the other hand, conversion of solar energy into chemical energy underlies some photoelectrochemical methods of degrading pollutants, e.g., in wastewaters. Among different methods of solar energy conversion, namely, thermal, thermochemical, photochemical, and photoelectric (photovoltaic), it is the latter that has a photoelectrochemical alternative. Moreover, it is the constantly growing requirement in cheap and practical device for solar energy conversion that stimulated the progress in the photoelectrochemical research during the 1980s. The most promising version of the conversion method involves photoelectrochemical (PEC) cells with semiconductor electrodes. Their principle of action, like that of the solid-state semiconductor solar cells, is based on the excitation of the electronic system in the semiconductor upon absorption of light. This results in particular in photogeneration of electron-hole pairs. In a solid-state cell photo-generated non-equilibrium charge carriers cross the semiconductor/metal interfaces and produce the electric current in the external circuit. In a PEC cell one of the semiconductor-to-metal interfaces in replaced by the semiconductor/electrolyte interface. As a consequence, the price of the device is likely to be reduced since the electrolytic system is much less sensitive to the crystalline perfection and purity of a semiconductor than the solid-state system. Hence, inexpensive and readily available materials can be made use of. (It is high price, not poor performance characteristics that narrows down the applicability of, e.g., silicon solar cells at present ) Moreover, passing of

418

current trough an electrolytic system naturally involves proceeding of an electrochemical reaction whose products may accumulate energy of light absorbed in the semiconductor electrode, in a form that makes it possible its subsequent utilization. Thus solar energy conversion and storage can be achieved at a single unit. The very first PEC cell with a Ti02 photoanode for water splitting was described by Fujishima and Honda [ I ] as long ago as 1972. After few years many electrochemists centered their attention at this paper, which gave way to numerous efforts to develop a new method for utilization of solar energy. Incidentally, this interest was stirred up by the "petroleum crisis" of the mid-1970s. One may regard the paper [ I ] as a catalyst for enhancing formation of a new trend in photoelectrochemical research and development. In this paper general principles of operation will be briefly outlined for PEC cells, and most important systems surveyed. (For detailed presentation of the semiconductor photoelectrochemistry, see, e.g., [2], and for description of PEC cells at greater length, see [3].) In the last Section, some works on the semiconductor dispersions for purifying wastewaters from pollutants will be reviewed.

Semiconductor/Electrolyte System in Dark and upon Illumination Main features of a semiconductor as an electrode material are:

- a gap in its electronic spectrum (i.e., forbidden energy band); - low current carriers concentration (determined by minor admixtures in the material); - two types of current carriers, namely, electrons in the conduction band and positive holes in the valence band. In the energy diagram (band diagram), Fig. 1a, characteristic energy and electrochemical potential levels are shown for an n-type semiconductor.

--

EV

EV

a

-It ----

5

@

b

Fig. I . Energy diagram of a semiconductor in the dark (a) and in the light (b).

(In what follows, the figures I , 2, 4, 7, 9, 10 being purely illustrative, are given for an n-type semiconductor.) They are: top of the valence band which is filled with electrons (Ev); bottom

419

of the conduction band, i.e., next band practically empty of electrons (E,); electrochemical potential of electrons (called the Fermi level, F) Their interconnections are as follows. Ec - Ev = Eg (Eg = band gap), E c - F = kT In (Nc/no) where Nc is the effective density-of-states in the conduction band, no - the equilibrium concentration of majority carriers (e.g., electrons in the n-type semiconductor), k the Boltzman's constant, T - absolute temperature. The Fermi level is a characteristic of the electronic system of the semiconductor at equilibrium. Upon illumination of the semiconductor, if the light quantum energy hv exceeds the bandgap Eg. the valence band electrons transit onto levels in the conduction band (Fig. Ib), leaving in the valence band the positively charged holes. Thanks to the existence of the bandgap, the interaction of the electronic states in the valence band with those in the conduction band is weakened; therefore, the holes appearing in the valence band as a consequence of the illumination, have quite a time before they recombine with the conduction band electrons. This time is well sufficient for the generated carriers to be transfered across the interphase boundaries, e.g., the semiconductor/electrolyte interface, where the electrochemical reaction thus proceeds Being illuminated the semiconductor is by no means at equilibrium. Nonetheless, under certains conditions both the electron ensemble and the hole ensemble, each one taken separately, may be assumed to be at equilibrium. Therefore, certain "electrochemical potential" is to be assigned to each ensemble separately; these are called "quasi-Fermi levels", Fp and ,Fn. They, naturally, do not coincide with F, the difference depends on how far the non-equilibrium (light) concentration of corresponding current carriers declines from the equilibriun~(dark) value. E.g., for holes the difference equals to: F - Fp = kT In [(po + Ap) / pol where po is the equilibrium hole concentration, Ap - increase in the concentration due to illumination. The interphase boundaries between a semiconductor and a metal or electrolyte solution can also be conveniently represented using an energy diagram (Fig. 2 ) . Here, energy and electrochemical potential levels are shown for all three phases making together the electrochemical cell, namely, the semiconductor photoelectrode, the metal counter-electrode, and the electrolyte solution in between, the latter containing a redox couple where an electrochemical reaction Ox + ne a Red takes place.

420

semiconductor lElectroiytel Metal Q

b

Fig. 2. Energy diagram of a liquid-junction solar cell: a - in the dark, b closing external circuit with the load).

- in the light (upon

These levels are: In the semiconductor, the abovementioned Ec, Ev and F. Note the band bending at the semiconductor surface, caused by the electrode charging. Owing to the low charge carriers concentration, electric field penetrates deeply the semiconductor, producing the so-called space charge layer which usually is ca. 10-6-10-5 cm thick. The near-surface barrier height is e I(aScl.Separation of photogenerated charges takes place in the space charge layer due to the electric field, and thus photocurrent is formed. The potential at which the electric field is absent in the semiconductor electrode (cpfi) is called flat-band potential; it is an analog of a zerocharge potential of a metal electrode. - In the metal electrode, its Fermi level, Fmet, is shown. - In the electrolyte solution, the distribution of the electronic levels filled with electrons (corresponding to the reduced form of the redox couple, Red) and those empty of electrons (corresponding to the oxidized form, Ox) can be also described by the electrochemical potential, Fredox:

-

where F O is a constant, Cox and Cred - concentration of the Ox and Red particles in the solution. Fredox is related to the reversible potential of the redox couple, 90,as Fredox = eq0 + const

42 1

where e is electron charge, and the constant depends on the nature of the solvent and reference electrode against which the cpo value was measured. (E.g., const = 4 43 eV for Normal Hydrogen Electrode in aqueous solutions). To formulate the condition of thermodynamic equilibrium, as well as that of (thermodynamic) possibility for the electrochemical reaction to proceed, either of quantities, F and cp, can be used, The equilibrium condition is F = Fredox, or, which is the same, cp

the the the the

= cpo,

while F < Fredox. or cp > cpo is the condition for the anodic reaction , and

F > Fredox. or cp < cpo is that for the cathodic reaction.

A distinctive feature of semiconductor electrodes, as compared to metal ones, is possibility for both types of charge carriers (conduction band electrons, and holes) to participate in the electrochemical reactions. In other words the charge transfer between the semiconductor electrode and electrolyte solution generally proceeds via both the conduction and valence bands. The photo-generated carriers induce electrode reactions at the semiconductor, i.e., holes induce an anodic reaction, and conduction electrons, a cathodic one, both reactions consuming energy of the light quanta. According to the quasi-thermodynamic approach, the condition for an anodic reaction involving holes to occur, in full analogy to what was written above for the "dark" reaction is now

and that for a cathodic reaction involving electrons is

These conditions are valid for any photoelectrochemical reaction. both for "useful" reactions that underlie the solar energy conversion in the PEC cells, and the deleterious reactions like (photo)corrosion reaction. Indeed, the latter reaction can be also assigned certain value of a reversible electrode potential, cpo dec (or electrochemical potential, Fdec). Hence, anodic decomposition of a semiconductor, with participation of holes is possible (in the dark) when

and cathodic decomposition, involving conduction electrons, when

Here, subscripts "p'l and "n'l denote type of current carriers taking part in the reaction

422

In full analogy with the redox reactions, see above, the photoelectrochemical decomposition (i.e., corrosion under illumination) is thermodynamically possible if Fp < Fdec>P and Fn

'Fdec,n.

A process on the semiconductor electrode is spontaneously set in depending on the comparative values of cpo for the usefir1 reaction, and 'Podec for the corrosion of the electrode material. Therefore, by selecting a combination of solvent, electrolyte, and semiconductor properly one can, at least basically avoid undesirable reaction. Indeed, the thermodynamic approach but defines thermodynamic possibility; while behaviour of a real system may be purely a kinetic one. There are two main types of PEC cells, the regenerative type cells, and the photoelectrolysis cells. The first type also called liquid-junction solar cells, serve like solidstate cells for solar energy conversion into electricity. The same reaction proceeds at both electrodes of the cell, directed forward at the anode, backward at the cathode. Hence, composition of the solution is preserved unchanged. The other type, the photoelectrolysis cells serve for conversion of solar energy into chemical energy. Here, two different reactions proceed at the electrodes. As a consequence, certain substance in the cell is transformed in the course of its operation. These two types of the PEC cells are described in more detail below.

PEC Cells for Solar-to-Electrical Energy Conversion Let us qualitatively describe the operation of the PEC cell. In such a cell a highly reversible redox couple is present in the electrolyte solution. Hence, equilibrium is set in the nonilluminated system at once, both electrodes (the semiconductor photoelectrode and the metal counter-electrode) taking the reversible potential of the couple: cp = cpo, or F = Fredox, i.e., the electrochemical potential levels in all three phases are now equal (see Fig. 2, a). The redox couple is to be selected in such a way that its reversible potential was more positive than the flat-band potential of the (n-type) semiconductor, cpo > cpfi, Therefore, the sign of the electric field within the space charge layer at the semiconductor surface is favourable for charge separation. Upon illumination, electron-hole pairs are generated in the semiconductor. EleLtrons and holes in the space charge layer move in the mutually opposite directions, namely, holes (the minority carriers) are driven to the surface, while electrons drift into the semiconductor bulk (Fig. 2, b). The charge separation results in the decreasing of the initial field in the space charge layer. In other words, band bending decreases. As a consequence of the unbending of the energy bands, other energy and electrochemical potential levels in the semiconductor change their position. Indeed, the difference of the conduction band and Fermi level is preset in the semiconductor bulk (see above), hence the band, unbending "pulls" Fermi level upwards as compared to its position in the non-illuminated sample, see Fig. 2, b. The difference between Fermi level in the dark and upon illumination determines photopotential:

423

Photogeneration of the charge carriers results in formation of quasi-Fermi levels, Fp and Fn in the space where light penetrates the semiconductor. Since Fp Fredox, and Fn E F > Fredox. both the forward and the reverse electrode reactions in the redox couple are accelerated. When external circuit is closed upon the load resistance (R) electrons are transfered to the metal counter-electrode and enter the cathodic reaction there, reducing Ox to Red; and holes cross the semiconductor/electrolyte interface and enter the (photo)anodic reaction, oxidizing Red to Ox. So the light-to-electrical energy conversion takes place; the potential difference across the cell is I h R (where I h is photocurrent), the output power is P Iph2R. In the simplest case where bani edges are “pinned” at the interface, the maximal (i.e,, when bands are fully unbent) open-circuit photopotential is equal in absolute value as can be concluded from Fig. 2 , b, to the potential drop in the space charge region, 10sclor, which is the same, Iqo - qfil. So, to increase the photopotential, hence, the efficiency of the cell one must select the semiconductor and solution composition in such a way as to make the above potential difference as big as possible. A good example of the regenerative cell is the n-GaAdalkaline solution of Se2- + SeZ2/metal cathode cell designed over 15 years ago yet retaining its importance. In the dark equilibrium is set in the cell se,Z + 2e-

22 Se

and both electrodes take the potential of the selenium couple. Hence, Fermi level of GaAs (F) and metal cat9:de are equal to the electrochemical potential of the redox solution, Fredox = Fse /Se2 . In the light, photogenerated holes move to the GaAs/solution interface where they oxidize Se2 ions to Se22 Photogenerated electrons move through the GaAs bulk to the ohmic contact at the electrode rear side, then through the external circuit (including the load resistance) to the cathode. There they reduce Se: to Se2 The PEC cell is designed as a thin-layer cell (Fig, 3) and this enables intensive mass transfer in the space between the electrodes, thus preserving the solution composition constant.

gmet)

,

,

current leads

o/ 0‘ hv t

Substrate

‘Glass ‘Sn 0,‘Electr

/

Sem conductor

Housing

Fig. 3

Thin-layer liquid-junction solar cell.

424

In its action, the regenerative type PEC cell is a full analog of the solid-state solar cell based on the semiconductodmetal junction called Schottky diode (see, e.g., [4]). Band diagram of the Schottky diode both in the dark and in the light is presented in Fig. 4, a and b respectively. In the dark, Fermi levels of the semiconductor and the metal are equal (cf. Fig. 2, a), F = Fmet, Upon illumination of the semiconductor, a photopotential emerges in it, pph. 0t

Semiconductor

Semiconductor R

Metal

b

Fig. 4. Energy diagram of a solid-state solar cell of the Schottky diode type; a - in the dark, b - in the light (upon closing external circuit with the load). We shall now briefly discuss current-voltage characteristics of PEC cells. For an ideal Schottky diode it is [4]:

I = I, [exp (eV / kT) - I ] - Iph where I, is the limiting current of the diode biased in reverse direction, V is external voltage. Under open circuit (i.e., when I = 0) photopotential is pPho.C. = (kT / e ) In [(Iph / lo) + 11 A photocurrent of a semiconductor electrode (both in solid-state and liquid junction cell) is composed of the flow of minority carriers generated within the space charge region (its thickness is Lsc) and those generated beyond the space charge region, that is, in the semiconductor bulk at a distance of the diffusion length, Lp. (Diffusion length is the average

425 distance covered by diffusing non-equilibrium current carriers for their lifetime, T i.e., before ?' they recombine; Lp = (Dp~p)1/2where Dp is the diffusion coefficient of the particles.) In the simplest case when all the rmnority carriers reaching the electrode surface are consumed by the electrode reaction (no surface recombination) the photocurrent is [S]: Zph

=e

J [ I - exp(- oc LSc) 1 l+ccLp

where a is light absorption coefficient of the semiconductor, J is flux of light quanta absorbed in the semiconductor. One can easily see by comparing Figs. 2 and 4 that in the solid-state cell photogenerated current carriers pass directly From semiconductor into metal while in the PEC cell this process is replaced by two electrochemical reactions at the cell electrodes, the forward one and the reverse one. (Ultimatelly, the charges from the photoanode come to the metal cathode via electrolyte solution altogether.) The replacement results, first, in increased efficiency because no light is lost due to the absorption by the metal film although thin on top of semiconductor, as in the Schottky diode. Second, the semiconductor/liquid electrolyte solution interface is notable for its very low surface recombination velocity as compared to solid-state systems (problably because of absence of the mechanical stress caused by misrnath of two solid crystal lattices). This makes it possible to use less perfect, hence, cheaper semiconductor materials. On the other hand, direct chemical interaction of a semiconductor with an electrolyte solution raises more problems. Among them most important is (photo) corrosion of semiconductor electrodes that may restrict service life of PEC cells drastically. Photocorrosion can be eliminated, at least partially by introducing a well-reversible redox couple into the electrolyte solution so far as, e.g., reduced form in the couple is more readily oxidizable than the semiconductor electrode material. Hence (see above) it well competes with the photocorrosion reaction for photogenerated holes. Many inorganic and organic redox couples were successfully tested in PEC cells, including ferrocene and other metallocenes [ 6 ] . Also, non-aqueous solvents were widely adopted for PEC cells since they are less corrosive toward many semiconductor compounds. This made it possible to widely use optimal band-gap but less stable semiconductors as electrodes of PEC cells - among them are GaAs, CdTe, CdSe, MoSe2, Si etc. The menace of photocorrosion compels to restrict efficiency of the regenerative PEC cells. In fact, to increase efficiency (through the increase in photopotential) one has to increase the initial (i,e,, preexisting in the dark) band bending in the semiconductor. For this reason, the reversible potential cpo of the redox couple in the cell with, say, n-type photoanode should be as positive as possible. Yet, at the same time it should not exceed the decomposition potential for the semiconductor, @dec,p. Thus, one is forced to deliberately diminish the cell photopotential. As a result, the cell efficiency and stability vary with the solution redox potential in an opposed way. In reality the cell characteristics always are a result of a compromise between the requirements of efficiency and stability. (Therefore, one should take cautiously higher values of efficiency and service life of PEC cells given in published papers: they might be measured not for the same cell.) It is the necessity to prevent photocorrosion that initiated increased interest in photocathodes during last decade since cathodes operate in reductive conditions and therefore

426

are less inclined to corrosion. (This holds for both regenerative and photoelectrolysis cells discussed below.) Before describing particular systems of PEC cells, let us formulate their main quantitative operating characteristics. The cell efficiency q is the ratio of the electric output power to the luminous input power, PI. The former is a product of photocurrent and photovoltage, Iph. Vph, and reaches its maximum value at a certain load resistance, R. Thus, rl = (Iph Vph)rnaxpl,

The fill factor is defined as the ratio of the above product (Iph Vph)max, to the product of the cell short-circuit current and open-circuit voltage: f = (Iph . Vph)max/(IphS'C'. Vpho"'). As one can see in Fig. 5, it is the "quality" of the cell operating characteristic that is appraised with the fill factor. Namely, the closer the curve to a "rectangular" one, the closer is (Iph)max to IphS.C.,and (V h)max to Vpho.C., so, the higher is f In other words, f is high when ohmic losses and overvoytages in the cell are low. Fill factor is as high as 0.75 in the most efficient solar cells both PEC and solid-state.

Fig. 5. Output power characteristics of a liquid-junction solar cell with n-CuInSe2 photoanode and KI-KI3-CuI-HI electrolyte under 100 mW/cm2 solar simulated irradiation [7]. Quantum yield is the ratio of the electron flux in the external circuit to the photon flux at the cell inlet. (Quantum yield can be defined either per incident photon, or per absorbed photon). In case of poor charge separation, or strong electron-hole recombination in a semiconductor, quantum yield is small.

427

In the Table data for most effective cells are summarized. The most effective semiconductor materials are' cadmium chalcogenides, and n-GaAs in polychalcogenide solutions, and transition metal dichalcogenides in iodine-iodide solutions as photoanodes; p-InP as a photocathode. A conclusion may be drawn from the Table that polycrystalline electrodes (e g., sintered tablets and evaporated films with grain size of microns) are nearly as effective as single-crystalline ones. This enables one to look forward to designing inexpensive and easy-to-manufacture PEC solar cells. With some types of cells corrosion was practically abolished, and service life up to one year was achieved for outdoor operation in the mid-Europe climat [20]. Yet certain degradation processes other than corrosion (e.g., ion exchange in the outmost layer of the semiconductor electrode [9]) restrict the service life of PEC cells which still are inferior to the solid-state solar cells whose service life comes to 10-20 years. To prevent photocorrosion and improve the cell efficiency one has to selectively increase rate of the "usefil" electrochemical reaction in the redox couple at the semiconductor electrode as compared to side reactions (including the electrode decomposition, and current carriers recombination) competing for the photogenerated carriers. To achieve this aim, several methods have been worked out; (i) Deposition of minor amount of a catalyst, e.g., noble metals (Pt, Rh, etc) or ruthenium dioxide onto the photoelectrode surface [9]. Being deposited as small "islands" (or clusters), the metals do not shade the semiconductor but accelerate the oxidation (reduction) reaction of the reduced (oxidized) form of the redox couple. For example, at the n-type silicon electrode covered with ultrathin layer of platinum (on top of equally thin "tunnel" oxide interlayer) very fast photooxidation of bromide or iodide ions proceeds without electrode degradation [ 151 (ii) Treatment of electrodes with solutions of Ru, 0 s . Ag, Pb ions, followed with mild oxidation of the semiconductor surface. Adsorption of these ions (as well as some organic molecules) hampers surface recombination and/or recombination at the intergrain boundaries in the polycrystalline samples [9, 22, 231. Incidentally, special "texturizing" etching of the surface [8, 91 is used to make surface "matt" and thus reduce its reflectivity. (iii) Surface modification of electrodes with charge carriers [24]. Such a catalyst usually is a highly reversible redox couple attached to the surface atoms of the semiconductor by chemical or adsorption bond, e.g., ferrocene-ferricenium couple on silicon [25]. Note that the "surface-attached" molecules take part in the electrode reaction with lesser overvoltage as compared with the dissolved molecules. This results in increasing the part of "useful" reaction in the total photocurrent. while the part of photocorrosion is decreasing. Moreover, silicon electrode being surfacely derivatized with ferrocene is capable to operate in aqueous solutions inspite the fact that ferrocene (and its derivatives) is insoluble in water. Thin films of conductive polymers like polypyrrole. polyaniline, etc. are also used for the surface modification of the semiconductor electrodes [26, 271. Their performance mechanism has not yet been deciphered. Most likely, the coating is a combination of a protective film and the charge carrier, the more so that reversible redox couples are used to be introduced into films [28], as well as catalytically-active admixtures like Ru02 [29]. Such films were used to stabilize a promising "solar" electrode material, amorphous silicon [30].

428 Efficiency of PEC cells for solar-to-electrical energy conversion

Table

Photelectrode

Electrolyte solution Aqueous solutions

n-GaAs

(singlecrystalline) KzSe-K2Se2-KOH

n-GaAs

(polycrystalline)

n-CdSe

Efficiency *, per cent

Reference

16

8

7.8

9

(singlecrystalline) Na2 S-Na2 S2-NaOH

7.5

10

n-CdSe

(polycrystalline)

5.1

10

n-CdSe

(singlecrystalhe) KjFe(CN)6-K4Fe(CN)6

16.4

33

n-Cd(Se,Te)

(singlecrystalline) K2Se-KzSe2-KOH

12.7

11

p-InP

(singlecrystalline) VC13-VC12-HCI

11.5

9

p-InP

(polycrystalline)

7

12

n-CuInSe2

(singlecrystalline) KI-KI3-HI

12

7

n-CuInSe2

(polycrystalline)

12

13

n-WSe2

(singlecrystalline) KI-KI3

14

14

n-Si

(singlecrystalline) KBr-KBr3

13

15

KzSe-K2Se2-KOH

NazS-Na2Sz-NaOH

VC13-VC12-HCI

KI-KI3-HI

16

n-Si Ferrocene-ferricenium (acetonitrile solution)

n-Ga(As, P)

(epitaxial)

n-Si

Ferrocene-ferricenium (singlecvstalline) (methanolic solution)

n-CdS

(singlecrystalline) KI-KI3 (acetonitrile solution)

13.2

17

14

18

9.5

19

* / Natural or simulated solar light The temperature dependence of the cell characteristics is vital problem with the cells for the outdoor application. With the increase in temperature the cell photocurrent is increasing while photopotential is falling down (Fig. 6). The product, hence, the cell efficiency is maximum at certain temperature. For example, the PEC cell with Cd(Se,Te)-photoanode and polysulphide electrolyte attains its maximum efficiency at ca. 55 "C [20]. On the whole,

429 variations in the ambient temperature do no harm to the cell performance. Passive cooling of the cell provides for spontaneously setting in favourable temperature conditions.

12h I,, h, m A/cm

8 4

0

0.2

0.4

0.6yppph,Y

Fig. 6. Output power characteristics of a liquid-junction solar cell with n- CdSe0,75Te0 25 photoanode (4.5 cm 2) and alkaline polysulphide electrolyte under 80 mW/cm2 solar illumination intensity. 1 - 28 "C, 2 - 48.5 "C. 3 - 61.5 "C [20]. The above discussion concerned semiconductor electrodes with optimal bandgap ( 1 . 1 - 1.5 eV) effectively absorbing solar light. Another possibility to create stable and efficient photoelectrodes for PEC cells is to sensitize wider bandgap semiconductors with adsorbed dyes which readily absorb visible light. The light-excited molecules of the dye exchange charges with the semiconductor. This may seem very attractive because wide bandgap semiconductors, e.g., metal oxides although insensitive to visible light, proved to be very stable against photocorrosion. The most recent example is the PEC cell with a sensitized Ti02 photoanode, transparent counter-electrode (a "conductive" glass promoted with deposited platinum) and a non-aqueous redox electrolyte [3 I]. According to a generally adopted view the dye-sensitizers cannot provide for large quantum yield because of poor light absorption in a thin layer of adsorbed dye. However, authors of [31] have overcome the difficulty by making use of highly dispersive Ti02 samples whose true surface area exceeds the geometrical area by 2 to 3 orders of magnitude. Thanks to multiple crossing the dye adsorption layers in the pores of the Ti02 film, light was entirely absorbed, and quantum yield of the photocurrent approached 100%. The cell efficiency was reported as 8.9% which seems to be not the final value. A propylenecarbonate solution of a lithium salt was used as electrolyte, and 2.2'-bipyridine-4,4'-dicarboxyliccomplex of Ru(II1) served as a sensitizer, distinguished with its strong visible light absorption, fast electrode kinetics, and good stability. The latter provides for long service life of the cell, approaching 9 months. Titanium dioxide is inexpensive and very stable as a photoanode. "Colloidal" Ti02 was deposited onto a substrate as a thin film by the "sol-gel" method. Ti02 particles are in good electrical contact with each other and with the substrate. There is good mass transfer in the electrolyte solution impregnating the film due to its small thickness. Another version of the Ti02 electrode was also elaborated for the same purpose [32].

430

Namely a thin film of polycrystalline Ti02 was deposited onto a substrate by thermal decomposition of titanium alcoholate solution. During the heat treatment the film was cracking and numerous pores and crevices were formed in it producing high surface roughness. Efticiency of the PEC cell with the above mentioned sensitizer and the Br- - Br3- redox couple amounted at 4%. On the whole, best regenerative-type PEC cells have fill factor of 0.55 to 0.6, open-circuit photovoltage 0.6 - 0.8 V (and even as high' as I . 2 V [33]) and photocurrent 25-30 mAlcm2 for a non-concentrated sunlight (AM 1). Is it now possible to evaluate the prospects of the liquid-junction solar cells? To do so, one must have in mind that they are approaching commercial solid-state solar cells in efficiency though are much inferior as to their service live. Their potentials seem not to be exhausted yet, however, since their life story is much shorter than that of solid-state cells.

PEC Cells for Solar-to-Chemical Energy Conversion Coming back to the cell operation mechanism (cf. Fig. 2) let us assume that the catode reaction is no more a reverse reaction to that proceeding on the photoanode. On the contrary, let two different reactions take place at the cell electrodes. This is the case, basically, when none of the redox couples present in the cell is reversible at the semiconductor electrode. Then. equilibrium does not set in across the cell, the electrodes remain ideally polarizable, their stationary potentials being determined, say, by oxygen or traces of fortuitous impurities present in the solution. An important point is that a space charge layer is to be formed in the near-surface region of the semiconductor, the sign of the electric field therein being favourable for the separation of the photogenerated charges. The case may be exemplified by an oxide semiconductor electrode in an indifferent electrolyte solution. Band diagram of the cell with the aqueous electrolyte solution is shown in Fig. 7.Now, two electrochemical potential levels in the solution are the characteristics of the two reactions of water transformation: its oxidation to oxygen, and reduction to hydrogen:

F H ~ / H ~ ofor 2 H20 + 2 e-

+ H2 + 2 OH-

where h+ denotes a positive hole. In the dark, Fermi levels of the semiconductor anode and metal cathode take arbitrary positions relative to the above levels (and to each other when the circuit is not closed). Upon illumination of the semiconductor, electron - hole pairs are generated therein, electrons and holes are separated in the space charge layer, as discussed in the preceding Section. Energy bands partly unbending, the Fermi level in the semiconductor is raised up, along with the quasi-Fermi level for electrons, Fn. Accordingly, quasi-Fermi level for holes, Fp, is coming down, the lower the more intensive is the hole generation rate. For certain light intensity the hole quasi-level at the electrode surface reaches the water oxidation level F ~ ~ o / while o ~ , electron quasi-level reaches the water reduction level, F H ~ / H0 (Note that this reaction is transfered onto the counter-electrode of the cell as the circuit %as been closed.) The two reactions proceed simultaneously at the cell electrodes,

43 1

stimulated with electrons and holes generated in the semiconductor by light. These partial reactions in the aggregate make up the process of water decomposition into hydrogen and oxygen. These products thus accumulate energy of light absorbed by the semiconductor.

0-t

1

I

Semiconductor Electrolyte Metal a

b

Fig. 7 . Energy diagram of a water photoelectrolysis cell: a - in the dark, b - in the light

In the case presented in Fig. 7 the conduction band is higher than the F H ~ / H ~level o in the energy scale. Therefore, conduction electrons are sufficiently rich in energy as to reduce water at the cathode, the reaction proceeds spontaneously upon illuminating the semiconductor. The O n-type photoanode; accordingly, for above situation is possible when 'phn< ' ~ O H ~ / H ~ for holes in the p-type photocathode to be able to oxidize water, the condition is 'pfiP > cpo If the above conditions do not hold, then, spontaneous decomposition of water is not possible. To decompose water, it is therefore necessary to apply external voltage to the illuminated cell, and so shift energy levels in the semiconductor relative to those in water. If the bandgap, Eg, is not sufficiently large as compared to the Gibbs energy, AG, of the reaction taking place in the cell, then, again it is necessary to apply external voltage to the cell in order to compensate the lacking energy, or else the reaction cannot proceed spontaneously upon illumination. This kind of process is called "photoassisted electrolysis". The application of the external voltage should be made alowance for in calculating the cell efficiency, namely:

where the Gibbs energy for the cell reaction, AG is given per 1 electron. If external voltage exceeds AG/e value, then q is negative, that is, no energy storage occurs in the cell. It should be stressed again that reaction other than water splitting can proceed in the cell at the same time, thus, decreasing the efficiency and, in the case of abovementioned corrosion

432

degrading the semiconductor photoelectrode. Indeed, unlike regenerative type cells, the photoelectrolysis cell contains no reversible redox couple which protects semiconductor. Therefore, it is of primary importance to select properly the electrode material and the solution composition and so provide for good stability, either thermodynamic or kinetic, of the semiconductor electrode in the cell. As far as cells for water splitting are concerned, semiconducting oxides were widely used as material for stable photoanodes because, being higher oxides, they well withstand further oxidation and decomposition in aqueous solutions at strong anodic potentials. The survey of the oxide photoanodes data is given, e.g., in [3]. Among them, the singlecrystalline SrTi03 anode demonstrated highest activity. The most popular oxide, a Ti02 (rutile) anode is as highly stable, yet it is inferior to SrTiOj because its flat-band potential is slightly more positive than the reversible potential of the hydrogen electrode. Therefore, hydrogen evolution on the cathode of the cell, and spontaneous water photo-decomposition is impossible other than with application of external voltage, which badly reduces the cell efficiency. At the same time, it was rutile that exemplified a polycrystalline semiconductor photoelectrode to be but slightly inferior to a singlecrystalline one. Indeed, for a sintered Ti02 anode the quantum yield of a photocurrent is as high as 0.75 as compared to ca. 1.0 for a singlecrystalline electrode [34]. However, oxide anodes like SrTiO3 and Ti02 (including also BaTi03, KTa03 and others) are wide bandgap semiconductors (Eg 2 3 eV), hence, they are sensitive to the UV light, and not to visible (solar) light. Efficiency of solar photoelectrolysis of water does not exceed 1-2%. Narrower bandgap oxides like WO3, Fe2O3, In203 etc. (Eg E 2 eV) being sensitive to visible light, still are not much suitable for water photoelectrolysis because their conduction band bottom, like that of Ti02 lies too low (in other words, flat-band potential is less negative than reversible hydrogen electrode potential) and, hence, application of the external voltage is required. Attempts have been made to sensitize the wide bandgap oxides to visible light by doping them with light-absorbing impurities (like Cr in Ti02). Thus, the very mechanism of nonequilibrium current carriers generation has been changed, since intrinsic light absorption in a semiconductor has now replaced by extrinsic (impurity) absorption of light. However, the impurity absorption is very poor (quantum yield of photocurrent proved to be of the order of few per cent), so no gain in efficiency has been obtained. Another trick was used [35, 361 with the object of providing for stable (oxide) anodes with properly arranged energy levels. For this purpose mixed oxides with two cathionic sub-lattices were synthesized, e.g., Cr2Ti207. In their band diagram the lower valence band was formed by the oxygen 2p-orbitals (which is in the usual run of things with oxides), the conduction band is formed with metal orbitals of 1st cathion (Ti), while the second (upper) valence band was formed with cathionic d- of f-orbitals of 2nd cathion (Cr). Energy of electron transition from this additional valence band to the conduction band is much less than the original bandgap of the "matrix" oxide (TiOz), e.g., 2 eV, hence, the mixed oxide is sensitive to visible light. Holes are produced in the additional valence band upon illumination of the electrode. At the same time, relatively high energy position of the conduction band edge, characteristic of wide bandgap oxides is preserved. Examination of a number of mixed oxides so prepared showed that the threshold energy for the photocurrent is reasonably low, and the flat-band potential is more negative than the hydrogen electrode potential, indeed. This makes it possible, at least basically to run

433

photoelectrolysis of water without application of the external voltage. However, quantum yield attained is still very low, probably owing to poor crystallinity of the materials, which resulted in intensive recombination. The above discussion was concerned with some drawbacks of particular classes of semiconductor photoelectrodes. Moreover, there exists a general problem of matching the electrolysis energetics with that of solar light spectrum Namely, main part of energy in solar radiation is composed by IR and visible light quanta, so optimal threshold energy, i.e.. semiconductor bandgap is 1.1 - 1 . 5 eV. This value somewhat exceeds the Gibbs energy for water decomposition reaction (AG = F H ~ / H ~ o - F H ~ o /=o ~1.23 eV per one electron). which makes it possible, basically, to run the process. In practice, however, energy losses in the cell shown in Fig. 7 amount to about I eV. Hence, quantum energy over 2 eV is necessary for spontaneous water photosplitting to hydrogen and oxygen. In other words, energy of the majority of the sunlight quanta is insufficient to split water in the single-quantum photoelectrochemical process. To overcome this difficulty, several ways were proposed: (i) Photoassisted electrolysis. Semiconductors with optimal bandgap are selected as electrode materials, and the "lacking" energy (up to ca. 2 eV, see above) is taken from an external battery. (ii) Two-quantum photoelectrolysis. When two photoelectrodes are used in the cell (an n-type semiconductor anode, and a p-type semiconductor cathode) their photovoltages are summing up. The cell photovoltage so produced is likely to be sufficient to split water. A hardly solved problem of the proper matching of the photoelectrodes characteristics arises, however. The other way is to split the process into two stages and to carry out each stage in a separate PEC cell. These "partial" cells are combined together into a plant using some redox system whose components (Ox and Red) serve as charge carriers between the cells; they are not consumed in the course of the plant operation. Thus the redox couple makes the two chemical potentials emerging in the "partial" cells to be "connected in series" like two electrical potentials. The plant simulates the well-known 2-scheme of natural photosynthesis with its two photosystems. In both cases described above two photoelectrodes are made use of, so the whole process is a two-quantum photoelectrolysis, which enables one to have got sufficient energy to split water. (iii) Two-stage conversion. Here the whole process is splitted into two consecutive stages: electricity production followed by electrolysis proper. Instead of a single unit (PEC cell), here we have two separate units: a (solid-state) solar cell and an electrolyzer. Accordingly, two different functions which are combined in a PEC cell, namely photovoltage production and water decomposition are now distributed between two specialized units. This makes it possible to attain the desired voltage by mere connecting solar cells in series. Also, a problem of combining good photoelectrical and electrocatalytic properties, and stability against corrosion in the same material is removed. For detailed discussion, see, e.g., [37] and the next chapter of this Volume. (iv) Dehydration of substances other than water. As hydrogen is the sole desired product, one can substitute certain process with lesser Gibbs energy for water electrolysis, e.g., dehydration of some organic compounds, hydrogen sulphide, hydrobromic acid, and others. These processes have been thoroughly studied during last decade.

434 Progress in research and development of the photoassisted electrolysis cells was caused by wider use of p-type photocathodes. As already mentioned cathodes are less subject to corrosion. Therefore, optimal bandgap semiconductors have been made use of, e.g., InP (Eg = 1.35 eV), Si (1.09 eV) and others. To increase their electrocatalytic activity toward hydrogen evolution, special surface treatment has been applied, namely deposition of noble metals (Rh, Ru, Pt) and Re as small islands (compare with the abovementioned surface treatment of the regenerative type PEC cell electrodes) [38]. These deposits acted as catalyst: they reduced the hydrogen overvoltage. Indeed, photoactivity of the electrode so prepared directly reflected electrocatalytic activity of metals used as catalysts, toward cathodic evolution of hydrogen. (Precisely, the gain in the silicon cathode photopotential caused by metallic deposits is a linear hnction of the logarithm of the exchange current for the hydrogen electrode reaction measured at these metals as electrodes [39]). Charge separation proceeds effectively in the InP-to-the noble metal juntion owing to the fact that the potential barrier height therein attains optimal magnitude upon absorption of the evolved hydrogen. Mild oxidation crowns the modification of the p-InP electrode surface: ultrathin oxide film favours reduction in the surface recombination velocity. Comparison of the current-voltage characteristics for the InP photocathode so prepared, and platinum (dark) cathode is given in Fig. 8. Although platinum is very effective as a cathode in electrolyzing water, the same process proceeds with even lesser overvoltage (to be precise, it is an underpotential hydrogen evolution) at the illuminated InP cathode. The gain in potential (i.e., the underpotential magnitude) given rise owing to the luminous energy consumption is as high as 0.6 V [38], and the cell efficiency is 6-8%.

NHE

I,,,,,mA/cm2 Fig 8. Current-voltage curves for p-InP cathode in the light (1) and Pt cathode in the dark (2) in 1 M HClO4 solution under 79 mW/cm2 solar illumination. Hatched area is proportional to the converted energy [38].

435

P-type silicon photocathodes promoted with the platinum deposit were also reported as very effective in the photo-splitting of water [40]. As to photoanodes, pyrites (FeS2) is thought [41] to be a very promising material due to its outstanding light absorption coefficient (over 105 cm-I) So, the FeS2 film as thin as 0 . I p absorbs solar radiation completely. Pyrites is very stable against corrosion in aqueous solutions. Unfortunately, photovoltage attained so far at the FeS2 photanodes (ca. 0.2 V) was insufficient to provide for efficient electrolysis, probably due to high density of crystal defects and impurities in the material used; this raised the problem of the FeS2 preparation refinement. Since protection of electrodes against corrosion in the photoelectrolysis cells is a question of vital importance, many attempts have been made to use protective films of different nature (metals, conductive polymers, or stable semiconductors, e.g., oxides). Of these, semiconductive films are less effective since they often cause deterioration in the characteristics of the electrode to be protected (laying aside heterqjunction photoelectrodes specially formed with semiconducting layers of different nature [42]). When metals are used as continuous protecting film (and not catalytical "islands" discussed above), a Schottky barrier is formed at the metal/semiconductor interface. The other interface, i.e., metal/electrolyte solution is as if connected in series to the former and is feeded with photocurrent produced in the Schottky diode upon illuminating the semiconductor (through the metal film). So, the structure under discussion is but a combination of the "solar cell" and "electrolyzer" within the photoelectrode. Unfortunately, light is partly lost due to absorption by the metal film. Progress in development of protective coatings for semiconductor photoelectrodes was bound with making use of materials other than metals, with "metallic" conductivity, namely, silicides (of platinum, or ruthenium), or degenerate semiconductors (Sn02: Sb. SnO2 - In2O3; BP, and others) [43 - 451. These coatings also provide for fast charge transfer between illuminated semiconductor and electrolyte solution. But, unlike metals they are transparent for visible light and do not attenuate the incident luminous flux. Very effective is the mixed oxide Ti02 - RuO2 (well known as an active material for the so-called dimensionally stable anodes, DSA for the electrolytic production of chlorine). Its stability is due to the Ti02 content, and catalytic activity, to that of Ru02. E.g., in the cell with n-silicon electrode protected with the Ti02 Ru02 film efficiency over 6% has been achieved for the electrolysis of the HCI solution [46]. We shall conclude the Section with brief description of the pilot plant for solar energy conversion and storage recently designed by Texas Instruments (USA). Hydrogen is produced in the course of photoelectrolysis of aqueous HBr solution. The plant is schematically pictured in Fig. 9. The PEC cell proper contains two photoelectrodes, namely an n-type silicon anode and a p-type silicon cathode. Each electrode is composed of a great number of small spheres (ca. 0.4 mm in diameter) casted of silicon according to a special manufacturing process (not yet disclosed) and embedded into a glass panel. The spheres' surface is coated with protecting film of platinum metals (Pt, Ir). The photopotentials of photoanode and photocathode are summing up, and the resulted voltage is sufficient to split HBr into H2 and Br2. The photoelectrolysis products are stored in tanks, and can be later converted to electricity in a fuel cell, The "sun-to-hydrogen'' efficiency was reported as 8.6%, the service life is 20 years (estimated) [46-48]. Thus, efficiency of the photoelectrolysis cells for hydrogen production approaches lo%, and the service life of protected semiconductor electrodes has been reported as many months

436 and, sometimes, several years.

5

0' Fig. 9. Scheme of the Texas Instruments plant with the HBr photodecomposition. A - PEC cell proper: 1 - cathode compartment, 2 - anode compartment, 3 -diaphragm, 4 - optical window, 5 - silicon spherical electrodes (dark regions denote n-type, white regions denote p-type), 6 - glass panel (the electrodes' holder), 7 - current conducting bus, 8 - protective film. B - storage tanks for hydrogen (9) and bromine (10). C - fuel cell: 1 1 - recovery of bromine-free electrolyte, 12 - membrane, 13 - the hydrogen electrode, 14 - the bromine electrode, 15 - excess H2 outlet, 16 - external load. Semiconductor Dispersions for Providing Cleaner Environments

In recent years great interest has aroused in photoelectrochemical behaviour of the semiconductor dispersions, e.g., suspensions, colloids, small particles embedded in membranes and vesicles, etc. Unlike PEC cells, they do not provide for special separation of products of the photoelectrochemical process. However, the microheterogeneous systems are superior in electrocatalytic activity (due to their giant surface area). Semiconductor particles suspended in a solution used to be thought of as microgalvanic couples on whose "electrodes" partial electrochemical reactions proceed. Indeed,

437

photoelectrochemical behaviour of suspensions and colloids can be well described this way (with few exceptions). A semiconductor particle serves as photosensitizer for a chemical process since light generates electrons and holes therein, and these non-equilibrium current carriers stimulate partial electrochemical reactions on the particle surface. To make the surface catalytically active toward these reactions special technique was used for deposition of catalyst onto the particle surface. For example, in the process of water photodecomposition with Ti02 dispersions each Ti02 particle was provided with two catalysts (see Fig. 10 a), one accelerating anodic oxidation of water to oxygen, the other accelerating its cathodic reduction to hydrogen. These catalysts may be Ru02 and Pt respectively [49], they operate as microelectrodes of the galvanic couple.

$eterof unction S S2

p - Cu,S a

n- ZnyCd,_yS b

Fig. 10. Band diagram of an illuminated semiconductor particle in the electrolyte solution. a - a homogeneous particle: CAT 1 and CAT 2 are catalysts for H2 and 0 2 evolution respectively. b - a heterojunction particle: CAT is catalyst for H2 evolution. Another vital problem is charge separation within the semiconductor particle [49, 501. if the particle can be regarded as "large", i.e., its dimensions well exceed the Debye screening length in the semiconductor, then a space charge layer is formed in the particle near its surface, and the charge separation in the electric field proceeds in the same manner as in the macroscopic electrode of the PEC cell discussed above. If, on the contrary, the particle dimensions are but few nanometers as is the case with colloidal solutions, no space charge is formed in the particle and the charge separation proceeds on the surface proper due to the selective activity of different spots to reactions consuming different kinds of current carriers. Hence, these spots (catalysts) act as "drain" either for electrons, or for holes. A peculiar case is a particle with an inner heterojunction, i.e.. contact of two different semiconductors (Fig. lob). When the particle is composed of materials with different bandgap (see Figure) its ability to absorb sunlight is superior to that of the homogeneous particle (Fig. 10 a), as now both short-wave and long-wave light is equally well absorbed by wider gap and narrower gap semiconductor respectively. Additional gain in the "particle efficiency" is brought in by charge separation in the heterojunction area. Milling is a common technique for preparation of suspensions, while colloids are used to be directly synthesized in the solution. Complicated systems, e.g., containing heterojunction. or inclusions of a catalyst are also prepared by chemical synthesis. Sometimes the semiconductor

438 particles are deposited onto inert carriers (like Si02, montmorillonite, etc.) or immobilized by embedding them into polymer membranes. These techniques are described at length in [5 1, 521 along with properties of so prepared microheterogeneous systems. Already at an early stage of the research in the semiconductor dispersions, attempts have been made to carry out water splitting, C 0 2 reduction, etc., in other words, the same photoelectrochemical processes as in the macroscopic PEC cells. The results obtained are summarized in [5 1-55]. We shall confine ourselves, however, to the processes that might underly some methods of purification, e.g., of waste waters etc. These processes are stimulated by electrons and holes produced in the particles by light. As only one type of the current carriers is consumed in the "usehl" reaction, measures should be taken to remove the other type from the particle in order to preserve its electroneutrality and sustain the process. For this purpose a sacrificial electron donor (or acceptor) is to be introduced into the electrolyte solution. Often it is the solvent that plays sacrifice, Some examples are listed below. Recovery of metals from their salt solutions. When aqueous suspension of Ti02 (or else, ZnO, CdS, SrTiOj) particles is illuminated in a metal salt solution, the metal is deposited on the particles' surface, e.g., according to the schemes

2Ag'

+(CH3)2CHOH

-+ 2Ag0 +

(CH3)2CO+2HS

Water and isopropyl alcohol serve as electron donors in the 1st and 2nd reaction, respectively. In this way traces of gold [56], silver [57], mercury [58], platinum [59] and other metals can be removed from solutions. Same procedure was used to purposely deposit noble metal "islands" as catalyst onto semiconductor particles (see above), e.g., to prepare platinized Ti02 suspensions by illuminating Ti02 particles in the H2PtCl6 solution. Destruction of inorganic compounds. Nitrite [60] and cyanide [61] ions are oxidized on platinized Ti02 suspension in the presence of dissolved oxygen. The former reaction has been proposed for removal of nitrites from the rainclouds (in order to prevent fall of acid rain). Destruction of organic compounds. Many deleterious organic compounds undergo deep photooxidation (to C 0 2 and H20) at the illuminated platinized Ti02 suspension, among them are: phenol, chlorophenol [62], trichlorophenol [63], chlorobenzene [64], nitrobenzene, chlorophorm [65], some surfactants [66], the dioxin and triazin based herbicides [67] - this list is by no means exhaustive. Figure 1 1 shows that the Ti02 suspension rapidly removes 2.7-dichlorodibenzodioxin from solution. When oxidizing species like periodate or peroxydisulfate is present at a concentration of 0.1 M, the degradation rate remarkably increases and more than 99% of dioxin has disappeared after 30 and 60 minutes, respectively. Splitting of hydrogen sulphide into hydrogen and sulphur. This process helps to solve two problems at one blow, namely to purify waters and produce hydrogen. Most effective as "photocatalysts" are suspensions of CdS [68] and those with heterojunction, e.g., CdS/CuxS [69] and (Zn,Cd)S/CuxS [70], all provided with catalytically active inclusions of Pt and Ru02.

439

When sulphite ions have been added to solution, thiosulphate is produced, in addition to hydrogen, in accord with the equation: so3'+

SO

+ ~20~'.

With this process, the system is freed from the solid sulphur precipitate.

M) with Fig. 11. Kinetics of degradation of 2,7-diclorodibenzodioxin(4 ppm = 1.56 x Ti02 dispersion (0.5 gA) and oxidizing species: 1 - none, 2 - 0.1 M S20$-, 3 - 0.1 MlO4- [67].

Kinetics of hydrogen photoevolution from the H2S saturated aqueous CdSRu02 suspension is shown in Fig. 12. Quantum yield of the process is as high as 0.35 [68] which approaches theoretical limit of 0.5."Commercial" efficiency (that is, calculated upon the Gibbs energy for hydrogen combustion in the oxygen ambient, i.e., taking no allowance for the Gibbs energy of the initial H2S [71]) is 2 to 3%. This value is large enough to encourage designing a pilot plant with a photo-reactor ofthe fluidized-bed type [55].

Concluding remarks With good reason one can believe that solar energy will occupy its niche in the energetics of the future, and the nowaday intensive research activities will be crowned with development of inexpensive and efficient PEC cells capable to compete with the traditional sources of energy, on the one hand, and other than electrochemical methods of solar energy conversion, on the other hand. This aim may well be achieved at the end of 20 century (although certain particular types of PEC cells might find an application even before this term). Of different types of the PEC cells more promising seem the liquid-junction solar cells with polycrystalline

440

semiconductor electrodes, photoassisted electrolysis cells with electrodes coated with protective films, and microdispersions for either production of chemicals at the expense of solar energy, or purification of sewage and so forth.

Fig. 12. Hydrogen generation from visible light ( h > 415 nm) by illumination of CdS dispersions (25 mg CdS loaded with 1% Ru02 in 25 ml solution of 0.1 M NazS, pH 3). Volume of hydrogen is plotted as a hnction of illumination time [52].

44 1

REFERENCES 1

2 3 4 5 6 7 8 9 10

II 12 13 14 15

16 17 18

19 20 21 22 23 24 25 26 27 28 29 30 31

Fujishima A, Honda K. Nature 1972; 238: 37-39. Pleskov YuV, Gurevich YuYa. Semiconductor Photoelectrochemistry. New York: Consultants Bureau, 1986. Pleskov YuV. Solar Energy Conversion: A Photoelectrochemical Approach. Berlin: Springer-Verlag, 1990. Sze SM. Physics of Semiconductor Devices, 2nd Ed. New York: John Wiley, 198 1 Gartner WW. Phys. Rev. 1959; 116: 84-87. Gronet CM, Lewis NS, Cogan G, Gibbons Y. Proc Nat Acad Sci USA, Phys Sci 1983; 80: 1152-1 156. Menezes S. Solar Cells 1986; 16: 255-282. Tufts BY, Abrahams IL,Casagrande LG, Lewis NS. J Phys Chem 1989; 93: 3260-3269. Heller A. ACCChem Res 1981; 14: 154-162. Miller B, Heller A, Robbins M, Menezes S, Chang KS, Thomson J. J Electrochem SOC 1977; 124: 1019-1022. Licht S, Tenne R, Dagan G, Hodes G, Manassen J, Triboulet R, Rioux J, Levy-Clement C. Appl Phys Lett 1985; 46: 608-610. Heller A, Leamy HJ, Miller B, Johnston WD. J Phys Chem 1983; 87: 3239-3244. Razzini G, Peraldo Bicelli L, Scrosati B, Zanotti L. J Electrochem SOC 1986; 134: 35 1-352. Tenne R,Wold A. Appl Phys Lett 1985; 47: 707-709. Howe AT, Fleisch TH. J Electrochem SOC1987; 134: 72-76. Switzer JA. J Electrochem SOC1986; 133: 723-728. Gronet CM, Lewis NS. Nature 1982-1983; 300: 733-735. Abrahams IL, Casagrande LG, Rosenblum MD, Rosenbluth ML, Santangelo PG, Tufts BY, Lewis NS. New J Chem 1987; 1 1 : 157-165. Nakatani K, Matsudaira S, Tsubomura H. J Electrochem SOC1978; 125: 406-410. Muller N, Cahen D. Solar Cells 1983; 9: 229-245. Tan MX, Newcomb C, Kumar A, Lunt SR, Sailor MJ, Tufts BY, Lewis NS. J Phys Chem 1991; 95: 10133-10142. Heller A. J Vac Sci and Techno1 1982; 21: 559-561 Razzini G, Peraldo Bicelli L, Pini G, Scrosati B. J Electrochem SOC 1981; 128: 2134-2137. Bard AJ. J Chem Educ 1983; 60: 302-304. Bocarsly AB, Walton EG, Wrighton MS. J Am Chem SOC1980; 102: 3390-3398. Noufi R, Nozik AJ, White J, Warren LF. J Electrochem SOC1982; 129: 2261-2265. Frank AJ, HondaK. J Phys Chem 1982; 86: 1933-1935. Malpas PE, Rushby B . J Electroanal Chem 1983; 157: 387-392. Noufi R. J Electrochem SOC1983; 130: 2126-2128. Harrison DJ, Calabrese GS, Ricco AJ, Dresner J, Wrighton MS. J Am Chem SOC1983; 105: 4212-4219. Nazeeruddin MK, Liska P, Moser J, Vlachopoulos N, Gratzel M. Helv Chim Acta 1990; 73: 1788-1803.

442 32 Vlachopoulos N, Liska P, Augustynski J, Gratzel M. J h e r Chem SOC1988; 110: 1216-1220. 33 Licht S, Peramunage D. Nature 1990; 345: 330-333. 34 Popkirov GS. Pleskov YuV. Elektrokhimiya 1980; 16: 238-241. 35 Bin-Daar G, Dare-Edwards MP, Goodenough J. Hamnett A. J Chem SOCFaraday Trans 1983, Ser I; 79: 1179-1 185. 36 Kraitsberg AM, Pleskov YuV., Mardashev YuS. Elektrokhimiya 1983; 19: 1435-1438. 37 Pleskov YuV, Zhuravleva VN, Pshenichnikov AG, Vartanyan AV., Arutyunyan VM, Sarkisyan AG, Melikyan VV. Geliotekhnika 1985; No. 4: 61-65. 38 Aharon-Shalom E, Heller A. J Electrochem SOC1982; 129: 2865-2866. 39 Contractor AQ, Szklarczyk M, Bockris JO'M. Electroanal Chem 1983; 157: 175-177. 40 Bockris JO'M, Szklarczyk M. Appl Phys Commun 1982-1983; 2: 295-299. 41 Ennaoui A, Fiechter S, Jaegermann W, Tributsch H. Electrochem SOC1986; 133: 97-106. 42 Kulak AI. Elektrokhimiya poluprovodnikovykh geterostruktur (in Russian). Minsk: Universitetskoye Izdatelstvo, 1986. 43 Decker F, Fracastoro-Decker M, Badawy W, Dobohofer K, Gerischer H. J Electrochem SOC1983; 130: 2173-2178, 44 Fan FRF, Hope GA, Bard AJ. J Electrochem SOC1982; 129: 1647-1649. 45 Gingey DS, Baughman RY, Butler MA. J Electrochem SOC1983; 130: 1999-2003 46 Pleskov YuV, Kraitsberg AM, Kolbasov GYa, Taranenko NI, Lipyavka VG. Solar Energy Mater 1991; 22: I 19-126. 47 White YR, Fan FRF, Bard AJ. J Electrochem SOC1985; 544-550. 48 Luttmer YD, Konrad D, Trachtenberg I. J Electrochem SOC1985; 132: 1054-1057. 49 Hodes G, Gratzel M. Nouv J Chim 1984; 8: 509-520. 50 Gerischer H. J Phys Chem 1984; 88: 6096-6097. 51 Gratzel M, ed. Energy Resources through Photochemistry and Catalysis. New York: Academic Press, 1983. 52 Gratzel M. In: Modern Aspects of Electrochemistry 1983; 15: 83-165. 53 Fox MA.Topics in Current Chem 1987; 142: 71-99. 54 Gruzdkov YuA, Savinov EN, Parmon VN. In: Fotokataliticheskoye preobrazovanie solnechnoj energii (in Russian). Novosibirsk: Nauka, 1991: 138-186. 55 Gruzdkov YuA, Savinov EN, Makarshin LL, Parmon VN. Ibid. 186-228. 56 Serpone N, Borgarello E, Barbeni M, et al. J Photochem 1987; 36: 373-388. 57 Ito K, Fujishima A. J Am Chem SOC1988; 110: 6267-6269. 58 Domenech J, Andres M, Munoz J. Electrochim Acta 1987; 32: 773-776. 59 Kraeutler B, Bard AJ. J Am Chem SOC1978; 100: 5985-5992. 60 Hori Y, Suzuki S. Chem Lett 1987; 1397-1400. 61 Rose TL, Nanjundiah C. J Phys Chem 1985; 89: 766-371. 62 Matthews RW. J Catal 1988; 113: 549-555. 63 Al-Ekabi H, Serpone N. J Phys Chem 1988; 92: 5726-573 1 , 64 Matthews RW. J Catal 1986; 97: 565-568. 65 Matthews RW. J Phys Chem 1987; 91: 3328-3333. 66 Hidaka H, Kubota H, Gratzel M, Serpone N, Pelizzetti E. New J Chem 1985; 9: 67-69. 67 Pelizzetti E, Carlin V, Minero C, Gratzel M. New J Chem 1991; 15: 35 1-359. 68 Kalyanasundaram K, Borgarello E, Gratzel M. Helv Chim Acta 1981; 64: 362-368

443

69 Fedoseev VI, Savinov EN, Parmon V N Kinetika i kataliz 1987; 28: 1 1 1 1 - I 1 15. 70 Savinov EN, Gruzdkov YuA, Parmon VN. Int J Hydrogen Energy 1989; 14: 1-9. 7 1 Parmon VN. In: Fotokataliticheskoye preobrazovaniye solnechnoj energii (in Russian). Novosibirsk: Nauka, 1985: 42-57.

This Page Intentionally Left Blank

445

PROSPECTIVE USAGE OF PHOTOELECTROCHEMISTRY FOR ENVIRONMENTAL CONTROL

S.K. Haram and K.S.V. Santhanam Chemical Physics Group, Tata Institute of Fundamental Research, Colaba Bombay 400 005, India

1.

INTRODUCTION: Solar energy has been playing a vital role in maintaining a

cleaner environment ever since this universe was formed. The photosynthesis which is occurring in green plants has been converting COz into useful products; however, the

deforestation during the last two decades has

caused an alarm as the process of C02 removal has been reduced. A global warming due to an

increasing concentration of CO

2

and the green house

effect are of considerable concern in today’s environment. The ideal control of the concentration of C02 in the atmosphere would be to convert

it at the point of its entry into the atmosphere. If we examine today’s input to this toxic waste (C02 is less toxic than CO; toxicity level of CO = 5000 ppm, toxicity level of CO= 50 ppm) it arises from industrial 2 wastes and automobile exhausts. The decontamination of this can be brought about by mimicing the photosynthetic process in a modified form i.e., utilising the photons of the sun in an appropriate manner to convert CO 2 into useful or less toxic waste. Among other toxic wastes entering into our atmosphere are H S, N 0, NO, HCN, hydrocarbons etc.. 2 2 Photoelectrochemistry can play a significant role in creating a cleaner environment by converting the above chemicals into other value added products. Although this concept has been in discussion in the last two decades, the practicality of it is slowly building up in recent times as a result of semiconductor/electrolyte

a large number of fundamental studies at the interface.

Several

oxidative

and

reductive

processes have been brought out in the last several years (1-10) at the semiconductor electrodes. An excellent review ( 1 ) on photoelectrochemical oxidation processes summarises the developments in this area. The present review discusses the photoelectrochemical reductive processes that have

446

been carried out at the semiconductor electrodes which have a bearing on the environmental control by the polluting gases.

2.

a)

PHOTOELECTROCHEMICAL REDUCTION OF C02

At bare semiconductors: The semiconductors which are suitable for photoelectrochemical

reduction of C02 are shown in Figure 1.

-'T w I v,

ui

>

> +2

AlAs

1

--

Eo= -0.096V CO*/CH30H

Lo= -0.16V C02/HCH0

CdS

EO=

+3 --

-0.1 1 v

co,/co

Figure 1. Band diagram of selected semiconductors for C02 reduction

A wide selection of semiconductors is capable of effecting this reduction. Out of these several semiconductors, atleast three of them have been examined in detail. p-GaAs single crystal electrode has been used by Zafrir et a1 (11). The photocurrent onset potential for CO was observed at +O.O5V 2 in 3.1 M CaC12 containing 0.87 M HC1 and 0.07 M V ( I I ) chloride. The

V(II)/V(III)

couple

was

used

to

bring

about

efficient

electron

transfer reaction at the semiconductor electrode. A faradaic yield of 1.2% for HCOOH has been obtained at a potential of -0.7V at a current density 2

of 8 . 5 mWcm

.

The optical to chemical energy conversion efficiency is

447 estimated at 0 . 2 % . The other products have the following faradaic yields; HCHO = 0.3% and CH OH = 0.13%.The following step wise conversion of C02 is 3

proposed at the semiconductor electrode.

co2

+

COO-'

->

e +

H +

HCOO'+ e

COO-'

(1)

-> HCOO'

(2)

->

-

HCOO

(3)

This conversion produces H2 as the bye product through direct reduction of H+. The

p-CdTe and p-InP electrodes have also been investigated ( 1 2 ) ;

while the reaction scheme is essentially as indicated above, the net formation of formic acid is demonstrated to be a function of the pH of the medium. Hence using the solutions of supporting electrolytes such as carbonates, sulphates, phosphates and perchlorates of alkali salts or tetra alkyl ammonium salts, the reaction ( 4 ) has been effected.

co-'+ 2

H ~ O + e ->

HCOO-+ OH-

(4)

playing a role in the photoelectrochemical

The possibility of adsorbed CO',

conversion at p-CdTe has been reported (13). The following steps are indicated C02+

7 co,.

e

-

co2 ' + co2--

(5)

o=c-0-co-0-

(6)

-> o=c-o-co-6

(7)

5

O=C-0-CO-O-+ e

o=c-0-co-0-

>

coz- + co

(8)

This mechanism has been invoked to account for the different products obtained at different electrodes such as Pb and p-CdTe. Table 1 summarises the conversion efficiencies of CO

2

at the semiconductor electrodes. It is

interesting to note that at p-GaP in Na2C03 electrolyte, a very high conversion of CO

2

to HCOOH is

reached, while at p-CdTe in the same

electrolyte a very low conversion is

obtained. A lso the conversion of

448

C02 to CO is efficient

at p-CdTe in non-aqueous electrolyte. The

conversion of C02 to CH OH is very high at p-GaAs at a neutral pH. While no 3 rationalisation is possible on the basis of the band diagram, it appears some specific adsorption of the electrolyte and C02 might be playing a role in the conversion.

Table

1:

Photoelectrochemical

reduction of

C02

at

semiconductor

electrodes.

Semiconductor

Incident power mW/cm

D-GaAs

Medium Bias Applied

Effi Ref c i ency

Faradaic yield

/o

2

V

1

2

4

3

-

0.045

0.012 0.003 0.0013 -

0.10

980

0.32 M -0.50 0.032 0.0180.007 VClz in 4 M HC1 -0.65 (8.4hours)

90

0.07 M

11

11

VC12 in 4 M HC1

-

p-GaAs

0.5

M

-1.05 0.07

0.003

-

-

1 la

0.1 to 0.8

-

llb

-

KC 1 p-GaAs

-

Na2S04

-1.15

-

-

-1.6

-

-

-

-

-

15

-

-

-

-

-

13

-

-

-

12

H2°

p-CdTe

0.1 M

TBAP in DMF 0.1M NH4C104

p-CdTe

3.2

M TBAF in

0.1

CH3CN p-CdTe

0.8

0.1 M

-1.20 0.48

0.066

449 Table 1 ( c o n t d ) Na2C03

p-CdTe

0.8

0.1 M TEAP

-1.20

0.20

-

-

0.65

-

0.1 M TEABr

-1.20

0.076

-

-

0.78

-

0.1

M

TEACl p-CdTe

-

p-GaP

p-GaP

-1.20

0.088

-

-

0.51

-

-

0.78

0.1 M TBAP H20/DMF

-1.35

-

-

0.5 M Na2C03

-0.75

0.78

-

R4N'C10;

-1.00

0.12

-

-

0.1 M TEAP

-1.20

0.36

-

0.1 M

-1.20

0.28

0.1 M TEAC 1

-1.20

0.1 M Na2C03

12

1Id

-

lla

0.03

-

llc

-

0.26

-

12

-

-

0.31

-

0.31

-

-

0.28

-

-1.20

0.17

-

0.08

-

-

0.1 M Na2S04

-1.20

0.04

-

0.16

-

-

0.1 M Na3P04

-1.20

0.023

-

0.10

-

-

0.1 M LiC104

-0.75

-

-

0.63

-

-

0.024 0.005 -

H2° p-InP

-

TEABr

p-si

-

CH CN/H20 3

The numbers i n t h e t a b l e header r e p r e s e n t s : 1.HCOOH

2.HCHO

3. CH30H

4. CO

17

450 The

photoelectrochemical

electrodes also produces H2

reduction

C02

semiconductor H2 yield the in appreciable amounts (14); of

at

depends on the nature of the semiconductor surface. The

mediation

of

photoelectrochemical

reduction

of

CO,

by

ammonium ions has been reported by Bockris et a1 (15). The first step in this mediation is the reduction of NH; and follows sequences ( 9 ) through

I

0.S niAcm-'

I

I

I

I

-0 5

-1.0

-1.5

1 -20 -2.5

wv vs SCE Figure 2. Current-voltage curves in DMF containing 5% H20 containing NH4C104in C02 A. 0.1 M NH C104 (in Ar atmosphere). B. 0.1 M atm~sphere,~C. 0.1 M TBAP (in Ar atmosphere). Taken from reference (15).

45 1

The current efficiency for CO formation in the presence of NH C10 depends 4 4 on the water content. Figure 2 shows the current-potential curves at p-CdTe

It is interesting to note

electrode in DMF in the presence of NH4C104.

that the onset of photocurrent is substantially shifted to a negative potential in the presence of 0 . 1 M TBAP. The ammonium ion mediated reduction is obvious in the current-voltage curves.

b)

Semiconductors with deposited Catalysts: Several

semiconductors

coated

with

catalysts

have

been

investigated (14-18)for increasing the efficiency. At a metal coated p-GaP (14) the faradaic efficiency for C02 reduction is about 50% in propylene carbonate containing tetra alkyl ammonium salt; in comparison the aqueous medium produced a faradaic yield of a few percent. The products in non-aqueous solvents are (COOH)2, HCOOH, CO and H2, The earliest attempt to catalyse

the

C02

reduction

at

p-Si

was

carried

tetra-aza-macrocyclic complexes of Co(I1) or Ni(II1

out

by

(17,191. At

using this

modified semiconductor, the catalysed electrochemical reduction of CO 2 to CO occurs at -0.55 V vs. SHE; the photoelectrolytic reduction produces H 2

besides CO in the ratio of CO/H

of 2:l. The modified electrodes can bring

2 about the reduction of C02 to CO in 100% faradaic yield. However, poisoning of the semiconductor surface by either carbon or CO

has been postulated

for the decreasing efficiency with time. Table 2. The photoelectrochemical reduction of CO

2

at a metal coated

semiconductors SerniIncident Medium conductor power Wcm p-CaP

1.7

(uncoated)

p-GaP/Au

1.7

2

Bias

Faradaic

voltage

yield ( % I

vs.Ag/ AgC 1

1

2

3

*

Efficiency

Ref

%

4

0.1 M TEAP PC

-1.W

2.6

-

2.6 65

0.1 M TEAP PC

-1.2v

8

6

94 trace

-

14

108

14

452

Table 2 (contd) p-GaP/In

1.7

0.1 M TEAP PC

-1.2

7

7 107 trace 121

14

p-GaP/Zn

1.7

0.1 M TEAP PC

-1.5

1

8 100 trace 109

14

p-GaP/Pb

1.7

0.1M TEAP PC

-1.2

38

11 41 trace

90

14

0.1 M NaC1O4

-0.2

-

83 16

-

16

p-GaP/ 2+ Ni(cyclam1

-

p-GaP/ 2+ Ni(cyclam1

0.1 M NaC lo4

-0.75

3

3 85

8

-

16

2+ p-GaAs/ Ni(cyclam1

0.1 M KC104

-0.95

-

- 47

30

-

16

-0.75

-

- 63

32

-

17

H2° p-Si/Tetra azo macro cyclic complexes of Ni

0.1 M LiC104

-

CH3CN/H20

The numbers in the table header represents

*

1. (COOH)2 2.HCOOH 3.CO 4.H2

Faradaic efficiency

Bockris et a1 (20a) proposed a model for photoelectrocatalysis at metallised semiconductors.

3.

PHOTOCATALYTIC REDUCTION OF C02 The

use

of

photocatalytic

semiconducting powders

for

effectively converting C02 into one organic such as HCOOH or HCHO or CH30H or CH3CH0 or C H OH has been carried out by a large number of investigators 25 (21-25). Ulman et a1 (21,221have used SrTi03 or Ti02 powders for effecting

453

the above conversion. At Sic the reduction of C02 proceeds to yield CH3CH0 and C H OH (23); by metallising SIC the photocatalytic reduction follows a 25 different route as shown in Table 3.

Table 3. Photocatalytic reduction of GO at semiconductors 2

Semiconductor

Metal mol%

HCOOH HCHO CH30H p~

p~

p~

CH CHO EtOH Effici pil p~ ency

Refer ence

%

SiC/Pb

0.44

1.2

1.1

0.20

0.018

4.4

24

SiC/Pd

0.011

1.6

1.3

1.6

0.014

9.8

24

SiC/Pd

0.020 2 . 0

1.6

1.3

0.14

10.0

24

SiC/Fe

0.80

1.6

0.95 1 . 1

6.8

24

SiC/Fe

1.03

2.3

0.86 1.5

0.05

11.0

24

SiC/Pt

0.30

2.1

1.2

0.03

8.0

24

si c/cu

0.25

1.5

0.90 0.78

0.016

5.8

24

sic/cu

0.93

1.3

0.74 0 . 6 5

-

4.8

24

1.9

0.40 0.50

-

4.2

24

sic

1.1

Heat of combustion of products

Eff lciency The

=

formation

Incident 1 ight energy

x

of

is

methanol

and

ethanol

100

considered

through

photoexcitation of Sic powder and HCHO formed in the reaction. For example, the ethanol formation is sequenced as

w

HCHO +

HCHO

->CHO-CHO

+ 2

H+ + 2 e (13)

454

CHO-CHO + 2 H++ 2e >CH~CHO + 2 H++ 2 e

-

CH3CHO + 1/202

(14)

C~H~OH

(15)

where the electron is contributed by the photoexcited Sic powder; H+ is supplied by H20 via. the oxidation by Sic.

4.

PHOTOELECTROCHEMICAL DECOMPOSITION OF H2S The photoelectrochemical conversion of H S to H 2

2

was reported by

Kalyanasundaram and Gratzel (26a1, Crzyll et a1 (26b) and Chun et a1 (83). Kalyanasundaram and Gratzel

carried out the conversion of H S to H2 2

efficiently with CdS loaded with 0 . 5 % Ru02 at a quantum efficiency of 35%

.

Chun et a1 conducted the photoelectrolysis of aqueous sulphide at n-Si coated with polypyrrole films and then loaded it with Ru02. The following 2sequence of reactions is postulated for the decomposition of S

-

+hv >-

n-Si

2 H++ 2e

h++ e

H~

(16)

(17)

The photocorrosion of the semiconductor has been reduced by the polymer coating. The simultaneous generation of H2 is an advantage of this process. This method gives a power conversion efficiency of 0.6% and a quantum efficiency of nearly 1%.

5.

PHOTOELECTROCHEMICAL CONVERSION OF SO2

The photoassisted reduction of SO at p-WS2, p-Si and p-InP has 2 been reported by Calabrese et a1 (27). The light assisted reduction produces S20z- via. the reaction SO2

+

SO2

+ 2e

->

s20y

(19)

455

The current efficiency for this conversion is reported to be >90% at all the electrodes; the naked semiconductors such as p-InP show sluggish kinetics and hence requires metallisation. Figure 3 shows current-voltage curves for SO2 reduction at these semiconductor electrodes. At p-InP modified by Pt, a power saving efficiency of

I Illuminated

T

' Pt

b)-lmMS02 [o-Euq N] C 104 / CH,CN 100m V/s

05

/

Illuminated R

I

n (platinize Illurninal p

Q -1nP (naked)

!

1

-I

I

,

1

1

1

2 -08 -04 Potential, V vs A g / A g t

FIGURE 3. Current-voltage curves for 1 mH SO2 in 0.1 bl TBAP in CH3CN at Pt, P-si p-InPzand P-usz. Irradiation wavelength: 632.8 nm(40 mW/cm 1. Taken from reference (27). 9

2

about 11% has been reported using 514.5 nrn (100 mWcn efficiency

for

significant.

the

conversion of SO2

to S20:-

).

The solar

is expected

to

be

456 6.

PHOTOELECTROCHEMICAL REDUCTION OF

O2

The photoelectrochemical reduction of O2 t o H20z

is an

indirect process.The early investigation into it was carried out by Calabrese and Wrighton (28). The photoelectrochemical process of reduction of 2-t-butyl-9,lO anthraquinone in CH3CN was carried out in the presence of CH3COOH using 632.8 nm radiation. An ECE (electron transfer-chemicalelectron transfer) process occurs resulting in the formation of the corresponding hydroquinone. A 2% photoelectrochemical energy conversion efficiency was obtained for this process. The hydroquinone reacts with O2 to yield >0.18M H202 by the following scheme

Keita and Nadjo

(29) used

9,lO-anthraquinone 2,6-disulphonate

a single redox couple, sodium

in

the

regenerative

constructed with p-WSe2 and p-Si photocathodes where

photocells

the following

reaction occurs 2 HI

+

->

O2

H202+

(20)

I2

The synthesis of H202 by the following route has also been proposed by the above authors Cata1yst AQ >-

AQHZ

(21)

H2 AQH,

+

02->

AQ

+

H2°2

(22)

where AQ represents anthraquinone. This scheme is similar to the one

457

proposed earlier; however, there is a possibility of a side reaction of reducing the overall yield. The formation of H 0 at the 22 illuminated Ti02 film electrode by the sol-gel method has also been

H202 with AQH2 proposed (301.

7.

PHOTOELECTROCHEMICAL GENERATION OF H2 Perhaps one of

the exciting

interests in this field when

Fujishima and Honda reported (31) photoelectrolysis of H 0 at n- Ti02 was 2 the generation of H2 at the cathode. As this electrolysis uses a wide bandgap semiconductor ( E =3.2eV), the practical utilisation of it has been g limited. The prospects of using small band gap semiconductors has been explored in the last few years by a large number of investigators (32-35). The previous attempts by others have already been reviewed (36-43)in the literature. The photoelectrochemical hydrogen evolution using metallised semiconductors have been investigated (44,55-60)as the metal catalysts increased the Gibbs free energy efficiency upto about 13%. Aspnes and Heller (44)proposed that the work function of the metal catalyst with reference to the semiconductor brings a true ohmic or nearly ohmic contact with a substantial collapse of the barrier between the metal and the semiconductor

upon hydrogenation. Figure 4 depicts the work function of

metals and band edges of semiconductors and the processes occurring at the semiconductor. It is observed (44) that there is a finite delay time when no gas evolution takes place until the separate hydrogen and oxygen enveloped

the

catalyst

semiconductors produce

islands are

established. However, the

oxide

trapping effects which are not reproducible. The

hydrogen evolution at small band gap semiconductors such as n-CdS would be accelerated by Rh or Ru than Pt. This conclusion is well supported by the hydrogen evolution on n-CdS deposited with Ru02 (44)and absence of it when deposited with Pt (for this discussion the work function of Ru02 is 4.8 eV which is close to Ru metal). The formation of ohmic contact

due to the

surface damage during sputtering of metals onto the semiconductors has also been reported (46); this has been demonstrated for n-CdS where surface regions are sufficiently damaged. The hydrogen evolution at a Pt modified p-InP electrode has been investigated in detail (32). This modification produced an efficient and stable solar to chemical energy conversion efficiency of 9.2 % by etching the electrodes in Conc. HC1. The formation

458

A

I

Cds

z 3 a

-6

IV

W J

W

-1

-1; ?

-I3

I

-8

c

B

(1 a

x

ANODE

2 h++ n20

0 z

-

CATHODE

2C-+ZH++#--

zn'+vz o

HZ

I-

V W w -I

I

n-TiOp

FIGURE 4. A. Semiconductor band edges and work function of metals B. The band bendings of a semiconductor in assymnetric junctions Taken from reference (441.

of minute Pt islands on InP is recommended where the InP is in direct contact with the solution. The deposition of Pd on the semiconductor i n the place of Pt produces a very poor response. The efficient cell performance is attributed to the interface states produced by hydrogen; the adsorption of hydrogen on the metal surface and consequently decreasing the metal

work function and the introduction of high barrier have been considered to be ineffective.

459

Several doped oxide materials have been considered for efficient production of hydrogen Somarji's

group

in the photoelectrolysis experiments

(35) produced

hydrogen

from

water

(34,351.

electrolysis by

illuminating Mg doped (p-type) and Si doped (n-type) iron oxides. The photoelectrochemical properties of p-CaFe204 and p-SrFe10022 with band gaps of 1.8 eV and 1.9 eV respectively have been considered for hydrogen generation (34). Figure 5 shows the band diagram for these semiconductors.The deposition of Pt on the surface of both electrodes increases the hydrogen evolution reaction. The conversion efficiency from light to H2

and O2 was estimated to be about

The long term

stability of CaFe204 is higher than Sr7Fe10022. Several other oxides having Fe in their lattice have been considered as photoanodes with band gap energies of about 2.1 eV on the basis of the stability considerations (36-43). The kinetics of photoelectrochemical oxidation of water has been

considered at n-Ti02 semiconductor electrode (46-48). The possibility of

I

Current (log i)

FIGURE 5. A description of the electron transfer process at CaFe 0 and band structure description for CaFe 0 and Sr2Fe100z24 e band formed by crystal field spligting 4 of iron g . s and d orbitals is shown. e 1,and a I hopping bands 8 1g with trigonal distribution of the lattice are also indicated in the figure. Taken from reference (34).

5.

A

460

photointercalation in semiconductors has been considered (49,501 and it has been shown that the intercalation in p-WSe2 causes a high overpotential for hydrogen evolution reaction. The quantity of hydrogen corresponding to 10’

monolayers

coverage

has

been

obtained

by

heating

p-WSe

2’

Tungsten-silicon and other modified electrodes have been examined for the stability in the continuous generation of H2 in photoelectrolysis (51). p-Si surface stabilised by hydrogen has been shown to produce H2 for more 2 than 48 h at a c.d. 240 mA/cm with no degradation. The characteristics of

Ni modified p-Si have also been examined in KOH solutions (52). The metal catalysts incorporated into p-Si have been investigated for the hydrogen evolution (53) in acidic solutions. It has been shown that Pt/p-Si contact is nearly ohmic; good rectifying character develops by etching this electrode in alkali. A similar result has been obtained with a Au layer covering p-Si. In effect this indicates the barrier height is increased at the junction. The structure of the metal layer plays a role in the photo activity. If the semiconductor is coated with minute islands of the metal

it causes H2 evolution at positive potentials than the bare Pt or bare p-Si. A discontinuous metal covering produces a good photovoltaic activity (53). Photosensitised reduction of water at Ru02 has also been examined (54) and

the

activation

energy

for

hydrogen

evolution

has

been

substantially changed by the addition of anthracene carboxylates. The photoevolution of H2 at metal catalysts has

been examined by a number of

investigators (55-60); conclusive evidence exists that the metal islands are necessary for a good activity. A short circuited photoelectrochemical cell has

been used with sintered CdS pellets with Pt deposits for the

oxidation of HCOOH and simultaneous evolution of H2 (61);3 p o l of H2 in 180 minutes was generated with an input energy of

1.2 kJ m

simultaneous generation of H2 and 0 using Ti02 and Ru(bpy):+ 2

-2 -1 s

.

The

has been

considered as a function of grain size and amount of Ti02 (62). The addition of thiourea and 2-aminopyridine has been shown to suppress the photocorrosion of Ti02 (65).

A physical model for water splitting to H2

has been described using semiconductors (63). The need for developing new materials for producing H

2

has also been emphasised (64,661. Popkirov and

Pleskov (67) proposed a new system for hydrogen production by water electrolysis and Siege1 and Schott (68) have suggested the optimisation of photovoltaic cells for hydr0gen.A derivatized p-InP has been used in the photoelectrochemical production of H2 (69).

The

generation

of

hydrogen

using

small

bandgap

semiconductors while stabilising from photo-corrosion has been quite challenging and has been innovatively solved by the concept of vectorial charge transfer in vertical and circular configurations (70,711.In this method a higher driving force for the decomposition of water has been obtained by the vectorial addition of voltages. The ultimate decomposition of water to hydrogen and oxygen occurs at the two Pt electrodes. A long term stability of the semiconductor is reached by using a polysulphide or polyselenide medium (72). A polymer coated semiconductor electrode in place of the naked electrode in the vectorial charge transfer assembly has also

been used

successfully

(73-75). The

production

of

H2

using

p-type

transition metal phosphides and p-type molybdinium sulphides have also been reported (76,77). The function of Co and Pt on p-InP in the evolution of H

2

has been explored by

(80). Pt

Goodman and Wessels (78,791and Kobayashi et a1

intercalated to K4Nb6017

has also been examined for water

decomposition (81).

8.

REFERENCES

1.

M. A . Fox in "Topics in Organic Electrochemistry", Eds., A. J. Fry and W.E. Britto, Plenum, New York, 1986

2.

A. J.

Bard, The Robert A .

Welch Foundation Conferences on Chemical

Research XXVI I I , , November 5-7, 1984

105,27

3.

M.D. Ward. J . R . White and

4.

S . Sato, Chem. Commun., 26

5.

T. Kanno, T. Oguchi, H. Sakuragi and K. Tokumura, Lett.,a, 467 (1980)

6.

R.A. Barber, P. de Mayo and K. Okkada, Chem. Comm., 1073 (1982)

7.

M. Gratzel, Acc. Chem. Res., l4, 376 (1981)

8.

J . Cunningham and

A

J. Bard, J. Am. chem.soc.,

(1983)

19821

B.K. Hodnett, J. Chem. Soc., Faraday

Tetrahedron

I, 7 8 ,

3297

462 ( 1982

9.

E.M. Kuliev, A.S. Sulelmanov and U.Kh. Agaev,

Elektrokhimiya, 22,

1069 (1986) 10a. C. Grutierrez and P. Salvador, J. Electrochem. SOC.,

133, 924

( 19861

lob. K.S. V. Santhanam in "Photoelectrochemical Solar

K.S. V. 11.

Cells", Eds

Santhanam and M. Sharon, Elsevier, Amsterdam, 1988

M. Zaf'rir, M. Ulman, Y. Zuckerman and M. Halmann, J. Electroanal. Chem.,

159, 373

(1983)

lla. B. Aurian-Blajeni, M. Halmann and J. Manassen. Solar Energy Mater., 8, 425 (1983) llb. D. Cranfield and K.W. Frese, J. Electrochem. SOC.,130,7772 (1983)

llc.

S.

Ito, S. Ikeda, M. Yoshida, S. Ohta and T. Iida, Bull Chem. SOC.,

Jpn., 5 5 , 583 (1984) lld. S. Tariguchi, B. Aurian-Blajeni and J. 0.M. Bockris, Electrochim. Acta,

29, 923 (1984)

12. H. Yoneyama, K. Sugimura and S.

Kuwabata, J. Electroanal. Chem.,

249, 143 (1988) 13. B. Aurian-Blajeni. M. A. Habib, I. Taniguchi and J.O'M.Bockris,

J.

Electroanal. Chem., 157, 399 (1983) 14. S. Ikeda, Y. Saito, M. Yoshida, H. Noda, M. Maeda and K. Ito, J. Electroanal. Chem., 260, 335 (1989) 15. I. Taniguchi, B. Aurian-Blajeni and J.O'M. Bockris, J.

Chem.,

Electroanal.

161,385 (1984)

16. J.P. Petit, P. Chartier, M. Beley and J.P. Denlle, J.

Electroanal.

463 Chem., 269, 267 (1989) 17. G. Bradley, T. Tysak, D.J. Graves and N.A. Vlachopoulos, Chem.

Commun., 349 (1983) 18. S.I. Ikeda, M. Yoshida and K. Ito, Bull. Chem.

Soc.,Jpn., 5 8 ,

1353

( 1985

19. G. Bradley and T. Tysak, J. Electroanal. Chem., 135, 153 (1982) 20. R.O'Toole, L.D. Margerum, T.D. Westmoreland, W.J. Vining, R.W. Murray

and T.J. Meyer, Chem. Commun., 1416 (1985) 20a. J.O'M. Bockris, S.V. Khan, O.J. Murphy and M. Szklarczyk, Int. J.

Hydrogen Energy, 3, 243 (1984) 21. M. Ulman, A.H.A. Tinnemans, A. Mackor, B. Aurian-Blajeni and M. Halmamm, Int. J . Hydrogen Energy, 3 l , 429 (1983)

22. M. Halmann, M. Ulman and 8. Aurian Blajeni, Sol. Energy, 31, 429 ( 19831

23. S. Yamamura, H. Kojima, J. Iyoda and W. Kawai, J. Electroanal. Chem.,

225, 287 (1987) 24. S. Yarnamura, H. Kojima, J. Iyoda and W. Kawai, J. Electroanal. Chem.,

247, 333 (1988) 25. S . Yamagata, B.H. Loo and A. Fujishima, J.

Electroanal.

Chem.,260,

447 (1989) 26a. E. Borgarello, K. Kalyanasundaram and M. Gratzel, Helv. Chim. Acta.,

65, 243

(1982)

26b. L.R. Grzyll, J . J. Thomas and R.G. Barile, Int. J . Hydrogen

Energy,

l4, 647 (1989). 27. C. S. Calabrese, T.J. Sobieralski and M.S. Wrighton, Inorg.

Chem.,

464

22, 1634 (1983) 28. G.S. Calabrese and M.S. Wrighton, J. Electrochem. SOC.,128, 1014 ( 1981 1

29. B. Keita and L. Nadjo, J. Electroanal. Chem.,

163, 171

(1984)

30. T. Yoko, K. Kamiya, A. Yuasa and S . Sakka, J. Electroanal. Chem.,

209, 399 (1986) 31. K. Honda and A. Fujishima, Nature (London

, 238, 37 (1972)

32. H. Kobayashi, F. Mizuno, Y. Nakato and H. Tsubomura, J. Phys.

93, 819

Chem.,

(3991)

33. K. Sayama, A. Tanaka, K. Domen, K. Maruya and T. Onishi, J. Phys.

Chem., 93, 1345 (19911 34. Y. Matsumoto, M. Omae, K. Sugiyama and E. Sato, J. Phys. Chem.,

9l,

577 (1987) 35. J.E. Turner, M. Hendewerk, J. Parmeter, D. Neiman and

G.A.

Somarji,

J. Electrochem. SOC.,131, 177 (1984) 36. K.L. Hardee and A. J. Bard, J. Electrochem. Soc.,

123, 1024

(1976)

37. D.S. Butler and M.A. Butler, J. Appl. Phys., 48, 2019 (1977) 38. D.E. Scaife, Solar Energy, 25, 41 (1980) 39

J.H. Kennedy, R. Shinar and J.P. Ziegler, J. Electrochem SOC., 127, 2307 (1980)

40. N. Hatanaka, T. Kobayashi, H. Yoneyama and H. Tamura.

Electrochim.

Acta, 27, 1129 (1982) 41.

A. Shami and W.E. Wallace, Mater. Res. Bull., l8, 389 (1983)

465

42. Y. Matsumoto, M. Omae, I. Watanabe and E. Sato, J. Electrochem.

133, 43 *

SOC.,

711 (1986)

E. Pollert, J.

Hejtmaneck, J.P. Doumere, J.

Clawerid

and

P.

Hagenmulle, J. Phys. Chem. Solids, 44, 237 (1983) 44. D.E. Aspnes and A. Heller, J. Phys.Chem., 87, 4919 (1983) 45.

D.J. Meissner, R. Memming and B. Kastening, Chem. Phys. Lett., 96, 34 ( 19831

46. M. Pujadas and P. Salvador, J. Electrochem. SOC.,136, 716 (1989) 47.

P. Salvador, J. Phys. Chem., 89, 3863 (1985)

48. F. Vanden Kerchove, A. Praet and W.P. Gomes, J.

132, 2357 49. C.

Levy

Electrochem.

SOC.,

(1985)

Clement

in

"Ion-sensing Electrodes

and

Electrochemical

Instrumentation",Ed.,K.S.V.Santhanam,World Scientific, Singapore, 1990, p.444 50. A. J. McEvoy and F. Boroumadnd, J . Electroanal. Chem., 209, 395 51. B. Keita and L. Nadjo, J. Electroanal. Chem.,

199,229

(1986)

(1986)

52. C. Li and S . Wang, J. Electroanal. Chem., 227, 213 (1987) 53. Y. Nakato, H. Yano,

S.

Nishiura, T. Ueda and H. Tsubomura, J.

Electroanal. Chem., 228. 97 (1987) 54.

J.M. Kleign and G.K. Boschloo, J. Electroahal. Chem., 300, 595 ( 1991 1

55. Y. Nakato, S. Tomomura and H. Tsubomura, Ber. Bunsenges, 83, 1289 ( 1976 1

56. K. Ohashi, J .

McCann and J.O.M. Bockris, Energy Research, 1, 259

466 ( 1977)

57

R.N. Cominey, N. S. Lewis, J.A. Bruce, D.C. Bookbinder and M.S. Wrighton, J. Am. Chem. SOC.,104, 467 (1983)

58. A. Heller and R.G. Vadimsky, Phys.Rev. Lett., 46, 1153 (1981) 59. W. Kautek, J. Colrecht and H. Gerischer, Ber. Bunsenges. Phys.

Chem.,

84, 1034 (1980) 60. M. Szklarczyk and J.O.M. Bockris, Appl. Phys.Lett., 42, 1035 61

(1983)

C. Bamwenda, A.M. Mika and T.2. Winnicki, Electrochimica Acta,

x,153

(1991

62

R.M. Quint and N. Cetoff, Int. J. Hydogen Energy,

63. S.U.M. Khan, R.C. Kainthal and J.O.M. Bockris, 64. T.Ohta, Int. J. Hydrogen Energy,

13, 333

13, 269

(1988)

225 (1988)

(1988)

65. C. Prasad, N.N. Rao and O.N. Srivastava, Int. J. Hydrogen Energy, l3, 399 (1988) 66. K. Uosaki and H. Kaita, Chem. Lett., No.2, 301 (19841 67. G.S. Popkirov and Yu. V. Pleskov, Elektrokhimiya, 24, 907 (1988)

68. A.Q. Siege1 and T. Schott, Int. J. Hydrogen Energy,

s, 659 (1988)

69. T.E. Mallouk, K.A. Daube, A. Spool and M.S. Wrighton,

Electrochem.Soc., 169th Meeting, Vo1.86, p.767 70. E. Smotkin, A.J. Bard, A. Campion, T. Mallouk, S.E. Uebber

and

R.E.

White, J. Phys. Chem., 90, 4604 (1986) 71. V. Sundaram, S . Aravamuthan and K.S.V. Santhanam. Ind. J. Tech., 613 (1987)

&

467 72

M. Sharon, P. Veluchamy, C. Natarajan and D. Kumar,

Electrochimica

Acta, 33, 1107 (1991) 73,

S. Aravamuthan, C. Levy Clement and K.S.V. Santhanam, J. Electroanal. Chem., 248, 233 (1988)

74.

S. Aravamuthan, S. Madhavan, C. Levy Clement and K.S.V. Santhanam, J. Power Sources, 32, 1 (1990)

75.

K.S.V. Santhanam, unpublished results

76.

H. Von Kanel, L. Cantert, R. Hanger and P. Wacher,Int. J. Energy,

77.

lo, 821

Hydrogen

(1985)

M. Szklarczyk and J.O.M. Bockris, Int. J. Hydrogen Energy,

831

(1984)

78.

C.E. Goodman and B.W. Wessels, Appl. Phys. Lett., 49, 829 (1986)

79.

C.E. Goodman and B. W. Wessels, Electrochem. SOC. , 170th Fall meeting, Vol. 86-2, p. 918

80.

H. Kobayashi, F. Mizuno, Y. Nakato and H. Tsubomura, J . Phys.

Chem.,

93, 819 (1991)

81.

KI. Sayama, A. Tanaka, K. Doman, K. Maruya and T. Onishi, J.

Phys.

Chem., 93, 1345 (19911 82.

H.J. Chai, K. Sun Kim, D. Hyunyoon, S . Nan and K. Hoson, Int. J Hydrogen Energy, l6, 379 (1991)

83.

B.C. Chun Yu, 934 (1986)

S.

Kapusta and N. Hackerman, J. Electrochem. SOC.,133,

This Page Intentionally Left Blank

469

ELECTROCHEMICAL STORAGE OF SOLAR ENERGY Yu.1.Kharkats. Yu.V.Pleskov A.N.Frumkin Institute of Electrochemistry, The Russian Academy of Sciences, 1 I707 1 Moscow, Russia

SUMMARY Electrochemical methods of energy storage as applied to photovoltaic solar energy conversion are discussed. A state-of-the-art of the plants consisting of solar cells, electrolyzers, and/or secondary batteries is presented. Recent progress in the simulation of the plant performance as well as its design optimization is surveyed.

INTRODUCTION The problem of solar energy conversion is tightly bound to the problem of energy storage. Indeed, the moment of the peak solar irradiation does not coincide with the peak energy consumption. The method of storage is determined by the particular kind of solar energy conversion. E.g., by thermal conversion, energy is accumulated in the form of heated water; by thermochemical, or photochemical conversion, energy is stored as chemical energy, in the products of the thermo - or photochemical reaction, and can be used, when necessary, by, e.g., combustion of the chemical fuel so produced. (The same is true for the photoelectrochemical conversion, see the preceding paper.) In the photoelectric (photovoltaic) conversion, the immediate product is electric current. Electrical energy can be converted into chemical energy to be stored, using electrochemical devices, namely water electrolyzer and/or secondary battery. so, generally the whole unit consists of (photovoltaic) solar array, water electrolyzer, and, possibly, secondary battery. The problem of converting solar energy to the chemical energy of hydrogen was discussed at great length in the literature, both from technical and economical view point (see, e.g., [ l,2]), Solar hydrogen systems were shown to be very promising not only for sun-rich regions like Saudi Arabia [3] and Libya [4] but also for mid-european countries [5] and Canada [ 6 ] . During the last decade, quite a few solar-hydrogen plants have been designed. One of the most urgent problems emerged in the course of research and development is mathematical simulation of the operation of the plant and optimization of its design and performance. In this paper the state-of-the-art in the field of designing plants for solar (photovoltaic) water electrolysis will be briefly reviewed, and the mathematical approach to the simulation and optimization of the plant will be presented.

Solar-Hydrogen Plant: Principles of Design and State-of-the-Art Solar (photovoltaic) plant with "hydrogen" energy storage consists essentially of the solar cells array (either stationary or with tracking-the-sun heliostat) and water electrolyser, supplemented with facilities for hydrogen storage. In the plants discussed below, singlecrystalline silicon solar cells were used, sometimes provided with the Fresnel lenses for the

410

sunlight concentration, and alkaline as well as solid-polymer-electrolyte electrolysers. Hydrogen was stored either in tanks (being compressed to some hundred bar), or as a solid hydride of some metal alloy (e g., CaNi5) which releases hydrogen back upon heating. The complete plant should comprise devices for later use of hydrogen, i.e., fuel cell for generating electricity, and/or a hydrogen torch for generating heat. Finally, facilities for monitoring performance of the plant and it different parts, and a control unit must be included. Attempts have been made (see, e.g., [7]) to supplement the plant with secondary battery, which plays a dual role in it. What lies on the surface, battery can be directly used, along with the electrolyzer, to store energy: this can be of practical advantage, for instance in using solar energy for the life-support of independent consumers in remote areas (by consumption of hydrogen, to obtain heat for, e.g. heating buildings, whereas the electricity from the battery is taken for feeding radio- and TV-equipment, etc.). Of course, with an effectively operating both he1 cell and electrolyzer, the two kinds of stored chemical energy, in the battery and "in hydrogen", are inconvertible, so that the consumer could choose the method of energy storage to suit the circumstances (proceeding from technological advantages of maintenance, possible limitations on specific power, etc). In practice, however, the technology of he1 cells is not yet sufficiently refined for them to find wide use for domestic and economic purposes, so that the convertibility of electrical to chemical energy still is more or less unidirectional, Therefore, the problem of combining the electrolyzer and storage battery for storing the energy produced by solar arrays remains to be vital (and the relative significance of the one or other of them in a particular plant is specified by the consumer). But what is still more important, the role of electrolyzer in such a team is not only to store energy; it also serves as a regulator to match the operation of a solar array to that of secondary battery in such a way that the former should operate under favourable conditions at each instant of the light day regardless of the current picked up by the latter. As is known, in the course of battery charging the power consumed for this purpose drops with time, sometimes drastically (see the last Section). The electrolyzer continuously "takes off' the difference between the electrical power produced by the solar array (which is a function of the time of day and the irradiation intensity) and the power consumed by the battery (which is a function of its state of charge) and converts this difference to the chemical energy of hydrogen. So, matching of the three main parts of the plant, i.e., solar array, electrolyzer, and secondary battery is of primary importance. For a simpler case, that is, a plant consisting of only solar cell and electrolyzer, it is used to quantitatively characterize the efficiency of the plant by the so-called "sun-to-hydrogen" efficiency:

where qs is the efficiency of the solar cell and qe is the "efficiency of the electrolyzer". The latter takes account of energy losses both in the electrolysis process proper and due to poor matching of the two parts of the plant. As qs is in any case preset by the chosen solar cell, the task of making an efficient plant amounts to increasing qe in every possible way. To this end, besides using a high quality electrolyzer, the characteristics of the electrolyzer and solar array should be well matched.

47 1

The current-voltage characteristcs of a solar cell are schematically given in Fig. la. The cell output, i.e., the product of photocurrent and voltage attains its maximum at a certain value of an external load.

Fig. 1 . Matching of current-voltage characteristics of solar cell and electrolyzer: (a) at constant light power density: 1, 2, 3 - characteristics of solar cell at Ns = 1,2, and 4 (at Ss = const); 4 - characteristic of electrolyzer (Ne = 1); o - maximum power point (MPP); dashed line shows the locus of maximum cell output power at the given radiation power density; (b) at varying radiation power density: l', 2'. 3', characteristics of solar cell; dashed line - locus of MPP; hatched area - variations of MPP in the most probable limits of variation of the light power density and temperature; 4 - characteristic of electrolyzer.

412

The operation point of the system as a whole conforms to the intersection of current-voltage characteristics of the solar cell and the electrolyzer, i.e., to the conditions (here, for simplicity, the ohmic voltage drop in the connecting wires has not been allowed for) Is = I, = I and Vs = Ve = V, where I is the current; V is the voltage (the subscrips s and e refer respectively to the solar cell and electrolyzer). As the solar cell operates most effectively at its maximum power point, therefore, the current-voltage performance characteristic of the electrolyzer should exactly pass through this point. This aim is easily attained when the solar cell and the electrolyzer have a modular structure, i.e., consist of a number of identical parts (modules). By connecting these modules in series circuits over N, and Ne of the modules and combining these circuits in parallel, the current-voltage curves can be arbitrarily changed within the given dimensions (areas) of both components of the plant, maintaining the power constant, This matching scheme is illustrated in Fig. la. Dividing the total area Ss of the solar cell into 2,4, and more parts, we get, instead of curve 1, curves 2,3, etc., and their points of maximum power lie on the hyperbola Pma, = (I*V)max = const. The same can be done with the current-voltage characteristics of the electrolyzer (not shown in Fig. 1). Adjustment of Ns and Ne is continued until the obtained curves intersect at the solar cell maximum power point. The procedure is readily applicable to the secondary battery as well (for more detail see the last Section). Even after setting up the values of Ns and Ne which are optimum for a certain preset illumination intensity, the optimization problem cannot be considered to be conclusively solved. In fact, the illumination intensity varies depending on the day time, season (these are periodic variations) and cloudiness (random variations). The geometric place of maximum power points on the curves corresponding to different illumination intensities is usually an "almost vertical" line in Fig. 1 b. For every illumination intensity there exists a pair of optimal values of Ns and Ne. And if these values are rigidly specified, then a disbalance (since the operation point of the plant does not coincide with the maximum power point of the solar cell) appears from time to time. This disbalance somehow decreases the energy conversion efficiency. This can be well seen from Fig. 2: matching being done to gain highest efficiency for the earliest and the latest parts of the light day, there is a fall in efficiency at mid-day hours. This point is to be discussed later. Some plants "solar cell + electrolyzer" designed and constructed during the last decade, and their characteristics are listed in Table 1, Of these, the HYSOLAR 350 kW plant is the world's first plant for solar hydrogen production on such a large scale. It is erected near Riyadh under the Saudiarabian-German joint program named HYSOLAR. (The program includes, besides constructing solar hydrogen plants in Riyadh, for 350 kW, Stuttgart, for 10 kW, and Jeddah, for 2 kW, also undertaking fimdamental research, system and utilization studies, and staff training [16]). Its scheme is given in Fig. 3 . The photovoltaic power system consists of 160 single pedestal concentrator arrays, each containing 256 circular silicon solar cells (64 modules). Each module contains 4 Fresnel lenses, for a sunlight concentration of ca. 33 times over solar cells. These 64 branches are connected in parallel to provide 3 5 0 kW of peak power. The electrolysis system is a commercially available HS 2000 electrolyzer with the cell area 0.2 m2 (No. of cells: 144). Electrolyte is 30% KOH solution, system pressure: 6 bar, working temperature: 100 O C . The plant also contains a grid operated rectifier for initial start-up and special testing.

41 3

P, $ l h

- 0.15 - 0.1

0-U

6 8

lo 72 7 4 2 , h

Fig. 2 . Comparison of computed (solid lines) and measured (dashed lines) characteristics of the solar-hydrogen plant throughout the day [22]. 1 - hydrogen production rate; 2 sun-to-hydrogen efficiency; 3 incident radiation power density.

-

Solar radiation

so Zar generator,

ElectPo-

lyzer

rl+l I Rectt fier

I

Gas-treatment+ compresszon

U t i 1i zot ion

Auxiliarfes

dDQtQ““9’ + I cont7-0

Fig. 3 . Block-diagram of the 350 kW solar-hydrogen plant at Riyadh (Saudi Arabia). The coupling of the electrolyzer and solar array was either direct, or through some matching device, e.g., the DC-DC converter. The converter efficiency is ca. 90%. Some arguments in favour of either coupling mode will be given below.

474

Table 1. Characteristics of Solar-Hydrogen Plants

Year of manufacture,

Peak output power,

andcountry

kW

Etficiency, % Remarks rls

qe

'IS-H Use is made of a heliostate to track the sun and of a solar concentrator presnel lenses). Electrolyzer is of a filter-press type. Electronic matching device.

1982,

France

1

14

50

The plant design envisages storage of hydrogen. Connection - direct or via a matching device on operational amplifiers.

1982,

Greece 1986,

USSR 1986, USA

8 1

1

6.1

65

3.7-4

10

8

9

Diaphragm-type electrolyzer. Storage of hydrogen.

10

Filter-press type electrolyzer

11

Mobile plant installed on the trailer chassis. Collapsible solar array. Electrolyzer with a solid polymer electrolyte. There are water storage tanks and a water purification (deionization) unit.

1986,

Japan

Ref.

12

1988,

Saudi Arabia

350

8.1

77

1989,

Germany

10

1991,

USA

9.2

10

70

6.3

See text

13

Three different types of matching devices are used. Alkaline electrolyzer;, operation temperature 80- 100 C. Hydrogen stored in standard gas bottles at 201 bar.

14

Alkaline high pressure electrolyzer. Fuel cell (eff. 60%). Air Compressing unit as a load.

I5

Some aspects of the plants' design and operation were discussed in [ 17-19]. Various types of secondary batteries were tested as storage batteries for photovoltaics. Lead-acid battery, as well as nickel-cadmium (alkaline) battery proved most suitable [20,2 I].

475

Solar-Hydrogen Plant: Simulation Principles of simulation have been elaborated by Carpetis [22] who introduced the currentvoltage characteristic of the solar cell in an implicit form, Q, (Is, V,, T, J) = 0 where Is and Vs are the cell output photocurrent and voltage respectively, T is the cell temperature, and J is the solar radiation intensity. Besides four variables the hnction @ contains also five constants which are to be determined from five independent sets of experiments. The electrolyzer characteristic can be taken, for simplicity, as a linear fbnction of voltage. For numerical computation it is also approximated with a polynoms of 5th order. The characteristics generally depend on temperature, which can be made allowance for by using several sets of coefficients for the polynomial expansion, taken for several discrete values of temperature and interpolating the data for intermediate values. Numerical calculations of the performance characteristics have been done for a plant with the numbers of consecutively connected solar cells (N,) and electrolyzers (N,) chosen empirically. In particular the influence of insolation and the cell temperature was investigated in detail. When radiation intensity varies, the maximum power point of the solar cell moves along "almost vertical" line in the photocurrent-voltage coordinates, already shown in Fig. 1 b. In other words, the cell photocurrent is nearly proportional to the radiation intensity. For all radiation levels of practical importance, the "working point" of the plant turned out to deviate no more than 6 per cent from the maximum power point of the solar cell: (ImVm IpVp)/lmVm I 0.06 where Ip and Vp are real values of the current and voltage, while 1, and Vm are the maximum possible values attained for the "ideal" matching. This calls in question profitability of using special devices like DC-DC converter for matching solar cell and electrolyzer because these devices consume up to 10 per cent of the transformed power. On the contrary, variation of the solar cell temperature results in a drastic declination of the plant's working point from the solar cell maximum power point. Indeed, with raising temperature the MPP photocurrent (I,) changes but slightly, yet the cell voltage, Vm (as well as open-circuit photovoltage) falls down markedly. Increase in the temperature within certain limits does not violate "optimal" matching conditions, however, further rise causes sharp drop in the cell output power. Moreover, when temperature has exceeded certain critical values the electrolysis process ceased altogether. The computation program worked out in [22] made it possible to calculate hydrogen production per day and per year for preset solar cell and electrolyzer characteristics and insolation, for any geographical location and climatic conditions. General scheme of computation was as follows. The sunlight flux values recorded during meteorological monitoring were taken for certain instants, and instantaneous current and voltage data were calculated for the intersection point of the solar cell and electrolyzer characteristics using the iteration procedure. The hydrogen production rate was immediately obtained from the current magnitude. To obtain daily production, the light day lenght was divided into many small intervals; the plant output and efficiency were calculated for each interval. Then total amount of produced hydrogen was obtained, to be compared to the theoretical limit for the case of ideally matched solar cell and electrolyzer. Ultimately, the yearly plant capacity was counted up by making the above calculations for all the days with their particular insolation-vs.-time characteristics and

476

integrating the obtained data. Additional sub-program enabled one to calculate cost of the hydrogen production, including storage. Local climate characteristics (e.g., air mass) can be made allowance for either by using (averaged) statistical data, or by monitoring real transparency of the atmosphere and introducing necessary corrections into calculations at every integration step. The computed data were compared with those measured experimentally at a plant composed of solar arrays (Solarex Model 260, peak power 6 W) and an electrolyzer with solid polymer electrolyte (General Electric Model 15 EHG 1A) [22]. The experiment has been performed in Albuquerque, New Mexico, USA (lattitude 35'05') in March, the solar array angle to the horizontal was 47". There was a cloudless day with slight obscurance at the afternoon hours. The temperature increase rate was assumed to be 4.6 'C per hour, the maximum array temperature having been reached at 15 h. The comparison of the calculated efficiency q S-H and instantaneous hydrogen output with those measured experimentally is shown in Fig. 2. The latter reached its maximum at approx. 9 h and levelled off during the middle part of the day (inspite still increasing insolation). So, the decay in the production rate is in all probability due to the array temperature rise, rather than to lower radiation. The plant efficiency measured as well as calculated was minimal at noon, its maximum being at about 9 h. Thus, the remarkable behaviour of the system (levelling of the hydrogen production rate and decrease in the system efficiency during the middle part of the day) is well predicted by the described [22] calculation method, using averaged (meteorological) solar radiation data. Annual hydrogen production for the plant located at the lattitude 35" (warm summer/ /mild -~ winter) amounted at 4.43 kg H2/m2, with peak production rate of 2 ~ l O kg/(m2.hour). Annual mean efficiency was about 7%. For the plant located in Central Europe (lattitude 48') the efficiency was about the same, yet the hydrogen production was about 50% of the prodution at the first location. These results were obtained for fixed solar array facing South with an inclination to the horinzontal equal to the lattitude. Small variation in the inclination (+ 10") proved to influence but slightly the annual hydrogen production. An important result of the study [22] is that it is possible to empirically adjust rather effectively the characteristic of the electrolyzer to the locus of maximum power points of the solar array for a given temperature has been already stated. It was recommended to make this "design adjustement" (for optimal power matching) for highest radiation level at the maximal operating array temperature. The reason is that for higher temperature the plant efficiency diminishes rapidly and the electrolysis cut-off could occur (see above) within a few degrees of hrther temperature increase. On the contrary, at lower temperature the efficiency slightly decreases. (A method of more accurate "design adjustement" will be presented in the next Section.) The same calculation scheme was used [23] to predict the solar hydrogen plant performance from real climat data recorded for two locations: Cape Canaveral, Florida (USA) and Stuttgart (Germany). The experiment was carried out either with a real plant (using single crystalline silicon solar cells) or with a computer-driven electrolyzer (with aqueous alkaline as well as solid polymer electrolyte). Figure 4 presents monitoring of the radiation intensity, the solar array temperature, and hydrogen production rate for a September day at Cape Canaveral. An important result here is that daily variations in the plant efficiency are very small even for days with markedly varying claudiness. Annual variaton is also small especially for the sunrich areas. This means that estimations based on the product of the mean efficiency and radiation intensity are highly satisfactory.

417

0

200

400

600 qmin

Fig. 4. Daily variation of electrolytic hydrogen production rate (l), the solar array temperature (2) and radiation power density (3). Single crystalline silicon solar cells, SPE electrolyzer, location: Cape Canaveral, FA. The time scale denotes minutes elapsed from 5 a.m. [23].

Fig. 5. Accumulated annual specific hydrogen production per 1 m2 of the solar array: 1 - for the real plant, 2 - assuming zero matching losses. Single crystalline silicon solar cell; SPE electrolyzer; location: Stuttgart (Germany); the design parameter t = Ns/Ne = 4 . 3 ~3 I .

478

Again an empirical adjustement of the solar array characteristics to those of the electrolyzer proved to give rather good matching (see Fig. 5). The simulated data conformed rather well to the experimental results. Solar-Hydrogen Plant: Optimization A schematic diagram of the plant "solar array + electrolyzer + secondary battery" is shown in Fig. 6. As already mentioned the plant has a modular structure. Thus, the solar array is a set of Ns series-connected identical units (moddules) whose total area is S,; fo(J,v) is the specific (i.e. per 1 cm2) photoelectrical characteristic of the array. Here J is the irradiation intensity depending on the time of day T. Similarly, Ne is the number of series-connected electrolyzer units, Se is the total area of their electrodes, go(v) is the specified electrical characteristic ; Nb is the number of series-connected battery units, Sb is the total area of their electrodes, ho(h, '5, V) is the specified electrical characteristic of the battery depending on the time and parameter h, characterizing the state of discharge of the battery at the initial instant of time, T ~ Everywhere V stands for the voltage at the electrodes of the corresponding module (battery, electrolyzer, solar array)

Fig. 6 . Schematic diagram of the plant. 1 - solar array, 2 - secondary battery, 3 - electrolyzer. A rectifier is connected into the battery circuit, which hinders the battery discharge through the electrolyzer when the solar array is out of operating.

It should be stressed that areas Ss and Se considered at this stage to be arbitrarily preset (as regards choosing area Sb, see below); the numbers of modules to be connected in series N,, Ne, Nb are not known a priori finding their optimum values (at each instant of plant operation) is the aim of the analysis.

-

,

479

In this Section we shall discuss the performance of a plant consisting of a solar photovoltaic array and a water electrolyzer, whose optimization has been theoretically analyzed in [24]. The optimization of the plant parameters was carried out in three stages. First, the algorithm of design optimization was found for a given solar radiation power density allowing the relation of the number of the series-connected solar cells to that of the electrolysis cells to be determined. Next, the case of the irradiation power density varying with time was considered and the regime of switching of these cells was computed so as to maintain the hydrogen production rate at this maximum level. For the case when no switching was possible the system parameters providing the maximum daily capacity were found. Finally, the ratio of the total array to electrolyzer areas was calculated, which in the previous two stages was chosen arbitrarily. This was calculated on the basis of the relation between the specific costs of these parts of the plant to ensure the maximum capacity per unit of the investment. The current-voltage characteristic of a solar array can be written as

and that of the electrolyzer, as

Dividing the total area of the solar array into Ns series-connected cells gives Is = ( S S ~ S ) f O ( J > V S ~ &

(3)

Here Vs and Is are the voltage and current at the array output. Similarly, dividing the total area of the electrolyzer electrodes into Ne series-connected cells gives

Here Ve and I, are the total voltage and current at the electrolyzer. During the operation of the solar array-electrolyzer system, as mentioned above,

hence it follows

The solution of the optimization problem for the solar array-electrolyzer system consists of finding the Ns and Ne values that would provide, at given values of the areas S, and Se and the solar radiation level J, the maximum hydrogen production rate of the system expressed (in electrical units) by the quantity P = IeNe, It can be readily seen that though the initial Equations (3)-(6) depend on Ne and Ns separately, the optimum depends only on the ratio of these parameters, NsNe. For this purpose Eq. (6) is transformed to

480

where t = (NSMe) and y = VMe. Solution of Eq. (7) defines the function y(t,J,Ss/Se) The maximum of P = IeNe = Sdo(y), i.e., the maximum hydrogen production rate is realized when the solar array operates at its MPP. The expression for the electric power generated by the solar array may be written asW(V)=IsV=Ssfo(V)V. From the condition of the maximum power, dW/dV=O, it follows that fo(v)+vfol(v)

=0

The optimization problem for the case of constant radiation power density was solved using the following algorithm: (i) from the known characteristic of the solar array and by solving Eq. (8), the maximum power value at the array output and the value of y/t are found; (ii) from Eq. (7) the value of T* corresponding to the optimum theoretical NSMe ratio is found (in the general case t* is a transcendental number), along with the value of the maximum theoretical limit of the capacity of the "solar array + electrolyzer" plant; (iii) finally, at a certain given maximum value of (Ns)max the integer values of Ns*,Ne* giving the Ns/Ne relation closest to t* are found and by means of Eq. (6) the corresponding plant capacity values is calculated. (Note that it is evident from general considerations that breaking up of the areas Ss and Se into a very large number of modules is unpractical. Therefore, it is recommended to restrict Ns with certain value (N,), and choose integer values of Ns and Ne which fit as closely as possible the calculated value oft). When using the algorithm described one has to solve transcendental equations only three times, whereas in the alternative method in which sifting of the values of Ns and Ne is used for determining the maximum hydrogen production rate the number of times the solution of Eq. ( 6 ) is resorted to depends on the choice of (Ns)max and may be two orders higher. The computer program used in [24] provides analytical approximations for tabulated fo and go characteristics in the form of 6th order polynomials, and permits Ns* and Ne* values giving the maximum P to be chosen. In computations, has been already pointed out, the maximum admissible value of (Ns)max is preset, thus limiting the search area. It is chosen for convenience in implementing the plant design. This particular computation (just as other computations below in this article) was performed for the plant described in [25], for which Ss= 950 cm2, Ss = 50 cm2. The plant characteristics are shown in Fig. 7. With the chosen value of (Ns)max = 15, the optimum values of Ns and Ne proved to be equal to 9 and 2, respectively. The gain in capacity as compared with the design parameters Ns = 4 and Ne = 1 used in [25] was over 7% (compare the points of intersection of Curve 2 by Curves 1 and 4). The second stage in the solution of the optimization problem was computation of the outimum design parameters at varying (e.g.. during day light) irradiation power density at the solar cells surface. The area Ss and Se and the dependence J(T) (where T is the time) are preset. In the general case, variation in the solar radiation during the day in the absence of cloudiness depends on the latitude of the locality and the date.

481

1-

Fig. 7. 1, 2 - Characteristics of solar array and electrolyzer, respectively, for the plant described in [25]; 3 - maximum power curve for the solar array; 4 - the dependence of I Ne on VMe corresponding to the optimum system (N, = 9, Ne = 2); - the maximum power points for characteristics 1 and 4 [24].

The problem consisted of computing the dependences

and

In the computations, which were only of an illustrative nature, it was assumed for simplicity that fo(J,v) depends linearly on the irradiation power density J. Taking into consideration the more complex dependences off, on J presents no fbndamental difficulties and can be carried out of analytical approximations of the tabulated fo(J,v) characteristics for several different irradiation levels. The t*(T) curve shows how the automatic tracking servo system vary the design parameters Ns and Ne in the course of the day. The calculated t*(T) dependences and also the J(z) curve used in the computation are shown in Fig. 8. In computations the value of (Ns)ma = I5 was again taken. The instants when the system should be switched over to the new optimum parameters Ns/Ne according to the scheme (Ns/Ne)opt

= 411

+ 1313 -+

912 -+ 1313 + 411

were determined from Curve 3. It follows from the results of this computation that with a chosen limited number of series-connected solar cells (Ns)max = 15, to provide the computed switching regimes the plant must actually consist of 4 x 13 x 9 = 468 separate solar cells and 1 x 3 x 2 = 6 separate electrolysis cells.

482

7

Fig. 8. Optimization of the system for varying solar radiation power density: 1 - dependence of the relative radiation power density J/Jmax on the time of day; 2 - dependence of the plant capacity P on t (at smoothly varying t); 3 - dependence of the optimum ratio t* = Ns/Ne on T. Dashed lines separate the regions with different optimum sets of Ns and Ne [24].

A version of this optimization problem was to compute the optimum values o f t * and (at given (N,),, of Ns and N e in the absence of an automatic tracking system for switching over the design parameters. The optimality criterion was the maximum daily integral hydrogen production by the plant,

Tmax - Tmin

’ tmin

An example of the computed (P(t)) dependence is shown in Fig. 9. It is clear that the maximum daily capacity is achieved at NS/Ne = 9/2, which is the solution of the optimization problem for the maximum daily radiation power density J found above. This is not surprising since it is the lightest time of day that makes the main contribution to the integral hydrogen production of the plant. One can judge from above discussion (see the preceding Section) that losses in the plant capacity as compared to the above case o f continuously switching over the plant parameters does not exceed 10 per cent.

483

6-7.

4

4.5

5t

Fig. 9. Dependence of the integral plant capacity per day on the ratio t = Ns/Ne [24] The optimum economic characteristics of the “solar array + electrolyzer” system were computed as the third step in solving the optimization problem. The condition of the maximum plant capacity referred to the cost of the system (i.e., to the investment) was used as an optimality criterion. The cost of the system was calculated conditionally from that of the unit area of the solar cell us and the cost of the electrolyzer per unit area of its electrodes us. Then

The plant capacity defined by Eq. (12) was calculated as the optimum capacity (in the sense that Ns and Ne were fitted to the values of Ss and Se) at a certain given solar radiation power density J. The F(Ss/S,) dependence found at given us and Ue has a maximum at a certain (Ss/Se)*. The optimum value of (Ss/Se)* and, accordingly, that of F((Ss/Se)*) depends on the current price of the solar photovoltaics and that of electrolyzer. An example of the computed dependence F(Ss/Se) is given in Fig. 10.

2

20 30 Ss/Se

Fig. 10. Dependence of the optimum specific capacity of the plant per unit investment on the ratio of the areas SslSe at ue/us = 2 [24].

484

Here the value of ue/us was assumed to be 2. It is clear that at such a ue/us value, the minimum investment and hence the minimum hydrogen cost for the plant described in [25], are reached not at the ratio of the surface areas, SSlSe, equal to 950150 = 19, as has been arbitrarily chosen, but at Ss/Se = 25. The current prices depend on the type of electrolyzer or solar cells, and vary quickly with improvements in technology and changes in the market situation. As an illustration, the dependence of the optimum ratio of the total areas of solar array and electrolyzer on the ratio of specific costs ue/us was computed (see Fig. 11) for the same plant [25] by which means it is possible to predict the design parameters of the plant depending on the market prices.

Fig. 11. Dependence of the optimum ratio of the areas of solar array and electrodes of the electrolyzer on the ratio of specific costs of electrolyzer and solar array [24]. It should be noted that this approach is applicable for optimization not only of the cost, but also of any other characteristic of the plant (weight, in particular), that can be taken to be proportional to the solar array and the electrolyzer areas. Optimization algorithms slightly different from the described above were developed in [26, 271. In particular, experimentally measured characteristics for the HYSOLAR plants in Kiyadh and Stuttgart (see the preceding Sections) were used [27] for simulating the optimized plant. Suitable dimensionless parameters were introduced characterizing solar-hydrogen plant configurations, and a general curve for optimally adapted configurations of the plant has been derived. Neither re-distribution of the solar array and electrolyzer modules, nor power conditioning devices (like the above-mentioned DC-DC converter) were used in designing the optimized plant. Nonetheless, annual hydrogen productivity for a directly coupled plants so designed was reasonably close to the theoretical limit, i.e., to the productivity of a plant operating under the condition that solar array constantly operates at its maximum power point. Optimization of the Plant "Solar Array + Electrolyzer + Secondary Battery" Optimization methods for non-steady-state operation conditions of the solar energy conversion plant were studied in [28]. The aims of this investigation were: (1) to look into the principles of interaction between the individual components of the plant and to compute the regime of their operation for some particular practical cases; ( 2 ) to find the relation between

485

the amount of energy stored in the battery to that “in hydrogen” at arbitrary battery and electrolyzer dimensions and also to establish the limits within which these amounts can be varied in practice, choosing for this purpose the optimum operating conditions and the plant design parameters. The total currents passing through the components of the plant (See Fig. 6) are given by Eqs. (3) and (4) for the solar array and electrolyzer respectively, and by

for the secondary battery; Vb is the voltage at the battery. Taking into consideration the scheme of connecting the system components shown in Fig. 6, we can now interrelate the currents I,, I, and Ib and voltages V,, Ve and Vb by the formulas:

(cf Eq. ( 5 ) ) . Substituting Eqs. (3), (4), (13) into Eq. (14) and using Eq. (15), one has

The battery can be charged either at given constant voltage or at constant current conditions (or by combining the two regimes). For practical calculations charging of a secondary battery at constant voltage was chosen in [28]. Each cell is under fixed voltage vb. Thus quantities V and Nb are linked by the relation V/Nb = y,= constant. The state of the system at given specific characteristics fo, go and ho and areas S,, S, and s b is characterized by three design parameters N,, Ne and Nb which can be interrelated with the dimensionless parameters t = Ns/Ne and r = Ns/Nb and a variable y = V/Ne. Then the basic Equation (16) defining the conditions of plant operation can be represented as

The argument y,of the function ho is omitted here since vb (as pointed out above) is a fixed quantity. This condition leads to a strict interrelation between y, rand t:

The aim of the consideration is to compute the optimum conditions of the plant operation regime, whereby the array operates at the maximum power point:

486

Is V = max

(19)

For simplicity, the specific characteristic of the solar array was assumed to be proportional to the irradiation intensity. Condition (1 9) leads to the equation

-

-

fo (v> + (v) fo' (v)= 0, Ns Ns NS Y

where fo (v) is the specific characteristic of the solar array recorded at the maximum

NS irradiation intensity J = Jmax (where J (7) is the irradiation intensity depending on the time of the day). The solution of Eq. (20) is a certain value of

Thus, when the plant operates under optimum conditions Eq. (17) defines the fhnction

on which two constraints (18) and (21) are placed; the first of them is caused by the battery charging regime chosen and the second by the condition of optimum system operation (solar array in the maximum power regime). Conditions (18) and (21) unambiguously define the value of design parameter r:

The value of the second parameter t, however, just as the quantity y varies with time, first, due to the fact that the charging characteristic of battery ho (A, z) is a fhnction of time and, second, because of the time dependence of the irradiation intensity of the solar array J(z). Using relation (21), one can express y in Eq. (17) in terms o f t : y = vs*t, and obtain an equation for determination of t(z), i.e., the optimum behaviour in time of the design parameter O\rsme)opt:

where f o is the instant of switching on charging of the battery.

481 Numerical calculation for the particular plant. The investigation of the system behaviour involves numerical solution of Eq. (24), which requires detailing of the fimctions contained in it. namely J(z), fo (vs*), ho(h, z), and fo (v). For numerical calculations below the fo and go characteristics measured experimentally in 11251 were taken, as in previous Section. An acid (lead) battery was chosen for delineation of the ho characteristic. Figure 12 shows the charging curve for a model negative electrode of the lead battery (whose area may be called the area of battery electrodes, Sb) at a given voltage vb = 2.4 V; the initial state of discharge was equal to 0.9.

I

0

I

2

3

r

1

4 'E;h

Fig. 12. Charging curve of the negative electrode of a lead storage battery at constant voltage of 2.4 V. The electrode area is 1.62 cm2. The state of discharge h is 0.9.

The following specific features of the charging process can be mentioned, which are important for calculations: (i) the initial charging current depends little on the initial state of charge of the battery; this current value defines the maximum area of the battery electrodes, Sbmax, which can be charged from the particular solar array; (ii) with the regime chosen, the battery is practically completely charged in 2 hours; (iii) at the end of the charging process the charging current reaches a constant value, which is about 4% of the current at the initial instant. On this plateau battery charging does not undergo any hrther change, and the electrical power supplied is wasted (consumed in the electrolyte decomposition). It should be emphasized that the plant must operate under optimum conditions: the solar array is at the maximum power point, and the voltage at each battery module is vb, for which purpose the optimum values of design parameters r and t should be maintained. The charging curves calculated when the whole battery area is switched on for charging on one occasion are shown in Fig 13. Charging of the battery (to be more precise, of a set of Nb series-connected modules) of the maximum total area (Sbmax = 28 cm2) stars at 12:OO h and ends at the instant when "the tail" of the battery charging curve intersects the solar array current curve (about 17.00 h, Curve 2).

488

t

Fig. 13. Charging of the battery set total area on one occasion. 1 - solar array current; 2 charging curve of battery with total area of 28 cm2; 3 - t ( 7 ) curve for the battery with total area 28 cm2; 4 - the same in absence of battery [28]. (To facilitate comparison of the operation of solar array and battery under different conditions in Figs. 13-1 5 are given not their real currents Is and Ib (at certain optimum Ns and Nb), but the quantities IsNs and IbNb, respectively). Let us follow the time behaviour of design parameter t. In absence of battery, at the solar array and electrolyzer characteristics chosen, during the day t assumes values up to 4.49 depending on the solar array current Is (T), passing through a maximum at the highest irradiation intensity at noon (Curve 4). If, however, a part of the solar array current is taken off for charging the battery, then the value o f t decreases during its whole charging time (Curve 3). This is not surprising since for the design optimization of the subsystem "solar array + electrolyzer" only the net current is of importance, which is used in the operation of the electrolyzer, i. e., Is - It,. Comparison of the areas under Curves 1 and 2, Fig. 13 shows that when battery even of maximum possible area is switched on for charging at one occasion, it is possible to use for this purpose only an insignificant part (less than 10 per cent) of the amount of electricity produced by the solar array during the light day. Analysis of the operating conditions of a plant with successive switching on of the battery sets. Since the charge received by a set of battery modules of the total area equal to Sb is relatively small as compared to the total amount of electricity produced by the solar array in one light day (even for a battery of a maximum permissible area Sbmax), it was of interest to investigate the operating conditions of a system whereby new sets of batteries of the same type (with the same Sb values) were successively put into operation.

489

For calculation of the operating conditions of such a system the generalized equation (24) was used:

-

Se t go (Vs*, t) = Ss J (T) Jmax

mb (z) fo (Vs*) - s b ( v h C ho (T Toi) vs i=l

(25)

Here zoi is the instant of switching on the i-th battery set and mb ( 5 ) is the number of battery sets switched to the system which depends on the elapsed time, the other designations are the same as used above. In the general case the condition of switching on of the i-th battery set can be written as follows i- 1 C J (toi) fo (vS*) Sb Ss ho (toi - ~ o j =) ( b, .) ho (0) (26) VS Jmax vs j=l

-

Figures 14 and 15 show the results of calculation of the time behaviour of the system under consideration, whose evolution was described by Eq. (25) and the instants of switching on of the battery sets were given by condition (26).

t

/

10

5Y!

06

8

10

12

14

76

Fig. 14. Successive switching on of battery sets, Sb = 21 cm2. 1 - solar array current; 2 battery charging curve; 3 - t (T) curve; 4 - the same in absence of battery. Charging time of each set is limited to two hours [28].

490

I,A 70 -

5-

a

b

Fig. 15. Successive switching on of battery sets, Sb = 14 cm2. The charging time of each set is limited to two hours (a), unlimited (b) [28]. Curves are numbered as in Fig. 14. Thereby the area Sb was taken arbitrarily (0.75and 0.5 of Sbmax), It is evident that the charging of the battery system is described by a jagged curve inscribed into the solar array current curve. Analysis of the results leads to the following conclusions: a) Limiting the charging time of each module to two hours (as been already pointed out, battery thus realizes up to 95% of its charging capacity) increases considerably the charge accepted. Comparison of a and b in Fig. 15 shows that it is thus possible to increase the number of charged sets (sb = 0.5Sbmax) from 8 to 1 1 . Further calculation for the sets, e.g., s b = 0.1 SbmaXshowed increase from 64 to 98 sets, i.e., 1.5 times. This is due to the cutting off of ballast "tails" of the charging curve (see Fig. 1). These tails, accumulating as new and new battery sets are switched on, add up to form a significant, as can be seen from Fig. 15 b, ballast current taking off an appreciable portion of the solar array current. b) With decreasing total area of the set of battery modules Sb relative to Sbmax, the number of these sets charged in one light day mb increases rapidly (Fig. 16, Curve 1). Thereby the product (mbSb)/Sbmax, and together with it the total capacity of battery, also increase (Curve 2). The amount of electricity stored by the battery asymptotically approaches the limit the total amount of electricity produced by the solar array during a day light (dashed line in Fig. 16). This increase is due both to the approach of the battery charging curve Ib(T) to the solar array current curve is(^), and also to earlier switching on and later switching off the battery system. Thus, at Sb = 0.1 Sbmax the contribution of the battery to the converted solar energy stored is 94%, and the contribution of the electrolyzer is 6%. But, as we see, in this case as well the use of an electrolyzer in a solar energy conversion plant is of advantage since it permits an increase in efficiency, if only by a few per cent.

49 1

5

0 1

10 S Y / S b

Fig. 16. Dependence of the number of battery sets charged (1) and the amount of electricity stored in them (2) on Sbma/Sb. Dashed curve - total amount of electricity produced by solar array in a light day [28].

Thus, the conditions of time operation of the plant can be described by two dimensionless parameters: one of which, r = Ns/Nb, is constant and the second, t = Ns/Ne, varies during the operation (though not in wide limits). Thereby, although the calculated value of t varies continuously, N,, Ne and Nb are, naturally, integers. Therefore, in working out the practical design of the plant, the user, reasoning from his objectives, first decides how to distribute the accumulation of the electrical energy received from the solar arrays between the electrolyzer and storage battery, and using the plot of Fig. 16, he finds the "degree of breaking up" of the maximum permissible battery area (i.e,, Sbmax/Sb). The next problem is to determine explicitely the values of three design parameters N,, Ne and Nb (all of which generally are functions of time T). It was already mentioned that breaking up of the total areas S,, Se and Sb into a very large number of modules is unpractical. It is, therefore, recommended to select a certain constant value o f t and to look for integer values of N,, Ne and Nb which would fit as closely as possible the calculated values of r and t (thereby the efficiency of the plant, of course, would be somewhat lower than theoretical). Thus, for example, for the specific characteristics fo, go and ho used in the above calculations assuming values of r z 7.2 and t I 4.3, the following combinations of N,, Ne and Nb could be chosen:

or

*S

Ne

Nb

r

t

1548

360

215

7.200

4.300

154

36

21

7.333

4.278

Thus, the plant control unit has but to switch battery sets for charging in compliance with the compiled program.

492

REFERENCES 1 Bockris J 0' M. Energy: The Solar-Hydrogen Alternative. Sydney: Australia and New Zealand Book Company, 1975. 2 Bockris J 0' M. Energy Options. Real Economics and the Solar-Hydrogen System. Ibid., 1980. 3 Zahed AH, Bashir MD, Alp TY, Hajjar YSH. Int J Hydrogen Energy 1991; 16: 277-281. 4 El-Osta WB, Veziroglu TN. Ibid. 1990; 15: 33-44. 5 Winter CJ, Nitsch J. hid. 1989; 14: 785-796. 6 Bilgen E, Bilgen C. Ibid. 1983; 8: 441-451. 7 Salamov OM, Bakirov MYa. Geliotekhnika 1986; Nr. 1: 43-46. 8 Esteve D, Ganibal C, Steinmetz D, Vialaron A. Int J Hydrogen Energy 1982; 7: 7 1 1-7 16 9 Koukouvinos A, Ligerou V, Koumoutsos N. Ibid. 1982; 7: 645-650. 10 Bychkovskii SK, Konev VG, Negreev BM, Strebkov DS, Samojlova LA, Svinarev SV, Trushevsky SN. Geliotekhnika 1986; Nr. 4: 29-33. 1 1 Hancock OG. Int J Hydrogen Energy 1986; 11 : 153-160. 12 Morimoto Y, Hayashi T, Maeda Y. In: Hydrogen Energy Progress VI, New York: Pergamon Press, 1986; 1: 326-332. 13 Steeb H, Weiss HR, Khosaim BH. Ibid.: 406-413. 14 Brinner A, Siegel A. Proc European Photovoltaic Conference (25-29 September, 1989 Freiburg). 15 Lehman PA, Chamberlin CE. Int J Hydrogen Energy 1991; 16: 349-352. 16 Steeb H, Aba Oud H, Brinner A, Grasse W, Hansen J. In: Hydrogen Energy Progress VII, New York: Pergamon Press, 1988; 615-625. 17 Costogue EN, Yasui RK. Solar Energy 1977; 19: 205-210. 18 Dini D. Int J Hydrogen Energy 1983; 8: 897-903. 19 Salamov OM, Bakirov MYa, Rzaev PF. Geliotechnika 1988; Nr. 1: 67-71, 20 Shimizu K. JPower Sources 1987; 19: 211-214. 21 Jivacate C. Ibid. 1989; 28: 181-186. 22 Carpetis C. Int J Hydrogen Energy 1982; 7: 287-310. 23 Carpetis C. Ibid. 1984; 9: 969-991. 24 Kharkats YuI, German ED, Kazarinov VE, Pshenichnikov AG, Pleskov YuV. Ibid. 1986; 1 1 : 617-621. 25 Pleskov YuV, Zhuravleva VN, Pshenichnikov AG, Vartanyan AV, Arutyunyan VM, Sarkisyan AG, Melikyan VV. Geliotekhnika 1985; Nr. 4: 61-64. 26 Salamov OM, Bakirov MYa, Rzaev PF. Ibid. 1987; Nr. 3: 15-19. 27 Siegel A,, Schott T. Int J Hydrogen Energy 1988; 13: 659-675. 28 Kharkats YuI, Pleskov YuV. Ibid. 1991; 16: 653-660.

SECTION SIX

MISCELLANEOUS ENVIRONMENTALLY ORIENTATED ELECTROCHEMICAL PROCESSES

This Page Intentionally Left Blank

495

ELECTRODIALYTIC MEMBRANE PROCESSES AND THEIR PRACTICAL APPLICATION H. Strathmann, University of Twente, Faculty of Chemical Technology, Enschede, The Netherlands

INTRODUCTION The desalination of brackish water by electrodialysis and the electrolytic production of chlorine and caustic soda are the two most popular processes using ion-exchange membranes. There are, however, many other processes such as diffusion dialysis, Donnan dialysis, electrodialytic water dissociation, etc. which are rapidly gaining commercial and technical relevance. Furthermore ion-exchange membranes are vital elements in many energy storage and conversion systems such as batteries and fuel cells. Although the large scale industrial utilisation of ion-exchange membranes began only 20 years ago, their principle has been known for about100 years [I]. Beginning with the work of Ostwald in 1890, who discovered the existence of a "membrane potential" at the boundary between a semipermeable membrane and the solution as a consequence of the difference in concentration. In 191 1 Donnan [2] developed a mathematical equation describing the concentration equilibrium. The first use of electrodialysis in mass separation dates back to 1903, when Morse and Pierce [3] introduced electrodes into two solutions separated by a dialysis membrane and found that electrolytes could be removed more rapidly from a feed solution with the application of an electrical potential. With the advent of ion-selective membranes it became feasible to transport ions against a concentration gradient. In 1940 Meyer and StrauB suggested a multicell arrangement, in which anion-selective and cation-selective membranes were installed in alternating series between two electrodes [4]. With such multi compartment electrodialysers, demineralisation or concentration of solutions could be achieved in many compartments with only one pair of electrodes. Thus the irreversible energy losses due to the decomposition potentials at the electrodes could be distributed over many demineralizing compartments and thus be minimised. With the development of highly selective ion-exchange membranes of low electric resistance in the late 40's by Juda and McRae [S] electrodialysis rapidly became an industrial process. The main use envisaged for ion-exchange membranes was in electrodialysis for the desalination of brackish water. In Japan electrodialysis was used also for concentrating sodium chloride from sea water to produce table salts [ 6 ] . A significant step towards the efficient application of electrodialysis was the introduction of a new operating mode referred to as electrodialysis reversal by Ionics. In this operation mode the flow streams and the polarity in an electrodialysis stack is reversed in certain time intervals [7] and membrane fouling and scaling can be reduced to a minimum.

496

With the development of a chemically extremely stable cation-exchange membrane based on sulfonated poly-tetra-fluor-ethyleneby Du Pont in the late 60 's the chlor-alkali ne membrane electrolysers were introduced [81. The practical use of bipolar membranes for the recovery of acids and bases from the corresponding salts by electrodialytic water dissociation in the early 80's by Liu et al. [9] opened a multitude of new applications in chemical industry and in waste water treatment. The combination of electrodialysis with conventional ion-exchange technology and the use of conducting spacers are both commercially and technically very interesting variations of the basic process [ 101. Diffusion dialysis through anion exchange membranes is used today on large scale to recover acids from pickling solution. Donnan dialysis is used for water softening or for recovering organic acids from their salts. PROPERTIES AND STRUCTURES OF ION-EXCHANGE MEMBRANES The most important parts in any electrodialitic process are the ion-exchange membranes. Their properties determine to a very large extent the technical feasibility and the economics of the processes. Therefore some fundamentals concerning the properties and structures of ion-exchange membranes shall be discussed. 1. Properties of ion-exchange membranes Ion-exchange membranes are ion-exchange resins in film form. They consist of highly swollen gels carrying futed positive or negative charges. There are two different types of ionexchange membranes: (1) cation-exchange membranes which contain negatively charged groups fixed to the polymer matrix, and (2) anion-exchange membranes which contain positively charged groups fixed to the polymer matrix.

negative fixed ion

Fig. 1:

~

negative co-ion

@positive counter-ion

Schematic diagram illustrating the structure of a cation-exchange membrane and the distribution of fixed and mobile ions in membrane matrix

In a cation-exchange membrane, the fixed anions are in electrical equilibrium with mobile

491

cations in the interstices of the polymer, as indicated in Figure 1, which shows schematically the matrix of a cation-exchange membrane with fixed anions and mobile cations, the latter are referred to as counter-ions. In contrast, the mobile anions, called co-ions, are more or less completely excluded from the poylmer mamx because of their electrical charge which is identical to that of the fixed ions. This type of exclusion is called Donnan-exclusion in honor of his pioneering work [2]. Due to the exclusion of the co-ions, a cation-exchange membrane permits transfer of cations only. Anion-exchange membranes carry positive charges fixed on the polymer matrix. Therefore, they exclude all cations and are permeable to anions only. Thus the selectivity of ion-exchange membranes results from the exclusion of co-ions from the membrane phase. For a cation-exchange membrane in a dilute solution of a strong electrolyte, the concentration of the cations is generally higher in the membrane than in the solution, because the cations are attracted by the negatively charged fixed ions of the cationexchange membrane. On the other hand, the concentration of mobile anions is higher in the solution than in the ion-exchange membrane. Thus concentration gradients are established between the membrane and the solution. These gradients act as driving forces for the mobile cations to move into the solution and the mobile anions to move into the membrane. Because electroneutrality is required the permeation of cations into the solution and of anions into the cation-exchange membrane leads to a counteracting space charge due to uncompensated ionsand an equilibrium is established between the attempt of diffusion on one side and the establishment of an electrical potential difference on the other. This electrical potential difference between an ion-exchange membrane and an adjacent salt solution is referred to as Donnan potential. It can be calculated but not be measured directly. The electrical potential gradient and the concentration profiles of mobile cations and anions in a cation-exchange membrane and a dilute adjacent salt solution is illustrated in Figure 2. The thermodynamical treatment by Donnan and Guggenheim [ l l ] is based on an equilibrium between the membrane phase, indicated by superscript M, and the outer phase, indicated by superscript 0 of the electrochemical potential Tli of all ions which are able to permeate through the

In both of the adjacent phases, the concentrations as well as the osmotic pressure and the electrical potential can be different. For the distribution of a specific ion, an established electrical potential difference (PM - (PO, the Donnan potential A q ~ ~ ~ be c adescribed n as a function of the different activities ai and the swelling pressure Ki [l8]:

where Zi is the valency of the ion species i, F the Faraday's constant, R the gas constant, T the absolute temperature, Vi the partial molar volume of the component i, and Xi the swelling

498 pressure. The numerical value of A ~ is negative D ~ for ~ the cation-exchange membrane and positive for the anion-exchange membrane. The Donnan potential cannot be determined by direct measurement, however, it can be used for the calculation of the distribution of the mobile ions between the solution and the membrane and hence for the determination of the membrane permselectivity

.

Membrane

Solution

)Qoo I

Q

10 1

8

0 Mobile Anions Q Mobile Cations - Fixed Charges

%

-Fixed

Charges Concentration Cation Concentration 0 Salt Concentration /

-Anion

concentration

Directional Coordinate

b

L irectiona oordinate

- t

Fig. 2:

Schematic diagram illustrating the electrical potential gradient and the concentration profiles of mobile cations and anions in a cation-exchange membrane and a dilute adjacent salt solution

The swelling pressure is proportional to the concentration of the fixed ions and inversely proportional to the concentration of the electrolyte. In ion-exchange membranes with high fixed ion concentrations and dilute solutions it can reach very high values well in excess of 100 bars [13]. The relationship between the electrolyte concentration in the solution and that in the membrane can be obtained by assuming chemical equilibrium between the two phases and electro neutrality in both the solution and the membrane. For negligible small pressure

499

differences between the membrane phase and the outer solution, and for membranes with a high fixed charge density compared to the salt concentration in the outside solution the coion concentration in the membrane for a monovalent salt, such as sodium chlorine, can be calculated to a first approximation by [ 141:

where Cco is the co-ion concentration, and COthe electrolyte concentration, Cfixed is the concentration of the fixed ions in the membrane and are the average activity and coefficients of the salt in the solution and in the membrane, respectively. This fundamental equation is based on the theories of Teorell and Meyer and Sievers [15,16]. However, the more complex structure of modem membranes cannot be adequately described exclusively by this theory [12]. The remaining differences in the observed and expected membrane behaviour are mainly due to a non-uniformity in the distributions of molecular components in the membrane. This results from structural irregularities on a molecular level and from the influence of the elecmc field. Additionally, the practical application of thermodynamics is rather limited by the difficulties in the experimental rneasm'ment of independent interaction, diffusion, resistance, and frictional coefficients. The Donnan exclusion equilibrium and thus the membrane selectivity depend on: (1) the concentration of the fixed ions, (2) the valence of the co-ions, (3) the valence of the counterions, (4) the concentration of the electrolyte solution, and (5)the affinity of the exchanger with respect to the counter-ions. Important parameters for the characterization of ion-exchange membranes are the density of the polymer network, hydrophobic and hydrophilic properties of the matrix polymer, the distribution of the charge density, and the morphology of the membrane itself. All these parameters do not only determine the mechanical properties, but they also have a considerable influence on the sorption of the electrolytes and the non-electrolytes and therefore on the swelling [131. The most desired properties for ion-exchange membranes are: High permselectivity - an ion-exchange membrane should be highly permeable to counter-ions, but should be impermeable to co-ions. Low elecmcal resistance - the permeability of an ion-exchange membrane for the counter-ions under the driving force of an electrical potential gradient should be as high as possible. Good mechanical and form stability - the membrane should be mechanically strong and should have a low degree of swelling or shrinking in transition from dilute to concentrated ionic solutions. High chemical stability - the membrane should be stable over a pH-range from 0 to 14 and in the presence of oxidizing agents.

It is difficult to optimize the properties of ion-exchange membranes because the parameters determining the different properties often have opposing effects. For instance, a high degree of crosslinking improves the mechanical strength of the membrane but also increases its electrical resistance. A high concentration of fixed ionic charges in the membrane matrix leads to a low electric resistance but, in general, causes a high degree of swelling combined with poor mechanical stability. The properties of ion-exchange membranes are determined by two parameters, namely, the basic polymer matrix and the type and concentration of the fixed ionic moiety. The basic polymer matrix determines to a large extent the mechanical, chemical and thermal stability of the membrane. Very often the matrix of an ion-exchange membrane consists of hydrophobic polymers such as polystyrene, polyethylene or polysulfone. Although these basic polymers are insoluble in water and show a low degree of swelling, they may become water soluble by the introduction of the ionic moieties. Therefore, the polymer matrix of ion-exchange membranes is very often cross linked. The degree of cross linking then determines to a large extent the degree of swelling, and the chemical and thermal stability, but it also has a large effect on the electrical resistance and the permselectivity of the membrane. The type and the concentration of the fixed ionic charges determine the permselectivity and the electrical resistance of the membrane, but they also have a significant effect on the mechanical properties of the membrane. The degree of swelling, especially, is affected by the concentration of the fixed charges. The following moieties are used as fixed charges in cation-exchange membranes:

In anion-exchangemembranes fixed charges may be:

These different ionic groups have significant effects on the selectivity and electrical resistance of the ion-exchange membrane. The sulfonic acid group, e.g., - S q is completely dissociated over nearly the entire pH-range, while the carboxylic acid group -COO-, is virtually undissociated in the pH-range < 3. The quaternary ammonium group -R3N+ again is completely dissociated over the entire pH-range, while the primary ammonium group -NH3+ is only weakly dissociated. Accordingly, ion-exchange membranes are referred to as being weakly or strongly acidic or basic in character. Most commercially available ionexchange membranes have -SO3- or -COO- groups, and most anion-exchange membranes contain -R3N+ groups [ 121.

2. Structures of ion-exchange membranes The structures of ion-exchange membranes are closely related to those of ion-exchange

50 1

resins. As with resins, there are many possible types with different polymer matrixes and different functional groups to confer ion-exchange properties on the product. Most commercial ion-exchange membranes can be divided, according to their structure and preparation procedure, into two major categories, either homogeneous or heterogeneous membranes. Homogeneous ion-exchange membranes are produced either by polymerization of functional monomers, e.g., by means of polycondensation of phenolsulfonic acid with formaldehyde [ 171, or by functionalizing of a polymer film by sulfonation [18]. The polycondensation with formaldehyde according to the following reaction scheme was one of the first procedures for making cation-exchange membranes : OH

Phenol is treated with concentrated H2SO4 at elevated temperatures which leads to the phenolsulfonic acid in para form. This acid is reacted with a solution of formaldehyde in water. The solution is then cast into a film which forms the membrane after cooling to room temperature. Excess monomer is removed by washing the film in water. The polymerizing styrene and divinyl benzene and its subsequent sulfonation according to the following reaction scheme [ 191 is widely used for the preparation of commercial cationexchange membranes:

For the preparation of anion-exchange membranes positively charged groups are introduced into the polystyrene by chloromethylation and amination with mamine according to the following reaction scheme:

502

There exist numerous references in the literature for the preparation of ion-exchange membranes by polymerization [19, 20, 211. In recent years, membranes based on perfluorocarbon-polymershave proved to be very useful in the chlor-alkaline industry [22]. They were introduced fist by DuPont as Nafion@ [23] and are prepared according to the following reaction scheme in a several-stepprocedure:

This intermediate is reacted with hexafluoropropylene oxide to produce a sulfonyl fluoride adduct.

By heating with sodium carbonate the sulfonyl fluoride vinyl ether is formed which is then copolymerized with TFE.

*

-(CF2-CF,)n-

c-

0

F-CF2-

0

( Cq

- CF- 0 a-

II

- CF;- S - F

I CF3

The resulting copolymer is extruded as a film and finally the ionogenic moiety is converted

503 to membranes which carry sulfone-groups as the charged moieties by reacting the -S02F groups with sodium hydroxide. The lifetimes membranes under the aggressive conditions of chlor-alkaline process are in the range of 3 years. Homogeneous ion-exchange membranes can also be prepared by the introduction of anionic or cationic moieties into a performed film. Starting with a film makes the membrane preparation rather easy. A typical example for this mode of preparing ionexchange membranes is the sulfochlorination and amination of polyethylene sheets according to the following reaction scheme [24]:

-

-

+ so, + CI

7 +

2NaOH

SQa

- HGI

- NaCI, - HO ,

S W

sQ- Na'

and

+ H,N-CI-$-N"' SQCI

L

'CH3

7 SQ-NH-CHyN H 3 d 'c%

Heterogeneous ion-exchange membranes consist of fine colloidal ion-exchange particles embedded in an inert binder such as polyethylene, phenol resins, or polyvinyl-chloride. Such membranes can be prepared simply by calandering ion-exchange particles into an inert plastic film [13].Also, ion-exchange particles can be dispersed in a solution containing a film-forming binder, and then the solvent is evaporated to give the ion-exchange membrane. Heterogeneous membranes with useful low elecmcal resistances contain more than 65% by weight of the cross-linked ion-exchange particles. Since these ion-exchange particles swell when immersed in water, it has been difficult to achieve adequate mechanical strength and freedom from distortion combined with low elecmcal resistance. In general, heterogeneous ion-exchange membranes have relatively high electrical resistances. Homogeneous ionexchange membranes have a more even distribution of fixed ions and often lower electrical resistances.

504

3. Special h p e r t y Ion-exchange Membranes In the literature, there are numerous methods reported for the preparation of ion-exchange membranes with special properties, e.g., to be used for the production of table salt, as battery separators, as ion-selective electrodes, or in diffusion and Donnan dialysis. Significant effort has also been concentrated on the development of anion-exchange membranes with low fouling tendencies. 3.1 Monovalent ion permselective membranes For the production of table salt by concentration of sea water monovalent cation selective membranes were prepared by forming a thin positively charged layer on the surface of a cation-exchange membrane. Monovalent anion permselective membranes have a thin highly cross-linked layer on the membrane surface have also been developed [25]. By such means the selectivity of sulfate compared to the one of chloride can be reduced from about 0.5 to about 0.01 and of magnesium compared to sodium from about 1.2 to about 0.1.

3 . 2 Anion-exchange membranes of high proton retention By means of traditional membranes, it is not possible to apply electrodialysis in the recovery of acid in order to reuse the acid because of high proton leakage through the anion-exchange membranes. In general, since protons permeate easily through an anion-exchange membrane, acids can not be concentrated to more than a certain level by electrodialysis with high efficiency. Recently developed membranes, however, exhibit low proton permeabilities and enable efficient acid concentration [25]. 3.3 Anti-fouling anion-exchangemembranes The anion-exchange membrane is more sensitive to fouling. The permeability of commercial anion-exchange membranes is limited in practical electrodialytical separations to components having a molecular weight of less than 100 Da [26]. A molecular weight of 350Da is to be considered as a maximum for any component to be transport through regular commercial membranes. Fouling of anion-exchange membranes often occurs when the anion is still small enough to penetrate into the membrane structure, but its mobility is so poor that the membrane is virtually blocked. To overcome this problem membranes were developed which are characterized by a high permeability for large organic anions. In general, the permselectivity of these membranes is lower than that of regular membranes. A membrane which is less sensitive to traces of detergents is commercially available today from Ionics. Another type of anti-fouling anion-exchange membrane is produced by Tokuyama Soda. The membrane is coated with a thin layer of cation-exchange groups causing electrostatic repulsion of organic molecules.

3 . 4 Bipolar membranes Bipolar membranes have recently gained increasing attention as an efficient tool for the production of acids and bases from their corresponding salts by electrically enforced

505

accelerated water dissociation. Bipolar membranes can be prepared by simply laminating conventional cation- and anion-exchange membranes back to back [27]. Laminated bipolar membranes often exhibit unsatisfactory water splitting capability. But special surface treatment of commercial ion-exchange membranes and subsequent laminating may yield bipolar membranes with satisfactory properties. Single film bipolar membranes and multilayer bipolar membranes fulfil most of the practical needs [28,29]. ELECTRODIALYTIC PROCESSES AS UNIT OPERATION

In electrcdialytic processes the membrane is the most important component determining the overall performance in a given application. However, process design and chemical engineering aspects also have a significant effect on the efficiency and the economics of a process in a given application. The different electrodialytic processes are rather different in their basic concept and system design. 1. Electrodialysis The technically and economically most important electrodialytical process used for the separation of ionic components from an aqueous solution is conventional electrodialysis. The main application of electrodialysis is the desalination of brackish water. However, other uses, especially in the food, drug, and chemical process industry as well as in biotechnology and waste water treatment, have recently gained a broader interest. In its basic form electrodialysis can be utilized to perform several general types of separations, such as the separation and concentration of salts, acids, and bases from aqueous solutions, or the separation of monovalent ions from multiple charged components, or the separation of ionic compounds from uncharged molecules.

1. 1 Elecwodialysis process principles The principle of the process is illustrated in Figure 3, which shows a schematic diagram of a typical electrodialysis ceU arrangement consisting of a series of anion- and cation-exchange membranes arranged in an alternating pattern between an anode and a cathode to form individual cells. If an ionic solution such as an aqueous salt solution is pumped through these cells and an electrical potential established between anode and cathode, the positively charged cations migrate towards the cathode and the negatively charged anions towards the anode. The cations pass through the negatively charged cation-exchange membranes but are retained by the positively charged anion-exchange membranes. Likewise the negatively charged anions pass through the anion-exchange membranes, and are retained by the cationexchange membranes. The overall result is an increase in the ion concentration in alternate compartments, while the other compartments simultaneously become depleted.

506

Fig. 3.

Schematic diagram illustrating the electrodialysis process

The depleted solution is generally referred to as the diluate and the concentrated solution as the brine. The technical feasibility of electrodialysis as a mass separation process, i.e., its capability of separating certain ions from a given mixture with other molecules, is mainly determined by the ion-exchange membranes used in the system. The economics of the process are determined by the operating costs, which are dominated by the energy consumption and investment costs for a plant of a desired capacity, which again are a function of the membrane area and various design parameters such as cell dimensions, flow velocities, etc.. 1.2 Electrodialysis energy requirements The energy required in an electrodialysis process is an additive of two terms: (1) the electrical energy needed to transfer the ionic components from one solution, i.e. the feed through membranes into another solution, i.e. the brine, (2) the energy required to pump the solutions through the electrodialysis unit. Depending on various process parameters, particularly the feed solution concentration, either one of the two terms may be dominating and thus determining the overall energy costs. At high feed solution concentrations energy requirements for the ion transfer are dominating. At very low feed solution concentrations energy for pumping the solution through the stack may be more significant [30]. a) Minimum energy required for the separation of a molecular mixture In electrodialysis as in any other separation process there is a minimum energy required for the separation of various components from a mixture. For the removal of salt from a saline solution this energy is given by:

G

AGO = RT l n y

aw

(3)

507

Here AGO is the Gibb's free enthalpy required to remove one mole of water from a solution, R the gas constant and T the temperature in

OK;

c0 and a,S

are the water activities in pure

water and the solution. Expressing the water activity in the solution with a monovalent salt by the concentration of the dissolved ionic components the minimum energy required to remove water from a monovalent salt is given by [31]:

I

--1

E C O11

(4)

Here AGO refers to the Gibb's free enthalpy required energy for the production of 1 litre of diluate solution. C is the salt concentration in moles per litre, the superscripts 0,('), and (") refer to the feed solution, the diluate and the concentrate. The reversible Gibb's free energy can also be expressed by:

AG =

ni zF Acp (i = 1,2,3

... n)

(5)

Here F is the Faraday constant, z the electrochemical valance, n is the number of moles, and Acp is the potential drop due to the concentration difference in the diluate and concentrate. This potential drop between two solutions separated by a semipermeable membrane is generally referred to as concentration potential. Practical energy requirements for the ion transfer b) The total electrical potential drop across an electrodialysis cell consists only partly of the concentration potential, the other part is used to overcome the ohmic resistance of the cell. This ohmic resistance is caused by the friction of the various ions with the membranes and the water molecules while being transferred from one solution to another, resulting in an irreversible energy dissipation in form of heat generation. The potential drop to overcome the ohmic resistance can be and in most practicle relevant applications is significantly higher than the concentration potential, thus in electrodialysis the practical consumed energy is generally much higher than the required theoretical energy. Furthermore, a considerable amount of energy is also necessary to pump the feed solution, the diluate and the electrode rinse solution through the electrodialysis stack. The energy necessary to remove salts from a solution is directly proportional to the total current flowing through the stack and the voltage drop between the two electrodes in a stack. The energy consumption in a practical electrodialysis separation procedure can thus be expressed by:

E m = l*nRet (6) Here E m is the energy consumption, I the electric current flowing through the stack, Re the

508

resistance of a cell pair, n the number of cell pairs in a stack, and t is the time. The electric current needed to desalt a solution is directly proportional to the number of ions transferred through the ion-exchange membranes from the feed stream to the concentrated brine. It is expressed as [321:

I=

z F QAC

5

(7)

Here F is the Farady constant, z the electrochemical valence, Q the feed solution flow rate, and AC the concentration difference between the feed solution and the diluate, and 6 the current utilization. The current utilization is directly proportional to the number of cell pairs in a stack. A combination of equations (6) and (7) gives the energy consumption in electrodialysis as a function of the current applied in the process, the electrical resistance of the stack, i.e., the resistance of the membrane and the electrolyte solution in the cells, the current utilization, and the amount of ions removed from the feed solution:

Eprod =

InRetzF QAC

5

Here EM is the energy requirement, I the electric current through the stack, n the number of cell pairs in a stack, Re the resistance of a cell pair, t the time, z the electrochemical valence of the components to be removed, F the Faraday constant, Q the volume flow rate of the feed solution, AC the ion concentration difference between the feed solution and the diluate, and 5 the current utilization. Electrical energy required in electrodialysis is not only directly proportional to the amount of salts, i.e. electrical charges that has to be removed from a certain feed volume to achieve the desired product quality but it is also a function of the electrical resistance of the cells. Electrical resistance of the cells, again, is a function of individual resistances of the membranes and of the solutions in the cells. Since, furthermore, the resistance of the solution is directly proportional to its ion concentration, the overall resistance of a cell will in most cases be determined by the resistance of the diluate solution. The concentration in the diluate cell, however, is decreasing during the desalting process and thus its resistance is increasing accordingly. Under the assumption, that the concentration in the diluate is much lower than that in the feed and brine the energy consumption can be expressed by 1241:

Here I is the electrical current passing through a cell stack, n the number of cell pairs in a

509

stack, Q the total volume of the diluate solution, C is the concentration and b a constant factor. The subscripts o and d refer to the feed and the diluate solution. A typical value for the resistance of an electrodialysis cell pair, i.e., the cation- and anion-exchange membrane plus the dilute and concentrated solution, e.g., in the desalination of brackish water, is within the range of 5 to 500 Q cm2. For other application the electrical resistance of a cell pair might be significantly higher or lower. Energy required to pump the solutions through the stack c) The operation of an electrodialysis system requires two or three pumps to circulate the diluate, the brine and eventually the electrode rinse solutions through the stack. The energy required for pumping these solutions are determined by the volumes to be circulated and the pressure drop. It can be expressed by:

Here Ep is the pumping energy, k a constant refemng to the efficiency of the pumps, Q volume flows, and Ap the pressure losses in the diluate; the subscripts D, B and E refer to brine and electrode rinse solution. The pressure losses in the various cells are determined by solution flow velocities and the cell design. The energy requirements for circulating the solution through the system may become significant or even dominant when solutions with rather low salt concentration are processed. Other energy consuming processes are the electrochemical reactions at the electrodes. In a stack with a multicell arrangement, however, the energy consumed at the electrodes is generally less than 1 % of the total energy used for the ion transfer and can therefore be neglected. d) Comparison of energy consumption in electrodialysis and competing processes In many applications electrodialysis is competing with other separation processes. While the theoretically required minimum energy is identical in all processes there are significant differences as far as the irreversible energy dissipation is concerned. For the desalination of a saline solution, e.g. different processes, such as reverse osmosis, ion exchange, distillation, are used in addition to electrodialysis. All processes require the same theoretical minimum energy. The irreversible dissipated energy is rather different in the different processes, as can be illustrated by comparing the basic principle of desalination by electrodialysis and reverse osmosis which is shown schematically in Figure 4.

510

+

Salt and water

Salt and water

Water

Water

Anions

AP

A P

AE

Water

Sah

Reverse osmosis

Fig. 4:

Electrodidysis

Schematic diagram illustrating the operating principles of reverse osmosis and electrodialysis

The basic difference between reverse osmosis and electrodialysis is that in reverse osmosis the water passes the membrane under a driving force of a hydrostatic pressure difference and in electrodialysis the salt is passing the membrane under the driving force of an electrical potential difference. The irreversible energy loss in reverse osmosis is caused by a friction loss of the individual water molecule on their pathways through the membrane matrix. This means, that the irreversible energy loss in reverse osmosis is independent of the feed water salt concentration. In electrodialysis the irreversible energy loss is caused by the friction of the individual ion on their pathway through the membrane from the diluate to the brine solution, thus in electrodialysis the irreversible energy loss is directly proportional to the concentration difference between the feed solution and the product water. For feed solutions with low salt concentrations the energy requirements are therefore generally lower in electrodialysis than in reverse osmosis, and at high feed solution salt concentration the situation is reversed. This is shown schematically in Figure 5 where the irreversible energy consumption is plotted versus the feed solution concentration assuming for both cases identical product water concentrations.

feed solution salt concentration

Fig. 5:

Schematic diagram showing the irreversible energy loss in electrodialysis, and reverse osmosis as a function of the feed solution salt concentration

511

A comparison of mass separation processes concerning their energy consumption has to take into account that in electrodialysis the energy is required in form of electricity, a relative expensive form, and in distillation, e.g., a relative inexpensive form of energy, i.e., heat can be used. In ion exchange very little energy is required directly. However, the chemical used for the regeneration of the resin required a significant amount of energy for their production. 1. 3

Electrodialysis process and equipment design Electrodialysis as a unit operation is determined by several process and equipment design parameters, such as feed flow velocities, cell and spacer construction, stack design etc. These parameters effect the costs of the process directly and also indirectly by means of the limiting current density and the current utilization [ 3 3 ] . Limiting current density and current utilization a) The limiting current density is the maximum current which may pass through a given membrane area without obtaining effects resulting in higher electrical resistance, lower current utilization, or other operational problems such as membrane fouling and scaling. The limiting current density is determined by the ion concentration in the dilute flow stream and by concentration polarisation effects as indicated in Figure 6, which shows the concentration profiles of cations in the boundary layer at the surface of a cation-exchange membrane during an electrodialysis process. Similar anion concentration profiles are obtained at the surface of an anion-exchange membrane.

t

Anode

Cathode

1 laminar boundary laier

Fig. 6:

Schematic diagram of the concentration profiles of cations in the laminary boundary layer at both surfaces of a cation-exchange membrane during electrodialysis. (C is the cation concentration, the subscripts b and m refer to the bulk solution and c and d refer to concentrate and diluate).

The transport of charged particles to the anode or cathode through a set of ion-exchange membranes leads to a concentration decrease of c o u n t w n s in the laminar boundary layer at the membrane surface facing the diluate cell and an increase at the surface facing the brine cell. The effect of concentration polarization due to a concentration increase in the

512

brine is less severe. The decrease in the concentration of counter ions directly affects the limiting current density and increases the electrical resistance of the solution in the boundary layer. The limiting current density is the current density at which the ion concentration at the surfaces of the cation- orJand anion-exchange membranes in the cells with the depleted solution will approach zero. The limiting current density can be calculated by a mass balance considering all mass transport through the membrane and the boundary layers. It can be described to a first approximation by [34]:

Here ilim is the limiting current density, C$ the bulk solution concentration in the cell with the depleted solution, D and z are the diffusion coefficient and the electrochemical valence of the ions in the solution, F the Faraday constant, Yb the boundary layer thickness, and TM and T the ion transport numbers in the membrane and the solution, respectively, and the subscripts + and - refer to cations and anions, respectively. The constant k is the mass transfer coefficient, taking into account the influence of the hydrodynamics of the feed solution flow, i.e. the flow channel geometry, the spacer design, the flow velocities, etc. According to equation (11) the limiting current density is proportional to the ion concentration in the diluate and the mass transfer coefficient, which is determined mainly by the cell geometry and the feed solution flow velocity. If in electrodialysis the limiting current density is exceeded, the process efficiency will be drastically diminished because of the increasing electrical resistance of the solution and because of water splitting which leads to both increasing energy consumption as well as pH changes in the solutions at the surface of the membrane causing additional operational problems. The mass transfer coefficient can be related to the Sherwood number, which again is a function of the Schmidt and Reynolds number [35].Introducing the proper relations in equation (1 1) leads to an expression which describes the limiting current density as a function to the feed flow velocity in the electrodialysis stack

d Here c b is the concentration of the solution in the bulk of the diluate cell, u is the linear flow velocity of the solution through the cells parallel to the membrane surface and a and b are constants, the value of which are determined by a series of parameters such as cell and spacer geometry. solution viscosity, ion-transfer numbers in the membrane and the solution, etc. The constants a and b are a function of the electrodialysis stack design and must be

513

determined experimentally. The limiting current density can be determined by several means [36], e.g. by measuring the electrical resistance of a cell pair or the pH-value in the diluate cell as a function of the current density. When the pH-value is plotted versus l/i a sharp decrease in the pH-value is noted when the limiting current density is exceeded. Likewise, when the total resistance of a cell pair is plotted versus l/i a minimum is obtained at the limiting current density. This is shown schematically in Figure 7 a and b. That usually a pH-drop in the diluate cell is observed is due to the fact that water splitting usually occurs first at the anion-exchange membrane probably because of the catalytic effect of the tertiary amine groups at the surface of the membrane [37] as will be discussed later when describing the water splitting mechanism in bipolar membranes.

al

G

0

al 3 -

2

Ia

r

.-

v)

2

I

1 '

4

1li

Limiting current density a)

Fig.7:

8 C

4

1 /i Limitin'g current density b)

Schematic diagram illustrating the determination of the limiting current density by plotting: a) the pH-value of the diluate cell versus l/i and b) by plotting the resistance of a cell pair versus l/i

The limiting current density determines the minimum membrane area required to achieve a certain desalting effect. Another very important parameter for the overall performance of the electrodialysis process is the current utilization. The current utilization determines the ponion of the total current that passes through an electrodialysis stack that is actually used to transfer ions from a feed solution. The current utilization which is always less than 100 % is affectea by three factors [38]: (1) The membrane selectivity, (2) osmotic and ion-bound water transport, and (3) current passing through the stack manifold.

Here 5 is the current utilization, q is an efficiency term and n is the number of cell pairs in a stack; the subscripts w, m and s are referring to efficiency losses due to water transfer, conductivity of stack components and membrane selectivity. The water transfer due to osmosis and ion hydration can be significant at higher brine salt concentrations. The membrane selectivity also depends on the salt concentration due to a Donnan equilibrium between the salt solution and the membrane as discussed earlier.

514

For a diluted feed or brine solution an ion-exchange membrane in general is more or less strictly semipermeable, i.e. the membrane is permeable to counter ions only. When, however, the ion concentration in the feed solution is of the same order as that of the fixed charges in the membrane co-ions may also enter the membrane and its selectivity will then be decreased, with the consequence that the current efficiency decreased, too. b)

The electrodialysis stack design A typical electrodialysis stack design is shown in Figure 8. An electrodialysis stack is essentially a device to hold an array of membranes between electrodes in such a way that the streams being processed are kept separated. Cation-exchange membrane

Spacer

Anion-exchange membrane Concentrate Diluate

Feed Dilu'ate cell Condentrate Cell

Fig. 8:

Exploded view of an electrodialysis stack

The gaskets not only separate the membranes but also contain manifolds to distribute the process fluids in the different compartments. The supply ducts for the diluate and the brine are formed by matching holes in the gaskets, the membranes, and the electrode cells. The distance between the membrane sheets, i.e. the cell thickness, should be as small as possible to minimize the electrical resistance. In industrial size electrodialysis stacks membrane distances are typically between 0.5 to 2 mm. A spacer is introduced between the individual membrane sheets both to support the membrane and to help control the feed solution flow distribution. The most serious design problem for an electrodialysis stack is that of assuring uniform flow distribution in the various compartments. In a practical electrodialysis system, 200 to 1000 cation- and anion-exchange membranes are installed in parallel to form an electrodialysis stack with 100 to 500 cell pairs. As any membrane separation process electrodialysis is effected by concentration polarization and membrane fouling. The magnitude of concentration polarization is largely determined by the electrical current density, by the cell and particularly spacer design, and by the flow velocities of the diluate and brine solutions [39]. Concentration polarization effects electrodialysis lead to a depletion in the laminar boundary layer at the membrane

515

surfaces in the cell containing the diluate flow stream and to an increase of ions in the laminar boundary layer at the membrane surfaces in the cell containing the brine solution. Concentration polarization is effecting the separation efficiency by decreasing the limiting current density 1401. More difficult to control are membrane fouling effects due to adsorption of polyelectrolytes, such as humic acids, surfactants, proteins etc.. The components often penetrate the membrane because of their size only partially and thus resulting in severely reduced ion permeability of the membrane 1411. In designing an electrodialysis stack several general criteria concerning mechanical, hydrodynamic, and electrical properties have to be considered. Since some of the criteria are counter effective, the final stack construction is generally a compromise between several conflicting parameters [38,391. A proper electrodialysis stack design should provide a maximum effective membrane area per unit stack volume. The dismbution of the solutions should ensure equal and uniform flow distribution through each compartment. Any leakage between the diluate, concentrate, and the electrode cells should be prevented. The spacer screen should provide a maximum of mixing of the solutions at the membrane surface and cause a minimum in pressure loss. Most stack designs used in today's large-scale electrodialysis plants are one of two basic types: tortuous path or sheet flow. These designations refer to the type of solution flow path in the compartments of the stack. In the tortuous-path stack, the membrane spacer and gasket have a long serpentine cut-out which defines a long narrow channel for the fluid path. The objective is to provide a long residence time for the solution in each cell in spite of the high linear velocity that is required to limit polarization effects. A tortuous-path and sheet flow spacer gaskets are shown schematically in Figure 9a) and b).

Fig. 9: Schematic diagram of a) a tortuous-path electrodialysis spacer gasket, and b) a sheet-flow electrodialysis spacer gasket

516

In stack designs employing the sheet-flow principle, a peripheral gasket provides the outer seal and the solution flow is approximately in a straight path from the entrance to the exit ports which are located on opposite sides in the gasket. This is illustrated in Figure 7 a) which shows the schematic diagram of a sheet-flow spacer of an electrodialysis stack. Solution flow velocities in sheet-flow stacks lie typically between 4 and 10 cm/sw, whereas in tortuous-path stacks solution flow velocities of 10 to 30 cm/s are required [42, 431. Because the higher flow velocities and longer flow paths, higher pressure drops in the order of 2 to 3 bars result in tortuous-path stacks than in sheet-flow systems where pressure drops of 1 to 2 bars occur. Electrdalysis process design and economics c) In addition to the actual stack and the power supply unit, an electrodialysis plant consists of several components essential for proper operation, such as pumps, process monitoring and control devices, feed solution pretreatment systems, etc.. There are two operating modes for the electrodialytic process described in the literature [44]. The first is referred to as the unidirectionally operated electrodialysis plant and the second is a reversed polarity operated electrodialysis plant [7]. A flow diagram of a typical unidirectional operated electrodialysis plant is shown in Figure 10. Feed Inlet

I

Concentrate Inlet

I Electrode Waste

Electrode Waste Product

Concentrate Recycle

Fig. 10:

-

....

cnncnntrate Blowdown -". ,"",

Flow diagram of a typical unidirectional electrdalysis desalination plant

After proper pretreatment, the feed solution is pumped through the actual electrodialysis unit, which generally consists of one or more stacks in series or parallel. A deionized solution and a concentrated brine is obtained. The concentrated and depleted process streams leaving the last stack are collected in storage tanks, when the desired degree of concentration or depletion is achieved, or they are recycled if further concentration or depletion is desired. Sometimes

517

acid is added to the concentrated stream to prevent scaling of carbonates and hydroxides. To prevent the formation of free chlorine by anodic oxidation the electrode cells are sometimes rinsed with a separate solution which does not contain any chloride ions. In many cases, however, the feed or brine solution is also used in the electrode cells. In the electrodialysis reversal operating mode the polarity of the current is changed at specific time intervals ranging from a few minutes to several hours. In this operating mode the hydraulic flow streams are reversed simultaneously, i.e. the diluate cell will become the brine cell and vice versa. The advantage of the reverse polarity operating mode is that precipitation in the brine cells are to a large extent prevented. Or if there is some precipitation, it will be redesolved when the brine cell becomes the diluate cell in the reverse operating mode. The flow scheme of a typical electrodialysis reversal plant is shown in Figure 11. Feed.Inlet , Feed

Electrode Waste

Electrode Waste

Fig. 11:

Flow diagram of a typical electrodialysis reversal plant

The process design and economics are closely related in electrodialysis. The total process costs are the sum of fixed charges associated with amortization of the plant capital costs and operating costs, such as energy and labour costs. Membrane replacement costs are sometimes regarded as a separate item because of their relatively short life of 5 to 7 years. Capital costs include depreciable items such as the electrodialysis stacks, pumps, elecmcal equipment, membranes etc., and non depreciable items such as land and working capital. The capital costs of an electrodialysis plant will strongly depend on the total membrane area required for a certain plant capacity. The required membrane area, however, is proportional to the number of ionic species removed from a given feed solution. It can be calculated by the

518

following relation [331: A =

zFQACn is

Here A is the membrane area, z the chemical valence, Q the volume of the produced potable water, A C the difference in the salinity of feed and product water, n the number of cells in a stack, i the current density which should be about 80 % at the limiting current density, and 5 the current utilization. The limiting current density is a function of the diluate concentration which is changing during the desalting process from the concentration of the original feed to the product solution concentration. The calculation of the minimum membrane area required for a given desalting capacity is based on an average limiting current density, which is a function of the average diluate concentration given by:

-

Here Flim is an average limiting current density, Cd is the average diluate concentration, a is a constant factor, which depends on the cell and spacer geometry and feed flow velocity. The subscripts o and d refer to the feed solution and the diluate. Introducing equation (15) into (14) leads to:

Here Amin is the minimum membrane area required for a certain plant capacity and feed and product solution concentrations, a is a constant for a given plant design and operating mode, z is the chemical valence, F the Faraday constant, Q the product solution volume, and COand Cd the feed and the product solution concentrations. For a certain plant capacity, the required membrane area is directly proportional to the feed water concentration. This is illustrated in Figure 12. For brackish water of ca. 3000 ppm TdS and an average current density of 12 mA/cm2, the required membrane area for a plant capacity of 1 m3 product per day is ca. 0.4 m2 of cation- and anion-exchange membrane. Other items such as pumps, electric power supplies, etc. depend on plant size. For desalination of brackish water with a salinity of ca. 3000 ppm the total capital costs for a plant with a capacity of 1000 m3/d will be in the range of US $200,000.- to US $300,000.-. The costs of the actual membrane is less than 30 % of the total capital costs. Assuming a useful life of 5 years for the membranes and 10 years for the rest of the equipment, a feed water salinity of 3000 ppm and a 24-hours operating day, the total amortization of the

519

investment is ca. US $0.10 to US $0.15 per m3 water with a salinity of less than 500 ppm.

1

10

100

Feed solution Concentration gll

Fig. 12:

Schematic diagram of the required membrane area in electrodialysis desalination as a function of the feed water concentration at constant current density, plant capacity and product water concentsation.

The operating costs are mainly determined by the required energy which is, as pointed out before, determined by the electrical energy required for the actual desalting process and the energy necessary for pumping the solution through the stack. The energy for the actual desalting process, i.e. the ion transfer from the feed solution to the brine is directly proportional to the number of ionic species to be removed, as indicated in equation (8) and (9), respectively. The energy requirements for the production of potable water as a function of the feed water concentration is shown in Figure 13. The case considered is a NaCl feed solution with the product having a salt concentration of less than 500 ppm.

x

P 0, C

UJ

1 I 1

1

I

10

100

Feed solution concentration (g/l) Fig. 13:

Energy requirements for the production of potable water with a solid content of 500 ppm as a function of the feed solution concentration (AU per cell pair = 0.8 V)

520

The pumping energy is independent of the feed solution salinity. Assuming a pressure drop in the unit of ca. 400 KPa (4 bar), a pump efficiency of 70 %, and 50 % product water recovery, the total pumping energy will be ca. 0.4 kWh per m3 product water. This indicates that at low feed water salt concenmtion the cost for pumping the solution through the unit might become quite significant. It should be noted, that according to equations (8) and (16),the energy costs increase with increasing current density while the required membrane area decreases with increasing current density. Thus the total desalination costs, which are the summation of capital, energy and operating costs, will reach a minimum at a certain current density as illustrated in Figure 14, where the total costs are shown as a function of the applied current density for a given feed solution. Total costs Energy cost

s

0

Capital costs Operating costs Current density

Fig. 14:

Schematic diagram of the electrodialysis process costs as a function of the applied current density

Quite interesting furthermore is a comparison of the cost of desalination by various processes as a function of the feed water salinity, as shown in Figure 15. O.Or

ion-exchange //

E!ectrodialysis

h

m

E

b

s

1.0

3

0

0.1

Na Ci Concentration (911)

Fig. 15:

Water desalination costs as a function of the feed solution concentration for ion- exchange, electrodialysis,reverse osmosis, and distillation

52 I

The graph in Figure 15 indicates that at very low feed solution salt concentration ion exchange is the most economical process. But its costs are sharply increasing with the feed solution salinity and at about 500 ppm TDS electrodialysis becomes the more economical process. While at around 5000 ppm reverse osmosis is the less costly process. At very high feed solution salt concentrations, in excess of 100 000 ppm multistage flash evaporation becomes the most economical process. The costs of potable water produced from brackish water sources are in the range of US $0.2 to US $0.5 per m3. 1.4

Technically relevant applications of electrodialysis

Electrodialysis was developed first for the desalination of saline solutions, particularly brackish water. The production of potable water is still currently the most important industrial application of electrodialysis. But other applications, such as the treatment of industrial effluents [45], the production of boiler feed water, demineralization of whey [46], de-acidification of fruit juices [47], etc. are gaining increasing importance with large-scale industrial installations. An application of electrodialysis which is limited regionally to Japan has gained considerable commercial importance. This is the production of table salt from sea water. Diffusion dialysis and the use of bipolar membranes have significantly expanded the application of electrodialysis in recent years [48]. a) Desalination of brackish water In terms of the number of installations the most important large-scale application of electrodialysis is the production of potable water from brackish water. Here, electrodialysis is competing directly with reverse osmosis and multistage flash evaporation. For water with relatively low salt concentration (less than 5000 ppm) electrodialysis is generally the most economic process, as indicated earlier. One significant feature of electrodialysis is that the salts can be concentrated to comparatively high values (in excess of 18 to 20 wt.%) without affecting the economics of the process severely . Most modern electrodialysis units operate with reverse polarity, i.e. the anode and cathode, and with that the diluate and concentrate cell systems, are exchanged periodically, preventing a scaling due to concentration polarization effects. In brackish water desalination, more than 2000 plants with a total capacity of more than 1,000,000 m3 of product water per day are installed, requiring a membrane area in excess of 1.5 million square meters [49]. b) Production of table salt The production of table salt from sea water by the use of electrodialysis to concentrate sodium chloride up to 200 g/Lprior to evaporation is a technique developed and used nearly exclusively in Japan. More than 350,000 tons of table salt are annually produced by this technique requiring more than 500,000 square meters of installed ion-exchange membranes. Key to the success of this technology has been the low cost, high conductive membrane with

522 a preferred permeability of monovalent ions [6]. However, it should be noted that in Japan this procedure of salt production is highly subsidized. c) Waste water treatment The main application of electrodialysis in waste water treatment systems is in processing rinse waters from the electroplating industry. Here, complete recycling of the water and the metal ions can be achieved by electrodialysisin some applications [50]. Compared to reverse osmosis, electrodialysis has the advantage of being able to utilize more thermally and chemically stable membranes, so that processes can be run at elevated temperatures and in solutions of very low or high pH-values. Furthermore, the concentrations which can be achieved in the brine can be significantly higher. The disadvantage of electrodialysis is that only ionic components can be removed and additives usually present in a galvanic bath cannot be recovered. The recovery of nickel salts from electroplating rinse waters is an application which has been pursued by several companies [42]. Here electrodialysis has the function of a "kidney" by removing the nickel salts which have been dragged out of the plating tank into a still-rinse [50].The concentrated nickel salts can often be directly fed back into the plating tank, while the diluate is recycled into the still-rinse. Dump leach waters containing heavy metal ions have successfully been treated by electrodialysis. The removal of nitrate from drinking water by electrodialysis has been studied extensively and seems to compete well in this application with other treatment procedures, such as ion-exchange or reverse osmosis [47]. An application which has been studied in a pilot plant stage is the regeneration of chemical copper plating baths [45]. In the production of printed circuits, a chemical process is often used for copper plating. The components which are to be plated are immersed into a bath containing, besides the copper ions, a strong complexing agent, for example, ethylenediamintetraacetic acid (EDTA), and a reducing agent such as formaldehyde. During the plating process, formaldehyde is oxidized to formate. After prolonged use, the bath becomes enriched with Na2 SO4 and sodium formate and consequently loses its useful properties. By applying electrodialysis in a continuous mode, the Na2 SO4 and formate can be removed from the solution, without affecting the concentrations of formaldehyde and the EDTA complex and the useful life of the plating solution is significantly extended. Several other potential applications of electrodialysis in wastewater treatment systems which have been studied on a laboratory scale are reported in the literature. In most of these applications the average plant capacity, however, is considerably lower than that in brackish water desalination or table salt production. Concentration of reverse osmosis brines d) A further application of electrodialysis is the concentration of reverse osmosis brines. Because of limiting membrane selectivity and the osmotic pressure of concentrated salt solutions, the concentration of brine in reverse osmosis desalination plants can not exceed

523

certain values. Often the disposal of large volumes of brine is difficult, and a further concentration is desirable. This further concentration may be achieved at reasonable costs by electrodialysis. [421 Electrodialysis in the chemical, food, and drug industry e) The use of electrodialysis in food, drug, and chemical industries has been studied quite extensively in recent years. Several applications have considerable economic significance and are already well established today. One is the demineralization of cheese whey [46]. Normal cheese whey contains between 5.5 and 6.5 % of dissolved solids in water. The primary constituents in whey are lactose, protein, minerals, fat and lactic acid. Whey provides an excellent source of protein, lactose, vitamins, and minerals, but in its normal form it is not considered a proper food material because of its high salt content. With the ionized salts substantially removed, whey provides an excellent source for the production of babyfood. The partial demineralization of whey can be carried out quite efficiently by electrodialysis. The removal of tartaric acid from wine is another possible application of electrodialysis. In the production of bottled champagne, it is necessary to avoid the formation of crystalline tartar in the wine and tartaric acid must therefore be reduced to a value which does not exceed the solubility limit. This can be done efficiently by electrodialysis. Several other applications of electrodialysis in the pharmaceutical industry have been studied on a laboratory scale [51]. Most of these applications are concerned with desalting solutions containing active agents which have to be separated, purified, or isolated from certain substrates [52]. Here, electrodialysis is often in competition with other separation procedures such as dialysis, solvent extraction, etc. In many cases, electrodialysis is the superior process as far as economics and the quality of the product is concerned. Especially the separation of amino acids and other organic acids by electrodialysis seems to be of interest to the pharmaceutical and chemical industry [53].However, the deionization of cheese whey with an installed capacity of more than 35,000 square meters of membrane area for the production of more than 150,000 tons of desalted lactose per year is economically by far the most important application of electrodialysis in the food industry today. f) Production of ultrapure water More recently electrodialysis is being used in combination with mixed-bed ion-exchange

resins for the production of ultra pure water. In this application electrodialysis is used as a pre-treatment step and is in direct competition to reverse osmosis which has the advantage to remove also neutral components in addition to the salts. For certain feed water sources, however, electrodialysis is preferred for economic reasons. Overall the electrodialysis industry has experienced a steady growth since it made its appearance as an industrial scale separation process about 15 years ago and new areas of application in the food and chemical process industry are gaining interest rapidly.

524

OTHER ELECTRICALLY DRIVEN MEMBRANE PROCESSES Although electrodialysis is today by far the most important industrial membrane separation process utilising ion-exchange membranes and an electrical potential gradient as driving force there are several other processes gaining industrial significance rapidly, such as regular electrolysis used for the production of chlorine and caustic soda 1541, the electrodialysis with bipolar membranes used for the production of acids and bases from the corresponding salts [55], or the combination of conventional electrodialysis with regular ion-exchange techniques to produce ultra pure water. Most of these processes have been developed only recently and their large scale industrial utilization is still in the beginning. The chlorine-alkaline electrolysis The electrolytic production of chlorine and caustic soda using a cation-exchange membrane as a separation medium is already a technically and commercially well established process [54, 563.The principle of the process is illustrated in the schematic drawing of Figure 16, which shows an electrolysis cell arrangement consisting of two chambers separated by an cation-exchange membrane. 1.

t Na CI

Fig. 16:

Schematic diagram illustrating the chlorine/alkaline production process

One compartment contains an anode and the sodium chloride feed solution. The other compartment contains the cathode and at the beginning of the process water. When an electrical potential difference between the two electrodes is applied, the positively charged sodium ions will migrate towards the cathode producing hydrogen and hydroxyl ions in an electrochemical reaction at the cathode. The negatively charged chloride ions move towards the anode and will be oxidized to form chlorine. Thus sodium chloride is converted into chlorine, caustic soda, and hydrogen. A migration of the hydroxyl ions is prevented by the cation-exchange membrane. Thus the current utilization in the electrolytic chlorine and caustic soda production is close to 100 %. The compartment containing the produced sodium

525

hydroxide is usually operated in a feed and bleed mode and its sodium hydroxide concentration is kept as high as possible. In industrial production processes sodium hydroxide concentrations in excess lOwt% are obtained. Since the sodium chloride concentration is also kept rather high the electrical resistance of the solutions is comparatively low, and the cell system can be operated at relatively high current densities up to a few thousand Mm2. The main problem in the electrolytic production of chlorine and caustic soda is the stability of the cation-exchange membrane which faces a strong caustic environment on one side and solution containing free chlorine on the other side. Today, membranes based on fluorinated hydrocarbone polymers have a useful life time of several years in operation at elevated temperatures [56]. 2. Electrodialytic water dissociation in bipolar membranes Bipolar membranes have recently gained increasing attention as efficient tools for the production of acids and bases from the corresponding salts by electrically forced water dissociation. The process, which has been known for many years, is economically very attractive and has a multitude of possible applications [%I. So far, however, the large-scale use of the process has been rather limited because of the shortcomings of today's bipolar membranes, which have to meet certain requirements as far as their water splitting capability, their electrical properties and chemical stability is concerned. But recent progress in the development of efficient bipolar membranes have increased the technical and industrial importance of this process. Principle of water dissociation in bipolar membranes a) The water dissociation in a bipolar membrane is illustrated in Figure 17 which shows a bipolar membrane consisting of an anion- and a cation-exchange layer arranged in parallel between two electrodes. anion-exchange

cation-exchange membrane

Y \

/

bipolar membrane

Fig. 17:

Schematic diagram illustrating the principle of the electrodialytic water dissociation in bipolar membranes

526

If an electrical potential difference is established between the electrodes all charged components will be removed from an aqueous interphase between the two ion-exchange layers. If only water is left in the solution between the membranes further transport of electrical charges can only be accomplished by protons and hydroxyl ions which are available in very low concentrations in completely de-ionized water. Protons and hydroxyl ions removed from the interphase are replenished because of the water dissociation equilibrium. A bipolar membrane thus consists of a cation- and anion-exchange layer laminated together. The theoretical energy required for the process is that for establishing the desired concentration of H+ and OH--ions in the outer phases of the membrane from their concentration in the membrane which is approximately 10-7mow. The free energy of this process is:

AG=zRTln

i i aH+ %HaH+ %H-

Here AG is the free energy, z the electrochemical valance, R the gas constant, T the absolute temperature, and a the activity. The superscripts o and i refer to the outside and the interphase between cation- and anion-exchange membrane, respectively. 0

0

For the generation of a one molar ideal solution of H+- and OH--ions, i.e., %+= 1, aOH-= 1, and z = 1, equation (17)reduces to:

i

i

AG = R.T-l n(a H+.a OH-)=R T h K w Here Kw is the dissociation constant of water. At 25 OC the negative logarithm of the water dissociation constant, - log Kw is 13.99. The free energy for the dissociation of one mole water and thus the production of one molar acid and base at 25 OC is: 79887 Joule or 0.0222 kwh. The generation of protons and hydroxyl ions via an electrolysis process, however, requires considerably more energy. This is evident from the very nature of the process which entails co-production of H2 and 0 2 or chlorine. This step requires some additional energy input. The theoretical energy in electrolysis varies slightly, depending on the particular salt being processed, the concentration of acid and base generated, and the temperature of operation. For production of one normal acids and bases at 25OC the theoretical free energy varies between 0.056 and 0.58 kwh./mol. To minimize the irreversible energy losses in a bipolar membrane its electrical resistance should be as low as possible. Furthermore in practical applications, bipolar membranes are exposed to an aggressive chemical environment. The cationic side of the bipolar membrane is facing a strong acid and the anionic side of the membrane is in contact with a strong base. The preparation of cation-exchange membranes with excellent

521

stability even in strong acid is relatively easy. Anion-exchange membranes with the required alkaline stability and electrical properties especially at elevated temperature are far more difficult to make [57]. But today bipolar membranes with long term stability at pH-values in excess of 13 are commercially available. These membranes can be operated at current densities in excess of lo00 A m-2 with high current utilization [55]. Applications of the electrodialytic water dissociation b) The main application of electrodialytic water dissociation in combination with regular ionexchange membranes is the production of acids and bases from the corresponding salts. In this application bipolar and regular cation- and anion-exchange membranes are installed in alternating series between two electrodes to form a stack of individual cells similar to those used in conventional electrodialysis. In this case, however, a repeating unit consist of three individual cells. A typical arrangement of an electrodialysis stack with bipolar membrane as used for the production of an acid and a base is shown in Figure 18.

Acid

Base

&Pt

A

0

C

A

-

3

Anod

0 Cathode

X-

Salt

Fig. 18:

t

Salt

Schematic drawing illustrating an electrodialysis cell arrangement with bipolar membranes used for the production of acids and bases from the corresponding salts

The electrodialytic water dissociation has been evaluated on a laboratory scale and a multitude of applications have been identified mainly in generating acids and bases from the corresponding salts. The process shows significant advantages in terms of energy requirements over conventional acids and bases production procedures. The purity of the products, however, is often unsatisfactory especially when high acid and base concentrations are required. The process can be integrated in chemical or biochemical production processes when an adjustment of pH-values is required. Bipolar membranes have also been integrated in acids and bases scrubbers used for the removal of waste gases from air such as SO2 [%I.

528

3 Diffusion dialysis Another process utilizing ion-exchange membranes in an electrodialysis stack cell arrangement is referred to as diffusion dialysis. This process can be used, e.g. to separate acid from mixtures with salts [58]. The principle of the process is illustrated in Figure 19. In this case a diffusion dialysis cell system contains anion-exchange membranes only. A feed solution is pumped through alternating cells while water is pumped in counter current flow through the other cells. Protons and the anions can penetrate the anion-exchange membranes due to a concentration gradient while the cations are rejected. The net result is the removal of acids from a mixture with salts. Like-wise bases can be removed from salt solutions using cation-exchange membranes. (Salt+acid) Acid

I

Acid

1 x

4 Fig. 19:

Schematic diagram illustrating the principle of diffusion dialysis

The process is used on a large scale to recover mineral acids from salt solutions obtained in pickling and etching processes. In this application only anion exchange membranes are installed in a stack as indicated in Figure 19. By feeding in alternating cells a mixture salt and acid and pure water in counter current flow more than 95% of the acids can be removed from the feed solution. Donnan dialysis D O M dialysis ~ is used to exchange ions between two solutions. The stack arrangement is identical to that used in diffusion dialysis. The principle of the process is illustrated in Figure 20, which shows a CuSO4 solution and H2SO4 separated by a cation-exchange membrane. Since the H+-ion concentration in the acid solution ' is significantly higher ( p ~ = 1) than the H+ ion concentration in copper sulfate solution " (PH = 7) there will be a driving force for the transport of H+-ions from solution I into solution II. Since the membrane is permeable to cations only, there will be a build-up of an elecmcal potential which will counter-balance the concentration difference driving force of the H+- ions. This electrical potential difference will cause a flux of Cu++-ions against their concentration

4.

529

gradient from solution " into solution '. As long as the H+-ion concentration difference between the two phases separated by the cation-exchange membrane is maintained, there will be the transport of Cu++-ions until their concentration difference is of the same order of magnitude as the H+-ion concentration difference. Solution

(PH = 7)

Solution'

(pH = 1)

"\cation exchange membrane

Fig. 20:

Schematic drawing illustrating the principle of Donnan dialysis by showing the transport Cu ++-ions through a cation-exchangemembrane utilizing an electrical potential build up by the flux of H+-ions

The process can be carried out accordingly with anions through anion-exchangemembranes. An example of anion Donnan dialysis is the sweetening of citrus juices. In this process hydroxyl ions furnished by a caustic solution replace the citrate ions in the juice. Electrodialytic regeneration of a cation- or anion-exchangeresins The process is illustrated in Figure 21 which shows a cation loaded resin placed between two electrodes. 5.

Salt free solution H+

I Cathode

Me

+

Anode

Cation exchange resin Me+ Feed solution

Fig. 21:

Schematic diagram illustrating the electrodialytic regeneration of a cationexchange resins

When an electric current is applied protons generated at the anode will move to the cathode into the resin and replace the metal ion on the resin which will then move from the resin to the cathode where they are collected, precipitated or concentrated. Electrodialytic regeneration of ion-exchange resins is less labour intensive than chemical regeneration and

530 adds no additional salts to the effluent. The process has been evaluated on a laboratory scale but low current efficiency due to the high mobility of the protons have hampered its practical application [59]. Recently introduced modifications in the process design, however, have eliminated part of this problem [60]. A anion-exchange resin is regenerated similarly by hydroxyl ions generated at the cathode. Using a mixed bed ion-exchangeresin completely de-ionized water is obtained. This process has recently been commercialised [61]. 6. Hybrid Processes Most separation processes are efficient only under certain conditions of feed solution concentration and required product quality. Therefore various separation processes are often combined, each operating its optimum range of application. A typical example for the combination of different processes is the combination of conventional ion-exchange process with electrodialysis. There are however, other combinations such as electrodialysis and reverse osmosis which are of growing technical and commercial interest. CONCLUSIONS The electrodialytic processes have experienced a steady growth since they made their appearance as industrial scale separation processes about 20 years ago. Currently the desalination of brackish water, the chlorine-alkaline elecwlysis and the production of table salt are still the dominant applications, but new areas of application in the food and chemical process industry and the use of hybrid processes are gaining interest rapidly. REFERENCES Ostwald, W. 1890. Elelctrische Eigenschaften halbdurchllssiger Scheidewiinde. Z. Physik. Chemie 6:71-82. Donnan, F.G., 1911. The theory of membrane equilibrium in presence of a nondialyzable electrolyte. Z. Electrochem. 12572. Morse, H.N., Pierce, J.A.. 1903. Z. physik. Chem., G:589. 795. Meyer, K.H., Strauss. 1940. Helv. Chem. Acta and W.A. McRae. 1950. Coherent ion-exchange gels and membranes. J. Am. Juda, W., Chem. Soc.22:1044. Nishiwaki, T. 1972. Concentration of electrolytes prior to evaporation with an electromembraneprocess. In: Industrial Processing with Membranes, Edts.: R.E. Lacey, and S. h b . Wiley & Sons, New York. Katz, W.E. 1979. The electrodialysisreversal (EDR) process. Desalination a:31-40. Connolly, D.J., and W.F. Gresham. 1966. Fluorocarbon vinyl ether polymers. U.S. Patent 3,282,875. Liu, K.J., F.P. Chlanda, and K.J. Nagasubramanian. 1977. Use of bipolar membranes

a:

531 for generation of acid and base: an engineering and economic analysis. J. Membr. Sci.

2:109. 10 Kedem, O., and Y. Maoz. 1976. Ion conducting spacer for improved electrodialysis. Desalination Ip( 1-3):465-470. 1 1 Donnan, F.G., and E.A. Guggenheim. 1932. Exact thermodynamics of membrane equilibrium. Z. Phys. Chem. A162:346-360. 12 Bergsma, F., Ch.A. Kruissink. 1961. Ion-exchange membranes. Fortschr. Hochpo1ym.Forsch. a:307-362. 13 Helfferich, F. 1962. Ion Exchange. New York: McGraw-Hill Book Co. 14 Gregor, H.P. 1951. Gibbs-Donnan equilibria in ion-exchange resin systems. J. Am. Chem. Soc. 22:642-650. 15 Teorell, T.Z. 1951. Z. Elektrochem. Z:460-469. 16 Meyer, K.H., J.F. Sievers. 1936. Permeability of membranes. Helv. Chim. Acta B: 649-677. 17 Adams, B.A., E.L. Holmes. 1936. Brit. Pat. 450 308. 18 Friedlander, H.Z. 1968. Membranes Encycl. Polym. Sci. Technol. &620-638. 19 Kusomoto, K., T. Sata, and Y. Mizutani. 1976. Modification of anion-exchange membranes with polystyrene sulfonic acid. Polym. J. &:225-226. 20 McRae, W.A., and S.S. Alexander. 1960. Sulfonating reagent and its use in preparing cation-exchange membranes. U.S. Patent 2,962,454. 21 Flett, D.S. 1983. Ion Exchange Membranes. Chichester, U.K.:E. Horwood Ltd. 22 Eisenberg, A. and H.L. Yeager. 1982. Perfluorinated Ionomer Membranes. ACS Symposium Series No. 180. Washington, DC: American Chemical Society. 23 Grot, W.G. 1973. Laminates of support material and fluorinated polymer containing pendant side chains containing sulfonyl groups. US.Patent 3,770,567. 24 Ishigaki, I., N. Kamiya, T. Sugo, and S. Machi. 1978. Synthesis of an ion-exchange membrane by radiation-induced grafting of acrilyc acid onto poly(tetrafluorethy1ene). PO~YIYI.J. J.Q(5):513-5 19. 25 Sata, T., R. Izuo, and Y. Mizutani. 1984. Study of membrane for selective permeation of specific ions. Soda to Enso 3(415):313-336. 26 Gudernatsch, W., Ch. Krumbholz, H. Strathmann. 1990. Development of an AnionExchange Membrane with Increased Permeability for Organic Acids of High Molecular Weight. Desalination B:249-260. 27 Leitz, F.B. 1972. Apparatus for electrodialysis of electrolytes employing bilaminar ionexchange membranes. U.S. patent 3,654, 125. 28 Chlanda, F.P., L.T.C. Lee, and K.J. Liu. 1976. Bipolar membranes and method of making same. U.S. Patent 4, 116,889. 29 Bauer, B., F.-J. Gerner, and H. Strathmann. 1988. Development of bipolar membranes. Desalination 48:279-292. 30 Schaffer, L.H., and M.S. Mintz. 1966. Electrodialysis. In: Principles of Desalination,

532

31 32

33 34

35 36 37 38

39 40 41

42 43 44

45 46

47 48 49

Ed.: K.S. Spiegler, pp. 3-20. New York: Academic Press. Spiegler, K.S. 1956 in: 'Ion-Exchange Technology", Edts.: F.C. Nachod, J. Schubert. pp.118-181, Academic Press, New York, N.Y. Strathmann, H., 1979. Trennung von Molekularen Mischungen mit Hilfe Synthetischer Membranen. Darmstadt: Steinkopff-Verlag. Mintz, M.S. 1963. Electrodialysis: principles of process design. Ind. Eng. Chem 2 5 (6):18-28. Lacey, R.E. 1972. Basis of electromembrane processes. In: Industrial Processing with Membranes, Edts.:. R.E. Lacey and S.Loeb. New York: John Wiley & Sons. Bird R.B., W.S. Stuart,E.N. Lightfoot, 1960. "Transport Phenomena". John Wiley & Sons, New York. Cowan, D.A:, J.H. Brown. 1959. Effect of turbulence on limiting current electrodialysis cells. hid. Eng. Chem. 51:1445. Simocs, R. 1984. Electric field effects on proton transfer between ionizable groups and 151-158. water in ion-exchange membranes. Electrochemica Acta 2: Huffmann, E.L., R.E. Lacey. 1972. Engineering and economic considerations in electromembrane processing. In: "Industrial Processing with Membranes". Edts.: R.E. Lacey, S. Loeb, pp. 39-55. John Wiley & Sons. Jonsson, G., C.E. Boezen. 1984. Polarization phenomena in membrane processes. In: "Synthetic Membrane Processes". Ed.: G. Belfort. Academic Press, New York. Spiegler, K.S. 197 1. Polarization at ion-exchange membrane-solution interfaces. Desalination 9367-385. Korngold, E., F. De KBriSsy, R. Rahav, and M.F. Taboch. 1970. Fouling of anion selective membranes in electrodialysis. Desalination &(2):195-220. Ionics Inc. 1988. Product Bulletin, Watertown, MA 02172. Asahi Chemical 1988. Product Bulletin, Tokyo, Japan. Strathrnann, H., 1990. Membranes and Membrane Separation Processes. "Ullmann's Encyclopedia of Industrial Chemistry" Vol. A 16, pp. 187-263.Verlag Chemie, Weinheim, FRG. Korngold, E., IS.Kock, and H. Strathmann. 1978. Electrodialysis in advanced waste water treatment. Desalination B(1-3):129-139. Ahlgren, R.M. 1972. Electro membrane processing of cheese whey. In: Industrial processing with Membranes, Edts.:. R.E. Lacey and S. Loeb, pp. 71-81. New York: John Wiley & Sons. Kneifel, K., G. Liihrs, H. Wagner. 1985. Nitrate removal by electrodialysis for brewing water. Desalination 68:203-209. Sata, T. 1986. Recent trends in ion-exchange research. Pure & Appl. Chemistry 58:1613-1626. Wangnick, K. 1985. Desalting plant inventory report. International Desalination Association (IDA) Report.

533

50 Deuschle, A., E. Kiibler. 1984. Elektrodialyse: Wertstoffriickgewinnungaus galvanischen Spiilwlssern. Galvanotechnik x:968:97 1. 51 Reed, Ph.B. 1984. Electrodialysis for the purification of protein solutions, Chem. Eng. Progress, Dec. 1984: 47-50. 52 Voss, H. 1986. Deacidification of citroacid solutions by electrodialysis. J. Membrane Sci. =:165-171. 53 Itoh, H., T. Yoshizumi, M. Saeki. 1986. Sieving effect in electrodialysis with an ionexchange membrane. J. Membrane Sci. z:155-163. 54 Coulter, M.V. 1980. Modem chlor-alkali technology. Ellis Honvood Ltd., Chichester, U.K. 55 Mani, K.N. 1991. Electrodialysis water splitting technology. Journal of Membrane Science 3(2):117. 56 Seko, M., S. Ogawa, K. Kimoto. 1988. Perfluorocarboxylic acid membrane and membrane chlor-alkali process developed by Asaki Chemical Industry. In: "Perfluorinated Ionomer Membranes". Edts.: A. Eisenberg, H.L. Yeager. ASCSymposium Series L8Q,American Chemical Society, Washington. 57 Bauer, B., F. Effenberger, H. Strathmann. 1990. Anion-Exchange Membranes with Improved Alkaline Stability. Desalination 2125-144. 58 Kobuchi, Y., H. Motomura, Y. Noma, F. Hanada. 1987. Application of ion-exchange membranes to recover acids by diffusion dialysis. J. Membrane Sci. u:173-179. 59 Strathmann, H., Kock, K. 1982. Effluent free electrodialytic regeneration of ionexchange resins, in: "Polymer Seperation Media", Edt.: A.R. Cooper, Plenum Press, New York, N.Y. 60 Johan, J., G. Eigenberger Chem. Eng. Techn. in Press 61 Millipore Corp. Bulletin. 1987. No. CJ 001. Bedford, Ma.

This Page Intentionally Left Blank

535

Removal of H,S Through an Electrochemical Membrane Separator Steven Alexander

School of Chemical Engineering Georgia Inetitute of Technology 225 North Avenue Atlanta, Georgia 30332-0100

United States of America

Introduction Use of selective membranes for separating gaseous componenta h m mixturee is a common unit operation. The thermodynamic basis for the separation is very eimple: a component will only move down a chemical potential gradient, Ap:

where the prime ' refers to the extracted phase. The effect of preeeure on the activity ie emall,

and the pressure difference must be large enough to overcome the concentration effect:

Thus, for a separation &om a phase with 1% into a pure phaee, a minimum preesure ratio of about 100 is needed. In actual practice a higher preesure drop is needed to promots a aimrificantflux. These procesees do not produce a highpurity product, nor do they remove one component with perfect selectivity. The situation is different for a charged species in the preeence of an electric potential, A@. Here, the electrochemical potential, F, is the driving force:

536

So,for a charged species With a charge of +/- 2, a potential difference of 60 mV can maintain the aame concentration difference that requires 100 atm for an uncharged species. The effect is more pronounced as the concentration in the feed drops to the levele encountered in contaminant removal, e.g. 10 ppm. Here a pressure-driven separation to a pure stream would require more than 10' atm while an eIectrochemicaI aeparation requires only 160 mv."] This principle can be, and has been, applied to high temperature gas mixtures. Specifically, the removal of &S from fuel gas streams (natural gas and coal gasification product streams). Since the membrane is exposed to the same pressure on both aides, there is no theoretical limit to the pressure at which the process operates. Also, since there is no reaction or absorption equilibrium, there is no theoretical limit to removal.

Technical Discussion A hot-gas electrochemical membrane process is illustrated schematically in Figure 1. The process gas is paesed by the cathode. Here, the most easily reduced component, that is, the strongest Lewis acid, w i l l be electronatad. With

natural gas and coal synthesie gas, it is H,S:

H#

+ 2c-

-

(6)

H, + S2-

A membrane which contains sulfide ions in a molten state will act to transport sulfide across to the anode where, in the simplest case, hydrogen can be supplied to form H a . If the membrane is capable of preventing the diffusion of hydrogen from the cathode side, an inert sweep gas such as N, can be used at the anode to carry away oxidized sulfide ions as vaporous sulfur, S,. The situation can become complicated when realistic gas mixtures are processed. Carbon dioxide and water vapor compete in the reduction reaction by:

CO, + H,O

+ 2a-

- C0,l-

+ H,

(6)

The ionic flux through the membrane depends upon the relative mobilities of carbonate and sulfide a8 well ae their concentrations. The direct oxidation of carbonate:

occurs at a standard potential aome 700 mV more positive than that for sulfide:

537

If the half-cell reactions (6) and (8) are summed, the resulting cell reaction and standard potential at 900 K is:

and if the half-cell reactions (6) and (7) are summed, the resulting cell reaction and standard potential are: H20

-

H2

+

1 zOz

E o = -1.03OY

(10)

If the relative extent of each of these reactions is determined by chemical equilibrium,then each will occur at the same potential as predicted by the Nernst relation: 1

If a process gas is supplied to the cathode with an H S level of 2000 ppm, a CO, level of 196, and an H,O level of 12% (a saturated natural gas composition), it is assumed that 99% of the H,S is removed by reaction (51, and if the process and sweep gas flowrates are equal, then there will exist an activity ratio of ucos/crsof 665 in the anolyte before significant (e.g. 1%)of the carbonate is oxidized. This assumes equivalent electrodekinetics for the cathodic and anodic reactions. When compared to the activity ratio of a d a s of 26.9, this shows the thermodynamic preference for the oxidation of'S to elemental sulfiv by equation (8) when there is an absence of reductant at the mode. This mode of operation is preferable for commercial application, with direct production of elemental sulfur vapor, eliminating the need for a Claw reactor for sulfiv production. The net effect, under these conditions, is continuous removal of Ha from the process gas accompanied by enrichment of the process gas with H, and direct generation of elemental sulfur. The only reagent required is electric power at a potentially attractive rate, which w i l l be shown.

538

The equilibrium potential for a single cell, given by equation (111, for the cathodic and anodic reactions (5) and (81, is -406mV for a process gas containing 2000 ppm H,S and an anode product of pure sulfur vapor. To this must be added the overpotentials needed for both electrode reactions and ohmic loss. The electrode reactions have been studied in free electrolyteon graphite Potential-step erperimente showed very rapid kinetics, with exchange currents in both cathodic and anodic direction near 40 d c m a . Cyclic voltammetry verified a 'catalytic' reaction mechanism with &sulfide as the electro-active epee&. At the cathode:

and

At the anode:

s,

-

+ 2s2-

$

-

S, +

2s;-

(16)

PC-

Enhanced cathodic H,S removal is found with CO, and &O in the gas, probably due to another 'catalytic' scheme, reaction (6) followed b y

@- + H# * S2-

+

H,O + CO,

(17)

Concurrent removal of COS down to below the analytical limit (ca. 2 ppm) is also observed. This apparently occurs due to rapid equilibrium between &S, CO, Cop COS at these temperatures. H# + CO H#

+ CO,

-

COS + H,

(18)

H 2 0 + COS

(18)

A study of potential cathode materials"' showed several promising alternatives. It was conducted using a configuration similar to anticipated designs (see Figure 2). Several electrode materials were found acceptable, including nickel and cobalt, formed from powdere, and allowed to go to either their equilibrium sulfide or oxide in-eitu Studies of membrane electrolyte compositions have also been performed. By analyzing the equilibrium of reaction (171, it is possible to know the electrolyte

539

compoeition which would be in equilibrium with a given process gas at a given procese temperature (see Figure 3Y6]. Theoretical membrane compositione were calculated by thermodynamic analysis of the membrane electrolyte equilibrium reaction (20). Since membranes M a r to those ueed in the Molten Carbonate Fuel Cell (MCFC) were used in this d y e i s , the catione preeent were K and Li in a ratio comsponding to the low melting carbonate eutectic (Li,,a&M).

This analyaie was perfomed by 5 d i n g the Gibb's fiee energy of reaction (20) at the process temperature and relating this to the equilibrium constant, y,by the relation:

-hK,

AGO RT

with K. defined as:

By this analysis, a procees gas with a composition of 0.88% CO,, 1760 ppm Ha, 12% H,O, and the balance methane with a run temperature of 610°C would have an equilibrium constant of 6.9 (based on thermodynamic data of Barin and Knacke)"]. If the activity coe5cient.s of the molten phase conetituents (namely the sulfide and carbonate in the electrolyte) are aesumed to be unity, this translates to an equilibrium composition of 19.596 sulfide and 80.596 carbonate. If a membrane is manufactured so as to be already in equilibrium with the gas to be treated, it w i l l not have to undergo the stresees inherent in the density changes associated with 'sulfiding' a carbonah membrane or 'carbonating' a eulfide membrane. While techniques for manufacturing such a membrane are still under study, the concept has been successlll used in both the coal gasification process cell"] and the natural gas process cell&

.

Experimental Cell Geometry The cell housings were machined !?omMACOR (machinableceramic)blocks and type 316 stainless steel bar stock. The housings were 3" diameter and 1" deep cylinders. Gas flow channels were machined into the large surface faces with dimensions of 0.3 QII by 0.3 cm. Gas flow tubes were connected to supply process and sweep gasee to the cell. Once the electrode and membrane materials

540

were ready for testing, the electrodes were set onto gold or platinum current collectore placed on top of the gas flow channels on one aide and contacting the surface of the membram on the other (see Figure 4) when the MACOR housings were used. When stainless steel housings were used, the electrodes were placed directly onto the flow channels and the housings acted as the current collectors. The full cell was then assembled by placing the membrane between the MACOR blocks and connecting the gas supply lines to the assembly. E l e c W Preparation Several potential electrode materials have been identified. Of these; Carbon, CoS,, and lithiated NiO were used in the experiments described here.

Carbon While useful for free-electrolytekinetic studies and bench scale removal cell testing, graphite would not be acceptable for commercial electrodes in a removal cell. At the cathode it would eventually be eroded by steam and CO,: C+

H20 CO + H,

m, + c 2co

(23) (24)

and at the anode it can act as a reductant for carbonate: CO;- +

-

c co +

CO, + 2c-

(26)

cos, High purity COS, was mixed with hydroxyethyl cellulose (HEC). Void percentages as high as 60% were obtained using a mixture of 10 wt% HEC and 90 wt% CoS, powder. This mixture was loaded into a 1 l/4" stainless steel die and pressed at 8000 psi using a hydraulic ram. The resulting electrode wafer was then heated at 350°C for 30 minutes to burn out the HEC. This final electrode was then cooled, weighed, and stored for use in the electrochemical cell.

NiO High purity partially aintered nickel sheeta were obtained from ERC (Electric Research Corp.). The porosity of these sheets varied from 60% to 70%. The proprietary nature of these materials prohibits any release of the pore size distribution. Disks with a thickness of 30 mils and a diameter of 1 l/4"were cut from this material. Thie disks were soaked in 1molar EOH, dried at lorn, and oxidized in room air at 660°C for at leaet 8 hours to form a lithiated NiO structure. Gravimetric analyeis of these electrodes ehowed that they were greater than 9896 converted to NiO.

541

Membrane Preparation Three techniques have been used for the manufacture of the membranes utilized in the experiments discussed here. The first technique involved manufacturing a sintered ceramic matrix of MgO without electrolyte present and then 'wicking' the molten electrolyte into the matrix by capillary action. The inert ceramic matrix which holds the electrolyte in place between the cathode and the anode serves two purposes; first, it holds the electrolyte by capillary action and prevente the molten salta from completely flooding the porous electrodes; second, the membrane acts to prevent the bulk diffusion of gases between the cathode and the anode side of the cell. If the electrolyte was not in chemical equilibrium with the process gas, localized d e d t y changes in the electrolyte caused by reaction (20) would cause the membrane to crack and allow bulk mixing of the process and sweep gas streams. The second technique of membrane manufacture involved making a composite structure consisting of woven zirconia cloths which had been densified with MgO powder. The structure consisted of three mats of ZYW-30A zirconia cloth (purchased from ZIRCAR Jnc.) layered with three tapes of MgO ceramic powder suspended within acrylic binder K565-4 (purchased from Metoramic Sciences, Inc.). The electrolyte was layered into the structure as pressed powder disks during set-up. Due to the amount of handling outside the cell run conditions involved during setup, the electrolyte was initially all eutectic Li/K carbonate which is etable (though hygroscopic) in normal room air. A nitrogen sweep was applied to both the process and sweep sides of the cell and the cell was loaded into the furnace for heat-up. The binder from the MgO tapes was volatilized out at 376°C overnight. The temperature was then ramped up to the run temperature and the electrolyte wicked into the MgO powders and zirconia cloth at process temperature. Processgas was then supplied to the cell and the electrolyte was allowed to reach the equilibrium composition described by reaction (20). Since the ceramic matrix was no longer a rigid sintered structure, localized density changes in the electrolyte did not cause the cracks Been with the more rigid structures. The third technique of membrane manufacture also involved using tapes of MgO suspended within Metoramic's acrylic binder KS65-4.However, in thie case, one mat of ZYW-30A zirconia cloth was layered with two tapes of MgO. This structure was loaded into the cell assembly which was heated to 350°C with pure 0,flowing across both sides of the cell. Thie burned out the acrylic binders leaving behind the ceramic matrix within the cell. A nitrogen eweep was then applied to both the process and sweep side of the cell. Electrolyte was then added in-situ by allowing Li/K eutectic composition carbonate electrolyte to melt down the reference electrode hole in the process side of the cell housing and wick into the ceramic matrix already in place between the electrodes by capillary action. Once the electrolyte was in place, process gas was applied to the cathode side of the cell and the electrolyte composition was allowed to go to equilibrium as described above.

542

Cell Assembly and Heating , The electrodes were arranged on either side of the membrane as shown in Figure (4). Current was carried to the electrode surfaces either 0.009" diameter platinum wire. The electrodes were separated from the ceramic surfaces by resting atop platinum current collectors. Once assembled, the cell was loaded into a custom-made furnace and connected to the procees and sweep gas supply lines. The exit gas 60m the cathode was routed to a Beckman IR BcBMBr for reading COa levels. A HewletWackard gas chromatograph fitted with a thermal conductivity detector was used for reading H a levels greater than 100 ppm and a flame photometric detector was used for H@levels less than 100 ppm. A gold reference electrode was placed on the t3ulfBce of the membrane away from either process electrode and supplied with a low flow rate of a 16% COO/ 3%0 9 / balance NImixture to maintain a stable thermodynamic reference potential by reaction (26).

co, +

30, + 2s-

c@-

(26)

The cell assembly was heated at a rate of 100Whr. Melting of the electrolyta was verified by a sudden improvement in the seals formed by the contact of the membrane with the cell housing surfaces and observed electrical conductivity through the cell. Frocess gas, consisting of specified levels of &S, CO,, -0,and the balance CH, (for natural gas studies) or H& CO, COa, and the balance & (for coal gas studies), was then supplied to the cathode side and the cell was heated to the run temperature. Test Procedure Once the cell had reached rn temperature, conductivity across the cell was measured by the current interrupt method. The equilibrium potentials at the cathode and anode were measured with respect to the reference electrode. Baseline exit cathode gas compositions were also measure at this point. Current was then applied to the cell in a stepwise fashion and the cell was allowed to equilibrate after each current step. Once stabilized,potentials with respect to the reference electrode and the exit gas compositions were measured.

Electrolyte Analyeis Actual electrolyte compositions were measured by gravimetric analysis of total s u l f v species present after oxidation with hydrogen peroxide. A sample of membrane material was weighed and then dissolved in water. The insoluble matrix materials were filtered and the filtrate treated with excess hydrogen peroxide which oxidizes all sulfur species to sulfate. It was assumed that only sulfur in the form of sulfide was preeent in the membrane under run conditions. This solution was then acidified with hydrochloric acid to decompose the carbonate to carbon dioxide and water. The solution was then boiled to de-gas the mixture

543

and barium chloride was added, causing the sulfate to precipitate as barium sulfate. The solution was then filtered, and the precipitate was rinsed, ignited, and weighed. The moles of barium sulfate precipitated is equal to the moles of sulfide in the electrolyte. The mass of the soluble electrolyte present in the sample was known by difference. It was assumed that carbonate and sulfide species were the only components of the electrolyte, and since the mass of sulfide present was determined by the above analysis, the mass of carbonate present in the sample was also known by difference.

Experimental Run Results While fifty five experimental runs have been performed; only four representative runs are presented here. These are Run 24 (representative of very sour natural gaa, 1.33 9% H,S), Run 25 (representative of moderately sour natural gas, 2000 ppm Ha),Run 40 (representative of polishing application to slightly sour natural gas, 100 ppm as>, and Run 49 (representative of polishing application to moderate quality coal gas, 100 ppm €I$ Reproducibility ). of removal trends has been observed with all runs which successfully removed H&3 from the process gas.

Run 24, Scrubbing Application with 1.33% H@ Natural Gas This experimental run used a membrane manufactured by wicking an equilibrium composition electrolyte into a partially sintered MgO matrix. The cathode in this cell was carbon and the anode was CoS,. The process gas supplied to the cathode had an €&S concentration of 1.33% and a CO, concentration of 19.3%. Ha concentrations were driven as low as 2000 ppm as shown in Figure 5 (corresponding to 84.9% removal), but CO, removal was also observed as shown in Figure 6. Condensed sulfur was recovered from the anode sweep lines, but poor cell seals on the anode side made complete sulfur recovery impossible. While Ha removal at the cathode, with sulfur production at the anode, was shown, the process was not yet completely selective as was shown by the removal of COP. This was due to hydrogen -ion through the membrane which would cause reactions (7) and (8) to simply become the reverse of reactions (5) and (6). In this situation, there is no net cell reaction for species transport and IIpS, CO,, and G O would simply be concentrated on the anode side of the cell as a function of applied current. Cell polarization data is presented in Figure 7. The sudden jump in anodic overpotential at the highest current levels was caused by a break-down on the anode material, most likely due to oxidation of COS, to non-conductive COO,. This would have occurmi due to the high oxidizing potentid at the anode and the presence of 0, from oxidized COsa. Selectivity is defined by the following equation:

544

If selectivity is equal to one then removal of H,S and CO, are equivalent. If the selectivity is greater than one then H,S is preferentially removed. If selectivity is less than one then COI is preferentially removed. For this run selectivity is found to be 21.1 at the highest applied current. Thue, Ha is removed preferenThe current efficiency ie defined as: tially to co,.

defined as the percent removal of both &S and CO, at the applied current level and 9b Species R e m o v h u , defined as the gas phase removal required to eupport the applied current at steady state operation. At applied currenta less than 100 mA, current efficiency is 100%. At 2200 mA, current efficiency dropped to 91.1%. This wae likely due to back diffusion of gas species on the anode side of the cell to the cathode side of the cell.

with 96 Species Removal,,

Run 26, Scrubbing Application with 2000 ppm HJ Natural Gas The membrane used in this run also wed a membrane manufactured by wetting an equilibrium composition electrolyte into a partially sintered MgO matrix. As in Run 24, the cathode was carbon and the anode was Cob. The process gas supplied to the cathode had an &S concentration of 1927 ppm and a CO, level of 0.874%. &S concentrations were driven a8 low as 94 ppm (correeponding to 95.7% removal) as shown in Figure 8. CO, was driven to 0.089% with the same applied current level (86.4% removal) as shown in Figure 9. Cell polarization data is preeented in Figure 10. Sulfur was recovered from the sweep lines on the anode side of the cell. Electrolyte analysis showed a final composition of 20.4% sulfide and 79.6% carbonate. This compares with a theoretical composition of 18% sulfide and 82% carbonate. An overall sulfur balance showed 49.6% of the H# removed either retained in the electrolyta as sulfide or oxidized to elemental sulfur and collected on the sweep tube wall. The remaining 60.6% is assumed to have been blown through the marginal seals on the eweep side of the cell and lost in the furnace, or converted back to HpS at the anode eince there was 5 diffusion through the membrane as evidenced by the transport of CO, at cross-cell potentials above that required for reaction (10). Selectivityin this experiment was 4.9. Current efficiency went from nearly 100% at low applied currents to 35.9% at 160 mA This is a more dramatic example of gas phase back m i o n effecta on cell removal performance.

545

Run 40, Polishing Application with 100 ppm H,S Natural Gaa This experimental run used the layered membrane structure described earlier. Aluminum foil gaskets were also cut and laid into the wet seal area between the membrane and the MACOR housings. This wae done in an attempt to improve the wet eeal of the cell by intimately binding the membrane structure to the MACOR houings with a layer of LiAlO,formed in-mtu. During heat-up to run temperature, the aluminum was converted to Al,O, and then to LiAlO, through a subsequent reaction with Li,CO,'"':

ym,

+ 4 0 3 SLUIO, +

CO,

(29)

This run u88d carbon at the cathode and Ni (converted to NiO in-situ) at the anode. The process gas for this run had a composition of 98 ppm €I 1.45% $, CO,, and 3.9% H,O. &S levels in the process gas were brought as low as 2 ppm (below GC analytical limits, H a level corresponds to 98.W removal) over the course of the run with application of as little as 5 mA (cathode flow rate = 450 cdmin) with total cell potentials of only around 0.8 volta and no detectable CO, removal. H,S levels with applied current are presented in Figure 10. Cell polarization data is presented in Figure 11. Selectivity in this run was nearly infinite, signifyingnearly completely selective removal of H,S &om the process gas stream. Current efficiency was 55.9%. At these concentration ranges, no CO, removal could be observed. The excess current either went to transport CO, (though concentrations changes would only be around 100 ppm, to small to be detected by our equipment) or was carried through the cell by eome other current path than sulfide transport. ARer 135 hours of continuous operation, the cell was shut down for postmortem analysis. The carbon cathode, while still operational, had degraded and was showing obvious signs of erosion. An analysis of the membrane showed an actual sulfide level of 7.5 mole96 and a carbonate level of 92.3 mole%. Theoretical analysis predicted a sulfide level of 3.7 mole96 and a carbonate level of 96.3 mole%. An analyais of the anode showed that the structure was only 62.9% flooded. The electrolyte which was wetting the pores of the electrode had an approximate compoeition of 4.8 mole% sulfide and 95.2 mole% carbonate. Examination of the electrode material under x-ray diffraction shows that the primary species are Ni data is and NiO,with a Bm8u amount of Ni,h present also. X-ray -action presented in Figure 13.

Run 49, Polishing Application with 100 ppm H,S in Coal Gaa This experimental run was the fiRh run using coal gas. It ueed two tapes of MgO and one mat of zirconia cloth as the membrane matrix material. The electrodes were both lithiated NiO. The acrylic bindets were burned out under an 0, atmosphere and the Li/K eutectic-composition electrolyte was added with the cell at run temperature. T h e inlet gases were passed through a stainless steel

546

shiR reactor to allow them to come to their equilibrium compositionbefore passing through the cell. This experimental run was divided into two sections: the first confirmed ionic transport through the membrane by removal of CO, (and H,O) h m the syngas at 625°C (Run 49A)and the second was an attempt at removal of &S from the syn-gas at 700°C (Run4 9 0 . The results of these studies are presented below: Run 49A CO, removal from the process gas as a function of applied current was recorded and is presented in Figure 14. Examination of this data showe that the removal of CO, from the cathode aide of the cell and production of CO, at the anode aide of the cell is stoichiometric across the range of applied cvfents examined. Fuel gas flow was set at 75 d m i n and N, sweep was set at 63 dmin. Seals were good and no cross-flowbetween the two process streams was observed. Cathodic CO, removal is shown in Figure 14. Anodic CO, production is shown in Figure 15. Cell overpotentials are shown in Figure 16. Examination of this data shows that removal of COPwas close to stoichiometric over the range of applied currents. This shows that the cell was functioning properly with respect to ionic transport of carbonate through the electrolyte.

Run 49C The cell temperature was increased to 700°C. At this temperature, analysis of the limiting current densities within the system shows that the gas phase limiting current density was 1.15 &cm2 while the membrane limiting current density was 3.29 m A f d . This shows that the maximum flux of material through the membrane is three time greater that the maximum flux of material through the gas phase to the membrane surface. H,S removal at a variety of flowrates was observed and is presented here in Figure 17. The measured cross-cell resistance was measured by current interrupt and was found to be about la.With the maximum current applied to the cell of only 20 mA, this corresponds to only 20 mV of ohmic loss. This is slight compared to the overall cross cell potential, which includes concentration effects and the potentials required to drive the electrochemical reactions. The exit H a composition is plotted against m n event for 88 cdm h data in Figure 18. Note that initially, the membrane was showing process-stream crossflow. 1.5 grama of electrolyte was added to atop the cross flow and 5 mA were applied to the cell. This current level corresponds to five times the theoretical current required for complete H# removal. After driving the H.$ down to 16 ppm (81.2%removal, zero current basis) the current was turned off. The exit H# level returned to 89.5ppm. The lowest level to which the €&S level was driven was 9.7 ppm (89.1% removal, mro current basis). This data shows good response of the system to applied m n t . The overpotenial required to accomplish this removal is shown by Figure 19 to be negligible.

547

The H2S removal versus run event for the 210 cdmin data is shown in Figure 20. This data stdl ShOWS good response of the system to applied current. More electrolyte had to be added to repair membrane damage, and thus the initial exit &S cathode level with no current applied was down to 57 ppm. This was due to a build-up of carbonate caused by the excess electrolyte which had been added to the system. The data taken at a flowrate of 400 d m i n is presented in Figure 21 and the data taken at 600 cc/mia ie presented in Figure 22. Comparison of this data with the overpotential results shown in Figure 19 shows that the efficiency of the system dropped off with time. At eeveral points through the run, as marked on Figures 18 through 22, electrolyte was added to stop cross-over between the cathode side of the cell and the anode side. The increase in anodic overpotenital ehows that this excess electrolyte had flooded the anode, thus decreasing the reactive surface area from the interfacial area of the electrolyte wetting the walls of the electrode capillaries to the superficial area of the electrode when the pores were fully flooded. This was verified in the post-mortem analysis of the cell when the assembly was taken apart and the components examined. The anode flooded because it is phymcally on the bottom of the assembly. A total of 18.7 grams of electrolyte were added to the membrane during the course of the run in addition to the 11grams that were initially added to fill the ceramic matrix of the membrane. Post-mortem examination showed a small fracture in the matrix around the edge of the electrodes. This fracture would be temporarily flooded with electrolyte to form a gas impermeable barrier. However, aggressive attack by the electrolyte on the MACOR housings would deplete the membrane of electrolyte and lead to cross-flow. The run was terminated due to flooding of the anode and poor membrane integrity after 216 hours of operation.

Economic Projection Natural Gae Procerreing Plant Accurate cost figures for processes early in development are impossible to project. However, it is possible to roughly estimate the power and capital requirements to asmss viability. The power consumption is overwhelmingly due to the cell current, which is near stoichiometric. Cell voltage, as shown earlier, can be estimated with reasonable accuracy. Capital costs can be eetimated by analogy with MCFC eta&, whose design these membrane cells will mimic. The Electrochemical Membrane Separator (EMS) technology being developed is compared to a 'wet' removal process with subsequent Claw Plant processing to elemental sulfur and SCOT Tail Gas treatment of flue gases. This 'wet' procees utilities aqueous Methyldiethanolamine (MDEA) as an absorbent in a scrubbing operation to bring the &S level &om 1.7% to 4 ppm. The flow rate to these processee is 6OMM SCFD and the process pressure is 715 psi. The

548

process gas in this study has a COOcontent of 18.2%[81.Marshall-Swiftequipment cost indexes were used to estimata the expenses for the EMS fa&@ in 1987 dollars. On stream time is 360 days in both cases. Table 1breaks out the direct and indirect costs of the present and proposed technologies for comparison. The capital cost is more difficult to estimate than the power consumption. In the MCFC, currept densities greater than 160 d c m * are routinely achieved. There are however, two major differences between the MCFC and the EMS (Electrochemical Membrane Separator). In the MCFC the gases are relatively rich, as compared with the dilute reactanta treated in the EMS. H e r , there is no competing reaction to dilute the current-cafiying anion. Thus,gas-phase or sulfide migration in the membrane may limit the current diffusion of density and define the needed active membrane area for a given duty. Gas-phase transport CBP be controlled through proper design of the gas channel^''^; pore a s i o n in the electrodes has been found not limiting in similar designs for CO, removal to very low levels"o1. The limiting step for removal in this analysis is gas d i h i o n of H,S to the cathode of the cell. This is assuming that the process gap has been dehydrated to the point that the equilibrium of equation (20) provides a sulfide rich tile. The operating temperature of the EMS thus must be high enough to insure that the sulfide rich electrolyte is molten (850" C for this analysis). The current density was based on 2 faradaye of charge transferred for each mole of H@ diffusing to the surface of the cathode. The capital cost estimation assumes that the tiles are available as 4 ft by 4 R squares (as used in MCFC units) and are arranged in tack^' of parallel removal cells with the process gas equally divided to each cell[111.Each 'stack' removes approximately 90% of the H,S fed to it. In this analysis, there are four 'stacks' in series; the first bringing the H.#composition from 1.7% to 2000 ppm, the second from 2000 ppm to 200 ppm, the third &om 200 ppm to 20 ppm, and the fourth from 20 ppm to 4 ppm. There is assumed to be a 0.3 cm deep and 4 R wide gas flow path above each cathode. This analysis also assumes the use of a regenerative heat exchanger and a h c e fbeled by polished natural gas to heat the process gas to the cell temperature (see Figure 23). The heat-exchanger system was priced using standard costing figures"'! A break-down of the costs aesociated with the EMS stacks are provided in Table 2. Under these assumptions and at an operating temperature of 850" C, the stacks show gas phase limiting current densities of 234 d c m pfor the first, 26.1 d u n ' for the second, 2.61 d c m ' for the third, and 0.332 Wcm' for the fourth. Once the limiting current density of the stack is known, the total stack area (or number of cells required in the stack) can be calculated by dividing the required stack current by the stack limiting current density. The total etack current wae found by assuming a stoichiometric amount of electricity for the moles of Ha removed. In other words, 2 faradays of charge in required for each mole removed. Stack power requirements assume that a potential of 1Volt drives the removal d e . With these assumptions in mind, the first stack requires 2150

549

kW,the second stack requires 258 kW, the third stack requires 26.8 kW, and the fourth stack requires 2.29 kW. This ~ u m eto a total of 2436 kW for the EMS system. Under these conditions,the EMS plant can performthe scrubbing operation for $ O . l W l O O O SCF of gae treated. This favorably compares to $0.259/1000 SCF for conventional removal technology. The cost of operation for the EMS plant could be further reduced by optimizing the design of the regenerative heat exchanger in order to cut back the fuel requirementa of the gas heating furaace. The cell stack aize could also be redud by optimizing the gas flow channel design thereby increasing the gas phase mass transport of &S and thus increasing the cell limiting cwrent density.

5 50

Table 1. Operation Cost Comparison between Current Technology and ERC (1087 dollars) Fixed Capital Investment (x 10"):

Conventional 11.168

EMS 6.368

0.518 0.392 0.000 0.000

0.000 1.110 0.006 0.420

Direct Operations Costa Utilities (x 10") stem (8 $2.44/10001b~) Electricity (8 $O.O524/KWH) Raw &O (8 $0.76/1000gal) Gas Losses (@$2.00/100OSCF) Chemical Losses

0.021

m

Operating Labor (8 $10.30/hr) Maintenance (8 4% FCI) Plant General (40% Labor)

0.180 0.446 0,072 0.698

0.089 0.216 0.036 0.340

Total Direct Coeta:

1.629

1.876

Indirect Operating Costs Depreciation (@ 10%FCI) Tax & Insurance (8 2.6% FCI) Total Indirect Cost

1.116 9.279 1.396

0.107

Total Operating Costa:

3.024

2.520

Cost of Profit (8 25% FCI) (Includes income tax,interest on investment, and reasonable profit)

2.790

1.342

Grand Total Treating Cost

5.814

3.862

Sulfur Credit ($lOO/LT)

1.162

1.152

Net Treating Cost (GrandTotal Credit)

4.662

2.710

Treating Cost / 1000 SCF

$0.259

$0.151

0.931

-

1.636

0.537 0.644

55 1

Table 2. Break-down of Capital Inveetment for EMS (1987 dollars) Electrochemical Removal Cell Stacks ($ x lo4): stack1

Ion Exchange Area (tt'x lod): 9.867 617 Tiles in Stack:

@tack2

Stack 3

&su

10.63 664

10.62 664

7.421 464

0.091 0.037 0.068 0.005 0.021

0.064 0.026 0.048 0.003 0.015 0.058 0.214

Anodes: Cathodes: Bipolar Hardware: Tiles: Auxiliaries: Assembly stack cost:

0.085 0.035 0.004 0.020 0.077 0.285

0.091 0.037 0.068 0.005 0.021 0.083 0.305

Rectifier: Controls & Misc.: Aseembly Total Stack Cost:

0.155 0.216 0.170 0.826

0.166 0.233 0.183 0.887

0.064

7

0.083 0.305 0.166 0.233

-

0.183

Total EMS Cost: Blowers: Heat Exchanger: Plant c o s t Project Contingency (16%) Fixed Capital Investment

3.222 0.101 1.344 4.668 9.700 6.368

0.887

0.117 0.163 1 0 128 0.622

552

Figure 1.

Electrochemical Process Schematic.

Figure 2. Experimental Configuration.

553

Figure 3. Theoretical Electrolyte Composition (Simulated Natural Gas).

554 Mole 46 HZS

1.2

c 0' 0

I

200

I

I

400 600

I

I

I

I

L

I

I

loo0 1200 1400 1600 1800 2000 2u10 Applied Current (mA)

800

Inlet 1.33% HZS Cathode Flow = 350 cc/min

Figure 5. H..$ Level versus Applied Current,Run 24.

0

200

400 600 800 loo0 1200 1400 1600 1800 u)o 2200 Applied Current (mA)

Inlet: 19.3% C02 Cathode Flow = 350 d m i n

F'igure 6. CO, Level versus Applied Current,Run 24.

555 overpotential (volts)

89

‘i

i I

I / /

4

/

*

-

Cathode Reference

+k. Anode-Reference 0

400

UMO

800 1200 1600 Applied Current (mA)

Figure 7. Cell Polarization versus Applied Current, Run 24.

0

I

I

20

40

1

I

I

60 80 100 Applied aUnnt (mA)

I

I

120

140

Inlet: 1927ppmv H2S Cathode Flow = 50 cc/min

Figure 8.

Hps Level versa Applied Current, Run 25.

160

556

I

I

I

Inlec 0.874% C02 Cathode Flow = 50 CC/&

Figure 9. CO, Level vereua Applied Current, Run 25.

-

Cathode Reference

-

Anode RCfOXCWC

Figure 10. Cell Polarization vemu8 Applied Cumnt, Run 25.

557

0

1

2 3 Applied Cunrnt (mA)

4

5

Inlet: 98 ppmv H2S Call& Flow = 225 Cc/&

Figure 11. K S Level versu6 Applied Current, Run 40.

0.3

Overpotential (Volts)

#- Cathode - Reference

+k 0

1

2 3 4 Applied Current (mA)

'

Anode - Refercnce

5

Figure 12. Cell Polarization v e m Applied Current, Run 40.

558

..so* 4.00'

¶.so' S.00'

2.S' 1.00'

$.so' 1.00' 0.50'

N I W L SIN A- u4

100.0 BO.0 60.0

rO.0

Figure 13. Anode Materid X-Ray m a c t i o n Pattern, Run 40.

c

' + Cath

C O actual ~

.-fl-. Cath C 0 2 calculated I

14'

I

100

0

200

300

Applied Current (mA)

-

Inlet C 0 2 P 19.1YO Cathode Flow 75 cc/mln

Figure 14. COsh v e l versus Applied Current, Run 49A.

559

f

+t+ 0 0

100 200 Applied Current (rnA)

Anode C02 actual Anode CO2 calculated

300

Inlet C 0 2 = 0.00% Anode Flow = 63 cdrnin

Figure 15. Anodic CO, Production versw Applied Current, Run 49A. Overpotential(Volts)

la4

a 1

-# Cathode - Reference -0.4

'

0

-RI

Anode- Reference

I

100 200 Applied Current (rnA)

300

Figure 16. Cell Polarization vewus Current, Run 49A.

560 Conc. H2S (ppm) .. . . 100 I

I

U

5

0

10 Applied Current (mA)

15

20

inlet = 97 ppm Temp = 700 C

Figure 17. H,S Level versa Applied Current and Flowrate, Run 49C. Exit Conc. H2S (ppm) 100 I

0' 96

I

I

I

98

100

102

I

104

I

1

I

106 108 110 Hours Into Run

I

1

I

I

112

114

116

118

Temp I700 C Cathode Flow I 88 dmln inlet H2S = 97 ppm

Figure 18. Level vemw Event, 88 dmin, Run 49B.

56 1

0.4

-0.1

Overpotential (Volts)

I

1

'

I

0

5

I

I

I

10 15 Applied Current (mA)

20

Temp = 700 C

pigure 19. Cell Polarization v e n n ~Current and Flowrate, Run 49B. 100

Exit Conc. H2S (ppm)

118

I

I

I

I

1

I

I

I

I

I

I

120

122

124

126

128

130

132

134

136

138

140

Hours Into Run Temp I 700 C Cathode flow P 210 cdmin Inlet H2S = 97 ppm

Figure ao.

Ha Lad vemu Event, 210 &,

Run 49C.

142

562 Exit Conc. H2S (ppm)

100

1

80

I

60 .:

8

0

1

:

1

7

7

q

q

a

40 .-

20

1.11

I

0

1

I

I

I

I

I

1

I

I

Hours Into Run

Temp P 700 C Cathode Flow E 400 cdmln Inlet H2S = 97 ppm

Figure 21. H2S Level versus Event, 400 cchnin,Run 49C.

80 -

60 .

40

a

'

20

I

0' 186

I

188

190

Hours Into Run

Temp P 700 C Cathode Flow = 600 Cdmln Inlet H2S = 97 ppm

Figure 22. H2S Level vemus Event, 600 cchnia,Run 4W.

I

4

563

Sul

Pro

r

Recuperative Heat Exchanger

Polished Natural Gas Product

Figure 23. Proposed Electrochemical Membrane Separator Plant Layout.

Conclusions Selective removal of H,S from simulated natural gas streams has been demonstrated for three representative gas compositions representative of a very 60ur natural gas (1.6% &S), a moderately sour natural gas (2000 ppm H,S), and a slightly sour natural gas (100 ppm &S). Selective removal of has also been demonstrated for polishing application to a coal gadcation synthesis gas (100 ppm H,S). With respect to the natural gas processing plant economic study presented in this paper, these results support the design specificationsfor the first three removal cell eta*. Even with current efficiencies below 100%caused by gas phase back-diffusion limitations through the membrane, the process is still economically favorable. Doubling the required current for the specified removals (running at only 50% current efficiencyin all four removal cell stacks) would raise the operating casts to $0.21Y1000 SCF (as compared to $0.259/1000 SCF for conventional technology). Gas pham diffusion limitations to the membrane surface can be overcome by designins the gas flow channels for optimum gas species mass transport to the electrode surface and selectivity can be made tramport by preventing H,diffusion through infinitely favorable for exclusive the membrane.

564

[U

J. Winnick, b 1 edited by H. Geriecher and C. Tobias, VCH Publiehere, Weinheim, aermany, 1990.

123

K.A.White and J. Winnick,Electrochem. Ada, 80,(1985).

131

E.K. Banke and J. Winnick,J. Appl. Electmchem., 16,(1987).

[41

D. Weaver and J. Winnick,J. Electrochem. Soc., 184,(1987).

151

S.Alexander and J. Winnick,Seph Sci. and Teich., 25, (1990).

[61

I. Barin and 0.Knacke, Th 1 hemi and BuDdement, Springer-Verlag, Berlin, 1977.

171

D. Weaver and J. Winnick, J. Electrochem. Soc., 189,(1992).

181

D. Borio, ABBICombustion Engineering, Pereonal Communication,(1990).

[91

D. Townley and J. Winnick,I.E.C. h. Design, 80, (1981).

[lo] M.Kang and J. Winnick,J. Appl. Electrochem., 16, (1985). [ll] A. Appleby and F. Foulkes, fie1 Cell Handbook, Van Noetrand Reinhold, New York, (1989). [121 M.Petere and K. Timmerhaus, Plant Deei Fneineers, 3rd ed., McGraw-Hill, New York, (1982).

565

Electrochemical treatment of mixed and hazardous wastes Joseph C. Farmer University of California Lawrence Livermore National Laboratory Livermore, California 94550

INTRODUCTION Electrochemical processes are being developed throughout the world for the ambient temperature destruction of hazardous waste and for the conversion of mixed waste to low-level radioactive waste (Figure 1). Here, the status of the technology is discussed, including measurements of process efficiency and models than can be used for scale-up. Mlxed

Electrolysls

waste

Carbon

I

Water

Figure 1. Schematic of ME0 process at LLNL showing both anolyte and catholyte loops.

STATE-OF-THE-ART

M E 0 process Several of these processes are based on Mediated Electrochemical Oxidation (MEO). A mediator, generated by anodic oxidation, is used to destroy organics dissolved in an aqueous phase. Dissolved organics are converted to CO, and H,O. Ambient temperature oxidation of organic components of mixed wastes in a condensed phase avoids the possibility of hightemperature volatilization of radionuclides during incineration. Thermodynamic calculations indicate that plutonium and uranium may be converted to volatile PuO,(OH), and UO2(OH),, respectively, at the elevated temperatures and high partial pressures of water found inside incinerators and other thermal treatment systems [ 1-31.

566

Chevron Early work on the ME0 process was done by Clarke and coworkers at the Chevron Research Center [4-71. Petroleum coke, sewage sludge, o-dichlorobenzene, phenol, and other residues were oxidized under pressure by Fe(1II) in H2S04 at 180-250T. In a similar study, Dhooge and Park studied the MEOof coal slurriesby Fe(II1) and Ce(1V) 181. A preliminary economicanalysis was also done by Clarke [7]. The cost for destroying aniline was estimated to be approximately $1.78/kg, assuming a cell voltage of 3 V, a 95% current efficiency, a power cost of $O.O7/KWH, and that each molecule requires 28 electrons. Since ME0 is more expensivethan other processes such as biological treatment and incineration, it should only be used when other technologies are prohibited. Battelle More recently, Molton and coworkers have investigated ME0 at Battelle Pacific Northwest Laboratory [9-lo]. They demonstrated the destruction of hexone, methyl isobutyl ketone (MIBK), 5-methyl-2-pentanone, phenol, toluene, o-xylene, diethylene glycol dimethyl ether (diglyme), quinoline, chloroform, lI1,2-trifluoro- 1,2.2-trichloroethane (Photorex). hexachlorocyclohexane(Lindane), as well as trichloro-and perchloroethylenes (TCEand PCE). The electrolytic cell (ElectroSynthesisMicrocell) had a Ti anode (20 cm2) and a steel cathode (20 cm2), separated by a Raipore 1035 anion-exchange membrane. An anolyte volume of approximately300 ml was used. The anolyte consisted of 1 M Ni(N03), or Co(N03)2in HNO,. The anolyte temperature and cell current were maintained at 22-50°C and 6 A, respectively. Destruction rates of 2.74-3.10 g/hr were achieved. During the ME0 of hexone, acetone and acetic acid were formed as intermediates. It was also found that quinoline could be destroyed by direct anodic oxidation in NaOH (no mediator). A treatment cost of $0.4/kg was quoted. Dounrey Steele and his coworkers of the United Kingdom Atomic Energy Authority have used ME0 for the treatment of mixed waste in a pilot unit [ 111. Cellulose, 20% n-mbutyl phosphate (TBP) in kerosene, rubber gloves, polyurethane manipulator gaiters, epoxy resins, hydraulic fluid, lubricating oil, and ion-exchange resins were destroyed by Ag(II) in HN03. Destruction rates greater than 5 L/d were achieved. Lawrence Livermore National Laboratory (LLNL) The oxidations of several real and surrogate organic wastes, including ethylene glycol (EG), isopropanol (IPA), 2-monochloro-1-propanol(MCP), 1,3-dichloro-2-propanol(DCP), benzene (BZ), and Trimsol cutting oil have been studied in detail by the author and his coworkers [12151. Mediators such as Ag(II), Co(III),and Fe(1II)were used in supportingelectrolytesof HNO, or H2S04. In contrast to Ag(I), both Co(I1) and Fe(II) remain soluble in the presence of halide anions liberated during the destruction of halogenated organics. Co(II1) and FeOII) are referred to as “halide-tolerant’’ mediators. Several of these studies were conducted in an electrochemical batch reactor that had a rotating-cylinder anode (RCA). Since the anode was operated below the limiting current for mediator generation, relatively high coulombic efficiencies were achieved. Ion-exchange membranes were used to separate electrodes in Ag(I1)-basedprocesses, but were eliminated in processes based upon Co(II1) and H2S04. Rates of CO, generation were measured and used to

567

calculate destruction and coulombic efficienciesfor the process. Reaction intermediates were identified by gas chromatography with mass spectrometry (GCMS). A simple model, which accountsfor the sequentialformation of intermediates,was developed to predict the unusual time dependance of C 0 2 evolution. Two ME0 pilot plants have been built and demonstrated at LLNL. The largest system is constructed around the plate-and-frame electrolytic cell shown in Figure 2 (ICIModel FM21). Each anode and cathode has an active area of 0.85 m2 and is made of Ir-coated Nb. Two anodes and three cathodes are arranged in a parallel configuration, providing 1.7 m2 of total anode area. Typically, anolyte and catholyte flows are approximately0.73 and 0.88 Us (1 1.5 and 14.0gpm), respectively, and are regulated to minimize pressure drop across the ion-exchange membrane separators. Turbulance promotors are used to enhance the mass-transfercoefficient. At 0.15 M AgN0, in 4 M HNO,, the limiting current is approximately 1500 A. At higher anolyte concentrations, the cell current is limited to 3000 A by the capacity of the power supply. Total inventories of anolyte and catholyte are 16 and 20 L, respectively. At a cell current of 750 A, a destruction rate of 3 Wd and a coulombic efficiency of almost 100% have been achieved. Destruction rates of 12 L/d are believed to be possible at 3000 A.

Figure 2. Photograph of process showing the ICI FM21 plate-and-frame cell, which serves as the heart of the pilot unit.

568

Electrocinerator ME0 has been commercialized for the purpose of air cleaning by Electrocinerator [16]. Organic, inorganic, and biological contaminants have been scrubbed from streams of air by various anolytes. Organics removed by the commercial system include alcohols, aldehydes, amides, amines, aromatichydrocarbons,carboxylic acids, esters, halogenatedorganics, ketones, nimles, olefms, phenols, and thiols. Inorganicsremoved include ammonia, hydrazine, hydrogen cyanide, hydrogen sulfide, nitrogen oxide, phosphine, and sulfurdioxide. Biological substances removed include adenovirus, echovirus, pneumoniae, rhinovirus, influenza virus, staphylococcus aureus, streptococcus, and hemophilus influenza. EXAMPLES OF MIXED-WASTE FEED STREAMS Benzene High-level radioactive waste (HLW) will be converted from an alkaline slurry to a durable at the Savannah River Plant borosilicateglass in the Defense Waste Processing Facility (DWPF) (SRP) in South Carolina [17]. This waste is the residue from thirty years of reprocessing of irradiated nuclear fuels for national defensepurposes and is currently stored in large carbon-steel tanks. The waste at SRP exists in three forms: sludge, saltcake, and salt-supernate solution. The sludge, comprising approximately 10%of the stored waste, consistsprimarily of precipitates of the hydroxides of Fe, Al, and Mn. This waste contains most of the radioactivity, including small amounts of actinides not recovered in the reprocessing plants and most of the fission products, except for 137Cs.The salt is largely NaN03, NaAlO,, and NaOH. Most of the 137Csis contained in the salt-supemate solution. The salt-supemate solution is decontaminated for disposal as low-level waste by removing the radionuclides by precipitation and sorption. A solution of sodium tetraphenylborateis added to precipitate K+, Cs+, and NH4+ as insoluble tetraphenylborate salts. These salts are further processed to remove most of the organic carbon. About 90%of the phenyl groups on the salt are converted to an immiscible BZ phase by formic acid hydrolysis. Currently, the BZ is steam distilled, further decontaminated if necessary, and incinerated as a low-level radioactive and hazardous (mixed)waste. Since there is tremendous public concern about incineration of mixed waste, alternative technologies are of particular interest. Kerosene The Purex process is a typical solvent-extractionprocedure used for the treatment of spent nuclear fuel [181. Both U(V1) and Pu(1V) can be easily extracted from aqueous nitrate solutions into kerosene by TEiP, while fission products remain in the aqueous phase. After extraction, Pu(1V) is reduced to Pu(III), which is removed from the kerosene phase into a fresh stream of HN03. U(V1) remains in the kerosene until it is extracted into an aqueous stream which is fiec of HN03. Dissolved species are usually removed from aqueous solutionsby ion exchange. More than 70,000 gallons of kerosene, contaminatedwith TBP, 242Pu,and 243Am,have accumulated at U.S. Department of Energy @OE) sites over the past several decades.

569

Trimsol cutting oil Approximately 30,000 gallons of contaminated Trimsol cutting oil have also accumulated at DOE sites. The organic fraction of this mixed waste is believed to contain a hydrocarbon oil, 4-chloro-3-methyl phenol, diethyleneglycol ether, and propylene glycol. PROCESS CHEMISTRY Silver-based process The classical MEOprocess uses Ag(I1) as an oxidant. Ag(II), in the form of Ag2+,is generated by the anodic oxidation of Ag+ [19].

The oxidation andreduction represented by Equation l are illustratedby the cyclic voltammogram shown in Figure 3.

Potential (V)

Figure 3. Cyclic voltarnmetry with a stationary Pt working electrode in 0.05 M AgNO, and 3.25 M HN03 at 21.5OC. The potential scan rate was 100 mV/sec. Potentials were measured relative to apt reference in pure 3.25 MHN03. The Ag(I)/Ag(II)redox couple is visible between 0.5 and 1.0 V.

510

In the absence of organics, most of the Ag(II) is present as a dark brown nitrate complex, AgN03' [201. Ag2'

+ NO3- <->

PI

AgN03+

The anodic reaction is balanced by the cathodic reduction of nitric acid, HN03. HN03 + 2H+ + 2e- ->

HNO,

+H20

[31

Nitric acid can be regenerated by the chemical reaction of nitrous acid, HNO,, with oxygen. 2HN02 + 0, ->

2HN03

[41

Note that HNO, has a beautiful blue color. After generation, Ag(I1) reacts with H20 and organics. The reduction of Ag(I1) by H20 is represented by Equation 5. 4AgN0,'

+ 2H2O ->

4Ag'

+ 0 2 + 4HN03

151

The kinetics of this reaction have been thoroughly studied by Po, Swinehart, and Allen [21]. Hydroxyl free radicals generated by the reaction of Ag(I1) with water may play an important role in the oxidation of organics. The oxidations of EG and BZ by Ag(I1) are given as examples of organic destruction, and are represented by Equations 6 and 7, respectively. 10AgN03++ (CH20H)2 + 2H20 -> 30AgN03"

+ c6H6 + 12H20 ->

10Ag'

+ 2C0, + lOHNO3

161

+ 30HN03

[71

6C02 + 30Ag'

Formaldehyde and formic acid are the two primary intermediates formed during the conversion of EG to CO,. Phenol, hydroquinone,benzoquinone, and benzoquinoneepoxide are a few of the intermediates formed during the initial stage of BZ oxidation. EG has been used as a surrogate waste in detailed investigations of the ME0 process [13] since studies of its partial oxidation by Ag(II) had been previously published [22-241. BZ has been studied [ 131since it will be aprimary constituent of mixed waste generated by the DWPF [ 171.

Cobalt-based process Co(1II) is generated by the anodic oxidation of Co(I1) and is reduced by reactions with water anddissolvedorganics. Reactionsof Co(1II) with IPA, MCP, andDCP, represented by Equations 8, 9, and 10, are given to illustrate the destruction of both non-halogenated and halogenated organic solvents [ 151. C3H7(OH)+ 5H20 + 18C03+-> C3H6(OH)C1+ 5H20 + 16C03+-> C3H5(OH)Cl, + 5H20 + 14C03+->

3 c 0 2 + 18H+ + 18C02+

+ 17H' + 16C02' + C13 c 0 2 + 16H+ + 14C02' + 2C1-

3C0,

181

r91 [ 101

57 1

The number of electrons required for Co(II1) regeneration during the ME0 of P A , MCP, and DCP are 18,16, and 14, respectively. Note that C1-can be converted to gaseous C1, at the anode.

Iron-based process Fe(II1) is an attractive mediator due to its low toxicity, even though it is a relatively weak oxidant. Oxidizing power depends upon redox potential. The reversible potentials for Ag(II)/ Ag(1). Co(III)/Co(II), and Fe(III)/Fe(II) redox couples are 1.987,1.842, and 0.770 V relative to a standard hydrogen electrode (SHE), respectively. Despite differences in redox potentials, reactions involving Fe(II1) are similar to those with Co(II1) and Ag(I1). ELECTROCHEMICAL GENERATION OF OXIDANTS IN BATCH REACTOR Electrochemical batch reactor Numerous treatability studies have been performed in small electrochemical batch reactors [ 13-15]. A typical batch reactor has a rotating cylinder anode (RCA) that is operated well below the limiting current for Ag(I1) generation (Figure 4). The RCA enables the scientist or engineer to use a small apparatus to mimic mass-transport conditions in a pilot plant, without using massive flow-through electrochemical cells and pumps.

Figure 4. Rotating cycling qnode (RCA) used to generate Ag(II), Co(III), and Fe(II1) can be used to simulate mass-transport conditions in pilot-scale systems.

5 72

Generation of mediator Anodic oxidation is used to generate Co(III),Fe(III), and Ag(II), aggressive oxidizing agents which are known as mediators. The anodic oxidation is represented by the following generalized equation. M"+ ->

&+I)+

+ e-

[I11

Mediators are produced at the surface of the rotating cylinder anode (RCA) shown in Figure 4. The rate of generation,R, is proportionalto the current density, i, and the surface areaof the RCA.

where4andLare thediameter andlengthof theanode,respectively,andFisFaraday'sconstant. The current density, i, should always be maintained below the limiting current density, i,, to achieve the maximum possible current efficiency. The limiting current density is determined by the flux of M"+ to the anode and is defined by Equation 13. i, = FNuD[M"+]/d,

[I31

where Nu is the Nusselt number, D is the diffusivity of M"+, and [M"+] is the concentration of M"+in the bulk anolyte. The dimensionlesscorrelation for the Nusselt number of a RCA is given as Equation 14 [251. Nu = 0.0791Reo.7Sco*356

1141

where Re and Sc are the Reynolds and Schmidt numbers, respectively. The Reynolds number is defined by Equation 15.

where p and p are the viscosity and density of the anolyte, respectively, and o is the angular frequencyof rotation. The Schmidt number representsthe intrinsic transport properties of Ag(I), Co(II), or Fe(II) in the anolyte and is calculated from p, p, and D.

The Reynolds and Schmidt numbers can also be written in terms of the kinematic viscosity, v, which is defined as Np. Empirical expressions for p and p were established by fitting data published in various references [26,27].

p = (1+3.2043~'.~~~~)exp[1723.24(1/r-1/293)] (CP)

[I71

p = ( 1+0.6741X1~0988)exp[49.20( l/r-l/293)] (gm/cm3)

[I81

573

where T is the anolyte temperature in degrees Kelvin and X is the weight fraction of HNO, in the anolyte. Typically, the value of D is approximately cm2/s [27]. In the case of Ag(I), D has been determined from measurementsof i, for anodic oxidation at a rotating disk electrode ( W E )[28]. The RDE was polarized at +0.85 V relative to Pt in 3.25 M HNO,, a stable “pseudo” reference electrode. The rotation speed was varied from 25 to loo0 rpm. The diffusivity was estimated by fitting measurements of i, to the Levich equation. The slope of the fitted line, B, is proportional to D ~ B . B = 0.62(2~)’~FD2/3(p/p)-*~[Ag(I)]

[191

Data has been obtained for several temperatures (4, 21, and 4OOC). Results are illustrated by Figures 5a and 5b and summarized in Table 1. The diffusivity of Ag(1) was found to obey the following empirical relationship [29]. - 0.304 x l F 7 T - 0.279 x lV9T2+ 0.113 x

D = 0.581 x

(cm2/s)

[201

where T is the anolyte temperature in degrees Centigrade. The temperature dependance was found to be greater than expected. It should be noted that models have also been developed for other reactor configurations, such as the annular-flow electrochemical cell [29]. 20

18 16

% a“E .-

Y

l4 l2 lo

8 8 4

2 0 0

1

2

3

4

5

6

7

0

1

2

3

4

5

6

7

Figure 5. (a) Limiting current of a Pt RDE vs the square mot of the rotation speed. The electrode had an active diameter of 0.127 cm. The electrolyte consisted of 0.01 M AgN0, and 3.25 M HNO, at 21OC. (b) Plot similar to that in (a), except the AgNO, concentration was 0.05 M.

574

Table 1 Estimation of diffusion coefficient from RDE data T ("C)

[Ag(I)](M)

B(UGcm2)

Regression coefficient

D(cm*/s)

4

0.01 0.033 0.05

1.100 10-3 2.765 x 4.494 x 10-3

0.976 0.994 0.995

0.686 x 0.527 x 0.494 x

21

0.01 0.033 0.05

1.137 x 3.877 x 10-3 4.928 x lW3

0.982 0.995 0.990

0.602 x 10-5 0.694 x 0.533 x

40

0.01

1.737 x lW3

0.997

1.142 x

DESTRUCTION AND COULOMBIC EFFICIENCIES FOR M E 0 PROCESS Electrolyte Rates of CO, evolution have been measured during the destruction of organics in candidate electrolytes and used to estimate coulombic efficiencies [ 13-15]. In experiments with nitrate solutions, the anolyte compartmentof the electrochemicalcell was charged with 40 ml of 0.5 M AgN0, and 3.25 M HNO,, 0.5 M Co(NO,), and 3.25 M HNO,, or Fe(N03), and 8 M HNO,. The catholyte compartment was charged with 3.25-8 M HNO,, depending upon the concentration of acid in the anolyte. In experiments with sulfate solutions, an undivided cell was charged with 80 ml of 0.5 M CoS04 and 1.63M H,SO,. Electrode separator Electrode separators were needed to separate a stationary Pt cathode from the RCA in experiments with HNO,. This precaution prevented cathodically-generated HN02 from reducing most of the Ag(I1) produced at the anode. Vycor microporous glass (nominal pore size of 40 A) and Nafion 117 cation-exchange membranes were found to be effective barriers to HNO,. Separators made of macroporous ceramics and polymeric anion-exchangers failed to prevent reduction of Ag(I1) by cathodically-generated species such as HNO,. No electrode separator was needed in experiments with Co(II1) in H2S04 since the cathodic reaction was H2 evolution. Rotating cylinder anode (RCA) The RCA had a diameter of 1.2 cm, a length of 1.78 cm, and was made of Au or Pt (Figure 4). The rotation speed of the anode was maintained at 1500 rpm. Experiments were performed at 336 mA (25%iL),673 mA (40% iL),and 1346mA (60%iL). Ohmic heating in the cell resulted in elevations of the anolyte temperature. In the case of 0.5 M AgN0, and 3.25 M HNO,, temperatures were 27,33, and 41-52OC at 336,673, and 1346 mA, respectively.

575

Measurements of CO, evolution As shown in Figure 6, reactions were conducted in a closed vessel so that all CO, could be captured and measured. Measurements of CO, were used to calculate substrate conversion. The volume of this container was 38.2 L, which corresponds to approximately 1.5 moles of gas at ambient temperature and pressure. In early experiments, concentrations of CO, were periodically determined by mass spectrometry of bomb samples. In more recent experiments, concentrations of CO, were monitored continuously with a flow-through dual-beam infrared analyzer (Horiba PIR-2000R, path length of 15 mm), as shown in Figure 7. The absorption of light obeyed Beer’s law, allowing determination of CO, concentration. The analyzer provided

Figure 6. Closed reaction chamber with electrochemical cell and RCA. In this case, the cell is undivided and contains C0S04 and H2SO4.

516

Figure 7. Continuous monitoring system for determination of CO, concentrations in reaction chamber.

an analog voltage signal to a 12-bitanalog-to-digital(AD)converter (National Instruments ATM10-16 Multifunction UO Board) installed in an IBM-compatible personal computer (Intel 80386 microprocessor). Typically, the analog signal from the infrared analyzer was sampled once every minute. Commercial software was used for data acquisition (LabTech Notebook, Wilmington, MA). A peristaltic pump (Cole-Parmer Instrument Company Masterflex, cam speed of 6-600 rpm) provided a gas flow of 1.5 L/min to the infrared analyzer. Before entering the analyzer, the gas was passed through a stainless steel coil (3/8" i.d. tubing, 7 coils, 2" dia.) that was immersed in an ice-water bath (Cole-Parmer Instrument Company Huber Model 126870, temperature range of -20 to +lOO°C). The bath was maintained at 0.2OC to condense corrosive gases that would be harmful to the analyzer. Continuous monitoring of the CO, concentration in the reaction chamber required complete and instantaneous homogenization of the gas phase by a piezoelectric fan.

Calibration of analyzer prior to an experiment, the entire system was flushed with Ar for approximately 15 min and the voltage output from the infrared analyzer was adjusted to zero. The system was then filed with acalibration gas that contained 5.23% CO, so that the gain ofthe measurementsystemcould be determined. Measurements made with the infrared analyzer were validated by periodic comparison with mass spectrometry results. Complete oxidation of EG by Ag(II) in HNO, Aplotofconversionvscumulativechargefor 8.8~10-~rnolesofEGin40mlof0.5MAgN0~ and 3.25 M HN03is shown in Figure 8. The initial concentrationof EG was 0.22 M. Coulombic efficiency was 83-88% at 336-673 mA, but decreased to 39-44% at 1346 mA. Losses in coulombic efficiency were probably due to the parasitic reduction of Ag(I1) by water. The anolyte remained clear during the oxidation of EG and its intermediates, indicating that the bulk AgN03+ concentration was essentially zero and that the organics were oxidized near the surface of the RCA. After complete conversion of EG and its intermediates to CO,,the anolyte turned dark brown, indicative of AgN03+ in the bulk anolyte. Note that AgN03+ has a strong absorbance at approximately 390 nm.

120

I

I

I

I

I

*.'-

-

..

Unexpected rionllnearlty

I

-

-

Theoretical 336 -A673mA 1346mA

a

I 5

I

I

_.-.-.-

J

0

I

100% 7

.

4

-

LI

15 20 10 Total coulombs (thousands)

25

30

Figure 8. EG conversionto CO, by Ag(1I)in HN03plottedagainstcumulativecharge(integrated current). Experiments wereconductedat 24%iL(336mA and27OC),40%iL(673mAand33'C). and 58% i, (1346 mA and 42OC). The electrode separator was a Nafion 117 cation-exchange membrane.

578

Complete oxidation of EG by C O W )in HNO, A plot of conversion vs cumulative charge for 8.8x1W3 moles of EG in 40 ml of 0.5 M Co(N03), and 3.25 M HN03 is shown in Figure 9. The coulombic efficiency was approximately 55% at 673 mA, and decreased to approximately 38% at 1346 mA. Data obtained from experiments with 0.5 M Co(N03), and 8 M HN03 were very similar.

Figure 9. EG conversion to CO, by Co(II1) in HN03 plotted against cumulative charge (integrated current). Experiments were conducted at 336 mA, 673 mA, and 1346 mA. The electrode separator was Nafion 117 cation-exchange membrane.

Complete oxidation of EG by Fe(II1) in HNO, A plot of conversion vs cumulative charge for 8 . 8 ~ 1 moles 0 ~ ~ of EG in 40 ml of 0.5 M Fe(N03), and 8 M HN03 is shown in Figure 10. The coulombic efficiency was approximately 28% at 673 mA, but decreased to 20% at 1346 mA. Complete oxidation of EG by Co(1II) in H,SO, Measurements of EG conversion to CO, were plotted against both time and cumulative charge, as shown in Figures 11 and 12,respectively. The undivided cell was charged with 90 ml of 0.5M C0S04and 1.63 M H2SO4, followed by 8 . 8 ~ 1 0 -moles ~ of EG (initial EGconcentration of 0.11 M). The coulombicefficiency was 72-76% at 336-673 mA, but decreased to57% at 1346 mA. During the oxidation of EG by Co(III), the color of the bulk anolyte was light pink, characteristic of Co(II) solutions (peak absorbance at 513 nm). The concentration of Co(II1) in the bulk anolyte was essentially zero, indicating that the organics were oxidized near the surface of the RCA. After complete conversion of EG to COP the color of the bulk anolyte turned dark

579

Figure 10. EG conversion to CO, by Fe(II1) in HNO, plotted against cumulative charge (integrated current). Experiments were conducted at 336 mA, 673 mA, and 1346 mA. The electrode separator was Nafion 1 17 cation-exchange membrane. 100

1346 mA 80

\

30 20 10 O +

0

I

2

4

6

8 Time (hrs)

I

1 1

I

10

12

14

16

Figure 11. EG conversion to CO, by Co(II1) in H,SO, plotted against time. Experiments were conducted at 336 mA, 673 mA, and 1346 mA. No separator was used.

580 100

90 80 70

2 60 E

E

50

E 40 0

0

30 20

10

0 0

10000

20000

30000

40000

50000

60000

70000

Total coulombs

Figure 12. EG conversion to CO, by Co(1II) in H2S04 plotted cumulative charge. Experiments correspond to Figure 11. purple, characteristic of Co(1II) solutions (peak absorbance at 342 nm). There was no evidence of Co electrodeposition on the cathode. The cathodic reaction was H2 evolution.

Complete oxidation of DCP by Co(I11) in H,SO, Measurements of DCP conversion to CO, were plotted against both time and cumulative charge, and are shown in Figures 13 and 14, respectively. The undivided cell was charged with 90 ml of 0.5 M CoSO, and 1.63 M H2S04, followed by 5.2x1W3 moles of DCP (initial DCP concentrationof 0.065 M). Experimentswere conducted at 336 mA, 673 mA, and 1346mA. The coulombic efficiency was constant at 43% for all three levels of current. Approximately 80%of the DCP charged to the system was converted to CO, at 336 mA. Conversion increased to about 90%at 673-1346mA. The failure to achieve 100% conversion is atmbuted to volatilization of DCP from the anolyte. Complete oxidation of BZ by Ag(I1) in HNO, The cell was charged with 40 ml of 0.5 M AgN03 and 3.25 M HN03, followed by 3.Ox1W3 moles of immiscible BZ. A maximum conversion of 8648% was achieved at 6731346 mA. The failure to achieve 100% conversion is attributed to the volatilization of BZ or an intermediate from the anolyte. The coulombic efficiency was about 67% at 1346 mA. The anolyte remained relatively clear (very light yellow) during the oxidation of BZ and its intermediates, indicating that the bulk AgN03+concentration was essentially zero. Dissolved organics were oxidized near the surface of the RCA, as they were during the ME0 of other organics. The anolyte turned dark brown after complete conversion of BZ to C02, which indicated the presence of AgN03+.

58 1

1 T

loo

80

fi

50

5

40

1346

+336mA

0

30 20 10 1

0

2

4

6

I 8 Time (hrs)

10

12

14

16

Figure 13. DCPconversion to C02by Co(II1)in H2S04 plotted against time. Experimentswere conducted at 336 mA, 673 mA, and 1346 mA. No separator was used.

336 and 673 mA

90 80 70 h

S 60

Y

9s 50 Q)

40 0

30 20 10 0 0

10000

20000

30000

40000 Total Coulombs

50000

60000

70000

Figure 14. DCPconversiontoCO, by Co(111) in H2S04 plotted cumulativecharge. Experiments correspond to Figure 13.

582

Destruction of Trimsol by Ag(II) in HNO, Since Trimsol contains 11% chloride by weight, initial experiments were performed with halide-tolerant Co(II1). Unfortunately, the conversion of Trimsol to CO, by Co(1II) was very inefficient. Poor efficiency may have been due to insufficient oxidizingpower of Co(III), as well as low solubility of the organic waste in acid. SinceTrimsol forms stableemulsions in base, direct anodic oxidation in NaOH was attempted. Unfortunately, this process was also inefficient. The best results were obtained with an electrolyte of AgN03 and HN03, elevated temperature, and a cell current near the mass transport limit. Under these conditions, complete destruction of the Trimsol was achieved at approximately 30% coulombic efficiency. As expected, chloride liberated during the destruction of Trimsol precipitated as AgC1. Results for AgOI), Co(III), and NaOH are compared in Figure 15. 100 90

80

T

Ag(ll), 1346 mA, 25 C I

t

Aafllb 1346 mA. 70 C

70

..

A

5

60

t

\'

Ag(ll), 673 mA, 25 C

I I

Co(lll), 673 mA, 25 C

I

I

z 50

8 40

30 20

10 0 0

2

4

6

8

10 Tlme (hrs)

12

14

16

18

20

Figure 15. Treatability data for the ME0 of Trimsol oil mixed waste.

Problems inherent in Ag(II)/HNO, process Obvious problems with the Ag(II)/HN03 process include: precipitation of Ag as a halide salt by anions liberated during the destruction of halogenated organics; leakage of cations and anions through the electrode separator; rupture, fouling, chemical attack, and radiation-induced degradation of the polymeric separator; excessive corrosionof the anode in the presence of HN03 and

583

HC1; and generation of HN02 and NO, at the cathode. Diffusion of HNO, into the anolyte has been found to lower the coulombic efficiency of the process to less than lo%, but can be prevented by cation-exchange membranes.

Advantages of Co(III)/H,SO, process Co(III)/H,SO, has been investigated as an alternative to Ag(II)/HN03. In addition to eliminating the problem of mediator precipitation, this process chemistry also eliminates the need for an electrode separator. Gaseous H2 is evolved at the cathode, so there is no deposition of metallic Co. Furthermore, Co(II1) reacts with dissolved organics in close proximity to the anode, so it never reaches the cathode. If it did, it would be reduced. Unfortunately, this mediator system is not sufficiently oxidizing to destroy all organic wastes. REACTION INTERMEDIATES FORMED DURING M E 0 OF GLYCOL AND BENZENE

Partial oxidation of EG and BZ at the RCA Knowledge of reaction intermediates and associated mechanisms is needed to thoroughly understand the ME0 process [ 131. To obtain reaction intermediatesfor identificationby GC/MS, EG and BZ were partially oxidized by Ag(I1) with a RCA at 40% i, (673 mA and 33OC). Anolyte samples obtained during the partial oxidation of EG were analyzed without extraction (neat). Anolyte samples obtained during the partial oxidation of BZ were prepared by extracting with solutionsof 75%methylenechloride and 25%isopropyl alcohol. After three extractions,extracts were combined and evaporated under N, at 60OC. Sufficient methylene chloride was added to the wet organic residue to increase the volume to approximately 100 pL. Partial oxidation of BZ at stationary anode BZ was also partially oxidized by Ag(I1) in a small H-cell that had stationary Pt electrodes separated by a porous ceramic frit. The relatively large quantity of BZ used in this experiment enabled intermediates formed during the initial stage of ME0 to accumulate in the cell. First, the catholyte compartment was filled with 35 ml of 3.25 M HN03. Next, the anolyte compartment was filled with 25 ml of 0.5 M AgN03 and 3.25 M HN03, followed by 10 ml of BZ. The cell was polarized under galvanostaticcontrol at approximately 1 A. A magneticallycoupled stir was used for agitation. The anolyte turned light brown (teacolor) immediately. After approximately 20 min of electrolysis, the anolyte and residual BZ phase changed from light brown to bright yellow. Both phases remained yellow during several hours of electrolysis. Samples of the anolyte and catholyte were neutralized and extracted with methylene chloride prior to analysis by GC/MS. Methylene chloride extracts were concentrated to 100 pL volumes by evaporation. The residual BZ phase from the anolyte compartment was analyzed without extraction. Identification of intermediates by GC/MS GC/MS was used in an attempt to identify reaction intermediates in samples of anolyte, catholyte, extract, and residual BZ. GC/MS analyses were performed with a Hewlett-Packard 589015970 benchtop GC/MS system running on RTE/6.

584

Samplesof unextracted electrolyte obtained during the partial oxidation of EG by Ag(II) and Co(1II) were injected directly onto a 30 m long Stabilwax-DA column. This column had an inner diameter of 0.25 mm and a film thickness of 0.5 pm. The head pressure and injection port temperature were 5 psi and 25OoC, respectively. After injection of 2-3 pL of sample, the temperature of the column was ramped from 40 to 24OOC at a rate of 2’C/min. Electron-impact spectra were obtained by scanning from 29 to 450 daltons ( d z ratio of 29 to 450). Samplesof electrolyte extract and residual BZ obtained during the partial oxidation of BZ by Ag(II) were injected onto a 30 m long Quadrex 007 (methyl silicone) column or a 25 m long DB-1 column. Both columns had inside diameters of 0.25 mm and film thicknesses of 0.25 pm. The head pressure and injection port temperature were maintained at 5 psi and 25OoC, respectively. After injection of 2-3 pL of sample, the temperatureof the column was held at 40°C for 10 min and then ramped at a rate of 4’C/min to 300OC.

Intermediates formed during EG oxidation CO,, N20,and formaldehyde were found to elute simultaneously at 1.8-3.8 min. The mass spectrum coincident with the elution at 1.8-2.8 min was characterized by a peak at 29 daltons, which is atmbuted to a CHO fragment formed from formaldehyde, CH20. A large amount of formic acid was detected at 43.52 min, with a smaller amount of acetic acid detected at 39.79 min. Formaldehyde and formic acid were found to be the primary intermediates formed during the ME0 of EG. The same results were obtained with Co(1II). Intermediates formed during BZ oxidation Compounds found in anolyte extracts from experiments with the RCA were acetone, methyl acetate, acetic acid, isopropyl acetate, dioctyl phthalate, and an unknown nitro species characterized by a base peak at 46 daltons. Isopropyl acetate was probably formed by the acid-catalyzed esterification of P A (extraction solvent) and acetic acid (found in the anolyte) in the presence of HN03. Similarly, methyl acetate could have been formed from the reaction of methanol with acetic acid. Methanol could have formed during the destruction of BZ. Since no residual methanol was found, it is assumed that the conversion to methyl acetate was complete. Results are summarized in Table 2. Table 2 Intermediates formed during final stage of oxidation Compound name

Molecular formula

Reten tion time (min)

Place found

acetone acetic acid methyl acetate nitro species (46 m/z ion)

CH3COCH3 CH3COOH CH3COOCH3 unknown

30.5,32.7-33.0 40.0 31.5 45.6,50.4

A A A,C

A,C

Note: “A” denotes anolyte, “B”denotes residual BZ phase, and “C” denotes catholyte.

585

BZ was also partially oxidized by Ag(II) in a small H-cell with stationaryplatinumelectrodes. Compounds identified in anolyte extracts included phenol, hydroquinone, benzoquinone, benzaldehyde, benzoic acid, methyl benzoate, benzonimle, benzonimle aldehyde, and 4-nitro butylnimle. The yellow color of the anolyte was probably due to benzoquinone, which had a relatively high concentration. A compound which was tentatively identified as benzoquinone epoxide (C6H403)was present at the highest concentration and is believed to be a product of the oxidation of benzoquinone. Numerous nitrated aromatics were also detected and include nitrobenzene,dinitrobenzeneisomers, nitrophenol isomers, and dinitrophenolisomers. Intermediates are summarized in Table 3 and classified as: I. BZ substrate; 11. nitrated BZs; HI. phenols, quinones, and epoxides; IV. nitrated phenols; V. BZ substituted with aliphatic and aromatic Table 3 Intermediates formed during initial stage of oxidation Class

Compound name

I

BZ

I1

nitrobenzene 1,Zdinimbenzene 1,3-dinimbenzene 1,4-dinimbenzene

19.0 32.5 31.4 31.6

III

phenol h ydroquinone benzoquinone benzoquinone epoxide 2-nitrophenol 4-nitrophenol 2,4-dinitrophenol 2,6-dinimphenol 33-dinitrophenol

4.4 12.4 8.2 16.0 36.9 39.2 36.1 34.6 36.3

V

biphenyl toluene

31.2 4.0

VI

benzaldehyde benzoic acid methyl benzoate

12.0 25.4 30.4

VII

benzonitrile benzonimle aldehyde 3-nitrobenzo-nimle 4-nitro butyl-nitrile

13.2 31.5, 31.7

IV

Molecular formula

Retention time (min)

Place found

1.8-3.2

28.7 10.0

Note: “A” denotes anolyte, “B” denotes residual BZ phase, and “C” denotes catholyte.

586

groups; VI. aromatic aldehydes, carboxylic acids, and related compounds; or VII. benzonitrile and related compounds. Species identified during this experiment are believed to be representative of intermediates formed during the initial stage of destruction.

REACTION MECHANISMS FOR M E 0 OF ETHYLENE GLYCOL AND BENZENE Plausible mechanism for EG oxidation by Ag(I1) A plausible reaction mechanism for the complete ME0 of EG has been postulated [131. The oxidation of EG probably generates two molecules of formaldehyde or an aldehyde-like intermediate [22]. (CH,OH)2

+ 2AgNO3+->

2CH20 + 2HN03 + 2Ag+

[211

This intermediate would then oxidized by Ag(I1) to produce formic acid [30]. CH,O

+ H 2 0 + 2AgN03+ ->

HCOOH + 2HN0,

+ 2Ag+

[221

The last step toward the complete conversion of EG to C02 probably involves the oxidation of formic acid [31]. HCOOH + 2AgN03+ ->

CO,

+ 2HN03 + 2Ag+

1231

Since ten AgN03+ ions are reduced during the oxidation of a single EG molecule, ten electrons are required forregeneration. Reactionsinvolving Co(II1) are believed to be similar to those with Agm.

Plausible mechanism for BZ oxidation by Ag(I1) The ME0 of BZ is much more complicated than the ME0 of EG. One of the first intermediates formed during the oxidation of BZ by Ag(1I) has been identified as phenol. Referencespertaining to this specific reaction have not been found. However, the conversion of BZ to phenol by Fenton's reagent, a mixture of hydrogen peroxide and ferric ions, is well known [32]. This synthesis is initiated by cathodic reduction of the ferric ion, Fe3+. Fe3+ + e- ->

Fe2+

~241

The ferrous ion, Fe2+, then reacts with hydrogen peroxide to produce hydroxyl free radicals, hydroxyl ion, and femc ion. H202 + Fe2+->

*OH+ OH-

+ Fe3+

Next, the hydroxyl free radical attacks the aromatic ring.

1251

587 Subsequently, the ferric ion is reduced while extracting a proton from the hydroxyl group. (C&j)oH* + Fe3+->

(C6Hs)OH + Fe2++ H+

[271

Note that the kinetic mechanism proposed by Po, Swinehart, and Allen for the reduction of Ag(I1) by H20 involves hydrogen peroxide as an intermediate [21]. Hydroxyl free radicals formed by the decomposition of hydrogen peroxide could react with BZ to produce phenol. Subsequently, phenol could be converted to hydroquinone by the reaction given as Equation 28.

The next step in the reaction sequence would probably be the conversion of hydroquinone to benzoquinone.

An analogous reaction has been studied by Pelizzetti, Mentasti, and Baiocchi [33]. They used Mn(III), Ce(IV), and Ag(I1) to oxidize 4,4'-biphenyldiol to 4,4'-biphenoquinone. Following the conversion of hydroquinone to benzoquinone, benzoquinone would probably converted to benzoquinone epoxide. Unfortunately, little information exists pertaining to this reaction. Nitrobenzene, dinitrobenzene, nitrophenol, and dinitrophenol have been identified as reaction products in HN03. The nitration of aromatic compounds has been studied extensively [34451. Kargin et al. have studied the anodic nitration of aromatic compounds in acetonitrile and sulfolane on a Pt electrode [34]. They attribute the formation of nitrated BZ to a reaction involving either nitronium ion, NO2+, or anode-generated $NO3+, as shown in Equations 32 and 33, respectively.

They produced NO2+ by the anodic oxidation of nitrogen dioxide, NO2. Moodie and Stephens have studied the nitration of aromatics in solutions of dinitrogen pentaoxide, N 0 ,and HN03 2. 5 [35]. In this case, NO2+ is also considered to be the species responsible for nitration of the ring. Dissociation of N205 PFoduces NO2+ and NO3-. The number of nitrated aromatics formed during the ME0 of BZ in the H-cell implies that NO2+ was produced at the stationary anode. Obviously, it is desirable to avoid such nitration reactions. During the complete ME0 of BZ, we found that this problem could be avoided by operating the RCA well below i, for Ag(I1) generation. As the oxidation of BZ approached completion, the predominant species found in the anolyte included acetone, acetic acid, and methanol. The reaction path leading from benzoquinone epoxide to acetone, acetic acid, and methanol has not been established. However, the reaction path leading from acetic acid to CO, is believed to be represented by Equations 32 through 35. CHSCOOH + H2O + 2AgN03' ->

C02 + CH3OH + 2HN03 + 2Ag'

[321

588

CH30H + 2AgN03+ ->

CH20 + 2HN03 + 2Ag+

CH20+ H,O + 2AgNO,+ -> HCOOH + 2AgN03+ ->

HCOOH + 2HN03 + 2Ag+

CO, + 2HN03 + 2Ag+

1331 1341 1351

The final stepsof BZ oxidation are probably the same as those involved in EG oxidation, Kinetic rate constants for the reaction of Ag(I1) with acetic acid, methanol, formaldehyde, and formic acid are given by Mentasti and Baiocchi [30]. Reactions involving Co(II1) are believed to be similar to those with Ag(I1). THEORETICAL MODEL FOR ELECTROCHEMICAL BATCH REACTOR

Differential mass balance for Ag(I1) The dynamic model for therotating-anodeelectrochemical reactor is based upon adifferential mass balance for Ag(1I) and organics [13]. The rate of accumulation of the i-th reactant in the anolyte, VdCddt, is equivalent to the difference between generation and loss terms: VdCi/dt = Generation - Loss

[361

where V is the volume of the anolyte. The anolyte is assumed to be turbulent and well mixed. Obviously, the model can be applied to other mediators.

Generation term for mediator The rate of generation of Ag(I1) at a cylindrical anode, RAg(II),is proportional to the current density, i, and the surface area of the anode.

where d, and L are the diameter and length of the anode, respectively, and F is Faraday’s constant.

Loss of mediator in bulk anolyte After generation, Ag(I1) can be reduced by any of several homogeneous reactions, each represented by a rate, rj Recall that in general rj = kjCj[Ag(II)] The maximum overall rate of loss of Ag(II) is calculated from the sum of individual rates of reaction rates. Equation 36 can be rewritten for Ag(I1) as m Vd[Ag(II)]/dt = (Zd&)(i/F)

+ v(x mjrj) F1

1391

589

whererj is therate of oxidation of thej-th species byAg(I1) and mj is the stoichiometriccoefficient for AgW) in the corresponding chemical equation; e.g., the number of Ag(II) molecules consumed to oxidize a single substrate. This formulation of the model is best at low concentrations of organic when significant levels of Ag(II) can exist in the bulk anolyte.

Reaction of mediator and organics at anode surface At high organic concentrations, the model assumes that all Ag(II) generated at the surface of the anode reacts with the organic substrate or its intermediates before reaching the bulk anolyte. All reactions are assumed to occur in a thin film of anolyte at the anode surface which we refer to as the “reaction layer.” The rate of generation of Ag(I1) at the anode, RAgOIp is balanced exactly by the sum of all rates of organic oxidation in the reaction layer, as shown in Equation 40. rn

‘Ag(1Q

= (dfA)(x mjrjavg) j= 1

[401

where df is the thickness of the reaction layer, A is the surface area of the anode, m, is the stoichiomemc coefficient for the j-th reaction, and rjav.g is the average rate of reaction in the anolyte film, based upon an average reactant concentration. In this case, the loss of Ag(II) due to reduction by H 2 0 is assumed to be insignificant. The fractional consumption of Ag(I1) by the i-th reaction is m ~i = (qriavg)/@ mjrjavg) F1

1411

Since all rates, rjavg, are first order in Ag(II) concentration, [Ag(II)J appears in both the numerator and denominator of Equation 4 1 and can be eliminated. Therefore, Oi is independent of AgOI) concentration. If the averageconcentrationof areactant in the reaction layer is assumed to be approximately ( 1/2)Cj,the factor (1/2) also appears in both the numerator and denominator of Equation 41 and can be eliminated. Thus, the approximate value of Oi can be calculated from the concentrations of organics in the bulk anolyte, Cj, and corresponding rate constants. The total rate of oxidation of the i-th organic in the anolyte film, (dfA)ri,ay, can be expressed in terms of the generation rate of Ag(II), RAg(II),by combining Equations 0 and 41.

The volume-average rate of oxidation used for predicting the concentration of the i-th reactant in the bulk anolyte, (ri), is derived by dividing (dfA)riavgby the total volume of anolyte, V. (ri>= (lN)(dfA)ri,avg

[431

By substituting Equation 37 into Equation 42 and Equation 42 into Equation 43, an expression for (ri) in terms of Oi, geometry, and current density is derived. (r1) = ( 1 N ) ( Q ~ q m q A r n

[441

590

Note that (ri) is only the loss term for the i-th reactant. The loss term for one intermedate, (rj-l), is identical to the generation term for the next intermediate in the reaction sequence, (ri). Equation 45 defines the rate of accumulation in terms of volume-average rates defined by Equation 44. dCj/dt = (ri-l) - (ri)

1451

Numerical integration of differential equations The system of ordinary differential equations represented by Equation 45 can be integrated numerically with a fourth-order Runge-Kutta algorithm, as discussed by Carnahan and Wikes [46]. Ci,m+l = (W(kl,i+2kzj+2k3j+k4,i)

r461

where Ci,m+lrepresents the concentrationof the i-th reactant at time tm+1, h is the time increment from t, to tm+l, and

This algorithm is executed simultaneously for each reactant, including Ag(II), ethylene glycol, formaldehyde, formic acid, and carbon dioxide.

APPLICATION OF MODEL TO THE ME0 OF GLYCOL BY SILVER@) Observed nonlinearity in conversion-time curves The instantaneous conversion of EG or any other organic to CO, would be characterized by the absence of chemical intermediates in the anolyte and a linear conversion-time (CO, evolution) curve. In such a case, conversion would increase linearly with time to a maximum of 100%. Contrary to expectations, nonlinearity (curvature) was found to be one of the most obvious features of actual conversion-timedata. A simple model has been formulated to explain thisnonlinear behavior by taking into account the sequential oxidation of known and suspected chemical intermediates [ 131. Application of model to ME0 of EG Consider applicationof the model representedby Equation45 to the conversionof EG to C02 Formaldehyde and formic acid are the two primary intermediates. Partial oxidation of EG to formaldehyde by Ag(II) in acidic perchlorate solutions has been investigated by Mentasti and

59 1

Kirschenbaum [22]. Kumar has performed similar studies [23-241. These investigators assume that the first step in the reaction involves complexation of Ag2+ and EG.

The species in Equation 51 are in equilibrium and obey Equation 52. [Ag-(CH20H)22+l= KqK1[Ag2+lI(CH20H)21

[521

where Kq and K, are equilibrium constants. The rate of Ag-(CH20H)22+decomposition determines the rate at which EG is converted to the first intermediate, which has been shown to be formaldehyde. d[(CH20H)2]/dt = -k[Ag-(CH20H)22+]

[531

where k is a first-order rate constant. This rate equation can be expressed in terms of [Ag2+]and [(CH20H)2]by substituting Equation 52 into Equation 53. d[(CHzOH)2]/dt = -kKeqKl[Ag2+][CH20H)2]

[541

Mentasti andKirschenbaumdetermined the value ofk from transient absorbance data. They used spectrophotometryto monitor the decay of Ag2+ due to its reaction with EG. Ag(IX), which is indicative of Ag2+,has a strong absorbance maximum at 470 nm in acidic perchlorate solutions. Before Equation 54 can be applied to situations with acidic nitrate solutions, the equilibrium between Ag2+andall formsof Ag(II) must be taken into account. The equilibrium between Ag2+ and AgNO,' is represented by Equation 55. K4

Ag2+ + NO3- <->

AgN03+

Other forms of Ag(II), such as Ag-(CH20H)22+and AgOH+, must also be considered. The equilibrium between Ag2+ and all forms of Ag(II) should obey Equation 56. [Ag2+l = tAg(WIA1 + Kd[H+l + K4[N03-l + K,qKI[(CH20H)21)

1561

where Kh and K, are equilibrium constants. The rate equation applicable to acidic nitrate solutions is derived by substituting Equation 56 into Equation 54. dt(CH2OW2l/dt = -kKqK 1[Ag(II)I[(CH20H)21/ (1 + Kh/[H+l+ K,[NO,-l

+ KeqKI[(CH20H)21)

[571

The reaction of Ag(I1) and EG eventually produces formaldehyde. The conversion of Ag(CH2OH)? to formaldehyde requires a second electron-transfer step which probably involves another Ag2+ion.

592

Mentasti andBaiocchi [30] givean empiricalrateequationfortheoxidationofformaldehyde.

where 4''is a second-order rate constant. Equation 58 can be used to calculate the rate of conversion of formaldehyde to formic acid. The next step toward the complete conversion of EG to CO, probably involves the oxidation of formic acid. The oxidation of formic acid by Ag(I1) is relatively simple and the kinetics have been investigated by Miller and Morrow [31].

where the first oxidation of the formate anion is assumed to be rate limiting. The rate expression for formic acid oxidation is d[HCOOH]/dt = -kl[Ag2+][HCOZ-]

r611

The dissociation of formic acid is assumed to be at equiIibrium. K3

HCOOH c->

H++ HC0,-

Consequently, the concentration of HC02- can be defined in terms of the concentration of HCOOH.

The rate expression for formic acid oxidation becomes d[HCOOH]/dt = -k K3[A& [HCOOHI/[H+l

[MI

By substituting this equilibrium expression for [Ag2+], the rate equation for HCOOH can be written in terms of [Ag(II)].

By ignoring [Ag-(CH20H)p] and [AgOH'], the expression becomes

593

which is the equation appearing in the literature [31]. The rate of CO, evolution is equivalent to the rate of formic acid oxidation.

Equilibrium and rate constants Values of equilibrium and rate constants for the oxidation of EG, formaldehyde, and formic acid are available in the literature, for at least one temperature. Data are summarized in Table 4. The temperature dependance of Kh is characterized by an enthalpy of reaction of approximately 12 kcal/mol. Other equilibrium constants were assumed to have identical enthalpies of reaction. All activation energies were assumed to be approximately 25 kcal/mol. Table 4 Equilibrium and rate constants from literature Constant Kh

KlK, K3

K4 k kl k2"

Value 0.15 x 1 F3 0.31 x l F 3 2.32 x 105 1.77 x l F 7 0.94 x 103 3.90 x 103 13.8 3.9 x 108 1.6 x 107

Units

T ("C)

Reference

10 20 10 25 25 25 10 25 22

Predictions for ethylene glycol Predictions and experimental measurements of EG conversion to CO, by Ag(II) at 40% i, (33°C and 673 mA) are compared in Figure 16. Calculated concentrationsof EG, formaldehyde, and formic acid are also shown. Comparisonsfor 24% and 58% i, can be found in the literature [13]. Conditions are summarized in Table 5. According to the model, EG is consumed by its conversion to formaldehyde. Formaldehyde reaches a maximum level when its rate of generation, due to the oxidation of EG, is equivalent to its rate of conversion to formic acid. Similarly, formic acid reaches a maximum level when its rate of generation, due to the oxidation of formaldehyde,is equivalent to its rate of conversion to CO,. The observed nonlinearity in the conversion-time curve was successfully predicted by accounting for the sequential formation of reaction intermediates. Conversion-timecurves for the oxidation of formic acid, a reaction with no accumulation of intermediates, exhibited no such nonlinearity.

594 0.5 -

I

I

I

1

I

~

I

120

Formic acid 0.4

- Formaldehyde

E 0

n

-60

9 C

0

0

3

2

1

4

5

7

6

Tlme (hr)

Figure 16. Comparison of model predictions and experimental data for ME0 of EG by Ag(I1) in the electrochemical batch reactor. The cell was maintained at 40% i, (673 mA and 33OC). Square (M) and triangular (A)symbols represent data obtained during experiments with Nafion 117 cation-exchange membrane and Vycor microporous glass, respectively.

Table 5 Conditions assumed for predictive calculations Current(mA)

T(OC)

336 673 1346

27 33 43

Re

12,650 14,120 16,790

Sc

Nu

iL (mA/cm2)

i/iL

1275 936 519

750 726 664

212 250 346

0.24 0.40 0.58

MAXIMUM MEDIATOR CONCENTRATION- NO ORGANICS Maximum possible mediator concentration in anolyte The maximum possible Ag(I1) concentrationcan be reached in the absence of organics. This steady-state concentration can be calculated by setting the accumulation term in the differential

595

steady-state concentration can be calculated by setting the accumulation term in the differential mass balance, d[Ag(II)]/dt, to zero. This condition requires that the rates of Ag(1I) generation and loss be equivalent.

Substitution of the expressions for i, and k,, leads to the quadratic:

where [Ag(I)], is the initial concentration of AgN03. Note that at any instant, [Ag(I)], is equivalent to [Ag(I)] and [Ag(II)]. The solution of this quadratic is simply [Ag(II)] = ((B2- 4AC)lR -B)/(2A)

[701

where

Reduction of Ag(II) by reduction with water The kinetics of the reduction of Ag(I1) by water have been thoroughly studied by Po, Swinehart, and Allen [21]. They explored a broad range of conditions: 2-9 M HNO,; 104-10-2 M AgN03; and 298-308'K. The rate was found to be proportional to [Ag(II)I2for Ag(T1) concentrations greater than lo4 M. In order to explain the observed second-order kinetics, Po, Swinehart, and Allen postulated a mechanism which assumes that two Ag(I1) ions react together and produce Ag(1) and Ag(1II). The rate equation for the reduction of Ag(I1) by water is d[Ag(II)]/dt = -kII[Ag(II)]2 (Ms-')

[741

where

Note that k/ = kJH20]. From the data of Po, Swinehart, and Allen, we have determined the rate constants quantitatively [21,29].

596

u

= kxp[-10.9186

+ 7921.3/(1.987")]

(M-'s-')

1781

p = exp[-48.2540 + 32746./(1.987T)l (M-3)

[791

As previously stated, second order kinetics is applicable for cases where the concentration of Ag(II) is greater than 1 e M . For lower concentrations of Ag(II), the rate has been found to be first order.

The conclusions of Po, Swinehart, and Allen have been verified by monitoring the decay of the peak absorbanceof AgN03+located between 390 and 400 nm [121. Spectraillustrating the decay of Ag(I1) by reduction with water is shown in Figure 17. As expected, the stability of Ag(II) in the bulk electrolyte was found to increase with decreasing temperature, increasing concentrations of AgNO,, and increasing concentrations of HN03.

1.753 1.461

-

-

I

I 2 mln

I

I

,.,

n nc u A n u n 1.11

A

I

I I"-"

.."..(

31s. uIuun

-

-

1.168 3768

r

II

Wavelength (nm)

Figure 17. UV-visible spectra showing the decay of Ag(II), in the form of AgN03+, due to reduction by water. Note the strong absorbance at approximately 390 nm. The electrolyte was 0.05 M AgN03 and 3.25 M HN03 at ambient temperature.

Steady-state Ag(I1) concentration After complete oxidation of all organics, Ag(II) begins to accumulatein the bulk anolyte and reaches a steady-state concentration (Figure 16). The steady-state Ag(II) concentration is established when the rates of Ag(II) generation and loss are equivalent. At this point, the only loss of Ag(I1) is due to the reduction by H,O. Steady-stateconcentrations of Ag(I1) in the bulk

591

anolyte can be calculated by setting the accumulation term in the differential mass balance for Ag(II) to zero. Results of thesecalculations forthe small electrochemicalbatch reactorare shown in Figure 18. Similar calculations have been done for flow-through electrochemical cells with recirculating anolyte loops [29,47]. Measurement of the steady-state concentration of Ag(I1) is useful in characterizing reactor performance.

0.25

I

I

I

I

I

I

I

I

I

Predictions at various temperatures ("C) with no organics

-

n n Y

2 0.10 Y

0

20-

i 0

4050 , 6070 I

I

200

400

I

I

I

I

I

I

I

600 800 1000 1200 1400 1600 1800 2000 Speed (rpm)

Figure 18. Maximum-possibleAg(I1) concentration possible after all organics are converted to COP Refer to Fig. 23. Predictions are given for rotation speeds from 100 to 2000 rpm and for temperatures from 20-70°C.

OTHER RECOMMENDED READING In addition to references already cited, several other publications are recommended for reading. These include works on: electron transfer reactions in organic chemistry by Eberson [48]; use of Ag(I1) as a coulometric titrant by Lingane and Davis [49-501,as well as by Schotherst and Den Boef [511; current efficiency losses by Comninellis, Grissen, and Platner [52-531;and various studies of mediator and radionuclide electrochemistry [54-561.

SUMMARY Incineration may not be an acceptable process for the treatment of mixed wastes due to possible high-temperature volatilization of radioactive materials. Electrochemical oxidation may be a good alternative. Several laboratoriesthroughout the world are exploring the possibility

598

of treating mixed and hazardous wastes electrolytically. Numerous organic wastes have been completely converting organic components of mixed wastes to CO, at low temperature by reaction with Ag(II), Co(III), Fe(II1) and other mediators. Pmcess efficiency has been determinedquantitativelyand severalimportantreaction paths have been established. Principles of mass and kinetics essential for process modeling and scale-up have been discussed in detail. This new waste treatment technology has been successfullydemonstrated on a pilot-plant scale (5 L/d capacity).

ACKNOWLEDGMENTS This work has been supported by the Advanced Process Technology (APT)Program, which is part of the Laser Program (Y-Division). The author is grateful to Zoher Chiba, Steve Gordon, Patricia Lewis, Leslie Summers, and Francis Wang for valuable collaborations. Raymond Bedford, Ruth Hawley-Fedder, and Linda Foiles are thanked for various analytical work. The CO, detection system was developed by Chris Ebbers. This work was done under the auspices of the United StatesDepartment of Energy by Lawrence LivermoreNational Laboratory under Contract No. W-7405-Eng-48.

REFERENCES 1 0. H. Krikorian, Proc. Incineration Conf., Knoxville, TN, May 13-17,1991; Lawrence Livermore Lab. Rept. UCRL-JC-106163 (1991). 2 J. F. Cooper, W. Brummond, J. Celeste, J. C. Farmer, C. Hoenig, 0. H. Krikorian, R. Upadhye, R. L. Gay,A. Stewart,S. Yosim,Proc. Incineration Conf., Knoxville,TN, May 13-17, 1991; Lawrence Livermore Lab. Rept. UCRL-JC-107288(1991). 3 0. H. Krikorian, Lawrence Livermore Lab. Rept. UCRL-ID-107877 (1991). 4 R. L. Clarke, A. T. Kuhn, E. Okoh, Electrochem. Brit. (1975). 5 D. Anderson, R. L. Clarke, Proc. Water Effluent Treat. Conv.,Birmingham,UK (1978). 6 R. L. Clarke, P. C. Foller, “Electrochemical Hydrogen Technologies,” H. Wendt, Ed., Elsevier, Amsterdam, The Netherlands (1990). 7 R. L. Clarke, “Iron(II1) as a Catalyst for the Anodic Destruction of CarbonaceousWaste,” Dextra Associates, Orinda, CA (1990). 8 P. M. Dhooge, Su-Moon Park, J. Electrochem. Soc., 5,1029-1036 (1983). 9 P. M. Molton, A. G. Fassbender, S. A. Nelson, J. K. Cleveland,Proc. 13th Ann. Environ. Qual. R&D Symp., Battelle Pac. Northwest Haz. Waste RD&D Center, Richland, WA (1988). 10 E. J. Wheelwright, L. A. Bray, J.L. Ryan, Intl. Pat. Appl. No. PCT/US89/01374,Intl. Publ. No. WO 89/10981 (1989). 11 D. F. Steele, Platinum Met. Rev., 2,10-14 (1990). 12 J. C. Farmer et al., Extended Abs. 178th Meeting of the Electrochem. Soc.. Electrochem. Soc.,Pennington, NJ (1990). 13 J. C. Farmer, F. T. Wang, R. A. Hawley-Fedder, P. R. Lewis, L. J.Summers, L. Foiles, Lawrence Livermore Lab. Rept. UCRL-JC-107043(1991);J. Electrochem. Soc., 3, 654-662 (1992).

m,

m,

599

14 J. C. Farmer, F. T. Wang, P. R. Lewis, L. J. Summers, Lawrence Livermore Lab. Rept. UCRL-JC-109134 (1991); Roc. Intl. Sym. Electrochem. Engr. Envir., U. Tech., Loughborough, UK, April 7-9,1992. 15 J. C. Farmer, F. T. Wang, P. R. Lewis, L. J. Summers, Lawrence Livermore Lab. Rept. UCRL-JC-109633 (1992); J. Electrochem. Soc.,in press. 16 Electrocinerator, Technologies, Inc., P. 0. Box 430, East Amherst, NY 14051. 17 “Defense Waste Processing Facility (DWPF) Process Description: Overview of DWPF Process,” DPSOP 257-8, Part 2, Item 100, Rev. 5 , pp. 1-6 (1989). 18 S. Glasstone, A. Sesonske, “Nuclear Reactor Engineering,” 3rd Ed., pp. 517-521, Van Nostrand Reinhold, San Francisco, CA (198 1). 19 M. Fleischmann, D. Pletcher, A. Rafinski, J. Appl. Electrochem., 1,1-7 (1971). 1698-1701 (1961). 20 J. A. McMillan, B. Smaller, J. Chem. Phys., 21 H. N. Po, J. H. Swinehart, T. L. Allen, Inorg. Chem., 2,244-249 (1968). 283-288 (1987). 22 E. Mentasti, L. J. Kirschenbaum, Inorg. Chim. Acta, 23 A. Kumar, J. Phys. Chem., 86, 1674-1678 (1982). 24 A. Kumar, J. Am. Chem. Soc.,1111,5179-5182(1981). 25 J. S. Newman, “Electrochemical Systems,” pp, 324-326, Prentice Hall, Englewood Cliffs, NJ (1973). 26 A. V. Wolf, M. G. Brown, P. G. Prentiss, “CRCHandbook of Chemistry andPhysics,” 61st Ed., p. D-247, R. C. Weast, M. J. Astle, Eds., Chem. Rubber Co. Press, Boca Raton, FL (1980). 27 C. 0.Bennet, J. E. Myers, “Momentum, Heat, and Mass Transfer,” pp. 775-777, McGrawHill, San Francisco, CA (1974). 28 A. J. Bard, L. R. Faulkner, “Electrochemical Methods, Fundamentals, and Applications,” pp. 283-298, John Wiley and Sons, New York, NY (1980). 29 J. C. Farmer, Z. Chiba, R. G. Hickman, Lawrence Livermore Lab., UCRL-LR-106858 (1991). 30 E. Mentasti, C. Baiocchi, Coord. Chem. Rev., S, 131-157 (1984). 31 L. Miller, J. I. Morrow, Inorg. Chem., 15,1797-1799 (1976). 32 H. Wendt, Electrochim. Acta, 29, 1513-1525 (1984). 33 E. Pelizzetti, E. Mentasti, C. Baiocchi, J. Inorg. Nucl. Chem., B,557-561 (1976). 34 Y. M. Kargin, Y. V. Rydvanskii, G. A. Evtyugin, D. A. Semanov, P. V. Yagin, L. N. Punegova, G . A. Marchenko, Zh. Obshch. Khim., 54, 2112-2115 (1986), translated by Plenum Publ. Corp. (1987). 35 R. B. Moodie, R. John Stephens, J. Chem. Soc.,Perkin Trans. 11,1059-1064 (1987). 36 A. Boughriet, M. Wartel, J. Chem. Soc.,Chem. Commun., 13,809-810 (1989). 37 A. Boughriet, M. Wartel, J. Electroanal. Chem., 262,183-194 (1989). 38 Y. V. Rydvanskii, G. A. Evtyugin, D. A. Semanov, Y. M. Kargin, Zh. Obsch. Khim., 2140-2143 (1986), translated by Plenum Publ. Corp. (1987). 39 Y. M. Kargin, Y. V. Rydvanskii, G . A. Evtyugin, D. A. Semanov, P. V. Yagin, Zh.Obsch. Khim., s,2557-2561(1987), translated by Plenum Publ. Corp. (1988). 40 Y. M. Kargin, Y. V. Rydvanskii, G. A. Evtyugin, D. A. Semanov, Zh. Obshch. Khim.,Z, 2562-2565 (1987), translated by Plenum Publ. Corp. (1988). 41 G. A. Evtyugin, D. A. Semanov, Y. V. Rydvanskii, V. Z. Latypova, Y. M. Kargin, Zh. 1864-1870 (1988), translated by Plenum Publ. Corp. (1989). Obshch. Khim., 3,

a,

m,

s,

600

42 L. Eberson, F. Radner, Acc. Chem. Res., 2,53-59 (1987). 43 A. J. Bloom, M. Fleischmann, J. M. Mellor, Electrochim. Acta, 2,785-790 (1987) . 44 0. Hammerich, V. D. Parker, “Advances in Physical Organic Chemistry,” V. Gold, D. Bethell, Eds., Vol. 20, pp. 55-189, Academic Press, San Francisco, CA (1984). 45 J. M. Achord, C.L. Hussey, J. Electrochem. Soc.,l2& 2556-2561 (1981). 46 B. Camahan,J. 0.Wikes, “DigitalComputing and Numerical Methods,” pp. 389-391,John Wiley and Sons, New York, NY (1973). 47 Z. Chiba, C. Dease. Lawrence Livermore Lab. Rept. UCRL-JC-105655 (1991); Roc. AIChE 1991 Summer Natl. Mtg., Treat. Rad. Mix. Wastes, Pittsburgh, PA, Aug. 18-21, 1991, Am. Inst. Chem. Engr. (1991). 48 L. Eberson, Proc. 29th IUPAC Congr., Cologne, FRG,June 5-10, 1983,Pergamon F’ress, New York, NY (1983). 49 J. J. Lingane, D. G. Davis, Anal. Chim. Act., 15,201-206 (1956). 50 D. G.Davis, J. J. Lingane, Anal. Chim. Act., U,245-252 (1958). 51 R.C. Schothorst, G.Den Boef, Anal. Chim. Act., 99-107 (1985). 52 Ch. Comninellis, E. Plattner, J. Appl. Electrochem., l2,1315-1318(1987). 1,72-76 (1985). 53 Ch. Comninellis, Ch. Griessen, E. Platmer, J. Electrochem. Soc., 54 L. A. Bray, J. L. Ryan,E. J. Wheelwright,Proc. Am. Inst. Chem. Engr. Ann. Mtg., Miami, FL, Nov. 2-7, 1986, Am. Inst. Chem. Engr. (1986). 55 C. Madic,M. Lecomte,B. Vigreux, Roc. Waste Mgmt. 90,Tucson,Feb. 25 toMar. 1,1990, Numatec, Bethesda, MD (1990). 56 N. Furuya, M. Shibata, J. Electroanal. Chem., 262,321-324 (1989). 57 Y. Zundelevich, Lawrence Livermore Lab. Rept. UCRL-JC-106889 (1991); J. LessCommon Metals (1991).

m,

m,

601

Magnetic Field Effects in Environmental Control Involving Electrolytes Thomas 2. Fahidy Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada I.

PREAHELE AND MOTIVATION

The magnetic properties of electrolytes and the interaction of externally imposed as well as internally induced magnetic fields on electrolyte properties and ionic transport characteristics has been the subject of steadily increasing interest since Faradaic times. Apart from well-established applications in analytical chemistry (e.g. NMR techniques), magnetic fields have been shown to exert a theoretically intriguing and practically important influence on electrolyte behaviour on a microscopic and a macroscopic scale. Environmental control, a topic of rapidly expanding importance, is perhaps the newest dominion of application, where much progress can be foreseen in the utilization of magnetically supported force fields. Waste water and effluent treatment is one of the most likely growth areas, although with a currently still modest presence among more mature techniques (Table 1). Magnetoelectrolysis itself has not yet acquired a distinct place in the Table 1 Comparison of literature in selected subfields of waste water and effluent control (Source: Computerized search in the Chemical Abstracts, August 19, 1991, Davis Centre Library, University of Water loo 1

Number of citations

Search key words Effluent Wastewater Purification Water purification Electrolysis and effluent Electrolysis and wastewater Magnetic field Magnetic field and effluent Magnetic field and wastewater Magnetoelectrolysis Magnetoelectrolysis and effluent and/or wastewater TOTAL

17089 89145 107521 40197 266 1236 31207 17 137 28

Relative number of citations, per cent 5.96 31.08 37.48 14.01 0.093 0.431 10.89 0.0059 0.048 0.0098

0

0

268843

100

realm of environmental control, but the potential of magnetic fields in this area has been clearly demonstrated. The objective of this chapter is twofold. The first goal is to describe, albeit in a condensed manner, the physical foundations of the magnetic field effect from a microscopic (ionic-scale), and from a macroscopic (fluid continuum-scale) viewpoint employed in the pertinent literature. The second goal is to illustrate the current state of art in three particular areas of environmental control: waste fluids and effluents, liquid phase corrosion and natural bodies of water (e.g. groundwater and water used for agricultural purposes). In these areas magnetic fields may well serve in the future as one of the control variables. The approach employed here is to indicate by representative examples and selected literature the reach of the subject matter; no claim of a comprehensive and all-encompassing treatment is made. If the reader's appetite is sufficiently whetted for further enquiry and reading, the raison d'&tre of this chapter has essentially been achieved. II. THEORETICAL FOUNDATIONS 11.1. The Microscopic (Ion-Scale] Approach The simplest model accounting for the magnetic field effect is an adaptation of the classical noninteracting-hard-ball approximation where the number of hard ball-like ions is NAci. In homogeneous coupled E l and magnetic (HI electric ( - fields (i.e. grad E = 0; grad H = 01, the a-priori thermal motion of an ion is deflected by the magnetic field if the magnitude of its interaction energy, is comparable with its

bH.

thermal energy, kT. If the interaction is sufficiently small, the ionic velocity is essentially thermal and the magnetic field effect is manifested by the classical Lorentz force

F = p ez e( V -L -T x u )

(1)

If the ionic atmosphere effect can be neglected, V 1/2

of a free ion, (kT/ui)

.

T

In general, the sum of external forces

acting on an ion comprising the electric force F

Ef, and

is essentially that

%'

the frictional force

the magnetic force F&, determines the change in momentum. In the electrolyte bulk, and under stationary conditions, the magnetic and the friction forces balance each other, since F g 0. In the vicinity

-E

of a solid e.g. of an electrode-electrolyte interface (and notably in the electric double layer) the ionic velocity due to a nonzero net electric field, is considerably smaller than the velocity of the centre of the total ionic mass given by the expression

603 whose experimentally observed values lie in the 4

5 < 10 A h ;

- 1 m/s range

2

(10 < H 100 < j < 1000 A h ; 1 < E < 10 V/cm). Although ionic momentum changes, especially the time instant of change, cannot be evaluated exactly, the view that imposed magnetic fields exert a short term shock wave-like effect on ions until the onset of viscous deceleration by the solvent has been proposed [1,2] to explain the observation of relatively large-scale effects triggered by weak magnetic fields. It is more likely, however, that magnetohydrodynamic (MHJI) considerations treated in Section 1.2 are also necessary for at least a semi-quantitative estimation of such effects. The boundary layer/interface behaviour in imposed magnetic fields is especially interesting in the case of capillary flow, where the calculation of the tangential velocity in a single rigid capillary depends on the size of the diffusional portion of the double layer relative to the pore radius [31. In nonhomogeneous magnetic fields imposed on paramagnetic ions or microscopic particles, the velocity equation [41 (3)

i 2

indicates that magnetic field gradients in the order of 1 GA/m2 - 1 TA/m would be required for sizeable magnetic effects. Such conditions could exist at liquid-ferromagnetic solid surfaces, i.e. at magnet surfaces, even if the mean magnetic field strength is relatively small (0.1 - 1 MA/m). The microscopic approach has been particularly successful in the treatment of the Hall effect in electrolytes, summarized in an earlier overview 151. As in the case of Hall conductivity, the magnitude of the magnetic field effect on diffusion is very small [6,71 but not negligible in a rigorous sense. The Lielmezs-Musbally formula [61 based on the theory of irreversible thermodynamics for bi-ionic systems:

in terms of ionic and counterionic velocities -1' V V predicts e.g. -2 D /D I 1.006 in a 0.01 mol/L aqueous NaCl solution at 25OC, subjected m o to a magnetic flux density of one tesla. Classical Hall-effect calculations have been further refined by considering electrohydrodynamic forces [81 opposing the Lorentz force acting on a ion. 11.2.

The Magnetohydrodynamic (MHD) Approach

MHD theory [9-141 has been applied extensively 1e.g. 15-211 in conjunction with convective diffusion theory [22-261 to the analysis of external magnetic field effects in the hydrodynamic and concentration boundary layer existing at the electrode/electrolyte interface [2,5,271.

604 The central tenet of the MHD approach is the rotational contribution to the overall force field by the magnetic force term in the classical vorticity equation, whose expanded form

indicates clearly that even if curl (J x B) = 0 (e.g. in the case of negligibly small induced magnetic fields), the second term on the right-hand side of Eq. 5 cannot be neglected in a boundary layer, unless J and are strictly collinear. If the electric and/or magnetic field is nonhomogeneous, the first term will not be zero, hence the imposition of a nonhomogeneous magnetic field on an electric field spanning an electrolyte may generate excessive turbulence in the boundary layer. As shown experimentally [181, if J and are sufficiently large, surface ripples and instability can easily be generated. It follows directly that ionic mass transport rates can be enhanced, or suppressed [281 according to fieldand geometric configurations in magnetoelectrolytic cells. The dimensionless mass transport rate (known as the Sherwood number) has been related in experimental and theoretical studies [e.g. 2, 17, 27, 29-341 to various powers of the magnetic field strength, the most frequent numerical value being cited as 1/3 and 112. The discrepancy does not appear to be, however, statistically significant [351. Laser interferometric studies [361 on copper deposition from aqueous CuS04 electrolytes provide clear experimental evidence for the magnetic field effect on the growth the diffusion layer at solid cathode surfaces. These findings are in agreement with theoretical predictions of mass transport improvement due to magnetic field-induced modifications of the diffusion layer structure. 11.3.

Kinetic and Thermal Aspects of the Magnetic Field Effect

The effect of magnetic fields on electrode kinetics has not yet been clearly demonstrated, although the modification of Tafel slopes (37, 381 and interaction with the catalytic activity of ferromagnetic catalysts [39-411 has been reported in certain cases. Conversely, the absence of magnetic field effects on Tafel parameters has equally been observed in the anodic dissolution of copper in aqueous sodium chloride [421, and other solutions [431. The existence of thermal gradients in aqueous electrolytes due to electric and magnetic field interactions has been extensively investigated by applying the theory of statistical mechanics [44, 451 to experimental observations [46-481. Direct visualization studies [e.g. 4, 49-521 furnish further evidence for turbulence generation by magnetic fields, and the existence of strong local vortices in electrolytic cells 14, 52, 531. The Magnetic Field Effect in Environmental Electrochemistry: Preliminary Notions

11.4.

A logical question to ask upon the reading of Section I is how magnetic fields can be employed efficiently to improve electrochemical

605

means of environmental protection. Water purification, effluent treatment and corrosion prevention are the most important target areas. Magnetoelectrolysis has not yet been shown to be a globally effective method, however, magnetic manipulation ("magnetization"1 has been remarkably successful in certain areas of environmental control and holds a definite potential in areas not sufficiently explored so far. The following sections describe avenues of magnetic field applications of current and future interest. A possibly major impediment to the rapid expansion of this field lies in the limited understanding of the magnetic field effect on solution properties. A case in point are experimentally observed local extrema of certain physical quantities of water employed in injection-based oil recovery [541. According to laboratory tests with imposed magnetic field strengths ranging up to about 400 A h , water hardness, dissolved oxygen, alkalinity, viscosity, pH and the concentration of suspended solids reach a minimum value at H = 380 A/m, shown in Table 2. While these and similar quantitative data permit, at least in principle, to design optimal magnetic flocculators (Section 1111, no currently available mechanistic/mathematical models can predict and fully interpret such phenomena. As seen in Section 111, quantitative relationships describing the magnetic field effect on environmental control variables are mostly statistical regression equations, indicating a vast territory ripe for fundamental studies of process mechanisms. Table 2 The Effect of Magnetic Fields on Certain Properties of Water Using In Injection Oil Recovery 154, Figs. 1-61 Magnet ic Field Strength

A/m

354 380 384 400

Quantity (approx.) Dis so 1ved Hardness O2 mg-eqv/L mg/L 9.3 9.0 6.7 8.4 8.2

24.6

22.7 22.0 24.2 27.1

Suspended Particles

Alkalinity Viscosity mg/L 36.7 32.9 28.0 29.5 30.3

cSt 0.993 0.970 0.966 0.988 1.000

pH 9.0 8.9 8.5 8.7 8.9

mg/L 383 286 12 33 37

111. The Magnetic Field Effect in the Treatment of Waste Fluids and Effluents The currently most important applications of magnetic fields to environmental protection are in the area of waste-fluid and effluent treatment, involving several industrial sectors: open-hearth furnaces, gas scrubbers, steel mills, converters, mines, power- and nuclear plants, leaching operations and metal recovery are major examples. The removal of particles via filtration and coagulation in magnetic fields has, in fact, a well established technology on its own.

606 Magnetic Filtration and Coagulation

111.1.

Table 3 contains a representative example of the breadth of technological applications of magnetic purification, and Table 4 illustrates devices and apparatus employed. The two lists are by no means exhaustive, but they indicate the width of activities taking place in the last two decades. The underlying theory has been treated by Watson 1951 for filters containing ferromagnetic wires, used for the Table 3 Selected Examples of the Technological Use of the Magnetic Filtration and Coagulation Principle Utilization 1.

References

Wastewater (general)

55, 69, 70, 72

2. Municipal and industrial wastewater High gradient magnetic fields

56-59, 82

Filtration/removal of chemical particles and compounds

62, 68, 71, 72, 74

3. Wastewater from metallurgical plants

63

4. Wastewater from power plants

60, 73, 76

5.

Clay and mineral processing

61, 78

6. Wastewater from steel mills, converters and dust collectors

64, 65, 77

7. Purification of mine waters

66, 79

8. Coagulation of suspensions and emulsions

67, 80

9. Leaching

81

extraction of micron-sized paramagnetic particles in a fluid medium passing through the filter; the latter is made up e.g. of magnetic steel wool and the magnetic flux density is typically about one tesla. The analysis allows the computation of the particle-capture cross section, flow orbits and a quantitative evaluation of filter performance. Under laminar-flow conditions in the filter the efficiency of a filter bed of length L, deflned as the fraction of particles f retained by the bed

P

with respect to the number of relationship f

P

= 1

-

exp (-kgL)

inlet particles is given by

the

607 Table 4 An Illustrative List of Filtration and Coagulation

Devices

and

Apparata

Used

Description

in

Magnetic

References

Monochromatic electromagnetic radiation for solid/liquid separation

83

Continuous magnetic filte

84

Electromagnetic filter wi h double flow

63

Coagulating apparatus

85

Ion-exchange resins

86

Magnetic resonance

87

High frequency magnetic field; a.c. electromagnetic fields; rotating magnetic fields

88, 89, 94

Re-elixirization for the removal of heavy metals

90

Circular flow-path filtration equipment

91

Electromagnet assembly with ferromagnetic filter packing for hot wastewater treatment from power plants

92

Flocculator using stray magnetic fields

93

Combined electromagnetic/polyelectrolyte coagulator

64

Emulsified oil removal

80

where the bed coefficient kg is a parameter depending on various geometric parameters of the filter. A similar equation was earlier proposed (961 in a less general approach. It appears on the basis of such studies that filters employing superconductivity magnets [e.g. 971, and operating over a twelve hours per day duty period at an average rate 3 2

of about 12 m /m .min of activated sludge flow (containing e.g. Fe203 3

-

colloidal particles), could process about 95,000 m /day throughput. Semi-empirical relationships between bed geometry and filtration cycle duration (981, for the computing of the efficiency of magnetic purification of liquids with flocculated iron-containing particles [991, and cascade filter beds (631, and the performance of a wetted dust removal from waste water process [lOOl employing magnetic coagulation have also been proposed. The beneficial effect of high frequency

608 electromagnetic field? on electrolytic effluent purification at temperatur s above 65 C and d.c. electrolytic current densities of 2.5-3 A/dm 5 has been attributed [881 to a particular magnetite-forming reaction whose rate is apparently higher than the rate of the "conventional" process Fe(OHI2 + 2Fe(OHI3 Fe304 + 4 H20. Thermodynamic conditions for magnetitoe formation are mtre favourable at 6,5'C than at lower temperatures (AG = -84.4kJ at 25 C; -96.9kJ at 65 C, per one mole of Fe(0H) 1, but the exact nature of thermal and oscillating-field 2 interactions is not known. Metal recovery from industrial waste has recently been shown [loll to be an active area of magnetic field applications, the base system ranging from mine-waste to cable scrap. Fe, Co, Ni, Cr, Mn, Mo, Ti, Va, W, Ta, U, rare-earth metals and precious metals are the most obvious candidates for recovery in magnetic fields, although other substances, e.g. Zn, Pb, Mg, Ca, have also been the subject of successful recovery studies [e.g. 711. IV. The Magnetic Field Effect on Corrosion In Electrolytes Since the corrosion of metals involves specific sets of electrochemical reactions at the corroding solid/liquid interface [102, 1031, the existence of the magnetic field effect on corrosion kinetics could be predicted by a-priori theoretical considerations discussed in Section 11. Under mass transport control e.g., the beneficial effect of magnetic field imposition on the electric field would be expected to increase the anodic dissolution rate on the corroding solid material, but as shown in Table 5 , experimental observations do not demonstrate consistently a deleterious influence of magnetic field on corrosion. The inhibition of corrosion by the formation of protective oxide layers is a plausible explanation in certain cases [e.g. 1061, but seemingly opposite findings by various researchers strongly indicate that the exact nature of experimental conditions is the predominant factor in determining the overall magnetic field effect. The complexity of the topic is further aggravated by the generation of magnetic fields during electrochemical corrosion 11191; in fact corrosion-induced fields may offer a quantitative measure of corrosion, utilizing e.g. modified electrochemical impedance spectroscopy [1201. Typical quantitative relationships between the relative increase in the corrosion rate and the imposed magnetic field strength, shown in Table 6, indicate a weak influence of electrolyte concentration in the case of anodic copper dissolution into acidified chromate solutions [110], but the range of the magnetic field exponent is close to the often-quoted value of 1/2 in magnetoelectrolysis. The nature of interaction of magnetic fields with scale formation on solid surfaces (e.g. pipes) exposed to liquid flow is not understood fully at present. Retardation of scaling, reported by various experimenters [e.g. 5 4 , 5 5 , 121-1231, has not been explained by a coherent theory, although the effect of magnetic fields on various physiochemical properties of solutions [541, the dipole moment of water molecules affecting CaO cementation [551, and the dissociation constant

609

Table 5 A Representative Overview of Experimentally Observed Magnetic Field Effects on Corrosion Qualitative effect of corrosion A: accelerating; D: decelerating; I: indifferent

System (Solution)

A

Low C-steel in seawater Iron in hydrochloric acid

Reference 104

Depends on degree of purification

105

Cu; Cu-Zn (40%) in nitric acid

D

106

Zn in nitric acid

I

106

Cu in ferric chloride

essentially I

107

Fe and A1 in sodium chloride

A; D (pH-dependent)

108

Iron and alloys in sulphuric, nitric and phosphoric acid

A

109

Cu in acidified dichromate

A

110

Cu, Fe mounted on a permanent magnet in nitric acid

D

111

CT3 Steel, Ni, Al, Sn, Cu, Zn in hyperchloric acid, sulphuric acid, nitric acid and hydrochloric acid

CT3, Ni: D Others: essentially I

Metals in nitric acid-sodium nitrate, sulphuric acid-sodium

HN03/NaN03: A

nitrate; sulphuric acid-sodium sulphate

H2SO4/Na2S04: A HCl/NaCl: I

112

113

Zn in sulphuric acid-nickel sulpha te

A

114

Cu, Ni, A1 in sulphuric acid

A, D

115

Fe, Al, Zn in hydrochloric acid

Fe: D; Al, Zn: A

116

Steel in sodium chloride

D

117

Steel in sulphuric acid

A

118

610

Table 6 Summary of Regression Analysis of Magnetically Assisted Corrosion In the Anodic Dissolution of Copper into Acidified Dichromate Solution (Fig. 2 , [1101)Regression line: h ( R ) = A + b h ( B ) K2Cr207 Concentration 9

R:

B:

mol/L

a

b

rL

0.3 0.5 0.7

3.753 3.726 3.644

0.606 0.567 0.701

0.878 0.939 0.945

Combined

3.708

0.625

0.901

Relative increase in corrosion rate, % magnetic flux density, mT

of water, hence the strength of the 0-H bonds [1231, have been suggested as (at least partial) causes. According to Martynova et al. 11221, the magnetic field retains ferromagnetic particles in water which form a "fluidized" layer on a solid surface, and this layer acts as an adsorbent of salts and gases at concentrations above the saturation level. The flow through the cell washes out ultimately adsorbed material (as discrete crystals or free gas bubbles), while the ferromagnetic particles retained in the cell act as crystallization centres. The effectiveness of such a "magnetic cell" must depend on many parameters, e.g. retention time, flow hydrodynamics, magnetic field strength and configuration, solute dispersion etc. whose incomplete knowledge may well explain discrepancies in behaviour between laboratory-scale and industrial equipment. The proposal of a "magnetic memory'' of water, on the basis of its infrared spectra showing an intensity dependence on the magnetic field strength and solute concentration [1231 adds a further interesting aspect to the somewhat speculative elucidations found in the literature.

V.

The Magnetic Field Effect on Natural Bodies of Water

Natural bodies of water is a generic term, encompassing groundwater, springwater and other kinds of water of hydrological importance (i.e. "natural aqueous objects", [ 2 ] English summary). Due to their specific chemical compositions, absorbed-gas content, and flow characteristics associated with their propagation in nature, their behaviour in magnetic fields is considered in a separate category. This category may be regarded as a specific area of electromagnetic hydrophysics, where mass transport intensification, capillary and jet flow, micro- and macro-bubble phenomena, wave generation and infiltration are major areas of interest both in induced and externally imposed magnetic fields. Potential applications to irrigation, purification and anti-pollution measures can readily be envisaged.

61 1

Uagnetic Field Effects in Capillary Flow and Filtration

V.1.

Under hydrodynamic pressure gradients across capillaries, electrical current may be generated due to the formation of free electric charges along the capillary surface due to its geometric irregularities [1241, in addition to current generated due to bulk flow through the capillary. The induced magnetic field strength due to these conditions depends on the geometry of the capillary, and attention must be paid in its calculation to the extremely small length scale (1 - 100 nm), where Maxwellian "macroscopic" electrodynamics does not apply. In the Lorentz-Maxwell approach intrinsic ("true" measures of electrodynamic field variables are employed instead of continuum-based mean values. In the range of capillary diameters of 1-2 run the induced magnetic field strength and in the range of 1-10 nm the gradient of the induced magnetic field strength is inversely proportional to the square of the radial position inside the capillary, and its interaction with current flow generates an axial torque along the capillary in the viscous sublayer at the capillary surface (rigid cylinder model). If filtration through a porous medium is considered as an additive process of flow through a network of individual capillaries, the induced magnetic field is a function of the current sum/mean capillary radius ratio and its value is given by

H

(7)

= (ihb1tan-l(b/Zh)

*

In externally imposed inhomogeneous magnetic fields (VB 0 ) the rotational forces may generate sufficiently strong vortices in the solid/liquid interface sublayers to destroy the layer structure on account of large MHD-stresses appearing on the capillary walls (also on the surface of microscopic particles and entrapped gas bubbles in the flow, if impurities are present). The critical flow velocity for structure destruction is 11251

2 range [31. The magnitude of VC depends on cavitational phenomena (e.g. vortex formation), surface and liquid properties and the amount of dissolved gas. In the neighbourhood of magnet poles continuous vortex formation acts as a sustained turbulence generator. falls in the 10-15 N/m

where

t

V.2.

The Deformation of Jet Structure in Uagnetic Fields

If a jet of water is created between two vertical electrodes separated by a sufficient distance (about 25 cm, [1261) with the nozzle embedded in the lower electrode, the upper expansion width of the jet decreases and the size of the water drops increases in the d.c. electric field strength range of 20-40 V/cm. At higher field strengths, the expansion width decreases gradually and self-induced oscillations appear. The frequency of such oscillations reaches a local maximum at a

612

certain field strength (e.g. 1.2 Hz at 220 V/cm, distilled water; about 1.45 Hz at about 160 V/cm, 5% sodium chloride solution). The Jet structure is also sensitive to magnetic fields, presumably due to induced electric fields. Experimental evidence at B = 0.52 tesla, [1271 indicates a slight increase in drop size and a decrease in the number of drops. Jet stabilization and the appearance of self-induced oscillations have also been observed elsewhere 131. V.3.

The Magnetic Field Degasification of Water

Effect

on

Free

Gas

Content

and

the

Evaporation experiments [1281 conducted in vacuum at Reynolds numbers exceeding 5000 at magnetic field strengths in the 20-30 kA/m range indicate an 11-25% relative increase in free gas content, with respect to the absence of a magnetic field, as a function of the vacuum cycle number. It has been proposed [31 that magnetically supported agitation may increase the removal of nitrogen and oxygen microbubbles from irrigation waters at the expense of dissolved C02, whereby increasing the C02 content and decreasing the pH in them. A secondary effect of such interaction could be the appearance of C02 in alkaline waters (where there is normally no free C02 at pH > 8.5) affecting adversely the precipitation process Ca(HCO3I2

t- CaC03 + 4

C02

+

H20

(9)

Magnetically manipulated dehydration of irrigation waters may reduce the number of ferromagnetic or paramagnetic microparticles carrying iron in organic and inorganic complexes existing at high pH; as the pH is decreased, a certain proportion of the complexes would break up and insoluble hydroxides would form. These oxides absorb ions on their surface, hereby promoting the presence of calcium and iron compounds in irrigated soils. V.4.

Precipitation and Crystallization In Magnetic Fields

Ultramicroscopic investigations of nuclei formation in imposed magnetic fields of the 0.01-0.2 T flux density range indicate a retarding effect on solid phase formation in groundwaters, distilled water, and calcite solutions. As shown in Table 6, the suppression of active CaC03 nuclei becomes increasingly stronger at elevated temperatures; at 85'C and B = 75 mT the number of active nuclei is about one-third of the number observed in the absence of the magnetic field. The relative increase in nucleus density passes through a local maximum at about 68 mT under isothermal conditions [Table 71, and falls rapidly to an essentially constant low level as the magnetic field strength is increased. Since as the temperature is increased, and the degree of supersaturation of calcium carbonate in water is increased, the average radius of the precipitate also increases, the magnetic field

613

effect appears to depress supersaturation, and to increase the surface activity of the already existing nuclei, possibly by breaking up the hydration layer on solid particles. Similar results obtained in ultrasound and irradiation experiments 11311, and the study of seawater [1321 also indicate interaction with surface activity. The existence of a locally maximum magnetic field effect has not yet been explained theoretically. Table 7 The Effect of Magnetic Fields on the Precipitation of CaC03 from a 100 mg/L

C02/Tapwater

Solution: Temperature Dependence "1291;

the

tabulated approximate values are based on Fig. 2.24 [31).

N x 106 (particles/cm3 solution) at T, (Temperature, "C)

a linear solution velocity of 1 m/s

B = 0 (mT) 40 66 73 78 85

B = 75 (mT) 0.2

0.6 2.0 4.0 7.1 9.0

1.0 1.7 1.7 2.8

6 B = 0: h (N x 10 1 = -13.78 + 3.56 tn(T); r2 = 0.935 B = 75 mT: h (N x 106) = -14.25 + 3.42 .!n(T); r2 = 0.990 Table 8 The Effect of Magnetic Fields on the Precipitation of CaC03 From a 100 mg/L

C02/Tapwater

Solution: Isothermal Flux Density Dependence

([1301; the tabulated approximate values are based on Fig. 2.24 [3]).

B, (magnetic flux density, mT) 0 24 50 68

104 127 150 174 200

Relative increase in nuclei density at a linear solution velocity of 1 m/s 1 1.21 1.77 2.29 1.21 1.08 1.08 1.08 1.06

614 V.5.

Thin Layer Magnetodynamics

The dynamic behaviour of thin water layers has been the subject of various studies by laboratory-scale facsimile experiments and MHD theory. A basic factor in thin-layer dynamics is the deceleration of the horizontal bulk motion over a thin liquid layer of thickness h by the laminar layer; a measure of this deceleration is the friction 2

coefficient A = 2 u/h . In a sufficiently wide thin layer cell, the twodimensional vorticity equation (x: coordinate along the gravity force, y: coordinate along the flow direction) 2 2 2 awz/at + u awlax + v aw/ay = v(a wz/ax2 + a 0Zlay I - AW + vz x % (10) indicates, that if F

-M

is a rotational force field, vortex formation is

expected, provided that the magnitude of Vz%

is stronger than viscous

losses in the liquid, and friction forces at the liquid/solid interface. In experiments 1133-1351 using a magnet assembly with alternating north/south poles in cascade, a sinusoidal MHD force in the x-direction was generated, producing a sinusoidal MHD force field in the (x,y) plane, thus resulting in various vortex structures sustained by the induced oscillations. The vortex shape and size depends on the electric current flow through thin layer, the mean flow velocity, the layer thickness and the shape of the imposed magnetic field, as indicated by photographic techniques employed in visualization experiments (e.g. Figs. 4.3, 4.4, 4.7, 4.11, 131). More complex vortex structures which may exist in water reservoirs have also been studied via visualization 11361. Theoretical analysis of Alfven wave formation 1101 in aqueous electrolytes [1371 has been complemented by capillary-gravitational wave formation experiments in channels [e.g. 137-1391.

V.6.

Miscellaneous Investigations

Studies of the role of magnetic fields on the stability conditions of mineral-containing subsurface waters [1401, the formation of certain biological entities in subsurface streams 11411, groundwater quality [142], plant life and crop yield [143, 1441 are indicators of a growing recognition of the importance of magnetic field effects in the environment. The suppression of certain ions (e.g. chloride, sulphate, calcium, magnesium and sodium) in soils (1431 by magnetically treated water, yield increases due to magnetically-treated irrigation waters [144; 8% for tomato and cucumber varieties], more recent work on the dependence of irrigation-water and drinking-water quality [1451, and irrigation with brackish water through magnetizing nozzles [1461 suggest the increasing relevance of magnetic fields in agricultural research. The importance of magnetic field measurements in detecting underground water locations and the movement of large bodies of water occurring in nature has also been documented [147-150].

615

VI.

Critique and Concluding Remarks

Magnetic fields are capable of generating, even at modest flux densities, i.e. less than one tesla, important interactions with flow and structural phenomena. Since the magnetic force is often considerably less than force magnitudes associated with forced and natural convection in liquid electrolytes, classical MHD theory cannot be employed exclusively to explain all experimentally observed magnetic field effects. On the other hand, there exists at present no other cohesive theory for the treatment of electrichagnetic phenomena in liquids. The current lack of a comprehensive framework of analysis remains a serious challenge to all theoreticians, be they mathematicians, physicists or chemists. For the environmental electrochemist and electrochemical engineer, the major question is whether "anthropogenic" magnetic fields (i.e. magnetic fields designed by technical personnel) can be employed to improve the course of natural phenomena, and to protect the environment. Progress in certain areas, e.g. waste water/effluent treatment and corrosion kinetics justifies a cautiously affirmative, but not an absolutely certain answer. Further research to be conducted in this area will have to bring together several disciplines and enjoin cooperation of the intellectual rigorist with the practical designer. The childhood of environmental magnetoelectrochemistry promises an exciting state of adolescence.

List of Symbols B -

b C

D

magnetic flux density; BM maximum magnitude of the induced flux density in a gradient field width of a capillary layer with current i flowing through it concentration; c nominal ionic; c electrolyte i S diffusivity: D. ionic; D electrolyte in the absence of a magnetic 1

field; D electrolyte, in a magnetic field

m

-F

electric field strength; E its magnitude electronic charge (0.1602aC) force vector; F electric; F frictional; F magnetic; F Lorentz; -E -f -H -L

f

fraction of (magnetizable) particles

-H

magnetic field strength; H its magnitude height measured in a liquid from the surface of a capillary layer with current i flowing through it electric current electric current density vector; j its magnitude

E

e

MHD

P

h i

-I k kg L

Boltzmann constant (1.3805 x filter bed coefficient

J/K)

characteristic length of a physical unit (e.g. filter bed)

616

e m

i

NA U

i

V

-T

vi m' wi

z

length of a capillary ionic mass Avogadro number (6.023 x

ion/g-atom)

ionic mobility mean thermal ionic velocity vector; VT its magnitude ionic velocity vector velocity of the centre of the total ionic mass ionic size valency

Greek Symbols

A

friction coefficient Bohr magneton (9.273 x

B'

C/m2)

magnetic permeability of the electrolyte (usually taken as the

e' 'ss

value for vacuum, 1.2566 pV.s/A.m) partial derivative of the electrolyte chemical potential with respect to the electrolyte concentration c

S

xi P 7

ionic magnetic susceptibility solution density surface tension

REFERENCES 1. Vonsovskii SV, Vestnik Akad Nauk SSSR 1966; 4: 77-92. 2. Bondarenko NF, Gak EZ, Elektromagnitnie Yevlenia V Prirodnikh Vodach, St. Petersburg: Gidrometeoizdat, 1984, Chapter 1, 13-55. 3. Bondarenko NF, Fizika Dvizhenia Podzemnikh Vod, St. Petersburg: Gidrometeoizdat, 1973. 4. Gak EZ, Inzhen Fiz Zhurn 1982; 43: 140-153. 5. Fahidy TZ, J Appl Electrochem 1983; 13: 553-563. 6. Lielmezs J, Musbally GM, Electrochim Acta 1972; 17: 1609-1613. 7. Dumargue P, Humeau P, Penot F, Electrochim Acta 1973; 18: 447-458. 8. Hubbard JB, Wolynes PG, J Chem Phys 1981; 75: 3051-3054. 9. Hughes WF, Young FJ, The Electrodynamics of Fluids, Wiley, New York, 1966. 10. Shercliff JA, A Textbook of Magnetohydrodynamics, Pergamon Press, Oxford, 1965. 11. Kendall PC, Plumpton C, Magnetohydrodynamics with Hydrodynamics Vol. I, Pergamon Press, Oxford, 1964. 12. Ferraro VCA, Plumpton C, An Introduction to Magneto-Fluid Mechanics, 2nd Edn., Clarendon Press, Oxford, 1966.

617 13. Bopp GR. Chem Engrg Progr 1967; 63: 74-79. 14. Womack GJ, MHD Power Generation: Engineering Aspects, Chapman and Hall, London, 1969, Chapter 3, 44-63; Chapter 4, 64-85. 15. Blum EY, Magnitn Gidrodin 1970; 6: 69-76. 16. Fahidy TZ, Electrochim Acta 1973; 18: 607-614. 17. Fahidy TZ, The Chem Engrg J 1974; 7: 21-27. 18. Mohanta S, Fahidy TZ, J Appl Electrochem 1976; 6: 211-220. 19. Blum EY, Magnitn Gidrodin 1978; 3: 11-17. 20. Blum EY, 0201s RY, Magnitn Gidrodin 1978; 4: 42-48. 21. Kelly EJ, J Electrochem SOC 1977; 124: 987-994. 22. Levich VG, Physicochemical Hydrodynamics, Prentice Hall, Englewood Cliff, 1962; Chapter 1 1 , 39-184. 23. Newman JS, Electrochemical Systems, Prentice Hall, Englewood Cliffs, Chapter 1 1 (1st Edn. 1973) 217-238; Chapter 11 (2nd Edn. 1991) 241-264. 24. Coeuret F, Storck A, Elements de Genie Electrochimique, Lavoisier, Paris, 1984, Chapter 2, 45-90. 25. Fahidy TZ, Principles of Electrochemical Reactor Analysis, Amsterdam: Elsevier, 1985; Chapter 7, 165-189. 26. RouSar I, Micka K, Kimla A , , Electrochemical Engineering I Parts A-C, Amsterdam: Elsevier 1986; Section 1.3, 27-29. Blum EY, Mikhailov YA, 0201s RY, Teplo i Massoobmen V Magnitnom 27 Polye Riga: Zinatne, 1980. 28. Fahidy TZ, Rutherford TS, J Appl Electrochem 1980; 10: 481-488. 29. Aogaki R, Fueki K, Mukaibo T, Denki Kagaku 1975; 43: 504-508. 30. Aogaki R, Fueki K, Mukaibo T, Denki Kagaku 1975; 43: 509-514. 31. Aogaki R, Fueki K, Mukaibo T, Denki Kagaku 1976; 44: 89-94. 32. Gak EZ, Krylov BC. Elektrokhimya 1986; 22: 829-834. 33. Aksel’rud GA, Molchanov AD, Gavriskevich LN, Inz Fiz Zhurn 1976; 31: 1013-1016. 34. Blum EY, Magnitn Gidrodin 1978; 3: 11-17. 35. Fahidy TZ, Electrochim Acta, 1990; 35: 929-932. 36. O’Brien RN, Santhanam KSV, J Electrochem SOC 1982; 129: 1266-1268. 37. Sviridova LN, Korshunov VN, Elektrokhimya 1978; 14: 99-100. 38. Gak EZ, Rokhinson EKH, Elektronn Obrab Mater 1973; 4: 73. 39. Gelles E, Pitzer K , J Am Chem SOC 1955; 27: 1974-1978. 40. Justi E, Wieth J, Zeitsch Naturforsch Ser A 1953; 8: 538-546. 41. Schwab J, Kaiser A, Zeitsch Phys Chem 1955; 4: 220-231. 42. Gu ZH, Olivier A , Fahidy TZ, Electrochim Acta 1990; 35: 933-943. 43. Olivier A, priv corn. 44. Tronel-Peyroz E, Olivier A, Comptes Rendus Acad Sc (Paris) C 1974; 278: 997-1000. 45 Tronel-Peyroz E, Olivier A, Fahidy TZ, Laforgue-Kantzer D, Electrochim Acta, 1980; 25: 441-446. Olivier A, Tronel-Peyroz E, Comptes Rendus Acad Sc (Paris) C 46 1974; 278: 1027-1029. Tronel-Peyroz E, Doctorat d’Etat thesis, Univ Rheims, 1978. 47 48 Olivier A, Doctorat D’Etat thesis, Univ. Rheims, 1979. Guerin-Oueler D, Nicollin C, Olivier A, Comptes Rendus Acad Sc 49 (Paris) C 1970; 270: 1500-1503. 50 Quraishi MS, Fahidy TZ, J Electrochem SOC. 1980; 127: 666-669. 51 Quraishi MS, Fahidy TZ, Chem Engrg Sci 1982; 37: 775-780.

618 52. Bondarenko NF, Gak EZ, Izv Nauk SSSR, Fiz Atm i Okeana 1984; 20: 640-645. 53. Lau A, Fahidy TZ, J Electrochem SOC 1989; 136: 1401-1408. 54. Kutzenko AN, Neftpromysl Delo, 1974; 8: 43-45. 55. Eremkin A, Sorokin A, Dobrov AG, Vodosnabz Sanit Tekhn 1978; 12: 12-13. 56. Harland JR, Oberteuffer JA, Goldstein DJ, Chem Engr Progr 1976; 72(10) 79. 57. Mitchell RG, Bitton G, Oberteuffer JA, Separ Purif Meth 1975; 2: 207. 58. Oder RR, Horst BI, Filtr and Separ, 1976; 13: 363-369. 59. Mitchell R, Allen D, Proc Intern Conf, Franklin Pierce College, Rindge NH (IEEE N.Y., Ed. YA Liu), 1978; 142. 60 Heitmann HG, Proc. Intern Conf, Franklin Pierce College, Rindge NH (IEEE N.Y., Ed. YA Liu), 1978: 115-120. 61. Clark NO, Proc Intern Conf, Franklin Pierce College, Rindge NH (IEEE N.Y., Ed. YA Liu), 1978; 62-63. 62. Peperstraete H, Janssens J, Trib Eau, 1990; 42: 32-42. 63. Topkin YV, Khirn Tekhnol Vody, 1984; 6: 555-559. 64. Czajkowski M, Gega J, Jaskowiec Z, Kowalski W, Zesz Nauk Akad Gorn-Hutn Stan Stasz, Sozol Sozotech 1977; 10: 125-136. 65. Kishi K, Hamabe M, Tsukada Y, Ito K, JP 50155403 15.12.1975. 66. Simakov AE, Rodionov SI, Zadvirnov YN, Nauchn Tr Permsk Nauchno-Issled Ugoln Inst 1973; 16: 61-67. 67 Gega J, Worbel J, Zesz Nauk Akad Gorn-Hutn Krakow Zesz Spec, 1972; 32: 197-213. 68. Tanaka M, Hirakao M, Migarni K, JP 62140695 A2 24.06.1987. 69. Muto E, JP 60039440 B4 05.09.1985. 70. Hayata F, Yugawa T, JP 60166017A2 29.08.1985. 71.. Jasionek TM, PL 115016 B2 31.07.1982. 72. Kovalev VV, Band MI, Elektron Obrab Mater, 1982, 1: 61-64. 73. Fedotkin IM, Sandulyak AV, Lazarenko LN, Garaschenko VI, Kuzrninskii AA, Khim Tekhnol (Kiev), 1980; 5: 62-64. 74. Gubenkov MG, Gravrilyuk LM, Shishkov NI, Denzanov GA, Slonchak VP, Malysheva EA, Binshtein AB, Martynyuk LP, SU 701955 05.12.1979. 75. Kurihava Y, Oda M, Okugawa K, Sunaoka Y, Kleper MH, Markind J,

Goldsmith RL, Ryan JM, Proc Intern Wat Conf Eng SOC. West Pa, 1978; 39: 135-146. 76. Glebov VP, Belau FI, Heinikov GI, Tr Tsentr Nauchno Issled Proektno-Konstr Kotloturb Inst, 1977; 141: 82-89. 77. Kishi K, Harnbe M, Tsukade Y, Ito K, JP 50155403 15.12.1975. 78. Chernyak LP, Nichiporenko SP, Zaionts RM, Kolloidn Zhurn 1975; 37: 110-114. 79. Simakov AE, Rodionov SI, Zadvornov YN, Nauchn Tr Perrnsk Nauchno-Issled Ugoln Inst, 1973; 16: 61-67. 80. Soda F, Yugawa T, JP 54065179 25.05.1979. 81. Bondarenko NF, Nazarrnarnedov 0, Aramedov Kh, Gak EZ, Gidrotekhn Melior, 1981; 8: 73-77. 82. Wallin M, Vatten 1976; 32: 211-219. 83. Hora H, Potter H, Potter Ar, DE 3516970 A1 13.11.1986. 84. Henel V, Mucha P , Horacek M, Kolar 0, Cibulka J , CS 228174B 15.05.1986.

619

85. Ramik P, Kostalova H, CS 210860 B 15.04.1983. 86. Sushkovskii VD, Khlopotov MN, Shumilova GV, Kuznetsova IE, Khim Tekhnol Vody, 1984; 6: 166-169. 87. Jimenez MP, DE 3225806 A 1 12.01.1984. 88. Kovalev W, Band MI, Khim Tekhnol Vody, 1983; 5: 248-251. 89. Hama Y, Kamiyama T, JP 53049856 06.05.1978. 90. Anon, Ind Miljoe, 1975; 6: 33, 35. 91. Eggerids TL, US 4879045A 07.11.1989. 92. Fedotkin IM, Sandulyak AV, Lazarenko LN, Garaschenko VI, Kuzminskii AA, Khim Tekhnol 1980; 5: 62-64. 93. Wrobel J, Gega J , Konietzui M, Stahl Eisen, 1978; 98: 476-481. 94. Kuramzin AV, Tsantker KL, Logvinenko VD, Shelyakov OP, Nikitenko MI, Mauzhelii AP. Sizonova NT, GB 1363294 14.08.1974. 95. Watson JHP, J Appl Phys 1973; 44: 4209-4213. 96. Bean CP, Bull Am Phys SOC 1971; 16: 350. 97. Stekly ZJJ, Proc Intern Conf, Franklin Pierce College, Rindge NH (IEEE N.Y., Ed. YA Liu), 1978: 26. 98. Topkin YuV, Topkina NM, Khim Tekhnol Vody, 1986; 8: 22-25. 99. Sandulyak AV, Kochmarskii VZ, Bartosevich RD, Khim Techno1 Vody, 1981; 3: 74-77. 100. Bezvesil’nyi EV, Grabareva SD, Sanit Tekh, 1975; 15: 140-142. 101. Brooks CS, Metal Recovery From Industrial Waste, Lewis Publishers, Chelsea Michigan, 1991. 102. Uhlig HH, Corrosion and Corrosion Control, Wiley and Sons, New York; 1963. 103. Evans UR, An Introduction to Metallic Corrosion, Edward Arnold, London, 1963. 104. Peev T, Mandzhukova B, Mandzhukov I, Rusaun V, Werkst Korros, 1991; 42: 90-93. Shiryaev VI, Vysokochist 105. Kamenatskaya DS, Piletskaya I B , Veschestva, 1989; 5: 87-90. 106. Chiba A, Ogawa T, Boshoku Gijutsu, 1988; 37: 595-600. 107. Gabashy MA, Corros Prev Contr, 1988; 35: 73-75. 108. Kravchinskii AP, Levochkina IA, Korroz Zashch Korroz Met Splavov, 1985; 31-33. 109. Kravchinskii AP, Isaev NI, Baluev VN, Zakharov AI, Shumilov VN, Revyakin AV, Izv Akad Nauk SSSR Met, 1984; 6: 31-35. 110. Ghabashy ME, Sedahmed GH, Mansour IAS, Brit Corr Journ, 1982; 17: 36-37. 111. Sagawa M, Trans Jpn Inst Met, 1982; 23: 38-40. 112. Nguyen Q, Pham VB, Tap San Hoa Hoc, 1975; 13: 1-5. 113. Nguyen Q, Pham VB, Tap Chi Hoa Hoc, 1977; 15: 1-7. 114. Gavasheli DV, Pruidze VP. Tr Molodykh Nauchn Sotr Aspir - Akad Nauk Gruz SSR, Inst Khim Elektrokhim, Akad Nauk Gruz SSR, Tiflis Gruzia 1974; 10-14. 115. Klyuchnikov NG, Verizhskaya EV, Zashch Metal, 1972; 8: 700. 116. Grigor’ev VP, Ekilik W , Ekilik GN, Gantmakher NM, Issled 061 Korroz Zashch Metal, Elista SSSR, 1971; 106-113. 117 Tebenikhin EF, Pronina ZF, Rybalchenko VS, Teploenergetika, 1972; 10: 69-71. 118. Verizhskaya EV, Klyuchnikov NG, Tr Krasnodar Cros Pedagog Inst, 1969; 130: 50-53.

119. Bellingham JG, MacVicar MLA, Nisenoff M, Searson PC, J Electrochem SOC,1986; 133: 1753-1754. 120 Murphy JC, Hartong J, Cohn RF. Moran PJ, Bundy K, Scully JR, J Electrochem SOC,1988; 135: 310-313. 121. Das CR, Misra HP, J Inst Eng (India) Part CH. 1982; 63: 30-31. 01, Kopylov AS, Tebenikhin EF, Ochkov VF, 122. Martynova Teploenergetika, 1979; 6: 67-69. 123. Karavayeva AP, Marshakov IK, Zhidkonozhkina AA. Teor Prakt Sorbts Proc, 1977; 11: 78-83. 124. Cooper, UF, Brit Journ Appl Phys Suppl 2, 1953; 11-15. 125. Bondarenko NF, Gak EZ, Gak MZ. Rokhinson EE, Inz Fiz Zhurn, 1978; 35: 842-850. 126. Frenkel YI, Vager GP, Izv Akad Nauk SSSR Ser Geogr Geofiz, 1948; 12(3); 3-7. 127. Minenko VI, Kolloidn Zhurn, 1976; 38: 821-822. 128. Bondarenko NF, Gak EZ, Culkov AN, Zaslavskii YA, Inz Fiz Zhurn, 1980; 39: 64-69. 129. Mikhelson ML, Kolloidn Zhurn, 1977; 39: 302-306. 130. Mikhelson ML, Kolloidn Zhurn, 1977; 39: 577-578. 131. Terebikhin EF, Energiya, 1977; 182. 132. Slesarenko NN, Energiya, 1973; 247. 133. Gak ME, Izv Akad Nauk SSSR Fiz Atm i Okeana, 1981; 17: 201-205. 134. Pleshanova LA, Izv Akad Nauk SSSR Fiz Atm i Okeana, 1982, 18: 339-348. 135. Bondarenko NF, Gak ME, Dolzhanskii FV, Izv Akad Nauk SSSR Fiz Atm i Okeana, 1979; 15: 1017-1026. 136.. Znamenskii V, Trudi GGI, 1970; 183: 180-201. 137. Fahidy TZ, Electrochim Acta, 1976; 21: 21-24. 138. Gak ME, Gak EZ, Zhurn Tekhn Fiz, 1972; 42: 1992-1994. 139. Gak ME, Bondarenko NF, Zhurn Tekhn Fiz, 1976; 46: 634-637. 140. Nikoladze GI, Fed'kushov YI, Sborn Trudy Mosk Inzh-Stroit Inst, 1980; 174: 155-165. 141. Rodwanowski LJ, Tech Poszukiwan Geol, 1980; 19: 40-44. 142. Zhivotnev VS, Sukasyan BD, Shoimov Sh, Sborn Trudy Mosk Inzh-Stroit Inst, 1976; 148: 159-164. 143. Volkonskii NA, Chalenko VI, Neroznak BK, Anikushin VF, Vestn S-Kh Nauki, 1978; 7: 93-96. 144. Penchev P, Khristov V, Nikolov I, Prangov L, Asenova N, Gesheva A, Gradinar Lozar Nauka, 1976; 13: 82-87. 145. Lin IJ, Yotwat J. Journ Magn Magn Mater, 1990; 83: 525-526. 146. Shumakov BB, Bagnenko BK, Dudakov NK, Rarova MA, Dokl Vses Akad S-Kh Nauk Imeni VI Lenina, 1989; 4: 13-14. 147. Khmelevskii VK, Chekanov VL, Moscow Univ Geol Bull, 1989; 44: 84-88. 148. Schlinger CM, Bull Assoc Engrg Geol, 1990; 27: 37-50. 149. Abramov OK, Karimov FK, Negmatullayev SK, Prokhorov AA, Skvorodkin YP, Izv Akad Nauk Tadzik SSSR (Otdel Fiz Math Khimi Geol Nauk), 1983; 87: 74-82. 150. Haupt RW, Martin JR, Greenfeld RJ, Trans Amer Geophys Union, 1981; 62: 1053.

62 1

Soil Decontamination Using Electrokinetic Processing R. J. Ga1e.a Heyi Li,a and Y . B. Acarb

Whemistry Department and bCivil Engineering Department, Louisiana State University, Baton Rouge, LA 70803, USA. INTRODUCTION

In industrialized countries worldwide, the irresponsible or poorly engineered disposal of hazardous chemicals or chemicals that react to produce toxic byproducts in the environment has necessitated extensive (and expensive) clean-up efforts. Nowhere is this more true t h a n in the United States, where the problems are being well documented. Hazardous waste, as defined by the U.S. Resource Conservation and Recovery Act (RCRA), is a solid waste material which may cause, or significantly contribute to, a n increase in mortality, or serious irreversible, or incapacitating reversible illness: alternatively, it may pose a substantial present or potential hazard to human health, or to the environment when improperly treated, stored, transported, or disposed of, etc. [1,2]. The so-called Superfund, created by the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) of the U S Congress, provides a mandate for the Environmental Protection Agency (EPA) to take actions in responses to hazardous releases of pollutants and to require responsible parties to contribute to the remediation. Massive government and private actions currently are underway to implement full site restorations, to develop new remediation technologies, and to avoid as much as possible problems arising from future disposal methods. Electrokinetic soil processing is a n emerging remediation technique with the capability to decontaminate soils or slurries polluted with heavy metals, radionuclides, or certain organic compounds. The EPA general classification is that it is a physical remediation treatment for phase separation, although the process also may include pre-, concurrent-, or post-chemical treatments (3.41. This chapter contains a brief overview of the theoretical basis of electrokinetic soil processing, the results of some laboratory tests and engineering models, and a summary of experiences of some actual site applications. In principle, t h e contaminant may be a n inorganic/organic/ organometallic species a n d charged (ionic)/uncharged (polar/ nonpolar). The subject is a multidisciplinary one, encompassing basic electrochemistry, soil/colloid chemistry, a n d geotechnical/ environmental engineering. Electrokinetic processing derives its name from one of the four major electrokinetic phenomena. These arise from the coupling between electrical and hydraulic flows and gradients in suspensions and porous (soil) media, which can be responsible for electroosmosis, streaming potential, electrophoresis, and migration or sedimentation

622

Water&

C-

ELECTROOSMOSIS

Particles

ELECTROPHORESIS

T AH

I

STREAMING POTENTIAL

Figure 1.

MIGRATION POTENTIAL

Four major electrokinetic phenomena in soils (adapted from Mitchell, 161)

potentials [5,6].Electroosmosis and electrophoresis are the movement of pore water and charged particles, respectively, due to the application of an electrical field. Streaming potential and sedimentation potential are the generation of an electrical field due to the movement of an electrolyte under hydraulic potential and the motion of charged particles in a gravitational field, respectively (Figure 1). Of the electrokinetic phenomena, electroosmosis has been of primary interest in geotechnical engineering because it is used in practice to dewater and to stabilize saturated fine-grained deposits. The proposed uses of electrokinetics in waste treatment and disposal may include: (1) dewatering of sludge slimes or dredged spoils, (2) electroosmotic flow barriers [ 7 ] , (3) leak detection systems for disposal facilities, (4) injection of grouts to create barriers, (5) to provide nutrients for biodegrading microcosm, ( 6 ) electrochemical in situ generation of reactants such as hydrogen peroxide for clean-up. and (7) decontamination of soils and ground waters 181. Electrokinetic soil processing involves not only electroosmosis and some electrophoresis but also the dispersal/entrapment of electrolysis products and their adsorption/desorption interactions, a s outlined below.

623

THEORETICAL

Background Electrokinetic processing of soils, by application of a direct current through a wet soil mass, results in the development of electrical, hydraulic, and chemical gradients. The formation of a n acidic front at the anode from water electrolysis and the induced electroosrnotic flux of the pore fluid enable the removal of those contaminants that can be solubilized, desorbed from the soil, or simply carried by the pore fluid. The fundamental basis of each of these two main processes is described below. When an electric potential is applied across a wet soil mass by immersion or placement of two or more electrodes, cations in the pore fluid are attracted to the cathode and anions to the anode (Figure 2). For a uniform, initial concentration of ions, the application of an electric field will only create instantaneously a uniform electric field. The current flow requires faradaic reduction and oxidation reactions at the cathode and anode, respectively, as well as the transport of ions in the solution phase (it is assumed normally that any movements of soil particles and/or colloids do not contribute significantly to the ionic current). Since there is generally a n excess of positively charged cations in the system, to neutralize the net negative charge on the soil particle surfaces, these double layer cations migrate toward the cathode carrying not only their waters of hydration b u t also producing, by viscous drag, an electroosmotic flux of the pore fluid. The net transport of water molecules by the waters of hydration of anions and cations

+ + + + + + ANODE

CATHODE NEGATIVE CLAY OR GLASS SURFACE

Figure 2.

Schematic of electrolysis, adsorption/desorption, and electroosmotic flow.

624

(5-10 moles/Faraday maximum) is negligible compared to the induced electroosmotic flux (100-4,000moles/Faraday). Counterions moving in the opposite direction tend to oppose this flow but the force exerted by the cations in the inter-particle regions is toward the cathode when the excess surface charge is negative. However, an excess of anions in the double layer for a positively charged clay surface would result in a reverse net flow from cathode to anode. The migrating ions reach a limiting velocity instantaneously and the electrical force is equal and opposite to the frictional drag (Stokes-Einstein Law). Notice that only a fraction of the total electrical current contributes to the electroosmotic force and, as the ionic strength of the pore fluid is increased, the fraction of the current carried by the double layer excess ions is diminished. With elapsed time, therefore, and if the electrode processes result in the formation of charged species, the movement of these electrode products and the original ions by transport and the advective (convective) flow of pore water combine to determine the product distribution and efficiency of remediation. Importantly, the downstream flux of hydrogen ions from the electrolysis of pore water at the anode desorbs, by displacement, adsorbates from sites on the clay surface 191. Basic metal hydroxides, carbonates, etc.. present a s insoluble salts or complexes, may be dissolved. Hence, the overall distribution of chemical species is complicated by ionic and convective t r a n s p o r t , t h e electrolysis reactions, contaminant species adsorption/desorption energetics and kinetics, and localized acid/base reactions. The electrode reactions depend on the availability of reactants, the relative energetics and kinetics, and perhaps passivation or product accumulation effects. The Electroosmotic Effect (Large Pore Theory)

The phenomenon of electroosmotic flow was first observed by the Russian scientist Reuss in 1808. I t has been applied practically in soil improvement and stabilization applications for over 50 years, e.g. [ 10121. One of the earliest explanations of electroosmosis is a model introduced by Helmholtz (1879) and later improved by Smoluchowski (1914). For modelling simplicity and experimental design, it is convenient to consider a single glass capillary. rather than a porous particle bed or membrane. The capillary (or clay particle)/liquid interface may be considered as a simple capacitor, with an excess of charges of one sign on the insulator substrate and an equivalent layer of oppositely charged ions in the liquid phase, as illustrated in Figure 2. We may ask what is the reason that insulator materials, such as glass or clay, have a surface charge? The majority of clay surfaces, a t neutral pH. exhibit a negative surface charge firstly because of isomorphous substitution in the underlying structure. Isomorphous substitution is the replacement of the primary atoms comprising the lattice structure by others of the same type but different valence and size resulting in deficiencies in charge. Clay particles, or glass, also will have surface ionizable functional groups at the broken edges of the lattice, e.g.,

625

Heterogeneous surface sites exist also and they may have cationic and anionic exchange capabilities. However, clay minerals have a net charge deficiency, the cation exchange values being larger and typically 5- 150 milliequivalents/ 100 gram [6]. Lower values are found for 1:1 kaolinite minerals and higher values for 2: 1 montmorillonite minerals. The classical Helmholtz-Smoluchowski theory applies for the condition that the double layer thickness is small compared to the capillary diameter [5,6]. The thickness of the diffuse layer sometimes is approximated to the reciprocal Debye length, 1 / ~which . is of the order 10 A and 100 A for 10-1 M and 10-3 M 1:l electrolyte in water a t 25°C. respectively [13]. A discussion of the fractional potential drop with diffuse layer thickness at metal electrodes has been presented in ref. [14]; 99.99% of the potential drop occurs in the distance 9 . 2 / ~ . The physical significance is that this estimate of diffuse layer thickness is no greater than 1 micron for a 10-5 M 1:l electrolyte and smaller a t higher interpore ionic concentrations. The double layer a t the surfaces of clays and soils will be complicated further by the surface heterogeneity, A epitaxial crystal faces, adsorbed chemical species, etc. comprehensive review of the equilibrium double layer and associated electrokinetic phenomena was written by Durkin and Derjaguin [ 151 and rigorous treatments of the classical theory for electrokinetic flow in narrow cylindrical capillaries have been presented by Rice and Whitehead [16]. Newman [17], and Koh and Anderson 1181. Original references are provided in [15-18], to which the reader is referred, and only a brief summary of the classical theories and earlier literature are included here. In large pore theory, it is assumed that the double layers at clay surfaces do not overlap. Therefore, the theory is applicable for pores of the order of one micron or greater in soils. When the double layer has an excess of cations, for example, and an electrical field tangential to the surface perpendicular is applied, a net, directional force is exerted causing the cations to move. A convective flow of the liquid occurs and a steady state velocity is reached due to the momentum transfer from the cations to the liquid molecules. At steady state, the electrical force on a cation is equal and opposite to the frictional slip forces within the liquid. The electrical force is assumed to be transferred completely to mechanical energy and no allowance is made for heating or eddy current losses. Cations, in general, are more highly solvated than anions. The effective cross-sectional area of a solvated ion will be larger than that of an unsolvated ion, although no studies of electroosmotic differences due to the ion species nature are available. It is assumed, also, t h a t the background ionic strength consists of equal concentrations of negative and positive ions whose motions result in

626

forces which are equal and thus cancel. The total electrical force in the axial direction z in a capillary, F,. is given by the product of the electrical field gradient, E,, and the local charge density due to excess ionic charge, p(r). The electrical field gradient is assumed to be uniform across the capillary in the radial direction, i.e., EZz f(r). The solution of the Poisson-Boltzmann equations for the local charge density p(r) of a double layer at the internal wall of a cylindrical capillary is given by [16],

in which K is the reciprocal Debye length, Io( 1 functions are modified Bessel functions of the first kind, a is the capillary radius with point distance from the axis, r, and yfo is the capillary wall potential. This

choice of the potential plane requires careful consideration [5,15]. In double layer theory, it could represent 42, which is the plane potential a t the compact layer in the Gouy-Chapman-Stern model. However, it can be assumed that a quiescent layer of liquid exists at the surface and a hydrodynamic shear plane exists within the double layer. The boundary now is that where the liquid may move and the shear plane has a potential known as the zeta or electrokinetic potential, .C, In this model, the excess charge responsible for fluid motion resides between this boundary and the outer diffuse layer. In the absence of any external pressure, for the capillary case, the equation of motion is found by equating the externally applied electrical force, F,, to the internally opposing viscous drag forces, when

in which v, represents the fluid velocity in the axial direction z, Fz is the electrical force, and q is the coefficient of viscosity. Traditionally, this equation has been solved with the boundary conditions which assume that the flow velocity will be maximum in the center of the capillary and zero at the wall,

627

There are important assumptions physically. The boundary condition at the surface of the capillary is referred to as the no slip condition. This condition generally has followed from experimental observations of Newtonian fluids, when the fluid which contacts a solid body is assumed to adhere to the boundary and have zero velocity, e.g. 1191. I t is a n assumption which simplifies the problem and allows a mathematical treatment. The magnitude and profile of any surface discontinuity in flow is often not known. The solution for the velocity profile then reduces to,

This profile has a flat, maximum which extends across the central region of the capillary and rapidly approaches to zero at the wall (Figure 3). By way of contrast, in Figure 3, we show the profile obtained by numerical integration of equation (4) when full slippage is assumed. This constitutes the case when there is no friction a t the interface between the capillary wall and the pore fluid. The physical significances of these integrals are as follows: (i) in the no slip model, the integral is based on the assumption that the electrical force is exerted between the surface potential plane of shear and the centre of the capillary, with a minimum velocity nearest to the wall. The mechanism by which the momentum is transferred from the double layer liquid, on which the electrical force is exerted, to the bulk of the liquid is not addressed. A Hagen-Poisseuille type of flow with a flattened profile results; (ii) for the full slip condition, only the liquid within the double layer ions is mobile. This produces a much smaller velocity profile that will be strongly dependent on the ionic strength for a constant surface charge. Obviously, the assumptions regarding slippage and momentum transfer are critical to the overall shape of the velocity profile. Precise measurements of surface charges and double layers in capillaries are difficult, however. More studies of the electroosmotic flow over wide concentration ranges and microscopic investigations of internal velocity profiles are needed. Cases in which the diffuse double layer region extends to the axis are presented in the literature 116,171, as well a s that for a rectangular cross section 1201, elliptical pores [IS],nonuniformly charged walls I2 11, and parallel plates 1221. In view of the thinness of the double layer in moderately concentrated electrolytes and the fact that electroosmosis can indeed still occur, it is clear that the electrical force can be applied extremely close to the surface and the shear plane may be very close to the Stern layer. The volume transport for the capillary is obtained by integrating equation (6) for the cylindrical coordinates. If the diffuse layer thickness is a small fraction of the radius, the expression reduces to the classical Helmholtz-Smoluchowski result,

628

0.0020 n

0

al

z

0.0015

- +

- no Slip

u)

0

v

0.0010 0.0005 -

0 -0.0005

Figure 3.

slip

$ I

,

I

I

1

I

I

I

_

Velocity profiles calculated numerically using eq. (6)-no slip condition and equation(4)-slip condition Q2 = 26 mV, 10-4 M KC1, EZ = 10 V/cm.

in which qe is the flow rate under electrical gradients, Ac is the capillary cross sectional area, and ke is the electroosmotic coefficient of permeability. In summary, this result is stated to be applicable to systems with large pores, dilute electrolytes, and it predicts that the rate of flow depends directly on the area, &. It is noted that ke is not a function of the pore size. The above theory raises two major questions. The first, already discussed, concerns the role of slip/no slip at the wall surface and requires an experimental approach. The second concerns the nature of the field surrounding the ions in the double layer. This second question pertains to the effective force exerted on the ions (water) because the electrophoretic and relaxation force terms in the DebyeHuckel-Onsager conductance equation are neglected. These terms become increasingly significant at higher concentrations, e.g. 1231. As recognized by Koh and Anderson [18], the Debye-Huckel theory cannot be applied readily t o the double layer, since that region is not electroneutral. One should expect experimentally, then, that the normalized electrokinetic efficiency might diminish with increase in electrolyte and double layers concentration since the net electrical force is reduced by electrostatic drag. Since the ionic strength of the pore fluid will be increased to a concentrated level by the electrolysis reactions a t practical current densities, the classical theory needs to be modified for concentrated electrolytes. A first approximation would be

629

to correct for the background ionic strength, however, a more complete theory would require quantification of any electrostatic drag created between the double layer excess charge and the wall potential as well.

Other Theories of Electroosmosis An interesting subcase of the above is the small pore theory, sometimes known as Schmid theory 161. Unlike the preceding large pore theory, in which the assumption is made that the diffuse layer thickness is small compared to the capillary diameter, here the excess of ions is distributed throughout the entire void volume 1241, i.e. it is assumed that the double layers overlap. Hence, the electrical force can act more uniformly across the pore section, as is the case for an external hydraulic pressure. This condition is treated also by Rice and Whitehead [16],for example. The equation for the velocity, when Ka << 1. becomes

Velocity profiles across the capillary have a Poisseuille shaped flow and the expression predicts that the electroosmotic coefficient of permeability should vary with the square of the radius. In practice, it is found generally that this law is not as satisfactory a s the HelmholtzSmoluchowski approach for predicting electroosmotic behavior in soils. The failure of small pore theory may be because most clays have an aggregate structure with the flow determined by the larger pores [6]. Another theoretical approach is referred to as the Spiegler Friction theory [25,6]. Its assumption, that the medium for electroosmosis is a perfect permselective membrane, is obviously not valid for soils, where the pore fluid comprises dilute electrolyte. An expression is derived for the net electroosmotic flow, 9, in moles/Faraday,

in which the net flow refers to the difference between the actual solution flow and that due to transport of waters of hydration, C3 is the mole cm-3 concentration of free (unassociated) water, C1 is the mole cm-3 concentration of cations, and Xq is the friction coefficient between components i and j (W s2 cm-2 mole-1). Subscripts 1, 3 and 4 refer to the cations, the water molecules, and the solid ionic matrix (wall), respectively. This theory is of interest because it enables isolation of

630

parameters to quantify specific ion/water frictional drag. For a cation exchange resin at 25". X i 3 has a value 2.59 x 108 W s2 cm-2 mole-', which compares to X34 for Na+ ionjwater of 4.8x lo6 W s2 cm-2 mole-l. Incorporation of this model with the classical one for electroosmosis could provide quantitative testing of the slip boundary condition. In addition to the approach using phenomenological equations for modelling ion transport in soils, the theory of irreversible thermodynamics may be adapted to soils [26], a s for the case of ionexchange membranes. Spiegler [25] and Kedem and Katchalsky [27,28] are the prime examples of this approach to transport models. The detailed review by Verbrugge and Pintauro contains a number of other references to mathematical approaches for modelling the fundamental electrokinetic phenomena. Electrical Models of Capillaries and Soils

Schufle et al. [29.301 and Rutgers and de Smet [31] have evaluated the total conductance of solutions contained in capillaries, t o assess the excess surface conductance due to the presence of the electrical double layer. The model treats the system as two resistors in parallel; one comprising a bulk resistance afforded by an homogeneous cylinder of electrolyte and the other is attributable to the ion excess in the double layer at the capillary wall. A formula for the total resistance, R, is given by

where oo is the specific conductivity of the bulk solution, oS is a socalled specific surface conductivity (expressed as a 2-dimensional parameter, i.e. K a >> l),and 1 is the length of the capillary. Through an experimental measurement of the total resistance and accurate assessment of the dimensions and bulk conductivity, the surface conductivity may be derived. This equation is of interest since it may be applied to obtain the fraction of electrical energy that represents the maximum available for electroosmosis: in other words, the surface conductance is the fractional contribution of the total current applied to cause solution flow as the electrical energy is dissipated as mechanical and thermal losses. Schufle et al. have proposed, from surface conductance results of dilute HC1 in small diameter glass capillaries, that long-range ordering of the structure of water may exist near interfaces (diams. ranged 2 mm - 5 microns). This controversial issue is inherently linked with models for electroosmosis, if dc methods are employed for the conductivity determination and a concurrent electroosmotic effect occurs (electrometer measurement methods have presumably been dc).

63 1

Higher activation energies for conductance in the more dilute solutions and with the smallest capillary diameters, as described by Schufle et al., are consistent overall with favorable conditions for electroosmosis. It would be of interest to compare the results of ac and dc methods for conductance of electrolytes in microcapillaries. Problems of defining the conductance of ions in double layer regions, a s mentioned above, persist however. Also, the double layer itself could have some structuring effect on surface water molecules. Modelling the electrical conductance of soils is complicated by a variety of factors. The simplest approach is to consider the porous medium to be comprised of a bundle of capillaries of constant radius, or This does not provide a good model with a range of radii. hydrodynamically, since it does not allow for the converging-diverging nature of the flow pathways 132). Other complications include the charge heterogeneity of the solid particles, colloid transport, the presence of organic chemical matter. etc. Conductance measurements have been used to predict the diffusion coefficients for salt movement through soils 1331. The conductance is considered to be due to a homogeneous solution and a surface process,

in which ECa is the specific conductance of the soil, ECw the pore fluid conductance, 8 is the volumetric water content, and ECs is the surface conductance. Tortuosity may be corrected for with a transmission coefficient. T. It was thought that this model may not be valid for dryer soils. The surface conductance and transmission factor are assumed not to vary with change of the bulk solution strength and can be derived using simultaneous equations and measurements at two pore fluid concentrations. In this experimental study, good agreement was found for diffusion coefficients derived in this manner with those obtained by radiotracer methods. The surface conductance was typically 0.3 to 10% of the bulk conductances a t 1M salt concentrations and the transmission coefficient ranged 0.35-0.53. In ac conductance measurements, the conductances are a function of the frequency. Usually, the electrolysis reactions a t the electrodes are neglected since their effective resistance is small in comparison to that of the soil matrix. High frequency ac conductance methods, which measure both conductance and capacitance, give information on porosity, particle shape and orientation, and dielectric constants, e.g. 134,351. Conductance measurements are useful practically to monitor ionic distribution profiles in electrokinetic laboratory test cells and to characterize the electrical properties of soil samples.

632

RESULTS

Laboratory Studies Many earlier geotechnical studies exist of the electroosmotic effect in soils and clays with a n emphasis on dewatering; in this chapter we limit the discussion to the more recent activities aimed at decontamination of pollutants. Most laboratory tests have used a onedimensional geometry and inert carbon electrodes.

MARIOTTE BOTTLE

ACQUISITION

IH

w

ITRANSDUCER I Figure 4.

CARBON CATHODE

CARBON ANODE

ENDCAP

One-dimensional laboratory test apparatus,

Figure 4 presents a schematic diagram of a one-dimensional laboratory test for electrokinetic soil decontamination. Laboratory tests are highly recommended as a first step in any actual site remediation for feasibility testing, efficiency assessment, and optimization. The soil of interest, in a partially or fully saturated state, may be housed in a suitably inert, insulating cylindrical tube, of typically glass or plexiglass (note: some organic compounds can interact with plastics). In a n open configuration, free ingress of deionized water or a processing fluid is permitted at the anode, with or without a n hydraulic head. Using a Mariotte bottle to provide a small, constant 1-2 cm head, the typical hydraulic flow rate through a 10 cm diameter cylinder of 10 cm length packed with kaolinite clay might be of the order 0.5 to 2.0 ml/day. This quantity can increase as much as 100-200 fold to 50-100 mls/day by application of a small dc current ( 1 mfi to 5 mA). A closed configuration at the anode has been used for experiments investigating

633

dewatering phenomena. To facilitate mathematical modelling, it is convenient to apply constant current and a small, constant hydraulic head across a saturated, homogeneous clay sample. The former is used to establish initial constant flux boundary conditions for electrogenerated products at the anode. The electrode reactions a t the cathode usually are more complex and may change appreciably over extended periods of electrolysis. Constant potential conditions might be applied at one electrode by use of a reference electrode and potentiostatic control. Alternatively, a two-electrode system can be used with a n approximately constant applied voltage. The choice, constant voltage or constant current, will affect the type of model equations used to predict complete clean-up or acid breakthrough (see below). For sample preparation, a clay slurry usually is consolidated to a void ratio of about 2.0 to 3.0 in cylinders a t a maximum consolidation The cylinders may serve also as the pressure of 200 kPa. electroosmosis cells. Some variation in water content occurs across the specimens prepared i n t h i s m a n n e r a n d irregularities in consolidation/compaction cause fairly large variations in the typical electrokinetic parameters that are monitored. It is advisable always to test a minimum of two or more samples in view of this. Some physical characteristics of Georgia kaolinite are presented in Table 1. The economics and efficiency of the process usually depend on the degree of solubility of the contaminant species and the electroosmotic flow rate, although some charged species may be separated by mainly migration in silty deposits. The electroosmotic flow rate, qe, is usually defined empirically from cell tests,

where ke = coefficient of electroosmotic permeability (cm2 5-1 V-1). Qe = electrical potential gradient (V cm- I), ki = electroosmotic water transport efficiency (cm3 A-l s-l),I = current (A), and (T = average conductivity (siemens cm-1). The parameter ke vanes from 1 x 10-4 to 10 x 10-5 cm2 s - 1 V-1 for all soils and is higher at higher water contents. However, as noted above, it may vary as the ingress of charged electrode products alters the ionic strength of the pore fluid, or with clay surface composition. The flow parameter ki varies widely from 0 to 1.2 cm2 s-1 A-’ and is a function of the water content, the ionic strength, and the cation exchange capacity of the soil. Table 2 provides a general comparison of the coefficients of electroosmotic and hydraulic permeabilities for differing soil types. Hydraulic permeabilities can range six orders of magnitude from very porous sands to fine clays. Thus, electroosmosis is most advantageous for use in more nonporous clay media, where pump technologies are too slow to be practical.

634

Table 1. Characteristics of Georgia kaolinite [91

Mineralogical Composition (Yoby weight) Kaolinite Illite

98 2

Index Properties (ASTM D 4318)a Liquid Limit (5) Plastic Limit (5)

64 34

Specific Gravity (ASTM D 854)b

2.65

YOFiner than -2 pm Size

90

Activity

0.32

Proctor Compaction Parameters Maximum Dry Density, tons/m3 Optimum Water Content, Yo

1.37 31.0

Compression Index (C,)

0.25

Recompression Index (C,)

0.035

Permeability of Specimens Compacted at the Wet of Standard Proctor Optimum (x 10-8 cm/sec)c

6-8

aASTM Method for Liquid Limit, Plastic Limit, and Plasticity Index of Soils (D 4318-83)

bASTM Method for Specific Gravity of Soils [D854-58 (197911 CFlexible wall permeability at full saturation

635

Table 2. Comparison of the Coefficients of Electroosmotic and Hydraulic Permeabilities [6,9]

Material Na-montmorillonite (170)' London clay (52) Kaolin (68) Clayey silt (32) Mica powder (50) Fine sand (26)

ke(10-5 cm2

s-1 V-1)

2.0

5.8 5.7 5.0 6.9

4.1

kh(cm s-l) 10-9 10-8

10-7 10-6 10-5 10-4

(*) indicates approximate water content in ?lo

For kaolinite clay a t 50 k 10% water saturation, Acar et al. [8,36] and Putnam [91 have reported that the measured coefficient of water transport efficiency, ki, decreased from a maximum of about 1.8 L/A-hr a t 15 hours of processing to 0.2 L/A-hr at 150 hours, for an electrode current density 0.05 mA/cm2. In many tests, a short induction period with no flow occurred after switching on the current (this has even been observed in single capillary experiments 118, Fig. 41). The flow then quickly increased to a maximum value. Similar behaviour was found in some tests with low levels of contaminant present, e.g., for decontamination of 130 ppm Pbz+ ion from kaolinite a t a current density 0.04 mA/cm2, Hamed et al. [37,38]report a decrease in ki from a maximum of 2.7 L/A-hr. to approx. 0.4 L/A-hr. at 600 hours, Figure 5. These initial ki values are consistent with those found in earlier studies (61 b u t measured efficiencies will vary with the clay type and preparation, such a s the degree of prewashing, the residual ionic salt content, water saturation, etc. The marked decrease in the water transport with time probably is caused mainly by the accumulation and reaction of ionic electrolysis products.

636

1.o h

0

a

0.8

2a

0.6

u)

\

c,

E

0.4

0

v

.Y

0.2 n "0

400

200

600

800

TIME (hrs) Figure 5.

Representative flow behaviour at 0.037mA/cmz, porosity 0.7, with 120 ppm Pb(I1) loaded on kaolinite [37,38].

pH Variations

The occurrence of pH gradients in electrochemical processing of soils has been noted by a number of earlier investigations concerning dewatering [11,12,39-451 and Acar et al. 18,361 and Putnam [9] have made detailed studies of the development of pH gradients in electrokinetic cells with open anode geometry. Attempts to model the early stages of acid-base distributions were made by Acar and coworkers [461. The primary electrolysis reaction a t a n inert anode is the electrolysis of water, provided that very low concentrations of other oxidizable species are present,

and, a t the cathode, 4 H2O

+ 4e- +

2 H2

+ 40H-

114)

The production of H+ ion a t the anode decreases the pH a t this electrode. at a rate depending on the current density employed and the through-volume flow of pure liquid. The resulting voltage drop a t the electrodes should always exceed that necessary to electrolyze molecular water (2 2V),

637

4H20+4e-

-)

2H2+40H-

2H20 + 0 2 + 4 H + + 4 e -

E O

= - 0.8287 V

E O

= - 1.229V ~~~

Net

6H2O -+ 2H2

+ 0 2 + 4H+ + 40H-

E O

= - 2.057 V

and, in the event of a n open boundary condition a t the anode (free ingress of water), there is a n ample supply of water to sustain the primary electrolysis reactions; a t the anode equation (13) and at the cathode equation (14). The presence of other electroactive species in the system will alter the faradaic efficiency of these primary reactions. For example, organic compounds might be oxidized at the anode. Metal ions, hydrogen ion and dissolved oxygen might be reduced at the cathode. The contaminant of interest may or may not be electrolyzable. Overall, the quantity of water electrolyzed is very small relative to the net flux generated by the electroosmotic effect. As the faradaic reaction proceeds, the electrogenerated hydrogen and hydroxide ions contribute to the ionic species already present in the pore fluid as ionic current carriers and, hence, a nonlinear ionic strength conductor is produced. Typically, the residual ionic strength of pore fluid in a kaolinite clay is millimolar or less of chiefly sodium sulfate [9]. Sites requiring remediation may vary widely in their pore fluid, ionic strength and chemical constituents present, both natural and extraneous. The hydrogen ion generated a t the anode has a n essential role in electrokinetic processing, since a low pH can achieve displacement of adsorbed contaminant species at charged sites on the clay surface (this is analogous to flushing a n ion exchange resin with acid to remove adsorbates). At low levels of contamination, many toxic species are likely to be distributed mainly in an adsorbed form rather than in the pore fluid. For example, the adsorption isotherms for Pb2+ ion and phenol on Georgia kaolinite are shown in Figure 6 (37,471. This mineral can adsorb about 1.100 pg of Pb2+ ion per gram of dry clay, which limiting quantity is usually defined as the cation-exchange capacity (for Pb2+, 1.06 milliequivalents/lOO g dry clay). The phenol adsorptive capacity of kaolinite similarly is about 1 mg phenol per gram dry clay. Flushing the soil mass with a n acidic solution causes the displacement of adsorbed species, e.g.. 2H+ + Pbz+(clay)2- + 2H+(clayI2- + Pb2+

(15)

In the case of phenol, hydrogen ion will protonate the molecule, which can enhance its transport to the cathode. Other positively charged ions introduced at the anode, such as NH4+, Na+. Zn2+, etc., could also

638

achieve displacement of adsorbed cationic contaminants but might themselves be unsatisfactory additives to the soil.

-P

3

1

h

0)

1 0 4 ~

0

~

v

0,

c 0

4

v)

-

210 d a 1

-111111111

11l111111

111111111

I11111,1l

lW

a

10-1

1

10

lo2 lo3

Equilibrium Concentration (ppm) Figure 0.

Adsorption isotherms for Pb(II) ion and phenol on kaolinite [37,47].

The rate at which H+ ion is injected into the system may be controlled by varying the current density. For example, assuming the reaction in eqn. (13) is 100% faradaic efficient, the rate of H+ ion production is equal to twice the molar rate of water molecule electrolysis, kHzO. Thus, the steady state concentration of H+ ions will be given by,

Molar concentration =

2 kHZO (moles/sec) Qt (L/sec)

where Qt is the net flux of water passing the anode. electrolysis rate is calculated with equation (17).

kHzo = 1 2F = 5.18 x

I moles/sec

(16)

The water

7

639

if I is the total current in amperes. For example, the solution leaving the anode will be pH 1.83 for 1 square meter of electrode, 0.25 amps current, and 15 L/day water ingress (note that, as discussed above, the electroosmotic flow rate may decrease with time and this complicates species distribution in time). The minimum pH required to desorb and solubilize contaminants may be determined in laboratory experiments with samples from a site. In practical operations, it may be desirable to control the H+ ion production rate, depending on the buffer capacity of the contaminated soil because, a t low pH values, the clay itself will undergo acidic dissolution, e.g. [48,491; so excessive current densities can waste energy and cause chemical changes to the soil. Figure 7 illustrates pH profiles determined in a 130 ppm Pb2+ ion decontamination from kaolinite. a t short and long process durations. Note that a t short process times, the H+ ion disperses rapidly throughout the specimen due to advection (the electroosmotic, convective flux), migration (charged species transport in a n electrical field), and diffusion. The pH determined by in situ insertion of a glass electrode into the soil results in a quasi-thermodynamic determination of the H+ ion activity, since

7

Initial Pore Fluid pH

: I 0

5 -

Figure 7.

0-

,-o - - _----- - - - - --

6 -

I

I

0

0

0

0

a ’&-n-‘ ../-

pH profiles of test cells after 388 A-hr/rns (circles) at 0.012 mA/cm2 and 1982 A-hr/rns (triangles) at 0.037 mA/cm2; open symbolspore fluid: full-in situ.

640

solution phase H+ ions, double layer and adsorbed H+ ions may all contribute to the measured activity. The particulate matter in contact with the glass membrane has an additional unknown influence on the membrane potential. The close concurrence of in situ and pore fluid pH levels close to the anode indicates that the clay ion exchange capacity has been exceeded, whereas closer to the cathode, the large divergencies in these values reveal that the clay surface has not been fully saturated with H+ ions. To conclude this section, some comments are included concerning the nature and role of the cathode processes. Some metal ion contaminants, such as Pb2+ ion, when desorbed by downstream H+ ion will meet the zone of upstream OH- ions. This can cause local precipitation of gelatinous metallic hydroxides, possibly decreasing the cross-sectional area available for flow, as well as decreasing the ionic species available for transport, e.g., Pb2+ + 20H- + Pb(OH)2 soluble, conducting

(18)

insoluble, nonconducting

Since lead is amphoteric, in stronger base the Pb2+ ions will exist as soluble plumbite ion, from which it can be electrodeposited [50]. Not all metal contaminants can be electrodeposited from aqueous solution but most precipitate as hydroxides (exceptions are alkali metals, some alkaline earths, NH4+). Hydrogen ion, meeting the upstream base, reacts exothermically producing nonconducting, molecular water.

This reaction, similarly, reduces the ionic strength available for conduction and thereby assists maintaining the electroosmotic effect. Eventually, however, with sustained electrolysis and the downstream loss of OH- ions, the upstream base will be completely neutralized. The cell then will mainly transport H+ ion from the anode to the cathode where it is reduced preferentially to those metal ions with more negative reduction potentials,

As for the case at the anode, other electrochemical reactions can be chosen a t the cathode by flushing, or choice of electrode material.

64 1

Acid-base effects on the soil surface conductivity and efficiency of electroosmosis need to be understood and quantified better. It is possible to control the anode current density such that the influx of H+ ions just saturates the available surface sites and not add to the pore fluid concentration. In this way, the excess H+ ion in the pore fluid available for migration is minimal and the electroosmotic advective flux should be a maximum possible. Using a n initial cell flow rate of 80 mL/day for 80 cm2 area electrode, we estimate that a current density of 0.007 mA/cm2 should be optimum if the ion-exchange capacity is 1 milliequivalent/lOO g clay and the porosity is 0.7. With changing flow, the current density might be adjusted to maintain the H+ ion influx constant. Higher current densities may be needed, however, if migration is needed to remove contaminant excesses in the pore fluid when the surface sites are fully saturated. At the completion of remediation. the processed volume of soil will have been acidified and natural re-equilibration may be possible. Alternatively, the area can be treated with acceptable alkalis such as lime or ammonia. The system total acidity is determined best with an acid-base titration. If alternative reactions are chosen at the electrodes to the production of acid and base, the effects of these on the whole system need to be evaluated in terms of the toxicity of electrode reactants and products, their influence on the soil/contaminant ionexchange efficiency, the effectiveness of electroosmostic flushing, solubilities, etc. Leach tests can be used to assess if the residual contaminant achieves a level acceptable for ground water quality. Conductivity In electrokinetic tests, the apparent conductivity, Ka, can be estimated from the electrical potential drop across the electrodes and the current, Ka(siemens/cm) = It(amp) Lkm] / Vt(vo1t) AJcm2)

(21)

where It is the current, Vt is the applied cell voltage, L is the specimen length, a n d & is the cross-sectional area. It is assumed that overvoltages due to electrolysis reactions are small relative to the potential drop across the soil and can be neglected. In laboratory tests with kaolinite, for example, at 0.05 mA/cm2, the value of & decreased from about 110 to 50 pS/cm [91. For experiments over longer periods, with kaolinite loaded with 130 ppm Pb2+ ion [37,38]or 100 ppm Cd2+ ion, the & values ranged 80-15 pS/cm and 100-13 pS/cm, respectively [51]. Higher initial conductances result from higher loadings of ionic contaminant. Note that the trend in conductance correlates well with the decrease in water transport efficiency and pH. Pore fluid conductances are higher than in situ soil measurements (uncorrected for porosity) and vary in time across the cell from anode to cathode.

642

Highest values are found close to the anode, where H+ ion has transported into the soil from the anodic water electrolysis reaction. In tests conducted for an intermediate time, a minimum conductance is found close to the cathode, presumably because of the acid-base neutralization reaction, eqn. (19). and possibly clogging, eqn. (18). With sustained processing, however, the conductance is high in the anode section (high H+ ion and anion concentration) and low close to the cathode (anion depleted region). Figure 8.

g- 10,000

"r

v

1,000

,oo

w i-"

ua_s/' 0

"

0

0.2

0.4

0.6

0.8

1.0

NORMALIZED DISTANCE FROM ANODE Conductivity profiles found in situ with 388 A-hr/ms (full circles) at 0.012 mA/cm2 and 1982 A-hr/ms (open circles) at 0.037 mA/cm2 processing, 144 and 117 ppm Pb(II), respectively [37,381. Electrochemical Modelling of the Electrokinetic Process

The purpose of modelling is to develop predictive equations t o assist in establishing the optimal processing conditions and the fullest extent of remediation. Verbrugge and Pintauro have provided a review of transport models for ion-exchange membranes and much of this theory might be adapted to electrochemical soil processing [26].Two fundamental approaches have been identified in this review: the first is based on the Nemst-Planck flux equations and hopes to provide a detailed molecular-level account of the processes: the second approach uses irreversible thermodynamics, without specifying molecular-level interactions. This latter approach uses the Stefan-Maxwell transport equations and concentrated solution theory. In general, it is desirable to obtain a kinetic model to predict the overall system behaviour

643

because this will be the most useful practically for planning and engineering electrokinetic site remediations. Acar and coworkers I461 and Shapiro et al. [521 have presented general models based on the first of these two approaches. These models predict that the contaminant and the electrolysis products a t inert electrodes will be transported and dispersed by advection, migration, and diffusion. Modelling in this manner provides only a first-order, mathematical framework to examine the flow patterns and chemistry generated in the process: adsorption/desorption kinetics, a c i d / b a s e chemical r e a c t i o n s , complex equilibria, a n d precipitation/solubility factors may heavily influence the model accuracy and outcome of any site remediation. Two approaches for mathematic modelling are the use of analytical solutions or numerical, finite element methods (FEM). Both models require adequate definitions for the boundary conditions (nature of electrolyses, flow behaviour). Total mass transfer, qt, due to electrical and material gradients in a soil, according to the Nernst-Planck equation, is

where n = charge on the ion, F = Faraday's constant, R = Universal gas constant, T = temperature, Cp = electrical potential, DJ = diffusion coefficient of species J. z = flow direction, v, = average seepage velocity, p = porosity, and A = average cross-sectional area. The three terms in equation (22) represent chemical diffusion, advective (electro-osmotic + hydraulic) flow, and ion migration, respectively. Acar et al. [46] solved this equation analytically for hydrogen ion transport downstream from the anode. I t was assumed the advective flow, v,. was constant. With this condition and a constant boundary flux, the local H+ ion concentration becomes,

{

c(z,t) = co erfc

(z-kt) 2(D*t)' I 2

+

exp

+1

(I3

erfc

+

kt) 2(D*t)'12} (z

(23)

in which k, taken to be constant, is a linear combination of the migration, electroosmotic and hydraulic velocities. Adsorption/ desorption kinetic effects can be approximated with a retardation coefficient, R, but the role of the nature of the surface species on the electro-osmotic effect is not known. Since it is known now, from numerous bench studies, that the total fluid flow rate decreases with time of processing, numerical methods are preferred for modelling. Approximate models could be employed with bulk constants derived from laboratory tests and the effects of diffusion (slow kinetically) might

644

be omitted. The decrease in the total flow with time presumably is linked with the enhancement of pore fluid ionic strength due to the formation of H+ and OH- ions at the boundaries. Acar et al. (461 provided also an analytical solution for the upstream transport of base from the cathode and discuss some physical consequences of this. Results of FEM simulations for acid-base distributions are provided in refs. [36.46.53,54]. Shapiro et al. [52] similarly have calculated pH profiles for the downstream H+ ion flux but included multicomponent ion fluxes, when two kinematic waves result. Neither model includes any complications arising from nonlinear, radial concentration fluxes or from the acid-base neutralization reaction eqn. (19). The latter will create a heated, dilute zone of neutral pH close to the cathode. Some mathematical modelling papers have been published also concerning the use of electroosmotic processes for ground water pollution control, e.g. 155-581. Selected Examples of Bench Scale Tests Runnels and Larson have studied the use of electromigration to remove Cu2+ from pure quartz. silty sand [59]. Applied voltages were maintained constant at fairly low levels, 1.5-2.5 V, and currents ranged 10-50 @ cm-2. At 0.01 M CuSO4 concentration, pHs in the range 24.50, about 7-53% of the copper was removed by processing to 30 days. These authors did not attempt optimization of the process b u t concluded that electromigration held promise a s a remediation technique to prevent further contamination of ground waters. Daniel and Eykholt 153,601 similarly attempted the recovery of Cu2+ ions. Tests employed voltages of 0-5 V, 0-320 ppm Cu2+ ion on kaolinite clay, and the test cell used anode and cathode reservoirs. Variations in soil pH similar to those reported by Acar et al. were found I8,36,46] but a high accumulation of basic Cu2+ salt resulted in the soil adjacent to the cathode reservoir. I t is concluded that future efforts were needed t o reduce the high pH values at the cathode, Power consumption was typically 7 kWhr m-3. Pamukcu et al. (611 have attempted to remove zinc ions by electroosmosis from kaolinite clay. The experimental cell was designed to have relatively large volume cathode and anode electrolyte chambers. The total zinc concentrations were found to increase from 175 mg/L to 745 mg/L in the anode chamber and to 440 mg/L in the cathode chamber. Zinc ion. similarly to lead, is amphoteric in aqueous media and can form negatively charged zincates. which accounts for the enhancement of zinc at the anode. Quantitative interpretation of these experiments was complicated due to periods with no current application and the addition of NH4OH and NaCl solutions to the anode chamber. Hamed et al. t37.38.511 have investigated the electrokinetic remediation of kaolinite containing Pb2+, Cd2+, and Cr3+ ions. High removal efficiencies. to 95% were achieved for Pb2+ ions present to

645

1000 ppm levels, Figure 9. In all tests, however, the section closest to the cathode had a large amount of precipitated hydroxide salt. Also, the removal efficiencies were > 92% for Cd2+ ions at 120 ppm. However, Cr3+ ions loaded at 120 ppm of dry soil indicated only 60-70% removal efficiencies, which may reflect both difficulties in desorbing this trivalent species and increased complex ion equilibria. Many metal ions react with water producing hydroxide complexes, which will adsorb or precipitate,

C$+

+ H2O

--L

Cr(OHI2++ H20

Cr(OHI2++ H+ .-L

Cr(OH)2++ H+ etc.

Cr(III), in addition, is amphoteric, forming chromites with base, Cr(OH)3 + OH-

+

Cr(OH)4-

(26)

--z 1.o

2% zz u-0

0

0

0.2

0.4

0.6

0.8

NORMALIZED DISTANCE FROM ANODE F'igure 9.

Profile of Pb(II] after processing 388 and 1982 A-hr/ms at 0.012 and 0.037 mA/cm2, respectively. Full circles indicate the shorter processing condition and note accumulation due to displacement 137,381.

1.o

646

Further complications ensue if Cr(V1) species are present. The standard potentials for Cr(VI)/Cr(III)couples are strongly dependent on the pH. H Cr Cr

04-

042-

+ 7H+ + 3e- e-

+ 4H20 + 3e- ==

Cr3+ + 4H20

E" 1.38 V

Cr(OH)3+ 50H-

Eo -0.11 V

Tables of acid/base interactions and thermodynamic references are available for predictions of homogeneous behaviour, e.g. [62.63]. Another complication is the result of soil chemistry. This can be especially important at trace levels when speciation is difficult because analytical methodologies may not have the required molecular sensitivities. A key aspect of the process recommended by Acar et al. I641 is to continue electrolysis for a sufficient time to completely acidify the soil mass. The effects of hydroxide ion formation a t the cathode can be alleviated further by the addition of a weak acid. Anions of this weak acid, for example acetic acid, will migrate upstream but upon meeting the downstream H+ ions, associate to form nonconducting molecules, H+ + CH3 COO-

CH3 COOH

(27)

Thus, the ionic strength increase is offset by the formation of associated acid. Acetic acid additionally is a good choice of cathode depolarizer since most metal acetate salts are soluble. In addition to metals, a number of laboratory tests have explored the feasibility of decontaminating organic pollutants from soils. Shapiro et al. [52] have demonstrated that phenol at 450 ppm and 0.5 M acetic acid can be removed in excess of 90% by the passage of 1.2 pore volumes of effluent. Acar et al. 1471 similarly have studied phenol remediation at a loading 500 pg/g dry kaolinite using electrokinetic soil processing and reported > 95% recovery in the effluent by 2.0 pore volumes. In this latter study, the energy expenditures were calculated to be in the range 18-39 kwh per cubic meter of soil processed. Bruell et al. [661 were able to successfully remove BTEX [benzene, toluene, ethylene, a n d xylene) compounds in gasoline as well as trichloroethylene on kaolinite at concentrations below the aqueous solubility limit. Studies currently are underway at LSU to remediate soils contaminated with higher levels of nonpolar organics using micelle additives (671. U.S.Patents using electrical currents/voltages for soil remediation have been issued to Probstein et al. [651and Acar and Gale 1641. The problems of radionuclide contamination of soils are particularly serious in the U.S. [68-701. There are 33 radioactively contaminated

647

sites listed or proposed'for listing on the National Priorities List. These were caused by uranium mining, the commercial radium industry, or the Federal nuclear weapons research and production programs. The quantity of contaminated soils is very large and since the radioactivity, unlike organic materials, cannot be destroyed and is decreased only by natural decay, the development of remediation technologies h a s focussed on separation/concentration techniques. Consequently, the EPA has initiated the Volume Reduction and Chemical Extraction (VORCE) project to select technologies that can reduce the volume of radioactively contaminated soil [7 1I. Electrokinetic soil treatment has been included as a potential VORCE technology 1721. and as a low cost, in situ method may be favored over soil/acid washing approaches. Bench scale decontamination of uranyl ion from kaolinite has been demonstrated and a yellow uranium hydroxide precipitate collected at the cathode (731. Thorium salts showed limited removal a t the same operating conditions, perhaps because of its larger ionic charge and tendency to hydrolyze. Radium was extremely insoluble and nonextractable from kaolinite. Clearly, the specific chemical properties of each species has to be considered and perhaps pre- or post-chemical treatments used in conjunction with the electrokinetic process. Short reviews of the available bench-scale and pilot-scale studies of electrokinetic soil remediation have been presented by Acar et al. [74]. Field Scale and Pilot Tests

One of the earliest studies of alkali metal ion removal from soils was carried out by Puri and h a n d 175). h r i proposed that monovalent ions will move faster than divalent due to the former's higher dissociation from sites on the clay surface. The volume treated was relatively small, 4.5 m2 and 0.3 m depth. Jacobs and Mortland demonstrated in a later bench study that Na', K+, Mg2+, and Ca2+ ions are leached from Wyoming bentonite by electroosmosis [76], the monovalent ions eluting a t a faster rate than the divalent, in agreement with Puri's suggestion. Krizek et al. [42] showed similarly that the soluble ions content increased appreciably in eluent from electroosmotic consolidation of polluted dredgings, but reported that heavy metals were not extracted. This result may have been due to the formation of insoluble hydroxides. On the other hand, Segall et al. I771 discuss the process leachate obtained from electroosmotic dredgings in a dewatering project and note an increase in heavy metals, organic materials, and alkalinity. All of the processing conditions were not available from this study but the distance between electrodes was 3-5 metres. Hamnet (781 also has studied the reclamation of agricultural soils by removal of unwanted salts by electroosmosis and a U.S. Patent exists for a process termed electrodrainage to reduce alkali and saline salts in soils [791. An interesting use of electrochemical processing is for exploring for minerals in deep soil deposits [80]. Shmakin notes that this method has been used in the USSR since the early 1970s for Cu, Ni, Co and Au prospecting. A porous, ceramic pot with nitric acid is placed as a

648

cathode vessel and the migrating ions are extracted and analyzed. The quantities of extracted metals and their rates of accumulation can be correlated to the ore compositions and the distance from the cathode to the deposits. A field study was attempted for control of radionuclide migration, specifically 9oSr species [Sl].The area treated was 11 x 5 metres with 25 anodes and a central cathode. Although a slight accumulation of 9oSr was found a t the cathode, the study was largely inconclusive since speciation and a comprehensive analytical survey of contaminant transport were not attempted. Another inconclusive study is that of Banerjee et al. [821 for decontamination of a chromium site. Combined hydraulic and electrical potentials were applied and steel electrodes were used, which could react chemically with the effluent. The complexities of chromium speciation have been alluded to above. However, it is stressed that for field and pilot study results to be meaningful, extensive monitoring of the contaminant and its speciation is necessary before, during, and subsequent to the electrochemical process. Perhaps the most detailed accounts of field studies presently Table available are those conducted by Lageman and coworkers [83,84]. 3 contains a summary of the sites and the metal contaminants remediated. The anode and cathode housings are interconnected in this process and form two separate circulation systems filled with different chemical solutions, details of which are not provided in the publications. However, the extent of remediation of many of the metal contaminants tested is quite high. These studies demonstrate that electrochemical processing of soils is a viable and practical technology. There is a need to engineer field studies better in terms of the placement of electrodes, the maintenance of electrode conditions, and the separation/concentration of contaminants. Feasibility studies and practical data used in electrokinetic dewatering technologies may be of assistance, e.g. [ 8 5 ] . Undoubtedly, however, chemical changes in the soil mass must be carefully monitored and understood for efficacious remediation.

Interest in soil decontamination by electrokinetic processing has been increasing steadily, as shown by the volume of scientific studies. There have been two workshop/conferences dedicated to this topic in the U.S. in recent years [86,87] and a number of companies now have offered specialized services in this area. The technology is an emerging one and it is not yet fully mature. There exists a need to conduct further basic and applied studies before it achieves its full potential. In environmental remediation, the site chemistry before and after any application of technical processes need to be fully evaluated. It is recommended that bench scale, laboratory tests be undertaken prior to any site work to help optimize the process parameters. Although the

Table 3.

Field and Laboratory Studies of Electrokinetic Soil and Sludge Processing I83.841 ~

Site Weser River Mud Germany

Contaminant (ppm) Cd

As

10 143 173 56 901 72 0.5 13

Pb

> 3204

Q1

Pb Ni

Zn Cr Hg

Paint factory] (Sediment)

a

Galvanising plant2 (Sandy clay1 Timber impregnation plant3 (heavy clay)

~~

~

~~

Removal YO

After processing (ppm)

5

50 71 54 91 94 64 60 66

41 80 5 54 26 0.2 4.4

587 158

> 82

510

Zn

2080

1624

22

As

143

c 61

57

69

~

~

~~

1 Cathode-anode 2m; 26 sampling locations: mean of 9 values 2 Cathode-anode 1.5 m; soil resistivity 5R cm falling to 2.5R cm: c.d. 0.8mA/cm2 average: mean of 12 values 3 Clay resistivity 10R cm falling to 5R cm: mean of 10 values: c.d. 0.4mA/cm2 average, 96 days.

o\

P W

650

fundamental principles of the process are now better understood and they can be expressed in theoretical formalism, it is essential to conduct studies in assessment of the theoretical models. We note that much of the basic data on sorption, diffusion, dispersion, and migration in capillaries and soils obtained in the last three decades are still insufficient to accurately predict the conduction phenomena in soils, in part due to the added complexity of the electrode reactions: however, in order to provide engineering design/analysis guidelines, implementation procedures and to assess the technical feasibility of alternatives, it is essential to provide a well-established theoretical basis. Nevertheless, the method is highly effective in many laboratory tests and has the potential to provide a low cost, in situ acid wash, which may be effective for remediating many sites contaminated with inorganic / organic pollutants.

ACKNOWLEDGEMENTS Electrokinetic studies at LSU have been supported by the Board of Regents of the State of Louisiana (LEQSF RD (86-89)BlO). the National Science Foundation (MSS-9014711). the Hazardous Waste Center of LSU, the Environmental Protection Agency, and Electrokinetics. Inc. Funds provided by these agencies and organizations are gratefully acknowledged. The authors acknowledge the following graduate students at LSU: A. Alshawabkeh, J. Hamed, S . Puppala, G. Putnam. N. Tran, and A. Ugaz as well as assistance in preparing this chapter from the office and drafting staff of the LSU Chemistry Department, Henry Hurtado and Barbara Marquette.

65 1

REFERENCES

1. Resource Conservation and Recovery Act Orientation Manual, J a n u q , 1986. EPA/530-SW-86-0011 2. Superfund Research P1a n 1989- 90, E PA / 600 / 8- 90/ 037, December, 1989. 3. Guide to Treatment Technologies for Hazardous Wastes a t Superfund Sites, EPA/540/2-89/052, March, 1989. 4. Evaluating Electrokinetics as a Remedial Action Technique, Second Intl. Conf. on New Frontiers for Hazardous Waste Management, EPA/600/9-87/0 18F; ORD, September, 1987. 5. Aveyard, R., Haydon. D. A., An Introduction to the Principles of Surface Chemistry, Cambridge University Press, 1973:52-57. 6. Mitchell, J. K., Fundamentals of Soil Behaviour, J o h n Wiley and Sons, Inc., NY, 1976;353-359. 7. Mitchell, J. K., Yeung. A. T., Transportation Research Records, National Academy of Science No, 1288,Geotechnical Engineering 1990;1-9. 8. Acar. Y. B., Gale, R. J., Putnam, G.. Hamed, J . , Second Intl. Symp. on Environmental Geotechnology, Shangai, China, May 14-17, Envo Publishing, Bethlehem, PA, Vol. 1, 1989:25-38. 9. Putnam, G. A., M. S . Thesis, Louisiana State University, December, 1988. 10. Casagrande, L., Die Bautechnik 1937; 15(1), 14-16;USP 2,099,328, Nov. 16, 1937;Nature 1971: 160,470-471: J . Boston Soe. of Civil Engineers ASCE 1983;69,255-302. 11. Esrig, M. I., Gemeinhardt, J . P., J . Soil Mechanics and Foundations Division ASCE, Proc. Paper No. 5235, 1967: 93 (SM3). 109-128. 12. Mise, T., Proceedings, 5th Intl. Conf. on Soil Mechanics and Foundation Engineering 1961;1, 255-258. 13. Bard, A. J., Faulkner, L. R., Electrochemical Methods, Fundamentals and Applications, John Wiley and Sons, NY. 1980; 504,506. 14. Mohilner, D. M., In: Electroanalytical Chemistry. A Series of Advances. Bard, A. J . , ed., Marcel Dekker, Inc., NY. 1966;Vol. 1, 247,318-321. 15. Dukhin, S . S . , Derjaguin, B. V.; In: Surface and Colloid Science, Matijevig, E., (ed.), Vol. 7. 1974;49-272. 16. Rice, C. L..Whitehead, R.. J . Phys. Chem. 1965;69(11).40174024. 17. Newman, J. S . , Electrochemical Systems, Prentice-Hall, Inc., N J , 1973;Chapt. 9,190-207. 18. Koh, W.-H., Anderson, J. L., AIChE Journal 1975: 21(6), 11761188. 19. Stulik. K., Pacakova, V., Electroanalpcal Measurements in Flowing Liquids, John Wiley and Sons, NY 1987;Chapt. 2,pp. 27-81. 20. Burgreen, D., Nakache, F. R., J. Phys. Chem. 1964:68,1084.

652

21. Anderson. J. L., Idol, W. K.. Chem. Eng. Commun. 1985; 38, 93106. 22. Ohshima, H., Kondo, T., J. Coll. and Interface Sci. 1990; 135(2), 443-448. 23. Glasstone. S . , An Introduction to Electrochemistry, D. Van Nostrand Company, Inc. NY, 1942; 87-95; Justice, J-C. In: Comprehensive Treatise of Electrochemistry, Conway, B. E., Bockris, J.O'M., Yeager. E. (eds.), Plenum Press 1983; Vol. 5, Chapt. 3, 223-337. 24. Schmid, G. 2.. fur Elektrochem. 1950; 54, 425; ibid. (1951); 55, 684; Elektrochem. und Angev. Phys. Chemie 1950; 54, 424-430. 25. Spiegler, K. S,, Trans. Far. SOC.1958; 54, 1408-1428. 26. Verbrugge, M. W.. Pintauro, P, N. In: Modern Aspects of Electrochemistry, No. 19, Conway, B. E.. Bockris, J.O'M., White, R. E. (eds.), Plenum Press 1989; Chapt. 1, 1-67. 27. Kedem, 0.. Katchalsky, A., Biochim. Biophys. Acta 1958; 27, 229. 28. Kedem, O., Katchalsky, A. , J. Gen. Physiol. 1958; 45, 143. 29. Schufle, J. A., Yu, N.-T., J. Coll. and Interface Sci. 1968; 26(4), 395-406. 30. Schufle, J. A., Huang, C.-T., and Drost-Hansen, W., ibid. 1976; 54(2). 184-202. 31. Rutgers, A. J., de Smet, M., Trans. Far. SOC.1947; 43, 102-111. 32. Kozak. M. W., Davis, E. J., J. Coll. Interface Sci. 1986; 112, 403. 33. Palmer. C. J,. Blanchar. R. W.. Soil Sci. SOC.Am. J. 1980: 44. 925929. 34. CamDanella, R. G . . Weemees. I., Can. Geotech. J. 1990; 27, 557567.35. Arulanandan, K., J. Geotech. Engrg. Div. ASCE, 1991; 117(2),319330. 36. Acar, Y. B., Hamed, J.. Gale, R. J., Putnam, G., Bull. Transportation Research Record No. 1288, 1990; 23-34. 37. Hamed, J., Ph.D. Thesis, Louisiana State University, December, 1990. 38. Hamed, J., Acar, Y. B., Gale. R. J., J. Geotech. Engrg. ASCE 1991; 117(2), 241-271. 39. Cambefort, H., Caron, C.. Geotechnique 1961; 11(3),203-223. 40. Titkov, N. I., Petrov, V. P., Neretina, A. Y., Special Research Report Translation, Consultants Bureau, NY. 1965; 70 pp. 41. Gray, D. H., Geotechnique 1970; 20(1), 81-93. 42. Krizek. R. J., Gularte, F. B.. Hummel, P. B., ASCE Natl. Water Resources and Ocean Engrg. Convention, San Diego. California, Preprint 2641, April 5-8, 1976. 43. Wan, T. Y., Mitchell, J. K., J. Geotech. Engrg. Div.. ASCE. 1976; 102(GTS) 473-491. 44. Segall. B. A., O'Bannon, C. E., Matthias, J. A.. J. Geotech. Engrg. Div., ASCE, 1980; 106(GTlO), 1148-1152. 45. Lockhart, N. C.. Colloids and Surfaces, 1983; 6, 229-269.

653

46. Acar, Y. B., Gale, R. J.. Putnam, G. A., Hamed, J., Wong, R. L., J. Environ. Sci. Health 1990; A25(6), 687-714; Alshawabkeh, A.. Acar, Y.,J. Environ. Science and Health 1992 (in press). 47. Acar. Y. B., Li, H., Gale, R. J., J. Geotech. Engrg. ASCE 1992; 118(11). 48. Kittrick, J. A., Clays Clay Miner. 1969; 17, 157-167. 49. Carroll, S. A., Walther, J. V., Am. SOC.of Science 1990: 290, 797810. 50. Modern Electroplating, Lowenheim, F. A. (ed.), J o h n Wiley & Sons, Inc., 3rd Edn., 1974: 266. 51. Acar, Y. B., Hamed, J. T., Gale, R. J., Geotechnique 1992 (submitted). 52. Shapiro, A. P., Renaud, P. C., Probstein, R. F., PCH Physic0 Chemical Hydrodynamics 1989; 1l(5-6), 785-802. 53. Eykholt, G. R.. Ph. D. Thesis, The University of Texas a t Austin, May 1992. 54. Lewis, R. W., Humpheson, C., J. Soil Mechanics and Foundations Division, ASCE, No. SM8, 1973; 99, 603-616. 55. Kirkner, D. J., Reeves, M., Water Resources Research, 1988; 24(1). 1719-1729. 56. Lewis, R. W., Humpheson. C., Bruch, J., Ground Water 1975; 13(6),484-491: (Bruch, J. C. PB-272377 Report, National Science Foundation, UCSB-ME-76- 1, 1976; 80 pp (available NTIS). 57. Mangold. D. C., Tsang, C., Rev. Geophysics, February 1991; 29(1), 5 1-79. 58. Yeh. G. T., Tripathi, V. S . , Water Resources Research 1989; 25(1), 93-108. 59. Runnells, D. D., Larson, J. L., Ground Water Monitoring Review, Summer 1986: 81-91. 60. Daniel, D. E.. Eykholt, G. R., Progress Report, Gulf Coast Hazardous Substances Research Center, Beaumont, Texas, Oct. 30, 1989. 61. Pamukcu, S . . Khan, L. I., Fang, H-Y., Transportation Research Record No. 1288, 1990; 41-46. 62. Pourbaix, M. Atlas of Electrochemical Equilibria, Translated by Franklin, J. A., NACE. Houston, Texas, 1974. 63. Bard, A. J., Parsons, R., Jordan, J. (eds.), Standard Potentials in Aqueous Solution, IUPAC, Marcel Dekker, Inc.. NY, 1985; 453461. 64. U.S. Patent 5, 137. 608, August 11, 1992. 65. U S . Patent 5, 074. 986, December 24, 1991. 66. Bruell, C. J., Segall, B. A., Walsh, H. T., ASCE, J E E 1992 (submitted). 67. Tran. N., Gale, R. J., Acar, Y. B. (work in progress). 68. Assessment of Technologies for the Remediation of Radioactively Contaminated Superfund Sites, USEPA 520/ 1-89-004, December 1988. 69. Technological Approaches to the Cleanup of Radiologically Contaminated Superfund Sites, EPA 540/2-88/002, August, 1988.

654

70. Frankena, F., Frankena, J. K., Radioactive Waste as a Social and Political Issue: A Bibliography. AMS Press, NY, 1991. 71. Richardson, W. S . , Particle Size Distribution, Radiochemical Distribution, and Chemical Wash Studies on the Contaminated Soils from Montclair and Glen Ridge, New Jersey, U.S. EPA Report. Contract No. 68-02-4375, 1988. 72. Coe, L., Steude, J., VORCE and Other Technologies for Treating Sites Contaminated with Radioactivity, U.S. EPA Report, Contract NO. 68-02-4375, 1989. 73. Feasibility of Removing Uranium, Thorium, and Radium from Kaolinite. EK-BR-009-0292 Electrokinetics, Inc., Baton Rouge, Report to U.S. EPA, 1992. 74. Acar, Y . B., ASCE Special Technical Publication No. 30, ASCE Grouting, Soil Improvement and Geosynthetics Conf., New Orleans, Feb. 28. Vol. 2, 1992, 1420-1432; Acar, Y . B., Alshawabkeh, A., Gale, R. J., Proc. Mediterranean Conf. on Environ. Geotech., May 25-27, 1992 Cesme, Izmir, Turkey, Balkema Publ. Inc.. Rotterdam, Netherlands (1992). 75. Puri, A. N.. Anand, B., Soil Science 1936; 42, 23-27; Puri, A. N., Soils, Their Physics and Chemistry, Reinhold Publ. Corp., NY, 1949: 134-140. 76. Jacobs, H. S., Mortland. M. M., Proc. Soil Science SOC.1959; 2342, 77. Segal, B. A., O’Bannon. C. E.. Matthias, J. A., J. Geotech. Engrg. Div., ASCE, No. GT10, Oct. 1980; 106. 1143-1147. 78. Hamnet, R., M.Sc. Thesis, University of Manchester, England, 1980; 84 pp. 79. U.S. Patent 2, 831. 804, Callopy. J. P., April 22, 1958. 80. Shmakin, B. M., J. Geochem. Explor. 1985; 23(1), 35-60. 81. Case, F. N., Cutshall, N. H., Symp. on Scientific Basis for Nuclear Waste Management, Boston, MA, USA, Nov. 26-29, 1979. Conf. 791112-28, 5 pp. 82. Banerjee, S., Horng, J., Ferguson, J. F., Nelson, P. 0.. Unpublished Report presented to U.S. EPA, Land Pollution Control Division, RREL, CR811762-01. 1990, 122 pp. 83. Lageman, R., NATO/CCMS Pilot Study, Copenhagen, Denmark, May 9, 1989; 18 pp. 84. Lageman, R., Pool, W.. Seffinga, G., Forum on Innovative Hazardous Waste Treatment Technologies, Atlanta, GA, 19-2 1 June, 1989; Chemistry and Industry, 1989; 18, 585-90. 85. Sprute, R. H., Kelsh. D. J.. Thompson, S. L.. Bureau of Mines Report of Investigations/1988, RI 9137; 75 pp. 86. U . S . EPA Workshop on Electrokinetic Treatment and Its Application in Environmental-Geotechnical Engrg. for Hazardous Waste Site Remediation, Univ. of Washington, Seattle, Washington, August 4-5, 1986. 87. U S . DOE Workshop on Electrokinetics, Atlanta, Georgia, January 22. 1992.

655

Electrochemical Utilization of Rydrochloric Acid -Waste 01 Chlororsanic ProBuction G.A.Tedoradze and V.Z.Kazarinov The A.N.Frumkin Institute of Electrochemistry, R~ssianAcadmy of Sciences, Lenirlsky prcspect 31, Moscow 117071 R i i s s l a The world prociuation of ohlorine, 2s of: 1330, is hbout 43 mln tons [II . Edcre than 70% of ohlorine prodaced, is h o m t o be spent f o r production of ohlororganic compounds. For example. socording to the reaotion of chlorination of ethylene (addition )

C12

'ZH4

-

-

c2H4c12

or t o the reactior, of substitution

RH t C12

RC1 t HCi

or according to other reactions; part of them is presented lower. I. Halogen replacement:

R-C1 2.

t

HF

+

R-F

-

t

HC1

Production of silicons:

4R-C1

251

t

2R$3IC12

R

2R2sic1

H20

t

R

Cl \

Sl

I

/

- 0 - si I \

/

R

c1

R

According to these reac-ions some par. of chlorLne used fonns the waste !iycirogen chloride. Liiount of the waste depericis on the type of reaotion. In the case of adcilti.cn this n n o ~ m tl a nct very big. A s for the substitution half of the chlorhe reacted, and in some cases the whole of the chlorine, turns to hydrogen chloride. ObtaLying of isocya?ates by phosgenating of corresponding mines can serve as an example of such reaction:

co

t

C12

coc12 t

-

q

-

COCl2 RNCO

t

wc1

R here is acyclic or aromatic radical. The majority of chlorine is s t i l l spent t o produce chlorri-

656

organic compounds by mean5 of the substitution reactions. A s it was already mentioned, 1 mole of chlorine produces 1 mole of hydrogen chloride. Thus, no less than 30 per cent of the total chlorine, produced in the world twns to the hydrogen chloride, which o a n be utilized in different ways. Firstly, HC1 together with aoetylene gives valuable product vinyl chloride:

C2%

t

HC1

-

H$=CHCl

Secondly, hydrogen chloride can be used in the reaotion of oxidizhg chlorination producing, f o r example, chlorobenzene.

2C&

t

2HC1 t

o*

-

2C6HgC1

t

2H20

The method of oxidizing ohlorination seems t o be rather powerful and uriversal one. However, high terngeratures and pressures, necessary f o r this technique, require complex equipment, as the prooesses take place in t h e corrosionactive media. Fmthemore, catalysts, used in this method being easily poisoned, require thorough purification of hydrcgen chloride. Regeneration of chlorine from waste hydrogen chloride by chemical means (Deacon process modified) requires good purification also, as catalysts become inactive i:i t!ie presence of admixtures, that hydrogen chloride contains. And, at last, shortcomings of techniques, mentioned above, have caused a number of attempts to use hydrochloric acid ac an electrolyte f o r !Wdrogen and chlorine obtaining. A s far back as 1964, the firm "9oechst" obtained 100 tons of chlorine per day with the help of a cell, worked out by the firm "Friedrich Uhde". This technique is widely used nowadays: about 1 mln ton8 of chlorine has been obtained by hydrochlorio acid electrolysis by 1990. In spite of the obvious successes in the field of hydrochloric aoid electrolysis, utilization ol" acid solutions, containing admixtures of organic compomdB, remains to be an ixsol-red problerr,. As it will be shown in the Chapter 1 , attempts to utilize such solutions have failed: eleotrolysis of waste hydrochloric acid at the diaphragin cell requlres expensive thorough purification from the organic admixtures, otherwise diaphragm appears to be out of action. In the USSR problem of hydrochloric acid u2ilication was studied by 3 team of scientists: one of them Worked in Moscow and - two in Sumgaite (now Republic Azerbaijan). They have worked out an original idea: the solutions of waste hydrochloric acid, that hadn't undergo stripping were electrolyzed. Electrolysis was held in the absence of diaphragm Kith the simultaneous obtaining of val*able chloroorganic compounds. Parameters of the process were ohosen 80, that accumulation as well as atmospheric emission of chlorine didn't take plaoe. The chlorine generated was entirely utilized to chlorinate organic compounds, presenting in the solution. These organic oompounds were specially selected to produce after their chlorination

657

ohloroorganic compounds with high efficiency and eeleotivity. The utilization of hyeochloric acid for chlorination prooess is complicated by the fact, that nature of the final product depends,on the aoid conoentration (at least for the process of olephines chlorination). For example, at the 75-30 per cent range of HC1 concentrations olephines produce, while elecrolyzing, almost only dichloroparaffines. Eowever, diluted HC1 solutions produces electrolysis of more chloroalcohols v;ith hi@ seleotivity. Such features of H C 1 solutions obliged to create complex teclzniqiie of acid utilization that differs f o r concentrated and diluted E I C l solutions. TheEe technologies have been installed o n a factory of Simgaite. I: should be emphasized, that dichloroethane obtained as a result of this process was p w e r than that produced by traditional means. The p~esent review distinguishes from other materials on this problem due to the fact, that ail the results obtained by three mentioned teams are generalized and tecl-iiological principals of production of the most valuable chloroorganic compound - dichloroethane is given here. Literature on electrochemical chlorination of organic oompounds was exhawtive represented in monograph. published recently 123. Possible ethylene chlorination n?chaniBm 5.n ooncentrated hydrochloric acid Goiutions tha:, results In producin& 1,2-dichloroethane was also discuseed there Gi detail. Potentials, mentioned i n this review are related to saturated calomci electrode. 1. Hydrochloric acid electrolysis producing chlorine and hydrogen.

Hydrochloric acid electrolysis producing both chiorbe and Pidrogen is possible only in the diaphmgrn cell. This is an essential shortcorn-hg of the teohnique under oonsideration, as voltage drop on the diaphragm plays a signifioant role in the tot31 balance. At the 20% concentration. 80 C tcrrgerature and 2 kA/m2 cwrent density, voltage clrop on the diaphragm reaches the value of 0,32 V, that makes g p more than :OX of total voltage. Since zhlorine dissolve& in hydrochloric acid nuch brine diaphragm or membrane fo? hydrochlcric better than acid electrolysis must be more close than tkat for brine electrolysis. Otherwise current yield of chlorine mu16 decrease as chlorine easily reduces on cathode. The best materials f o r diaphragm o r membraqe are o l m e asbestos, polypropylene, vinylohloride copolymer with acrylonitril, polydnylhdenchloride, polytetrafluorethylene. For a long time the problem of diapbagm, stable u d e r conditions of hydrochloric aoid electrolycis have been the most difiicult one. However, in 1960-1970 a new membrane "Naloion" based on the 2erfluorosulfonic acid was worked out. The membrane's stability even in hot solutions of hydrochloric acid appeared to be satisfactory one. Another difficult problem while HC1 electrolyzing is 8 problem of electrodes. The choice of anode material for

658

electrolysis of H C i is more limited than in the case 0-4 NaC1. A number of oxidic electrodes successfully used at brine eleotrolysis, are unsuitable at HCI electrolysis. For example, electrodes DSA (dimensionally stable anodes) type dissolve well at the HC1 concentrations exceeding 4-6%. In the acidic solutions platinum is unstable as well - it dissolves with 100% current efficiency at the potentiel value lower than 0.7-0.W. Though at the high potential !glues platinum passivates, in the 15% H C 1 solution almost 10 X of total current la spent to d i s s o l v e platirm (at 1.2 V). It should be noted, that plattnum alloys with iric?ium are considerably more stable. The rate of alloy dissolution is 5 times slower at the 10% iridim content. However, at the great irldium content mechanical features of electsode becomes worije. Graphite Beems to be the best material f o r electroly~isof ooncentrated Wdrochloric acid. Corrosion stability of solid graphite anode enables it to work for several years under following conditions: absence of strong oxidizers a d state 18-27% concentratim of hydrochloric acid 131. Graphlte locses per 1 ton of chlorhe generated accounts f o r 0,l-1,s kg acoording to the different data 141. The drawback of graphite electrode must be considerable electric resistance, that causes great ohmic loe5es in the anode body. At the ollrrent density value of 2 kA/m voltage drop in the anode b o Q reaches the value of 0,26 V. Concentration of hyclrochloric acid in the ~olutioni6 h-mvn to be one of the most important factors, influencing eleotr*olysisprocess. Firstly, as graphite electrodes stability depends on HCl omcentration, it shouldn't bz allowed to decrease more than to 75-18%. Secondly, HCL concentration determines chlorine yield. The yield close t o 100% can be gotten only if the concentration is not less than 8% i31. However, under conditfons of real eleatrolysis the yield is always less than 100% For example at the 4 mm hterelectrode space, 75°C and H C 1 concentration of 20-26%, the yield of chloriqe accowit f o r 95,6X, current density being equal to 1 kA/;n,; and correspondingly 97% at the currenk density of 4 k A / m . This phenomenon m u s t be caused by the reduction of chiorhe, difzused to the cathode space. W-iie usirg graphite electrodes of hcreased porosity in the bipolar cell, yields appear to be even lower, less than 30-91%, that must be suggested by t h e reduction at the cathode side of chlorine, diffused tknough the electrode pores. Thirdly, solution electric conductivity and, 30 the voltage drop on the cell ourrsnt colleotors, conoiderably 2epends Gpon the concentration of electrolyte. It should bc noted, that conductivity achieves its maximum at hjrdrczhloric Loid concentration about 13-22%, at 66°C. The increase In the temperature results in narrowing of concentmtgon range, correspondent to conductivity maximum. Thus, at 88 C electric conductivity maximu? corresponds to the 18-21X concentrations [31. Due to the mutual overlapping of different factors 141 (among them theTe are factor, tnat determine overvoltage of chlorine and hydrogen discharge as well as solution filling with gas) voltage on the cell current collectcrs appears to

.

659

achieve m i n i m u m at thp concentratiom 2 3 4 6 % (1,39 V, 75"C, current density 3 U / m , interelectrode space being e q m l to 4 mm). Further rise of concentrations is followed by sharp increase in amount of hydrogen chloride, leavhg the Bolution due to the considerable increase in its vapor pressure. e'T temperature appreciably influences the voltage drop on the cell. For example, increase in temperature from 40 to 50°C at 20-26% concentration of hydrochloric acid caises 0,14 V decreaze ir, voltage while those 70 to 80°C - additional +xrfase 0,04 V (current density in both caEes is equal to 4 kA/m ) 141. T a k i n g into account the difficulties, concerned with corrosion stability of equipment and anodes at high temperatures, the optimal range of temperatures m u s t be that from 60 to E O C . The current density influence on the process of hydrochloric acid electrolysi3 seem5 to be varied. On the one hand, the rise of cwrent density i- followed by increase in both expenditure of eleztric power and chlorhe cost, - since it will cause an increase in operation expenses, in particular, expense of cold for cooling HC1 before electrolysis. On the other hand, the production rated at the work at increased current densities requi-es iesser investments. It should be taken into accomt also9 that while increasing the current density from 1 t o 4 IcA/m. the chloriye yield rises from 95,6 to 97% 141. Sometimes it is necessary to hold the elestrolysis of diluted hydrochloric acid, which concentration is less or equal to 17%. Graphite electrodes being rather stable ?n the concentrated €El, begin to destroy catastrophiaaily jn the diluted acid solutions. The graphite corrosioa reaches the value of 50 kg per 1 tor, of chlorine, that is 10 t-hes more than that of NaCi electrolysis 151. The destruction of paphite electrode6 is caused by 2 reasons: firstly, graphite forms C02 while bmi,ng, seconGly, it cruzbles. Under the industry conditions, the solution, containing graphite particies, should be filtered or sedimented, otherwise graphite slag woald fall out, making it, difficult to feed the acid. Partly conversion of graphite to CO, leads to the undesirable oonsequences, sinoe it contaminates chlorinegas aqd decreases chlorine yield (by 5 7 % ) . Another fact of great importance is the change of interelectrode sp.ace, tamed by anode corroaion. Thus, increase in current density lead to considerable deterioration of process parameters. There are no any data h the literature on using of other material, besides graphite, f o r industry electrolysis of hydrochloric acid. The most popular during recent years s e e m to be the electrolysis, using solid polymer eiectrolyte {SPE). Sevei-a1 works, showkg the SPE effectiveness in the case of HCI electrolysis have appeared recently [6-91. SPE is an ior, exchange membraqe, both sides of whioh are covered by oatalytio active layer. This layer is usually represented by reduced oxides of the metals f?om platinum group, or their mixtine. They are often awornpznisd by the recluced oxides of titaniuq, niobium, tantaium m d othera. Such L

660

additions are t3 increase stability of catalyst layer against infhence of chlorine and ozygen. Polytetrafluorethylene usua;ly serves as a bindagent. Anode mass usualiy contains fine dispersed graphite. Usual thickness of a layer varies from 10 to 300 pm. DescripLion of SPE o e i l ao well as teohnique of electrodes preparation is given in [10-141. There ai-e almost no ohmic drops related to the current passage through electrolyte in the SPE cells, since the charge transfer prooemes proceed via ior, exchange membrane to the metal porous electrode. Thc gain in a voltage drop when applying the SPE electrodes natunally, grow with a dilution extent of the used acid. T h i s gain, however, I s concerned not only with a voltage drop in the electrolyte, but also with the fact, that overvoltage of both oathocle and anode processes io decreased on suoh electrodes with rather developed surface. Distribution of current in SPE is also improved, since electrio field between the electrodes becomes more uniform. B i n d i n g of different hydrophilio groups to the sdrface of membrane (for example - P0Cl2, S03H and others) is often m e d to improve the SFE features [10,15l. So, device with SPE is an electrolyte made of m l i d polymer, located betwem two porous electrodes. If one places the hydrochlorio acid solution into the anode space and water in the cathode space of mentioned device and apply constarit voltage to it, the electrolysis of hydroohloric acid begins. Chloride ions iyhile diffusing to the interface betweea membraie and porous anode, dimharge. Protons discharge as well, when having migrated from the anode space via ion exchange membrane, they reaoh the interface between membrane and cathodc. Hydrogen is generated in the space of cathode, chlorine that of anode. These compounds diffuse via the porous electrodes to the solutior, and give o f f as Motion of hydrated prGton through the membrane leads o the electroosmotio transfer of water to the cathode space. Due to the imperfect membrane selectivity partial chloride ions tramfer takes place. As a result, diluted hydrochloric acid appears in the zathode space, while in the ancde spaoe its amount decreases. Porous anode 'I, used in SPE electrolysis (rig.: ) aonsiste of mixture of ruthenium dioxide (75%) and iridium oxide bound to the graphite layer. Thickness of such layer depends on the mount of graphite, coverkg the ,anode surface unit. In particular, If this amount is 40 g/m, oorreRpondent thickness to anode is held with the reaohes 100 pm [123. Current feed' help of point collector 6 (the z a l net can serve 3s this collector). Anode space is separated from the cathode one with the help of membrane SPE ( 2 ) . Platinum blaok, serving as an anode 1 , is 100 pm wi8e. Cathode point collector S i~ connected with graphlte zlate 5 , that maintain direct contaot with cathode. A weak point of SPE technique is an opportunity to impoverish electrolyte in the electrode pores. Decrease in hydrochloric acid concentration iE known to cause decrease in chlorine yield owing to the formatlon of oxygen and chlorine oxyacids [ I h l . Only in the case, when concentration is not l e E s than b X , accessory processes are almost absent. One might avoid

. Yes

66 1

decrease in the yield, oawed by eleotrolyte irripoverishment in the pores only by considerable decrease of poi"oils electrode thickness. While using anode of 50 pm or less wide, the yield of oxygen can be considerably reduced on the SPE even 5 n the diluted solutions of hydrochloric acids. SPE cell with cathode depolarized by oxygen is considered Graphite aotivated by to give the be8t features [8,9,12,15,171. mixture, oontaicing the oxides of ruthenium, iridiun, and titanium, serve as an qode. Mixture, containing 1 0 Q of catalyst is req'ciired per 1 m of anode. Cathode contains plrtinum black. The ce;l assembly to Fig.;! [12-141 inoludes of cation permeselective SPE aembraqe 3 , that separates the oell into anode and ccthode chambers. A gas permeable cathode 4 consisted of a layer of electrooatalytic particles and partioles of a polymerio binder is bonded to the upper surface of membrane 3 . A gas and liquid permeable oatalytic anode (nut shown) is bonded tc; otlrdr side of membrane 3 . Each of electrodes physioally fom. a part of membrane 3 and is in the intimate electrical oontxt with the membrane. The oell assembly is clampcd between respectively Cathode ( 1 ) and m o d e (11) end plates. Plates aFe connected to the terminals of a power supply. Xiobiurn ourpent distributing .%Green elements 2 and 5 m e positimed between oathode end plate and anode current collecting zlernent (1 and 6 respectively). Collector 6 which is molded graphite and fluorocarbon oom2osite consists of a main body 7 1 , a ohamber 7 within the m a i n body, and of array of oonductive contaot elaments 10 which contacts either screen 5 or the anode electrode bonded to the underside of membrane 3 . There m e inlet (8)znd outlet (9) conduits for the introduction fresh and for tke removal spent hydrochloric aoid. Electrical contact to the bonded anode is made though the array of contact elements 10. Cathode chamber is supplied with oxygen at the r w m temljerature. On the oathode oxygen receives electrons:

-

O2 t 2H;zO + 4e 4aKIons of hydroxyl are neutralized by cations of hydrogen, that have diffused through the cathode rnembrane.21n suoh a cell one o m achieve current density more than 4 kA/m spending less than 1000 kwh of electric energy per 1 ton of chlorine. It is well seen from the inr"ormation mentioned above that parameters of SPE electrolytic cell for hydrochloric acid electrolysis have a number of zdvantages over those of industry application. However two Pacts cannot fail to be mentioned. Firstly, during prolonged elaotrolysis ion exchange diaphragm depades and loses selectivity, that leads t o the worsening of the electrolysis showings. Secondly, admixtures presenthg in the electrolyte can lead to the deorease in activity of eleotroda catalyct; that muses the rise of overvoltage on the electrodes followed by the increase in electric energy expenses. It should be noted, that there are no mentions about opportunity to electrolysis waste HC1, that contabs oFganio oonpounds in any patent, Euggesting SPE. Tha data o n the electric energy expenses per obtaining cf 7

662

ton of chlorine by means of different teohI?iques are summed up in the Fig.3. It is well seen from the figure, that after 7 months of continuous work the electric energy expen6es of SPE apparatus achieve the level of the best industry C B L ~ S , that prooess concentrated waste hydrochloric acid. All the above material oonfhn that the problem of industry electmlysis of pure hydrochlorio acid is solved in general. Raste HC1 from several chlororganio productions ( f o r example, the chloromethane production [181) can be electrolyzed as well. Electrolysis of such acid cause no difficulties unless an organic phase appears, as its presence sharply reduces the diaphmgn iifetime. Chlorine, the m a i n purpose product, oontains only tetraolorooarbon - the final prodwt of methane chlorination. However, in the process of electrolysis the diaphragm work takes a turn to the worse if the eleotrolyte oontains benzene, chlorobenzene, phenol and its derivatives; as concomitant chloroderivatives fall out on the diapl-Lz;agm,sharply increaeing its rcsiatance. To avoid undesirable phenomena, it is suegested [I81 t o held adiabatic absorption of HC1 to purify it with the help of activated carbon. Tne technological aspects of the prooesses dizcussed above am described in detail in the monographs [2,'l9,,?3,211.

2. Hydrochlorlc acid electrolysis producing hydrogen and chloroorganic compounds. 2.1 Electrolyals of concentrated hydrochloric acid.

.

2.1 1 Electrochemical chlorination of aromatic hydrGCWbGn6 t 22-31 I A s i t was stated above, production of

chloroarornatio compothe wacte - hydro&en ohloride, difficut for utilizirg. That is why attempts to chlorinate electrochemically benzene and its hornologues in order to utilize ''in situ" hydrogen chloride, that; is formed as a result of substituted chloririation of aromatic OompQundS, seem to be natural ones. Whrle chlorinating of benzene homologues, the n a b question is the place of chlorine binding: aromatic nucleus or the Bide chah. In the case of chemical ohlorination these processes a r e mastered well [ 2 ] . It was necessary to determine the oonditions of selective chlorination of either the nucleus or the side chain in the case of electrochemical chlorination. The research of the influence of hydrochloric zcrd concentration on the prrrpose products yield has shornl that graphite electrodes are unctable at the acid conoentration less than 10-728. In the further eq~erhents of aromatic hwbocarbom chlor-hation in the electroohemical ~ystem, the hydrochloric aoid conoentration varied from 15 to 30%. It is necessary to note that temperature considerably influences the process of ohlorination. At the temperatures, higher than 80°C, the l o s ~of hydrogen chloride a?d decrease in graphite electrodes stability take plaoe. At the temperatures unds nowadays is a main supplier of

663

close t o the mom teinperature the yield dscraases apparently the hydrocarbon reactive due to the I-eduction of ability - chlorine is evaluated. The determining of the ratio of the amour!, of aromatic compound loaded into the cell to the aiiount of the chlorine electrochemically obtained, seems to be a matter of great importance. Special researches were held to study the influence of this ratio on the current and substance yieles. This ratio m a t be adjusted in such a way to prevent the chloripe appeanance the evaluated gases. On the other hand, a great swpl.uz of aromatic compounds is inudmissible for keeping of the chiorinatio:; products yield on the hi 1 level. It mas stated, that at the molar ratio of chlorine aromatic compomd (benzene, toluene, xylene and others) being eqmi to 1 :I, the chloriiiation process current yield is about 80%. Further increaGe in the aromatic compound concentration almost doesn': effect on the c'mrent efficiency, but refiuoes the substance yield. The influence c7f the current density on the pmcesses uncier consideration is ofl tmo kinds. D n the one hand, inorease ix the owrent density enhances the productivity of the oell. Dn the other hand, at the sharp incTease in the rate of electrochemical ohlorine evolution aromatic molecule have no enough time to bond it. Atmospheric emlssion of shlorine that takes plaoe r ~ s u l t ~in both decrease of the current efficiency and ;nviropent pollution. At the ou-rent density being equal 1-1,5 kA/m the current efficiency of the chlorinated alkjrlaromatic compounds was 90%, while that of the moncchlorbenzene was 70% When the teinparature is riaen, the yield of monochiorobencene decreases at the expense of diohlorobemene content in the mixture of ohlorahydrooarbons. Addition cf Cimethylformmide make6 it possible t o achizve essential increase i.1 monOChlGFObenZene yield. In this case the total current efficiency of the benzene ohlorination products is 90%, 98-99% substance yield, while the cnlorobansene current efficiency 75%. This effect might have been concerned with the enhancement of benzene solubility in the water phase. Purthermore dimethylfomainide depresses side processes, concerned with both electrochenical conversion of chlorine, f w m e d on the electrodes and formation of dichlorobenzene. A s far as chlorination of benzene derivatives to the Eide chain iE ooncerned, the flsctors that influence the shemical chlorination (W and more harct radiation, the presence of dir'fermt initiators of fpee radical6 fornation) favorably affect t!ie isolation of benzyl chloride, 0 - , m-, p-xylylchlorii-. The current efficiency is more than 85%. By now the pilot scale apparatim rated at cum?nf; 5008 has been designed. Tlie diaphrae in the electrolyzer is absent. aromatic Vith the help of t h k cell chlorination cf hydwcarbons to both nucleus and side chain was held with high current efficiency of the chlorination products ( 9 0 % ) . The posstbilitg of the waste hgdroohlorio acid using was examined on the above mentioned apparatus. That acid had been f c m e d in the prooess of epichlorhpirine p r c d u o t i m 2,113

?

.

664

contained ally1 chloride (30Omg/l) and dichloropropane (4Omg/l) as admixtures. The purpooe produot current efficiency was about 80%. The admixtures didn't influence the process of aromatic compounds ohlorination and was separated from the chlorination product6 by vacuum distillation after the synthesis. Both mathematical modeling and optimization of the process were made f o r aromatic compounds chlorination by the concentrated hydrochloric acid electrolysis. It was stated, that declination of experimental data from the rated values didn't exceed 2-10%. Ad60 tion and electrochemical behavior of the aromat'pc compounds on the solid electrode surl'ace

2.1.1.1

[24*25,27*29-351

Adsorption cf benzene and its derivatives on the smooth plathum eleotrode if?. slow. The coverage of the electrode surface by chemisorbed partioles, at different values of potential and bulk conoentrations linearly increase with the growth of time logarithm. The adsorption rate of both benzene and its derivatives can be well described by the Roginslcy Zeldovioh equation. The o w e of the dependence on the potential of platinun electrode stationary coverage by benzene and its derivatit-es chemisorbed particles have a dome like shape with the max-&num in the potential range of 0.2 0.4 V. The coverage decreases at the potential values higher than 0.4 and lower than 0 . 2 V . In the range of potentials lower, than 0,2V the adsorption value6 of benzene coincide with those of derivatives, while at the potentials, highaer than 0.4V. adsorption of benzzne excel the adsorption of its derivatives. The decrease in extent of surface coverage at the potentials higher than 0 . 0 is conoerned with both oxidation of aromatic eubstance and aborption opportunity of background electrolyte ar,ions. The experimental data suggest, that the surface coverage by chemisorbed particles of both benzene and its derivatives are just Einilar; but maximal value of pl3tinW surface coverage of benzene is more than that of its derivatives. The range of inaximal coverage gets narrow while turning f m m beazene t o toluene and xylenes. Furthermore, these data suggest that when benzene substituted oompounds are adsorbed on platinum, the substituents create additional steric difficulties that decrease the surface ooverage. Potentiodynarnic curveE were constructed f o r the oxidizing processes of benzene, chlorobenzene and dichlorobenzene xith the help of rotating disk electrode. At the range of potptials from 1.20 to 1.45V these curves pass the maximum (at 65 C). At more positive potentials one can see a current s l m p on the curvec; this fzot testifies to a hindering of electrooxidation prooess. Aocording to the experimental data, when increasing the rate of potential sweep from 0.01 to 0.04 V/seo, hindering of electrooxidizhg procees appears at more positive potential values. At rates of potential sweep being equal or more than 2.5 V/sec, current slump on the benzene oxidation curves disappears at a l l

...

665

(at 25Oc). As it is stated, the tempnrature growth oauses increase in the rate of benzene and benzylohloride oxidation, decrease in that of the dichlorobenzene, whereas the rate of ohlorobenzene electrooxidation almost does not depend on the temperature. The above data testify to the fact that temperature dependence of the electrooxidation rate is determined in gengral by the influence of the chlorine atoms, that substitute hydrogen atoms in the benzene ri-?g. It Euggests the temperature influence on the electrooxidating reaction of those organic ccmpounds to be of t.i;.o kinds: the tempepature growth increases the rate constant, but deci-eases the electrode surfaoe ccverage by the organic particles that undergo oxidation. Chlorine introduction t o the benzene rnoleoule causes not only the change of the temperature dependence character of the electrooxidation -i.eaction rate, but also essential decrease of maximal current f o r substanoe oxidizing. When plathum electrode is polarized from - 0 , O l to - 0.3 V at the presence of chlorobenzene and dichlorobenzene the cathode current increases synchronously with electrode rotation rate and temperature; it testifies to the electrochethese compoundE. In the case of mical reduction of benzene and beiizyl chloride reduction current considerably loTer than that of chlorobenzene and dichlorobenzene. In all the cases one can see a loop of hysteresis. The adsorpticn of benzene and its derivatives takes p l a c e on platinum in both chloride solutions and colutions free from chloride ions. Disthction is in the fact, that in the ohloride solutions, chemisorbed particles leave the surface due to r,ot only their oxidation, but also their chiorination forining different oompomds On the graphite eleotrode, in contrast to platinum, one can't note the potential delay on the ohronopotentiogrm d u r i n g neither anode nor cathode polarization. In all the potential range that have been researched ( 0 , O - 0,8 V ) the charging c u r ves in the preEence of benzene and its derivatives are higher than those in the oase of background solution. It testifie~ to considerable adsorption of benzene and its homologues as ?ell as t o verj low m t e of electrochemical oonverslon of studied organic compounds on the graphite electrode. In the case of graphite, in contrast to platinu!, there $3 absence of oxidation Eave of studied compounds on the potentiodynamic benzene, ourve6, corresponding to the !solutions of ohlorobenzene, dichlorobenzene, toluene, benzyl chloride and xylenes. Anodic potentiodynamic curves on graphite in the presence of above organic compounds is situated lowep than the background curve. It testifies to the slowness of the eiectroreduction and electrooxidation processes at the considerable adi3orption of these substances on the graphite electrode surfaoe On tie DSA ( O R ' P P type 1201) both charging curves and polarization curves obtained in the background solution a3 well as in the solution containing benzene and its derivatives a r e similar. It suggest, that studied organio compounds neither adsorb not undergo electr.mhemioa1conversion on ORTA.

.

666

T'ns data, mentioned above make it possible to conclude that plathum ia worthless a8 an eleotrode material f o r the prooerjses of aromatio oompounde chlorination (to the nuoleus as well as to the side o h a h ) be chlorine, obtained by hydroohlorio aoid electrolysis. Sinoe oxidation as well as reduotion besides chlorination of initial and f i n a l products take place on the platinum, the using of platinum electrode deoreases the selectivity of the chlorination process. It 2s advisable to uf3c grashite as an eleotrode material since in this case the lcsses caused by oxidation of initial and f i n a l products are almost absent. 2.1.2 Ethylene chlorination by hydrochloric acid electrolySis 36-583

Ethylene is known to be easily ohlorhated by free chlorine and it is this method that is used in the industry f o r dichloroethne production. It was this faot that had, prompted the of electroohemical utilization of waste possibility hydrochloric aoid in the course of electrolysis without diaphragm [321. This :echnique wa5 than patented in the VSA, Gemariy and Italy [50-521. The only waste of this procem, ethylene chlcrination uschlorine generated electrochemically, is hydro&en. Un3er the proper experimental conditions chlorine doesn't evoluted and explosive mixture doesn't form. Number of features make this technique to be coxpatible to the traditional ohernice1 method, they a r e : eaainem of ethylene chlorination, process independence on the presence of organic admixtures in waste hydrochlorio acid, ability to electrolyze at rather high current densities using graphite electrodes. To determine optinurn conditions of ethylene chlorination prooem in the cell without diaphragm at current up to 200 A , it waE studied the influence on the purpose product ourrent and substanoe yields of the number of factors, such as hydrochloric aoid concentration, temperatwe, current density and ratio of electrochemioalljr generated amount of chlorine tc the amount of ethylene, put into reaction. The influence of hydrochloric acid ooncentration appeared to be sknilar to that in the case of aromatic compounds chlorination in the electrochemical system. Whila working with aphite electrodes 5x1 27 - 3056 hydrochloric a d d solptions at g e range of current density values fron 1 o t o 4 kA/m , and at the range of temperatures from 35 to 80 C, the dominating product of ethylene chlgrination reaotion was stated to be I,2-dichloroethane. A t 35 C, the yield of ethylene chlorination products is 68430% at rather low current densities. $he increase in the temperature of chlorination process up to 50 C, at the same cilrrent densities cause the owth of the dichloroethane content hothe mixture up to 7 7 r - - 32%. Though ternpe7atw-e growth to 65 C leads to some increase in total mixtwe mass of ethylene chlorination prGdUCtS, however, the substance yield of 1,2-dichloroethane is only 75-80%, as lateral processes take plaoe. While tempeTature rises to 8OoC,

667

the yield of the pLzpose product in the mixture of ethylene chlorination products considerably decreasec. The same regulTity can be seen at rather high current densities ( 4 5 kA/m ) as well. Thus, while ethylene cqlorinating, at :he current 350 A (current density w a s 4,2 kA/m ) and at 80 - 85 C the current efficiency was 36% and 1.1,2-trichloroethane oontent in the mixture was about 5096. It was deteA-mLTed that range of temperatwes at which the maximum substance yield and current efficiency of 1,2-dichlorgethane (93% and 37% respectively) can be achieved i3 50 - 60 C (ht current density 4 kA/m ). Further temperature growth leads to a decrease in the substance yield of purpose px-oduct due to the formation of deep chlorinated products. The increasing of current density from 'I to 4 kA/mZ c a m e 5 considerable grGwth of the I ,2-dichlo?oethane substance yield as well as current efficiency at all theotemperatwes. However, f u r t h e r growth of ourrent density at 50 C, the composition of ethylene chlorkation products changes considerably. A number high boiling points appears whereas 1,2of compounds, ha7dichloroethane sontent decreases. It was found, that the optimu! conditions, that correspond to maximal substance yield of the pwpose product (90-9376) a@ current efficiency (95-9796) are: current density 3-4 kA/m , temperature 50-60 C. hydrochloric acid concentration 15-30s. It is advisable to hold t h e electrolysis at moisr ratio of electrogenerated chlorine: ethylene being equal to 1:1,05. The ethylene chlorination was held in the pilot scalc ci=li without diaphragri. Cell was rsted at the ourrent 7 ~ pto 530 A . 1,P-Dichlorcethane mas sjrllthesized electrochemioally from both t e c h n i w l and waste hydrochloric acid - the waste of the epiohlorhydrbe production prooesses. The chlorination was held in the presenoe (3,O - 5,0 weight per cent per Teactiom mass) of iron trichloride. Addition of iron trichloride to the electrolyte enabled to hold, the electroly~is at rather high current densitiqs - 1 1 U / m [36,56--581. h e cannot note the atmospheric emission of chlorine under such conditions. The 1,2-dichlorcetham content ir, the mixtwe of chlorohydrocarbons exceed 37%. The totai material balance of the dichloroethaqe isolation procecs was dravm, comparative data o n cost prices o f 1 ton of dichloroethane have been calculated. Taking i n t n account the costs of mea~-mes,necessary f o r the environment preserving from the waste hydrochloric acid, Sichloroethane cost price appears to be considerably loner than that of chemically synthesized product. The process of dichloroethane manuf'actixe was held on the industrial scale cell of filter-press type by electrolysis of waste hydrochloric acid with current up t o 25 kA. For holding the reaction of ethylene chlorination by electrolysis of concentrated hydrochloric acid, the cell designed by firm "Feibzal Leipzig" (DDR) for the hydrochloric acid eleotrolysis, was reoonstructed, only one cell section being m e d . That is the right-angled section with woPk% voluiie 85 liters, which consisted of cathode pnd anode graphite plates (eacn hay-ing the working swfaoe 3.2 rn ) Cell is divi-

.

668

ded by diaphragm made of ftorolone fabric. 3iaphragn is located between rubber lists of aofd-proof washers that serve as space paokings between anode and cathode blocks. Electrodes are connected to form a comnon filter-preSE constmctiori with the help of ,oummed electrohsulation material, pressing steel frame and bolts. To facilitate the gas mixture leaving of reactionary zone, there are ledges Gn the working surfaces of anode and cathcde. Through all the bottom wide of anode there is a groove (260 m high) f o r layirg and I'astenirj of a polyprcjpylene chute with 323 holes, of 1 n! dimetsr. Ethylene gets to the ohute through the polypropylene tube. The holes in the chute enable t o provide even ethylene distribution on the graphite electrode surlaca.

There 2re lit-clegrooves in the top of the electrode plates. They serve f o r initial and final prociuots to leave of both anode a d czthode chambers. The products of ethylene chlorkation as a vapor are sent to condense via carbine, wherezs the wastes of hydrochloric acid (with omcentration up to 15%) nixed Kith the hydrogen, fomed on the csthode is sent to the gas draifi via cwbine. Filter-pr-ess,?ell is fastened on the supportiq gummed frame anc! k s u l s t o = . s . Technological sahenie of I,2-~iichloroethanesyztheses at the indust,ry-experimntal cell of filter-press type by electrolysis of waste hydrochloric acid is given in the Fig.4. 27-30% waste hydrochlorio acid is sent from the tank ( 3 ) to the cathock spaos of cell ( 1 ) via q-aphite heatexchanger ( 4 ) , where it is heated by steam to 35-49 C. Than, depencting on the teEperature changes in the cell [I) , temperature of fed, hydrOChlGriC acid changes in accordance with t h e given olf heat. Zthylene is sent to the cell via spacial device [distributive vcllve ( 1 1 ) and ratatorneter (811 during 3-5 minutes. The leeding of ethylene is regulated (with 2-376 exceed over stochicmetry). Only after this the power supply is switched on. The reaction temperature is kept owing to the heating or c o o l h g of the hydrochloric acid in the heatexcbzznger ( 4 ) . Gas mixture, formed as a result of electrolysis, oonsists of ethylene chlorirlation products and ethylene exoeed. The gas mixture gets from anode spacg to thi. vertical graphite heatesohanger ( 5 ) having 15 m of heat exohaie surface. Dichloroethane and other products of ethylene chlorination condense in the heatexchanger ( 5 ) and than get to the collector ( 7 ) f o r raw storage. Uncondensed part of the procucts that gets to the raversal condenser (61, where further condensation of ethylene ohlorination products takes plaoe. Condensation in the graphite heatexchvlgers ( 5 ) and ( 6 ) as well as in the collector (7) is made by the brlke of calcium C h l O r k k coGled to -1 5=C. After oondensatton of ethylene chlorkation produots, the ethylene exceed gets from the heatexchanger ( ' 7 ) via absorber ( 9 ) (being purified from the vapor of hy-ogen chloride there) baok to the cell ( 1 ) through the distributive valve.

669

The reacly product of ethylene chlorination - ram I-2-dichloroethane gets f o r further treatment and realization. the Pjdrochloric acid Electrolysis is held until concentraticn reaches the 15% value. After that the mixture of gas and liquid, that contains hydrogen and unreacted H C 1 gets via CatizGde space to the waste-dilator ( 2 ) . In the latter the unreacted ECl is separated from the hydrogen arid gets to the tank (10). When the rate of ethylene feeding is decreased. and waste HCL conoeritration is lower than an optimal value, the automatic indicztors switzh on and electrolysis stops as rectiZiers switch off. The obtained raw material was chrornatographicaly analyzec! and it was stated, that mass portion of 1,2-dichloroethane is 97X, while that of ?,1,2-trichloroethane - 3%. Current efficience and substance yield of 1,2-dichloroetha?e a r e equal correspond ly to 90 a d 93%. Above men toned results have detected the officlenoy of industry scale cell of filter-press type for the obtaining of 1 ,2-dichloroethm-e by electrolysis of waste hycbochloric acid as well as advisability of this method use in the industry to of chloroorganic industry utilize the great amount waste -hydrochloi-io aci.6.

?

2.2.

Utilization of diluted hydrochlorlc acid.

The ohapter 2.1. was concerned with the discussion of works, aLied at the utilization of concentrated (1 5-30%3 hydrochloric aoid. Whereas n m b e r of plants have more diluted hydcnochloric acid solutions (2-10%) as wastes. The utilization of these solutior~s is rather difficult problem too, while it is expensive to ccncentrate them. Thus it seems to be advisable to work out the tzohniques of diluted hydrochloric m i d utilization by means of electrolysis in the presence of organic CorripOlinds. Electrochemical processes, taking place in the diluted solutiJns of hyitrochloric acid considerably difI'er from those in the concentrated solutions [23. In particuiar, as almost the only process -k the concentrated solutions is chlorice evolution, graphite electrodcs a r e stable enough there. But while diluting these solutions (startirg with approximately 12%) oxygenous o:irbon compounds appear ammg the anode products and graphite anodes become corrode. Eleotrolysis of hydrochloric acid diluted solutions in the presence of olephlnes result3 in formation of chla-oalcofiols besides ~chlorides. This process of chlorhydroxilation appears to be enhanced when the concenx-atiori of hydrochloric acid decreases and at the concentrations lower than 7% the chloroba*oxilation becomes almost the only process in the oase of determined opthum conditions. Thus, it seem to be interesting t o determine the olephines chlorFkvcb*oxllatim ability in the diluted hydroohloric acid 159-89 1

.

670

The following unsaturated cornpounds served research:

%=CHC#5;

q-rCK-CH=C%;

q=CH-CH=CH%;

E ~ E objects

CH$XC

of

(%

r'rGL=cHcH2c1; c€i2=cc1cH2c1; ~ C C l = C H C ~ C l ;( X p H ~ O H The presence of electrondonor and electronacceptor functional substituents near the double-bond enable to study the influence of steric and electrostatic factors on the reaction ability of olephines bi the chlorohydroxilating reaotions as well as on the yield of corresponding chlorohydrines and other chlorinated hydrocarbons. The reaction of ohlorohydrhe formation by electrolyBis of diluted hydrochloric acid in the cell without diaphragm in the presecce of' msaturated compounds oan be presented by a oommon scheme : CI

2 >C=C%

t 2HC1 t 2%0

+ 4F -1

2H2

Undoubtedly the meohanisrn of above reaction is considerably affected by the processes of both olephines and their chlcrohydrines electrochemical oxidation 01: the anode. Furthermore, the above compounds might have taken part in the reducing reactions OP, the cathode. Thus, it would be interesting to study eleotrochernical and adsorption behavior of olephines and corresponding chlorhydrines on the eleotrodes.

2.2.1 Electrochemical and adsorption behavior of ~ n B a t W 8 t e d corn ounda and their chlorhydrines on the electrodes 166.7'1, 77* g4.87-89 3

Electrochernioal and adsorption behavior of misaturated compounds and corresponding ohlorohydrines were studied by the method of construction ol" both chronopotentiogram and n the solutions of 0,l N mlphuric m i d potentiodynmic curves i and, 0,l N potassium ohloride on the platinum and graphite electrodes. A s organic compounds were chosen: propylene, ally1 chloride, divinyl and corresponding chlorohydrines: propylenechlorol.y~ine (PCH ) , glyoerol dichlorohydrine (CDC ) dichlorobutanediol (DCBC) The potential delay, that is 6een on the anode galvanostatic

.

67 1

curves, obtained in sulphuris scid on platinum at the presence a11 the abrJve mentioned organic compounds, testifies t o their oxidation. Chronopotentiogrsms on the graphite anode vier8 obtained iLn the preserice of propylene, divinyl and allyl chlorida. Charging curves on graphite differ radically from those on plat-hum. Chronopotentiograms on graphite in she presence of sulphuric m i d have shorn, that there is no any delay cn graphite in contrast t a p l z t i n m , that is polarized by the sane current density in the presence of Gleph-hes. On graphite corresponding charging curve is considerably higher than that of background soluticn. Such a result suggests that, firstly, mentioned olephines hdsorhy on the graphite and, secondly, the rate of their oxidation is much more on platinum than o n gaphite. Curves in the pesence ~f olephines are sitmted lower than those of the background solution qnlg in the o8sc of very low (1 < 10-JA/cn 1. Whereas on platinum the current densities d e l q is well e e m even at the highest current densities (i=0.6 A/am ) . All the above data suggest that true oxidation rate of studied olephines on platinum is, in any oase,three orders more than that on qaghite. T h m oxidatitjn of propylene, diyinyl a d ally1 chloride on the graphite anzde takes plage oQy at the polarization by very low ourrent densities (i=13 A/cm ), that are almost two orders lower, than those of PCH, GDC, DCBD synthesis and one car, disregzrd these processes under the synthesis conditions. Electrochemical behavior of PCB, GDC and D2UD on platinum and graphite electrodes was researclied as well in the presence of 0 , l N sulpliurio acid. These studies have shown that mentioned chloro’hydrines give delays, that can be well seen on platimrn. FotentLals G f these delays a m c l o s e to those of oleFhines oxidation chronopotentiogr2rr.s. The dependencies of adsorption of propjrlene, divinyl, ally1 chloride, PCH, CCC a d DCBD iFig.5) on potential of platinum eleatrode x m e cmstructed on the base of chronGpotentiornetric curves. It is well seen, that all the curves i r e of d m e like shape having a m s x i ~ min the range of potentials from -0,l to 0.2 V. At the potentisls, more positive than 0,6 V or close to the potential of reversible hywogen electrode, adsorption of olephines doesn’t take place. Dependence cwves of chlorohydrine adsorptioa on platimn testify to good adsorption of these substances even at low concentrations, though the range of chiorohydrine adsorption almost coincide with range of olephines adsorption. DCBD is r considerable larger amounts and its adsorbed on plat,inum i adsorption range much wider than that of PCH and SDC. The sharp rise of adsorption must be caused by increase i?n number of hydroxyl groups in the DCBD moleoule, and not by the presence of two chlorine atoms, sinos GDC molecule contains the same number of chlorine atoms. data were oonfimcd by the Chronopotentiometric potectiometric ourves, obtained with the help of rotating disk platinum and grziphite electrodes. A s it is well seen from the m o d e potantiodynanic curves on plathum electrode, there are of

672

maxima in the presence of div-1 a d propylene, - it can be eqlained by their oxidation. The presence of chloride ions in the solution moves the wave of orgvlic compounds oxidation in positive potentials direction. A current slump in the range of potentials E = 0,4 - 0,55 V might testify t o the su&estion about electrode passivation by adsorbing produots that take part in the eleotrochemical reacticn. The wave height in the presence of olephines is independent on the rotation rate, that suggests almost total absence of diffusion limitations. In contrast t o olephines, on platinuni chlorohydrines are oxidized at lower rate under potentiodynamic conditions. Under potentiodynmic conditions propylene, divinyl aid ally1 chloride are oxidized easily on both graphite and platinum. In contrast to them, PCH, GDC and DCBD cannot be oxidized on graphite electrode. Cathode potentiodynamio curves on plat hum and graphite testify to the proceeding of olephines reduction processes at potentials that are less negative thzn the potential of hydrogen evolution on the mentioned electrodes. That is w h y at the Dotential values, close to that of processes of elect6oohemical chlorohydroxilation of olephines one oan ignore the olephines 03 reactions both on the graphite and on the platinum eleotyodes as their rate seems to be rather negligible. As for the possible chlorohydrines reduction cn graphite and platinum electrodes, it appeared to be impossible to obtain the curve, that appreciably differs from the curve of background in the solutions of PCH m d CCBD neither on graphite nor on plathum electrode. Thus the studying of electrode prooesses has shorn that at the potential values close to the chlorine evolution potential, adsorption of olephines considerably deoreases, it suggests that the majority of chlorohydrines forms in the electrolyte volume according to the ohemical mechanism. The data, presented above confirm an obviow. advantage of graphite electrode over platinum while synthesizing chlorohydrines. In this case an ability of side electrachemical conversions of both initial and final products is almost absent.

Chlorohydroxylation of unsaturated co om& by the electrolysis of diluted hydmchloric acid I59- 0, 72-76, 78-

2.2.2

03, 85, 861

?

Xonooiephines C3-G7, that have electrodonor substituents near double bond turn to a- and p-ohlorohydrines at hydrochloric acid electrolysis in cell without diaphragm. CorTelation of isomers in the mixture depends on the size of substituents. So in the case of propylene, p-isomer content is lo%, in the case of hexen-l and hepten-1 it is only 2-3%, and a6 for isobutylene and styrene, - this isomer is almost absent. While ohlorohjdroxylating of nonoolephines , oorresponding dichlorides appear besides chlorohydrines. The yield of fonners depends, in general, on the ohloride ions concentration. It should be noted, that k the case of olephines, ha7;Yg

613

iso-structures, besides mentioned reactions, the reaction of substitutional chlorination takes place to a great extent. So, while chlorohydroxylating of isobutylene one can note an addit ional formation of 3-chloro-2-methylpropylene-I Thus the main products of isobutylene chlorohydroxylating reaction are I-chloro-2-methylpropanol-2 and 1,3-dichloro-2methylpropanol-2 (weight ratio is 7 0 : 3 0 ) . Possible mechz~nism of chlorohydrine formation in solution is chemical interaction of reacting substances via chloronium complex. Another evident of chlorohydrines primary formation in the electrolyte volume is the reaction of low-temperature oxidizing olephine chlorohydroxylation in the hydrochloric acid applying hydrogen peroxide instead of electrolysis; final products, as f o r their chemical composition, don't differ from the products obtained electrochemioally. Conjugated diolephines C4-C5: divinyl, isopren and piperilen under the conditions of electrochemical chlorohydroxylating processes seem to have a high reactive ability. It is stated, that both double bonds of mentioned compounds can easily be chlorinated in both 1,2- and 1,4- positions. A s a result chlprohydrines evaporated when boiling in the range from 95 tG 125°C (at 3 mm of Hg) appear. A n intengive wide band at the is well seen on the frequencies range from 3200 to 3600 cm products IR spectra. This band must be caused by primary and secondary hydroxyl groups. Such wide range of boiling temperatures confirms, that final products consist of isomeric mixture of chlorohydrines. In particular chlorohydroxilating of divinyl in hydrochloric acid solution gives 1,4,-dichlorobutanediol-I , 3 (24%) and 3,4-dichlorobutandio1 (25%). Let's ,suppose, that reaction of 3-chloropropen-I chlorohydroxylating with electrogenerated chlorine proceeds via stage of chloronic complex, that transform to a core stable form ( A ) .

.

. b

.c1. 4-

.c1. :+ -. $ X + HC Z YC .

e

**

C%.ZXH-C%Cl

A

*

(A 1

*.+: ..

c1

In this case the reasons of glycerol dichlorohydrine isomers formation can be easily explained according to the scheme:

2 A t 250

-

r

HOC%CHClCH$l

-

+

2Ht

Cl%CH(OH)CH$l

Since the water has greater possibility to open chloronio complex from the side of methylene groups, one can expect preferential formation of compound with primary hydroxyl group. Indeed, the ratio o f 2,3-dichloropropanol-1 to l,3-dichloropro-

674

is equal to 65:35. Side produots of reaction of 3-ohloropropen-I chlorohydroxylation are: 1,2,3-triohloropropane and tetraohlorodipropyl (isopropyl) ether. Usually, the yield of former rise in connection with horease in ohloride-ions ooncentration, whereas the yield of latter - with n ohlorohydrines ooncentration in the reaothg increase i mixture It should be emphasized, that the reaction of 3-ohloropropen-I ohlGrohydroxylation was held a150 by low-temperature oxidizing method in the presenoe of hydrogen peroxide and h&xirochloric aoid. As f o r the ohemical isomerio content of the products obtained by low-temperature oxidizing method, it was almost similar to that, electroohemioally obtained. These results testify to the ohemioal formation of above oompounds in the eleotrolyte volume. a-chlorsubstituted olephkes also have reactive ability under the conditions of ohlorohydroxylation reaction k the prooess of the electrolysis of diluted hydroohlorio aoid. 2,?diohloropropene-7 and 2,3-diohlorobutene-2 while olorohydroxilating give chlorine-containing ketones besides corresponding chlorohydrines. Such effect can be explained by the formation of unstable oomplexes: paiol-2

.

+2cL-

Cl~CHCl~Cl 7 tZ13-GDC

’ cl~cHclcH20~mclcH2cl + 2H+

+ ,3-GDC I

+

’ (c1cH2)*cHocH2cHc1~c1 + 2H

Eleotrochemical ohlorohydroxilation of 1,3-diohlorobutene-2 results in the formation of chlorine-substituted ketone -1,:dichlorobutanon-3 (substance yield is 85-90%). The application of electrochemical chlorohydroxylating reaotion in the diluted hydrochloric acid solutions to the ally1 alcohol enabled to obtain glyoerol monochlorohydrine with a high yield. Obtained ohlorohydrhes serve as raw material f o r synthesis of corresponding epoxide oompounds acoording to the known methods; the yields acrhieve 85-904e. The results of studied olephineer ohlorhydroxilation are suwned up in the Table 1. It is well seen from the Table, that the current efficienoy of chlorohydrines oorresponding to monoaid diolephines C3-C5 is 76-80% on graphite, 72-78% on platinum and 78434% on DSA. Whereas at the electrolysis of olephines cg-c~,the yield of chlorohy&.ines is only 56-62% on graphite and 60-65s on the DSA. A 6 f o r the rest of the current, it is spent, 5n general, to form oorrespondbg diohlorides. Technological soheme for eleotrosynthesis of 1,2-dichloroethane was slightly modified and than applied to the eleotrosynthesis of PCH and GDC on the pilot oell.

675

Electrosynthesis k7as held at current 5,5 kA until the hydrochloric acid concentration becomes 3-4%. Row solution of PCH contain% 60 g/l of ohlomhydine (the c-ment efficiency is 80% and substance y k l d 92%); and solution of GDC oontaining 50 g/l of chlorohydrine (79% and 90%). The obtained products were m e d f o r production of oorresponding olryranea.

Table 1 !Rie inflwnce of eleotrode material on ourrent effioienoy of chlorohydrins. The hypchloric acid concentrztio?. 2.8 M, current density 10 A/& , temperature 5OoC Current efficienoy, % Glephinee

Chlc,mhydrins Graphite Flathe DSA

80 PropyZene Prop3lcnechlorohydrine Isobutylene 1 -Chloro-2-methylpro76 panole-2 and 1 ,3-dichloro-2-Ire thylpropanole-2 Hexenc-l I-Chiorohexanol-2 59 Iieptene-I 1 -Chloroheptm-ol-2 56 3-Chloro- Glycerol dichlorohyCrfne 78 propens-1 Divinyl I35chlorobutanedio 1s ao Piperilme Ciciiioropentanediols 78 Isoprene Dichloroisopentanediols 80 S t irene 1 -Pheny1-2-chloroetha62 nol-7 Ailyl Glycerol monochloroa5 hydrine a1coho1

7

2

3 4 5

6 7

79

82 80

74

65 6C 80 76

84 65

75 -

63 82

Scvbell RF. Bwneg Iis, Journal of the Electrochemical Society 7390; 7 137; 40: 485-503. Tedoradze Gd, Aver'yanova NM. Eleotrochemical Sgfithesis of Chloroorganio Compounds. Moscom: Nauka, 1387; 'I 81 . Hoelemann N. Chemie-I~enieur-Tec~ik 1362; E d 34; 5: 381-376. Janson HG Ct.,emie-IrbPenieur-Technik 1967; Bcl 33; 72: 729-734. Korczynski A , Dglewska J. Zeszity Na*owe Politecknki Slaskiej. Seriz: Chiria. 1973; V 65; 393: 165-175. FRG Offenlcgmgsschrift N 2856882 (C1. C25E 1 / 2 6 ] Agpl. 30.1 2. 'I 378 PJbl. 05.07.1 979. Dempsey RY, L3Conti AB Gezeral Electric: Co. USA Pat. N C247376 (C1. C25B 1 / 2 6 ) 27.07.1381/ App1.431

.

676

8

9 10 11

12

13

14 15 16 17 18

19 20 21 22 23.

24 25 26

27 28

02.01.1979. Dempsey RLI, LaConti AB. General Electric Go. Uehara I, K o w a n i Y, Wakabayashi N, Motone M, Takenaka H. Denki Kagaku 1990; V 58; 4: 360-367. Uehara I, Komani Y, Wakabayashi N, Fdotone M , Takenaka H. Denki Kagaku 1990; V 58; 5 : 459-464. FRG Offenlegungsschrift N 2844499 (C1. C25B 1/24). APPl. 12.10.1978. Publ. 13.06.1979. Laconti AB, Dempsey RM, Coker TH. General Eleotrio Co. USA Pat. N 4287032 (Cl.C25B 1/24) 01.09.198?. Appl. 121414 14.02.1980. Pellegri A. Oronzio de Nora Implant1 Elettrcchimici S.p.A. USA Pat. N 4311568 (C1. C25B 1/25) 19.01.1982. Appl. 136633 02.04.7980. Balko EN. General Electrio Go. USA Pat. N 4294671 (GI. C25B/ 1/26) 13.T0.1981. Appl. 14961 9. 1 4.05.1 980. Balko EN. General Eleotric Go. UK Patent Appl. GB N 2073251 (C1. C25/B 1/26) 09.03.1981. N 136603. Publ. 14.10.81. Balko EN. General Electric Co. Japan kokai tokkyo koho, 57/194284. 29.11.19E2. CA, 98 :I 3 4 3 6 6 ~ . Yalcimenko LM. Production of Chlorine, Caustic Sods and Anorganic Chloroproducts. Moscow: Khimiya, 1974; 600. Furuya N, Suzuki Y, Shirai T. Denki Kagaku 1989; V 57; 4: 332-333. Stockmanne. ldetallooberflaeche angewandte Elektrochemie 1973; Bd 27; 9 : 320-322. KuiyaEIiova AS, rdvrozova AS. J Khimiuhaskaya PPGr1-t~6hlmnOSt ' Za Rubezhom (Chemioal Industry Abroad) 1985; 1 (2651: 26-35. Yakirnenko LM. Eleotrode Materials in Applied Electrochemis t r y . Mosoow: Khimiya, 1977; 264. Zimin Wbf, Kamaryan GM, Mazanko AF. Electrolysers f o r ChloI-ine Production. Moscow: Khimiya, 1984; 304. Yubekov Y U . Tedoradze GA, Atamov MA!, Babaev WA. In: Sumgaite Scientiific-Teohnioal Conference on Organio arld Chloroorganio Synthesis.Abstracts. Baku: ELM., 1980; 40-41. SU Pat. N 857099 (C1.C 07G 21/24) 23.08.81. Appl. 2855?06/ /23-04; 18.12.1979. Yuzbekov Yuk, fedoradze GA, Guseinov MM, Atamov MM, Gorokhova LT. Institute of Electroohemistry, Academy of Soiences of t h e USSR; Institute of Chloroorgmic Synthesis, Academy of Sciences of Azerbaiclzhan. SSR. Atamov W, Maksimov KhA, Yuzbekov YuA. In: 6 t h All-Union CGnferenoe on Eleotrochemistry. Abstracts. Moscow: Naulra, 1982; 2 : 190. Yuzbekov YuA, Atamov Mhf, Maksimov KhA, Safarova MI. J herbaidzhanskii Khhnicheskii Zhurnal 1984; 4: 59-63. Atamov MM, Yuzbelcov YuA, Borisov SN, Sultanova ZIP. J Azerbaidzhanskii Khimicheskii Zhurnal 1984; 6: 24-28. Ashurov DA, Hakshov KhA, Tedoradze GA, Gorokhova LT, Yuzbekov YuA. J Eleotrokhimiya 1984; V 20; 5: 600-603. Tedoradze GA, Yuzbekov YuA, Babaev NB. In: Modern Status and PersDectives of DeVelODment of Theoretical Basises Manufact-wirgof Chlororgah Products Baku: E l i , 1985 ; 218-219. Maksimov KkA, Btamov W, Yuzbekov YuA, Rotenberg Zk. Depos i t a d VINITI. MOSCOY~ 19.02.1985; 1266-85: 21. Vasiliev YuB, Kazarimv VE, Andreev VN. Journal of E l e c t -

.

29

30

~

677

31 32 7 -J> 13

34

35 36 37

38

39

40 41

42

43

L-oanalgtical Chemistry and Iriterfacial Elcotrochemistry 1983; V 147: 1/2: 247-261. Grinberg VA, Zhdravleva VN, Vasiliev Y I B , Kazar?ncsv VE, J ElektroMilmiya 1983; V 19: 10: 1447. Horanyi G, hdreev VN, Kazarinov I-. J Electrckhimia 1985 ; V 21: 10: 1363-1366. Shchulrin IVY Kazarinov VE, Vhsiliev YuB, Babkin VA. J E l e ktrokhbiya 1985; V 21: 5: 725-732. KazarinGv VE, Horanyi 6 , Vasiliev YUB, Ancireev 3%. In: Itogi Nadci i Tekhnyiki, Seria: El&tmkhimiys. VINITI: MCSCGW, 198s; v 22; 97-140. 'xzzarL?ov VE, Maksimov W. J Elektrokhimiya 1984; V 20: 6: 782-789.

Atmov ME, Y-abekovYuA, Guliava XA. In: News cf Electrochemistry of Organic Substances. L'VGV: State University Publication, 1986; 206. SU Pat. N 652923 (C1.C 07C 2 5 / 0 6 ] 30.06.1978. Bppl. 23664561 /23-04. 27.05.1976. Ashurov DA, Abdullaava TM, Salakhov MS, Guseinov I@&, Uamedaliev YuG. Institute of Petrochemical Prozesses. SU Pat. N 88.4263 (C1.C 07C 19/02) 23.11.1954. Appl. 2565101/23-04. 24.01.1978. Tedoradze GA, Ponornare,&o EX, SOliOlOV Y W . Institute of Electrochemistry, Academy of Sciences of the USSR. SU Pat. N 900568 (C1. C 07C 19/02) 23.11.84. Appl. 2378718/ 23-04. 01.0701976. Tsdoradze GA, Paprotskaya VL, (rornilov AP. Institute of Electroohemistry, Academy of Scienoes Qf the USSR. TedoFadze G A Y Papmtskaya VA, Yuzbekov Y U , Vlzdbirova NB. In: D.I.Nend2leev All-Union Chemicai Szci?ty, 3 t h Conference. Abstracts. Moscow: 1976; 8. Yuzkekov YUa, Tedoradze G A Y Paprotskaya VA, Sokolov YuM, e t al. In: Ccntributions in a Synthesis of P o l y x e r and Monomer Materials ( i n Russian). Baku: EW, 1977; 35-38. Tedorzdza GA, Yuzbekov YuA, Fonornarenko FA. In: Modern Status and Persaectives of Development of Theoretical Easises MmufactwiYLg of Chlororganic Products. Baku ZLN, 7978; 7 34-1 44. Yubekov YuA, Tedoradze GA, Sokolov YuK, Ponornarenko El, e t al. J Azerbaidzhanskii Khimicheskii Zhurnzl 1979: 1: 143-1 5 2 .

Tedoradze GA, AS?U~OV DA, PoncmarenkG EA, Tonilov Ap, et al. In: Feoktistov LG, ed. Electrosynthesis cf Xonomers (.Ln R u s s i a n ) . Moscorv: Nauka, 1980; 209-219. 45 Yuzbekov YLLA, Tedoradze G A Y Sokolov YLLM, Eabzsv NE. et al. J Azerbaiclzhtinskii Khimicheskii Zhurnal 1981 : 3 : 1111-1 14. 46 Tedora&e GA, kshurov DA, Yuzbekov YuA, Babsev ID. In: 12th Mendeleev Congress on General and Applied Chemistry. Extended Abstracts (inRussian). Baku-Moscow: NTiulia, 1931; V 3 ; 299-300. 47 Tedoradze GA, A s h m v DA,Yuzbekov YUB, Mazanko AF, et al. In: 6th All-Union Conference on Electrochemistry.Abstract~. MOSCOW: N3uka, 1982; 183-184. 48 Tedcradze GA, Aver'yanova NM. In: Tomilov AP ed. Progress in Electroohcmical Synthesis of Organio Compcunds. 44

618

49 50 51

52

53 54

55 56

57

58

59 60 61 62

Electrosynthesis. Electrochemical Reactions with Participation of Organic Compounds. MOSCGW: Nauka, 1990; 87-100. Yusifov VG, Mustafaev TKh, Bairamov DN, Yuzbekov YuA, et al.In: News of Electrochemistry of Organic Substances. L'vov: State University Publication, 1986; 267. Yuzbekov YuA, Babaev NB, Sultanova ZT, Borisov SN. J Aaerbaidzhanskii Khimicheskii Zhurnal 1987; 1: 119-122. Babaev NB, Yuzbekov YuA. Izvestya Akademii Nauk Azerbaidjanskoy SSR, Seria Ekonomioheskaya (Proceedings of Azerbaidjan SSR Academy of Sciences, Series of Econcmics) (in R u s sian) 1990; 1 : 35-44. Yuzbekov YuA, Ashurav DA. In: News of Eleotrochemietry of Organic Substances. K a r a g a n d a , 1990; 55-56. Yuzbekov YuA, Babaev NB, Tomilov AP. J Khimicheskaya Promyshlennost (Chemical Indmtry) (in Russian), 1391 ; 1 :14-1 7. Tedoradze GA, Paprotskaya VA, Yuzbekov Yuk. In: News of Electrochemi3try of Organic Compounds. 9th All-Union Conferenoe on Electroohemietry of Organic Conpounds. Abstracts. Tula-Mosoow, Publ. of Institute of Electrochemistry of Aoademy of Soiences of the USSR, 1976; 22. Babaev NB. Electroohemical Chlorination of Ethylene in Hydrochloric Aoid Medium. Thesis of Candidate of Technical Sciences. Xoscow, 1990; 20. USA Pat. rJ 4334967 (Cl. C25B 3/06) 15.01.1982. Appl. PCT/ /Su79/00001. 23.01.1979. Tedoradze GA, Paprotshaya VA, Tomilov La,Sokolov Y W . Institute of Eleotrochemistry of the Aoademy of Soiences of the USSR. Italy Pat. N 1110283 (Cl. C25B 3/06) 23.12.1985. Appl.PCT/ /Su79/00007. 21.01.1979. ledoradze GA, Paprotskaya VA, Tomilov AP, Bokolov YUK. Institute of Electrochemistry of the Academy of Soiences of the USSR. FRG Pat. N 2953388 (Cl. C25B 3/06) 24.07.1986. Appl. PCT/ /Su79/'00001. 21.01.1979. Tedoradze GA, Faprotskaya VA, '20milov A€', Sol:olov Y W . Institute of Electrochemistry of the Acadeny of Sciences of the USSR. Ashurov DA, Akhmedov AN, Alumyan Zka, Sadykh-Zade SI. J Zhurnal Obshchei Khhnii (Journal of General Cherniatry) (inRussian) Leningrad, 1974; V 44; 8: 1842. Ashurov DA, Kyazirnov ShK, Akhmedov AM. J Azerbaidjanskii Kbimioheskii Zhurnal 1975: 5 : 58-60. Ashurov DA, I(yazimov ShK, Alumyan ZhR, Alihnedov AM. J Elektrokhimiya 1975; V 11: 12: 1901. A E ~ U ~ ODA, V Alumyan ZhR, Sadykh-Zade SI. J Elcktrokhhiya 1977: V 13: 12: 1898. Deposited VINITI 28.06.1977: 2596-77:

63

7.

Ashurov CA, tilumyan ZhR. J Zhurnal Obshchei Khimii (Journal of General Chemistry) 1978; Leningrad; V 48: 1: 166-1 69.

Ashurov DA, Alrhmedov Add, Sadykh-Zade SI, Tverdokhleb LP, Tedoradze GA. J Zhurnal Prikladnoi Khimii (Journal of Applied Chemistry) 1978; Leningrad; V 51: 2: 459-461. 65 Kyazimov ShK, Ashurov DA, Bairamov DN, Tverdokhleb I9. 2 Elektrokhimiya 1976; V 12: 8: 1342-1343. 66 Ashwov DA, Sairamov DN, Tverdokhleb LP. J Azerbaidzhanskii Khhicheskii Zhurnal 1977: 4: 63-66. 64

679

67 Ashurov DA, Alumyan ZhR. J Azerbaidzhanskii Khimicheskii Zhurnal 1978: 1 : 65-68. 68 Ashurov DA. J Zhurnal Clbshcheii Khimii (Journal of General Chemist-ry) 1980; Leningrad; V 50: 4: 953-954. 69 Teaoradze GA, Agaev NM, Yusifov VG, BaiTamov Lp, Kyazimuv ShX, Ashurov DA, Yuzbekov YuA. J Elektrokhimiya 1979:V 15: 3: 396-399. 70 Ashurov DA, Tverdokhleb LP. J BzerbaidzhanRkii mimicheskii Zhimal 1380: 2: 104-107. 71 Ashurov DA, Akhnedov Ab¶, Bairamov DN. J Azerbaidzhanskii Khirnicheskii Zhurnal 1980: 5 : 102-1 06. 72 Ashurov DA, Mamedov ZF. J Azerbaidzhanskii Khimicheskti Zhurnal 1981: 3: 115-178. 73 Aahurov DA, Shchegol IE, Kharitonova LN, Tedcradze GA. J Z h m a l Prikladnoi Khimii (Journal of Applied Chemistry) 1982; Leningrad: V 55: 4: 868-870. 74 Ashurov DA, Alumyan Z h i i , B a i m o v DN. J Axerbaidzhanskii Khimicheskii Zhurnai 1982: 3: 48-51. 75 AGhurov DA, Sharipova DV, Shafieva LG. J AzerSaidzhandlcii Khimicheskil Zhurnal i983 : 5 : 106-1 09. 76 Ashurov DA, Babaev NB, Kharitonova IN, Tedorsctze GA. J Zhurnal Prikladnoi Khimii (Journal of Applied Chemistry) 1383; Leningrad; V.56: 10: 2379-2381. 77 Ashurov DA, 'Pagiev AG, Babaev NB. J Azerbaidz!~~skiiK h i micheskii Zhculnal 1983: 4: 41-44. 78 SU Pat.N 1087595 (Cl. C07C 31/34) 15.06.1984. Appl. 25321 02/23-04. 31.03.7978. Achurov DA. Akhmedov AAf, Tedoradze GA, BaSaev NB. Institute of Eleotrochemistry of the Aoademy of Sciences of the USSR: Institute of Chloroorganic Synthesis of the Academy of Sciences of Azerbaija SSR. 79 SU Pat. N 1097596 (Cl. C07C 31/34) 15.06.1984. Appl. 2682639/23-04. 10.11.1978. Bairamov DN, Ashurov DA, Tedoradze GA. Institute of Electrochemistry of the Academy of Sciences of the LSSR; Institute of Chlcroorganic Synthesis of the Academy of Sciences of Azerbaijan SSR. so Azhurov DA, Bairamov DN, Ramazanzade IlbM, Bimatov RE. J Azerbaidzhanskii Khimicheskii Zhurnal 1387: 2: 118-120. a i Ashwov DA, X l u m y a n ZhR, Afinedov AN, Sadykh-Zade SI.In: News of Electrochemistry of Organio Compounds. 8th AllUnion Conferenoe on Electrochemistry of Organig Compomds. Abstraots. Riga: Zinatne, 1973; 212. a2 Ashurov DA, Akh!medov AM, Alumyan ZhR, Sadylch-Zade SI. In: Materials of 2nd Ail-Union Conference of Epoxi.de Monomers an9 Epoxide Resins. Abstract=. Dnepropetrovsk: Dnepropetrovsk University Publication, 1974; 68-69. 83 Ashurov DA, Akhinedov AM, Sadykh-Zade SI. Doklady Akademii Nauk AzerbaiL-hanskoi SSR (Reports of Azerbatjan SSR Acsdeiny of Sciences) 1975; V 31: 3: 43-46. 84 Ashurov DA, Akhmedov AM, Kyaeimov ShK, Tedoradze GA. In: News of Electrochemistry of Organio Conpoun&. 9th A l l Union Cenference on ElectrocherLstry of Crganic Compounds. Abstracts. Tda-Moscow: Publioation of the Institute of Electrochemistry of Academy of Sciences of the USSR, 1976; 25. 85 Ashnrov DA, Tedoradze GA, Ibraghova PM. In: News of El. '

680

86

€37 88

89

trocheniistry of GTganio Compouncb. 10th All-Union Conferenoe on Electrochcmistry of Orgaia Compounds. Abstracts. Novocherkassk: Publication of Novooherkassk Polytechnic Imtitute, 1980; 778. Ashurov DA. In: New6 cs9 Electroohemistry of clrganic Compounds. 11th All-Union Conference on Electroohemistry of GIganic Compounds. Abstracts. loscow-L'vov: L'vov State Untversi-by, 7385; 13. Bairarnov DN, Tedoracize GA, Gorokhova NT, Kazarinov TE, Aahurov DA. 3 Elektrokhimiya 1980: V 16: 8: 1177-1183. E a i r m o v DB, Tedoradze GA,Gorokhova NT, Kazarinov VE, AEhUTOV DA. J Elektrokhimiya 1980: V 17: 1: 90-97. Aahurov DA, Maksbov K h A , Tedoradze GA, Gorokbova NT. J 3lektrokhimiya 1984: V 20: 5: 600-603.

68 1

Figure 1. Schematic illustration of the electrodes with SPE membrane and major reactions proceeding. 1 - Porous anode. 2- SPE membrane. 3- Cathode ourrent oolleotor. 4- Porous oathode. 5 - Graphite sheet. 6- Anode 2-1 4 1. ourrent collector [I

682

e 2. An exploded perspeotive view of an eleotrolyze FiF wi h SPE. 1- Cathode terminal. 2- Niobium net from oathode side.

3- Cation permeseleotive SPE membrane. 4- Gas permeable oathode. 5- Niobium net from anode side. 6- Anode terminal. 7- Chamber in the oolleotor. 8- Fresh aoid inlet. 9- Outlet f o r a spent acid. 10- Contaot elements. 1 1 - Anode terminal main body [12-143.

683

w,

K

WA

IS00

1300 1100

900 700 I

25

I

I

I

29

/m2

I

I

I

I

31

33

HC! yo

Figure 3. Energy expenditure for 1 t chlorine production: I- Current density effect (SPE with oxygen depolarization; concentration of HC1 = 8.3 mole/lh. 11- Hydrochloric acid concentration effect (i = 4.3 kA/m ). 111- Current density effect in industrial condition [21.

684

685 cy

c

to

0.5

Figure 5. Variation of adsorption on platinum with potential. 1- of propylene; 2- of divinyl; 3- of propylene dichlorohydrine; 4- of dichlorobutanediol (the supporting electrolyte- 0.1 N sulfuric acid). 5- Of propylene dichlorobutanediol (the supporting electrolyte- 0.1 N sulfuric acid + 0.: N KC1) [71, 871.

This Page Intentionally Left Blank

687

ELECTROCATALYSIS FOR ENVIRONMENTALLY ORIENTATED ELECTROCHEMICALPROCESSES AND ENVIRONMENTAL PROTECTION

K.WIESENER, D. O H M S ~

INSTITUTE OF PHYSICAL CHEMISTRY AND ELEmROCHEMISTRY, DRESDEN UNIVERSITY OF TECHNOLOGY, D-0-8027 DRESDEN, MOMMSENSTR. 13

CONTENTS 1.

INTRODUCTION

2.

ENVIRONMENTALPROBLEMS IN ELECTROCHEMICALPROCESSES

3.

THE SPECIAL ROLE OF HYDROGEN AND OXYGEN ELECTRODES

3.1 3.2

GENERAL CONSIDERATION ELECTROCATALYSISOF OXYGEN CATHODES AND HYDROGEN ANODES IN AQUEOUS ELECTROLYTES CONSTRUCTION OF GAS-FED ELECTRODES

3.3

4. 4.1 4.2 4.3 4.4 4.5 4.6 4.7

EXAMPLES FOR ELECTROLYTICAPPLICATIONS OF OXYGEN AND HYDROGEN ELECTRODES APPLICATION OF GAS-CONSUMING ELECTRODES APPLICATION OF ELECTROCATALYSTSIN CURRENT-LESS PROCESSES METAL DEPOSITION AND REMOVAL OF METAL TRACES FROM ELECTROLYTES REMOVAL OF ORGANIC AND INORGANIC IMPURIFICATIONS DE-SALINATION AND SALT-SPLITTING SPECIAL APPLICATIONS FOR BATERY SYSTEMS APPLICATION IN ELECTROSYNTHESIS

5.

CONCLUSIONS

6.

REFERENCES

Deutsche AiitomobilgesellschaftmbH (DAUG), D-7300 Esslingen, Emil-Kessler-Sir.5

688

1.

INTRODUCTION

In the age of raising industrial production the problem of environmental protection becomes a vital aspect in all questions of the human society. The problem involves questions of careful utilization of natural resources (raw materials and energy), the prevention of useless or harmful side-products, environmentally-tayloring of products, and the environmentally acceptable elimination (better recycling) of consumed products. Chemical industry in the widest meaning plays a major role in the solution of these questions. Also electrochemistry can contribute as its main scientific interest lays in the interaction of energy and chemical matter. Thus, in this paper one aspect among several others is mentioned which may play a role in the efforts to find new ways of the protection of the environment. 2.

ENVIRONMENTALPROBLEMS IN ELECTROCHEMICAL. PROCESSES

Electrochemical processes may influence the environment mainly by the following factors: - production of harmful substances in the course of the electrochemical reaction - generation of harmful side-productsat the electrodes - indirect generation of harmful side-products at the electrodes or the electrolyte by chemical reaction sequences - energy losses (Joule heat) due to electric resistances included in the cell arrangement - energy losses due to hindrances in the electrode reactions - decomposition products of the cell and electrode materials - formation of residual electrolytesand electrode wastes. In general, electrochemical processes offer a certain number of advantages in comparison to classical chemical processes [l].Thus the electrochemical reaction can easily be controlled by the electric current. In this way also the electrical conditions, e.g. potentials, reaction rate, are observable and can be monitored. Therefore the level of disturbance of the environment should be lower than in classical chemical processes. On the other hand electrochemical processes are usually very selective due to the opportunity to control the electrode potential. They can be used to remove certain harmful substances from the environment. As electrochemical reactions demand the flow of an electric current they are restricted to such systems which show electric conductivity, e.g. electrolyte solutions. During the last years the types of electrolytes has been extended to new fields, e.g. solid state ionic conductors and the investigation of photo-electrochemical processes. Several pure chemical processes were investigated under the viewpoint of electrochemistry. So, the application of electric current is no longer the only criterion to consider a process as an electrochemical one. In addition to the classical electrochemical processes one has to consider such heterogeneous processes which follow similar dependencies of the rate upon the free energy of the reaction interface and

689

have a charge transfer across the interface. Thus in a certain sense also reaction steps in the photoelectrochemical water splitting have to be considered as electrochemical reactions. And finally, chemical reactions at certain catalysts proceeding without any electric current may be include electrochemical reaction steps. This would be visible only if one looks for the potential dependence of the reaction rate and ways to influence the course of the eletrochemical reaction. Nevertheless such processes may be influenced by electrochemical tools and therefore they can be tuned to special applications as needed in environmental protection. The application of electrochemical processes for the environmental protection should therefore be possible and higly recommended for future technologies. If one looks for possible applications to protect the environment from wastes caused by electrochemicalprocesses the main problem consists in the removal of exhausted electrolytes. The first approach should consist in the reduction of the total reactor volume in order to limit the amount of substances at all. But there are a number of restrictions from the engineering. A second facility consists in the choise of such electrode materials and electrolytes which minimize the formation of side products and lead to high exhausting levels of process solutions. Here also the energy demand for chemical reactions has to be taken into account. The last way is to treat the electrolytes formed. These electrolytes contain metal salts in small or medium concentrations. From the metal processing industry similar electrolytes are produced from pickling bathes and galvanotechnical processes as well. Besides the metal content the pH has to be taken into account. In addition, the salt content may be high due to conductivity addititives and chemical neutralisation reactions. Thus, one may summarize the possible ways to treat processing solution in the following principial reactions: - removal or of metal traces (heavy metals from direct reactions) - removal of salt contents (mainly from neutralization) - removal of gases harmful to the environment or the process - conversion of hazardeous substances into harmless components

The aim should consist in the research for processes in which the components are extracted in a form in which they can be used again (closed cycles). 3. THE SPECIAL ROLE OF HYDROGEN AND OXYGEN ELECl'RODES

3.1 GENERAL CONSIDERATION Among possible electrochemical reactions in aqueous systems only the oxydation of hydrogen and reduction of oxygen will not introduce other ions into the system than those which are already present. The only result in using these reactions consists in the change of the pH-value of the electrolyte, The reverse reactions, the evolution of gases will reveal only minor problems as oxygen is a component of the air and hydrogen is not poissoneous (safety considerations have to made with respect to ignition problem when hydrogen is formed in a

690

large scale). Both reactions proceed at inert electrode materials, thus the electrodes can be utilized for longer periods of time, As the reactions are hindered under most conditions electrocatalysts have to be applied to gain acceptable yields. Gas generating electrodes are widely known from technical electrolyses (e.g. TAINTON-anodes in galvanotechniques and insoluble platinum-catalyzed anodes for water splitting). A larger problem consists in the reduction of oxygen and oxydation of hydrogen. In these processes the gases are consumed. Therefore, the gas has to be continuosly supplied to the electrode/electrolyte interface. Special constructions are needed for such electrodes to gain high current densities at low polarizations.

3.2 ELECTROCATALYSIS OF OXYGEN CATHODES AND HYDROGEN ANODES IN AQUEOUS ELECTROLYTES

The electrochemicalreaction of the oxygen reduction in acid solutions is visualized in the following scheme: 4-electron path way:

0 2 + 4 H+ + 4 e-

<=>

2 H20

Eo = 1,223 V

<->

H202

Eo = 0,680 V

2-electron path way:

0 2 + 2 H+ + 2 e-

under certain circumstances followed by one of the following processes:

>

H202

I

2H20

+1/202

or

IH202 + 2 H++ 2 e'

<->

2 H20

Eo = 1,780 V

The oxygen reduction is in general more complicated. Reaction steps including radicals are discussed at different surfaces in alkaline, acid and neutral solutions [2-121. The thermodynamical treatment delivers values for the electrode potential which is not reached at practical oxygen electrodes. Reason for that deviation are the side reaction leading to hydrogen peroxide and additional reaction steps at certain materials (oxide layer formation even at noble metal surfaces). For carbon electrodes a large number of surface processes are known which lead to surface groups and influence the complete behaviour of the material in electrolytes. Thus, the application of electrocatalysts for the oxygen electrodes is demanded as the reaction is hindred at most electrode materials. Besides noble metals, which show high

69 1

electrocatalytic action but are expensive and sensitive with respect to catalytic poissons, certain non-noble electrocatalysts are under investigation: ELECTROCATALYSTS FOR OXYGEN REDUCTION

acidic aqueous electrolytes

alkaline aqueous electrolytes

platinum group metals Nq-chelates (macrocycles) and their annealing products, annealing products of certain nitrogen-containing compounds (e.g. PAN), spinels, chevrel phases, and other inorganic materials silver and alloys metal alloys oxides, spinels and perowskites heat treated macrocycles specially treated carbon materials

The anodic oxydation of hydrogen belongs to the most intensively studied electrochemical processes. The reaction steps are well known and kinetic parameters are compiled in the literature. Besides that the reaction proceeds in technical interesting rate only in presence of electrocatalysts. The highest activation energy is needed from the splitting of the hydrogen bond. Thus, materials which enable an activated adsorption of hydrogen catalyze the electrochemical reaction. Similar to the oxygen reduction noble metals are known as good electrocatalysts for the hydrogen oxydation. Especially at platinum and palladium electrodes the hydrogen reaction displays high current exchange current densities. In the case of palladium and nickel hydrogen is soluble to a considerable extent in the metal. Taking into account the sensivity of noble metals to other electrochemically active species non-noble metal catalysts with a high selectivity are needed in more complex systems. Specially synthesized inorganic compounds as e.g. carbides can activate the hydrogen bond and may therefore be employed as electrocatalysts [13-191. Some electrocatalysts for the reaction of hydrogen in acid and alkaline electrolytes are compiled in the following table:

692

ELECTROCATALYSTS FOR HYDROGEN OXYDATION

acidic aqueous electrolyte

platin group metals, tungsten carbide

alkaline aqueous electrolyte

nickel and alloys, palladium

3.3 CONSTRUCTION OF GAS-FED ELECTRODES

Gaseous reactants in electrochemical devices demand a special design of the electrodes in order to provide sufficient gas transport, fast electrochemical reaction, transport of the products, electrolytic contact and high electronic conductivity of the electrode and the current collector. Mainly gas-diffusion electrodes are used in technial applications. There are different types of gas-fed electrodes. In general these electrodes have layers of different properites with respect to electrolyte and gas transport. The catalyst is located in the electrolyte side of the electrode only. The gas-side is non-permeable for the electrolyte (often PTFE-membranes or PTFE-containing carbon materials are used). The collector grid may be located at different positions depending upon the producer and the electrolyte system under investigation. For alkaline electrolyte it may be pressed into the catalytic layer in order to minimize ohmic losses. For acidic solutions a position in or at the surface of the gas-side layer is prefered (corrosion problems). In order to minimize the demand of catalysts and to obtain reliable working conditions (gas transport, preventation of electrolyte penetration, large active surface area) our electrodes are composed of two different porous layers using different carbon materials as support for the catalyst and as main component in the gas-side layer [20-231.

-

electrocatalyticlayer (wetable catalyst-containing carbon material with PTFE-binder)

-

gas-side layer (non-wetable carbon material with PTFE-binder, specially designed for gas transport)

In addition the electrodes contain a metal grid as current collector. As different partial processes are combined in the operation of a gas-fed electrode its design demands a optimization with respect to its parameters. The operation can be tuned to the electrolyteheactant system by the choise of composition (binder. catalyst, su~uortcarbon.

693

collector material), the preparation procedure (processing) as well as the operation conditions (gas-pressure, electrolyte pressure temperature). The following figure shows the typical construction of a gas-fed electrode:

I

current collector

,

electmcotalytic layer

1

electrolyte

I

J

~~

~

~

PHYSICOCHEMICALAND CHEMICAL PROPERTIESOF A HYDROGEN ANODE

Electrocatalvtic laver

support: Acethylene black P1042 surface area ab. 60 m2/g (by nitrogen adsorption and BETequation), pore volume ab. 2.8 cm3/g (mercury-porometry) Catalyst : Tungsten Carbide WC [80 %], (specially prepared) Binder: 8%PTFE (in combination with the carbon material\ Gas-side laver Main component: Acethylene black P1042: (surface area ab. 60 m2/g (by nitrogen adsorption and BETequation), pore volume ab. 2.8 cm3/g (mercury-porometry) Binder:

35% PTFE (precipitated from PTFE-suspension)

694 1

I

PBYSICOCHEMICALAND CHEMICAL PROPERTIES OF AN OXYGEN CATHODE

Electrocatalvtic laver: Carbon P33 (surface area ab. 1000 m2/g (by nitrogen adsorption and BET-equation), pore volume ab. 2.3 cm3/g (mercuryporometry) Catalyst : Dibenzotetraazaannulen-cobalt(I1) (CoTAA), [15% on carbon], heat-treated in argon at 900K,5 h Binder: PTFE (precipitated from PTFE-suspension, 10%content in the electrode layer) Gas-side laver

Main component: Acethylene black P1042: (surface area ab. 60 m2/g (by nitrogen adsorption and BETequation), pore volume ab. 2.8 cm3/g (mercury-porometry) Binder: PTFE (precipitated from PTFE-suspension, 35% content in the electrode layer)

4.

EXAMPLES FOR ELECTROLYTICAPPLICATIONS OF OXYGEN AND HYDROGEN ELECTRODES

4.1

APPLICATIONOF GAS-CONSUMING ELECTRODES

There may be two ways to employ gas-consuming electrodes in technical processes: substitution of gas-evoluting electrodes in electrolysis cells A. B. current-less catalysis at electrocatalysts By this the energy demand of the electrolyses becomes reduced and processes may be designed which are not possible if the potentials of both electrodes are far from each other. Thus, substituting an hydrogen consuming electrode for a usual oxygen generating one the gain of voltage exeeds the difference of the standard potentials as the oxygen generation . second methode proceeds at higher polarizations than the hydrogen oxydation [a]The involves the electrocatalyst for processes without electric current based on the assumption

695

that several chemical processes include electrochemical reaction steps. Following processes may be used as examples for the application of gas-consuming electrodes [24-251: Hydrogen oxydation: 1. Recycling of copper etching bathes 2. Recycling of pickling bathes for steel processing 3. Deposition of extremely pure metals and alloys, electroplating 4. De-salination of electrolytes and salt splitting Oxygen reduction: 1. Recycling of manganese-containingelectrolytes 2. Synthesis of metal-salt solutions

4.2

APPLICATION OF ELECTROCATALYSTS IN CURRENT-LESS PROCESSES

Simultaneouselectrocatalysisof partial reactions takes place at a particular electrode material. Although reaction seems to follow rules of simple chemical catalysis electrocatalysts will influence the whole process. For processes without outer current the following processes may be considered: including oxygen reduction:

- metal dissolution, e.g. tin dissolution in HBFq. - oxydation state change of metal ions, e.g. oxydation of Cut - oxydation of impurities in waste-water (e.g. cyanide oxydation) - corrosion protection by self-passivation including hydrogen oxyda.tion:

- metal deposition - oxydation state change of metal ions, e.g. reduction of Cut+, Fe3+ The difference between purely chemical processes and the processes mentioned here consists in the way how reaction rate is influenced by the electrode potential and the fact that the reaction takes only place at the interphase of differently conducting materials (e.g. including an electrolyte).

4.3

METAL DEPOSITION AND REMOVAL OF METAL TRACES FROM ELECTROLYTES

Cathodic extraction of iron during recycling of pickling bathes plays a considerable role for metal working enterprises r261. Large quantities of iron sulphate solutions are formed and can

696

hardly be transfered into utilizable products or disposed without any formation of noxious matters. Substituting the conventional oxygen electrodes by hydrogen anodes the electrode potential becomes more negative and therefore any anodic oxidation of iron-(11) ions is prevented. That leads to - higher current efficiency and - prevention of the contamination of the iron cathode iron by basic iron sulphates. The following three-step process is suggested to make use of hydrogen-fed electrodes for processing iron pickling bathes. The first step (current-less reduction at a hydrogen (WC-catalyzed electrode) is a prereduction of iron (Fe3+) as iron can only be efficiently deposited from the Fe2+ oxydation state. In this step also other oxidizing compounds (e.g. nitrate) are removed. The second step (or a cascade of similar cells) is applied to shift the pH of the solution to higher values. Hydrogen is needed only due to minor iron deposition and losts. The third step contains the electrolysiscell for the deposition of iron. The process solutions transport can be arranged in such a way that the pickling electrolyte can directly be used again in the cell.

697 UGGESTED PROCESS SCHEME :

from pickling process

I

Fe

2H'

H,

to the pickling process

.. "2A Sep. IAEM

CEMl

Sep.

I

W

I

v

I

inert cathode

--

Fe

NO;, SO ;:

M'

2+

Fe3+

Fe?

I

+

*

H+>

(-

H+

'

__

2a

I

I H2

4

AEM

CEM

Sep.

v

I

inert

--

cathode

Fe

(*

+

H+>

'-H+

-5

2

2b

D

I

A

I I

T I

v

I

iron cathode

(

--

I

*

H+)

so'-

Fe+ .-

I I

A

*

H+

_' _4 2

3

698 RESULTS OF THE PRE-REDUCTION OF IRON CONTAININGSOLUTIONS

Current-less Reduction of Iron3+ In Sulphuric Acid at 5OC ' 32 g/l F a . 22 g/l HN03

I

I

lo

t (hl

20

I

CURRENT-LESSREDUCTION OF IRON (3 t) BY APPLICATION OF A HYDROGEN ANODE IN CHLORIDIC ELECTROLYTE

699

The properties of WC-catalyzed hydrogen electrodes offer the opportunity for another recycling process - the regeneration of cupric chloride etchants from printed circuit manufacturing [26-281. The conventional method to recycle acid cupric chloride etchants consists in an addition of hydrochloric acid and hydrogen peroxide. Thus, the volume of the etchant is continuously growing and it has to be discarded partly. Many attempts have failed to find a solution for the electrolytic treatment even with the use of membranes. The main reason consists in the evolution of toxic chlorine at the anodes as soon as mass transfer becomes limited at high anodic current densities. In the recycling of chloridic copper etching electrolytes both electrodes under consideration, hydrogen anodes as well as oxygen cathodes, may be employed. Thus, a certain portion of the copper-(11)-chloride is reduced to copper(1) in a current-less process step (step 1) using tungsten carbide-catalyzed electrodes until a precipitation of copper in an electrochemical cell using the same electrodes becomes possible (step 2). The final oxydation of the copper(1) is achieved by means of contact of the electrolyte with an oxygen-fed electrode in a current-less process (step 3). This leads to a complete regeneration of copper etching electrolytes. The currentless reduction process increases considerably the efficency of the electrolysis by formation of Cu+ before the electrolysis procedure. The porous partially-hydrophophized gas-fed electrodes may be applied as walls of the pipings through that the electrolyte is transported. The advantage of the currentless oxydation consists in the complete regeneration of the etchant. The re-oxydation of the Cu2C12 can be performed by inserting oxygen directly into the solution or by oxydation at electrocatalyzed O2-fed electrodes in a short circuited cell. The latter methode prevents the formation of aerosols from gas bubbling.

700

The kinetics of the reaction steps during the electrolysis was investigated and a mathematical model for the reduction process was established: etching:

CuC12 t Cu

step 1

PRE-REDUCITON

I

cu2c12

>

Cu2C12 t 2HC1

Cu2C12 t 2e-

>

2 c u t 2c1-

CuC12

t 2 e-

>

2 cu2c12 t 2 c1-

Cu2C12 t 2 e-

>

2cu

2CuC12+ H2

>

Cu2C12 t 2HC1

cu2c12

>

cu

H2

>

2Ht

2CuC12

;tep 2

+

>

H2

ELECLaOLYSIS

cathode

+

2c1-

electrolyte t Cuc12

anode

step 3

+ 2e-

RE-OXYDATION

Cu2C12 t 2 HCl t 1/2 0 2

>

2CuC12 t H20

A mathematical model of the kinetics using a RUNGE-KUTTA procedure was established and is in good agreement with the observations. Concentration profiles of all different copper ions are obtained and the redox-potential of the Cu+/Cutt-equilibrium in the electrolyte can be estimated.

70 1

In the figure the concentration profiles of different copper ions during electrolysis (Cu+,chem indicates the portion of Cu+ ions formed by chemical (current-less) reduction) are shown.

CONCENTRATION PROFILE FOR THE RECYCLING PROCESS

cu

\

I

experiment

702 CURRENT-LESS OXYDATION OF

AT AN OXYGEN ELECTRODE IN SULPHURIC

ELECTROLYTE AT 298 K

8.0

mCUP2 6.0

4.0

2.0

4.4

REMOVAL OF ORGANIC AND INORGANIC IMF'URIFICATIONS

There are two ways to utilize the high potential of the oxygen electrode to oxydize organic and inorganic traces in industrial waste water. First, oxygen may be generated at a inert electrode anodically. The formed oxygen can itself react with impurities as it is formed inan activated state at the electrode surface sites. A second oxidizing agent is the hydrogen peroxide as side product. The second way consists in the application of an oxygen cathode electrically connected to an inert electrode at which the oxydation takes place. The latter has the advantage of a separation of the oxygen process from the cathode. Thus, special materials may be employed for this process. The advantage of such a process consist in the fact that no side-products are formed at the gas electrode and the oxygen for the electrode is available at low costs. A disadvantage is the slow oxydation rate if no auxiliary energy is employed. Practically, such a process is not used in technical scale yet as it is difficult to find stable electrocatalysts of high activity with respect to oxygen reduction in very diluted aqueous solutions. A possible application couId be the removal of cyanides (galvanotechnique) and traces in agricultural waste water. Similar to an oxydation the hydrogen electrode may be

703

used to reduce organic materials in aqueous solutions. The advantages and disadvantages are in principle the same.

4.5

DE-SALINATIONAND SALT-SPLITTING

Using gas diffusion electrodes, aqueous NaCl- and Na2S04- solutions can be splitted into NaOH and HCl and H2SO4, respectively, at very low electrolysis voltage (resulting only from the ohmic drop and diffusion potentials in the cell) in a three-chamber cell with cation and anion exchange membranes.

SCHEME OF THE THREE-CHAMBER CELL WITH CATION CEM AND ANION AEM EXCHANGE

MEMBRANESFOR NACLSPLITIING

NaCI.

conc.

CEM

I

I

AEM

NaCI. dil.

4.6

I

I

SPECIAL APPLICATIONS FOR BATTERY SYSTEMS

Besides a direct application of the hydrogen and oxygen electrodes in power sources (fuel cells, metal-air batteries, metal-hydrogen batteries, hydride cells) a application of gasconsuming electrodes consists in the gas-recombination in sealed cells. An application of the electro-catalysts for acidic electrolytes mentioned above could be the elimination of

704

hydrogen traces in lead acid batteries by composite electrocatalysts. In this case an noble metal catalyst cannot be used as traces of noble metals in the electrolyte can shift the hydrogen overvoltage of the negative electrode and lead to self-discharge of the battery. A recombination of the gases, especially hydrogen enables a complete sealing of the battery and prevents the generation of acidic aerosols from the battery completely [30-341.

4.7

APPLICATION IN ELECTROSYNTHESIS

4.7.1 ZINC WINNING ELECTROLYSIS In the industrial zinc winning electrolyses, the purified zinc sulphate solution is electrolyzed by using aluminium cathodes and TAINTON-anodes (lead with 1% silver). At the TAINTON-anode oxygen is formed from the sulphuric electrolytic solution. In order to reduce the specific electric energy required, the oxygen anode can be substituted by a hydrogen anode, thus hydrogen oxydation takes place and the electrolysis proceeds according to the following equations:

1cathode:

zn2+

+

2e-

->

anode:

H2

cell :

Zn2+ t H 2 >-

Zn

U&= -

2H+ t 2e-

a=-0.oov (2)

Zn t 2H+

0.76 V (I)

(3)

In this reaction the standard cell voltage is 0.76 V, but additionally, one equivalent of hydrogen is necessary to produce on equivalent of zinc. The voltage difference between the oxygen anode and the hydrogen anode is always higher than the difference of the standard potentials of 1.23 V because the hydrogen electrodes overvoltage in the presence of an appropriate stable electrocatalystsis kept much lower than that of the oxygen electrode. Thus, a working cell voltage of 1.8 ... 2.0 V could be obtained. Substituting oxygen anodes by hydrogen consuming anodes in the process of zinc electrolysis in an industrial scale, a cell voltage gain of 1.5 ... 1.7 V can be expected for an acid zinc sulphate electrolyte. Thus, considering current efficiency in practice, the specific electric energy required for the zinc production is reduced from 3100 ...3300 k W t to 1500 ... 1800 k W t at the expense of the application of hydrogen [35-361.

I

705

4.7.2 SYNTHESIS OF TIN TETRAnUOROBORATE Gas-fed electrodes can be applied in the production of inorganic substances. For instance, the preparation of tin(I1)-tetrafluoroborate can be performed in the following way [37-381:

anode:

Sn

Sn

->

Sn2+

+

UR=-

2e-

0.14 V (la)

~ n 4 ++ 4e-

( 9

possible consecutive reaction sequences in the anodic chamber: 2H+ + H202 + Sn2+

->

Sn4++2H20

-> ->

Sn2++2H20 2~n2+

02+4H++4e-

->

2H20

02+2H++2e-

->

2H202

+ 2 HBF4

->

Sn(BF4)2

+

+

->

Sn(BF4)4

+ 2H20

->

2Sn(BF4)2

or 2H++ H202+Sn ( ~n4++~n

cathode:

Uf$= 1.230 V

(3b)

I

cell:

Sn + 1/2 0 2 Sn+02

4HBF4

Sn(BF4)4 +Sn

(3a)

H20

706

The reaction makes use of the potential difference between an oxygen and the tin electrode. It is not necessary to separate the electrode chambers as in a membrane process. In the following table some results with different electrocatalysts are compiled.

EXAMPLES FOR DISSOLVING TIN BY OXYGEN ELECTRODES IN GALVANOSTATIC

LONG-TERM EXPERIMENTS

i = 90mA/cm2, 298 K, 40% HBF4, depolarizing agent: air

electrode material'0 : P33/CoAc/PAN 800oC/3h time E *1 m(Sn), dissolved m(Sn), precipitated m(Sn), determined by titration m(Sn), calculated "2 efficiency +3 efficiency +4

P33/CoTAA 9000C/Oh

P33/"MeOPPCo 8OO0C/3h

h mV g g

17 -620 21 0.92

19 -660 24.2 2.1

16 -580 21.2 1.85

g g

14.4 30.8 0.47 0.71

14.9 34.5 0.43 0.76

15.4 29.0 0.53 0.79

........................................................................................................... *O electrode materials: different materials precipitated and heat-treated at a active carbon (P33) *1 *2 *3 '4

surface, CoAcPAN: cobalt acetate polyacrylonitrile mixture, C o T M dibenzotetraazaannulenecobalt (II), TMeOPPCo: tetramethoxyphenylporphyrinecobdt (11) potential of the oxygen electrode vs. SCE according to m=I+M/(z.F) efficiency= m Sn titrated 1m(sn) lculated efficiency = mb~dissolved/ m(s%alculated ~

An additional advantage of this process consists in the fact that the cell does not need electric

energy from outside. By connecting both electrodes the reactions proceeds by itself. This leads to the suggestion to perform the formation of the tintetrafluoroborate by applying a catalyst for oxygen reduction direct to the metal surface and providing the interface with the reaction gas (e.g. forming a suspension of metal powder, catalyst particles in an areated electrolyte). For this process electrocatalystsfor oxygen gas-diffusion electrodes may applied (even if the process does not proceed if electric current). Applying oxygen electrodes the production of the tetrafluoroborate delivers no side-products and needs less energy than compared to traditional processes based on the reaction of tin with copper tetrafluoroborate. An additional advantage of this process consists in the fact that the cell does

707

not need electric energy from outside. By connecting both electrodes with each other the reactions proceeds by itself driven by the potential difference. This leads to the suggestion to perform the formation of the tintetrafluoroborate by applying a catalyst for oxygen reduction direct to the metal surface and providing the interface with the reaction gas (e.g. forming a suspension of metal powder, catalyst particles in an areated electrolyte). For this process electrocatalysts for oxygen gas-diffusion electrodes may applied (even if the process does not proceed if electric current). A unwanted side-process is a partial formation of Sn(1v) in the electrolyte if most active electrocatalysts are applied.

5.

CONCLUSIONS

Gas-fed electrodes can be applied for energy saving processes and processes which are not disturbing the environment. In addition applications can be expected for recycling of process solutions and side-products of processes, For such applications electrocatalysts are demanded which require properties like surperior activity, stability selectivity. These catalysts should not be to expensive and easy ways to apply them in electrodes. This requieres to develope new electrode designs for stable long-term applications and to gain knowledge on the physico-chemical processes proceeding in them.

REFERENCES

D.Ohms, K.Wiesener; UNESCO Expert Workshop "Contrib. of Electrochem. to Energy Conserv. and Saving and Environm.Protect.", Gaussig Castle, Oct 30-Nov 3, 1989, Dresden University of Technology, Proceed., TU Dresden (1990) 153-164.. E.Yeager, Electrochim. Acta 29 (1984) 1527. K.Wiesener, Electrochim. Acta 31 (1986) 1073. H.Jahnke, M.Schoenborn u. G.Zimmermann in "Topics in Current Chemistry" 61, (1976) 235. F.Beck, Ber. d. Bunsenges. Phys.Chem. 77 (1973) 353. K. Wiesener, D.Ohms;in "Electrochemical Hydrogen Technologies- Electrochemical Production and Combustion of Hydrogen", (ed. H.Wendt) Elsevier Publ. Amsterdam, Oxford, New York, Tokyo, 1990, p.63-103. G.Gruenig, K. Wiesener, S.Gamburzev, I.Iliev and A.Kaisheva, J.Electroanal.Chem. 159 (1983) 155. E.Yeager et.al., ISE-meeting 1984 Berkeley, Calif., USA), Ext. Abstr. p.351. V.S.Bagotzky, M.R.Tarasevich, O.A.Levina, K.A.Radushkina and S.I.Andruseva, Dokl. Akad.Nauk SSSR 233 (1977) 889. J.A.R.vanVeen,H.A.Colijn,J.F.Baar, Electrochim. Acta 33 (1988) 6,801-804

708

[ 11 ] K.Wiesener, D.Ohms, V.Neumann, R.Franke; Mat.Chem.Phys. 22,3-4 (1989),457-

475. [ 12] I.Iliev, S.Gamburcev, A.Kaisheva, E.Vakanova, I.Muchovski, E.Budevski; Izv.Otd.Khim. Nauki, Bulg. Acad.Nauk 7(1974)233. [ 13 3 I.Nikolov, M.Svata; J.Power Sources 3 (1978),237-244. [ 14] LGrigorov, T.Vitanov, ZZabransky, I.Nikolov, T.Vitanov; J.Power Sources 3 (1978),

273. [ 16 ] K. Wiesener, E.Winkler,W.Schneider; Z.phys.Chemie 266 (1985),579-588. [ 15 ] E.Winkler, W.Schneider, K.Wiesener; Z.phys..Chemie 266 (1985),589-594.

[ 17] K.Wiesener, W.Schneider; Chem.Techn. 37 (1985),373-375. [ 18 ] K.Wiesener; 6th World Hydr.Energy Cod., Vienna, 20-24.7.1986publ. , in "Hydr. Energy Progr.VI", Pergamon Press, (ed. T.N.Veziroglu, et.al.) 3 (1986), 1301-1317.

[ 19 ] G.V.Bojkova, W.Schneider, K.Wiesener, G.V.Shutajeva, M.Tarasevitsch; Electrokhim. 22 (1986),1135-1138. [ 20 ] D.Ohms, W.Schneider, K.-H.Ulbricht, K-Wiesener;Z.phys.Chemie, Leipzig, 267

(1986)45. [ 21 ] [ 22] [ 23 ] [ 24 ]

D.Ohms, K.Wiesener; Chem.Technik 33 (1981),312-314. R.Franke, D.Ohms, K.Wiesener; J. Electroanal. Chem. 260 (1989)63-73. U Dok Hi, D.Ohms, R.Franke, K.Wiesener; Chem. Techn. 42(1990)294-297. K.Wiesener, D.Ohms, A.Mobius; DECHEMA, Society of Chemical Industry Electrochemical Technology Group (London), European Federation of Chemical Engineering (Working Party, 432Event) Conference "ElectrochemicalCell Design and Optimization Procedures", Bad Soden, 24-26.9.1990Abstr., , and publ. in DECHEMA-Monographien 123 (1991),95-109. [25] D.Ohms, R.Franke, K.Wiesener; Vortragstagung "Elektrochemie in Energie- und Umwelttechnik" der GDCh (Fachgr. Angew. Elektrochemie), Kassel, 10.-12.10.1990, Abstr.26, and publ. in DECHEMA-Monographie,Band 124,ed. A.Winse1, VCH Verlags-AG Weinheim, Basel, Cambridge, New York, 1991. [ 26 ] A.Mobius, K.Wiesener; 1nt.Symp. "Fuel cells, their appl. and electrochem. sensors", Bonn, 1990.

-

[ 27 ] K.Wiesener, D.Ohms, R.Franke, D.Wahl, Chr.Gruhnwald; 42nd ISE-Meeting , Montreux, Aug. 25-30,1991, Abstract 5-22. [ 28 ] patent DD 294513,1991. [ 29 ] patent DD 294515,1991. [ 30 ] D.Ohms, L.Dittmar, H.Dietz, K.Wiesener; Chem.Techn. 43 (1991)187-191. [ 31 ] I.Nikolov, G.Papazov, V.Naidenov, T.Vitanov, D.Pavlov; LABAT-89,Int.Conf. on Lead Acid Batteries, Druschba (Bulgaria), 1989.

709

[ 32 ] H.Dietz, M.Radwan, L.Dittmar, D.Ohms, K.Wiesener; UNESCO Expert Workshop "Theory and Practice of the Lead Acid Battery System", Gaussig Castle, Dresden University of Technology, April 2 - 5,1991, Proceedings; TU Dresden (1991). [ 33 ] H.Dietz, L.Dittmar, D.Ohms, M.Radwan, K.Wiesener; to be published in J.Power sources. [ 34 ] D.Ohms, L.Dittmar, H.Dietz, K.Wiesener; Chem.Techn., 43 (1991) 187-191. [ 35 ] A.Mobius, ICWiesener, B-Brandt; Wiss. 2. TU Dresden 38 (1989), 127-130 [ 36 ] K.Wiesener, W.Schneider, A.Mobius; J.Electrochem SOC.136 (1989), 3770-3772. [ 37 ] D.Ohms, R.Franke,S.Herzog, K.Wiesener; "Elektrochem. in Energie- und Umwelttechn." der GDCh, Fachgr.Angew.Elektrochem.,Kassel, 10.-12.10.1990, and publ. in DECHEMA-Monographie 124, (ed. A. Winsel), VCH Verlags-AG Weinheim, Basel, Cambridge, New York, 1991.

[ 38 ] D.Ohms, S.Herzog, R.Franke, K. Wiesener; Jahrestagung der Fachgruppe "Angewandte Elektrochemie" der GDCh,"ElektrochemischeStoffgewinnungGrundlagen und Verfahrenstechnik", LudwigshafedRhein 23.-25.10.91, and publ. in DECHEMA-Monographien 125 (1992), 287-302.

This Page Intentionally Left Blank

71 1

AUTHOR INDEX A m Y.B. Alexander S. Al-Turkait A. Araujo L.P.S. Asiri A. Brito P.S.D. Bro P. Comninellis C. Fahidy T.Z. Farmer J.C. Fenton J.M. Gale R.J. Guilbault G.G. Harm S.K. Ismail M.I. Kajiyama F. Kazarinov V.E. Kharkats YU.1. Kordesch K. Levy S.C. Li H.

62 1 535 347 223 347 203 131 77 60 1 565 103 62 1 273 445 347 365 655 469 163 131 62 1

Micka K. Nakahara T. Ohms D. Pacheco A.M.G. Pleskov Yu.V. Reade G.W. Rocchini G. Rousar I. Santhanam K.S.V. Seiyama T. Sequeira C.A.C. Strathmann H. Suleiman A.A. Takeuchi T. Tatapudi P. Taucher W. Tebbutt P. Tedoradze G.A. Walsh F.C. Wiesener K.

45 233 687 223 417,469 3 377 45 445 233 203,223 495 273 233 103 163 305 655 3 687

This Page Intentionally Left Blank

713

SUBJECT INDEX

A Acceptable degradation 104 Acetoin detection 277 Adsorption/ desorption 623,664 Adsorption isotherms 638 Advanced batteries 133,189 Advanced oxidation processes 11 1 Aerobic corrosion 365 Air pollution 233 Alkaline fuel cells 2 10 Alkaline manganese batteries 134, 163 Aluminiudair cell 2 18 Amines detection 278 Ammonia detection 278 Anaerobic corrosion 365 Anaesthetic gases 327 Analytical methods for gas analysis 237 Anodic deposition 82 Apparent corrosion density 384 Apparent polarization resistance 392 Archimedes number 59

B Backrial ecology 365 Band bending 457 Battery life cycles 134 Battery recycling 144, 158, 162, 193 Battery separators 169, 177 Battery service life 173 Battery shelf life 173, 187 Battery waste management 140,

142, 159, 191 Benzene 86,568,580,583 Bipolar flat plate cells 180 Bipolar membranes 525 Bipolar trickle tower reactor 34 Bromide detection 296 Bundle cells 181 Buried pipes 366

C CaCO3 precipitation 6 13 Cadmium toxicology 198 CaFe2O4 photoelectrode 459 Capillary flow 611 Carbon electrodes 540 Carbon monoxide detector 280 Catalyst coated semiconductors 45 1 Catalytic combustion gas sensors 240 Cathodic dehalogenation 73 CdS dispersions 440 Cellcapacity 172 Cementation 3 Cheese whey demineralization 523 Chemical oxidation of organics 96, 103 Chemical oxygen demand 78, 103 Chlorinated hydrocarbons 72 Chlorine 116 Chlorine-alkaline electrolysis 524 Chlorine dioxide 117 Chlorine production 657 Chlorofluorocarbon sensor 258 Chlorohydrines 670 Chlorohydroxylation 672

714

Chlororganicproduction 655 Clark type sensor 313 COaI gasification process cell 539, 547 Cobalt-based mediated process 570,583 Concentration polarization 5 1 Concentric cylindrical cell 30 Constant radiation power density 480 Copper deposition 19,54 Copper detection 296 Corrosion-induced fields 608 Corrosion rate 392 C02 reduction 446,451,453 CoS2 electrodes 540 C@ evolution 575 C02 monitoring 576 C02 sensors 253,279,3 15 Coulombic efficiencies 574 Current efficiency 77,675 Current interruption technique 405 Current-less processes 695 Cyanide detection 297 Cyclic voltammetry 569

D Dc pulse technique 347 Degree of oxidation 85 Desalination 703 Desalination plant 5 16 Diaphragmless electrolysers 62 Dichloro propanol conversion 580 Diffusion dialysis 528 Dimensionally stable anodes 80 Discharge curves 172 Donnan dialysis 528 Donnan potential 497 Double layer representation 389

Double-wall electrochemical cell 380

E Electrocatalysis 687 Electrocatalystsfor hydrogen oxidation 692 Electrocatalystsfor oxygen reduction 691 Electrochemical batch reactor 571 Electrochemical cells (reactors) 10, 16,20,31,45, 506,566,575 Electrochemical chlorination 662 Electrochemical impedance spectroscopy 227,370 Electrochemical membrane separator 535,563 Electrochemical oxidation index 83 Electrochemical oxidation of organics 86,96,438, 565 Electrochemical oxygen demand 85 Electrochemical parameters, 395 Electrochemical potential 535 Electrochemical techniques 8, 226,305,347,367,377,445 Electrode geometry 23 Electrodeposition 18,49 Electrodialysis 3, 505, 514, 527 Electrodialysis reversal plant 5 17 Electrokinetic processing 62 1 Electrolysis 9, 524,623,657, Electroosmosis 622 Electroosmotic flow 623,629,633 Electrophoresis 622 Endotoxin detection 297

715

Energy diagrams 4 18,420,43 1 Encrgy parameters 189,507,509 Engiteo process 149 Environmental protection 605, 687

Ethylene chlorination 666 Ethylene glycol conversion 577, 583,593

F Fermi level 419 FET sensors 241 Fill factor 426 Flow-through channel electrolyser 54 Fluidized bed electrolyser 58 Formaldehyde detection 280 Fractional conversion of metal-ion 21 Fuel cells 133,203,208

G Galvanostaticprocesses 706 Galvanostatictechnique 402 Gas-fed electrodes 692,694 Gas sensors 233,294,305 Greenhouse effect 204,234

H Halide-tolerant mediators 566 Halothane detection 33 1 Harmful battery components 137, 193

Hazardous wastes 565 HBr photodecomposition 436 High-Gequency alternating current 407 Household batteries 153

Hybrid membrane processes 530 Hydrocarbons detection 28 1 Hydrochloric acid-waste 65 5 Hydrogen chloride detection 282 Hydrogen cyanide detection 283 Hydrogen electrodes 689,693 Hydrogen generation 457,477 Hydrogen peroxide 110 Hydrogen production 657 Hydrogen sulphide 283,438,454, 535

Hydrometallurgical

processing

148, 151,155

Hypochlorite 116

I IB corrosion 372 InP photocathode 434 Internal resistance of cells 173 IOB corrosion 371 Iodide detection 297 Ion-bound water transport 5 13 Ion-exchange membranes 496, 504,529 Ion permeable separator 13 Irodair cell 2 17

Iron-based mediated process 57 1 Iron detection 297 Isoflurano detection 334

J Jelly-roll cells 178

K Kaolinite clay 634 Kerosene 568 Kinematic viscosity 53

716

L Large pore theory 624 Lead-acid batteries 135, 146 Lead detection 298 Lead toxicity 199 b l a n c h 6 cells 165 Limiting current density 54,5 11 Liquid electrolyte gas sensors 240, 338 Liquid-junction solar cell 423 Lithiated NiO electrodes 540 Lithium manganese batteries 135 Local charge density 626 Luggin capillary hnctionality 403

M Magnesium polymer electrolytes 226

Magnetically assisted corrosion 610 Magnetic coagulation 606 Magnetic field effects 601,604, 609,611 Magnetic filtration 606 Magnetoelectrochemistry 6 15 Magnetohydrodynamic approach 603 Manganese dioxide cathodes 168, 175 Mass transport control 18 Maximum mediator concentration 594 Mediated electrochemical oxidation 565 Membrane covered sensors 307 Membrane selectivity 5 13 Mercury cells 135

Mercury detection 284,298 Mercury reduction 170, 197 MetaVair batteries 2 13 Metal deposition 695 Metal ion concentration 18,54 Metal-ion liquors 5 Metal ion removal 17, 438, 644, 695 Microbial detection 298 Microscopic (ion-scale) approach 602 Migration potential 622 Mixed wastes 565, 568 Mn02-Zn cell 164 Molten carbonate he1 cells 213, 539 MPB corrosion 373

N Natural gas process cell 539 Nernst equation 50,311,537 Nickel-cadmium batteries 135, 148 Nickel-metal hydride batteries 136 NIFE process 150 Nitro-aromatic compounds 285 Nitrogen oxides 206, 224, 234, 286 Nitrous oxide detection 328 N 4 , sensor 244,278 Numerical methods 387,487

0 Ohmic drop 377,392,403 Optical gas sensors 240 Organic pollutants 77 Organophosphorous compounds 286

717

Oscillator type gas sensors 24 1 Osmotic transport 5 13 Oxygen electrodes 689,694 Oxygen flow rate 78 Oxygen reduction 215,456 Oxygen sensors 260, 3 12 Ozone 105 Ozone sensors 256,290

P Packed bed electrolyser 68 PAM cells 165 PEC cell efficiency 426,428 Phenol 88,96 Phosgene detection 290 Phosphoric acid fuel cells 2 11 Photocorrosion 425 Piezoelectric crystal detectors 273 Plate electrolyser 57 Polarization curves 52,395,455, 555

Pollution control 261, 35 1 Potable water production 5 19, 52 1 Potassium permanganate 115 Power series expansion 397 Primary batteries 132, 163 Primary degradation 103 Propylene glycol dinitrate detectors 290 Ptanodes 93 Pulsating bed electrolyser 7 1

Q

Rainfkll pollution 35 5

RAM cells

174

Recycling of metals 24 Recycling processes 70 1 Recytec process 156 Recytec recycling process 196 Reserve batteries 132 Reticulated vitreous carbon cell 37 Reverse osmosis 5 10, 522 Reynolds number 53, 5 12,572 Rolling layer cell 70 Rotating-cylinder anode 57 1 Rotating cylinder electrode reactor 30 Rotating ring disc electrode 3 14

S Salt-splitting 703 Schmidt number 574 Seawater pollution 360 Secondary batteries 133, 163,478 Semiconductor dispersions 436 Semiconductor/electrolyte systems 418 Semiconductor gas sensors 237, 336

Semiconductor photoelectrochemistry 417,445 Sheet-flow electrodialysis spacer gasket 5 15 Sherwood number 5 1,5 12 Silver-based mediated process

Quantum yield 426

569, 582,595

R

Silver deposition 299 Silver oxide cells 135 Small pore theory 629 SOB corrosion 373

Radionuclide electrochemistry 597,647

718

Soil contamination 357 Soil decontamination 62 1 Solar array+electrolyzer+secondary battery 484 Solar energy 417,445,469 Solar-hydrogen plants 469,474 Solid electrolyte gas sensors 239 Solid oxide fuel cells 2 12 Solid polymer electrolyte fuel cells 212 Solid-state solar cell 424 Solid state batteries 223 Solution resistance 399 SO, sensors 251,291 Spirally rolled cells 178 Spray pyrolysis technique 82 SRB corrosion 369 Streaming potential 622 Sulphur oxide 206,224,234,454 Sumitomo process 154 Sun-to-hydrogen efficiency 470

T Table salt production 52 1 Tartaric acid removal 523 Thermal batteries 132 Thermal decomposition technique 80 Thermal processing 146, 148, I 53 Thin layer magnetodynamics 6 14 Thionyl chloride batteries 135 Threshold limit value for short term exposures 137 Ti/Ir02anodes 80 Time weighted average exposure 137 Tin tetrafluoroborate synthesis 705 Ti02 electrode 429,457

T i P Q a n o d e s 82 Ti/Ru02 anodes 80 Ti/Sn02 anodes 82 TNO process 152 Toluene diisocyanate detection 293 Tortuous-path electrodialysis spacer gasket 5 15 Total organic carbon 102 Trimsol cutting oil 569,582 Two-electrode technique 381

U Ultimate degradation 104 Ultrapure water production 523 Underground corrosion 365

V Varta process 147 Vinyl chloride detection 294 Voltage limited taper chargers 190

W Wastewater treatment 7,45,77, 98, 103,522,601,702 Water degasification 6 12 Water photoelectrolysis cell 430

Z Zindair cells 135,2 16 Zinchubon batteries 134 Zinc powder anodes 169,176 Zinc winning electrolysis 704