European Federation of Corrosion Publications NUMBER 38...

European Federation of Corrosion PublicationsNUMBER 38

Corrosion ofreinforcement in

concrete

Mechanisms, monitoring,inhibitors and rehabilitation

techniques

Edited byM. Raupach, B. Elsener, R. Polder and J. Mietz

Published for the European Federation of Corrosionby Woodhead Publishing and Maney Publishing

on behalf ofThe Institute of Materials, Minerals & Mining

CRC PressBoca Raton Boston New York Washington, DC

W O O D H E A D P U B L I S H I N G L I M I T E DCambridge England

i

© 2007, Institute of Materials, Minerals and Mining

Woodhead Publishing Limited and Maney Publishing Limited on behalf ofThe Institute of Materials, Minerals & Mining

Published by Woodhead Publishing Limited, Abington Hall, AbingtonCambridge CB21 6AH, Englandwww.woodheadpublishing.com

Published in North America by CRC Press LLC, 6000 Broken Sound Parkway, NW,Suite 300, Boca Raton, FL 33487, USA

First published 2007 by Woodhead Publishing Limited and CRC Press LLC© 2007, Institute of Materials, Minerals and MiningThe authors have asserted their moral rights.

This book contains information obtained from authentic and highly regarded sources.Reprinted material is quoted with permission, and sources are indicated. Reasonableefforts have been made to publish reliable data and information, but the authors andthe publishers cannot assume responsibility for the validity of all materials. Neitherthe authors nor the publishers, nor anyone else associated with this publication, shallbe liable for any loss, damage or liability directly or indirectly caused or alleged to becaused by this book.

Neither this book nor any part may be reproduced or transmitted in any form or byany means, electronic or mechanical, including photocopying, microfilming andrecording, or by any information storage or retrieval system, without permission inwriting from the Woodhead Publishing Limited.

The consent of Woodhead Publishing Limited does not extend to copying forgeneral distribution, for promotion, for creating new works, or for resale. Specificpermission must be obtained in writing from Woodhead Publishing Limited for suchcopying.

Trademark notice: Product or corporate names may be trademarks or registeredtrademarks, and are used only for identification and explanation, without intent toinfringe.

British Library Cataloguing in Publication DataA catalogue record for this book is available from the British Library.

Library of Congress Cataloging in Publication DataA catalog record for this book is available from the Library of Congress.

Woodhead Publishing ISBN-13: 978-1-84569-210-0 (book)Woodhead Publishing ISBN-10: 1-84569-210-1 (book)Woodhead Publishing ISBN-13: 978-1-84569-228-5 (e-book)Woodhead Publishing ISBN-10: 1-84569-228-4 (e-book)CRC Press ISBN-13: 978-1-4200-4401-0CRC Press ISBN-10: 1-4200-4401-XCRC Press order number: WP4401

The publishers’ policy is to use permanent paper from mills that operate asustainable forestry policy, and which has been manufactured from pulpwhich is processed using acid-free and elementary chlorine-free practices.Furthermore, the publishers ensure that the text paper and cover board usedhave met acceptable environmental accreditation standards.

Typeset by Replika Press Pvt Ltd, IndiaPrinted by T J International Limited, Padstow, Cornwall, Englandii


http://www.woodheadpublishing.com

Contents

Contributor contact details xi

Series introduction xvii

Volumes in the EFC series xix

Preface xxiii

1 Corrosion of metals in contact with mineral buildingmaterials 1U. NÜRNBERGER, University of Stuttgart, Germany

1.1 Corrosion behaviour during contact with buildingmaterials that contain cement 1

1.2 Reaction in case of contact with aqueous cementsolutions and alkaline waters 6

1.3 Corrosion performance in chloride containing alkalinebuilding materials 7

1.4 Corrosion behaviour during contact with buildingmaterials containing magnesia cement 8

1.5 Corrosion behaviour during contact with gypsum products 81.6 References 9

2 Corrosion and electrochemistry of zinc in alkalinesolutions and in cement mortar 10K. VIDEM, University of Oslo, Norway

2.1 Introduction 102.2 Experimental methods 112.3 Results 122.4 Discussion 172.5 Conclusions 242.6 References 25

iii


3 Corrosion behaviour of galvanized steel rebars in thepresence of coating discontinuities 27T. BELLEZZE, R. FRATESI, and F. TITTARELLI, Università Politecnicadelle Marche, Italy

3.1 Introduction 273.2 Experimental methods 283.3 Results and discussion 293.4 Conclusions 363.5 References 36

4 Influence of scale and rust on steel activation inmodel concrete pore solution 38P. NOVÁK, R. MALÁ and M. KOURIL, Institute of ChemicalTechnology, Prague, Czech Republic

4.1 Introduction 384.2 Experimental methods 384.3 Results 394.4 Conclusions 434.5 Acknowledgements 434.6 References 43

5 The surface of iron and Fe10Cr alloys in alkaline media 44A. ROSSI, G. PUDDU and B. ELSENER, University of Cagliari, Italy

5.1 Introduction 445.2 Experimental methods 455.3 Results 465.4 Discussion 545.5 Conclusions 595.6 Acknowledgements 605.7 References 60

6 Risk of galvanic corrosion induced by CFRPstrengthening in reinforced concrete 62L. BERTOLINI, M. GASTALDI and M. P. PEDEFERRI, Politecnico diMilano, Italy

6.1 Introduction 626.2 Experimental procedure 636.3 Results and discussion 656.4 Conclusions 726.5 Acknowledgements 736.6 References 74

Contentsiv


7 Macrocell corrosion of steel in concrete –experiments and numerical modelling 75S. JÄGGI and H. BÖHNI, ETH Zürich, Switzerland and B. ELSENER,ETH Zürich and University of Cagliari, Italy

7.1 Introduction 757.2 Experimental methods 777.3 Experimental and modelling results 787.4 Discussion 837.5 Conclusions 877.6 References 87

8 Modelling of chloride-induced corrosion ofreinforcement in cracked high-performanceconcrete based on laboratory investigations 89M. RAUPACH and C. DAUBERSCHMIDT, Aachen University, Germany

8.1 Background 898.2 Test programme 908.3 Results 918.4 Numerical simulation 1008.5 Conclusions 1038.6 References 104

9 Influence of stray currents on corrosion of steel inconcrete 105L. BERTOLINI, M. CARSANA and P. PEDEFERRI, Politecnico di Milano, Italy

9.1 Introduction 1059.2 Experimental tests 1079.3 Results and discussion 1099.4 Conclusions 1179.5 References 119

10 Assessment and monitoring of corrosion velocity ofrebars and prestressing cables of a bridge 120D. BINDSCHEDLER, Swiss Society for Corrosion Protection, Switzerland

10.1 Introduction 12010.2 Results of detailed corrosion inspection 12010.3 Repair 12210.4 Monitoring system 12210.5 Results of the monitoring 12410.6 Corrosion velocity and further service of the bridge 13010.7 Conclusions 13110.8 References 132

Contents v


11 On-line monitoring of corrosion in reinforcedconcrete structures 133Y. SCHIEGG, L. AUDERGON and H. BÖHNI, ETH Zürich, Switzerlandand B. ELSENER, ETH Zürich and University of Cagliari, Italy

11.1 Introduction 13311.2 Instrumentation 13311.3 Field tests 13511.4 Modelling of the temperature dependence of RW and Icorr 13511.5 Results and discussion 13811.6 Conclusions 14511.7 Acknowledgements 14511.8 References 145

12 Integrated system for corrosion monitoring ofreinforced concrete structures 146U. SCHNECK, T. WINKLER and S. MUCKE, Concrete ImprovementTechnologies, Germany

12.1 The task 14612.2 The solution 14612.3 The displays 15312.4 Monitoring in use – the results 15412.5 Acknowledgements 15712.6 References 157

13 Use of portable equipment to determine thecorrosion state of concrete structures 159R. BÄßLER and A. BURKERT, Federal Institute for Materials Researchand Testing, Germany and T. FRØLUND and O. KLINGHOFFER, ForceTechnology, Denmark

13.1 Introduction 15913.2 Background 15913.3 Experimental setup 16213.4 Results 16213.5 Discussion 16713.6 Conclusions 16813.7 Outlook 16813.8 Acknowledgements 16913.9 References 169

Contentsvi


14 Corrosion inhibitors for reinforced concrete – an EFCstate of the art report 170B. ELSENER, University of Cagliari, Italy and ETH Zürich, Switzerland

14.1 Introduction 17014.2 Mode of action of corrosion inhibitors 17214.3 Corrosion inhibitors to prevent or delay corrosion initiation 17314.4 Corrosion inhibitors to reduce the propagation rate of

corrosion 17614.5 Field tests with corrosion inhibitors 17814.6 Transport of the inhibitor into mortar or concrete 17914.7 Critical evaluation of corrosion inhibitors 18114.8 Conclusions 18214.9 References 182

15 Mixed-in inhibitors for concrete structures 185F. BOLZONI, G. FUMAGALLI, L. LAZZARI, M. ORMELLESE andM. P. PEDEFERRI, Politecnico di Milano, Italy

15.1 Introduction 18515.2 Service life 18715.3 Experimental methods 18815.4 Results 19015.5 Discussion 19615.6 Conclusions 20015.7 References 201

16 Effectiveness of mixed-in organic corrosion inhibitorson extending the service life of reinforced concretestructures 203R. CIGNA, Consultant, Italy, A. MERCALLI, Autostrade S.p.A, Italy,L. GRISONI, Sika Italia, Italy and U. MÄDER, Sika A. G., Switzerland

16.1 Introduction 20316.2 Experimental methods 20416.3 Discussion and conclusions 20416.4 References 210

17 Migrating inhibitors on corrosion in reinforced concrete 211F. BOLZONI, G. FUMAGALLI, L. LAZZARI, M. ORMELLESE andM. P. PEDEFERRI, Politecnico di Milano, Italy

17.1 Introduction 21117.2 Experimental methods 21217.3 Results and discussion 21417.4 Conclusions 22217.5 References 223

Contents vii


18 Effectiveness of corrosion inhibitors – a field study 226Y. SCHIEGG, F. HUNKELER and H. UNGRICHT, Swiss Society forCorrosion Protection (SGK) and Technical Research and Consultingon Cement and Concrete (TFB), Switzerland

18.1 Introduction 22618.2 Field study in Naxbergtunnel 22618.3 Investigation 22718.4 Results 22918.5 Conclusions 23718.6 Acknowledgements 23818.7 References 238

19 Corrosion protection of steel rebar in concrete usingmigrating corrosion inhibitors 239B. BAVARIAN and L. REINER, California State University, USA

19.1 Introduction 23919.2 Experimental procedures 24119.3 Results and discussion 24219.4 Conclusions 24419.5 References 248

20 Determination of coating permeability on concreteusing EIS 250J. VOGELSANG, G. MEYER and M. BEPOIX, Sika GmbH, Germany

20.1 Introduction 25020.2 Experimental design 25220.3 Results and discussion 25620.4 Conclusions 26020.5 References 261

21 Chloride extraction from reinforced concrete – a newdefined way of application 263U. SCHNECK, T. WINKLER and H. GRÜNzIG, Concrete ImprovementTechnologies, Germany

21.1 The task 26321.2 The solution 26321.3 Description of configuration 26721.4 Application to a highway bridge abutment 26821.5 Results of the follow-up survey 27221.6 Conclusions 27321.7 Acknowledgements 27521.8 References and further reading 275

Contentsviii


22 Microscopy study of the interface between concreteand the conductive coating used as an anode forcathodic protection 277R. B. POLDER and W. H. A. PEELEN, TNO Building and ConstructionResearch, The Netherlands and J. LEGGEDOOR and G. SCHUTEN,Leggedoor Concrete Repair, The Netherlands

22.1 Introduction 27722.2 Theoretical background 27722.3 Samples and microscopy examination 27922.4 Results 28022.5 Discussion 28522.6 Conclusions 28622.7 Acknowledgements 28722.8 References 287

23 Protection of reinforced concrete piles in marinestructures with sacrificial anodes 288L. BERTOLINI, M. GASTALDI, M. PEDEFERRI and E. REDAELLI,Politecnico di Milano, Italy

23.1 Introduction 28823.2 Experimental procedure 28923.3 Results and discussion 29023.4 Conclusions 29523.5 References 298

24 Renovation of the cathodic protection system of aconcrete bridge after 12 years of operation 300G. SCHUTEN and J. LEGGEDOOR, Leggedoor Concrete Repair,The Netherlands and R. B. POLDER and W. H. A. PEELEN, TNOBuilding and Construction Research, The Netherlands

24.1 History 30024.2 The CP installation of the southern bicycle path (1986) 30024.3 Replacement of the northern bicycle path (1996) 30124.4 CP system behaviour in 1998 30124.5 System upgrade 1999 30224.6 Cost aspects 30424.7 Durability aspects 30424.8 Conclusions 30524.9 Acknowledgement 30524.10 References 306

Contents ix


Contributor contact details

(* = main contact)

Editors

Prof Dr-Ing M. RaupachAachen UniversityInstitute of Building MaterialsResearchSchinkelstr. 352056 AachenGermany

E-mail: [email protected]

Professor Dr B. ElsenerETH ZurichInstitute for Building MaterialsETH HönggerbergCH-8093 ZürichSwitzerland


Dr R. PolderTNO Dr Built Environment andGeosciencesPO Box 492600 AA DelftThe Netherlands


Dr-Ing J. MietzFederal Institute for MaterialsResearch and Testing (BAM)Unter den Eichen 8712205 BerlinGermany


Chapter 1

Professor U. NürnbergerMaterialprüfungsanstalt UniversitätStuttgartPfaffenwaldring 470569 StuttgartGermany


Chapter 2

Dr K. VidemCentre for Materials ScienceUniversity of OsloGaustadalleen 21N-0349OsloNorway



mailto:[email protected]







Chapter 3

Dr. T. Bellezze*, Dr R. Fratesi andDr F. TittarelliDipartimento di Fisica e Ingegneriadei Materiali e del TerritorioUniversità Politecnica delle MarcheVia Brecce Bianche60131, AnconaItaly


Chapter 4

Professor P. Novak*, Dr R. Maláand Dr M. KouřilDepartment of Metals and CorrosionEngineeringInstitute of Chemical TechnologyPragueTechnická 5CZ-166 28 Prague 6Czech Republic


Chapter 5

Professor A. Rossi*, Dr G. Pudduand Professor B. ElsenerDepartment of Inorganic andAnalytical ChemistryUniversity of CagliariI – 09042 MonserratoCagliariSardiniaItaly


Chapter 6

Dr L. Bertolini*, Dr M. Gastaldi andDr M. P. PedeferriDipartimento di Chimica, MaterialiIngegneria Chimica ‘G. Natta’Politecnico di MilanoVia Mancinelli 720131 MilanoItaly


Chapter 7

Professor B. Elsener*, Dr S. Jäggiand Professor H. BöhniETH ZürichInstitute for Building MaterialsETH HönggerbergCH-8093ZürichSwitzerland


Chapter 8

Prof Dr-Ing M. Raupach* andDr C. DauberschmidtAachen UniversityInstitute of Building MaterialsResearchSchinkelstr. 3,52056 AachenGermany


Contributor contact detailsxii









Chapter 9

Dr L. Bertolini*, Dr M. Carsana andDr P. PedeferriDipartimento di Chimica, MaterialiIngegneria Chimica ‘G. Natta’Politecnico di MilanoVia Mancinelli 720131 MilanoItaly


Chapter 10

Dr D. BindschedlerSwiss Society for CorrosionProtectionTechnoparkstrasse 1CH-8005 ZürichSwitzerland


Chapter 11

Dr Y. Schiegg*, Dr L. Audergon, DrB. Elsener and Professor H. BöhniSchweizerische Gesellschaft fürKorrosionsschutzTechnoparkstrasse 1CH-8005 ZürichSwitzerland


Chapter 12

Dr U. Schneck*, Dipl-IngT. Winkler and Dipl-Ing (FH)S. MuckeCITec Concrete ImprovementTechnologies GmbHDresdner Strasse 42D-01462 CossebaudeGermany


Chapter 13

Dr Ralph Bäßler* and Dr AndreasBurkertFederal Institute for MaterialsResearch and Testing (BAM)Unter den Eichen 87D-12205 BerlinGermany


Thomas Frølund and OskarKlinghofferForce TechnologyDK-2605 BrøndbyDenmark

Chapter 14

Professor B. ElsenerETH ZürichInstitute for Building MaterialsETH HönggerbergCH-8093 ZürichSwitzerland


Contributor contact details xiii









Chapter 15

Dr F. Bolzoni*, Dr G. Fumagalli,Dr L. Lazzari, Dr M. Ormellese andDr M. PedeferriDipartimento di Chimica, Materiali eIngegneria Chimica ‘G. Natta’Politecnico di MilanoVia Mancinelli 720131 MilanoItaly


Chapter 16

Dr R. Cigna*, Dr A. Mercalli,Dr L. Grisomi and Dr U. MäderHaward TechnologySmith MooreLLP Suite 7501355 Peachtree StreetAtlantaGA 30309-3214USA

E-mail: [email protected];[email protected]

Chapter 17

Dr F. Bolzoni*, Dr G. Fumagalli, DrL. Lazzari, Dr M. Ormellese and DrM. PedeferriDipartimento di Chimica, Materiali eIngegneria Chimica ‘G. Natta’Politecnico di MilanoVia Mancinelli 720131 MilanoItaly


Chapter 18

Dr Y. Schiegg*, Dr F. Hunkeler andDr H. UngrichtSchweizerische Gesellschaft fürKorrosionsschutzTechnoparkstrasse 1CH-8005 ZürichSwitzerland


Chapter 19

Professor B. Bavarian* and Dr L.ReinerDepartment of MSEMCalifornia State University,Northridge18111 Nordhoff StreetNorthridgeCA 91330-8332USA


Chapter 20

Dr J. Vogelsang*, Dr G. Meyer andDr M. BepoixSika Technology AGTueffenwies 16CH 8048 ZürichSwitzerland

E-mail:[email protected]

Contributor contact detailsxiv









Chapter 21

Dr U. Schneck*, Dipl-Ing T. Winklerand Dipl-Chem H. GrünzigCITec Concrete ImprovementTechnologies GmbHDresdner Strasse 42D-01462 CossebaudeGermany


Chapter 22

Dr R. Polder*, Dr W. Peelen,Dr J. Leggedoor and Dr G. SchutenTNO Built Environment andGeosciencesPO Box 492600 AA DelftThe Netherlands


Chapter 23

Dr L. Bertolini*, Dr M. Gastaldi,Dr M. Pedeferri and Dr E. RedaelliDipartimento di Chimica, MaterialiIngegneria Chimica ‘G. Natta’Via Mancinelli 720131 MilanoItaly


Chapter 24

Dr G. Schuten*, Dr J. Leggedoor,Dr R. Polder and Dr W. PeelenLeggedoorTuinstraat 58PO Box 39514 ZGGasselternijveenThe Netherlands


Contributor contact details xv







The EFC, incorporated in Belgium, was founded in 1955 with the purpose ofpromoting European co-operation in the fields of research into corrosion andcorrosion prevention.

Membership is based upon participation by corrosion societies andcommittees in technical Working Parties. Member societies appoint delegatesto Working Parties, whose membership is expanded by personal correspondingmembership.

The activities of the Working Parties cover corrosion topics associated withinhibition, education, reinforcement in concrete, microbial effects, hot gasesand combustion products, environment sensitive fracture, marine environments,surface science, physico–chemical methods of measurement, the nuclearindustry, computer based information systems, the oil and gas industry, thepetrochemical industry, coatings, automotive engineering and cathodicprotection. Working Parties on other topics are established as required.

The Working Parties function in various ways, e.g. by preparing reports,organising symposia, conducting intensive courses and producing instructionalmaterial, including films. The activities of the Working Parties are co-ordinated,through a Science and Technology Advisory Committee, by the ScientificSecretary.

The administration of the EFC is handled by three Secretariats: DECHEMAe. V. in Germany, the Société de Chimie Industrielle in France, and The Instituteof Materials, Minerals and Mining in the United Kingdom. These threeSecretariats meet at the Board of Administrators of the EFC. There is an annualGeneral Assembly at which delegates from all member societies meet todetermine and approve EFC policy. News of EFC activities, forthcomingconferences, courses etc. is published in a range of accredited corrosion andcertain other journals throughout Europe. More detailed descriptions of activitiesare given in a Newsletter prepared by the Scientific Secretary.

The output of the EFC takes various forms. Papers on particular topics, forexample, reviews or results of experimental work, may be published in scientificand technical journals in one or more countries in Europe. Conference

European Federation of CorrosionPublications: Series introduction

xvii


proceedings are often published by the organisation responsible for theconference.

In 1987 the, then, Institute of Metals was appointed as the official EFCpublisher. Although the arrangement is non-exclusive and other routes forpublication are still available, it is expected that the Working Parties of theEFC will use The Institute of Materials, Minerals and Mining for publicationof reports, proceedings etc. wherever possible.

The name of The Institute of Metals was changed to The Institute of Materialson 1 January 1992 and to the Institute of Materials, Minerals and Mining witheffect from 26 June 2002. The series is now published by Woodhead Publishingand Maney Publishing on behalf of the Institute of Materials, Minerals andMining.

P. McIntyreEFC Series Editor,The Institute of Materials, Minerals and Mining, London, UK

EFC Secretariats are located at:

Dr B. A. RickinsonEuropean Federation of Corrosion, The Institute of Materials, Minerals andMining, 1 Carlton House Terrace, London, SW1Y 5DB, UK

Dr J. P. BergeFédération Europénne de la Corrosion, Société de Chimie Industrielle,28 rue Saint Dominique, F-75007 Paris, FRANCE

Professor Dr G. KreysaEuropäische Föderation Korrosion, DECHEMA e. V., Theodor-Heuss-Allee25, D-60486, Frankfurt, GERMANY

EFC series introductionxviii


Volumes in the EFC series

1 Corrosion in the nuclear industryPrepared by the Working Party on Nuclear Corrosion

2 Practical corrosion principlesPrepared by the Working Party on Corrosion Education (Out of print)

3 General guidelines for corrosion testing of materials for marineapplicationsPrepared by the Working Party on Marine Corrosion

4 Guidelines on electrochemical corrosion measurementsPrepared by the Working Party on Physico-Chemical Methods ofCorrosion Testing

5 Illustrated case histories of marine corrosionPrepared by the Working Party on Marine Corrosion

6 Corrosion education manualPrepared by the Working Party on Corrosion Education

7 Corrosion problems related to nuclear waste disposalPrepared by the Working Party on Nuclear Corrosion

8 Microbial corrosionPrepared by the Working Party on Microbial Corrosion

9 Microbiological degradation of materials – and methods ofprotectionPrepared by the Working Party on Microbial Corrosion

10 Marine corrosion of stainless steels: chlorination and microbialeffectsPrepared by the Working Party on Marine Corrosion

11 Corrosion inhibitorsPrepared by the Working Party on Inhibitors (Out of print)

xix


12 Modifications of passive filmsPrepared by the Working Party on Surface Science and Mechanisms ofCorrosion and Protection

13 Predicting CO2 corrosion in the oil and gas industryPrepared by the Working Party on Corrosion in Oil and Gas Production(Out of print)

14 Guidelines for methods of testing and research in high temperaturecorrosionPrepared by the Working Party on Corrosion by Hot Gases andCombustion Products

15 Microbial corrosion (Proc. 3rd Int. EFC workshop)Prepared by the Working Party on Microbial Corrosion

16 Guidelines on materials requirements for carbon and low alloysteels for H2S-containing environments in oil and gas productionPrepared by the Working Party on Corrosion in Oil and Gas Production

18 Stainless steel in concrete: state of the art reportPrepared by the Working Party on Corrosion of Reinforcement inConcrete

19 Sea water corrosion of stainless steels – mechanisms andexperiencesPrepared by the Working Parties on Marine Corrosion and MicrobialCorrosion

20 Organic and inorganic coatings for corrosion prevention – researchand experiencesPapers from EUROCORR ‘96

21 Corrosion-deformation interactionsCDI ‘96 in conjunction with EUROCORR ‘96

22 Aspects of microbially induced corrosionPapers from EUROCORR ‘96 and the EFC Working Party on MicrobialCorrosion

23 CO2 Corrosion control in oil and gas production – designconsiderationsPrepared by the Working Party on Corrosion in Oil and Gas Production

24 Electrochemical rehabilitation methods for reinforced concretestructures – a state of the art reportPrepared by the Working Party on Corrosion of Reinforcement inConcrete

Volumes in the EFC seriesxx


25 Corrosion of reinforcement in concrete – monitoring, preventionand rehabilitationPapers from EUROCORR ‘97

26 Advances in corrosion control and materials in oil and gasproductionPapers from EUROCORR ‘97 and EUROCORR ‘98

27 Cyclic oxidation of high temperature materialsProceedings of an EFC Workshop, Frankfurt/Main, 1999

28 Electrochemical approach to selected corrosion and corrosioncontrol studiesPapers from 50th ISE Meeting, Pavia, 1999

29 Microbial corrosion (Proceedings of the 4th international EFCworkshop)Prepared by the Working Party on Microbial Corrosion

30 Survey of literature on crevice corrosion (1979–1998): mechanisms,test methods and results, practical experience, protective measuresand monitoringPrepared by F. P. Ijsseling and the Working Party on Marine Corrosion

31 Corrosion of reinforcement in concrete: corrosion mechanisms andcorrosion protectionPapers from EUROCORR ‘99 and the Working Party on Corrosion ofReinforcement in Concrete

32 Guidelines for the compilation of corrosion cost data and for thecalculation of the life cycle cost of corrosion – a working partyreportPrepared by the Working Party on Corrosion in Oil and Gas Production

33 Marine corrosion of stainless steels: testing, selection, experience,protection and monitoringEdited by D. Féron

34 Lifetime modelling of high temperature corrosion processesProceedings of an EFC Workshop 2001. Edited by M. Schütze, W. J.Quadakkers and J. R. Nicholls

35 Corrosion inhibitors for steel in concretePrepared by B. Elsener with support from a Task Group of WorkingParty 11 on Corrosion of Reinforcement in Concrete

36 Prediction of long term corrosion behaviour in nuclear wastesystemsEdited by D. Féron of Working Party 4 on Nuclear Corrosion

Volumes in the EFC series xxi


37 Test methods for assessing the susceptibility of prestressing steels tohydrogen induced stress corrosion crackingby B. Isecke of EFC WP11 on Corrosion of Reinforcement in Concrete

38 Corrosion of reinforcement in concrete: mechanisms, monitoring,inhibitors and rehabilitation techniquesEdited by M. Raupach, B. Elsener, R. Polder and J. Mietz

39 The use of corrosion inhibitors in oil and gas productionEdited by J. W. Palmer, W. Hedges and J. L. Dawson

40 Control of corrosion in cooling watersEdited by J. D. Harston and F. Ropital

41 Corrosion by carbon and nitrogen: metal dusting, carburisationand nitridationM. Schutze and H. Grabke

42 Corrosion in refineriesJ. Harston

43 The electrochemistry and characteristics of embeddable referenceelectrodes for concretePrepared by R. Myrdal on behalf of Working Party 11 on Corrosion ofSteel in Concrete

44 The use of electrochemical scanning tunnelling microscopy (EC–STM) in corrosion analysis: reference material and proceduralguidelinesPrepared by R. Lindström, V. Maurice, L. H. Klein and P. Marcus onbehalf of Working Party 6 on Surface Science

Volumes in the EFC seriesxxii


Preface

In 2001 the European Corrosion Conference, Eurocorr 2001, was organisedon behalf of the European Federation of Corrosion (EFC) by the AssoziationeItaliana di Metallurgia (AIM), a member society of the EFC, in Riva delGarda. Each of the 17 working parties of EFC prepared sessions on particulartopics in their respective fields of corrosion and corrosion protection.

Jürgen Mietz, the chair of Working Party (WP) 11 at that time, organisedsessions on corrosion of the reinforcement in concrete. Altogether 27 paperswere selected for oral presentation from many more abstracts in this field.The main topics were corrosion mechanisms, corrosion measurement methods,assessment and monitoring, inhibitors and electrochemical protection methods.

During the annual meeting of WP 11 it was concluded that the quality ofthese presentations was excellent and that the papers should be publishedtogether as an EFC book. Unfortunately, a number of problems meant thatpublication of the papers was delayed. In 2006 these problems were solved,allowing quick publication of the papers. As the papers still present currentknowledge in the field, the editors have no doubt in recommending publication,even after five years.

I am sure that this publication is a useful tool for all people interested inthe field of corrosion of reinforcement in concrete and hope that all readerswill receive new information, insights and ideas.

Prof Dr-Ing Michael RaupachChairman of EFC Working Party 11

xxiii


1

1.1 Corrosion behaviour during contact with

building materials that contain cement [1–5]

1.1.1 Iron and steel

In sufficiently moist oxygen-containing aqueous media that are nearly neutralto weakly basic, iron is transformed into iron(II) hydroxide with the help ofwater and oxygen and is subsequently oxidised into iron(III) hydroxide (rust,FeOOH). These rust layers have no corrosion inhibiting character at all and,therefore, iron and steel are extremely sensitive to corrosion in the mediumpH-region of the iron–water system. (Fig. 1.1).

1Corrosion of metals in contact with

mineral building materials

U. N Ü R N B E R G E R, University of Stuttgart, Germany

Co

rro

sio

n r

ate

Iron

Zinc

Lead

Aluminium

Copper

2 4 6 8 10 12 14 16pH-value

1.1 Effect of pH on corrosion rate of metals; reference data from astructural point of view [1].


Corrosion of reinforcement in concrete2

In Portland cement-based concrete, steel is protected against corrosionbecause of contact with the highly alkaline concrete pore water. As a resultof the strongly alkaline reaction during the hydration of cement, concrete hasa large proportion of alkaline ingredients. The pH-value of the aqueousphase of normal concrete is about 12.6 to 13.8 depending on the content ofingredients in the cement that have strongly basic characteristics (K2O, Na2O).In the region of pH ≥ 11.5, the steel surface is passive if there are no ingredientsthat are aggressive to steel and destroy passivity (causing pitting corrosion);there is, therefore, complete inhibition of the anodic partial reaction. Thepassivating layer of hydrated iron oxide is 2–20 nm thick. The corrosion-protective effect of the concrete for embedded steel is lost if the pH fallsshort of the above-mentioned value. However, it is not until below pH 9.0that severe corrosion starts, as, for example, after carbonation of the concreteand the neutralisation of all alkaline ingredients. In addition to this, thecement must contain as much free water in its pores as possible and oxygen fromthe structural element must get through the concrete to the steel, by diffusion.

1.1.2 Aluminium [6]

The generally good corrosion behaviour of aluminium with its very negativestandard electrode potential (–1.66 VH) results from the development ofpassivating oxide and hydroxide surface films. In the pH range 4 to 9, thesefilms are largely insoluble. Because of this, aluminium materials aredistinguished by good resistance to corrosion in nearly neutral to weaklyacidic aqueous media and also in humid atmospheric conditions. This explainsthe wide application of aluminium in constructional engineering.

Aluminium is an amphoteric metal. Because of this, the protective effectof the coating is lost as a result of its disintegration in strongly acidic andalkaline media. Aluminium and its alloys are mainly attacked by generalcorrosion in these pH regions and there is no question of applying thesematerials in such cases. However, noticeable disintegration can take placeeven in the more weakly alkaline region above pH 9 (Fig. 1.1). Aluminiummaterials then become as active as might be expected from their position inthe electrochemical series and, even in the absence of oxygen, react to evolvehydrogen and form soluble aluminates:

2Al + 6NaOH + 6H2O Æ 2Na3[Al(OH)6] + 3H2 (1.1)

Therefore, if there is prolonged contact with moist Portland cement-basedbuilding materials, aluminium and its alloys are attacked through a reactionwith the free alkali hydroxides of the cement solution within the pores becauseof general corrosion.

In Fig. 1.2 (on the left-hand side), the corrosion reaction of aluminium iscompared with that of other structural metals in wet cement mortar.


Corrosion of metals in contact with mineral building materials 3

In cement mortar and concrete, corrosion increases with increasing moisture.Considerable loss of mass occurs especially in the case of wet storage (Fig.1.3). Extremely voluminous corrosion products develop, that can burst andspall-off the concrete cover. Account must be taken of the fact that, dependingon time spent in formwork and the dry-out conditions, hardening buildingmaterials often evolve their excess water very slowly, so that considerablecorrosion damage can occur in fresh concrete before a low equilibrium moisturelevel is achieved. According to the information given in reference 5, therelative extent of corrosion in Portland cement mortar is in the ratio of1 : 6 : 25 for dry, moist, and wet states, respectively. Figure 1.3 shows thatthe intensity of attack decreases with time, since the corrosion products thatdevelop impede the transport of alkalis to the corroding surface.

Anodically produced anodisation layers (Eloxal) are also attacked by moistalkaline-reacting building materials and under these circumstances do notoffer effective protection. To preclude corrosion damage through contactwith moist alkaline building materials, aluminium must, in addition, beprotected by proper organic coatings.

In metals that are susceptible to alkalis, such as aluminium and its alloys,

PC-mortarPC-mortar +

1.6% Cl–(CaCl2) Gypsum plaster

1 – Steel2 – Aluminium

99.33 – Copper4 – Lead5 – Zinc

Liquidcement mortarin g m–2:50 95 45

1190 955

2

5

Water 30 cm

2

1 3 5

41

4

3

15

2 3 4

Mas

s w

eig

ht

loss

(g

/m2 )

600

500

400

300

200

100

1.2 Loss of mass of metals after twelve months of storage in moistbuilding materials (the mortar blocks were immersed to a depth of2 cm in the water during the tests) and in liquid cement mortar(numerical data) (in [1] according to the results of [2]).



the corrosion risk increases as the pH value of the building material increases.This means that the corrosion of aluminium can be limited by the choice ofa suitable bonding agent (Fig. 1.4). Concretes, that, depending on theirproduction route are only mildly alkaline, e.g. autoclave-treated pore concrete(gas concrete), are not to be classified as aggressive in respect of structuralmetals that are susceptible to alkalis, such as aluminium. For aluminium,carbonation of the building material will also have a corrosion-protectiveeffect.

1.1.3 Copper

On one hand, the excellent corrosion resistance of the copper-based materialsdepends on the ‘noble’ character (the standard electrode potential of copperis +0.34 VH) and on the inability to form protective layers in many normalenvironments and on contact with building materials. In water and neutralsalt solutions, copper-based materials have very good resistance to corrosionover a wide pH range (Fig. 1.1). In diluted (non-oxidising) acids and in thealkaline region, copper, above all, is superior to other non-ferrous metals.Copper and its alloys are unapplicable only if the formation of the protective

Al Mg Si 1Al Mg 3

Fresh PC-mortar

Wet exposureHardenedPC-concrete

Moist exposure

Dry exposure

0 3 6 9 12Exposure period (month)

Loss

in

th

ickn

ess

(mm

)

400

300

200

100

0

1.3 Corrosion behaviour of aluminium alloys in alkaline buildingmaterials resp. media (in [1] according to the results of [2]).wet exposure: concrete, partially immersed in water,moist exposure: concrete in 95% relative humidity,dry exposure: concrete in 65% relative humidity.



layers is hampered and the material is heavily attacked through the formationof complex salts, e.g. in contact with ammonia and ammonia-containingsolutions.

The layers of slightly soluble copper(I) oxide, that form in air, are virtuallyinsoluble in alkalis. Copper and its alloys therefore undergo negligible uniformcorrosion when embedded in moist concrete or cement mortar (Fig. 1.2, left-hand side and Fig. 1.4). When applied in cements with higher alkality (pHvalue of the cement pore solution >13.3), copper, especially brasses (e.g.CuZu37), that are rich in zinc, are not sufficiently resistant to corrosion.

1.1.4 Zinc [7–10]

Zinc, like aluminium, is characterised by a negative standard electrode potential(–0.76 VH) which renders it thermodynamically susceptible to corrosion.Zinc, however, also forms protective films of solid corrosion products inmany media, including building materials, by reaction with its environment.

Zinc is also an amphoteric metal and, therefore, is not resistant to corrosionin both acid (< pH 5) and alkaline regions (> pH 12) (Fig. 1.1). In morealkaline solutions, zinc hydroxide, which is transformed into readily solubleand non-protective zincates by reaction with the alkaline compound, developsunder the formation of hydrogen. In alkaline concrete with pore solution pHvalues of between 12.6 and 13.8, for alkali-enriched cements, galvanisationwould be expected to be susceptible to corrosion because of the amphotericreaction. However, in fact it is found that, at least for pH values £ 13.3, the

1.4 Loss of mass of metals after twelve months of storage in wetmortar with various bonding agents (testing arrangement as shownin Fig. 1.2) (in [1] according to the results of [2]).

1 = Portland cement2 = Hydraulic lime3 = Trass cement4 = Portland blast-furnace

cement5 = High-alumina6 = Gypsum

Lead

Iron

Aluminium

ZincCopper

Cement 1 2 3 4 5 6pH 12.7 12.6 12.3 12.3 11.7 8.7

Mas

s w

eig

ht

loss

(g

m–2

)

250

200

150

100

50

0



dissolution rate of zinc under the formation of hydrogen quickly diminishesbecause of passivation. If Ca(OH)2 exists together with Zn(OH)2, a furthercorrosion product, the slightly soluble calcium hydroxozincate Ca[Zn(OH)3]2

· 2H2O, develops, which is held responsible by some researchers for thepassivation of zinc in concrete. Because of this, zinc is uniformly attacked inalkaline building materials to a somewhat greater extent than copper materials,but much less than aluminium and lead (Fig. 1.2 left-hand side and Fig. 1.4).Above pH 13.3, the ease of passivation becomes more and more restrictedwith rising alkality, and the corrosion of zinc increases greatly. In carbonatedconcrete, the corrosion rate of zinc can, indeed, be a little higher than inalkaline concrete, but it is still considerably below that of e.g. steel. Becauseof that, galvanised reinforcing steels may be applied where prematurecarbonation is expected.

Heightened contents of chromium in the cements have a beneficial effecton the corrosion of zinc in the alkaline region because they promote rapidinitial passivation of the zinc.

1.1.5 Lead [11]

Lead is relatively ‘noble’ in regard to corrosion, because of its position in theelectrochemical series; the standard electrode potential is –0.13 VH. However,lead also owes its good corrosion resistivity to the ability to form impervious,tightly adherent and slightly soluble coatings made of lead compounds,depending on the corrosive medium.

Lead ranks among the amphoteric metals that can be dissolved not only inacids, but also in alkalis (Fig. 1.1). In alkaline electrolytes, lead is heavilyattacked above pH 9:

Pb + 1/2O2 + Ca(OH)2 Æ CaPbO2 + H2O (1.2)

Thus, primarily developed lead hydroxides are transformed into readilysoluble plumbate, mainly calcium plumbate. Because of that, lead, incomparison to aluminium, is very susceptible to corrosion in moist alkalinebuilding materials (Fig. 1.2, left-hand side). Intensified attack results as thepH value of the aqueous phase of the building material increases (Fig. 1.4).The high rate of attack of the mainly uniform corrosion decreases with timeand as the moisture level of the building material falls.

In wet concrete or mortar, lead can be protected, e.g. by insulation withthick bituminous coatings, plastic sheets or the like.

1.2 Reaction in case of contact with aqueous

cement solutions and alkaline waters [1, 2]

In practical constructions, it occasionally happens that structural metals becomemoistened by aqueous cement solutions (fresh concrete) or by aqueous extracts



that have been in contact with hardened mortar or concrete for a longer time.The latter may contain components of the cement in a dissolved state, andbehaves in an alkaline manner. In such media, aluminium materials and leadin particular suffer much worse attack than in moist, solid building materials(see the numerical data in Fig. 1.2), in the course of which the corrosion ratehardly slows down. In zinc and copper as well, a much stronger attack thanin hardened wet building material can be determined. Thus the corrosionreaction in alkaline solutions can not be compared with that in solid phasesof a building material. In the case of contact of the metals with alkaline-reacting electrolytes, irregular alterations in color can occur even after short-term attack and, under certain circumstances, this can strongly affect theexterior of construction elements.

1.3 Corrosion performance in chloride containing

alkaline building materials [1–3, 6]

Chloride ions in sufficiently moist mortar/concrete, normally resulting fromsalt penetration and of more than 0.5 to 1.0% content by mass (referred tothe cement content), will cause extensive pitting corrosion of steels embeddedin it and in a passive state (Fig. 1.2). In addition to the chloride content, theintensity of chloride-induced corrosion depends on other parameters of theconcrete (pH value and kind of steel, aeration and water content).

In structural elements made of zinc or in galvanized items with otherwisealmost passive behaviour, a small increase of corrosion in alkaline buildingmaterials must be expected if the chloride content exceeds about 1.5% bymass related to the weight of cement. The higher critical content of chloridecompared with steel results from the fact that the chloride ions are partiallybound as slightly soluble basic zinc chlorides.

In the presence of chlorides in the concrete/mortar, serious susceptibilityto widespread pitting exists for aluminium, which in any case is sensitive tocorrosive attack. The presence of chloride can intensify the corrosion inalkaline building materials by a factor of several times (Fig. 1.2). Even thesmallest additions of chloride are corrosion-promoting.

Lead, which is also strongly corrodible in moist alkaline building materials,does not sustain further aggravation of corrosion in the presence of additionalchlorides (Fig. 1.2). This can be explained by the formation of slightlysoluble reaction products with a protective effect.

Copper is also largely unsusceptible to the influence of chloride salts,since the primarily developing copper(I) chloride is only slightly soluble.




building materials containing magnesia

cement [2,12]

A number of building materials, such as stone wood (for flooring) and lightbuilding boards, are made of wood-shavings that contain magnesia cementas a binding agent. Magnesia cement is tempered by mixing lightly burntmagnesite and concentrated magnesium chloride solution and hardened bythe formation of a compound of low solubility (probably 3MgO·MgCl2·11H2O).The pores of the building material are filled with a magnesium chloridesolution. If the building materials that are bound with magnesia cementcome into contact with metals they can heavily attack them. Above all, iron(steel), aluminium and zinc, including galvanised steel, and, to a lesser degree,copper and lead, are attacked.

The distinct aggressiveness of such building materials is explained by thestrong hygroscopic character of magnesium chloride (pores containingmagnesium chloride do not dry up at relative humidities >32%) on the onehand, and the corrosion-promoting characteristics of concentrated chloridesolutions on the other. Therefore, corrosion is possible even in a comparativelydry environment.

In the case of a reaction between the moisture that is always present instone wood floorings and magnesium chloride, e.g. in warm pipe walls,hydrochloric acid is split off and the pH value is reduced. In this way, veryaggressive corrosion conditions are created for steel pipes and, therefore,pipes made of unalloyed steels or galvanised steels may not be laid in stonewood floorings without external protection.


gypsum products [1–3, 6, 12]

Structural gypsum (e.g. gypsum mortar) that is mixed with water forms acompound of needle-shaped dihydrate crystals CaSO4·2H2O, as soon as thefluid pulp reacts. Because of the unusually high surplus water resulting fromfresh gypsum, the porosity of the hardened building material is quite high. Ifthe hardened gypsum products (gypsum plasters, gypsum pasteboards, gypsumwall-construction boards) are kept moist, the pores are filled with a saturatedcalcium sulphate solution. Since this salt has a corrosion stimulating effectin neutral building materials, gypsum and gypsum mortar attack zinc andiron (or steel) very strongly in combination with humidity (Fig. 1.2 on theright-hand side and Fig. 1.4). Steel pipes and galvanised steel pipes that arein contact with gypsum, which is moistened long-term, are attacked by thickrust products and can be destroyed even after a few years. At relative humiditiesof <99%, gypsum mortar completely drains with time and no longer causescorrosion of steel and zinc.



Aluminium materials and lead in general are not likely to be attacked bythe more neutrally reacting gypsum building materials. In the case of lead,gypsum forms slightly soluble lead sulphates, that hamper surface corrosion.In aluminium that is free of copper, moist gypsum promotes limited pittingcorrosion. However, aluminium alloys that contain copper sometimes corrodequite intensively in wet gypsum.

Copper materials are also largely resistant to gypsum because the surfacesare coated with an oxide film that is stable to sulphate.

1.6 References

1. U. Nürnberger, Korrosion und Korrosionsschutz im Bauwesen. Bauverlag Wiesbaden,1995.

2. A. Bukowiecki, Über das Korrosionsverhalten von Eisen- und Nichteisenmetallengegenüber verschiedenen Zementen und Mörteln. Schweizer Archiv, 1965, 31, 273–293.

3. W. Wiederholt and J. Sonntag, Korrosion von Metallen im Bauwesen. Berichte ausder Bauforschung, 1965, 44, 1–62.

4. H. Woods, Corrosion of embedded material other than reinforcing steel. Res. Dev.Lab. PCA Bull. 1966, 198, 230–238; s. a. Zement-Kalk – Gips, 1967, 120–122.

5. O. V. Franqué and W. Huppatz, Korrosionsverhalten von Bauteilen ausNichteisenmetallen bei Berührung mit Baustoffen. Werkstoffe Korrosion, 1986, 37,318–322.

6. E. Fischer and H. Voßkühler, Verhalten von Aluminiumlegierungen gegenüberMörtelmischungen. Aluminium, 1957, 33, 602–612.

7. H. Kaesche, Zum Elektrodenverhalten des Zinks und des Eisens inCalciumhydroxidlösung und in Mörtel. Werkstoffe und Korrosion, 1969, 20, 119–124.

8. C. Andrade, J. D. Holst, U. Nürnberger, J. D. Whiteley and N. Woodman, Protectionsystems for reinforcement. Bull. ‘Information No 211’, CEB, Lausanne, 1992.

9. K. Menzel, Zur Korrosion von verzinktem Stahl in Kontakt mit Beton. DissertationUniversität Stuttgart, 1992.

10. U. Nürnberger, Besondere Maßnahmen für den Korrosionsschutz und zur Sanierungvon Stahlbeton und Spannbeton. Schriftenreihe Otto-Graf-Institut Stuttgart, 1988,79.

11. W. Hoffmann and R. Reinert, Korrosionsverhalten von Blei. In: D. Grimme, K. A.Von Oeteren, M. Pötzschke and W. Schwenk, Korrosion und Korrosionsschutzmetallischer Werkstoffe im Hoch- und Ingenieurbau. Verlag Stahleisen mbH Düsseldorf,1976, 235–246.

12. U. Nürnberger, Korrosionsverhalten von feuerverzinktem Stahl bei Berührung mitBaustoffen. Werkstoffe und Korrosion, 1986, 37, 302–309.


10

2.1 Introduction

According to the Pourbaix diagram [1], shown in Fig. 2.1, zinc is expectedto corrode at pH >10.5 with the formation of the soluble hydrogen zincateions, HZn O .2

– Above pH 13.1, the dominating soluble product is zincateions, Zn O2

2– . Under normal conditions concrete is alkaline at pH 12.5 to 13.Although pore water often has a pH above what zinc normally tolerates, thebehaviour of zinc-coated steel reinforcement in concrete is usually good.Galvanized steel reinforcement is used successfully in situations where steelwill corrode due to chloride or carbonation. However, the corrosion rate inaerated, strongly alkaline solutions is too high for most practical applications.Therefore, this subject has received little attention. Bocris et al. [2] describecathodic and anodic reactions in the presence of fairly high levels of zincateions – Tafel slopes, exchange current and the mechanism of dissolution anddeposition being the main subjects. Chloride induced passivity breakdownand pitting are treated in detail by Guo et al. [3]. The anodic behaviour atabout the same pH as in the present work is described by Augustynski et al.[4] and in more concentrated alkaline solutions by Baugh et al. [5, 6]. Thepresent report describes experiments with zinc in KOH solutions, synthetic

2Corrosion and electrochemistry of zinc in

alkaline solutions and in cement mortar

K. V I D E M, University of Oslo, Norway

Corrosion

Pas

sivi

ty Corrosion

Immunity

E [

V(N

HE

)]

1

0.6

0.2

–0.2

–0.6

–1

–1.4

–1.8–2 0 2 4 6 8 10 12 14 16

pH

2.1 Simplified potential pH diagram for zinc [1].


Corrosion and electrochemistry of zinc 11

concrete pore water and in cement mortar, including studies of some selectedphenomena by cyclic voltammetry and electrochemical impedancespectroscopy (EIS).

2.2 Experimental methods

Static corrosion experiments were carried out in 0.5 L glass vessels open toair. Test conditions are briefly described together with the results. Page andVennesland [7] have found that the pore water in new concrete can consist ofabout 0.2M KOH – 0.1M NaOH saturated with Ca(OH)2. The present testsolutions had KOH and NaOH in the molar ratio 2:1 unless otherwise stated.Some Ca(OH)2 powder was placed at the bottom of the test vessels. Thechemicals were analytical grade. The electrodes used for electrochemicalexperiments in both solutions and in cement mortar were made from super-pure zinc in the form of 6-mm-diameter discs masked off with evacuatedepoxy, giving a flat, circular working area of 0.5 cm2. The surface wasprepared by abrasion with 1000 grit paper followed by etching in 0.05 wt %HCl. Electrochemical experiments in solutions were carried out in 0.6 Lclosed bottles with apertures for the entry of the working electrode, referenceelectrode and AISI 316 counter electrode, and with a stirring magnet and airbubbled though the solutions. Similar zinc electrodes were cast into cementmortar cylinders of about 8 cm diameter and 15 cm high. Sand and Portlandcement were mixed in the ratio 3:1 and cast with a water/cement ratio of 0.6.One slab had mixed-in NaCl with a chloride content corresponding to 1% ofthe cement weight.

The cement mortar slabs were covered with water for the first 10 days andthen stored in air with 95% relative humidity in a stainless-steel container at22 ± 2∞C. The response to different moisture levels was of interest. Therefore,after 130 days exposure, the mortar was saturated with water. The slabs werethen allowed to become dryer for a period of 200 days and then wetted again.

The zinc electrodes in cement mortar were accompanied by embeddedreference electrodes made of copper at a separation of about 3 mm. Thecement mortar slabs also had embedded copper wire counter electrodes atthe periphery. This significantly improved the electrochemical measurementswith applied current as they could be performed without wetting the surface.The potentials of the copper reference electrodes and corrosion potentials ofthe zinc electrodes were measured at intervals with a calomel electrodecontacting the cement mortar surface via a small, moist piece of paper. Asthe zinc electrodes sometimes had polarization resistance in the MW-range,a normal multimeter would draw far too much current. An electrometer withas low bias as 10 pA was used. Electrochemical impedance spectroscopywas performed with a Gamry 900 EIS instrument. A Gamry CMS 100 computercontrolled potentiostat and galvanostat was used for the other measurements.



2.3 Results

2.3.1 Static corrosion exposures

Specimens of 15 cm2 surface area were exposed in 0.5 L glass vessels opento air under static conditions. The corrosion product dissolved when thehydroxyl concentration was 0.03M and above. A period with a nearly constantcorrosion potential and constant corrosion rate occurred shortly after thestart of the exposure. The weight loss after 18 h exposure was the same forexposures in 0.03 to 1M hydroxyl (pH 12.5 to 13.9) and corresponded to acorrosion rate of 0.4 mm y–1. Corrosion potential and linear polarizationresistance (LPR) were measured by contacting the specimens with a zincneedle of about 1 mm diameter. LPR was about 400 W cm2 after exposure for3 and 9 h and independent of pH. Gentle stirring of the solutions and evenwork at the bench with the test vessels leading to vibrations reduced thecorrosion potential. Therefore, these corrosion rates are not generally applicable,but apply for the test geometry and conditions in this lab. The solutions werenot replaced in these tests and soon contained high levels of hydrogen zincateions. After 10 days, specimens exposed to hydroxyl concentrations of 0.1M

and below had a white corrosion film and gained weight. As the corrosionfilm did not protect the complete surface, pitting occurred. The corrosionproduct liberated CO2 when dissolved in acid, proving the presence of zinccarbonate. It is assumed that CO2 from the air had removed the Ca2+ in thesolutions, despite solid Ca(OH)2 having been placed at the bottom of the testvessels. These experiments revealed interesting aspects. However, as CO2 isabsent in uncarbonated concrete, the details are not described in this report.

2.3.2 Exposures in solutions with CO2-free air

Laboratory air was passed through a double Ca2+ trap before it was continuouslybubbled though the test solutions contained in closed glass vessels. Figure2.2 shows the corrosion potential as a function of time at pH 13 in a Ca(OH)2

saturated solution, (containing 10–3 M Ca2+). It is seen that a sudden step toa higher corrosion potential took place in the absence of an applied current.LPR varied between 20 and 30 W cm2 before the potential step. After thestep, LPR varied between 600 and 800 W cm2, showing a change from activeto passive corrosion. A small constant anodic current was applied for a shortperiod to determine LPR and the potential shift measured with an electrometer.To avoid disturbance of the electrode, currents that gave potential shifts ofonly about 5 mV were used.

Passivity at pH 13 is not expected from the Pourbaix diagram [1], Fig. 2.1.Calcium ions may affect the process by formation of calcium zincate. Tocheck this, experiments in 0.1M KOH were carried out. The change from theactive to the passive state also took place in this case, but appeared to require



slightly longer time. Any effect of calcium on the corrosion potential andLPR in the active, as well as in the passive state, was hardly detectable. It ispossible that calcium affects corrosion, but is not the cause of the passivity.These measurements were carried out under agitation. The corrosion rate inthe active state was as high as 5 mm y–1. and thus much higher than in thestatic exposures.

The high corrosion rate led to a significant release of HZn O2– to the test

solution. When passivity occurred after 210 minutes in the experiment shownin Fig. 2.2, the HZn O2

– concentration was about 0.03M. The possibility wasconsidered that the shift from active to passive was initiated by accumulationof this corrosion product. To explore this, the experiments were repeated inthe used solutions. In this way, the hypothesis that hydrogen zincate ionaccumulation caused the passivity was rejected.

2.3.3 Cyclic voltammetry

Figure 2.3 shows current density as a function of potential at pH 13. Theeffect of hydroxyl concentrations from 0.01 to 1M is illustrated in Fig. 2.4. Inthe last case, the measurements were taken in deaerated KOH solutions. Thesweeps started in the positive direction from a slightly cathodic potential. Itis seen from these figures that the corrosion potential decreased with increasedalkalinity and that the current density at a given potential increased. Theanodic dissolution took place in different regions, as indicated at Fig. 2.4. Upto about 30 mV above the open circuit potential, the current increased sharplywith the potential. This region is indicated as ‘Tafel’ in Fig. 2.4. At higherpotentials, the current went over a flat maximum in a region with currentlittle affected by potential. As the anodic current density in this region increasedwith increased agitation, it is assumed that the anodic dissolution was restricted

E [

V(S

CE

)]

–0.400

–0.600

–0.800

–1.000

–1.200

–1.400

–1.600

Active

Passive

0 200 400 600 800Time (min)

2.2 Corrosion potential as a function of time at pH 13 in a solutioncontaining KOH and NaOH in the proportion 2:1 and saturated withCa(OH)2. Experiments in KOH at the same pH gave rather similarresults.



by the slow escape of dissolved oxidation products by diffusion and forcedconvection. This region is marked ‘Diffusion’ in Fig. 2.4. At still higherpotentials, passivity occurred. However, passivity reduced the dissolutionrate only to a factor of 8 at pH 14 and to a factor of 5 at pH 13. At pH 12,passivity was hardly detectable. The anodic current in the passive region wasalso agitation dependent. An unusual aspect was that the passive oxidationrate in this case was much higher than for active corrosion without polarization.The cathodic peak shown in Fig. 2.3 is due to the reduction of corrosionproducts formed in the anodic period of cyclic voltammetry.

2.3.4 Chloride

Corrosion of zinc in the real life is usually little affected by chloride. However,it is well known that passive zinc suffers local film breakdown and pitting inalkaline solutions under anodic polarization. A detailed study, including also

Cathodic peak

Up

Reversed

–1.5 –1.3 –1.1 –0.9 –0.7 –0.5 –0.3E [V(SCE)]

CD

(m

A c

m–2

)

1

0.1

0.01

0.001

2.3 Current density as a function of potential at pH 13. Same solutionas for Fig. 2.2. Sweep rate 1 mV s–1. Curves obtained with automaticcompensation for the potential drop between working and referenceelectrode.

2.4 Current density as a function of potential in deaerated 0.01 to 1M

KOH. Only data from the first half-cycle shown. Sweep rate 1 mV s–1.

CD

(m

A c

m–2

)

1000

100

10

1

0.1

0.01

0.001

Diffusion

Passive

Tafel0.1 M

0.01 M

–1.6 –1.4 –1.2 –1E [V(SCE)]

1 M



the morphology of pits on single crystals, has been published by Guo et al.[3]. An example from the present work is given in Fig. 2.5. No effect ofchloride was detected under conditions where the corrosion product dissolved.In the experiment referred to in Fig. 2.5, local passivity breakdown andpitting occurred at –0.46 V (SCE) leading to a current increase of almostthree orders of magnitude. Very interesting aspects were observed in thereverse sweep. A current reduction took place in the reverse sweep at about–0.66 V (SCE). However, the film damage was not fully repaired as thecurrent at a given potential was more than an order of magnitude higher thanbefore passivity breakdown occurred. An astonishing aspect is that the cathodiccurrent was even higher. It appears that the breakdown event has createddefects with very high anodic as well as cathodic activity. This indicates thatso called active zinc is not ‘naked’ in the solution, as the increased cathodicreaction rate would then be difficult to explain.

Figure 2.6 shows corrosion potential without applied current as a functionof time at pH 13 and with 0.15 M NaCl added to the solution of KOH, NaOH

CD

(m

A c

m–2

)

1000

100

10

1

0.1

0.01

0.001

CathodicPitting

Reverse

Forward

–1.5 –1 –0.5E [V(SCE)]

0 2000 4000 6000 8000Time (min)

E [

V(S

CE

)]

–0.400

–0.600

–0.800

–1.000

–1.200

–1.400

2.5 Current density as a function of potential at pH 13 and a solutioncontaining 0.15M NaCl. Sweep starting from a slightly cathodicpotential and with a rate of 1 mV s–1.

2.6 Corrosion potential as a function of time at pH 13 and a solutioncontaining 0.15 M NaCl.



and Ca(OH)2. It is surprising that chloride did not prevent the set up ofpassivity. Just at the onset of passivity, the corrosion potential was about thesame as without chloride for a short time (see also Fig. 2.2). Thereafter, asudden drop in the potential took place. It is assumed that the passive potentialoccuring after 1500s was higher than the pitting potential. Therefore, defectswere formed. At the lower potential after the drop some pits must havestifled, so the potential ended up only slightly below the passive region.

2.3.5 Zinc in cement mortar

Figures 2.7 and 2.8 show corrosion potential as a function of time. Figure 2.7applies for two electrodes cast into mortar of sand and cement, while Fig. 2.8relates to mortar with a chloride content equivalent to 1% of the cementweight. To study the response to moisture level, the mortar was water saturatedin some periods. This is indicated with ‘w’ at the top of Figs. 2.7 and 2.8. It

2.7 Corrosion potential as a function of time for two electrodes inmortar of sand and cement. Periods with water saturated mortar areindicated with ‘w’ at the top of the figure.

0 500 1000 1500Time (days)

E [

mV

(SC

E)]

200

–200

–600

–1000

–1400

W W W W W

2.8 Corrosion potential as a function of time for two electrodes inmortar with 1% Cl–. Periods with water saturated mortar areindicated with ‘w’ at the top of the figure.

0 500 1000 1500Time (days)

E [

mV

(SC

E)]

200

–200

–600

–1000

–1400

WW W W W



is seen that moisture level is a very important parameter, leading to lowcorrosion potentials. As will be described later, this is indicative of highcorrosion rates. The addition of chloride to the mortar had a large effect andmade corrosion more severe. Without chloride, the electrodes seldom hadpotentials below –0.7 V (SCE). For mortar with 1% Cl– the potentials usuallywere below –0.7 V (SCE) and many reading below –0.9 V (SCE), as seenfrom Fig. 2.8. The average of all measurements was reduced by about150 mV with 1% chloride.

In contrast to exposures in alkaline solutions, hardly any well-definedpolarization resistance exists for zinc in cement mortar. Different valueswere obtained with different techniques. Figure 2.9 shows the polarizationresistance obtained by a potential step of 20 mV. Polarization resistance inthis case is the ratio between this potential shift and the current density. Thevalues obtained in this way increased by about an order of magnitude whenthe polarization time was increased from 10 to 100 s. As the lines in Fig. 2.9are nearly parallel, the electrode with the corrosion potential of –1133 mV(SCE) had LPR values two orders of magnitude lower than that at –51 mV(SCE) regardless of the polarization time. This reflects a much higher corrosionrate for the electrode with the lowest corrosion potential.

2.4 Discussion

2.4.1 Tafel relationships in the active state

Similarly to that shown in Fig. 2.3, the active regions of anodic polarizationcurves have no straight parts for experiments in solutions with oxygen. The

–51 mV(SCE)

–656 mV(SCE)

–1133 mV(SCE)

1 10 100 1000Time (s)

LPR

(M

W c

m2 )

100

10

1

0.1

0.01

0.001

2.9 Examples of values for polarization resistance for electrodes incement mortar obtained with potential steps of 20 mV and variouspolarization times. Electrodes with corrosion potentials of –1133 mV,685 mV and –51 mV (SCE).



curvature is due to corrosion rates being so high that the anodic reaction ratetakes place at the border of the Tafel region. As shown from Fig. 2.10, goodfit with the Tafel region in Fig. 2.3 was obtained by assuming a corrosioncurrent of 0.034 mA cm–2, a cathodic Tafel slope of –110 mV decade–1 andanodic Tafel slope of 27 mV decade–1. The cathodic slope fits with oxygenreduction as the cathodic reaction. The anodic current in the Tafel region wasindependent of agitation. Increased agitation raised the Tafel region to higherpotentials due to reduced restrictions caused by mass transfer. The Tafelslope varied somewhat with pH. Sweep rate of 1 mV s–1 led to slopes of 27and 34 mV decade–1 at pH 14 and 13, respectively. More accurate measurementswith instantaneous potential steps resulted in 22.5, 31.0 and 31.5 mVdecade–1 for pH 14, 13.5 and 13, respectively. At pH 12.5 and below, theTafel region became so small that a determination was hardly possible.Kabakof [8] reports an anodic Tafel slope of 30 mV decade–1, in good agreementwith this study, while Bocris et al. [2] obtained a value of 49±13 mVdecade–1 in a carefully purified solution. The present Tafel slopes for theactive state give a Stern–Geary constant of 10.0 mV for estimation of corrosionrates from LPR, corresponding to a corrosion rate of 0.145 mm y–1 for anLPR of 1 kW cm2. Using this value for the static corrosion experiments inalkaline solutions, gave acceptable agreement with corrosion rates obtainedby weight loss. As mentioned, the corrosion rate was independent of pH inthese static corrosion tests despite a large pH effect on the anodic kinetics.Corrosion being limited by oxygen supply rate is probably the mainphenomenon responsible for this.

2.4.2 Passivity

Figures 2.3 to 2.4 show sudden current reductions occurring at potentials afew hundred mV above the corrosion potential. The existence of passivity

Modelled

Observed

–1.45 –1.43 –1.41 –1.39 –1.37 –1.35E [V(SCE)]

CD

(m

A c

m–2

)

1

0.1

0.01

0.001

2.10 Measured current as a function of potential at pH 13 andmodelled current by assuming a corrosion current of 0.034 mA cm–2,an anodic Tafel slope of 27 mV decade–1 and a cathodic Tafel slope of–110 mV decade–1. ‘Observed’ is a part of the polarization curveshown in Fig. 2.3.



under anodic polarization in alkaline solutions is well known [3–6]. However,attention is drawn to some unusual aspects:

∑ The gradual shift between active and passive state.∑ Passivity leading to a rather small reduction of anodic rate.∑ Passive current increasing significantly with increased potential.∑ Passive current affected by mass transfer.

For most combinations of metal and environment, a sharp transition fromthe active to passive state takes place at the passivation potential. However,Fig. 2.3 shows a gradual change in a potential span of about 50 mV. As thecurrent in the reverse sweep followed the same trace, this phenomenon isreal. It should be noted that these curves were obtained with automaticcompensation of the resistance potential drop in the gap between the workingand the reference electrode. The reduction in anodic rate was about 10 timesat pH 14 and 5 times at pH 13. From this it is concluded that the passive filmis only slightly protective, compared to typical passive film that often reducesthe reaction rate many orders of magnitude.

In the classical description of passivity, the current at steady state isindependent of potential. From Fig. 2.3 it is seen that the current was raisedby increased potential both for a positive as well for a negative sweep direction.As the passive current was sensitive to agitation, it is concluded that the filmwas attacked by hydroxyl ions, and that charge was consumed to maintainthe film. This view is supported by the observation that the passive currentat a given potential was more or less proportional to the hydroxide concentration.The various observations indicated the presence of a passive film that isporous (at least at the outer part), becomes thicker at increased potential, andhas a larger area in contact with the solution due to the pores. The last pointis meant to take care of the potential dependence of the passive current.

Passivity occurring without polarization in aerated solutions is verysurprising. The required passivation current density was rather high underpolarization. From the experiments with cyclic voltammetry (see Figs. 2.3and 2.4), oxygen reduction appears to be far too slow to supply the necessarycurrent. A period in the active state between 5 and 40 ks was necessary tobring the electrodes to the passive state without polarization. An interestingaspect is what goes on in this induction period before the onset of passivity.No gradual changes were identified in this period with the techniques used.Corrosion potential as well as polarization resistance were nearly constantuntil just before the onset of passivity.

2.4.3 EIS indicates complex processes

A Nyquist diagram from the active period at pH 13 in a solution also containingCa2+ is shown in Fig. 2.11. This diagram resembles two arcs. The first arc



extrapolates to about 100 W. As the Nyquist diagrams are not corrected forthe specimen area (being 0.5 cm2), this indicates a LPR value of about 50 Wcm2 in good agreement with the value obtained galvanostatically, being 20–30 W cm2, as described earlier. The low frequency part of Fig. 2.11 has aninductive tail that could indicate an adsorption phenomenon. EIS is difficultto interpret in this case, as the change of modulus and phase angle withfrequency is not caused by this parameter alone. Many hours were requiredfor the measurements at the lowest frequencies, leading to changes of theelectrodes. However, the theory of an adsorption phenomenon triggeringpassivity appears reasonable as no other phenomena are identified. Figure2.12 shows a Nyquist diagram from the passive state and has a shape inharmony with a metal with a passive film. The zigzag path at frequenciesbelow 0.001 Hz is due to variations of the corrosion potential and illustratesthe difficulties with EIS at such low frequencies that the measurements takemany hours.

Two examples of Nyquist diagrams for zinc in cement mortar are shownin Figs. 2.13 and 2.14. Figure 2.13 applies for an electrode with a very low

Imag

(W

)

25

201510

50

–5–10

Max. 10000

1 0.1

0.01

0.001

0.0001

0 50 100 150Real (W)

2.11 Nyquist diagram at the corrosion potential in the active periodfor exposure at pH 13 with air and CO2 free conditions. Exposure 2 hwhen EIS was started. Corrosion potential was –1.34 V (SCE).Specimen area 0.5 cm2. Numbers are frequencies in Hz.

0 20000 40000Real (W)

Imag

(W

)

20000

15000

10000

5000

0

–5000

0.2

0.02

Max. 20000.002

0.0002

2.12 Nyquist diagram at the corrosion potential in the passive stateand same solution as in Fig. 2.11. Corrosion potential was –0.430 V(SCE). Specimen area 0.5 cm2. Numbers are frequencies in Hz.

10



corrosion potential, –1.009 V (SCE), while Fig. 2.14 is for one with a muchhigher potential, –0.613 V (SCE). All Nyquist diagrams for electrodes inmortar consisted of constant phase angle lines at the low frequency side. Asthe constant-phase-angle elements were not observed in solutions, it isassociated with the conditions in the mortar. The phenomena responsible forthis have not been identified. Warburg impedance theory encounters difficultieswith the slope in the Nyquist diagrams and the frequency response ofimpedance. The slope in the Nyquist diagram in Fig. 2.13 is about 4 and thusvery different from the unity predicted by Warburg theory for homogenousenvironments. The impedance of the constant phase angle elements in Figs.2.13 and 2.14 increases with the frequency to the power of –0.78 – the valuefor Warburg impedance being –0.5.

LPR values cannot be found from these diagrams because arcs forextrapolations are missing. However, qualitatively EIS responded to the severityof corrosion. The values for the real and imaginary components at a given

Imag

(kW

)

200

150

100

50

0

0.00019

0.0019Max 30000

0 20 40 60Real (kW)

2.13 Example of Nyquist diagram for zinc in cement mortar.Electrode with a low corrosion potential, –1.009 V (SCE). 425 daysexposure. Specimen area 0.5 cm2.

0.00019

0.0019

Max 30000

0 100 200 300 400Real (kW)

Imag

(kW

)

500

400

300

200

100

0

2.14 Example of Nyquist diagram for zinc in cement mortar.Electrode with a much higher corrosion potential, –0.613 V (SCE).425 days exposure. Specimen area 0.5 cm2.



frequency were much higher for electrodes with high corrosion potentialsthan those with low, as seen from Figs. 2.13 and 2.14.

2.4.4 Polarization resistance and corrosionrate for zinc in mortar

As already described, active electrodes corroding in solutions had clearlydefined values for Tafel slopes and polarization resistance. The corrosionrate can be obtained from electrochemical measurements in these cases.Both the Stern Geary treatment as well as modelling of polarization curvesappear to function satisfactorily. The polarization resistance for passiveelectrodes in solutions varied with the measuring parameters, but to a muchlesser extent as was seen for electrodes in cement mortar.

In contrast to alkaline solutions, the assessment of the corrosion rate forzinc in cement mortar from electrochemical measurements is subject to severedifficulties. The problem is that the metal lacks a clearly defined polarizationresistance, the values varying substantially with the method and parametersof the measurements. Figure 2.9 gives one example, showing polarizationresistance values varying by two orders of magnitude depending on theduration of polarization. An earlier report describes polarization resistanceobtained by galvanostatic polarization [9]. The apparent values for polarizationresistance varied in the same way with the duration of polarization, but thevalues for the electrodes were nearly ten times higher. As could be expected,polarization resistance obtained by potentiodynamic sweep increased withreduced sweep rate. Therefore, quantitative values for the corrosion ratecannot be obtained from any of these measurements. No theory exists forassessing the severity of corrosion from these electrochemical measurements.Also, for the time being, the engineering database for doing this empiricallyis lacking.

It is tempting to suggest that the lowest potentials in Figs. 2.7 and 2.8apply for active zinc and the highest for passive metal. This was done in aprevious report [9]. However, both the EIS and dc methods indicate thatcorrosion becomes gradually more severe with reduced corrosion potentialwithout any threshold for change from the passive to active state. The varioustypes of measurements agreed qualitatively in that they indicated that lowcorrosion potentials were linked to higher corrosion rates than high potentials.The difference between LPR for the electrodes with corrosion potentials –1.133 and –0.51 V (SCE) in Fig. 2.9 is nearly two orders of magnitude. Thisindicates that corrosion rates of zinc in this study exhibit very large variations,possibly two orders of magnitude. However, it is premature to draw anyconclusion about the applicability of zinc under conditions giving very lowcorrosion potentials.

It is well established that moisture is a very important variable for the



corrosion of galvanized steel reinforcement in concrete. Moreno et al. [10]have observed corrosion potentials in the region –350 to 200 mV (SCE) forgalvanized steel in concrete without added chloride stored in air, and potentialsbetween –600 and 400 mV (SCE) under wet conditions. This is in goodagreement with the measurements shown in Fig. 2.7. An important differencebetween zinc and steel is that steel is passive in mortar without chloride andhas rather high corrosion potentials independent of moisture level. The cementmortar slabs used in this work contained embedded steel electrodes. Wettingreduced the corrosion potential for steel in mortar with chloride only [11]. Asalready described, chloride reduced the corrosion potentials of zinc. Figure2.9 demonstrates that low corrosion potentials are linked to low polarizationresistance and hence higher corrosion rates, as confirmed by Moreno et al.[10]. Due to differences in the experimental techniques, their results are notdirectly comparable. Figure 2.8 shows potentials as low as –1000 mV (SCE)under moist conditions for mortar with 1% Cl– even after more than 1000days exposure. This is far below the potentials reported by Moreno et al.[10] for concrete without chloride additions. Reliable relationships betweencorrosion potential and corrosion rate are missing both from laboratoryexperiments and from real life. Therefore, it is premature to draw any conclusionabout the applicability of zinc under conditions giving very low corrosionpotentials. Nevertheless, the low potentials in water saturated mortar is anaspect that needs attention.

A really puzzling aspect is that zinc alters from a fairly plain polarizationbehaviour in alkaline solutions to a very complex one in cement mortar. Thepolarization resistance for zinc, shown in Fig. 2.9, varies in the same manneras for the steel electrodes embedded in the mortar slabs being used for thepresent work [11]. Potential steps gave currents decreasing with time to thepower of about –0.5, as for zinc. In a similar way, constant phase angle lines,like those shown in Fig. 2.13 were the normal result for steel with EIS. Asiron has oxides with different valence, redox of corrosion products can takeplace. The charge consumed by this has been suggested as the reason for thevariation of current with the duration of polarization as well as the constantphase angle elements with EIS [12]. As zinc is only two-valent, redox ofcorrosion product cannot occur. The mechanism behind the changes of theperformance with duration of the polarization is far from understood. Thecurrent change with the slope of –0.5 for a potential step, as well as theconstant phase angle elements illustrated in Figs. 2.13 and 2.14 could indicatediffusion limited kinetics. To agree with the experimental results, the reactionrate must be speeded-up by moisture. The data hardly permit further speculation.

2.4.5 Corrosion products

Feitknecht [13] reported that the solid corrosion product is ZnO in KOHsolutions at the same pH as the present experiments. The same product has



also been reported for the anodic film by Rudd and Breslin [14], who alsostate from photo-chemical measurements that it is a highly doped n-conductor.The corrosion potentials recorded in Figs. 2.7 and 2.8 show that the metal inmortar formed corrosion films. Belaid et al. [15] reported that a film ofcalcium hydroxyzincate, Ca(Zn(OH)3)2 · 2H2O, passivates zinc in freshconcrete. In addition to this component, they also identified ZnO, Zn5(OH)8Cl2· H2O and eZn(OH)2. ZnO has a lower volume than the zinc metal fromwhich it is formed. Zn5(OH)8Cl2 · H2O occupies a 3.62 fold higher volumethan that of zinc and has been held responsible for cracking of concrete witha high chloride content [15]. When zinc passivates in KOH solutions underanodic polarization, there are hardly any other possible passive films thanzinc oxide or hydroxide. In the absence of polarization, passivation occurredfaster when Ca2+ was present. This is consistent with calcium hydroxyzincateas the film material. However, absence of calcium also led to passive filmsthat appeared to be equally protective.

2.5 Conclusions

The corrosion rate was about 0.4 mm y–1 in static corrosion tests conductedin alkaline solutions open to air. The rate was independent of pH in the range12.5 to 13.9 during the initial period because the corrosion products weresoluble and the corrosion rate was controlled by oxygen reduction kinetics.After some days of exposure, carbonate-containing corrosion films formeddue to the reaction of CO2 from the air.

Zinc exhibited an active/passive transition during cyclic voltammetry inalkaline solutions without CO2. The Tafel slope for active anodic dissolutionwas about 30 mV decade–1 and for cathodic oxygen reduction –110 mVdecade–1. The passive current density was about 0.5 mA cm–2 at pH 13 andincreased in proportion with the hydroxyl concentration. Increased agitationalso raised the passive current, indicating that the passive film was continuouslydissolved or destroyed by attack from hydroxyl ions.

Zinc became passive after prolonged exposure without applied current insolutions that are in contact with air from which CO2 had been removed. AtpH 13, passivation occurred after a few hours in solutions with Ca2+. WithoutCa2+, passivation set in after slightly longer exposures. The polarizationresistance in the active state was 20 to 30 W cm2 and between 600 and 800W cm2 in the passive state. The passive reaction rate is thus much lower thanthat observed in cyclic voltammetry. The processes taking place in the inductionperiod that suddenly trigger passivity are not understood. Adsorptionphenomena are suggested from EIS.

Chloride caused passivity breakdown and pitting under polarization athigh potentials. The film damage was not fully repaired when the potentialwas reduced. A surprising observation was that the metal had a higher anodic



as well as cathodic reaction rate at low potentials after the pitting event athigh potential.

Zinc cast in cement mortar had corrosion potentials in the range –1.4 to–0.1 V (SCE). Low potential indicated more severe corrosion. Water-saturatedmortar contributed towards low corrosion potentials. The corrosion potentialswere consistently lower in mortar with 1% Cl– than in mortar without andwere reduced by 150 mV on average. It is concluded that the mixed-inchloride made corrosion more severe.

In contrast to electrodes in solutions, Nyquist diagrams for electrodes inmortar consisted only of constant phase angle lines. The phenomena responsiblefor this were not identified. LPR values could not be determined from thesediagrams because arcs for extrapolations were missing. The values for thereal and imaginary components at a given frequency were much higher forelectrodes with high corrosion potentials than those with low.

In contrast with exposure in alkaline solutions, a well-defined polarizationresistance hardly exists for zinc in cement mortar. Very different values wereobtained with different techniques. Therefore, the corrosion rate in mortarcould not be obtained from the electrochemical measurements. Polarizationresistance achieved by the same measuring technique could be two orders ofmagnitude lower for electrodes corroding with a low potential than thosewith a high. Thus, a very large span of corrosion rates occurred in theseexperiments. It is premature to draw any conclusions about whether the ratewas always acceptably low for practical applications in concrete. Cautionregarding the use of zinc in very humid concrete is recommended until moreknowledge is available.

2.6 References

1. M. Pourbaix, Atlas of Electrochemical Equilibria. Pergamon Press, Oxford, 1966,p. 147 and p. 406.

2. J. O’M. Bocris, Z. Nagy and A. Damjanovic, J. Electrochem. Soc., 1972, 119, 285.3. R. Guo, F. Weinberg and D. Tromans, Corrosion, 1995, 51, 356.4. J. Augustynski, F. Dalard and J. C. Sohm, Corros. Sci. 1972, 12, 713.5. L. M. Baugh and A. Higginson, Electrochim. Acta, 1985, 30, 1163.6. L. M. Baugh and A. R. Baikie, Electrochim. Acta, 1985, 30, 1173.7. C. L. Page and Ø. Vennesland, Mater. Construct., 1983, 16, 19.8. B. N. Kabanov, Izv. Akad. Nauk SSSR, 1962, 980.9. K. Videm, Behaviour of zinc in synthetic concrete pore water and in cement mortar.

EUROCORR2001, Riva del Garda, Italy, October 2001.10. E. I. Moreno, A. A. Sagues and R. G. Powers, Performance of plain and galvanized

reinforcement during the initiation state of corrosion in concrete with pozzolanicadditions. NACE International, Houston Texas, CORROSION 96, 326.

11. K. Videm, Corrosion of steel in cement mortar with chloride and micro-silica.EUROCORR2005, 523, Lisboa, Portugal, September 2005.



12. C. Andrade, F. Bolzoni, A. Collazo, X. R. Novoa and M. C. Perez, Measurement ofsteel corrosion in concrete by electrochemical technique: influence of the redoxprocesses in the oxide scale, Corrosion, 2000, 56, 500.

13. W. Feitknecht, Met. Corros., 1947, 192.14. A. L. Rudd and C. B. Breslin, Electrochim. Acta, 2000, 45, 1571.15. F. Balaid, G. Arliguie and R. Francois, Corrosion, 2000, 56, 960.


27

3.1 Introduction

The limited durability of reinforced concrete structures has promptedconsiderable effort on research, aimed at the establishment of an adequatelevel of knowledge on the properties of this material. When concrete technologyis correctly applied, it is possible to increase the service life of reinforcedstructures, even if it is difficult to ensure complete protection from aggressiveagents, particularly when cracking of concrete or accidental causes ofdegradation occur. Therefore, under critical conditions for corrosion, onlyadditional protection to the steel reinforcement can guarantee the durabilityof the structure.

Among the possible methods for improving the corrosion resistance ofreinforcement in concrete, new consideration has been given to the use ofgalvanized rebars, because of their relatively low cost relative to other protectionsystems. It is clear that the galvanized bars increase the initial cost of theconcrete structures but, during their whole service life, this is not a majorcost considering the rising costs of restoration and maintenance. Althoughgood practical results have been reported in the literature [1–4], the benefitsof using galvanized steel in reinforced concrete structures are still uncertainbecause of some controversial laboratory test results [5–11].

However, Swamy [12] stated that the results of laboratory tests must beviewed with caution due to the fact that the simulated environment does notfully match the actual conditions. Furthermore, it has been observed thatzinc coating delays the onset of corrosion of reinforcing steel, as explainedby the conceptual model proposed by Yeomans [13].

One of the unanswered questions concerning the use of galvanizedreinforcement is the risk of corrosion where discontinuities are present in thezinc coating owing to bending of the bars or welding procedures which leaveuncoated spots.

In this work, the corrosion behaviour of galvanized reinforcement withdiscontinuities in the zinc coating was studied. For this purpose, several

3Corrosion behaviour of galvanized steel

rebars in the presence of coatingdiscontinuities

T. B E L L E Z Z E, R. F R AT E S I and F. T I T TA R E L L IUniversità Politecnica delle Marche, Ancona, Italy



reinforced concrete specimens were manufactured. The coating discontinuitieswere simulated by a small piece of black bar assembled with two lateralgalvanized bars, electrically isolated from each other, in order to measureexternally the galvanic corrosion current; these specimens were submitted towet–dry cycles by ponding both with tap water and with a sodium chloridesolution.

At the end of the tests, all specimens were broken and the embeddedreinforcements were examined to assess visually the corrosion attack.


Sixteen prismatic specimens (14 ¥ 12 ¥ 44 cm) were manufactured usingCEM II/A-L 42.5 R cement with a water–cement ratio of 0.70. The specimenswere reinforced in the longitudinal direction with a bar (diameter = 12 mm;cover = 15 mm) obtained by assembling three electrically isolated bars: twolateral galvanized bars (anodic parts) of the same length and a small piece ofblack bar (cathodic part) in the centre. Four different types of bars wereproduced with anodic to cathodic surface area ratios, Sa/Sc as shown inTable 3.1.

Figure 3.1 shows two assembled bars (Sa/Sc = 20 and 7.5) ready forcasting. The assembly and electrical insulation were performed with a PVCinsert (not shown in the figure) and epoxy resin between the two lateralgalvanized bars and the central black bar. In Figure 3.1, the electrical cablesfor the external current and potential measurements are also clearly visible.The Sa and Sc values were well defined by masking the bars with epoxy resin.

Four concrete specimens were prepared for each value of Sa/Sc (Table3.1): two for each type of exposure condition. All specimens were removedfrom their moulds three days after casting. The exposure conditions includedwet–dry cycles with tap water and wet–dry cycles with 5% NaCl solution.The wet–dry cycles were applied to the specimens after 28 days of air curingand they consisted of five days of drying and two days of wetting. During theexperimentation period, there were some longer drying periods.

During the whole test period (about 270 days), the galvanic couplingbetween the galvanized bars and the black bar was externally obtained by

Table 3.1 Assembled bars with four different Sa/Sc

Sa /Sc Lateral galvanized bars Central black bar(cm) (cm)

80 16 + 16 0.440 18 + 18 0.920 15 + 15 1.57.5 15 + 15 4.0


Corrosion behaviour of galvanized steel rebars 29

short-circuiting the electrical cables soldered to the single parts (Fig. 3.1).The short circuit current between the anodic part (the two galvanized bars)and the cathodic part (black bar) was monitored using a zero resistanceammeter AMEL Mod. 668. Furthermore, potential measurements wereperformed with a calomel electrode (SCE) as a reference both during thecoupling conditions and in free corrosion conditions; in this last case, theexternal electrical contacts between the bars were removed and the freecorrosion potential measurements were performed after 1.5 h at open circuit.

Two further specimens of the same type, but without reinforcement, weremanufactured and submitted to ponding with 5% NaCl solution, in order todetermine periodically the depth of chloride penetration during the wet–drycycles.

In the two different exposure conditions, concrete resistivity was alsoevaluated to study the possible ohmic control of galvanic corrosion. Fourcubic specimens (15 ¥ 15 ¥ 15 cm) were cast, two for each type of exposure,with the same cement and the same water–cement ratio as mentioned above.Two stainless-steel plates (18 ¥ 15 cm) were embedded in these specimensand positioned vertically at a separation of 11 cm. Measurements of conductivitywere performed during the whole corrosion test period using a digitalconductometer AMEL Mod. 160 where 50 mV sinusoidal peak-to-peak signaland a frequency of 1 kHz is set.

The values reported in the following section are the average of themeasurements carried out on each type of specimen.

3.3 Results and discussion

Straight after casting the concrete, the galvanized steel assumed an activestate with a free corrosion potential of about –1350 to –1400 mV, while the

3.1 Two assembled bars with Sa/Sc = 20 (top) and Sa/Sc = 7.5(bottom), used to study the galvanic protection on discontinuities.



black steel reached corrosion potential values of –400 to –520 mV (Fig. 3.2).After galvanic coupling of the anodic and cathodic parts, the galvanizedsteel cathodically polarises black steel, which assumed potential values inthe range –850 to –1050 mV, very close to the thermodynamic immunityconditions for steel. A day after casting, the zinc coating became passivewith a relative free corrosion potential value of –750 to –850 mV; the galvaniccorrosion was under cathodic control in the case of Sa/Sc = 80 and Sa/Sc = 40;it became slightly anodic for Sa/Sc = 20 and completely anodic for Sa/Sc = 7.5(Fig. 3.2).

The short circuit currents were very high only during the early days aftercasting and they changed from a minimum value of 300 mA up to a maximumvalue of 1000 mA as the ratio Sa/Sc decreased, implying that an increase inbare steel surface leads to a higher consumption of the adjacent zinc coating.

The wet–dry cycles started after 28 days of curing.

3.3.1 Wet–dry cycles with tap water

During wet–dry cycles with tap water (Fig. 3.3), the specimens showedanodic macrocell control in the case of Sa/Sc = 20 and of Sa/Sc = 7.5, whilefor Sa/Sc = 40 the cathodic polarisation was almost equal to the anodicpolarisation. For Sa/Sc = 80 there was cathodic macrocell control; this lastresult is quite clear considering the low value for the black steel surface. Asa consequence, the cathodic protection is much more effective for a smalldefect in the zinc coating.

The short circuit currents became very low with time and assumed valueslower than 10 mA. The resistivity was low and ranged between 5 and 20 Wcm during wetting periods, and between 20 and 75 W cm during normaldrying periods. Because of the low values both of the short-circuit currentand of the resistivity of concrete, ohmic drop control of galvanic corrosionhas to be excluded.

During wet–dry cycles, free corrosion potential values ranging between–100 to –200 mV and –500 to –600 mV were monitored for black andgalvanized bars, respectively.

3.3.2 Wet–dry cycles with 5% NaCl solution

Differently from wet–dry cycles in tap water (Fig. 3.3), in the presence ofsodium chloride solution the galvanic corrosion came under cathodic controlfor every Sa/Sc value (Fig. 3.4), due to the damage of the zinc coating passivefilm.

In particular, after 3 wet–dry cycles, the galvanized bars exerted ‘cathodicprevention’ [14] toward the chloride attack with respect to the coupled blackbars. In fact, considering that at a depth of 15 mm (equal to the cover of the


Corrosion behaviour of galvanized steel rebars

31

Sa/Sc = 80 Sa/Sc = 40

Sa/Sc = 20 Sa/Sc = 7.5

0 7 14 21 28Time (days)

0 7 14 21 28Time (days)

Po

ten

tial

(m

V v

s S

CE

)0

–200

–400

–600

–800

–1000

–1200

–1400

–1600

–1800

–2000

Po

ten

tial

(m

V v

s S

CE

)

0

–200

–400

–600

–800

–1000

–1200

–1400

–1600

–1800

–2000

1000

900

800

700

600

500

400

300

200

100

0

Sh

ort

cir

cuit

cu

rren

t (m

A)

1000

900

800

700

600

500

400

300

200

100

0

Sh

ort

cir

cuit

cu

rren

t (m

A)

3.2 Trend of the free corrosion potential, short circuit potential and short circuit currents, during 28 days of curing inatmosphere: ■ free corrosion potential of black steel; ● free corrosion potential of galvanised steel; æ■æ short circuitpotential of black steel; æ●æ short circuit potential of galvanized steel; ···▲··· short circuit current.


Corrosion of reinforcem

ent in concrete32

Sa/Sc = 80 Sa/Sc = 40

Sa/Sc = 20 Sa/Sc = 7.5

28 56 84 112 140 168 196 224 252 280Time (days)

Po

ten

tial

(m

V v

s S

CE

)0

–200

–400

–600

–800

–1000

–1200

–1400

–1600

–1800

–2000

Po

ten

tial

(m

V v

s S

CE

)

0

–200

–400

–600

–800

–1000

–1200

–1400

–1600

–1800

–2000

400

360

320

280

240

200

160

120

80

40

0

Sh

ort

cir

cuit

cu

rren

t (m

A)

28 56 84 112 140 168 196 224 252 280Time (days)

400

360

320

280

240

200

160

120

80

40

0

Sh

ort

cir

cuit

cu

rren

t (m

A)

3.3 Trend of the free corrosion potential, short circuit potential and short circuit currents, during wet-dry cycles with tapwater: ■ free corrosion potential of black steel; ● free corrosion potential of galvanised steel; æ■æ short circuitpotential of black steel; æ●æ short circuit potential of galvanized steel; ···▲··· short circuit current; ····wetting periods.


Corrosion behaviour of galvanized steel rebars

33

Sa/Sc = 80 Sa/Sc = 40

Sa/Sc = 20 Sa/Sc = 7.5

28 56 84 112 140 168 196 224 252 280Time (days)

Po

ten

tial

(m

V v

s S

CE

)0

–200

–400

–600

–800

–1000

–1200

–1400

–1600

–1800

–2000

Po

ten

tial

(m

V v

s S

CE

)

0

–200

–400

–600

–800

–1000

–1200

–1400

–1600

–1800

–2000

2000

1800

1600

1400

1200

1000

800

600

400

200

0

Sh

ort

cir

cuit

cu

rren

t (m

A)

28 56 84 112 140 168 196 224 252 280Time (days)

2000

1800

1600

1400

1200

1000

800

600

400

200

0

Sh

ort

cir

cuit

cu

rren

t (m

A)

3.4 Trend of the free corrosion potential, short circuit potential and short circuit currents, during wet-dry cycles with 5%NaCl solution: ■ free corrosion potential of black steel; ● free corrosion potential of galvanised steel; æ■æ short circuitpotential of black steel; æ●æ short circuit potential of galvanised steel; ···▲··· short circuit current; ···· wetting periods.



assembled bars) the measured chloride concentration (by weight of cement)was slightly under 4 % (Fig. 3.5), the short circuit potentials of the black bar,for the different Sa/Sc values, were under –300 mV vs SCE (Fig. 3.4), whichapproximately corresponds to the pitting potential of a black bar embeddedin concrete with a chloride concentration of 4 % by weight of cement [15].

This effect was much more evident when the Sa/Sc value was 80, whichsimulates the smallest defect in the series. After 10 wet–dry cycles, the‘cathodic prevention’ exerted by the galvanized bars was probably still effective:the short circuit potential of the black bars remained in all cases at valueslower than –600 mV vs SCE, which assures steel cathodic prevention, evenif the measured chloride concentration has reached, at this exposure period,values slightly higher than 4 % [14,15].

During wet–dry cycles, free corrosion potentials ranging from –200 to–400 mV and from –700 to –1000 mV were monitored for black steel andgalvanized steel, respectively. In this case, the low values of free corrosionpotential of zinc coating indicate its weak passive state.

The short circuit currents were much higher than in the previous case(Fig. 3.4 and Fig. 3.3, respectively) due to the higher electromotive forcebetween the galvanized and bare steel. Furthermore, the short-circuit currentincreased as the Sa/Sc decreased, implying that an increase in bare steelsurface leads to a higher consumption of the adjacent zinc coating.

The resistivity, which contributes to the possible ohmic drop control, wasvery low: 3 to 5 ohm cm during wetting periods and 5 to 15 W cm duringdrying periods. Therefore, even if the short-circuit currents in this case aresignificantly high, the ohmic drop control of the galvanic corrosion has stillto be considered ineffective.

Ch

lori

des

co

nte

nt

(% b

y w

eig

ht

of

cem

ent)

6

5

4

3

2

1

0

Concrete cover

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5Concrete depth (cm)

3.5 Trends of chlorides content at various depths in the concretespecimens, after 3 wet-dry cycles (—◆—) and 10 (—●—) wet-drycycles with 5% NaCl solution.



3.3.3 Examination of the reinforcement

At the end of the exposure period, all bars were examined in order to comparetheir visible corrosion conditions with the data obtained by the electrochemicalmeasurements. No red rust was observed on any black bar embedded in themanufactured specimens exposed both to wet–dry cycles with tap water andto wet–dry cycles with NaCl solution. These simulated discontinuities of thezinc coating were protected against corrosion by the cathodic protectionoffered by the adjacent zinc, even if the potential values did not reachthermodynamic immunity.

The galvanized bars embedded in the specimens exposed to wet–dry cycleswith tap water were not corroded, with the exception of those bars coupledwith the largest black bars (Sa/Sc = 7.5). In this case, near the bar joints, thegalvanized bars appeared dark due to the consumption of the external purezinc layer of the coating (h phase), which permits the underlying Zn–Fealloy layer to appear on the surface (z phase visible from the cross-sectionreported in Fig. 3.6a).

In the presence of chloride ions, the corrosion attack of the zinc coatingclose to the pieces of black bar was detected for all Sa/Sc values: very low forsamples with high Sa/Sc ratio (80, 40) and high for the lower Sa/Sc ratios (20,7.5). In particular for samples with Sa/Sc = 7.5, the zinc coating close to thejoints with black steel was heavily corroded and the localised attack hadpenetrated into the Zn–Fe alloy layers leading to local damage of the zinccoating (Fig. 3.6b). This shows that the larger the galvanized coatingdiscontinuity to be protected, the higher the corrosion of adjacent zinc.

Concerning the specimens manufactured in this work, the experimentalresults obtained demonstrate the beneficial protective effect against steelcorrosion by zinc coating, even when discontinuities as large as 4 cm are

3.6 Metallographic pictures of galvanized bar cross-section in a zoneclose to the joint with the black steel bar when Sa/Sc = 7.5: (a) thegalvanized bar was embedded in the specimen exposed to wet–drycycles with tap water; (b) the galvanized bar was embedded in thespecimen exposed to wet–dry cycles with NaCl solution.



present and a high chloride level has reached the bars. However, a largerdiscontinuity produces higher corrosion in the adjacent zinc than a smallerone and the galvanized rebars offer more effective cathodic protection whenthe discontinuity is smaller.

3.4 Conclusions

Experimental tests were performed to simulate field conditions for reinforcingbars in concrete structures. Pieces of bare steel immersed in chloride-freeconcrete and in concrete contaminated with chlorides were coupled withgalvanized steel with the aim to simulate the defects of galvanized surfacedue to bending or welding, that might be cathodically protected by the adjacentzinc coating.

The short circuit potentials measured indicate that the zinc of galvanizedsteel exerts cathodic protection on bare steel for all Sa/Sc ratios examined. Inparticular, it exerts a ‘cathodic prevention’ against chloride attack by increasingthe chloride concentration threshold able to induce the localized corrosionon bare steel.

The damage to the zinc coating depends on the aggressiveness of theenvironment surrounding the bars and on the dimensions of the defect. Indetail, in the absence of chlorides the ‘macrocell effect’ is very low, independentof the Sa/Sc ratio. However, in more aggressive environments, because of thepresence of chlorides, zinc coating always increases the durability ofreinforcement by a beneficial protective effect against steel corrosion, evenwhen large discontinuities on the coating are present and a high chlorideslevel has reached the bars. In these conditions, even though the discontinuitiesup to 4 cm resulted protected from the adjacent zinc, discontinuities greaterthan 1 cm can be considered critical for corrosion phenomena of the zinccoating.

3.5 References

1. D. Stark and W. F. Perenchio, Final Report Project No. 2E-206, Costr. Technol. Lab.,1975, 80.

2. J. E. Slater, Mater. Perf., 1979, 18(6), 34.3. D. Stark, Corrosion of Reinforcing Steel in Concrete, (Eds. D. E. Tonini and J. M.

Gaidis), ASTM STP 713, American Society for Testing Materials, Philadelphia,1980 p. 132.

4. K. W. J. Treadaway, B. L. Brown and R. N. Cox, Corrosion of Reinforcing Steel inConcrete, (Eds. D. E. Tonini and J. M. Gaidis), ASTM STP 713, American Societyfor Testing and Materials, Philadelphia, 1980, p. 102.

5. I. Cornet and B. Bresler, Galvanized Reinforcement for Concrete – II, InternationalLead Zinc Research Organization, New York, 1981, p. 1.

6. C. Andrade, A. Macias, A. Molina and J. A. Gonzales, Technical Symposia —Corrosion 85, Boston, 25–29 March 1985, NACE, Houston, Paper N∞ 270.



7. C. Andrade and A. Macias, Surface Coating-2 (Eds. A. D. Wilson, J. W. Nicholsonand H. J. Prosser), Elsevier Applied Science, London, 1988, p. 137.

8. G. Sergi, N. R. Short and C. L. Page, Corrosion, 1985, 41, 418.9. E. Maahn and B. Sorensen, Corrosion, 1986, 42, 187.

10. A. J. Gonzales and C. Andrade, Br. Corr. J, 1982, 17, 21.11. W. G. Hime and M. Machin, Corrosion, 1993, 10, 858.12. R. N. Swamy, Corrosion of Reinforcement in Concrete Construction, (Eds. C. L.

Page, K. W. J. Treadway and P. B. Bamforth), Elsevier Applied Science, London,1990, p. 586.

13. S. R. Yeomans, Proc. Int. Conf. Corrosion and Corrosion Protection of Steel inConcrete (Ed. R. N. Swamy), 24–28 July 1994, Sheffield Academic Press, Sheffield,1994, Vol II, p. 1299.

14. L. Bertolini, F. Bolzoni, P. Pedeferri, L. Lazzari and T. Pastore, J. Appl. Electrochem.,1998, 28(12), 1321.

15. P. Pedeferri and L. Bertolini, La corrosione nel calcestruzzo e negli ambienti naturali,McGraw-Hill, Milano, 1996, p. 70.


38

4.1 Introduction

Many authors have used laboratory tests in model solutions to determinecritical conditions for steel activation, using steel specimens with bare steelsurfaces [1–3]. Nevertheless, in real conditions the steel reinforcement isutilized with a scaled surface that, depending on the duration of atmosphericexposure, is covered with rust to a varied extent. Laboratory tests, as well aslong-term exposure tests in concrete, have revealed the negative influence ofa rust layer on the corrosion resistance of steel reinforcement in concrete.The corrosion rate of pre-rusted reinforcement in moist non-carbonated concretewas found to be unacceptable even in the case of concrete that was notcontaminated with chlorides [4, 5]. The published literature on the criticalchloride concentration causing steel activation reports a wide range of valuesover several orders of magnitude (0.04 to 35 g L–1, pH = 12.5) [6]. The so-called critical concentration ratio of Cl–/OH– is believed to be the determiningfactor for steel activation [7, 8]. However, certain theories suggest that thelaboratory determined dependence of critical chloride concentration on pHcannot, for various reasons, provide any practically usable specifications oflimiting conditions for steel activation in concrete. In regard to the acceptablechloride content, the limiting criterion for construction use of reinforcement innon-carbonated concrete is an empirical value of 0.1 to 0.2 wt. % Cl– in thecement. Depending on the degree of water saturation and the quality of theconcrete, this value corresponds to chloride concentrations in the pore solutionof approx. 2 to 10 g L–1. In carbonated concrete, chlorides do not determinesteel activation, although they affect the corrosion rate in the active state.


The corrosion behaviour of carbon steel (0.2% C) in a model pore solutionof concrete with chloride content from <0.01 to 165 g L–1 was observed.Exposure tests were carried out in a testing apparatus equipped with a

4Influence of scale and rust on steel

activation in model concretepore solution

P. N O V Á K, R. M A L Á and M . K O U Ř I L,Institute of Chemical Technology, Prague, Czech Republic


Influence of scale and rust on steel activation 39

continuous-flow system ensuring constant values of oxygen concentration(from 1 to 8 mg L–1) and alkalinity (pH 12.5 sat. Ca(OH)2, pH 13.3 sat.Ca(OH)2 + KOH) in the model solution during the whole measurement. Theaim was to determine – on the basis of electrochemical measurements ofpolarization resistance (LPR) – the critical chloride concentration at whichthe steel surface becomes activated. Measurements were performed for threetypes of steel surface: bare (as received clear metallic surface after degreasing),scaled (650∞C, 10 minutes) and pre-rusted surface (five months in an outdooratmosphere with corrosion category C3 according to ISO 9223). The averagethickness of the scale was 15 mm, the average thickness of the rust was22 mm. The testing apparatus allowed for the parallel exposure of 15 specimensand statistical data evaluation. After 4 h of spontaneous passivation in alkalinesolution with a particular pH value (without chlorides), the steel surface wasexposed for 6 h to a solution containing chlorides. Measurements of polarizationresistance were carried out by means of a Gamry measuring system with theCMS100 program.

4.3 Results

The dependence of polarization resistance on the chloride concentration(Figs. 4.1–4.4) at each value of pH show that the critical chloride concentrationscan be most precisely determined, for the bare surface at both values of pHand with a balanced concentration of oxygen, to be 1 g L–1. The pairs ofidentical lines in Figs. 4.1–4.4 express the ranges of measured values for

BareScaledPre-rusted

pH = 12.5O2 = 1 mg L–1

Rp(W

m2 )

100

10

1

0.1<0.01 0.1 1 10 100

Cl– (g L–1)

4.1 Influence of chloride content in model pore solution onpolarization resistance data of carbon steel at pH 12.5(oxygen 1 mg L–1).



each chloride concentration and each superficial state. In the case of thescaled surface, the value reaches 3 to 6 g Cl– L–1, again independent of pH.The pre-rusted surface shows a significant decrease of polarization resistanceonly in pH 13.3 solution (with chloride concentrations approx. > 2 g L–1),


pH = 12.5O2 = 8 mg L–1

Rp(W

m2 )

100

10

1

0.1<0.01 0.1 1 10 100

Cl– (g L–1)

4.2 Influence of chloride content in model pore solution onpolarization resistance data of carbon steel at pH 12.5 (oxygen8 mg L–1).


pH = 13.3O2 = 1 mg L–1

Rp(W

m2 )

100

10

1

0.1<0.01 0.1 1 10 100

Cl– (g L–1)

4.3 Influence of chloride content in model pore solution onpolarization resistance data of carbon steel at pH 13.3(oxygen 1 mg L–1).



whereas at pH 12.5 no considerable decrease was observed with increasingchloride concentration. Compared with the bare and scaled surfaces, thecorrosion resistance of the pre-rusted surface was the lowest in the region oflow chloride concentrations (<1 g L–1). This concentration region correspondsto the unstable passive state of scaled and metallic surfaces during short-term exposure; in the case of pre-rusted surfaces even the partial activationcan be expected.

With respect to a large scatter of polarization resistance values acquired(in the range of one order of magnitude), the results were evaluated on thebasis of comparison of all the values obtained in the respective chlorideconcentration interval (see Tables 4.1 and 4.2). The short-term tests in modelsolution show the value of 2 Wm2 that allows for distinguishing the unstable

4.4 Influence of chloride content in model pore solution onpolarization resistance data of carbon steel at pH 13.3 (oxygen8 mg L–1).


pH = 13.3O2 = 8 mg I–1

Rp(W

m2 )

100

10

1

0.1<0.01 0.1 1 10 100

Cl– (g l–1)

Table 4.1 Portion of values of polarization resistance £ 2 W m2, model pore solutionpH 12.5 [sat. Ca(OH)2], in bold when portion ≥50%

Surface Oxygen Portion of Rp £ 2 W m2 (%)(mg L–1)

<0.01 g L–1Cl– 0.05–1 g L–1Cl– >1–10 g L–1Cl– >10 g L–1Cl–

Bare 8 5 4 100 100

1 0 0 55 100

Scaled 8 13 25 50 100

1 0 8 42 100

Pre-rusted 8 67 67 100 100

1 83 100 100 100



passive surface from the partial surface activation. It is in contrast to thelong-term tests, where the polarization resistance value corresponding withtechnically acceptable corrosion rate of carbon steel < 1 to 2 mm year–1 wasdetermined to be >30 Wm2 [9]. The number of measurements in each intervalof chloride concentrations was 6 to 30. The average values of polarizationresistance for each interval are listed in Tables 4.3 and 4.4.

Table 4.2 Portion of values of polarization resistance £ 2 W m2, model pore solutionpH 13.3 [KOH + sat. Ca(OH)2], in bold when portion ≥ 50%

Surface Oxygen Portion of Rp £ 2 Wm2 (%)(mg L–1)


Bare 8 44 14 100 100

1 28 14 83 100

Scaled 8 40 29 67 77

1 17 29 67 70

Pre-rusted 8 83 67 100 100

1 100 100 100 100

Table 4.3 Average values of polarization resistance, model pore solution pH 12.5[sat. Ca(OH)2], in bold when Rp £ 2 W m2

Surface Oxygen Rp (W m2)(mg L–1)


Bare 8 20.8 8.2 0.2 0.1

1 39.7 54.8 13.3 0.5

Scaled 8 8.3 3.7 2.5 0.2

1 29.6 9.2 3.1 0.5

Pre-rusted 8 2.3 2.0 0.5 0.2

1 1.2 1.0 0.3 0.1

Table 4.4 Average values of polarization resistance, model pore solution pH 13.3[sat. Ca(OH)2 + KOH], in bold when Rp £ 2 W m2

Surface Oxygen Rp (W m2)(mg L–1)


Bare 8 4.5 4.6 0.5 0.4

1 4.2 3.0 1.0 0.5

Scaled 8 3.1 3.0 2.0 0.9

1 4.3 2.9 1.2 1.6

Pre-rusted 8 1.4 2.2 0.6 0.3

1 0.6 0.4 0.4 0.3



4.4 Conclusions

On the basis of short-term measurements it was found that a scale layer onthe steel surface increases the critical chloride concentration for steel activationcompared with that for a bare surface by a factor of 3 to 6 times. Thebreakdown of the passive state on the bare steel surface proceeds in alkalinesolution at a balanced concentration of atmospheric oxygen when the chlorideconcentration exceeds 1g L–1. Lower oxygen concentration leads to an increasein critical activation concentration of chlorides for both bare and scaledsurfaces. No significant shift in critical chloride concentration was observedon increasing the pH of the pore solution from 12.5 to 13.3, correspondingto a six-fold increase in the OH– concentration. The results prove that theCl–/OH– concentration ratio has no practical significance on the corrosionaggressivity of concrete towards steel reinforcement. In the region of lowchloride concentrations, the pre-rusted surface showed the lowest corrosionresistance of all the surfaces studied, and the results confirm the negativeeffect of the rust layer formed by atmospheric exposure. Increasing the solutionalkalinity did not explicitly lead to the spontaneous passivation of steel witha pre-rusted surface even in solutions with very low chloride content.

4.5 Acknowledgements

The authors aknowledge financial support of this research, which was thepart of the Czech Grant Agency Project 103/02/0282 and MSM 223100002project.

4.6 References

1. A. K. Suryavanski, J. D. Scantlebury and S. B. Lyon, Cem. Concr. Compos., 1998, 20,263.

2. K. Thangavel and N. S. Rengaswamy, Cem. Concr. Compos., 1998, 20, 283.3. J. F. Henriksen, Corros. Sci., 1980, 20, 1241.4. J. A. González, E. Ramirez, A. Bautista and S. Feliu, Cem. Concr. Res., 1996, 26, 501.5. P. Novák, R. Malá and L. Joska, Cem. Concr. Res., 2001, 31, 589.6. W. Breit, Mater. Corros., 1998, 49, 539.7. O. A. Kayyali and M. N. Haque, Mag. Concr. Res., 1995, 47, 235.8. L. Zimmermann, B. Elsener and H. Böhni, Corrosion of Reinforcement in Concrete,

Corrosion Mechanisms and Corrosion Protection, European Federation of CorrosionPublication No 31, IOM Communications Ltd, 2000, p. 25.

9. P. Novák and R. Malá, Corrosion of Reinforcement in Concrete, Corrosion Mechanismsand Corrosion Protection, EFC Publication No 31, IOM Communications Ltd, 2000,p. 41.


44

5.1 Introduction

An important part of our infrastructure is based on reinforced concrete, withconcrete taking the compressive load and the embedded steel the tensile loadof the structures. The durability of this composite material is based on theexcellent chemical stability of hydrated portland cement and the passivity ofsteel in the alkaline pore solution of concrete (pH 12.5–13.5) [1, 2]. Despitethis decisive fact for the durability of steel-reinforced concrete structures,comparatively little is known about the surface chemistry of iron and ironalloys in model alkaline media [3, 4], in synthetic pore solutions containingsodium, calcium and potassium cations as well as sulphates [5, 6], or inconcrete. A recent work of Joiret et al. [7] studied the dissolution and passivationof iron with electrochemical techniques and in-situ Raman spectroscopy,showing that with increasing potential a gradual oxidation of the oxide filmis observed. At very low potentials magnetite (Fe3O4) is present, at moreanodic potentials a-FeOOH and Fe2O3 are formed. These results are inagreement with short time experiments on sputtered iron immersed in alkalinesolutions [3].

From a thermodynamic point of view iron oxides or oxyhydroxides arestable compounds at high pH, leading to the formation of a thin protectiveoxide film (passive film) on the iron surface in alkaline media and concrete.This passive state of the reinforcement can be destroyed by carbonation(reaction of the alkaline pore solution with CO2 from the atmosphere) andthe subsequent drop in pH or by the ingress of chloride ions, leading tolocalised corrosion attack [2, 8]. It is well known from the literature andpractical experience that prolonged exposure of steel to alkaline media (ageingof the passive film) increases the critical chloride content for the initiation ofchloride-induced localised corrosion, as has been reported for pore solution,mortar and concrete [9, 10]. Ageing of the passive film at the same timedecreases the efficiency of cathodic oxygen reduction [10] and thus mayreduce the rate of corrosion [10]. In conclusion, the thickness, composition

5The surface of iron and Fe10Cr alloys in

alkaline media

A. R O S S I, G. P U D D U and B. E L S E N E R,University of Cagliari, Italy


The surface of iron and Fe10Cr alloys in alkaline media 45

and electronic configuration of these thin iron oxide films (passive films) inalkaline media are of great importance for the corrosion behaviour of steel inconcrete.

In this work, the results of a combined electrochemical and XPS surfaceanalytical study of the surface of iron and Fe10Cr alloys in alkaline mediaare reported. The results obtained are discussed with respect to the ironFe(II)/Fe(III) ratio in the film and its relation to the open circuit potentialand to the influence of the chromium addition on the stability of the passivefilm.


5.2.1 Materials and sample preparation

Iron sheets (99.99%) were supplied by GoodFellow and the Fel0Cr alloyswere prepared by the Institute of Physical Chemistry and Electrochemistryof the University of Düsseldorf. The samples – already fixed on the XPSsample holder – were ground in bi-distilled water using 200, 320, 500 and1000-grit silicon carbide paper and in ethanol to a 1 mm finish with diamondpaste (this state is called mechanically polished, m.p.). They were washedwith ethanol, dried under a nitrogen stream and transferred under nitrogen tothe spectrometer.

Samples were analysed by XPS immediately after immersion in deaeratedalkaline solutions of pH 13 and after exposure for prolonged time (up to 20days) to air at 35% relative humidity and subsequent immersion in alkalinesolution for 20 hours.

5.2.2 Reagents and solutions

Reagent-grade ethanol and bi-distilled water (l = 1.4 mS cm–1 at 20 ∞C; pHª 6.5) were used for mechanical polishing. The alkaline solutions of pH 13were prepared from NaOH of analytical grade (Carlo Erba). Solution pH wasmonitored using a Metrohm 654 pH meter. Deaereated solutions were preparedby argon gas bubbling for at least four hours.

5.2.3 Electrochemical experiments

The electrochemical experiments have been carried out in a cylindricalelectrochemical cell with an opening of diameter 1 cm (surface area0.78 cm2) at one side in order to expose the sample surface to the solution.Solutions saturated with oxygen and deaerated with argon gas were used.The open circuit potential values were recorded with a PAR 273 potentiostatunder computer control. A saturated calomel electrode (SCE) with a doubleprotection diaphragm was used for all experiments.



5.2.4 X-ray photoelectron spectroscopy

XPS analyses were performed in an ESCALAB 200 spectrometer (VacuumGenerators Ltd., UK). The vacuum system consisted of a turbomolecularpump, fitted with a liquid nitrogen trap, and a titanium sublimation pump.The residual pressure in the spectrometer during the data acquisition wasalways lower than 5 ¥ 10–7 Pa. The X-ray source was AlKa (1486.6 eV), runat 20 mA and 15 kV. The spectra were obtained in the digital mode (VGEclipse software on IBM 486). The electron analyser was operated in FixedAnalyser Transmission (FAT) mode with a pass energy of 20 eV Full Widthat Half Maximum height (FWHM) Ag 3d5/2 = 1.1 eV. The analysed area wasca. 0.5 cm2. The instrument was calibrated using the inert-gas-ion-sputter-cleaned reference materials SCAA90 of Cu, Ag and Au [11]. For calibrationpurposes the Au 4f7/2 line at 83.98 eV, the Cu 2p3/2 line at 932.67 eV, theCu LMM signal at 334.94 eV and the Ag3d5/2 at 368.26 eV were taken. Thespectra were resolved into their components after background subtractionaccording to Shirley and Sherwood [12]. The Gaussian /Lorentzian ratio andthe FWHM were determined on standards and held constant, the peak energyand height were fitted using a least-squares algorithm.

5.2.5 Quantitative analysis

From the integrated peak intensities the thickness and composition of thesurface film were determined with a three-layer model [13] taking into accountthe attenuation of the photoelectrons by the hydrocarbon contamination layerand the passive film.

5.3 Results

5.3.1 Electrochemical results

Open circuit potential (OCP) measurements have been carried out on ironand Fe10Cr samples in alkaline solutions at pH 13 immediately after mechanicalpolishing and after mechanical polishing and air oxidation for different timeperiods in a dessicator. In Fig. 5.1 examples of the OCP versus time curvesfor mechanically polished samples of Fe and Fe10Cr immersed in NaOHsolution at pH 13.0 are shown for solutions saturated with O2 and deaeratedwith Ar. The OCP of the mechanically polished iron samples (Fig. 5.1a)immediately after immersion is very negative, –0.6 V (SCE), and it increasesrapidly to reach a more positive value of –0.35 V (SCE, deaereated) and of–0.15V (SCE, O2) after immersion for 12 h. The Fe10Cr samples (Fig. 5.1b)exhibit similar behaviour but the open circuit potential of the mechanicallypolished samples after 12–14 h is more negative than those of iron both indeaerated and aerated solutions.



A second series of experiments has been performed using iron and Fe10Crsamples mechanically polished and then stored in a dessicator for timesranging between 1 day and 20 days at 25∞C ± 0.5 ∞C and relative humidityof 35%. These air-exposed samples were then immersed in alkaline solutionat pH 13.0 deaerated with argon gas and the OCP was measured for 20 h; theresults are shown in Fig. 5.1. The open circuit potential for both pure iron(Fig. 5.1a) and Fe10Cr samples (Fig. 5.1b) exposed to air immediately afterimmersion in the solutions is much more positive than for the mechanicallypolished samples. Iron samples exposed to the dry air start with an OCP at–0.4 V (SCE), at longer immersion times the potentials become identicalwith those for the mechanically polished samples. The air-exposed Fe10Crsamples start at an open circuit potential of ca. –0.3 V (SCE); in the firsthours of exposure to the solution the OCP decreases but remains more positivethroughout the experiment than the values for mechanically polished samples

OC

P v

s S

CE

(V

)

–0.1

–0.2

–0.3

–0.4

–0.5

–0.6

–0.7

O2

Ar

In air for 17 days

A

0 1.7 ¥ 104 3.4 ¥ 104 5.1 ¥ 104 6.8 ¥ 104

Time (s)

OC

P v

s S

CE

(V

)

–0.1

–0.2

–0.3

–0.4

–0.5

–0.6

–0.7

O2

Ar

B

0 1.7 ¥ 104 3.4 ¥ 104 5.1 ¥ 104 6.8 ¥ 104

Time (s)

In air for 4 days

5.1 Open circuit potential versus time for mechanically polished (—)and air exposed (- - -) iron (A) and Fe10Cr (B) samples immersed inNaOH solutions of pH 13.



in deaerated solution (Fig. 5.1b). Similar behaviour was found for air exposuretimes between 1 and 20 days.

5.3.2 X-ray photoelectron spectroscopy results

X-ray photoelectron spectroscopy was conducted on iron and Fe10Cr samples,both mechanically polished and immersed in NaOH solution and mechanicallypolished, air oxidised in a controlled atmosphere and immersed in NaOHsolution at pH 13.

Mechanically polished surface

The surface state after mechanically polishing in ethanol was chosen as areproducible reference and starting state of the two alloys under study. Ascan be noted from the Fe2p spectra of pure iron after mechanical polishing(Fig. 5.2a) the signal of iron metal is clearly detectable both in the Fe2p3/2

(706.7 eV) and in the Fe2p1/2 region of the spectrum. The peak maximum ofthe oxidised iron is found at 710.2 eV, suggesting a major contribution to thissignal of Fe(II) in the surface film. Similar results were found for the Fe10Cralloy (Fig. 5.2b).

A more detailed analysis of the iron Fe2p3/2 and the oxygen O1s signalafter background subtraction and curve fitting is shown in Fig. 5.3 (exampleof Fe10Cr alloy). The full widths at half maximum height (FWHM) and theposition of the individual peaks in the iron Fe2p3/2 region used for curvefitting were held constant (see Table 5.1), only the peak height was allowedto vary. As can be seen from Fig. 5.3a, the iron Fe2p3/2 signal can be resolvedinto different contributions: that of metallic iron (706.7 eV), a Fe(II) component(708.8 eV), Fe(III) oxide (710.3 eV) and Fe(III) oxyhydroxide (712.1 eV)[3]. Note that the Fe(II) component has a satellite signal at 714.5 eV (10%of the main peak) which has been fitted in this work. The oxygen O1s spectra(Fig. 5.3b) were fitted with three Gaussian/Lorentzian peaks found at 530.0± 0.05 eV (oxide), 531.6 ± 0.1 eV (hydroxide) and 532.9 ± 0.1 eV (adsorbedwater). Samples of mechanically polished iron showed similar results, thecontribution of the Fe(III) oxyhydroxide is higher than in the mechanicallypolished Fe10Cr alloy. Quantitative analysis of the oxide film on the Fe10Cralloy after mechanical polishing indicated a composition of 16% chromiumoxide (Table 5.2); the alloy beneath the film had the nominal composition.

Mechanically polished and immersed in alkaline solution of pH 13

The XPS spectra after immersion in alkaline solution of pH 13 changed: thesignal of metallic iron became much less pronounced, indicating a higherfilm thickness, and the maximum of the iron oxide peak was shifted to



Inte

nsi

ty (

a.u

.)In

ten

sity

(a.

u.)

Fe

Mech polished

m.p. and20 h pH 13

740 735 730 725 720 715 710 705 700Binding energy (eV)

740 735 730 725 720 715 710 705 700Binding energy (eV)

Fe10Cr

Mech. polished

20 h at pH 13

5.2 XPS spectra of iron Fe2p recorded on iron (upper) and Fe10Crsamples (lower) after mechanical polishing and after subsequentimmersion for 20 h in deaerated NaOH solution at pH 13.

Table 5.1 Peak assignment, position and FWHM of the individual iron Fe2p3/2

compounds in oxidised surface films of iron and Fe10Cr alloys

Assignment Position KE Binding Energy FWHM Comments(eV) (eV) (eV)

Iron metal 779.9 ± 0.1 706.7 1.3 exp 0.63Fe(II) oxide 777.80 708.8 2.6 G/L* = 0.45Fe(III) oxide 776.30 710.3 2.6 G/L = 0.45Fe(III) 774.50 712.1 3.0 G/L = 0.45oxyhydroxide

*Gaussian/Lorentzian



higher binding energies (Fig. 5.4). The more detailed analysis after curvefitting of the Fe2p3/2 spectrum of the mechanically polished Fe10Cr alloyfollowing immersion for 20 h at pH 13 (Fig. 5.4a) revealed that the Fe(II)component at 708.8 eV had markedly decreased in intensity (to <10%) and

Inte

nsi

ty (

a.u

.)Fe2p3/2/Fe10Cr mp

716 714 712 710 708 706 704Binding energy (eV)

Inte

nsi

ty (

a.u

.)


O1s/Fe10Cr mp

5.3 Fe2p3/2 and O1s spectra of mechanically polished Fe10Cr alloyafter background subtraction and curve fitting (for parameters seeTable 5.1).

Table 5.2 Thickness and composition of the oxide film and the composition of thealloy beneath the film of mechanically polished Fe10Cr alloys immersed fordifferent times in deaerated alkaline solutions of pH 13

Time Thickness Oxide film Alloy beneath the film(nm)

% Fe % Cr % Fe % Cr

Mechanically 2.0 84 16 90 10polished16 h 2.6 75 25 91 925 h 2.6 66 34 93 73 days 2.7 63 37 94 6



that the major contribution was then due to the component assigned to Fe(III)oxide. The same occurred on mechanically polished pure iron after immersionat pH 13, but the decrease of the Fe(II) component was less pronounced, theFe(II) component remained at ca. 20%. The oxygen O1s spectra of iron andFe10Cr alloy (Fig. 5.4b) after immersion were very similar to each other.The composition of the oxide film and of the interface beneath the oxide filmtogether with the thickness of the oxide layer were evaluated using the peakintensities of the different components and the three-layer model successfullyemployed in other samples [13]. The results of the quantitative analysis ofthe Fe10Cr alloy (Table 5.2) indicate an increase in the chromium content inthe oxide film from 10 to 25 % after 16 h of immersion. Prolonged immersionof the Fe10Cr alloy for up to 3 days led to an oxide film composition of ca.35% Cr, thus markedly enriched in chromium oxide (Table 5.2). In parallel,prolonged immersion led to depletion of the chromium content in the alloybeneath the oxide film (Table 5.2).

Inte

nsi

ty (

a.u

.)Fe2p3/2/Fe10Cr mp

20 h pH 13.0


Inte

nsi

ty (

a.u

.)


O1s/Fe10Cr mp20 h pH 13.0

5.4 Fe2p3/2 and O1s spectra of Fe10Cr alloy, mechanically polishedand immersed for 20 h at pH 13 after background subtraction andcurve fitting (for parameters see Table 5.1).



Mechanically polished and exposed to air (35% rh)

A series of experiments were performed to study the influence of air exposureon the surface film composition. The Fe2p3/2 and O1s spectra were resolvedinto their components, as described above, and the relative intensities(percentage) of the oxidised iron signals Fe(II), Fe(III) and FeOOH weredetermined. As can be seen from Fig. 5.5, the main effect of exposure to dryair is a marked decrease in the Fe(II) component and a corresponding increasein the intensity of the Fe(III) and FeOOH signals (Fig. 5.5a). This filmtransformation was complete only after one day, and prolonged exposure todry air for up to 20 days did not change the composition of the surface film.

Fe(II) %Fe(III) %FeOOH %

MOH %MO %H2O %

Inte

nsi

ty r

atio

%

80

70

60

50

40

30

20

10

00 1 5 10 15 20

Time of air exposure (d)(a)

Inte

nsi

ty r

atio

%

80

70

60

50

40

30

20

10

00 1 5 10 15 20

Time of air exposure (d)(b)

5.5 Effect of the exposure time to air on the composition of the airformed film on pure iron. (a) iron compounds, (b) oxygencompounds.



However, the film thickness increased from 1.8 nm to 2.4 nm after 20 days(Fig. 5.6). The relative intensity of the oxygen O1s signal for different timesof air exposure (Fig. 5.5b) shows a more gradual change: the percentage ofthe component at the highest binding energy (ca. 533 eV) decreases withtime whereas the hydroxide (binding energy ca. 531.8 eV) increases slightly.

The general trend of the Fe10Cr alloys is similar to that for iron, but it isworth pointing out that the Fe(II) intensities were lower from the beginningof air exposure. The same holds true for the oxygen component at 533 eV.The film thickness of the air formed oxide film on Fe10Cr alloys increasedfrom 2.0 to 4.0 nm after exposure to air for 20 days (Fig. 5.6). The thin oxidefilm formed after mechanical polishing was found to contain 16% of oxidisedchromium on average. After air exposure from 1 to 20 days, it remainedconstant at the nominal composition of 10 ± 1% oxidised chromium.

Air oxidised and immersed in alkaline solutions

The air-exposed samples of pure iron and Fe10Cr alloy were immersedsubsequently for 20 hours in alkaline solution of pH 13. The Fe2p3/2 and Olsspectra were resolved into their components, as described above, and therelative intensities of the oxidised iron signals, Fe(II), Fe(III) and FeOOH,were determined. Upon immersion of the air-exposed Fe10Cr alloys part ofthe air-formed oxide film dissolved (Fig. 5.6) and a film thickness of3.0 ± 0.2 nm resulted after 20 h of immersion. Parallel to the decrease of the

Air exposedImmersedAir exposedImmersed Fe10Cr

Fe

0 5 10 15 20 25Time of air exposure (d)

Oxi

de

film

th

ickn

ess

(nm

)

4.5

4

3.5

3

2.5

2

1.5

5.6 Thickness of the oxide film of mechanically polished iron andFe10Cr alloys after air exposure and air exposure with subsequentimmersion for 20 h in solution of pH 13.



film thickness the content of oxidised chromium in the film increased from10 to 20% (Table 5.3). The greatest increase in chromium content was foundfor the samples with long exposure to air and correspondingly high filmthickness (Fig. 5.6).

5.4 Discussion

5.4.1 Air-formed oxide films

After mechanical polishing in ethanol, a very thin oxide film (1.9 ± 0.1 nm)containing up to 50% Fe (II) is formed on the surface of iron and Fe10Cr alloy.After exposure to dry air, this film is further oxidised and the amount of Fe(II)ions in the film drops to about 15% with a corresponding increase of theFe(III) content in the film. This is in agreement with results that have beenreported for sputtered iron surfaces exposed to oxygen or air [14]. It isinteresting to note that only 24 h of air exposure are sufficient to induce thisoxidation reaction, a further air exposure for up to 20 days does not significantlyalter the Fe(II)/Fe(III) ratio in the film both for iron (Fig. 5.5) and Fe10Cralloy. Changes are found instead in the film thickness (Fig. 5.6): the filmthickness of the oxide film on pure iron increased from 1.9 ± 0.1 nm of themechanically polished sample to 2.4 ± 0.1 nm after 20 days. On the Fe10Cralloy, the increase in film thickness was more pronounced (Fig. 5.6); the oxidefilm composition was close to the nominal (10 ± 1% Cr) composition of thealloy. Thus, the presence of chromium in the alloy accelerates the formationof an air-formed oxide film. The plot of film thickness vs. log t (t = time ofair exposure) results in a straight line indicating a power law for film growth.

5.4.2 Corrosion potentials

The corrosion potential of passive reinforcing steel in concrete structures ismainly determined by the cathodic oxygen reduction reaction and, thus, by

Table 5.3 Thickness and composition of the oxide film and composition of thealloy beneath the film on mechanically polished Fe10Cr alloys air exposed fordifferent times and then immersed for 20 h in deaerated alkaline solutions ofpH 13

Time Thickness Oxide film Alloy beneath the film(nm)

% Fe % Cr % Fe % Cr

Air exposed see Fig. 5.6 90 ± 1 10 ± 1 92 ± 1 8 ± 11 day 2.8 88 12 92 85 days 3.2 88 12 92 810 days 3.3 88 12 93 715 days 3.2 84 16 92 820 days 3.1 79 21 90 10



the pH of the pore solution and the availability of oxygen; corrosion potentialsin the range of –0.15 to 0 V (SCE) are observed for aerated structures [2]. Inthis laboratory study, working with well-defined mechanically polishedsurfaces, the influence of the surface state before immersion (mechanicallypolished, air exposure) and of alloying with 10% chromium has beeninvestigated. As can be noted from Fig. 5.1, in the presence of oxygen morepositive potentials are found both for iron and the Fe10Cr alloy, in agreementwith the results for steel in concrete [2, 7]. Whereas freshly polished samplesexhibit very negative initial corrosion potentials of around –0.6 V (SCE),samples exposed to air for several days show more positive initial corrosionpotentials in deaerated solution due to the presence of the air-formed oxidefilm (Fig. 5.1).

This behaviour can be rationalised further by combining the electrochemicaldata with surface analytical information. As established by Haupt et al. [3]in a laboratory study on sputtered (oxide-free) iron exposed to alkaline solutionsat pH 13, the Fe(II)/Fe(III) ratio in the oxide film depends on the electrochemicalpotential of film formation (Fig. 5.7): at very negative potentials, below–0.9 V (SCE), only Fe(II) species are found in the oxide film, whereas atmore positive potentials, around –0.1 V, the films contain only about 10% ofFe(II). In this work, the oxide films on the surface of iron and Fe10Cr alloyswere formed naturally at the open circuit potential (without anodic polarisation)in alkaline solutions. By plotting the percentage of Fe(II) and Fe(III) oxidesin the surface film, as determined by XPS analysis in this work, versus thefinal value of the open circuit potential (Fig. 5.1), a clear trend can beobserved (Fig. 5.8): with increasing open circuit potential the percentage of

Fe (II)Fe (III)

–1 –0.8 –0.6 –0.4 –0.2 0Potential [V (SCE)]

Am

ou

nt

Fe(I

I) a

nd

Fe(

III)

(%)

100

80

60

40

20

0

5.7 Change of Fe(II) and Fe(III) content in the passive film on ironexposed to pH 13 solution with the potential of film formation(according to reference 3).



Fe(II) in the film diminishes. Despite the limited range of the OCP values(–0.4 V < E < –0.1 V SCE), the observed variations are between 20 and ca.5%, in good agreement with earlier XPS surface analytical results [3], withresults from in situ Raman spectroscopy [7], and with bulk oxides immersedin alkaline solutions at the same pH.

The Fe10Cr alloy follows the same trend as pure iron, showing a slightlylower Fe(II) content at the same open circuit potential. As additionalconfirmation, freshly polished iron samples showed very negative potentialsupon immersion in alkaline solution of pH 13 (Fig. 5.1). The percentage ofFe(II) in the surface film, as determined by XPS analysis, is ca. 50% (Fig.5.3 and Fig. 5.5) in very good agreement with results in the literature for afilm formation potential of –0.6 V (SCE). Thus, the initial surface film richin Fe(II) becomes oxidised in alkaline solutions and the open circuit potentialincreases in agreement with an increase in the Fe(III) content. The air-formedfilms instead show initial values of open circuit potential of around –0.3 V(SCE) due to the much lower Fe(II) content of ca. 15%.

5.4.3 Film growth and dissolution

The XPS data on mechanically polished iron and Fe10Cr alloy indicate thatimmersion of mechanically polished samples in alkaline solution for 20 hleads to an increase in oxide film thickness of 0.2 nm for pure iron and of0.5 nm for the Fe10Cr alloy (Fig. 5.6). Both materials are passive and thefilm growth is associated with the increase in open circuit potential(Fig. 5.1). This has been confirmed by recent experiments combining

–0.45 –0.4 –0.35 –0.3 –0.25 –0.2 –0.15 –0.1Potential [V (SCE)]

Am

ou

nt

Fe(I

I) a

nd

Fe(

III)

(%)

100

80

60

40

20

0

Fe (III)

Fe (II)

5.8 Open circuit potential and Fe(II) �� and Fe(III)ox �� content inthe surface films after exposure of iron (��), Fe 10Cr alloys (��)and bulk oxides (▼ ▲) [14] to alkaline solution at pH 13.



potentiostatic passivation and quartz crystal microbalance (QCM)measurements [15] on mechanically polished iron in alkaline solutions ofpH 13: in the potential region from –0.4 to 0 V a slight mass increase isobserved during 1 h of polarisation. On the Fe25Cr alloy the mass increasewas more pronounced [15]. The growth in film thickness observed in thiswork and the mass increase associated with a shift in the electrode potentialto more positive values can be explained by film transformation from magnetite[Fe3O4, 50% of Fe(II)] to Fe2O3 (only Fe(III)), found by in situ Ramanspectroscopy [7, 21], according to the reaction

2Fe3O4 + 2OH– = 3Fe2O3 + H2O + 2e– (5.1)

This reaction is reversible, thus it can contribute to film growth in theanodic direction and to film dissolution in the cathodic direction. In theoxidation direction the reaction is accompanied by the uptake of one oxygenatom into the lattice.

On the air oxidised surfaces, a reduction of the oxide film thickness wasobserved after 20 h of immersion in alkaline solution of pH 13 (Fig. 5.6).This was negligible for pure iron but up to 0.5 nm for Fe10Cr alloys. Duringthis process the open circuit potential remained constant or increased onlyslightly (Fig. 5.1), thus the constant film thickness of the pure iron is consistentwith the film transformation reaction discussed above. In the case of theFe10Cr alloy, additional reactions must occur that are not primarily associatedwith a change in the valence state of iron. As can be noted from Table 5.3,the immersion of air-formed oxide films of Fe10Cr alloys leads to an increasein the content of oxidised chromium in the film, thus the iron component ofthe oxide film is preferentially dissolved.

Summarising the results from this XPS surface analytical study, togetherwith results from the literature [3, 7, 21], a quite consistent picture of thesurface of iron in alkaline media is obtained (Fig. 5.9): at low potentials bothby ex situ XPS and in situ Raman spectroscopy [7, 21] mainly magnetite,Fe3O4 [Fe(II) to Fe(III) ratio 1:1], is found. With increasing potential (eitherimposed potentiostatically or from natural immersion at the open circuitpotential) the oxide film thickens and gradually transforms to a-FeOOH andFe2O3, as indicated by in situ Raman spectroscopy [7, 21], and the Fe(II)content practically disappears (Fig. 5.9).

5.4.4 Application to steel in concrete

In addition to naturally occurring wetting and drying cycles, electrochemicalrestoration techniques influence the potential of steel in concrete. Duringelectrochemical chloride removal or electrochemical realkalisation, the steelis polarised to very negative potentials. For example, after switching off thecurrent in a realkalisation experiment potentials of ca. –1.1 to –0.8 V (SCE)



have been measured [16]. Using half-cell potential measurements as a diagnostictechnique [17] for the success of electrochemical restoration techniques,these negative potentials could be misinterpreted as a corrosion state, ratherthan repassivated rebars [18]. The results of the surface analysis undertakenin this study allow this interpretation to be ruled out: the very negativepotentials during the electrochemical treatment change the composition ofthe oxide film to mainly Fe(II) and the long time required (in practice severalweeks [16]) to achieve a stable, more positive potential for steel in concreteis due to the reoxidation to Fe(III).

The question of pre-rusted steel embedded in mortar or concrete has notbeen addressed in this work as all the iron and Fe10Cr surfaces were preparedby mechanical polishing or by exposure to dry air. In the case of a localisedcorrosion occurring on the otherwise passive steel surface, the reduction ofthe Fe(III) in Fe2O3 to Fe(II) in Fe3O4 might be an additional reductionreaction in parallel with oxygen reduction, as proposed for atmosphericcorrosion [19], that might increase the corrosion rate or sustain a corrosionreaction in the absence of oxygen. The amount of cathodic current of this

5.9 Schematic summary of the electrochemical and surface analyticalresults of iron and Fe10Cr alloy in alkaline solutions. Top: Fe(II) /Fe(III) ratio of this work and [3], middle: iron oxides identified byRaman spectroscopy [21], bottom: polarisation curve with initial andfinal OCP of mechanically polished samples (this work).

This work

Fe(II)

Fe(III)

–0.9 –0.8 –0.7 –0.6 –0.5 –0.4 –0.3 –0.2 –0.1Potential [V(SCE)]

100

80

60

40

20

0

Fe2O3

a-FeOOH

Fe3O4

Initial 24 h

Potential

Am

ou

nt

Fe(I

I) a

nd

Fe(

III)(

%)



origin is determined by the quantity of Fe(III) ions present in the oxide filmand, thus, by the thickness of the oxide or rust layer. A rough estimationshows that a (thick) rust layer of 1 mm corresponds to 300–600 mC cm–2,thus a cathodic reduction current of ca. 1 mA cm–2 for several days may beproduced (without taking account of possible kinetic hindrance). Furtherstudies are required in this respect.

When electrochemical measurements (e.g. polarisation resistance,impedance spectroscopy and pulse techniques) are applied, usually polarisingthe sample by ± 20 mV at the open circuit potential, the valence change inthe iron oxide film (oxidation, reduction) may produce a current flow due tothe Fe(II)/Fe(III) redox couple. This redox couple current could lead to a lowfrequency time constant in impedance spectra (pseudo-capacitance) andinfluence the corrosion rate values obtained [20, 21], but the magnitude ofthis charging current is not yet known. In the light of the surface analyticalresults obtained in this work, a small potential change of ± 20 mV would notbe expected to affect the Fe(II)/Fe(III) ratio (Fig. 5.8).

Anodic polarisation (polarisation curve, cyclic voltammetry) leads to anadditional oxidation of the film, film thickening and the presence of onlyFe(III) can be expected. During the reverse cycle of the voltammetry, anadditional contribution in the cathodic curves will be measured that can beattributed to the film dissolution and reduction reaction.

The corrosion resistance of pure iron and Fe10Cr alloys has not beenstudied in this work. The results of this surface analytical study neverthelessallow some of the results in the literature to be interpreted. Ageing of thepassive film of pure iron or mild steel has been found to improve the resistanceto pitting attack by increasing the pitting potential [10] and to decrease theefficiency of the cathodic reduction of oxygen [10]. Ageing has been shownto influence markedly the pitting potential of steel in alkaline solutions [10]and might also influence the ‘critical chloride content’. This may be explainedby the film growth asssociated with a decrease in the Fe(II) content in thepassive film, leading to an electronically less defective passive film. Theimproved corrosion resistance of chromium steels in concrete, and thus theirmuch higher critical chloride content for depassivation [22], may be explainedby the marked enrichment of the passive film in oxidised chromium found inthis work.

5.5 Conclusions

The results have been reported of an electrochemical and XPS surface analyticalstudy on the model systems iron and Fe10Cr alloy in NaOH solution of pH13, simulating the concrete pore environment. As a starting point, iron andFe10Cr samples were studied after mechanical polishing in ethanol. TheXPS results show that a thin oxyhydroxide film is present, the oxidation state



of iron being mainly Fe(II). These mechanically polished samples weresubsequently subjected to conditions simulating natural exposure.

Immediately after immersion in alkaline solutions of pH 13, very negativeOCP values were recorded. Continuing immersion resulted in asymptoticallyincreasing, more positive OCP values. The OCP values depend on alloycomposition and on the degree of aeration (presence or absence of oxygen)of the solution. XPS surface analysis shows that the surface film after immersionis mainly formed of Fe(III) oxyhydroxide. After prolonged immersion of theFe10Cr alloy up to 35–40% of oxidised chromium is present in the passivefilm.

Samples exposed for a prolonged time (up to 20 days) to air (35% relativehumidity) showed progressive oxidation of the surface film and a gradualchange of Fe(II) to Fe(III). The chromium oxide content of the Fe10Cr alloyremained constant at the nominal composition 10 ± 1%. After immersion ofthese oxidised samples in alkaline solutions, the initial OCP was much morepositive than that of mechanically polished samples.

Immersion of air oxidised samples in alkaline solutions of pH 13 led to atransformation of the oxide film. Especially for the Fe10Cr alloy, the oxidefilm thickness decreased and after 20 h of immersion a chromium oxidecontent of 20% was reached.

The results of this study allow the electrochemical behaviour of iron andFe10Cr alloy in alkaline solutions to be correlated with the surface chemistryof the oxide films. It can be concluded that the value of the open circuitpotential (OCP) in alkaline solutions is strongly related to the percentage ofFe(II) and Fe(III) in the film, more positive OCP values corresponding to ahigher Fe(III) content in the film.


The authors are pleased to acknowledge the financial contribution of RegioneAutonoma della Sardegna (RAS), of the Italian National Research Council(CNR) and of the Italian Ministry of University and Scientific and TechnologicalResearch. The Institute of Physical Chemistry and Electrochemistry of theUniversity of Düsseldorf is acknowledged for having supplieding the Fe10Cralloy.

5.7 References

1. M. Collepardi, Science and Technology of Concrete, Ed. Hoepli, 1991.2. B. Elsener, ‘Corrosion of Steel in Concrete’, in Corrosion and Environmental

Degradation, Vol. 2, 389–436, Materials Science and Technology Series, John Wiley,2000.

3. S. Haupt and H. H. Strehblow, Langmuir, 1987, 3, 873.



4. J. T. Hinatsu, W. F. Graydon and F. R. Foulkes, J. Appl. Electrochem. 1991, 21, 425–429.

5. B. Elsener, L. Zimmermann, D. Fluckiger, D. Burchler and H. Bohni, ‘Chloridepenetration – Non destructive determination of the free chloride content in mortarand concrete.’ Chloride Penetration into Concrete, ed. L. O. Nilsson and J. P. Olivier,RILEM, 1997, 17–26.

6. A. Rossi, B. Elsener, M. Textor and N. D. Spencer, ‘Combined XPS and ToF-SIMSanalyses in the study of inhibitor function–organic films on iron’, Analusis, 1997,25(5), M30.

7. S. Joiret, M. Keddam, H. Perrot, H. Takenouti, X. R. Novoa and M. C. Perez,‘Anodic Behaviour of Fe in 1 M NaOH in the presence of Cl– and NO 2

– , Proc. VIIISymposium Passivity of Metals and Semiconductors, Eds. M. B. Ives, J. L. Luo, andJ. R. Rodda, Electrochem. Proc, 99–42, The Electrochemical Society, PenningtonNY, 2001, 799–805.

8. P. Schiessl, (Ed), Corrosion of Steel in Concrete, RILEM, Chapman and Hall, London,1988.

9. L. Zimmermann, B. Elsener and H. Bohni, ‘Critical Factors for the Initiation ofRebar Corrosion’, Corrosion of Reinforcement in Concrete: Corrosion Mechanismsand Corrosion Protection, EFC Publication No. 31, IOM Communications, London,2000, 25–33.

10. S. Jaggi, B. Elsener and H. Bohni, ‘Cathodic oxygen reduction on passive steel inalkaline solutions’, Corrosion of Reinforcement in Concrete: Corrosion Mechanismsand Corrosion Protection, EFC Publication No. 31, IOM Communications, London,2000, 3–12.

11. M. P. Seah, Surf. Interface Anal., 1989, 14, 488.12. P. M. A. Sherwood, Practical Surface Analysis, Eds. Briggs and M. P. Seah, Appendix

3, p. 445, J. Wiley, N.Y., 1983.13. A. Rossi and B. Elsener, Surf. Interface Anal., 1992, 18, 499.14. C. R. Brundle, T. J. Chuang and K. Wandelt, Surface Science, 1977, 68, 459–468.15. P. Schmutz and D. Landolt, Corros. Sci., 1999, 41, 2143–2163.16. B. Elsener, L. Zimmermann, D. Burchler and H. Bohni, ‘Repair of reinforced concrete

structures by electrochemical techniques – field experience’. Corrosion of reinforcementin concrete – monitoring, prevention and rehabilitation, Eds. J. Mietz, B. Elsenerand R. Polder, EFC Publ., No. 25, IOM Communication, London, 1998 125–140.

17. B. Elsener, S. Muller, M. Suter and H. Bohni, ‘Corrosion monitoring of steel inconcrete – Theory and Practice’, Corrosion of Reinforcement in Concrete, Eds. C. L.Page, K. W. Treadaway and P. B. Bamforth, Elsevier Applied Science, London,1990, 348–357.

18. B. Elsener, ‘Half-cell potential mapping to assess repair work on RC structures’,Constr. Build. Mater. 2001, 15, 133–139.

19. M. Strattmann and K. Hoffmann, Corr. Sci., 1989, 29, 1329–1352.20. C. Andrade, L. Soler and X. R. Novoa, Mater. Sci. Forum, 1995, 192–194, 843–856.21. S. Joiret, M. Keddam, X. R. Novoa, M. C. Perez, C. Rangel and H. Takenouti, ‘Use

of EIS, ring disk electrode, EQCM and Raman spectroscopy to study the film ofoxides formed on iron in 1M NaOH’, Cement Concrete Compos. 2002, 24, 7–15.

22. U. Nurnberger (Ed.), ‘Stainless Steel in Concrete – State of the Art Report’, EFCPublication No. 18, The Institute of Materials, London, 1996.


62

6.1 Introduction

Galvanic phenomena often occur in reinforced concrete structures whencorroding steel bars in chloride-contaminated concrete are connected withpassive steel bars embedded in concrete of lower chloride content. The passivereinforcement is the cathode of this macrocouple, while the active reinforcementis anodically polarised and experiences an increase in the corrosion rate.1,2

Galvanic coupling can also be induced when the corroding structure is repairedby replacing damaged concrete with alkaline and chloride free mortar. If theconcrete in the area surrounding the patch repair contains a significant amountof chloride, a macrocouple takes place between repassivated bars in contactwith the repair mortar and depassivated bars in the original chloridecontaminated concrete3. Consequently, early corrosion damage can occur inthe area surrounding the patch repair. For this reason, in the rehabilitation ofstructures damaged by chloride-induced corrosion, it is recommended thateven the mechanically sound concrete should be removed, if it containssignificant amounts of chlorides.3,4

The risk of galvanic corrosion can also arise when new materials are usedfor repair. For instance, in the past great concern was expressed with regardto the use of stainless-steel bars. It has now been clearly shown that stainless-steel bars do not increase the risk of galvanic coupling in reinforced concretestructures. In fact, because of the high overvoltage of the cathodic reactionof oxygen evolution, stainless steel is a poor cathode compared with normalcarbon steel.5,6

Even non-metallic materials can provide sites for the cathodic reaction totake place and, thus, can generate macrocouples with corroding carbon steel.This is the case for composite materials with carbon fibres (carbon fibrereinforced plastics, CFRP). It has been shown that reinforcing bars made ofCFRP can induce a macrocouple on steel reinforcement in chloride-contaminated concrete.7 For several reasons, mainly related to concerns abouttheir long-term performance and durability, CFRP reinforcing or prestressing

6Risk of galvanic corrosion induced by CFRP

strengthening in reinforced concrete

L. B E R T O L I N I, M. G A S TA L D I andM. P. P E D E F E R R I, Politecnico di Milano, Italy


Risk of galvanic corrosion induced by CFRP strengthening 63

bars are rarely used; on the contrary, CFRP is often used for structuralstrengthening (e.g. for seismic retrofitting) and rehabilitation of deterioratedstructures.8,9 High-strength carbon fibres are bonded to the concrete surfacein the form of laminates or sheets; epoxy adhesives are normally used forbonding. The low weight of the composite materials and high design flexibilitymake this technology quite attractive. Nevertheless, the durability of thiscomplex system, consisting of the combination of the composite, the adhesive,and the reinforced concrete, has to be studied. Furthermore, when the compositematerials are applied to corrosion damaged structures, it should be clear thatthey are only aimed at strengthening the structure and a proper repair shouldbe carried out before application of the composite to stop ongoing corrosionof the reinforcing steel. Indeed, if carbon fibre composites are used, even aharmful effect of the composite can be hypothesised, owing to possiblegalvanic effects.

In this work the consequences of using commercial CFRP are investigatedwith regard to corrosion of reinforcement in structures contaminated bychlorides. The effects of galvanic coupling produced by these materials werestudied and compared with those normally produced by the contact betweenactive reinforcement (that are corroding) and passive bars (that are notcorroding) made both of common carbon steel and stainless steel.

6.2 Experimental procedure

Tests were carried out using commercial laminates (Sika Carbodur) andsheets (Sikawrap HEX 230C) of unidirectional carbon fibre. To investigatethe potential risks of galvanic coupling induced by the composite, a series oftests were carried out on the specimen shown in Fig. 6.1a. In one part of thespecimen, a carbon steel bar (10 mm in diameter) was embedded in concretecontaining 3% of chlorides by mass of cement; a laminate of CFRP (15 mmwide and 1.4 mm thick) was embedded in chloride-free concrete in theadjacent part. Concrete was mixed with a water-to-cement ratio of 0.55, 350kg m–3 of portland cement, and 1900 kg m–3 of crushed limestone aggregate;chlorides were added as CaCl2 to the mixing water. The carbon steel bar waselectrically connected to the laminate. Macrocell current was evaluated throughthe ohmic drop on a 100 W shunt; potentials of steel and CFRP laminatewere measured versus embedded reference electrodes made of mixed metaloxide (MMO) activated titanium. Tests were carried out in a climatic chamber;humidity (95% and 65% relative humidity (rh) and temperature (20, 40 and60 ∞C) were changed in steps of at least 15 days. Comparison tests were alsocarried out using 10 mm bars of graphite, carbon steel or 316L stainless steel(EN 1.4401) in place of the CFRP laminate.

A second series of tests were carried out on the specimens of Fig. 6.1b and6.1c to evaluate the risk of galvanic coupling when the composite is externally



ent in concrete64

Carbon steel

R

20 20

CFRP laminate or graphite,AISI 316L or carbon steel bars

(a) (b)

5

Concrete with 3% of chloride

Concrete chloride free

20

R

R

CFRP sheet

Carbon steel

5

10

CFRP laminate

R

CFRP sheet

10

(c)

6.1 Schematic representation of the specimens (dimensions in cm). A carbon steel bar was embedded in concrete with3% chlorides by cement weight and was coupled with: (a) laminate embedded in chloride free concrete (otherspecimens had a carbon steel, stainless steel, or graphite bars in place of the CFRP laminate), (b) laminate and sheetbonded to the two opposite faces, (c) CFRP sheet wrapping.



bonded. In the specimens of Fig. 6.1b a corroding bar of carbon steel, embeddedin concrete with 3% chloride by mass of cement, was simultaneously coupledwith a laminate and a sheet bonded to the surface of the concrete. Both thelaminate and the sheet had a width of 25 mm. Two specific epoxy systemswere used to apply the laminate and the sheet, according to the manufacturer’sinstructions. Macrocell tests were carried out in the same environment as forspecimens with embedded laminate. Subsequently, a 1.5-cm-thick layer ofmortar was applied to coat both the laminate and the sheet, and changes inthe macrocouple current density were monitored. Galvanic coupling testswere also carried out on the specimen shown in Fig. 6.1c, which was wrappedwith the CFRP sheet for a length of 10 cm.

The cathodic behaviour of the materials in alkaline environments wasstudied with polarization tests in Ca(OH)2 saturated solution (pH 12.6).Specimens were immersed in the solution for 48 h before testing.Potentiodynamic tests were carried out with a cathodic scan rate of20 mV min–1 from the free corrosion potential to –1.2 V vs. SCE.


6.3.1 CFRP laminate embedded in concrete

Figure 6.2 shows the results of galvanic coupling between a corroding steelbar in concrete contaminated with 3% of chloride by mass of cement anddifferent cathodic materials. Before coupling, at 20 ∞C and 95% relativehumidity, the corroding bars of carbon steel had free corrosion potentials ofaround –500 mV vs. MMO and corrosion rates higher than 20 mA m–2

(evaluated with polarisation resistance measurements). Passive carbon steelin chloride-free concrete had a free corrosion potential of about –50 mV vs.MMO; potential measured on the stainless-steel bar, the CFRP laminate andthe graphite bar was about –100 mV vs. MMO.

Following the electrical connection, a macrocouple current flowed fromthe corroding carbon steel bar in 3% Cl– concrete (anode) to the passive baror the CFRP laminate (cathode). Changes in potential and macrocouple currentdensity occurred soon after the coupling, while rather stable values weremeasured during the following 15 days at rh of 95–98%. The current densitydecreased from initial values higher than 100 mA m–2 to lower and stablevalues within a few hours. Changes in potential produced by the macrocouplewere mainly confined to the cathode, which showed a decrease from about–100 mV vs. MMO to values below –300 to –400 mV vs. MMO. Conversely,the potential of the corroding carbon steel only showed a small increase.

Figure 6.2a shows that the macrocouple current induced by couplingbetween active carbon steel and passive carbon steel reinforcement was about10 mA m–2 during the period of exposure to 20 ∞C and 95% rh. With a 316Lstainless-steel cathode (Fig. 6.2b) the macrocouple current was significantly



lower (2 mA m–2). Conversely the CFRP laminate (Fig. 6.2c) led to a highercurrent density of 35–40 mA m–2. Therefore, the coupling with the compositematerial seems to be more dangerous than coupling with passive carbonsteel.

Comparing results obtained with the laminate with those of a graphite barof equal surface, a basic similarity can be observed (Fig. 6.2c and d). Thecurrents always maintained similar values throughout the entire time ofexposure. This can indicate that the epoxy matrix of the CFRP laminate doesnot insulate the fibres from the concrete and allows the cathodic process todevelop in the same way that an element of graphite of similar surface does.

Cu

rren

t d

ensi

ty (

mA

m–2

)

1000

100

10

1

0.195% R.H. 65% R.H.

Po

ten

tial

(m

V v

s M

MO

)

0

–100

–200

–300

–400

–500

–600

–7000 10 20 30 40

Time (d)(a)

Cu

rren

t d

ensi

ty (

mA

m–2

)

1000

100

10

1

0.195% R.H. 65% R.H.

Po

ten

tial

(m

V v

s M

MO

)

0

–100

–200

–300

–400

–500

–600

–7000 10 20 30 40

Time (d)(b)

6.2 Macrocouple tests between corroding bars of carbon steel inconcrete with 3% of chlorides and: (a) passive carbon steel inchloride free concrete, (b) 316L stainless steel, (c) CFRP laminate,(d) graphite. Potential of corroding carbon steel (D), potential ofcathode material (�), and macrocouple current (—).

Coupling



Differences in the macrocouple current density observed in Fig. 6.2 onspecimens of the same geometry exposed to the same environment can onlybe explained by a different cathodic behaviour of the materials coupled withthe corroding carbon steel. Figure 6.3 shows results of potentiodynamicpolarization tests in Ca(OH)2 saturated solution.

It can be observed that the cathodic polarization curve of stainless steel isshifted to more negative potentials with respect to the curve of carbon steel,while cathodic polarization curves of CFRP laminate and graphite are shiftedto more positive potentials. This confirms that stainless steel has higherovervoltage for the cathodic reaction of oxygen reduction than carbon steeland, thus, when it is cathodically polarized, for a given potential it cansupply a lower cathodic current density. However CFRP laminate and graphitehave lower overvoltages, and the cathodic current density for a given potentialis higher. Indeed, CFRP laminate is a quite effective cathodic material and

Cu

rren

t d

ensi

ty (

mA

m–2

)

1000

100

10

1

0.195% R.H. 65% R.H.

Po

ten

tial

(m

V v

s M

MO

)

0

–100

–200

–300

–400

–500

–600

–7000 10 20 40 40

Time (d)(c)

Cu

rren

t d

ensi

ty (

mA

m–2

)

1000

100

10

1

0.195% R.H. 65% R.H.

Po

ten

tial

(m

V v

s M

MO

)

0

–100

–200

–300

–400

–500

–600

–7000 10 20 40 40

Time (d)(d)

6.2 Continued



can produce large macrocell currents when it is cathodically polarized bycoupling with corroding steel.

Obviously, the macrocell current is influenced by the resistivity of concreteand, thus, by its humidity content. Figure 6.2 shows that when the relativehumidity was reduced from 95 to 65%, the macrocouple current densityprogressively decreased and potentials increased in time. However, thedifference in the macrocouple behaviour for the materials coupled with activecarbon steel was confirmed.

6.3.2 Bonded CFRP laminate and sheet

The case considered thus far, i.e. when composites are embedded in concrete,in reality can only occur if CFRP are used as reinforcing bars7. When thesecomposite materials are utilised to strengthen existing structures damagedby corrosion, they are usually bonded to the surface of the concrete structure.However, even in this case they could come in contact with the reinforcementand galvanic coupling may occur.

Figure 6.4 shows the macrocell current densities measured by coupling anactive bar of carbon steel in concrete with 3% chlorides with the CFRPlaminate and sheet externally bonded to the concrete surface. For comparison,the results on the CFRP laminate embedded in concrete are also reported.Tests were carried out at temperatures of 20, 40 and 60 ∞C and rh of 95%.

When the laminate was embedded in concrete (Fig. 6.4a) the macrocouplecurrent density at 20 ∞C had values of about 40 mA m–2; a slight increase was

Graphite

CFRPlaminate

Carbon steel

316L stainless steel

0.01 0.1 1 10 100 1000Current density (mA m–2)

Po

ten

tial

(m

V v

s S

CE

)

0

–100

–200

–300

–400

–500

–600

–700

–800

6.3 Cathodic polarization curves in Ca(OH)2 saturated solution for thematerials studied in this work.



observed on increasing the temperature to 40 ∞C. At the end of exposure at40 ∞C, the concrete cover cracked due to corrosion of the carbon steelreinforcement. This crack caused the current density to decrease, even duringthe test at 60 ∞C when values of 10-14 mA m–2 were measured.

The CFRP laminate bonded to the concrete surface (Fig. 6.4b) led to amacrocell current much lower than that measured with the embedded laminate.At 20 ∞C, it was lower than 1 mA m–2; in one of two replicate specimens itwas even of the order of 0.1 mA m–2. The externally bonded sheet alsogenerated a macrocouple current of the order of 0.1 mA m–2 (Fig. 6.4c and d).

Cu

rren

t d

ensi

ty (

mA

m–2

)

100

10

1

0.1

0.01

0.001

Embedded laminate

20 ∞C40 ∞C60 ∞C

0 5 10 15Time (d)

(a)

Cu

rren

t d

ensi

ty (

mA

m–2

)

100

10

1

0.1

0.01

0.001

Externally bonded laminate

0 5 10 15Time (d)

(b)

6.4 Macrocouple current density measured in time in the specimens,exposed at 95% rh and at different temperatures, in which a bar ofcarbon steel in concrete with 3% of chlorides is coupled with:(a) CFRP laminate embedded in chloride free concrete, (b) CFRPlaminate externally bonded, (c) CFRP sheet externally bonded,(d) CFRP sheet wrapped around the concrete specimen.



6.4 Continued

Cu

rren

t d

ensi

ty (

mA

m–2

)

100

10

1

0.1

0.01

0.001

Externally bonded sheet

0 5 10 15Time (d)

(c)

Cu

rren

t d

ensi

ty (

mA

m–2

)

100

10

1

0.1

0.01

0.001

Externally bonded sheet(wrapped specimen)

0 5 10 15Time (d)

(d)

Increasing in temperature to 40 and 60 ∞C led to ambiguous results in both thelaminate and the sheet bonded parallel to the steel bar; in fact, often a decreasein the current was observed on increasing the temperature (Fig. 6.4b and c).Such a decrease was actually due to the cracking of the specimen induced bycorrosion of the steel bar in concrete with 3% of chloride by mass of cementduring exposure at high temperature. The effect of temperature was clearer inthe specimen wrapped with the CFRP sheet, since the geometry of thisspecimen allowed the macrocouple current to flow even after cracking of theconcrete cover. Furthermore, the CFRP wrapping could also limit the crackwidth.

Figure 6.5 plots the average values of the macrocouple current densityduring different tests as a function of temperature. The laminate embedded



in concrete (Fig. 6.5a) led to a current density higher than 10 mA m–2 evenafter cracking of the concrete specimens. The current density generated bythe CFRP laminate bonded to the surface of the concrete was two orders ofmagnitude lower, as for the bonded sheet (Fig. 6.5b). In spite of severaldrops in the macrocouple current density due to cracking of the specimen, itcan be observed that, in general, the current increased as the temperatureincreased from 20 to 60 ∞C. Nevertheless, the current was always negligiblefor the composite bonded to the surface of the concrete; only at 60 ∞C on thewrapped specimen did the current density reach values of 1 mA m–2.

Cu

rren

t d

ensi

ty (

mA

m–2

)

100

10

1

0.1

0.01

0.0010 20 40 60 80

Temperature (∞C)(a)

Cu

rren

t d

ensi

ty (

mA

m–2

)

100

10

1

0.1

0.01

0.0010 20 40 60 80

Temperature (∞C)(b)

6.5 Effect of temperature on the macrocouple current density (steadyvalues) generated at 95% rh by: (a) CFRP laminate embedded inconcrete (∑) and externally bonded (�); (b) CFRP sheet bonded on aside ( ) or wrapped around ( ) the concrete specimens. Greysymbols show results obtained after cracking of the concrete cover.



These results show that when CFRP laminates or sheets are bonded to aconcrete surface, even when they accidentally come into contact with corrodingsteel, they should produce negligible macrocell currents. The epoxy used tobond the composite to the concrete surface greatly reduces electrolyticcontinuity with the concrete, even though it does not insulate them completely.

However the electrolytic continuity can be restored if a coating of cementmortar (for example a plaster) is applied to the CFRP strengthening system.In fact, in this case the CFRP may come into contact with the layer of mortarand, thus, with the underlying concrete. Figure 6.6 shows the effect producedby the application of a layer of 1 cm of mortar on specimens of the typeshown in Fig. 6.1b. The mortar layer was first applied on the side of thespecimens where the CFRP laminate was bonded (Fig. 6.6a) and, after 5days, it was applied on the opposite side where the CFRP sheet was bonded(Fig. 6.6b).

The application of the mortar layer led to a remarkable increase inthe macrocell current, especially in the laminate, where stable values of2–4 mA m–2 were reached (Fig. 6.6a). Nevertheless, values never reachedthose observed for the laminate embedded in concrete (35–40 mA m–2, Fig.6.2c).

In the case of the CFRP sheet the increase was lower and the currentdensity approached values of 0.2–0.3 mA m–2; in this case, the fibres wereembedded in epoxy resin and were thus insulated, even from contact with alayer of mortar applied afterwards.

6.4 Conclusions

Composite materials with carbon fibres are potentially able to generate galvaniccoupling with corroding steel in concrete. CFRP, like graphite, have verylow overvoltages for the cathodic reaction of oxygen reduction. A CFRPlaminate embedded in concrete and coupled with a steel bar in chloride-contaminated concrete could induce a macrocouple current several timeshigher than with passive carbon steel and more than one order of magnitudehigher than with stainless steel.

CFRP bonded to the concrete surface, in the form of laminate or sheets,led to a negligible macrocouple current (lower than 1 mA m–2), even attemperatures higher than 20 ∞C. Therefore, when an existing structure isrepaired by applying laminates or sheets on the surface of the concrete withan epoxy adhesive, the effects of galvanic coupling are negligible even if theCFRP is accidentally in contact with the corroding reinforcement. However,when a cement plaster is applied, electrolytic continuity with the concretemay be partially restored and the macrocouple current can be significant.

Externally bonded CFRP strengthening should be applied, in any case,only after a proper repair has been carried out to restore passivity to the



corroding reinforcement. Even in the absence of galvanic coupling, propagationof corrosion of reinforcement in chloride-contaminated concrete can lead tocracking of the concrete cover and to serious consequences on the effectivenessof the strengthening.


The authors are grateful to Sika Italia SpA for supplying the compositematerials.

6.6 Changes in the macrocouple current density in two replicatespecimens, exposed at 95% rh and 20 ∞C, on which a mortar layerwas applied on the externally bonded CFRP laminate (a) and sheet (b).

Cu

rren

t d

ensi

ty (

mA

m–2

)

10

1

0.1

0.010 5 10 15 20

Temperature (∞C)(a)

Application ofmortar layer

Application ofmortar layer

Cu

rren

t d

ensi

ty (

mA

m–2

)

10

1

0.1

0.010 5 10 15 20

Temperature (∞C)(b)



6.6 References

1. C Andrade, I Rz-Maribona, S Feliu and J A Gonzales, ‘Macrocell versus microcellcorrosion of reinforcements placed in parallel’, Corrosion ’92, NACE, Houston, 1992,paper No. 92194.

2. P Rodriguez, E Ramirez and J A Gonzalez, ‘Effect of galvanic macrocouples on thecorrosion of steel embedded in concrete’, Corrosion ’94, NACE, Houston, 1994,paper No. 94037.

3. RILEM, Technical Recommendation 124 SRC, ‘Guidelines to repair strategies forconcrete structures damaged by reinforcement corrosion’, 1993.

4. ENV 1504-9, ‘Products and systems for the protection and repair of concrete structures– definitions, requirements, quality control and evaluation of conformity – Part 9:General principles for the use of products and systems’, 1997.

5. L Bertolini, M Gastaldi, T Pastore, M P Pedeferri and P Pedeferri, ‘Experiences onstainless steel behaviour in reinforced concrete’, Int. Conf. Eurocorr ’98, EuropeanFederation of Corrosion, Event No. 221, Utrecht, 28 September–1 October 1998.

6. L Bertolini, M Gastaldi, T Pastore, M P Pedeferri and P Pedeferri, ‘Effects of galvaniccoupling between carbon steel and stainless steel reinforcement in concrete’, Int.Conf. Corrosion and Rehabilitation of Reinforced Concrete Structures, Federal HighwayAdministration, Orlando, 7–10 December 1998.

7. A Torres-Acosta, A Sagués, R Sen, ‘Galvanic interaction between carbon fiber reinforcedplastic (CFRP) composites and steel in chloride contaminated concrete’, Corrosion’98, NACE, Houston, 1998, paper No. 98648.

8. A Nanni, ‘CFRP strengthening’, Concrete Int., 1997, 6, 19.9. B E Dolan, H R Hamilton and C H Dolan, ‘Strengthening with bonded FRP laminate’,

Concrete Int., 1998, 6, 51.


75

7.1 Introduction

Reinforcing steel in good quality concrete does not corrode even if sufficientmoisture and oxygen are available. This is due to the spontaneous formationof a thin protective oxide film (passive film) on the steel surface in the highlyalkaline pore solution of the concrete. When sufficient chloride ions (fromdeicing salts or from sea water) have penetrated to the reinforcement orwhen the pH of the pore solution drops to low values due to carbonation, theprotective film is destroyed and the reinforcing steel is depassivated. Corrosionin the form of rust formation and/or loss in cross-section of the rebars thenoccurs in the presence of oxygen and water (humidity) [1–3]. The corrosionof steel in concrete is essentially an electrochemical process, where, at theanode, iron is oxidised to iron ions that pass into solution and, at the cathode,oxygen is reduced to hydroxyl ions. Anode and cathode form a short-circuitedcorrosion cell, with the flow of electrons in the steel and of ions in the poresolution of the concrete [1–3].

According to the different spatial location of anode and cathode, corrosionof steel in concrete can occur in different forms:

∑ as microcells, where anodic and cathodic reactions are immediately adjacent,leading to uniform iron dissolution over the whole surface. Uniformcorrosion is generally caused by carbonation of the concrete or by veryhigh chloride content at the rebars.

∑ as macrocells, where a net distinction between corroding areas of therebar (anode) and non-corroding, passive surfaces (cathode) is found.Macrocells occur mainly in the case of chloride induced corrosion (pitting).Generally the anode is small with respect to the total (passive) rebar surface.

On reinforced structures and in the experimental study of macrocells,coplanar or face to face situations of anode and cathode can be distinguished[4–6]. A typical coplanar situation is a localised corrosion attack in an otherwisepassive rebar (Fig. 7.1), a typical face to face situation is the corroding upper

7Macrocell corrosion of steel in concrete –

experiments and numerical modelling

S. J Ä G G I and H. B Ö H N I, ETH Zürich,Switzerland and B. E L S E N E R,

ETH Zürich and University of Cagliari, Italy



layer of the reinforcement in a bridge deck with the lower mat being passive.Macrocell corrosion is of great concern because the local dissolution rate(reduction in cross-section of the rebar) may be greatly accelerated due tothe large cathode/anode area ratio [4–6]. Indeed, values of local corrosionrates up to 1 mm per year have been reported for bridge decks, sustainingwalls or other chloride contaminated reinforced concrete structures [7–9].This rapid corrosion attack may lead – if not detected early – to structuralsafety problems.

Macrocell corrosion can be considered like a battery, the total currentflowing, IME, is given by the driving voltage (potential difference betweenuncoupled anode and cathode), DU, divided by the sum of the resistance ofthe electrolyte, REl, the resistance of the anodic, RA(i), and the cathodicreaction, RC(i):

IME = DU / [REl + RA(i) + RC(i)] (7.1)

In the literature, values betweeen 0.25 and 0.5 V are reported for the drivingvoltage DU [10, 11]. The resistance of the electrolyte, REl, contains thegeometry factor anode/cathode (e.g. increases for small anodes) and themortar or concrete resistivity. The influence of porosity (w/c ratio, hydration...),relative humidity and temperature on the resistivity of cement-based materialsis well known [12, 13]. The temperature dependence of the electrolyte resistancecan be written as

REl = REl,0 exp (b [1/T – 1/T0]) (7.2)

For reference temperature T0 = 20 ∞C values of the constant b in the rangeof 1700 K (synthetic pore solution, pH 13.5) and 3800 K (concrete exposed

7.1 Localised corrosion attack on a 20 mm rebar, loss in cross-sectionca. 30%.


Macrocell corrosion of steel in concrete 77

for a long time at 60% rh) have been reported [12, 13]. The cathodic oxygenreduction and especially its temperature dependence have not yet been studiedextensively and only few data on Tafels slope are available [14]. Correspondingdata for the anode reaction in mortar or concrete are missing. For the temperaturedependence of the two electrochemical reactions and of the total macrocellcurrent an equation similar to eq. (7.2) can be written with the constant in theArrhenius equation called a:

I = I0/exp (a [1/T – 1/T0]) (7.3)

The aim of this work was to investigate the temperature dependence ofthe cathodic and the anodic reactions in the macrocell and to provide thenecessary input data for the numerical modelling. The intensity and thetemperature dependence of the macrocell current, IME, and the currentdistribution on the cathode are evaluated by numerical modelling.


7.2.1 Cathodic and anodic reactions

The reaction kinetics were studied with potentiodynamic polarisation curvesin a conventional three electrode electrochemical cell in a thermostatic bathin order to vary the temperature. The reference electrode was a saturatedcalomel electrode. The cathodic reaction was studied on polished mild steelin synthetic pore solution with pH 13.5 [14] open to air or deaerated withargon gas. The anodic reaction was studied in 0.1M HCl.

Measurements of the cathodic polarisation curves in mortar were performedwith specially designed cylindrical mortar samples with a diameter of 4 cm.In the centre, a degreased rebar sample (Ø 1 cm) was mounted; the counterelectrode was a stainless-steel grid (Ø 2.5 cm) and, as the referenceelectrode, a small piece of activated titanium was used. The mortar samples(400 kg m–3 OPC, water cement 0.6, cement/sand 0.25) were cured for 28days at 80% relative humidity before starting the measurements.

All the potentiodynamic measurements started at the open circuit potential,the sweep rate was 1 mV s–1. The potentials reported are referred to that ofa saturated calomel electrode and corrected for the ohmic potential drop.

7.2.2 Macrocell investigations

The influence of temperature on macrocell corrosion has been studied forsteel in mortar with the experimental setup reported previously [15]. A linearmacrocell arrangement was prepared as a mortar block of 30 ¥ 30 mm witha length of 34 cm. It contained a segmented cathode (10 electrically isolatedsegments of 25 mm length) and, in the centre, an anode of 10 mm length(precorroded, embedded in a chloride-containing mortar). The cover depth



in this case was 1 cm. A more realistic two- and three-dimensional macrocellarrangement was prepared in mortar blocks of 55 ¥ 55 cm with a height of15 cm, the cover depth was 7 cm on both sides.

The total macrocell current and the currents to the individual cathodesegments were measured with a zero resistance ammeter, and the switchingwas performed on a programmable multimeter (Keithley). A specialswitchboard guaranteed a complete short circuit during the measurements.Data were recorded on a personal computer. The corrosion potentials of theanode and cathodes were measured with a saturated calomel electrode (SCE).

7.2.3 Numerical modelling

The numerical modelling was performed with a commercial boundary elementprogram BEASY Corrosion and Cathodic Protection Design (ComputationalMechanics, Ashurst, England). As input parameters the anodic and cathodicpolarisation curves determined in the experiments were used.

7.3 Experimental and modelling results

7.3.1 Cathodic oxygen reduction

The polarisation curves of the oxygen reduction reaction and its temperaturedependence are shown in Fig. 7.2 [14, 15]. The curves showed, as expected,a Tafel behaviour for low overvoltages followed by a potential range with adiffusion-limited current density. The Tafel slope increased with increasingprepassivation time to values higher than 240 mV per decade, leading to a

20 min24 h1 mo4 mo

Flow: 1 mL s–1

open to air

–1400 –1200 –1000 –800 –600 –400 –200 0Potential (mVSCE)

Cu

rren

t d

ensi

ty (mA

cm

–2)

1000.00

100.00

10.00

1.00

0.10

7.2 Influence of the ageing time of the passive film in synthetic poresolution on cathodic polarisation curves (flow rate 1 mL s–1, open toair) [14].



decrease in the reduction current at constant potential. It is interesting to notethat the diffusion-limited current density was practically independent oftemperature in the range of 5–50 ∞C.

The cathodic oxygen reduction in mortar showed lower current densitiesthan in solution [14, 15]. The current densities increased with increasingtemperature. The diffusion-limited current density was found only at potentials< 0.8 V (SCE) or at very low oxygen content. The temperature dependence– obtained by normalising to the cathodic current density at 20 ∞C – wasidentical in solution and in mortar (Fig. 7.3).

7.3.2 Anodic iron dissolution

The polarisation curves of the anodic iron dissolution in 0.1 M HCl areshown in Fig. 7.4. The Tafel slope is about 75 mV per decade at 20 ∞C andbecomes lower with increasing temperature. The temperature dependence,normalized for a current density at 20 ∞C, is shown in Fig. 7.3.

7.3.3 Macrocell corrosion current

The macrocell corrosion current measured between anode (rebar of 10 cm2

area embedded in chloride-contaminated mortar) and cathode (rebar in syntheticpore solution, aerated) is shown in Fig. 7.5 together with the temperaturevariation over time. The macrocell current for the five (in principle identical)individual macrocells differs in intensity (probably due to a different size ofthe effective anode area) but the temperature dependence is similar; it is

Cathodic, solutionCathodic, mortarAnodic, solution

0 10 20 30 40 50Temperature (∞C)

% C

ath

od

ic c

urr

ent

no

rmal

ized

to

20 ∞C 500

400

300

200

100

0

7.3 Normalised temperature dependence of the anodic irondissolution and the cathodic oxygen reduction in solution and inmortar.



Temperature increase

3 deg20 deg40 deg47 deg

–550 –500 –450 –400 –350Potential (mVSCE)

Cu

rren

t d

ensi

ty [mA

cm

–2]

1000

100

10

1

0.1

7.4 Anodic polarisation curves of rebar in 0.1M HCI at differenttemperatures. Scan rate 1 mV s–1.

120

100

80

60

40

20

0

Cu

rren

t (m

A)

50 100 150 200 250Time (h)

(a)

50 100 150 200 250Time (h)

(b)

50

40

30

20

10

0

Tem

per

atu

re (∞C

)

7.5 (a) Macrocell current of five macrocells in synthetic pore solutionand (b) temperature program versus time.



evident that the macrocell current increases with increasing temperature.The highest values measured were 120 mA at 50 ∞C, demonstrating very highcorrosion rates. Further, the macrocell current remains fairly constant overthe whole measuring period (e.g. compare the currents at 10 ∞C at the beginningand at 250 h).

As for the anodic and cathodic reactions, the macrocell currents werenormalised to the value at 20 ∞C; the resulting temperature dependence forthe five macrocells is shown in Fig. 7.6. The constant of the Arrheniusequation results in a = 4350 ± 80 K in the same range as for the anodic andcathodic reactions (Fig. 7.3).

7.3.4 Macrocell corrosion in mortar – one dimensionalmodel

The segmented one dimensional model macrocell bar allows the currentdistribution on the cathode to be determined as a function of the distancefrom the anode. The cathode current slightly decreases with time. The maineffect is the decrease of the cathode currents with increasing distance fromthe anode, the cathode segments at the end of the bars showing higher currents(Fig. 7.7); this is due to the fact that the mortar block is 2 cm longer then themacrocell bar.

Modelling the same geometrical arrangement of the linear segmentedmacrocell with the boundary element program BEASY, using the mortarresistivity, the cathodic polarisation curve for the cathode segments and theanodic polarisation curve for the anode as input data, the macrocell current

–10 0 10 20 30 40 50Temperature (∞C)

I ME n

orm

aliz

ed a

t 20

∞C

(%

)

400

350

300

250

200

150

100

50

0

7.6 Normalised temperature dependence of the macrocell currentmeasured in solution.



and its distribution were calculated. A very good agreement betweenexperimental and calculated values of the cathodic current was found (Fig. 7.7).

7.3.5 Macrocell corrosion in mortar – two dimensionalmodel

The macrocell current in the two-dimensional model macrocell was measuredas a function of temperature in several experiments. The macrocell currentincreases with temperature. It is interesting to note that even at temperaturesas low as –10 or –20 ∞C a small macrocell current of some pA was stillflowing (Fig. 7.8). The temperature dependence of the macrocell current inthe mortar block is identical with that found in solution (Fig. 7.6), the coefficientof the Arrhenius equation describing the temperature dependence is 4209 ±88 K.

The two dimensional macrocell was modelled numerically with the boundaryelement program BEASY, using as input data the mortar resistivity, thepolarisation curve of the cathodic reaction for the cathode segments and theanodic polarisation curve for the anode segment. The resulting currentdistribution is shown in Fig. 7.9. A gradual decrease from the centre (currentdensity at the cathode >2.9 mA cm–2) to the end of the 55 cm-long cathodebars (current density <2.3 mA cm–2) can be seen. The current distribution issymmetrical for all four directions; it can be noted that the current density atthe end of the bars is far from being negligible. The total macrocell currentcalculated with the numerical model as a function of temperature agrees wellwith the experimental results.

Model calculationExperimental

K5 K4 K3 K2 K1 Anode K1 K2 K3 K4 K5

Cu

rren

t (m

A)

3

2.5

2

1.5

1

0.5

0

7.7 Distribution of the cathodic currents as a function of the distancefrom the anode, results from experiments on the one-dimensionalmacrocell in mortar and from the numerical model.



7.4 Discussion

Macrocell corrosion between actively corroding areas of rebars and largepassive areas (either beside the active spot or behind in a second layer ofreinforcement) is of great concern because it results in very high local anodiccurrent densities with corrosion rates up to 0.5 to 1 mm per year. The resultinglocal loss in cross-section has dangerous implications for structural safety if

TemperatureCurrent

0 50 100 150 200Time (h)

Cu

rren

t (m

A),

tem

per

atu

re (∞C

)

50

40

30

20

10

0

–10

–20

7.8 Macrocell current and temperature versus time, two-dimensionalmacrocell in mortar. Anode area 3.1 cm2, cathode area 301 cm2,cover 7 cm.

7.9 Numerical calculations with the boundary element programBEASY CP of the current distribution in a two-dimensional macrocellin a mortar block (cover 35 mm). Temperature 5 ∞C, total macrocellcurrent 45.9 mA, anode area 3.14 cm2.



the corroded rebars are located in a zone of high tensile or shear stresses.Furthermore, these dangerous attacks very often do not manifest themselvesat the concrete surface by cracking or spalling because soluble iron chloridecomplexes are formed [16]. The implications of this very inhomogeneouscorrosion situation on different monitoring techniques that can be used todetect these locally corroding areas, and to quantify the extent of local attack– half cell potential mapping and polarisation resistance measurements –have been presented recently [17].

In this work, two new aspects arise: first, macrocell corrosion of steel inconcrete has been studied for the first time over a wide range of temperatures,and second, numerical modelling with the boundary element program hasbeen based not only on geometry (area ratio cathode/anode, cover depth) andconcrete resistivity but also on the actual polarisation curves of the cathodicand anodic areas. This allows more significant results to be obtained.

7.4.1 Temperature dependence

The temperature dependence of both the macrocell current and the anodicand cathodic partial reactions of the corrosion reaction of steel in concretehas been studied in the range 0 to 50 ∞C. The coefficients, a, of the temperaturedependence are summarised in Table 7.1. As can be noted, the temperaturedependence of the anodic iron dissolution and the cathodic oxygen reductionreaction in alkaline media as well as the total macrocell current in alkalinesolution and mortar agree very well, a = 4280 ± 150 K. A comparison withresults from the literature shows good agreement (Fig. 7.10), although it hasto be noted that the temperature interval may influence the calculated valueof the temperature coefficient a. Numerical modelling of the macrocell withthe polarisation curve of the anodic and cathodic partial reactions and themortar resistivity allowed the influence of temperature on the macrocellcurrent to be calculated and a very good agreement with the experimentalresults was observed (Fig. 7.10). It is very interesting to note that the coefficientsof temperature dependence calculated from macrocell currents measuredbetween an anode in instrumented cores and the rebar network of bridgedecks [22] is in good agreement with these laboratory results (Table 7.1).The higher standard deviation results from different exposure conditions.

Table 7.1 Temperature dependence coefficient a (eq. 7.2) for anodic, cathodic andtotal macrocell current in solutions and mortar experiments

Current/Media Solution Mortar 1-dim Mortar 2-dim Field 3-dim

Macrocell 4350 ± 80 4210 ± 90 4000 ± 250 [22]Anodic 4300Cathodic 4310 4250



A comparison of the temperature dependence of the anodic and cathodicreactions and the macrocell current of steel in concrete with the temperaturedependence of the mortar (at 85% RH) is shown in Fig. 7.11. As can benoted, the temperature dependence is much more pronounced for theelectrochemical reactions than for the mortar resistivity. This indicates that

ExperimentalArya (11)Schiessl/Raupach (70)Raupach (106)Weyers/Liu (155)Calculated

–20 –10 0 10 20 30 40 50Temperature (∞C)

I ME n

orm

aliz

ed a

t 20

∞C

(%

)

500

400

300

200

100

0

7.10 Temperature dependence of the macrocell current of steel inmortar and concrete. Comparison of literature values of Arya [18],Schiessl [19], Raupach [20] and Weyers and Liu [21] with the currentexperimental (∑) and numerical (�) results.

7.11 Temperature dependence of the anodic, cathodic and macrocellcorrosion currents compared with the temperature dependence ofthe concrete resistivity (b = 3080 K, rh 85%).

CathodicAnodicResistance [Ohm]Macrocell current

I ME n

orm

aliz

ed a

t 20

∞C

(%

)

500

400

300

200

100

0–20 –10 0 10 20 30 40 50

Temperature (∞C)



the proportionality between polarisation resistance and concrete resistivityoften reported in the literature [23, 24] cannot be assumed to be valid apriori or – in other words – for chloride-induced corrosion, the corrosionrates cannot be calculated from concrete resistivity.

7.4.2 Rate controlling reaction step

According to eq. (7.1), the macrocell current is controlled by the electrolyteresistance, Rel, and the two polarisation resistances of the anodic (Ra) andcathodic (Rc) reactions. Which one of these three resistances determines theoverall macrocell current is still under discussion. It is often stated that theconcrete resistance is the controlling factor for the corrosion rate of steel inconcrete. From the results of this work it can be concluded that for largecathode areas and in mortar with a quite low resistivity of about 100 W m (asfrequently occurs in chloride-induced corrosion), it is the cathodic oxygenreduction reaction occuring on the passive reinforcement that is controllingthe overall reaction (> 90% of the total resistance in eq. 7.1) whereas the partof the electrolyte resistance is only ca. 5–10%. The anode is practicallyunpolarised. Numerical simulation with a plan-parallel arrangement of theelectrodes [10] has also shown that the cathodic reaction controls the overallmacrocell corrosion by more than 60%. The difference to this work arisesfrom the different geometrical arrangement: the small anode in a relativelylarge mortar block (55 ¥ 55 ¥ 15 cm) has a lower resistive control than theplan-parallel arrangement.

It can be concluded that the geometry used in laboratory studies withmortar beams or blocks can greatly influence the experimental results; similarlythe cathodic polarisation curve (Tafel slope, exchange current density) usedas input data greatly influence the results of numerical modelling.

7.4.3 Advantages of numerical modelling

Numerical modelling of macrocell corrosion using a geometrical arrangement(size of the anode, cover depth, size and position of cathodes) and thepolarisation curve of the anode and the cathode together with the concreteresistivity as input data has been shown to be a powerful tool in studyingmacrocell corrosion. The total macrocell current, its distribution on the cathodeand its temperature dependence agreed very well with the laboratoryexperiments with identical parameters. This allows the usual way of doingexperiments to be changed: based on a small set of input data, numericalmodelling is first performed for different geometries (e.g. long slabs, decksetc.) and for different concrete resistivities. In a second step, laboratoryexperiments are designed and the results of numerical modelling areverified.



7.5 Conclusions

In real structures, localised chloride-induced corrosion macrocells are formedthat greatly accelerate the local dissolution rate of the anode. From thecurrent work, performed on active/passive model macrocells in the laboratorycombined with numerical modelling, it can be concluded that:

1. The temperature dependence of anodic dissolution of iron, of the cathodicoxygen reduction reaction and of the overall macrocell corrosion is nearlyidentical (a = 4200 K) and much higher than the temperature dependenceof the mortar or concrete resistivity (rh 85%).

2. Numerical modelling with the boundary element program BEASY providespractically the same results for the total macrocell current, currentdistribution on the cathode and temperature dependence as found in theexperiments in mortar for a given geometrical arrangement, concreteresistivity and cathodic polarisation curves of the passive steel in concrete.

3. Chloride-induced macrocell corrosion concrete of low to moderateresistivity is governed by the cathodic oxygen reduction reaction and notby the resistivity of the concrete.

4. This numerical approach allows parameter studies to be performed veryrapidly and experiments with macrocells in concrete to be designed in arational way.

7.6 References

1. K. Tuuti, Corrosion of Steel in Concrete, CBI Forskning/Research, April 1982,Cement och Betonginstitutet, Stockholm.

2. P. Schiessl, Corrosion of Steel in Concrete, RILEM Technical Committee 60-CSC,Chapman and Hall, New York (1988).

3. B. Elsener, ‘Corrosion of Steel in Concrete’, in Corrosion and EnvironmentalDegradation, ed. M. Schütze, Vol. II p. 389–436, Wiley-VCH, Weinheim, 2000.

4. C. Andrade, I. R. Maribona, S. Feliu, A. Gonzalez and S. Feliu Jr, Corros. Sci. 1992,33, 237.

5. M. Raupach, ‘Chloride-induced macrocell corrosion of steel in concrete – theoreticalbackground and practical consequences’, Constr. Build. Mater., 1996, 10, 329.

6. B. Elsener, A. Hug, D. Bürchler and H. Böhni, ‘Evaluation of localised corrosionrate on steel in concrete by galvanostatic pulse technique’, Corrosion of Reinforcementin Concrete Construction, ed. C. L. Page, P. S. Bamforth and J. W. Figg, SCI,Cambridge, 1996, 264–272.

7. B. Elsener and H. Böhni, ‘Potential mapping and corrosion of steel in concrete’, in‘Corrosion Rates of Steel in Concrete’, ASTM STP 1065, eds. N. S. Berke, V.Chaker and D. Whiting, American Society for Testing and Materials, Philadelphia,1990, 143.

8. F. Hunkeler, Assessment of Corrosion on RC Structures with Potential Mapping (inGerman) Schweiz. Ingen. Architekt, 1991, 109, 272.

9. B. Elsener, Corrosion Rate on Reinforced Concrete Structures Determined byElectrochemical Methods, Mater. Sci. Forum, 1995, 192-194, 857.



10. M. Raupach and J. Gulikers, ‘Investigations on cathodic control of chloride inducedreinforcement corrosion’, in Corrosion of Reinforcement in Concrete – CorrosionMechanism and Protection, EFC Publication No. 31, ed. J. Mietz, R. Polder and B.Elsener, IOM Communications, London, 2000, 13–23.

11. B. Elsener, ‘Corrosion rate of steel in concrete – from laboratory to reinforcedconcrete structures’, in Corrosion of Reinforcement in Concrete, Monitoring, Preventionand Rehabilitation, EFC Publication No. 25, ed. J. Mietz, B. Elsener and R. Polder,IOM Communications, London, 1998, 92–103.

12. D. Bürchler, ‘Der elektrische Widerstand von zementosen Werkstoffen’, PhD ThesisNo. 11876, 1996, ETH Zürich (in German).

13. D. Bürchler, B. Elsener and H. Böhni, ‘Electrical resistivity and dielectric propertiesof hardened cement paste and mortar’, in Corrosion of Reinforcement in ConcreteConstruction, ed. C. L. Page, P. Bamforth and J. W. Figg, SCI, Cambridge, 1996,283–293.

14. S. Jäggi, B. Elsener and H. Böhni, ‘Oxygen reduction on mild steel and stainlesssteel in alkaline solutions’, in Corrosion of Reinforcement in Concrete – CorrosionMechanism and Protection, EFC Publication No. 31, ed. J. Mietz, R. Polder and B.Elsener, IOM Communications, London, 2000, 3–12.

15. S. Jäggi, ‘Experimentelle and numerische Modellierung der lokalen Korrosion vonStahl in Beton unter besonderer Berücksichtigung der Temperaturabhängigkeit’,PhD Thesis No. 14058, 2001, ETH Zürich.

16. J. P. Guilbaud, G. Chahbazian, F. Derrien and A. Raharinaivo, ‘Electrochemicalbehaviour of steel under cathodic protection in medium simulating concrete’, Corrosionand Corrosion Protection of Steel in Concrete, ed. R. N. Swamy, Sheffield AcademicPress, Vol. 2, 1382–1391.

17. B. Elsener, ‘Macrocell corrosion of steel in concrete – implications for corrosionmonitoring’, Cement Concrete Compos., 2002, 24, 65–72.

18. C. Arya, Cement Concrete Res. 1995, 25, 989.19. P. Schiessl and M. Raupach, ‘Influence of concrete composition and microclimate

on critical chloride content in concrete’, in Corrosion of Reinforcement in Concrete,C. L. Page, K. W. Treadaway, P. B. Bamforth eds., 1990, London, Elsevier AppliedScience, 49–58.

20. M. Raupach, ‘Results from laboratory tests and evaluation of literature on the influenceof temperature on reinforcement corrosion’, Corrosion of Reinforcement in Concrete,EFC Publication No. 25, J. Mietz, B. Elsener and R. Polder, Eds., IOM CommunicationsLondon, 1998, 9–20.

21. T. Liu, and R. W. Weyers, Cement Concrete Res., 1998, 28, 365–379.22. Y. Schiegg, B. Elsener and H. Böhni, On-line monitoring of the corrosion in reinforced

concrete structures, this volume, Ch. 11.23. C. Alonso, C. Andrade and J. A. Gonzalez, Cement Concrete Res., 1988, 18, 687–

698.24. F. Hunkeler, Constr. Build. Mater., 1996, 10, 381–389.


89

8.1 Background

High-strength concrete with a compressive strength above 80 N mm–2 hasbeen used for many years. Besides increasing the load-bearing capacity ofthe concrete one important aspect for the development of high-performanceconcrete mixtures is the durability, especially the resistance against chloridediffusion. Numerous investigations have been performed with regard to theabrasion resistance, capillary water suction, permeability to gases and liquids,chloride diffusion resistance or resistance against frost and de-icing salts[1–5]. Nevertheless, questions remain with regard to the behaviour of thereinforcement in high-performance concrete structures when exposed toaggressive environmental conditions. Cracks in reinforced concrete structuresallow aggressive agents like chlorides from de-icing salts or sea-water topenetrate into the concrete.

As shown in Fig. 8.1, the corrosion rate of a macrocell is a function of theanodic polarisation resistance, RA, the cathodic polarisation resistance, RC,the resistivity of the electrolyte (concrete), Rel, and the difference betweenthe rest potentials at the anode (ER,A) and the cathode (ER,C). It can bededuced from the equation that the corrosion rate is reduced if only one ofthe three resistances (RA, RC and Rel), especially the most corrosion ratedetermining resistance, increases.

Due to the low permeability and high electrolytic resistivity of high-performance concrete, it might be expected that the corrosion rates of steelin the area of cracks in concrete are far lower than in normal concrete.Another reason for a possible reduction of corrosion rates in high-performanceconcrete may be the restricted volume expansion of corrosion products at theanode due to the limited space in small cracks causing the anodic polarisationresistance, RA, to increase with time.

In order to evaluate the corrosion behaviour of steel in cracked high-performance concrete, laboratory tests have been performed on cracked concretebeams.

8Modelling of chloride-induced corrosion of

reinforcement in cracked high-performanceconcrete based on laboratory investigations

M. R A U PA C H and C. D A U B E R S C H M I D T,Aachen University, Germany



8.2 Test programme

The design of the test specimens (Fig. 8.2) allowed the measurement ofvarious corrosion parameters such as macrocell corrosion currents (anodicand cathodic currents), corrosion potentials and electrolytic resistance of theconcrete. The main rebar intersecting the crack was designed to be the anode,therefore the active area was restricted by coating the reinforcement in theuncracked area. The cathodes consisted of 24 mild steel rebars, fixed at

8.1 Simplified electrical circuit model for the corrosion of steel incracked concrete [ER,C and ER,A: rest potentials at the cathode andanode; RA, RC and Rel: polarisation resistances at the anode, thecathode and resistivity of the electrolyte/concrete; Ie: macrocellcurrent (~ corrosion rate)] [2].

Crack

Concrete

Steel

Rel

RA

ER,A

RC

ER,CRSt ª 0DE

Ie

Anode Cathode

I e

R,C R,A

A C el=

– + +

E ER R R

ER,C

DE

ER,A

DEc = Rc Ie

DEel = Rel Ie

DEA = RA Ie

EC,C

EC,A

Ie

Section A-A Section B-B

100 100 50 350Chloride solution 1%

Uncoated anodic steel area Crack

Mild steel cathodes Activated titanium cathodesEpoxy coated rebar

700

150

25 100 25

Titaniumcathode

150

Top view�B

�B

A A

50 600 50

AnodeRebar Ø 12 mm

Cathodes:Mild steelEpoxy coated

(Measures in mm)

8.2 Design of test specimens.

700


Modelling of chloride-induced corrosion of reinforcement 91

defined distances from the crack (50 to 250 mm) and with two concretecovers (23 and 47 mm). After concreting, the specimens were stored for7 days in a fog-room and afterwards for another 21 days in a climate at20 ∞C and 65 % rh. The cracks were formed by fixing the specimens in steelframes and by bending them to the designed crack widths (0.1, 0.25 and 0.5mm at the surface of the beams). Corrosive conditions were created by wet–dry cycles (1 day immersed with a 1 % chloride solution followed by 6 daysdry in a laboratory atmosphere at about 65 % rh).

In order to evaluate the corrosion behaviour of reinforcing steel bars inthe cracked concrete beams, macrocell corrosion currents, corrosion potentialsand electrolytic resistances were measured over a period of 64 cycles (atleast 15 months). The concrete mixtures used in the tests are specified inTable 8.1.

8.3 Results

8.3.1 Measurement of the electrolytic resistivity of theconcrete, Rel

To measure the resistivity over time at different depths an embedded multi-ring electrode (MRE) was used [6]. The sensor consisted of several stainless-steel rings maintained at a defined distance from one another by insulatingplastic rings (Fig. 8.3).

Cable connections through the sensor enabled the resistivity of the concreteto be determined between each pair of neighbouring stainless-steel rings bymeans of impedance measurements. The ac resistance values (in W) can beconverted to resistivity values by a sensor-specific transfer factor determinedin aqueous solutions with known conductivity.

The standard type of the multi-ring electrode-sensor allowed themeasurement of eight resistances between nine rings, down to a distance of42 mm from the concrete surface.

The specimens were stored at 22 ∞C and 52 % rh (average values). In Fig.8.4–8.8 the results of the resistivity measurements of the different types ofconcrete are presented. The highest measurable value is 20 kW m. A significantincrease in the resistivity with time is noted for all depths for the C 35concrete mixture (Fig. 8.4). This increase for all depths can be explainedbecause this comparably permeable concrete had fully dried, whereas concretemixtures C 65 and C 85-0 (Fig. 8.5 and 8.6) had only dried to about 17 mmas a significant rise can be observed only for the curves of 7, 12 and 17 mm.For the concrete mixtures produced with silica fume (Fig. 8.7 and 8.8) dryingonly occurred to a depth of 12 mm.

All curves show a clear profile of decreasing resistivity with increasingdepth. There is a tendency for the resistivity of the humid inner area of



ent in concrete92

Table 8.1 Concrete mix proportions

Mixture Cement Cement Water Silica fume Super-plasticiser* Water-to-binder Compressive strength†

type content ratio 7 d 28 d

– – kg m3 % – N mm–2

C 35 300 150 – 1.7 0.50 36 42

C 65 450 160 – 2.7 0.36 – 69

C 85-0 OPC 500 135 – 3.1 0.30 – 88

C 85-SF 455 160 30 2.4 0.33 – 92

C 115-SF 550 135 45 6.1 0.23 – 122

*Addiment, FM 93†Determined on cubes 100 ¥ 100 ¥ 100 mm3



concrete (depth of ~42 mm) to increase with the strength of the concrete.The values show that the resistivity of C 115-SF is up to 10 times higherthan that of the normal strength concrete (C 35). On adding silica fume,the resistivity is increased considerably (resistivity of C 85-0 after 1 year:360 W m, C 85 with silica fume: 1080 W m) (Fig. 8.9).

Top view

Section A-A

CableConcrete surface

Ring (noble metal)

Cable

Electrolyticresistance in W

Distance fromsurface(mm)

2.52.52.5

712172227323742

8.3 Schematic presentation of the multi-ring electrode (MRE).

Res

isti

vity

of

con

cret

e ( W

m)

100000

10000

1000

100

Concrete mixture C 35

Max. measurablevalue

7 mm 12 mm 17 mm 22 mm 27 mm 32 mm

37 mm

42 mm

0 100 200 300 400 500Concrete age (d)

8.4 Concrete resistivity of specimens C 35 at various distances fromsurface.

A A



8.3.2 Results of potential measurements ER,A and ER,C

The average differences in rest potential between the anode and cathodes ofthe macrocells in each specimen were measured several times. The valuesdiffer between 370 mV and 460 mV for all corroding macrocells. No significantinfluence was observed for the different concrete mixtures or concrete covers(Fig. 8.10).

Res

isti

vity

of

con

cret

e ( W

m)

100000

10000

1000

100

0 100 200 300 400 500Concrete age (d)

42 mm37 mm32 mm

7 mm 12 mm

17 mm

22 mm

27 mm


Concrete mixture C 65


Res

isti

vity

of

con

cret

e ( W

m)

100000

10000

1000

1000 100 200 300 400 500

Concrete age (d)

42 mm37 mm

32 mm

7 mm 12 mm

17 mm

22 mm

27 mm


Concrete mixture C85-0




8.3.3 Results of corrosion current measurements IeThe currents for specimens made of the five different concrete mixtures withcrack widths of 0.10, 0.25 and 0.50 mm were measured for 64 wet-drycycles. The current has been recorded separately for three macrocells consistingof the four bottom cathodes (concrete cover: 47 mm) with different distances

Res

isti

vity

of

con

cret

e ( W

m)

100000

10000

1000

1000 100 200 300 400 500

Concrete age (d)

42 mm37 mm32 mm

7 mm 12 mm

17 mm

22 mm27 mm


Concrete mixture C 85-SF

8.7 Concrete resistivity of specimens C 85-SF at various distancesfrom surface.

Res

isti

vity

of

con

cret

e ( W

m)

100000

10000

1000

1000 100 200 300 400 500

Concrete age (d)

42 mm37 mm32 mm22 mm 27 mm


Concrete mixture C 115-SF

7 mm 12 mm

17 mm

8.8 Concrete resistivity of specimens C 115-SF at various distancesfrom surface.



(50, 150 and 250 mm) from the crack and the anode and for three macrocellsconsisting of the four top cathodes (concrete cover: 23 mm) with differentdistances from the crack and the anode.

To calculate the mass loss of the reinforcement it was necessary to integratethe current versus time. This cumulative charge is shown as a function of thenumber of wetting periods in Fig. 8.11.

By far the highest corrosion charges from 0.5 mm wide cracks wererecorded for the reference concrete mixture C 35 (average: 20369 mA d), thelowest for the silica fume containing concrete mixtures C 85-SF (5854 mA d)and C 115-SF (7235 mA d). For the concrete mixtures C 35 and C 65 it isobvious that the corrosion charge is only slightly dependent on crack width(e.g. C 65, crack width 0.10 mm: 6885 mA d, crack width 0.50 mm: 9505mA d).

Electrolytic resistivity of concrete (W m)100 1000 10000 100000

Concrete age t = 50 d

C 65

C 85-0

C 85-S

C 35C 115-S

Max.measurable

value

Dep

th u

nd

er c

on

cret

e su

rfac

e (m

m) 0

10

20

30

40

50

Electrolytic resistivity of concrete (W m)100 1000 10000 100000

C 65

C 85-0

C 85-S

C 35

C 115-S

Max.measurable

value

Dep

th u

nd

er c

on

cret

e su

rfac

e (m

m) 0

10

20

30

40

50

Concrete age t = 400 d

8.9 Profiles of concrete resistivity for all specimens at concrete age of50 days (upper) and 400 days (lower).



A decrease in the corrosion rate over time due to an increasing anodicpolarisation resistance was not observed. However, it can not be excludedthat this effect might occur in the course of time, which can only be verifiedby long-term investigations.

8.3.4 Cathodic current distribution

To evaluate possible changes in the current distribution of macrocells inhigh-performance concrete mixtures, the local currents for the different anode–

Top cathodes c = 23 mmBottom cathodes c = 47 mm

Vo

ltag

e (m

V)

500

450

400

350

3000.10 0.25 0.50 0.10 0.25 0.50 0.50 0.50 0.50

C 35C 65 C 85-0 C 85-SF

C 115-SF

Crack width (mm)Concrete mixture

8.10 Differences in rest potential between anode and cathode in mV(age of specimens: 187 to 383 days).

0.10 0.25 0.50 0.10 0.250.50

0.50 0.500.50

C 35

C 65

C 85-0C 85-SF

C 115-SF

Crack width (mm)

Concrete mixture

5640

248

Number of wettingand drying cycles

Cu

mu

lati

ve c

orr

osi

on

cu

rren

t (m

A d

)

25000

20000

15000

10000

5000

0

Specimens (average value of two macrocells)

8.11 Cumulative macrocell corrosion currents as a function of thenumber of wetting and drying cycles.



cathode distances have been calculated. Assuming that the total corrosionrate of the anode with the top cathodes is 100 %, the contribution of eachanode–cathode cell (three cells with different distances from the crack) wasevaluated. Figure 8.12 shows the determined current balances as percentagesof the total current for the top cathodes in the left part. The same evaluationwas made for the macrocells between anodes and the bottom cathodes (rightpart of Fig. 8.12).

No systematic differences between the corrosion current of the top macrocelland the current of the bottom macrocell could be observed for all testedspecimens. Furthermore, no change of the current distribution resulting fromthe use of high-performance concrete could be evaluated.

8.3.5 Visual examinations

To verify the results of the current measurement the specimens were brokenafter 64 cycles and the positions of the cracks in the specimens as well as thedegree of corrosion were determined. Whereas the cracks in specimens withC 35 and C 65 crossed the anode (main rebar) in the planned uncoated area,in specimens of high-performance concrete with small crack widths (0.10and 0.25 mm), the cracks were found to cross the anodes in the coatedregion. It seems that the more brittle the concrete the more the cracks formedat discontinuities in the coating of the reinforcement. The results for specimenswhere the crack crossed the coated region of the main rebar are neglected inthe further evaluation.

C 35; w = 0.10 mm

C 65; w = 0.50 mmC 85-0; w = 0.50 mmC 85-SF; w = 0.50 mm

C 115-SF; w = 0.50 mm

100%

75%

50%

25%

0%

Top cathodes(c = 23 mm)

Bottom cathodes(c = 47 mm)

Bal

ance

of

the

curr

ent

(%)

250 150 50 50 150 250Distance anode – cathode (mm)

8.12 Balances of the cathodic currents of the macrocells (w = crackwidth).



However, the degree of corrosion was clearly related to the measuredcumulative corrosion current according to Faraday’s law.

8.3.6 Influence of water-to-binder ratio on resistivity andcorrosion current

Figure 8.13 shows the water-to-binder ratio of the concrete mixtures (seeTable 8.1) versus the resistivity of the concrete measured with the multi-ringelectrode at a depth of 42 mm. As expected, a decrease of the water-to-binderratio leads to an increase in the resistivity of nearby water-saturated concrete.This increase is significantly higher when silica fume is used.

In Fig. 8.14 the results of the corrosion current measurement (mean valuesfor each concrete mixture) are related to the water-binder-ratio of the concretemixtures. The determined average corrosion rate is strongly related to thewater-to-binder ratio of the concrete in macrocell corrosion.

8.3.7 Relation between resistivity and corrosioncurrent

Figure 8.15 shows the resistivity against the measured average corrosionrate, as drawn from Fig. 8.13 and 8.14.

As can be seen from Fig. 8.15, the decrease of corrosion current is notlinearly related to the increase of resistivity. Thus, the corrosion rate is notcontrolled totally by Rel, but mainly by the polarisation resistances RA andRC.

(Multi-ring electrode: depth 42 mm)

Ele

ctro

lyti

c re

sist

ivit

y ( W

m)

10000

1000

1000.2 0.3 0.4 0.5

Water-to-binder ratio

C 35-0C 65-0

C 85-0

C 85-S

C 115-S

8.13 Water-to-binder ratio of the concrete mixtures versus measuredresistivity of the specimens at a depth of 42 mm.



8.4 Numerical simulation

To verify the results of Fig. 8.15 numerical simulations of the corrosionprocess have been carried out. Actively corroding steel may exhibit microcellaction in which anodic and cathodic sites are randomly distributed over theexposure surface of the steel electrode, giving rise to uniform corrosionattack. In this condition, the polarisation behaviour of corroding steel isdescribed by the following equation relating average current density, i, andpotential change (overvoltage), DU:

0.2 0.3 0.4 0.5Water-to-binder ratio

Ave

rag

e co

rro

sio

n c

urr

ent

( mA

)

50

40

30

20

10

0

C 85-S

C 65-0C 85-0

C 35-0

C 115-S

8.14 Water-to-binder ratio of the concrete mixtures versus measuredcorrosion current.

100 1000 10000Electrolytic resistivity (Wm)

(Multi-ring electrode: depth 42 mm)

Ave

rag

e co

rro

sio

n c

urr

ent

( mA

)

50

40

30

20

10

0

C 85-S

C 65-0

C 85-0

C 35-0

C 115-S

8.15 Electrolytic resistivity of the concrete (measured with multiring-electrode) versus average corrosion current.



i i U

bU

b = exp

ln(10)– exp

– ln(10)corr

a c

D DÊË

ˆ¯

ÊË

ˆ¯

ÈÎÍ

˘˚

(8.1)

with ln(10) = 2.303ba = anodic Tafel slope = 90.7 mV dec–1

bc = cathodic Tafel slope = 176.3 mV dec–1

icorr = self corrosion rate, here 1.0 mA cm –2 [4].

For passive steel it is assumed that anodic reactions can only proceed toa very limited extent. The electrochemical behaviour of passive reinforcingsteel under cathodic polarisation is given by:

i U

b i

Ub

i = 1 – exp

–ln(10) 1 – exp

– ln(10)

c corr

c

lim

DD

ÊË

ˆ¯

ÊË

ˆ¯ (8.2)

with ilim = limiting diffusion current density due to oxygen diffusionbased on data obtained from experimental investigations [4].

Based on these equations, the corrosion current of the macrocells withcoplanar arrangement of local anode and macro cathode as a function of thedistance of the cathodes from the anodes can be calculated. Figure 8.16shows the results for the C 35 and the C 115 concretes with Ue = 400 mV, Ra

= 0 W and with electrolytic resistivity of 500 W m and 5000 W m, respectively.Figure 8.17 shows the results of calculations according the equations

(8.1) and (8.2) with the potential as a function variable of the distance fromthe anode.

Furthermore, the corrosion current of the specimens can be calculated asa function of the density of reinforcement. In Fig. 8.18, the curves of the

El.

curr

ent

den

sity

(mA

cm

–3)

0.15

0.12

0.09

0.06

0.03

0.006 12 18 24 30

Distance from the anode/crack (cm)

I/b/h = 60/15/15 cmUe = 400 mV

C 35, Rel = 500 W *m

C 115, Rel = 5000 W *m

8.16 Calculated corrosion current of cracked specimens with C 35and C 115-SF concrete as a function of the distance between anodeand cathode.



calculation with Ue = 400 V and Ra = 0 W are shown. With increasing densityof reinforcement, the corrosion current increases too. For the high-performanceconcrete C 115-SF (Rel = 5000 W m) the calculated corrosion current islower than for the concrete C 35 (Rel = 500 W m), following a non-linearrelationship.

Figure 8.19 shows the ratios of calculated corrosion current of C 115-SFto the corrosion current of C 35. For concretes with a 10-fold electrolyticresistivity ratio, the corrosion current decreased only by a factor of two for

6 12 18 24 30Distance from the anode/crack (cm)

I/b/h = 60/15/15 cmUe = 400 mVC 35, Rel = 500 W *m

C 115, Rel = 5000 W *m

Po

ten

tial

(m

V)

0

–50

–100

–150

–200

–250

–300

–350

–400

–450

8.17 Calculated distribution of the potential of cracked specimenswith C 35 and C 115-SF concrete as a function of the distancebetween anode and cathode.

I/b/h = 60/15/15 cmUe = 400 mV

C 35, Rel = 500 W m

C 115, Rel = 5000 W m

Mac

roce

ll cu

rren

t ( m

A)

250

200

150

100

50

00 5 10 15 20 25 30 35 40 45 50

Reinforcement density (cm2 cm–1)

8.18 Calculated corrosion current of cracked specimens with C 35and C 115-SF concrete versus density of reinforcement.



low reinforcement densities. For high reinforcement densities this factor isabout 4.

8.5 Conclusions

Tests have been carried out to determine the corrosion mechanisms of specimensproduced with cracked high-performance concrete beams.

As expected, the resistivity of the electrolyte (concrete) increasessignificantly in high-performance concrete. The resistivity of concrete mixtureC 115 in humid conditions is about ten times higher than the resistivity ofconcrete mixture C 65 in humid conditions.

This increase in the electrolytic resistivity leads to a reduction in thecorrosion rates in specimens with high-performance concrete. Accordingly,the corrosion currents of macrocells in high-performance concrete mixturesare also reduced, e.g. the average corrosion rate of concrete mixtures C 85-SF and C 115-SF is about 1/3 of the current of C 35 (Fig. 8.11) under theconditions investigated.

It can be summarised that the mechanisms of corrosion do not change inhigh-performance concrete: the major type of corrosion is macrocell corrosion,the balances of currents as a function of the distance from the cathode to theanode remain nearly the same for normal and high-strength concrete andthere is no sign of a reduction in the current due to insufficient space forcorrosion products in specimens of high-performance concrete after 15 monthsof exposure to aggressive wet and drying cycling.

Comparing the ratio of the increasing resistivity and the ratio of reductionof the corrosion rate with regard to the equation of Fig. 8.1, it can be determined

8.19 Ratio of calculated corrosion current of cracked specimens withC 35 and C 115-SF concrete versus content of reinforcement.

Rel

. m

acro

cell

curr

ents

C 1

15/C

35 I /b/h = 60/15/15 cm

Ue = 400 mVRel,C115 = 5000 W *mRel,C35 = 500 W *m

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

00 10 20 30 40 50

Reinforcement density (cm2 cm–1)



that the corrosion rate is not controlled totally by Rel, but mainly by thepolarisation resistances RA and RC. These results have been confirmed bynumerical simulations of the corrosion process. These simulations show thatthe corrosion rate is mainly dependent on the polarisation resistances of theanode and the cathode.

The use of high-performance concrete with cracks exposed to severe chlorideattack leads to a reduction in the corrosion rate of the anode in comparisonto normal strength concrete. So the expected service lifetime of the structurecan be prolonged significantly by using high-performance concrete. Toinvestigate long term effects, like a possible anodic self polarisation, additionaltests over increased periods are planned.

8.6 References

1. Guse, U. and Hilsdorf, H. K., Durability Aspects of High Strength Concrete. in High-Performance Concrete. ACI International Conference, Supplementary Papers, Singapore,1994, Malhotra, V. M., (Ed.), American Concrete Institute, Detroit 1994. 229–250.

2. Raupach, M., ‘Chloride-induced macrocell corrosion of steel in concrete – theoreticalbackground and practical consequences. Constr. Build. Mater. 1996, 10(5), 329–338.

3. Raupach, M., ‘Corrosion of steel in the area of cracks in concrete – laboratory test andcalculations using a transmission line model. Corrosion of Reinforcement in ConcreteConstruction, 4th International Symposium, Cambridge, UK, 1–4 July 1996, Page,C. L.; Bamforth, P. B.; Figg, J. W. (Eds.), The Royal Society of Chemistry, Cambridge,1996, 13–23.

4. Raupach, M. and Gulikers, J., Electrochemical models for corrosion of steel in concrete– introduction for the planned new EFC-WP11 Task Group, EUROCORR 2000.

5. Raupach, M. and Gulikers, J., ‘A simplified method to estimate corrosion rates – anew approach based on investigations of macrocells, in 8th International Conferenceon Durability of Building Materials & Components – Service Life and Asset Management,Vancouver, May 30–June 3, 1999, Vol. 1, 376–385.

6. Schießl, P., Breit, W. and Raupach, M., ‘Investigations into the effect of coatings onwater distribution in concrete using multi-ring electrodes, in Concrete Bridges inAggressive Environments, Philip D. Cady International Symposium, Minneapolis,November 9–10, 1993, Weyers, R. E. (Ed.), American Concrete Institute, Detroit ACISP-151, 1994, 119–133.


105

9.1 Introduction

Stray currents, arising for instance from railways, cathodic protection systems,or high-voltage power lines, often induce corrosion on buried metal structures,leading to severe localised attack.1 They may be dc or ac depending upon thesource. Stray currents can deviate from their intended path because they finda parallel and alternative route. They may also find a low resistance path byflowing through metallic structures buried in the soil (pipelines, tanks, industrialand marine structures). For instance, underground pipelines can pick upstray current from a railway system at some point remote from the tractionpower substation and discharge the current to the soil and then back to therail near to the substation. Stray direct currents are known to be much moredangerous than stray alternating currents. In the case of dc, a cathodic reaction(e.g. oxygen reduction or hydrogen evolution) takes place where the currententers the buried structure, while an anodic reaction (e.g. metal dissolution)occurs where the current returns to the original path, through the soil. Metalloss results at the anodic points, where the current leaves the structure;usually, the attack is extremely localised and can have dramatic consequences,especially on pipelines. Effects of stray ac currents are more complex. It hasbeen shown that ac can influence the anodic behaviour of steel and, thus,may increase the corrosion rate of steel, as well as galvanic effects.2,3

Nevertheless, steel in soil is usually under cathodic protection and it was shownthat stray ac current can induce corrosion only in those particular circumstanceswhere very high currents are picked up by the buried structure, so that thecurrent density in the points where the current enters or leaves the structureis extremely high (a threshold ranging from 20–100 A m–2 has been proposed4).

Stray currents can also flow through reinforced or prestressed concreteand produce an alteration of the electrical field inside the concrete, whichcan influence the corrosion of embedded steel. Several types of structuresmay be subjected to stray current, such as bridges and tunnels of the railwaynetworks or structures placed in the neighbourhoods of railways. Here, the

9Influence of stray currents on corrosion of

steel in concrete

L. B E R T O L I N I, M. C A R S A N A andP. P E D E F E R R I, Politecnico di Milano, Italy



concrete, like the soil in buried structures, is the electrolyte and the reinforcingbars or prestressing wires can pick up the stray current (Fig. 9.1). It has beenshown that stray dc currents rarely have corrosive consequences on steel inconcrete, in contrast to their effect on metallic structures in the soil.5,6 Infact, steel in alkaline and chloride-free concrete is passive. Passivity, besidesbeing essential for protecting the reinforcement from the environmentalaggressiveness, also provides resistance to stray currents. Figure 9.2 depictsthe anodic polarisation curve of passive steel in concrete and shows that at

Dc electricsubstation

Cathodic reaction

Anodic reaction

Concrete

Anodic reaction

Cathodic reaction

¨ Reinforcement ¨

DV

9.1 Example of stray current from a dc railway line picked up by steelreinforcement in concrete.

Potential(mV vs SCE)

500

0

–500

–1000

Ecorr

Ec

Log (current density)(a)

DV

Potential(mV vs SCE)

500

0

–500

–1000

Log (current density)(b)

Ea

Ecorr

9.2 Schematic representation of electrochemical conditions in (a) thecathodic and (b) the anodic zones of reinforcement in non-carbonated and chloride-free concrete which is subject to straycurrent I.

I


Influence of stray currents on corrosion of steel in concrete 107

potentials below +600 mV vs SCE no iron dissolution or any other anodicprocess takes place and, thus, it is impossible for the current to leave themetal. Before the stray current is picked up by the reinforcement, a significantpotential difference (DV) has to be produced between the point where thecurrent enters the reinforcement (cathodic site, Fig. 9.2a) and the point wherethe current returns to the concrete (anodic site, Fig. 9.2b).

Experimental tests showed that passive reinforcement in non-carbonatedand chloride-free concrete offers a high intrinsic resistance to stray current,since the driving force required to produce the circulation of an appreciablecurrent density in the anodic areas (i.e. >2 mA m–2) is at least 500 mV.5

Furthermore, even if such a condition is reached and current circulates throughthe reinforcement, this does not automatically lead to corrosive attack. Theanodic process taking place at potentials higher than about +600 mV (SCE)is oxygen evolution, instead of iron dissolution.

Nevertheless, it has been shown that an attack may occur when the currentflows for sufficiently long periods of time.7 The initiation of corrosion wasascribed to the depletion of the alkalinity in the vicinity of the anodic areaspromoted by the anodic reaction of oxygen evolution (2H2O Æ O2 + 4H+ +4e–). In the case of structures contaminated by chlorides, even at levels toolow to initiate pitting corrosion, stray currents may have more seriousconsequences. In fact, stray currents stimulate the initiation of pitting corrosionby taking the steel potential to values higher than the pitting potential. Oncecorrosion has initiated on the reinforcement, for instance due to carbonationor chlorides, the effects of stray currents become similar to that experiencedby steel buried in the soil.

In this paper, the effects of stray currents on the corrosion of steel embeddedin atmospherically exposed concrete are studied. Both the mechanisms ofcorrosion initiation on initially passive reinforcement and the effects of straycurrents on reinforcement that was already corroding have been studied. Theinfluence of several factors has been investigated, such as: the type of current(ac or dc), the current density, the presence of interruptions in the circulationof current, and the chloride content in the concrete. The mechanism of initiationof corrosion has also been investigated by measuring changes in pH andchloride content induced by the stray current in the cement paste in thevicinity of the steel surface. This work deals with consequences of straycurrents on reinforced concrete structures; high-strength steels for prestressedconcrete, which can also suffer corrosion due to hydrogen embrittlement incathodic areas, are not taken into consideration.

9.2 Experimental tests

Tests were carried out on specimens of cement paste and concrete with twosteel inserts subjected to the circulation of dc and ac currents (Fig. 9.3).



Specimens of Fig. 9.3a had two parallel plates of carbon steel embedded incement paste made of portland cement and a water/cement (w/c) ratio of0.55. Only the two opposite surfaces of the plates were exposed to thecement paste (the rest was masked with epoxy). Fixed reference electrodes,made of a thin wire of mixed metal oxides activated titanium (MMO), wereembedded in the vicinity of each plate. Specimens of Fig. 9.3b had twoparallel bars of carbon steel embedded in concrete made of 350 kg m–3 ofportland cement, w/c ratio 0.55, and 1900 kg m–3 of crushed limestoneaggregate (the average strength of the concrete was 50 MPa). Chlorides wereadded in amounts ranging from 0.1 to 0.8% by weight of cement to some ofthe specimens of cement paste or concrete.

The effects of direct current were studied by applying a constant directcurrent between the two electrodes (steel plates or rods) in the specimens, sothat one was the cathode and the other one was the anode. Specimens wereexposed in a climatic chamber at 20 ∞C and 95% rh. Current densities of 1and 10 A m–2 were applied to specimens made of cement paste and8.6 A m–2 to specimens in concrete. In order to study the effects of theinterruption of current, tests were also carried out with cycles during whichcurrent circulated for 1 h and was then switched off for 1 or 3 h.

All the tests were carried out until cracking of the specimen occurred.During tests, potentials of steel against the fixed reference electrodes and thefeeding voltage were monitored. Periodically, depolarisation tests were carriedout by interrupting the current for 5 min. and measuring the subsequentchanges in the steel potential.

The potential of the activated titanium electrodes was regularly calibrated

Steel plate(35 ¥ 35 mm)

Metal mixed oxideactivated titaniumelectrode (MMO)

50

50

100

(a)

Steel barf10 mm

70

70

100

(b)

9.3 Specimens used for tests: (a) steel plates embedded in cementpaste, (b) steel bars embedded in concrete (dimensions are in mm).



against an external SCE. These measurements also allowed estimation of thepH of the pore solution near the surface of the steel, since the potential ofactivated titanium depends on pH.

On specimens of the type shown in Fig. 9.3a, whose geometry favours auniform distribution of current between the opposing steel plates, furtheranalyses were carried out at the end of tests, i.e. when cracking occurred onthe anodic side of the specimen. The steel plates were removed from thecement paste and weight loss was measured on the steel plate that wasanodically polarised and underwent corrosion. The pH of the cement pastewas analysed by spraying phenolphthalein and other commercial pH indicatorson a fracture surface perpendicular to the steel plates; the area close to thesteel plates was observed with a stereomicroscope. Chloride content wasalso measured by cutting the cement paste between the steel plates into sliceswith a thickness of 10 mm; powder from each slice was digested in nitricacid and chloride content was measured by means of potentiometric titration.Chloride analyses were also carried out on fragments of cement paste collectedwithin 1–2 mm of the steel surface.

Tests with alternating current were carried out on specimens of cementpaste with 0 and 0.2% of chloride by weight of cement; a current density of40 A m–2 was applied between the two plates for 2 months and the corrosionrate of steel was monitored, by means of polarisation resistance measurementscarried out after the interruption of ac for 4 h. Tests were also performed onthree specimens with the geometry of Fig. 9.3b, which had the two barsembedded in concrete with different chloride contaminations: 0–0.4, 0–0.8and 0.4–0.8% by weight of cement. Alternate current of 50 A m–2 wasinitially applied to the passive bars for 5 months and the corrosion rate wasmeasured. Afterwards, corrosion was initiated in the bar in concrete with thehigher chloride contamination of each specimen (by imposing an anodicdirect current of 3 A m–2). Consequently, a macrocouple was generatedbetween the two bars in each specimen. Alternating currents of 20 A m–2

were superimposed for various lengths of time and their influence on themacrocouple current was monitored.


9.3.1 Effects of direct current

Corrosion can be induced on passive steel in concrete when it is subjected tothe circulation of anodic current for a long period. Afterwards the anodiccurrent stimulates the corrosion rate and can lead to cracking of the concrete.Fig. 9.4 shows, as an example, the results of the test on the specimen withcement paste contaminated with 0.4% of chloride by weight of cement whichwas subjected to the circulation of a nominal current of 1 A m–2 (actual



current density was 1.1 A m–2). Both steel plates were initially passive andhad a free corrosion potential of around –200 mV against the fixed referenceelectrode of mixed metal oxide (MMO) activated titanium. When the externalcurrent density of 1 A m–2 was applied, the steel plate that was polarisedanodically reached a potential around +700 mV, showing that oxygen evolutiontook place at its surface. Conversely, the steel plate that was polarisedcathodically reached a potential of about –1.1 V (Fig. 9.4a). The feedingvoltage was 2.2 V (Fig. 9.4b). The initiation of corrosion on the anodic steelplate took place after 78 hours of testing, when a charge of 84 A h m–2 hadbeen circulated; it was detected by a sharp decrease in the potential (and thefeeding voltage as well). The initiation of corrosion also led to a significantchange in the depolarisation of the anodic steel when the current was interrupted.Comparison of Fig. 9.4c (test carried out before corrosion initiated) and Fig.9.4d (carried out after corrosion initiated) shows that after corrosion hadinitiated, a potential of about –500 mV vs MMO was quickly reached byswitching the current off. After corrosion initiation, the potential of anodicsteel also showed remarkable fluctuations. Cracks developed in the cementpaste in contact with the corroding steel plate after 190 days of testing.

Results similar to those shown in Fig. 9.4 were obtained with all the testscarried out on specimens made of cement paste or concrete with differentchloride contaminations and applied currents. The onset of corrosion wasassociated with a decrease in the potential of the anodic steel (both in the

Anode

Corrosioninitiation Cracking

Ecorr = –200 mV

Cathode

0 50 100 150 200Time (h)

(a)

Po

ten

tial

(m

V v

s M

MO

)

1000

500

0

–500

–1000

–1500

9.4 Results of the test on the specimen with cement pastecontaminated with 0.4% of chloride by weight of cement that wassubjected to the circulation of 1 A m–2. Potential of anodic andcathodic steel plates (a) and feeding voltage (b) in time.Depolarisation tests carried out after (c) 21 and (d) 93 h ofapplication of the current.



Anode

Cathode

0 1 2 3 4 5 6Time (min)

(c)

Po

ten

tial

(m

V v

s M

MO

)

1000

500

0

–500

–1000

–1500

Current switched off

Anode

Cathode

0 1 2 3 4 5 6Time (min)

(d)

Po

ten

tial

(m

V v

s M

MO

)

1000

500

0

–500

–1000

–1500

Current switched off

0 50 100 150 200Time (h)

(b)

Feed

ing

vo

ltag

e (V

)

2.4

2.3

2.2

2.1

2

1.9

1.8

Corrosioninitiation

9.4 Continued



presence of current and during depolarisation tests). Tables 9.1 and 9.2summarise the results, showing the time required for the onset of corrosionand the charge actually circulated (during some tests, current decreased becauseof the high resistivity of the concrete; this occurred on specimens withoutchlorides before the initiation of corrosion and in specimens with chloridesonly after the initiation of corrosion when a very high ohmic drop contributionoccurred in the vicinity of the anode). The times and charges passed from thebeginning of the test up to cracking of the specimen are also reported. Forspecimens made of cement paste (Table 9.1), the weight loss measured onthe steel plate is also shown. In general, the measured weight loss, at least forspecimens with chlorides, was in agreement with weight loss estimated byassuming that all the charge circulated after initiation led to iron dissolution.In chloride-free cement paste, the weight loss was only a fraction of thetheoretical value, showing that oxygen evolution occurred even after corrosion(probably depassivation initially took place only in certain areas).

Anodic current density

Corrosion initiated on steel embedded in cement paste without chlorideonly after more than 200 h of application of an anodic current density of10 A m–2 (Table 9.1). A current density of 1 A m–2 could not initiate corrosioneven after 14 months (>10 000 h) of continuous application, although thecharge that circulated (>10000 A h m–2) was much higher than the charge thatcould initiate corrosion during the test with a current density of 10 A m–2

(2200 A h m–2). These results show that a decrease in the anodic currentdensity can lead to a significant decrease in the aggressiveness of straycurrent.

Tests on steel embedded in concrete without chloride (Table 9.2) are inagreement with those obtained in cement paste, since a charge of about5700 A h m–2 was enough to initiate corrosion with an intermediate currentdensity (current density decreased in time from the initial value of 8.6 A m–2

to values of 2 A m–2, because a high ohmic contribution progressively generatednear the steel surface, and reached values close to the maximum voltage of thecurrent feeder, i.e. 24 V).

These results show that the risk of corrosion induced by stray current onsteel in alkaline and chloride-free concrete is extremely low. In fact, only ahigh current density circulating for a very long time can induce corrosion atanodic sites. Since the reinforcement is not coated (and thus current is notforced to concentrate in small areas of defects of the coating, as occurs onsteel in metallic structures), it is seldom that high current density can beinduced by stray currents in concrete.


Influence of stray currents on corrosion of steel in concrete113

Table 9.1 Results of tests on specimens made of cement paste: time (ti) and charge (Qi) for the initiation of corrosion on the anodic steelplate; time to cracking (tcrack) and charge to cracking (Qcrack); weight loss (Dm) measured at the end of tests (usually tests wereinterrupted after cracking)

Nominal current Chloride Cycle ti Qi tcrack Qcrack Dm(A m–2) (% wt cement) (h) (A h m–2) (h) (A h m–2) (g m–2)

1* 0 continuous >10000 >12000 – – –1 on – 1 off >10000 >5600 – – –

0.1 continuous 3885 3920 4070 4180 620.2 continuous 2200 2220 2325 2396 38

1 on – 1 off >10000 >5600 – – –0.4 continuous 78 84 190 204 301†

1 on – 1 off 2230 1135 2280 1430 3810 0 continuous 231 2206 263 2520 371†

1 on – 1 off 338 1690 670 3790 10700.1 continuous 22 220 43 437 151

1 on – 1 off 231 1155 262 1760 139†

0.2 continuous 12 120 20 898 247†

1 on – 1 off 43 215 101 508 2090.4 continuous 6 60 22 225 182

5.8 58 21 129 57†

1 on – 1 off 15 75 22 133 586 30 27 130 63

*Actual current density ranged from 1 to 1.2 A m–2 depending on the specimen. Charges have been calculated on the basis of the actualcurrent density monitored throughout the test.†Current was maintained for a certain time after cracking, so further charge circulated after tcrack and comparison of Dm with Qcrack is notpossible.



Chloride contamination

Stray dc currents may have serious consequences in chloride contaminatedconcrete. In Fig. 9.5, the results are plotted of tests carried out in cementpastes with chloride contents up to 0.4% by weight of cement. The chargerequired for the onset of corrosion shows a remarkable decrease as the chloridecontent increases. Figure 9.5 shows that even a current density of 1 A m–2

can initiate corrosion in the presence of small amounts of 0.1 and 0.2%chloride by weight of cement (i.e. not dangerous for pitting corrosion in theabsence of stray current). Figure 9.5 also confirms the higher risks connectedwith higher anodic current densities: the charges required for corrosion initiationwith a current density of 10 A m–2 are more than one order of magnitudelower than those due to 1 A m–2, i.e. times for initiation of corrosion are morethan 100 times lower.

Fig. 9.6 compares results with current density of 8.6–10 A m–2 obtainedon steel in cement paste and in concrete with chloride contents up to 0.8% byweight of cement. Results in cement paste and concrete are in good agreement;slightly higher charges for tests in concrete may be a consequence of theslightly lower current density. It can be observed that when the chloridecontent approaches 0.8%, the charge for initiation of the attack reduces to afew A h m–2. This is not surprising, since 0.8% is a chloride content that mayin itself be enough to initiate corrosion in non water-saturated concrete.Nonetheless, stray currents may have an adverse effect even under these

Table 9.2 Results of tests on specimens made of concrete: corrosion rate (icorr)before application of the nominal current of 8.6 A m–2, time (ti) and charge (Qi) forinitiation of corrosion on the anodic steel bar, and time (tcrack) and charge (Qcrack)for cracking of concrete

Chloride Cycle icorr ti Qi tcrack Qcrack

(% wt cement) (mm year–1) (h) (A h m–2) (hours) (A h m–2)

0 continuous 1.2 1377 5690 1515 60601.3 1370 5840 1515 6170

1 on – 1 off 0.3 1950 4750 2210 5000

0.4 continuous 0.2 7.5 65 46 3900.5 12.5 108 43 370

1 on – 1 off 1.6 20 86 135 6400.5 21 90 105 460

1 on – 3 off 0.2 48 103 265 5701.2 40 86 135 330

0.8 continuous 3.8 0.5 4 69 6001.5 0.6 5 22 190

1 on – 1 off 2.2 2.3 10 23 1053.2 0.5 5 97 492

1 on – 3 off 1.5 0.3 3 135 2905.4 0.6 5 97 212



conditions, since they can promote the initiation of incipient pits or extendthe corroding area.

Table 9.2 shows that, even before the tests began, several specimens hadaverage corrosion rates of 3–5 mm year–1 (measured by polarisation resistancetests); such values show that pitting corrosion had already initiated. Followingthe application of the external current, the steel potential reached values ofabout +700 mV and, in the case of passive specimens, and only after severalminutes, displayed the typical sharp decrease that has been related to initiationof corrosion (the time for initiation was estimated as 0.3–0.6 h, depending onthe specimen, Table 9.2). This result suggests that, although corrosion wasinitiated in some spots even before testing, the application of the anodiccurrent could extend the attack to most of the exposed surface of steel (thiswas confirmed by observation of the steel surface at the end of the tests).

Interruptions in the stray current

Stray currents produced by transit systems are non-stationary, and thus theeffect of interruptions of the current should be taken into consideration.Figures 9.5 and 9.6 also show the results of tests in which cycles of circulationof current are alternated with periods of interruption (1h on – 1h off or 1h on– 3 off). The periodical interruption of current had a beneficial effect, sinceit increased the charge required for the initiation of corrosion. This effectwas noticeable in cement pastes with chloride contents lower than 0.4% by

Ch

arg

e (A

h m

–2)

10000

1000

100

10

Cycle: cont 1 on–1 off

1 A m–2

10 A m–2

1 A m–2

10 A m–2

0 0.1 0.2 0.3 0.4Chloride (% by weight of cement)

9.5 Charge required for the initiation of corrosion on steel platesembedded in cement pastes with different chloride contents, thatwere polarised anodically with current densities of 1 A m–2 or10 A m–2.



weight of cement (Fig. 9.5). It could not be observed in tests in concrete with0.8% chloride (Fig. 9.6), since corrosion always initiated during the firsthour of the test (before any interruption could be made).

The role of the anodic current in promoting the corrosion of steel inconcrete has been explained by the production of acidity at the steel surfacedue to the anodic reaction of oxygen evolution6,7. In the presence of chlorides,there can also be an enrichment of these ions in the vicinity of the steelsurface, due to electrical migration, that further favours corrosion. The gradientsof ionic concentration in the pore solution produced by the anodic reactionand the electromigration near the steel surface can be mitigated by theinterruption of the current. The increase in hydroxyl ions and decrease inchloride ions near the steel surface, which occurs during periods of rest, maydelay the initiation of corrosion.

At the end of the tests on the cement paste, analysis of pH by means ofphenolphthalein and other pH indicators never showed any detectable changein pH near the steel surface (i.e. in the cement paste within 0.1 mm of thesteel plate).

Activated titanium electrodes fixed 1 mm from the steel surface did notshow any change in potential during the tests, confirming that the pH of thecement paste did not vary macroscopically. Chloride analysis did not showevidence of any appreciable change in the chloride content, even very closeto the surface of corroding steel. Therefore, possible changes of compositionof the cement paste induced by stray current, which can be responsible for

Ch

arg

e (A

h m

–2)

10000

1000

100

10

10 0.2 0.4 0.6 0.8 1

Chloride (% by weight of cement)

Cycle:

Concrete:

Cement paste:

Cont. 1 on–1 off 1 on–3 off

9.6 Charge required for the initiation of corrosion on steel inconcretes and cement pastes with different chloride contents (currentdensity was 10 A m–2 in specimens made of cement paste and 8.6 Am–2 in specimens made of concrete).



corrosion initiation, would be restricted to within a very narrow distance ofthe steel surface.

9.3.2 Effects of alternating current

The effects of alternating current on passive steel were studied by applying40 A m–2 ac (50 Hz) to the specimens of Fig. 9.3a made of cement pastesboth free of chloride and with 0.2% of chloride by weight of cement. Thecorrosion rate was monitored with polarisation resistance tests carried outduring interruptions of the ac current. No significant changes with respect tothe initial value of 1–2 mm year–1 were observed during 60 days of applicationof the ac current. At the end of the tests, the steel plates were removed fromthe cement paste and no corrosion could be observed.

Tests were also carried out on bars in concrete with up to 0.8% chlorideby weight of cement subjected to 50 A m–2 ac. No effects were observed onpassive steel in concrete with up to 0.4% chloride; the corrosion rate waslower than 1 mm year–1 even after 5 months of circulation of ac. The corrosionrate increased for steel in concrete with 0.8% chloride; in that case steel wascorroding before the application of ac, with a corrosion rate of 1.5–2 mmyear–1 and at the end of the test corrosion rates increased to 7.5–10 mmyear–1. These results suggest that high ac currents may have an adverseeffect on steel subjected to pitting corrosion in chloride-contaminated concrete.Such an influence was also evidenced by an increase in the macrocouplecurrent between corroding and passive steel. For instance, Fig. 9.7 showsthat the superposition of 50 A m–2 ac for 5 min led to a temporary noticeableincrease in the macrocouple current between corroding steel in concrete with0.8% chloride and passive steel in concrete with 0.4% chloride. Changes inthe macrocouple current density were always observed after the applicationof ac pulses (even if they only lasted 20 s), showing that ac actually influencesthe electrochemical behaviour of corroding steel, even for a certain timeafter it ceases. However, such influence appeared to be rather complex;sometimes ac also led to a temporary change in the direction of the macrocouplecurrent.

Further studies are required in order to clarify the actual role of stray accurrents on corrosion of steel in concrete. Attention should also be dedicatedto possible synergistic effects of ac and dc stray currents, not only in stimulatingthe corrosion rate of depassivated steel but also in promoting corrosion onpassive steel.

9.4 Conclusions

During this study, it was found that dc stray currents could induce corrosionon steel in contact with cement paste and concrete, in the areas where the



anodic reaction took place. Corrosion initiated only after the stray currenthad circulated for a certain time. The initiation of corrosion appeared not tobe simply related to the time, the current density or the circulated charge.The amount of charge leading to the initiation of corrosion showed a noticeableincrease if the current density decreased, if the concrete did not containchloride, or if the current was not continuous (i.e. if it was periodicallyinterrupted).

In chloride-free cement paste, corrosion did not initiate even after 14months of continuous application of 1 A m–2, after a charge in excess of10,000 A h m–2 had been circulated, while only 10 days and a charge of2,200 A h m–2 were enough to initiate corrosion with an anodic currentdensity of 10 A m–2. Results for steel in concrete were in agreement withthose for steel in cement paste. The presence of small amounts of chloridesled to a noticeable decrease in the charge required for the initiation of corrosion.There was evidence that stray dc current can also increase the corrosion rateon steel already corroding in chloride-contaminated concrete, since it canpromote the initiation of incipient pits or extend the corroding area.

Analyses of cement paste in the vicinity of steel subjected to stray current-induced corrosion could not detect any significant change in pH or chloridecontent related to the initiation of corrosion.

The ac current proved to be much less dangerous than the dc; in fact, evencurrent densities up to 50 A m–2 were not able to induce initiation of corrosionon steel. Nevertheless, it was shown that ac current may stimulate macrocouples

9.7 Effect of 50 A m–2 ac on the macrocouple current densitybetween corroding steel in concrete with 0.8% chloride by weight ofcement and passive steel in concrete with 0.4% chloride by weight ofcement. Potential of anodic steel is also shown.

Mac

roco

up

le c

urr

ent

(mA

m–2

)

10

8

6

4

2

0

50 A m–2

ac

Potential

Macrocouple current

0 5 10 15 20 25Time (min)

–200

–250

–300

–350

Po

ten

tial

(m

V v

s M

MO

)



that take place in the concrete between passive and corroding steel and canincrease the corrosion rate on corroding steel in chloride-contaminated concrete.

9.5 References

1. L L Shreir, R A Jarman and G T Burstein, (Eds.), ‘Corrosion’, 3rd edition,Butterworth Heineman, 1994.

2. D A Jones, ‘Effect of alternating current on corrosion of low alloy and carbon steels’,Corrosion, 1978, 34(12), 428–433.

3. A W Hamlin, ‘Alternating Current Corrosion’, Mater. Perf., 1986 25(1), 55–58.4. G Heim, T Heim, H Heinzen and W Schwenk, ‘Research on corrosion of steel under

cathodic protection due to alternate current’ (In German), 3R International, 1993, 32,246–249.

5. L Bertolini, F Bolzoni, T Pastore and P Pedeferri, ‘Stray current induced corrosion inreinforced concrete structures’, in Progress in the Understanding and Prevention ofCorrosion’, Eds. J. M. Costa, A. D. Mercer, Institute of Materials, London, 1993,658–664.

6. L Bertolini, L Lazzari and P Pedeferri, ‘Factors influencing stray current inducedcorrosion in reinforced concrete structures’, L’industria italiana del cemento, April1996, 709, 268–279.

7. L Bertolini, F Bolzoni, M F Brunella, T Pastore and P Pedeferri, ‘Stray current inducedcorrosion in reinforced concrete structures: resistance of rebars in carbon, galvanizedand stainless steels’ (in Italian), La metallurgia italiana, 1996, 88(5) 345–351.


120

10.1 Introduction

During the inspection of a 1 km-long twin road bridge with longitudinalprestressing, corroded and broken tendon wires were detected by chance.Further investigations showed that the corrosion had been induced by de-icing salts, which reached the girder through water-bearing aeration tubesand a leaking drainage system [1].

10.2 Results of detailed corrosion inspection

Potential measurements were used to assess the state of corrosion of thereinforcement of the bridge over a surface area of about 12 000 m2.

In order to evaluate the effective state of corrosion fully, 35 inspectionfields were selected in areas of increased corrosion risk. In these fields, theconcrete was removed to reinforce the first layer of prestressing cables,which had a concrete cover of between 35 mm and more than 100 mm. Thedegree of corrosion of the ducts containing the cables was assessed using thescheme given in Table 10.1. Nearly all of the inspected prestressing cablesof the first layer showed corrosion. In about 35% of the inspection fields theducts had been perforated. Broken tendon wires were detected in two morecases, but there was no evidence for hydrogen-induced cracking. In the otherfields, corrosion was observed generally on a few single wires. As far ascould be seen, not more than 8 of the 55 wires of one cable had been corrodedat the time. In most cases, the cross sectional loss was up to about 25%.

The rebars showed localised corrosion in about 25% of the inspectionfields. On the outside of the box girder, the reduction of cross section wasbetween 10 and 40%; inside the box girder it was up to 100%. It was estimatedthat pitting corrosion on rebars must be expected on 3% of the surface of theeastern bridge and on 1% of the western bridge.

In Fig. 10.1, the empirical correlation is shown between potential and thedegree of corrosion of the duct found in the 35 inspection fields of the

10Assessment and monitoring of corrosion

velocity of rebars and prestressingcables of a bridge

D. B I N D S C H E D L E R, Swiss Society for CorrosionProtection, Zürich, Switzerland


Assessment and monitoring of corrosion velocity 121

investigated bridge. This shows quite a sharp limitation in the extent ofattack towards more positive potentials. This allowed the definition of acritical potential with a certain safety margin. In areas with potentials morenegative than this critical potential the presence of perforated ducts waspossible at the time of the inspection. It must be mentioned that a correlationbetween potential and corrosion of the duct cannot be expected in any case,because the potential is dominated by the corroding rebars. This also explainsthe fact that, in some cases, the prestressing cables showed no corrosion orjust slight corrosion even at very negative potentials. There was no correlationbetween the degree of corrosion of the tendon wires and the potential. Ingeneral, it must be expected that there are corroded wires when the ducts areperforated.

From the correlation in Fig. 10.1 it was concluded that there are more than200 areas on the bridge with a latent corrosion risk for the prestressing cables.

Table 10.1 Degrees of corrosion

Degree of Condition of ductcorrosion

1 Rust points2 Rust stains3 Surface completely corroded4 Local perforations5 Extended perforations

Deg

ree

of

corr

osi

on

6

5

4

3

2

1

0–600 –500 –400 –300 –200 –100 0

U(mVCSE)

10.1 Correlation between potential and degree of corrosion of theduct of prestressing cables.



10.3 Repair

Based on the results of the detailed inspection of the bridge, the possibilityof a conventional repair with removal of the contaminated concrete, in someplaces to behind the first layer of prestressing cables, was studied initially.However, the considerations concerning the statics of the structure showedclearly that this would not be possible without detrimental effects on the loadbearing behaviour of the bridge. Furthermore, the application of other methods,e.g. cathodic protection, would have been possible only with restrictions andconsiderable risks.

Therefore, it was a question of what would be the remaining service lifeof the bridge without an ordinary repair. It was decided that a monitoringsystem should provide the necessary information about whether and to whatextent sealing of the bridge deck and the repair of the drainage system wouldlead to drying out of the concrete and, therefore, to a reduction of the corrosionvelocity. The aim of the test programme was, within three years, to collectinformation to allow an estimation of the remaining service life and/or toshow the need for further short-term repair measures.

10.4 Monitoring system

10.4.1 Concept

The monitoring system concept depended on the observation of changes incorrosion velocity based primarily on the measurement of macrocell currentsbetween isolated rebars and the reinforcement of the bridge as well as on themeasurement of the humidity-dependent concrete resistance which is inverselyproportional to the corrosion velocity [2].

This concept, which was realised in this way for the first time in Switzerland,consisted of the following investigations:

∑ Measurements on isolated rebars (resistance, current and potential differencebetween isolated rebars and reinforcement) every 6 months to giveinformation on the corrosion velocity;

∑ Repeated potential measurements in 25 test fields at the end of thesurvey period to give information on the corrosion behaviour of rebars(active/passive);

∑ Visual inspection of isolated rebars (corroded area) at the end of thesurvey period to give information on the current density and the corrosionrate;

∑ Measurement of specific concrete resistance on drill cores in the laboratoryas a function of relative humidity and chloride content to give informationconcerning the resistance of concrete during the drying phase.



10.4.2 Installation of test fields and laboratory tests ondrill cores

For the measurements described above, it was necessary to isolate rebarselectrically from the reinforcement of the bridge. The removal of drill coresat the crossing points of the vertical and horizontal reinforcement allowedthe isolation of rebars of the first and the second layer with a length of 100to 150 mm (Fig. 10.2). Electrical contact to the rebars was achieved bylocking a cable in a pre-drilled hole by means of a conical pin. The isolatedrebars were short-circuited to the general reinforcement during the whole ofthe investigation period by plugs placed in a measuring box within the boxgirder.

The short-circuit connections were opened only for measuring purposes.The contact areas of the cables and the cores were filled with an isolatingepoxy-mortar.

To allow temperature correction of the resistance measurements, some ofthe test fields were equipped with temperature probes for measuring thetemperature of the structure.

Aer

atio

n

tub

e

Bridge deck

–120 –200 –085Arh

BArv

T2

T1

–185

–190 –160 –010

–065

Girder

1.20

–1.

71

0.50 0.50

B Drill core (Ø 80 mm) for chloride analyses and laboratory testsF Drill hole (Ø 8 mm) for measurement of concrete humidityPt 100 Temperature sensor Pt 100 (Ø 8 mm)Arhv Isolated rebar, h: horizontal, v: verticalT1 Drill core for separation of the reinforcement Ø 80 mmT2 Drill core for separation of the reinforcement Ø 50 mm+–190 Potential of the reinforcement

10.2 Equipment for monitoring test fields.

T2

FO

–195Pt100



The laboratory tests were effected on drill cores of diameter 80 mm takenfrom the test fields. One half of the cores were used for chloride analyses insteps of 15 mm. The other half were equipped with copper pins so that thespecific resistance at different depths could be measured by means of a 4-electrode measurement. At the end of the test period the porosity andcarbonation were also investigated. The specific resistance was measured oncores which had been stored at five different relative humidities between 43and 95% and at temperatures of 15 to 20 ∞C, as well as after 30 days ofstorage under water.

10.5 Results of the monitoring

10.5.1 Field tests

Electric resistance of the concrete

The specific resistance of the concrete cannot be calculated exactly from themeasured resistance between isolated rebars and the reinforcement (spreadresistance of the isolated rebars). For a first approximation the formula forthe spread resistance of earthing rods can be used [3].

Figure 10.3 shows changes of the thus-calculated specific concrete resistanceduring the investigation period in some selected test fields. After an initialperiod all the curves show the same behaviour. The absolute values of theconcrete resistance vary, as expected, over a wide range. The strong temperature

6000

5000

4000

3000

2000

1000

0

r c (W

m)

TO1 TO2 SO21SW35

35

25

15

5

–5

–15

–25

T(∞C

)

Mar

95

Sep

95

Mar

96

Sep

96

Mar

97

Sep

97

Mar

98

Sep

98

Mar

99

Sep

99

Mar

00

Sep

00

Date

10.3 Concrete resistance as a function of time and temperature.

T



dependence of the concrete resistance is evident. Average resistance valuesmeasured in autumn (T = 1–8 ∞C) were about double those measured inspring and summer (T = 16–25 ∞C).

To allow a better assessment of the changes as a function of time, thespecific concrete resistance was temperature corrected according to equation10.1 [2].

r rc c,o

1 – 1

= o◊ÊË

ˆ¯e

bT T (10.1)

where rc,o is the specific electrical concrete resistance at temperature T[W m); T and To are temperature [K] (T between –25 and –40 ∞C); and b isa constant (K).

The constant b was determined for each test field by linear regressionfrom the measured values (To = 240 K). The calculated values for b arebetween 2700 and 5000 K. This procedure is an approximation, because thetemperature dependence is overlaid by the drying process. Furthermore, theconstant b is a function of humidity [4].

Figure 10.4 shows the specific concrete resistance as a function of time,normalised at a temperature of 20 ∞C and by the resistance values measured3 months (84 days) after the installation of the monitoring system.

Generally, the following observations were made:

∑ All spread resistances and, therefore, the specific concrete resistancesshowed a more or less continuous increase during the survey period. This

TO1TO2SO21SW35

Mar

95

Sep

95

Mar

96

Sep

96

Mar

97

Sep

97

Mar

98

Sep

98

Mar

99

Sep

99

Mar

00

Sep

00

Date

r c/r

c,84

6

5

4

3

2

1

0

10.4 Concrete resistance as a function of time (normalised to 20 ∞C).



indicates a decrease in concrete humidity. In most of the test fields,5 years after the sealing of the bridge deck, there is still a tendency for afurther increase of the concrete resistance.

∑ The increase of the concrete resistance shows big variations in the differentfields. The average increase was 240% for fields inside the box girder andabout 90% outside. In both cases 15% of the test fields showed an increaseof less than 40% outside and less than 100% inside the box girder.

∑ The spread resistance of the isolated rebars of the first and the secondreinforcement layer showed only small differences.

Macrocell currents

Figure 10.5 shows the macrocell current as a function of time for differenttest fields. Anodic (in the diagrams negative) currents led to corrosion on theisolated rebars with a weight loss of 10 mg mA–1 per year.

The following observations were made:

∑ In test fields with anodic currents the chloride content at the depth of therebar was generally above 1% (average value 1.2%) whereas it was lowerthan 0.73% in test fields with cathodic currents.

∑ The macrocell currents show a temperature dependence which is muchstronger than those for the resistance. At the end of the survey period,current and temperature were logged every 10 min in two test fields for1 month. With these data a temperature correction analogue of equation

TO1TO3SO5SW27T

30

15

0

–15

–30

–45

–60

–75

–90

–105

–120

Temp

erature ( ∞C

)IS

O27 (m

A)

Mar

95

Sep

95

Mar

96

Sep

96

Mar

97

Sep

97

Mar

98

Sep

98

Mar

99

Sep

99

Mar

00

Sep

00

Mar

01

Sep

01

Date

I (m

A)

10

5

0

–5

–10

–15

–20

–25

–30

–35

–40

10.5 Macrocell currents.



10.1 was applied (Fig. 10.6). There was a good correlation but the ‘b-values’ calculated were 30 to 90% higher than those calculated for thetemperature dependence of the resistance.

∑ The highest measured macrocell current was 94 mA (T = 20 ∞C). In thecase of homogeneous corrosion of the whole surface of the isolated rebarthis would correspond to a corrosion rate of only 0.04 mm year–1. On theother hand, if the corroding area were restricted to a length of 1 cm thecorrosion rate would reach 1.1 mm year–1.

∑ Within 6 years after the sealing of the bridge deck, all macrocell currentsdecreased considerably. However, in several test fields no noticeable changein corrosion velocity was observed during the first three years. After twoyears, in one test field an initially cathodic isolated rebar became anodicfor about 1 year (Fig. 10.5). Note that the macrocell currents in Fig. 10.5are not temperature corrected.

∑ The absolute values of macrocell current increase with decreasing potential,as shown in Fig. 10.7. A useful correlation between the two parameterswas reached only when the potential was measured directly above theanodic isolated rebar.

∑ At the end of the investigation period, changes in macrocell current werecaused in about equal amounts by changes in the concrete resistance andby variations of the potential difference between the isolated rebars andthe reinforcement (Table 10.2).

I (mA

)

45

40

35

30

25

20

15

10285 290 295 300

T(K)

10.6 Correlation between macrocell current and temperature.



19982000

–500 –400 –300 –200 –100 0 100U(mVCSE)

I( mA

)

20

0

–20

–40

–60

–80

–100

10.7 Correlation between macrocell current and potential.

Table 10.2 Changes in concrete resistance, macrocell current and potentialdifference between isolated rebars and reinforcement in selected time intervals(e end of time interval, b beginning of time interval)

Test field DUe/ DUb 1/(Re/Rb) (DUe/ DUb)/(Re/Rb) Ie/Ib(measured)

TO1 0.19 0.55 0.11 0.11TO1 0.36 0.90 0.34 0.33TO2 0.96 0.86 0.89 0.78TO2 0.84 0.88 0.74 0.84TO5 1.03 0.79 0.81 0.80SO21 0.31 0.59 0.18 0.10SW50 0.56 0.60 0.38 0.48SW50 1.40 1.47 2.06 2.64

Potential measurements

The potential measurements were repeated in the test fields after 3 and 5years and the following observations were made:

∑ Generally the potentials increased clearly during the observation period(Fig. 10.8), but in some cases quite negative potentials were measured atthe end of the test period in restricted areas. Unfortunately, prestressingcables were located in these areas.

∑ The extent of the potential shift toward more positive values showed largevariations between the different test fields. There was a tendency for thepotentials in the different test fields to approach each other with time.



∑ Statistical analysis of the potential measurements led to the conclusionthat, in most of the test fields, the rebars were still corroding. This meansthat the corrosion velocity decreased but corrosion did not stop.

Visual inspection

At the end of the investigation period, some of the isolated rebars wereremoved in order to determine the corroded area. This area was between 2and 6 cm2, which corresponds to localisation factors (ratio of total surface ofthe isolated rebar/corroding surface) from about 7 to 15.

10.5.2 Laboratory tests

The most important observations can be summarised as follows:

∑ The electrical resistance of concrete cores stored at relative humidities ofmore than 60% reached more or less constant values during the investigationperiod, whereas cores stored at relative humidities £ 60% showed a tendencyfor a further increase of resistance even after 940 days (Fig. 10.9).

∑ The concrete resistance decreases with increasing depth. The resistanceof the layer near the surface (0–15 mm) is generally considerably higher.This effect is due to carbonation.

∑ In alkaline concrete, the specific concrete resistance does not correlatewith the chloride content of the concrete. The fact that there is no correlation

U20

00 (m

VC

SE)

200

100

0

–100

–200

–300

–400

–500

–600

OutsideInside

–600 –400 –200 0 200U1994 (mVCSE)

10.8 Potentials at the beginning and at the end of the investigationperiod.



between chloride contamination and (spread) resistance was also observedin the field tests. Otherwise, it seems that, in carbonated concrete, chloridesinfluence the specific resistance (based only on a few results).

∑ The specific concrete resistance increases with decreasing relative humidity(Fig. 10.10). At humidities below about 70% a marked increase in resistanceis observed.

∑ The resistance of the cores from different test fields at a given relativehumidity shows a variation of about a factor of 3 (Fig. 10.10). This can beinterpreted as an indication of locally different concrete qualities.

10.6 Corrosion velocity and further service of the

bridge

One of the most difficult problems to solve in monitoring is the estimationof the absolute corrosion velocity. On the one hand, there is informationgathered in inspection fields and, on the other hand, results from macrocellmeasurements or even galvanostatic pulse measurements. In the first case,the cross-sectional loss of the rebars is known, but not the beginning of thecorrosion process; in the second case, the corrosion current is known, but notthe corroding area.

0 200 400 600 800 1000Time (d)

TO 2,45% TW 1,45%SO 21,60% SO 26,60%T

r c(W

m)

25000

20000

15000

10000

5000

0

25.0

20.0

15.0

10.0

5.0

0.0

Temp

erature ( ∞C

)

10.9 Specific concrete resistance of drill cores as a function of time.



10.10 Specific concrete resistance of drill cores as a function ofrelative humidity.

40 50 60 70 80 90 100Relative humidity (%)

r c(W

m)

30000

25000

20000

15000

10000

5000

0

AvgMinMax

In the present case, the estimation of the maximum corrosion velocitybefore sealing led to values of between 0.5 mm year–1 (visual inspection,cross sectional loss) and 0.75 mm year–1 (macrocell currents in combinationwith localisation factors and the yearly temperature variations). The repeatedpotential measurements, together with the correlation between macrocellcurrent and potential, allowed a good estimation of the actual corrosion rate,which was a maximum of 0.3 mm year–1, to be made at the end of theinvestigation.

The assessment of the actual corrosion risk and the estimation of theremaining service life of the bridge has to be done based on structuralconsiderations and corrosion models for the progress of corrosion with time,above all in the prestressing cables. A first worst case scenario based onsimple corrosion models and considering the actual corrosion velocities ledto the conclusion that there is no actual or short time risk for the load-bearingbehaviour of the bridge. For better estimation of the remaining service life,more sophisticated corrosion models will be developed.

An important point for the secure service of the bridge is the continuationof the monitoring. It is essential to guarantee that the humidity of the concretedoes not increase again (e.g. as a consequence of new leakages).

10.7 Conclusions

∑ The monitoring system that has been described has allowed estimates ofthe corrosion velocity in a way that, in combination with structural



considerations and corrosion models, provides a final assessment of theremaining service life of the bridge.

∑ Within five years, the corrosion velocity decreased as a consequence ofthe drying out of the concrete in all test fields (with chloride concentrationsnear the rebar of up to 1.9%) by a minimum of 50%, but, in general, thecorroding areas remained active.

∑ The drying out of the concrete is a slow process and is not complete afterfive years. Water infiltration (even over short periods) which leads to anincrease of the humidity of the concrete must, therefore, be avoided infuture.

∑ The corrosion velocity is not determined solely by the specific resistanceof the concrete. The different temperature dependencies of resistanceand macrocell current, as well as the fact that the potential differencebetween the reinforcement of the bridge and isolated rebars influencesthe corrosion velocity, are clear indications that electrochemical factorsand, particularly, their change with time, determine the corrosion velocity.

∑ The chloride contamination (concentration) has no measurable influenceon the resistance of alkaline concrete.

∑ The strong temperature dependence of the specific concrete resistanceand the corrosion velocity have to be taken into account in the design ofmonitoring systems. Regular, periodic measurements and, preferably, acontinuous data acquisition system are necessary. Isolated single measuringpoints do not often lead to reliable information.

∑ The time dependence of the concrete resistance and the corrosion velocityhas shown large variations along the investigated bridge. For reliable,meaningful monitoring sufficient measuring points or test fields arerequired, which leads to substantial effort.

10.8 References

1. D. Bindschedler and F. Hunkeler, Schweiz Ing Architekt, 1997, 115, 374.2. F. Hunkeler, Grundlagen der Korrosion und der Potentialmessung bei

Stahlbetonbauwerken, EVED/ASB, VSS-Bericht Nr. 510, 1994.3. W. von Baeckmann, Taschenbuch für den kathodischen Korrosionsschutz, Vulkan-

Verlag, Essen, 1987, 254.4. Y. Schiegg, L. Audergon, B. Elsener and H. Böhni, Online-monitoring of the corrosion

in reinforced concrete structures, Eurocorr 2001.


133

11.1 Introduction

In the last 10 to 15 years many reinforced concrete constructions, which hadsuffered damage by corrosion of the reinforcement, were repaired usingvarious procedures (local repair, sprayed concrete, coating, or electrochemicalprocedures). Until now, much priority has been given to methods of conditionassessment for reinforced concrete structures. The monitoring of concreteconstructions (repaired or new structures) is a new approach with only limitedexperimental data available.

In the case of unalloyed rebars, the time to initiate local corrosion attackand the stable propagation of the corrosion process is primarily influencedby environmental factors such as chloride concentration, pH, oxygen contentof the pore solution or the porosity of the concrete. Since these parameterscontinuously change as a function of time, the exposure condition of a concretestructure is of decisive importance for the prognosis of the service life andthe durability of reinforced concrete constructions. The long-term monitoringof important corrosion variables will give us a more precise understanding ofthe actual processes occurring during the corrosion of reinforced concreteand will lead to better knowledge for the evaluation of repair methods andfor the project engineering of new constructions.

The term ‘on-line monitoring’ means the continuous measurement of certainparameters for real structures, where sensor devices are built into the concrete.By the use of short measuring intervals (e.g. 10 min) repeated in a systematicway, both temporary changes and long-term differences can be measured.

11.2 Instrumentation

11.2.1 Instrumented cores and measurements

In order to measure the moisture content and chloride uptake as a functionof the concrete depth, up to eight chloride and resistivity sensors [1] are cast

11On-line monitoring of corrosion in

reinforced concrete structures

Y. S C H I E G G, L . A U D E R G O N and H. B Ö H N I,ETH Zürich, Switzerland and B. E L S E N E R,

ETH Zürich and University of Cagliari, Italy



into cylindrical concrete samples (Fig. 11.1). The cell factor of the resistivitysensors is first determined in the laboratory. The samples are drill corestaken from real structures or laboratory samples. A further core contains areinforcing bar (length 40 mm, Ø 8 mm) for corrosion current measurements,a chloride sensor as well as three PT1000 temperature sensors. The reinforcingbar can be removed at a later point in time for further investigations. Thelateral surfaces of the cores with the eight sensor elements are coated with anepoxy resin coating, in order to permit only one-dimensional water uptakeperpendicular to the concrete surface. The instrumented cores are then fixedwith a low viscosity mortar into boreholes at different locations of the concretestructure. The potentials of the chloride sensors are measured versus a MnO2-reference electrode embedded in the surrounding concrete [2].

With the instrumented drill cores and additional sensors for the measurementof climatic influences, the following parameters are obtained:

∑ Electrical concrete resistance, RW (W)∑ Potential of the reinforcement, chloride sensor, U [V (MnO2)]∑ Corrosion current, Icorr (A)∑ Air and concrete temperature, TA/C (∞C)∑ Relative humidity, rh (%)

205mm

12.520 27.5 35

42.550

R.1

R2 R3R4

R5

R6

R757.5

2.5100

Ø75

T115

35

60T2

T3

20

100

Ø75

Chloride andresistivity sensors

Rebar probe andtemperaturesensors

40 mm

2 5 4 6 3 1

1 = Silver wire Ø0.5 mm 4 = Stainless steel tube Ø2.5 mm2 = AgCl-coating 5 = Epoxy sealing3 = Teflon tube (insulation) 6 = Insulation

11.1 Instrumented cores for field tests, chloride and resistivity sensorand embedded cores in an edge beam of a bridge.

120 mm

Ep

oxy

-co

atin

g


On-line monitoring of corrosion in reinforced concrete structures 135

11.2.2 Data acquisition system

The data acquisition system, developed particularly for on-line monitoring,contains four different measuring modules (potential, current, resistance andan auxiliary module). These measuring modules are protected, by means ofa protection module, against overvoltages (e.g. lightning impact). Altogether32 potentials, 8 currents, 24 resistances and 5 auxiliary variables can bemeasured. The recording interval can be flexibly selected in minute intervals.There is the possibility of connecting together up to 16 data collection systemsby a RS485-network. The control and the current supply (12 V-battery) ofthe devices take place from a central, well accessible place. The selected dataare stored on the pc in a text file and can be further processed with normalstatistical software.

11.3 Field tests

In 1998, reinforced concrete structures on the national highway A13 in thecanton of Grisons were equipped with instrumented cores and data acquisitionsystems for on-line monitoring. A number of typical exposure conditionsand structures were chosen, where different weathering and corrosionpropagation characteristics were to be expected (Table 11.1). In order to beable to examine the influence of porosity, an instrumented core from thestructure and mortar cores from the laboratory (max. grain size ø 4 mm,w/c = 0.5/0.6) were used.

11.4 Modelling of the temperature dependence of

RW and Icorr

Concrete resistances and corrosion currents are strongly influenced bytemperature (as described in section 11.5). In order to describe the moisturecontent of the concrete and to be able to detect trends for the evolution of thecorrosion currents, it is necessary to compensate the temperature dependence

Table 11.1 Concrete structures and exposure conditions for the field tests

Structure Exposure

Bridge deck Beyond an asphalt layerEdge beam Direct weathering / splash waterPile Partially wetArch Direct weatheringUnderside bridge deck Indirect weatheringAbutment Partially wetGallery Splash water, indirect weatheringMain girder Indirect weathering



of these measured variables. The exponential relation between the electricalconcrete resistance and temperature (equation 11.1) was derived from theArrhenius equation. In the literature values, between approx. 1500 and 5000 Kfor the temperature coefficient b and between 2000 and 7000 K for thetemperature coefficients a have been reported [3, 4].

R RT T

b 1T

– 1T

1 01 0 = e◊

ÊË

ˆ¯

I IT T

–a 1T

– 1T

1 01 0 = e◊

ÊË

ˆ¯ (11.1)

where RT1 and IT1 are concrete resistance and corrosion current at temperatureT1 (K); R IT T0 0 and are concrete resistance and corrosion current at temperatureT0 [K] and b and a are temperature coefficients (K).

Since the b-value is dependent on the concrete humidity [5], the equation(11.1) may be applied only for short time periods, in which the humiditydoes not change or only slightly changes. The analysis of the performed fieldtests, where a measuring interval of 10 min was selected, indicated that thedetermination of a b-value is optimal over 24 h (day and night). As Fig. 11.2

concrete resistance are performed at the same depth from the surface in orderto obtain an exact determination of the b-value. It should be noted that thetemperature gradient is not constant in a concrete structure and, because ofthe good heat conductivity of the concrete, gradient changes take place veryfast. The heating and cooling rates of the concrete near the surface aredifferent (different time constants).

In (

RB)

(W)

9.6

9.5

9.4

9.3

9.2

9.1

9

R(5-12.5 mm), T(calc)

R(5-12.5 mm), T(60 mm)

b = 3250, R2 = 0.98

b = 4312, R2 = 0.78

Ponte Nanin, arch

0.0034 0.00342 0.00344 0.00346 0.00348 0.0035 0.00352 0.003541/T (K–1)

11.2 Determination of the b-value for two different temperatures withthe Ordinary Least Squares method. With an unfavourabletemperature, hysterisis loops and wrong b-values result.


and 11.3 show, it is essential that measurements of the temperature and the


Figure 11.4 shows the result of the temperature compensation (T0 = 293 K)of the concrete resistivity (depth 20–27.5 mm) in the abutment of a highwaybridge in a Swiss alpine region. The daily fluctuations of the temperaturecould practically be eliminated. Because of the constant concrete resistivity,it can be concluded that in the selected time period, January to July 1999, nodeep-going humidity changes as a result of water absorption or drying processescould be measured. This is a meaningful result because this structure isexposed to indirect weathering and, therefore, the transport zone (distancefrom the surface where humidity changes can clearly be observed) is onlysmall.

Ponte Nanin, arch

R(5-12.5 mm) 8h50

15h10

T(calc)T(15 mm)T(35 mm)T(60 mm)

9.4

9.2

9

8.8

8.6

8.4

In (R

a ) (W)

23h50

15h10

8h500h00

T (∞C

)20

18

16

14

12

10

8Mar/10 Mar/10 Mar/10 Mar/10 Mar/11

Time

11.3 Concrete resistance and concrete temperatures over time incase of cooling and heating the concrete. T(calc): calculatedtemperature for the depth of 8.75 mm.

Malabarba, abutment mortar w/c 0.5Tcalc 20-27.5 mm

RB 20-27.5 mmRB T-compensated

294

280

266

252

238

Co

ncrete tem

peratu

re (K)

Co

ncr

ete

resi

stiv

ity

(W m

) 3000

2500

2000

1500

1000

500

0Jan/1 Jan/24 Feb/16 Mar/11 Apr/3 Apr/26 May/19 Jun/12 Jul/5

Date

11.4 Concrete resistivity RB (raw data and with temperaturecompensation, T0 = 293 K) on a depth of 20–27.5 mm and calculatedconcrete temperature over time in the abutment of a Swiss highwaybridge, January to July 1999.




11.5.1 Corrosion current and temperature

Figure 11.5 shows the corrosion current of a corroding rebar (chloride content1.2 mass %/c) and the concrete temperature over time in the edge beam ofthe Nanin bridge between July 1998 and September 2000. There is a pronouncedtemperature dependence of the corrosion current (daily peaks and seasonaldifferences). In this example, the corrosion current differs between winterand summer by a factor 3.5 to 4. Within the daily fluctuations and theseasonal fluctuations, the increase of the corrosion current during a rise intemperature reaches a factor of 1.8 to 2.0 per 10 ∞C. The corrosion processesare also not stopped even at temperatures below 0 ∞C.

11.5.2 Concrete resistance and humidity

An important parameter of the corrosion is the humidity of the concrete. Theconcrete resistances measured by means of ac resistance measurement provideinformation about humidity changes in the zone near the surface of theconcrete. Figure 11.6 shows the concrete resistance over time at three depthlevels in the edge beam of the Nanin bridge, the concrete temperature andthe amount of rain for the selected time period. Like the corrosion current,the concrete resistance also shows a pronounced temperature dependence(2.5 kW per 10 ∞C). After a drying period lasting until 31 December 1998,precipitation caused a strong decrease of the concrete resistance 5–12.5 mm(approx. factor 3) in the following days, whereby the gradient of the resistance

Jan/6 Sep/30 Jan/24 May/19 Sep/12 Jan/6 May/1 Aug/24Date

Ponte Nanin, edge beam

T (concrete, 15 mm)

Co

rro

sio

n c

urr

ent

(mA

)

0.3

0.25

0.2

0.15

0.1

0.05

0

Icorr

40

30

20

10

0

–10

Temp

erature ( ∞C

)

11.5 Corrosion current of a corroding rebar in the edge beam of ahighway bridge (Nanin bridge) and the concrete temperature overtime, July 1998 until September 2000.



decreases with increasing concrete cover. Resistance measurements showthat it is possible to differentiate clearly between wet and dry periods.

11.5.3 Determination of b-values and temperaturecompensation of the concrete resistance

The b-values over 24 h (144 measured values) for the temperature compensationof the concrete resistances were extrapolated with an Ordinary Least Squares-method. To evaluate the quality of the model the correlation factor R2 (0 £ R2

£ 1) was used, whereby high values of R2 refer to a high quality of the model.The R2 serves also as a filter criterion. With the condition R2 > 0.98 it ispossible to eliminate false b-values, which can occur e.g. during a period ofrain precipitation occurring on the concrete structure. Figure 11.7 shows fora mortar with a w/c-ratio of 0.5 in the edge beam of the Nanin bridge thecalculated b-values, the value of R2, the amount of rain and the temperaturecompensated concrete resistances (reference temperature 20 ∞C). The b-values are situated between 2000 and 4000 K and are relatively constant. Thestrongest fluctuations occur within the first depth level (5–12.5 mm), whichis the result of the changing concrete humidity due to direct weathering. In

Amount of rainC

on

cret

e re

sist

ance

(W

)

10000

8000

6000

4000

2000

10000

8000

6000

4000

2000

20

10

0300

280

260

R (m

m) T

con

crete (K)Depth 9 mm

5–12.5 mm12.5–20 mm20–27.5 mm

27.5–36 mmPonte Nanin-edge

beam mortar w/c 0.6

12/16/98 12/21/98 12/26/98 12/31/99 1/5/99 1/10/99Date

11.6 Concrete resistances over time (depth: 5 to 35 mm) of a mortarcore (w/c = 0.6) in the edge beam of a highway bridge (Naninbridge), concrete temperature and the amount of rain over time (datafrom Swissmeteo), Dec. 1998 to Jan. 1999.



most cases, R2 is clearly above 0.98. Because of the small time fluctuations,only one b-value (average value over 2 years) for each depth level was usedfor the calculation of the temperature-compensated resistances. It is noticeablethat, apart from the initial phase, strong fluctuations of the concrete humidityoccur mainly in the proximity of the surface. However, the transport of waterand pollutants (e.g. chlorides) to larger depths within the concrete structuretakes place about twice a year.

5-12.520-27.5

5–12.520–27.5

b-value

Correlation factor

b v

alu

e (K

)R

2

5000

4000

3000

2000

10001.00

0.99

0.98

0.97

0.96

0.95

Amount of rain

5–12.512.5–2020–27.5

Precipitation

Resistivity

Ponte Nanin–edge beam

1/1/99 7/1/99 1/1/00 7/1/00 1/1/01 7/1/01Date

r (m

m)

Co

ncr

ete

resi

stiv

ity

(W

m)

80

60

40

20

0

500

400

300

200

100

0

11.7 Time development of the b-values, correlation factor R2, amountof rain and temperature-compensated concrete resistances of amortar core w/c = 0.5 in the edge beam of the Nanin bridge. To makethe graph clear, only 10% of the data are plotted in the upper twodiagrams. Data were recorded over 3 years.



For the concrete resistivity, at a depth of 5–12.5 mm, there is a cleardifference of more than 1000 K between the b-values before and after astrong precipitation of rain (Fig. 11.8). In the depth range 20–27.5 mm thedifference between the b-values is smaller (400 K) because the concretehumidity before the precipitation was higher than near the surface (sincedrying out of the concrete occurs from the outside of the structure).

11.5.4 Temperature dependence of thecorrosion current

The modelling of the temperature-dependence of the corrosion current wasexecuted in the same way as for the concrete resistances. The a-value inrelation to the corrosion current is calculated instead of the b-value (Fig.11.9). The comparison with Fig. 11.7 shows that a-values are situated in thesame range as the b-values. In Fig. 11.10, the a- and b-values of differentconcrete structures with different exposure conditions are represented. Thea-values are situated between 3000 and 5000 K. In many cases, the a-valuescorrespond relatively well with the b-values. In particular, for mortar coreswith a w/c = 0.6 ratio, a clear increase in the a-value is to be determined forconcrete structures with indirect weathering. This suggests that the a-value,similarly to the b-value, is humidity dependent. With the comparison between

5–12.5 mm before precipitation20–27.5 mm before precipitation5–12.5 mm after precipitation20–27.5 mm after precipitation

Ponte Nanin, edge beam

Co

ncr

ete

resi

stiv

ity

(W m

)

1200

1000

800

600

400

200

0

b = 3863 K

b = 3704 K

b = 3294 K

b = 2774 K

275 280 285 290 295 300Temperature (K)

11.8 Concrete resistivity and b-values vs. concrete temperaturebefore (March 2000) and after (April 2000) a strong precipitation ofrain, edge beam of the Nanin bridge.



a and b, it must be considered that the corrosion current flow takes placeboth over the mortar and over the concrete of the structure (macrocell corrosion),while the concrete resistance is measured only within the core because of thecoating on the lateral surfaces. This can entail, depending upon the quality ofthe concrete environment, stronger fluctuations of the a-value. The largerdispersion of the a-value is also probably to be attributed to the fact that thecorrosion current is limited not only by the concrete resistance but also bythe electrochemical resistances at the anode and at the cathode.

Jäggi found, in his laboratory tests with mortar blocks, b-values fromapprox. 2250 to 3530 K and a-values between 4000 and 4400 K, wherebythe stronger temperature dependence of the corrosion current in relation tothe concrete resistance became clearly recognisable only at temperaturesover 30 ∞C [6]. In these investigations the concrete resistance had only asubordinate influence on the temperature behaviour of the macrocell corrosion.Because of the results in Fig. 11.10, no clear conclusions can be drawn overwhich resistance controls the corrosion current. In addition, further analysisof the data must be done, where the voltage drop related to the corrosioncurrent and the resistance between the corroding rebar and the cathodicreinforcement is compared with the calculated potential difference betweenthe anode and the cathode (approximately 400 mV). It is to be expected thatthe proportions of ohmic and cathodic control vary with the humidity of theconcrete.

11.5.5 Corrosion rate and propagation

To allow prediction of future developments and evaluation of the expectedcross-sectional losses at the rebars, it is necessary to obtain informationabout the corrosion rate from differently exposed concrete structures. For

I co

rr (

mA

)

0.2

0.15

0.1

0.05

0

a va

lue

(K)

5000

4000

3000

20001000

Jun/16 Aug/31 Nov/16 Feb/1 Apr/18 Jul/4 Sep/19 Dec/4 Feb/19 May/6 Jul/21 Oct/6

Edge beam, Ponte Nanin

Edge beam, Ponte Nanin

Jun/16 Aug/31 Nov/16 Feb/1 Apr/18 Jul/4 Sep/19 Dec/4 Feb/19 May/6 Jul/21 Oct/6

11.9 Corrosion current over time (Icorr ≥ 10 mA) in the edge beam ofthe Nanin bridge and related a-values, R 2 ≥ 0.975.

Date

Date



this purpose, corroding rebars were removed from various constructions andthe active surface was determined visually after cleaning. The mass loss andloss in cross section were calculated on the basis of the measured corrosioncurrents using Faraday’s law (equation 11.2).

G MzF

I tt

= S D (11.2)

b (mortar w/c = 0.5)

a (mortar w/c = 0.5)

Direct weathering

Indirectweathering

a, b

val

ue

(K)

6000

5000

4000

3000

2000

1000

b (mortar w/c = 0.6)

a (mortar w/c = 0.6)

Direct weathering

Indirectweathering

a, b

val

ue

(K)

6000

5000

4000

3000

2000

1000

Structure

Bri

dg

e d

eck

1

Bri

dg

e d

eck

2

Ed

ge

bea

m 1

Ed

ge

bea

m 2

Arc

h

Ab

utm

ent

Un

der

sid

eb

rid

ge

dec

k

Pile

Th

rust

Structure

Bri

dg

e d

eck

1

Bri

dg

e d

eck

2

Ed

ge

bea

m 1

Ed

ge

bea

m 2

Arc

h

Ab

utm

ent

Un

der

sid

e b

rid

ge

dec

k(f

irst

2 m

on

ths)

Pile

Th

rust

11.10 Mean value and standard deviation of the a- and b-values forvarious concrete structures, exposure conditions and mortar types;measured values over a duration of approx. 2 years were used.



where G is mass loss (g), M is atomic mass (g mol–1), z is valency (Fe = 2),F is the Faraday constant (A s mol-1), I is the corrosion current (A), and t istime (s)

Figure 11.11 shows the increase of the cross-sectional loss of differentlyexposed structures over time. While the cross-sectional losses increase rapidlyin the edge beam and in the gallery column, the reinforcing bars in theunderside of the bridge deck corrode only slowly (icorr < 0.01 mm year–1).The propagation curves show clearly that the exposure conditions play adecisive role in the increase of the cross-sectional loss because the chloridecontent in each structure is higher than 1.0 mass %/c. The maximum corrosionrate in the edge beam measured during the summer months (high temperaturesand humidity) was approximately 0.4 mm year–1 (0.6 mm year–1 in an edgebeam of another bridge). In winter, the corrosion rate was about two timessmaller. From experience, these corrosion rates are realistic. In Switzerland,bridges that are 25 to 30 years old often show large areas with corrosionattack and high rates of corrosion have led to a decrease of about 10 mm ofthe rebar diameters. This means that in many cases the propagation periodwas much longer than the initiation period and the activation of the corrosionprocesses must have started only a few years after completion of the concreteconstruction.

11.11 Corrosion propagation over time of corroding rebar probes invariously exposed concrete structures indirect weathering (XD1)spray, partially wet (XD2) splash water (XD3) direct weathering(XD4).

Loss

in

cro

ss s

ecti

on

(m

m)

0.8

0.6

0.4

0.2

0

Underpass, abutment portal zoneBridge, underside bridge deckBridge, edge beamGallery, column

0.4 mm year–1

Summer

0.2 mm year–1

Winter

XD4XD3

XD2

XD1 <0.01 mm year–1

Nov/19 Jul/5 Feb/18 Oct/3 May/19Date



11.6 Conclusions

With the development and use of sensor-instrumented cores, as well as adata acquisition system for the continuous measuring of corrosion-relevantvariables, a new measuring technique is available for monitoring the statusof reinforced concrete constructions. The results of field tests at differentconcrete structures permit the following conclusions to be drawn:

∑ The monitoring concept of instrumented cores is well suited to practicalapplication and provides meaningful results.

∑ Concrete resistance and corrosion current are considerably influenced bytemperature. For high quality modelling of the temperature dependenceof these two parameters, characterisation of the temperature (by measurementor calculation) at the exact depth from the surface is necessary.

∑ The temperature coefficients b (concrete resistance) and a (corrosion current)are dependent on the moisture content in the concrete and are mainlydetermined by the exposure conditions of the concrete structures.

∑ At bridge structures with direct weathering and splash water, corrosion ratesup to approximately 0.4 to 0.6 mm year–1 are to be expected, while structureswith indirect weathering clearly corrode more slowly (< 0.01 mm year–1).Therefore, the exposure conditions and, to a lesser extent, the chloridecontent in the concrete are decisive in controlling corrosion propagation.


The authors are pleased to acknowledge the Swiss Federal Highway Authorities(ASTRA) and the canton of Grisons for supporting this research.

11.8 References

1. B. Elsener, L. Zimmermann, D. Flückiger, D. Bürchler and H. Böhni, ‘Non-destructivedetermination of the free chloride content in mortar and concrete’, Proc. RILEM Int.Workshop Chloride Penetration in Concrete, 1997, 17–26.

2. H. Arup and B. Sørensen, ‘A new embeddable reference electrode for use in concrete’,Corrosion 92, NACE Paper No. 208, Houston, 1992.

3. W. Elkey and E. J. Sellevold, ‘Electrical resistivity of concrete’, Norwegian RoadResearch Laboratory, Publ. No. 80, 1995, 1–35.

4. P. Schiessl and M. Raupach, ‘Influence of temperature on the corrosion rate of steelin concrete containing chlorides’, in Reinforced Concrete Materials in Hot Climates,Vol. 2, United Arab Emirates University, 1994, 537–549.

5. D. Bürchler, B. Elsener and H. Böhni, ‘Electrical resistivity and dielectric propertiesof hardened cement paste and mortar’, Materials Research Society, Electrically BasedMicrostructural Characterisation, Symposium Proceedings, Vol. 411, Boston, 1995,407–412.

6. S. Jäggi, ‘Experimentelle und numerische Modellierung der lokalen Korrosion vonStahl in Beton unter besonderer Berücksichtigung der Temperaturabhängigkeit’, PhDThesis No. 14058, Zürich, 2001.


146

12.1 The task

During a project dealing with electrochemical chloride extraction fromreinforced concrete one requirement was to design a generally usable, detailedand precise working system for describing and evaluating the state of corrosionin reinforced concrete structures. This should deliver, reliably, decision criteriaabout the state of corrosion and contain all of the data needed to configurea chloride extraction treatment. The required hardware should mostly consistof a minimal, commercially available configuration, added only by specificsensors, and the software surface should be able to manage all actions suchas data acquisition, data storage, data evaluation and data export.

Another requirement was to design both the hardware and software insuch a way that a single user would be able to undertake all of the fieldoperations (mainly data acquisition) conveniently and rapidly. In an extendeddesign of the system, contractors (operators), clients and consultants shouldall be allowed personal access to the data to allow better interaction betweenthe parties.

12.2 The solution

For receiving and storing all data in a common, reproducible way, any structureis to be divided logically into sections (e.g. an abutment, a column or abridge deck). Above the surface of such a section a grid overlay will bedrawn that splits the section into basic cells of 60 ¥ 60 cm. These cells willcontain the main data.

12Integrated system for corrosion monitoring

of reinforced concrete structures*

U. S C H N E C K, T. W I N K L E R and S. M U C K E, ConcreteImprovement Technologies, Germany

*The manuscript of this chapter was submitted originally for the EUROCORR 2001 andrepresents the state of knowledge and development in early 2001. Scientific and technologicalprogress since then has led to improved solutions.


Integrated system for corrosion monitoring 147

12.2.1 Data organisation

The appearance of the monitoring system follows the well-known explorationstructure: on the left side the main and sub-topics are listed and, on the rightside, the data from the selected topics are presented or can be edited. Aftersetting up a project file, which is usually named as the structure, the sectionscan be generated. There all the important time-independent data are to berecorded. Within the sections, the inspections can be added as needed, whereall the time-dependent data will be stored as shown in Table 12.1.

A database in the background organises all input values and allows thedata content to be extended e.g. for monitoring on-line sensors. Thedocumentation items allow the addition of any files that contain furtherinformation such as photos, drawings, other measurement data (e.g.galvanostatic pulse measurements).

Table 12.1 Data organisation of the monitoring system

Data layer Comment

Project File for structure

Globals General, non-time-dependent data

Data AdministrativeNotes General remarksDocu Photos, drawings, other

documents

Section 1..x Logical structural parts (tobe named)

Data Used materials, concreteand steel data

Rebar Scaled image of the toprebar layer

Docu Photos, drawings, otherdocuments

Inspection 1..y Time-dependent data of asection

Data Date, weather, circumstancesVisual Cracks, spalls, concrete

strength per cellPotential Potential mapping per cellHumidity Humidity values (weighed,

electrically)Chloride Five layers, content of

concrete and cementCarbonation DepthDocu Photos, drawings, other

documents



12.2.2 The interfaces

Some basic import and export functionality has been implemented whichmakes the interaction with several external devices possible. For dataacquisition, there are:

– import of *.csv (table structure) files and *.bmp (bitmap image) files fromthe HILTI Ferroscan 10 software [1] for creating the scaled rebar image andfor getting both the minimum and the average concrete cover of each cell;

– on-line potential mapping via PC Card on a notebook – with a connectedhalf-cell, potential readings can be taken directly into the monitoringsystem (one reading per cell);

– on-line humidity readings via PC Card on a notebook – similar to thepotential mapping, but with a connected 4-electrode sensor/signal generator(one reading per cell).

For continuing the surveillance into an electrochemical repair, all data areto be exported. The following related data currently will be used by a specialcontrol program.

The cell structure of a section including:– the rebar area [m2] which equals the cathodic electrode surface;– the minimum and the average concrete cover;– the coordinates of potential measurement and of selected mounting points.

As long as the internal evaluation of the data is under expansion, someimportant data, such as potential and humidity readings, can be exported forevaluation in spreadsheet programs.

12.2.3 Data input

The ‘Data’ subtopics for the global, sectional and inspectional items have aspreadsheet appearance and should be populated directly as marked. Otherspecial data inputs must be explained in more detail. Before being able toedit any further data, the grid system of a section must be generated. Adistinctive point of origin – any certain marker close to the section – can beused to give a reproducible geometric system. The cells are identified bytheir row and column numbers.

The rebar menu

Before editing a cell, the related data (*.csv and *.bmp) from the HILTIFerroscan must have been made accessible to the monitoring system. Theimported table of rebars will be simply orthogonal, so some orientationalcorrections may be necessary. For the purposes of comparison and adjustment,the HILTI scan bitmap can be displayed in the background (Fig. 12.1).



Automatically, the rebar area, its relation to the concrete surface, and bothminimum and average concrete cover are calculated. Furthermore, the locationof the potential mapping and the position of the electrode mounting in thecase of an electrochemical repair can be sketched in. If needed, this gives theopportunity to define a certain distance from the half cell to the rebars whichmakes the potential readings more readily comparable.

The measurement positions of the potential readings are kept and displayedin the potential mapping menu for each inspection. This means that the restpotentials can be obtained from exactly the same positions with an accuracyof +/–1 cm.

The visual menu

For every cell a conventional visual inspection can be included. Cracks,spalls and other features can be selected and sized with the local cell coordinates.

12.1 Rebar edit menu with table import file and bitmap display.



Each cell can contain two different cracks and three different spalls.Automatically, a sum of all cracks and spalls can be found in the inspectionsmain item and gives the total amount for a concrete repair. The value of theconcrete strength can be included as well (Fig. 12.2).

The damage codes used correspond with a paper by Browne and Pocockpresented at the Comett Course ‘The corrosion of steel in concrete’, 1992,which refers to ACI instructions (Table 12.2). An ‘Info’ card gives theopportunity to add specific information that does not fit into the main inputmask. A marker on the cell display points to such an additional information,so it will not be overseen.

The potential menu

Potential mapping values can be entered manually or automatically via thePC card of a notebook. There is no specific type of half cell required, but thiscan be set up as desired (Fig. 12.3).

The humidity menu

For a proper interpretation of potential mapping values it is essential toconsider the current humidity content of the concrete. The monitoring systemallows the insertion of humidity information in three ways:

12.2 Visual edit menu with input of cracks and spalls (example).


Integrated system for corrosion m

onitoring151

Table 12.2 Visual inspection codes – according to Browne and Pocock [2]

Code Feature Description Cause Details to be given

A1 Cracking (general) Jagged separations of concrete from Overload, corrosion, shrinkage Direction, widthno gap upwards

A2 Pattern cracking As cracking but formed as pattern Differential volume change between Surface area, widthinternal and external concrete

B1 Exudation Viscous gel-like material exuding ASR Severitythrough a pore

B2 Incrustation A crust (white) on the concrete Leaching of lime from cement Severity, dampnesssurface

B3 Rust stains Brown stains Corrosion of rebar Severity

B4 Dampness The extent of water on the surface Leakage, rundown Severity

C1 Pop-out Shallow, conical depression Local internal pressure i.e. expansion Area, depthof aggregate particle

C2 Spall A fragment detached from a larger Exertion of internal pressure i.e. by Area, depthmass rebar corrosion or external force

C3 Delamination A sheet spall Exertion of internal pressure over a Area, depthlarger area

C4 Weathering Loss of the concrete surface Environmental action wears away the Area, depthlaitance and paste

D1 Tearing Similar to cracks Adhesion to slipform shuttering Width, depth

D2 Honeycombing Voidage between coarse aggregates Lack of vibration Area, depth, severity



– estimated (dry – medium – wet)– measured by sampling and drying (mass %)– measured by electrical resistivity (on-line/off-line, W m)

According to the common evaluation of the related chance of corrosion,coloured icons will indicate this with ‘green – yellow – red’ standing for a‘low – indifferent – high’ chance of corrosion. For the on-line acquisition ofthe electrical surface resistivity a Wenner type electrode is needed. Its diameteris suitable for measurements up to 4–5 cm depth which usually representsthe steel environment. With the logged positions of potential mapping, theelectrode can be placed in the same position and gives a good correlationbetween both readings.

The chloride menu

Chloride contents can be inserted manually in five layers per cell, representinga layer thickness of 1 cm each. The input value is to be given in mass % (wt%) and is related to the sample mass (the concrete mass). With the knowncement content and concrete raw density – to be inserted in the section datamenu – the mass % related to the cement mass will be calculated and shownautomatically (Fig. 12.4).

12.3 Potential input menu (example).



The carbonation menu

Here the manually determined carbonation depths are to be inserted – oneper cell, given in millimetres. If they come close to the concrete cover value,this will be notified automatically.

12.3 The displays

The data displays are generated partly within the show-and-edit appearanceaccording to Table 12.1, and can be printed or will be available withinspecial evaluation menus. The basic previews and printouts are orientated onthe grid layout of a section.

12.3.1 Currently available displays

The global, sectional and inspectional data display

This gives a spreadsheet-like summary of its data and the data of the menusranking below, such as the cells with maximum and minimum chloride content,the sum of cracks and spalls per section and inspection. The number of cellswith additional information is notified as well.

The rebar display

This contains the numerical data in a table structure for each cell, including:the rebar surface; the average and the minimum concrete cover; the relativeposition of potential measurement and – for later use – the selected mountingpoint of a repair electrode. Furthermore, a scaled image of the outer rebarlayer can be generated. In this way, the true rebar layout can be viewed andcompared with the construction plans.

12.4 Chloride edit and input menu.



The visual display

Here the detected damage codes, the sum within the cell of cracks and spalls,the maximum depth of spalls and the concrete strength are shown.

The potential/humidity display

Both displays show the measurement readings and the location of measurementper cell, and, as a coloured icon, the case evaluation (good – intermediate –poor). A more intensive evaluation comes from separate displays.

The chloride/carbonation displays

In addition to the measured (and calculated) values, the displays give noticewith a coloured icon if the chloride content in the rebar depth is above 0.5 wt% cement and if the carbonation depth is in the range of the minimumconcrete cover per cell. For more information the info tab can be used.

12.3.2 Displays next-to-be integrated

Since it has been the first task to do the acquisition, the data organisation andcataloguing functions, the next steps for an integrated evaluation of all inputvalues will be, for instance:

– a graphic surface of the potential and the humidity mapping;– a coloured display to show the gradients between the neighbouring potential

readings that marks potential differences above 100 mV;– the sum frequency of the potential readings to find out the limits of the

critical, transient and non-critical potentials and to define the related colourborders;

– the change of potential, humidity, chloride or carbonation values of a rowor a column vs. the time of inspection;

– correlation of potential, humidity and chloride content per cell;– chloride profile per cell and per inspection or per section and depth.

12.4 Monitoring in use – the results

The opportunities given by the monitoring system will be shown for twohighway bridges in Saxony, Germany, where investigations for selecting areference object for the electrochemical chloride extraction have been made.From the six structures investigated, two shall be ‘discussed’ in more detail,using the monitoring system.



12.4.1 Neißebrücke Görlitz

The Neißebrücke Görlitz bridge spans the Neisse River and is a German–Polish border crossing. It was finished in 1992, and the southern bridge caphas been investigated. Despite very low potential readings (about –280 to–350 mV vs. CSE), no corrosion could be found as well as no chlorides. Theconcrete cover is fairly high – about 60 to 70 mm, but the rebar image looksvery interesting and shows imperfections introduced during concrete placement(Fig. 12.5).

12.4.2 Highway A4 – exit Uhyst

Another structure to be investigated is the highway bridge of the A4 exit atUhyst, 50 km east of Dresden (Fig. 12.6). It was finished in 1995 and,although of young age, up to 2.5 mass % chlorides were found in the splashzone of the abutments. This seems to contradict the observations made at theNeisse bridge, but obviously the chloride ingress is closely connected to thehumidity content in the concrete: a bridge cap, being exposed to the weatherstays relatively wet and has a low capillary suction ability; thawing salts willmostly be washed away by rain; the abutments, where a road is passing by,are protected from rain and have a much higher capillary suction ability.

12.5 Rebar display of the bridge cap showing slight imperfections.

12.6 Photo documentation of the Uhyst bridge.



A monthly surveillance of both abutments was started which includedpotential readings, humidity values and chloride measurements. There weresome repaired cracks to be seen (Fig. 12.6), and on the western abutmentthree zones of dampness showed a considerable potential drop of more than100 mV.

As an example, an image of the potential distribution of the westernabutment is given (Fig. 12.7). It shows a typical slope from the abutmenthead (back of the figure) down to the abutment bottom (front of the abutment).This appearance is disturbed by vertical potential breakdowns in the areas ofspotted repaired cracks and zones of dampness. The potential mapping of theeastern abutment shows a perfect slope from head to bottom over the wholewidth.

The potential values range from +80 to –240 mV vs. CSE and are fairlyhigh. If being interpreted by the advice given in [3], no corrosion activityshould be determined. However, a different evaluation results from a diagramdisplaying the sum frequency of all measured potential values vs. the potentialsaccording to [4] (Fig. 12.8). Here, the three ranges of corrosion likelihood liewithin the potential spectrum and show with the curve form of the westernpotentials that the vertical lower potentials do not result from chlorides.

The investigated chloride profiles show a higher content in the splashzone – up to 2.5 wt % cement – but with the thick concrete covers, there isdefinitely no corrosion damage yet. When evaluating the critical potentials

R1

R5

Res

t p

ote

nti

al (

mV

vs

CS

E)

100

50

0

–50

–100

–150

–200

–250

–300

C1 C2 C3 C4 C5 C6 C7 C8 C9

C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23 C24 C25 C26

12.7 Rest potentials of the western abutment.



down from –100 mV, a possible corrosion activity is concentrated on the row0–0.6 m above ground. The structure is perfectly suited for demonstratingelectrochemical chloride extraction – no concrete damage yet, different concretezones to be controlled, a high concrete cover and a rather difficult concretemade with CEM III.

It should be pointed out that the monitoring system is not designed toremove responsibility from the operator. It shall assist him in a very convenientand detailed way, but will not give any ultimate conditional evaluations orsuggestions about what to do. This is not possible because corrosion continuesto be a very complex matter. Nevertheless, by the unification and assistedreproduction of very different measurement data on a single location – thebasic cell – the monitoring system offers an extended scale of interpretationwithout much effort.


This project was carried out with the financial support of the Federal Ministryof Economics of the FRG within the FUTOUR program. Furthermore, theSaxonian Highway Administration provided assistance in the field monitoringof the highway bridges.

12.6 References

1. HILTI AG, Manual Ferroscan 4.0 Data Analysis Software, Liechtenstein, 1997.2. R. D. Browne and D. C. Pocock, ‘Assessments in Practice’, Proc. 2nd Comett Course,

Aachen, 1992.

12.8 Potential evaluation by the sum frequency.

100

10

1

0

Su

m f

req

uen

cy (

%)

Danger of corrosion Transition stage Corrosion safe stage

Widerlager WestWiderlager Ost

–250 –200 –150 –100 –50 0 50 100Rest potential (mV vs CSE)



3. H. Wojtas, Elektrochemische, zerstörungsfreie Prüfmethoden für Zustandsanalysenund Qualitätssicherung bei Instandsetzung von Stahlbetonbauten, Int. J. Restor. Build.,1997, 3, (6) 581–602.

4. ASTM C876-80, Standard Test Method for Half Cell Potentials of Reinforcing Steel inConcrete.


159

13.1 Introduction

Much of the infrastructure in Europe has reached an age where capital costshave decreased. However, inspection and maintenance costs have grown soextensively that they constitute the major part of the current costs [1].

During a Brite/Euram Project several European partners have developedand produced an integrated monitoring system to reduce inspection andmaintenance costs and disturbances to traffic. Additionally, the system willallow the operators to take preventative actions before damaging processes start.

A major part of this project involved determination of the corrosion stateof the rebars in new and existing structures depending on the deterioration ofthe concrete [2]. In addition to the evaluation of different types of sensors,newly developed portable equipment using the galvanostatic pulse techniquewas tested in laboratory conditions. The objective was to test the suitabilityof portable monitoring equipment for the unambiguous, non-destructivedetermination of reinforcement corrosion. Evaluation of the results fromdestructive testing in the laboratory provides background information ofvalue for on-site situations.

13.2 Background

The galvanostatic pulse technique was introduced for field application in1988 to overcome problems with interpretation of the corrosion risk toreinforcement occurring when half-cell potential readings are applied in wet,dense or polymer-modified concrete, where access of oxygen is limited [3,4]. Since the introduction of this technique, development work has beenconducted in order to allow quantitative evaluation of ongoing reinforcementcorrosion [5–9].

The galvanostatic pulse method is a rapid, non-destructive polarisationtechnique, which has been used for evaluating the corrosion of reinforcementboth in the laboratory and on site.

13Use of portable equipment to determine

the corrosion state of concretestructures

R. B Ä ß L E R and A. B U R K E R T, Federal Institute for MaterialsResearch and Testing (BAM), Berlin, Germany; and

T. F R Ø L U N D and O. K L I N G H O F F E RForce Technology, Denmark



A short duration anodic current pulse is impressed galvanostatically onthe reinforcement from a counter electrode placed on the concrete surfacetogether with a reference electrode. The applied current is normally in therange 5 to 400 mA and the typical pulse duration is up to 10 s. The smallanodic current causes a change of the reinforcement potential, which isrecorded as a function of the polarisation time. The reinforcement is polarisedin the anodic direction compared with its free corrosion potential. A typicalpotential transient response is shown in Fig. 13.1.

When the constant current Iapp is applied to the system, the polarisedpotential of the reinforcement Et, at a given time t can be expressed as:

E I R e Rt

R Ct app P

– = 1 – + P dl

ÊËÁ

ˆ¯

È

ÎÍÍ

˘

˚˙˙

W (13.1)

where: Rp = polarisation resistance, Cdl = double layer capacitance andRW = ohmic resistance

In order to obtain values of Rp and Cdl separate from the ohmic resistance,RW, this equation can be transferred to the linear form:

ln( – ) = ln( ) – max t app pP dl

E E I R tR C (13.2)

where Emax is the final steady potential value.Extrapolation of this straight line to t = 0, using least square linear regression

analysis, yields an intercept corresponding to ln (IappRP) with a slope of(RpCdl)

–1.The remaining overpotential corresponds to IappRW, which is the ohmic

voltage drop. After the polarisation resistance, Rp, is determined by means of

Po

lari

sati

on

(m

V)

300

200

100

0

–100

–200

Emax

I · RP

I · RW

Ecorr

1 2 3 4 5 6Time (s)

13.1 Typical polarisation pattern.


Use of portable equipment to determine the corrosion state 161

this analysis, the corrosion current, Icorr, can be calculated from the Stern-Geary equation (3):

I BRcorr

P = (13.3)

where B is an empirical constant determined to be 25 mV for actively corrodingsteel and 50 mV for passive steel.

The dc polarisation resistance technique with calculation of the instantaneouscorrosion current (Icorr) from the Stern-Geary equation has been appliedextensively since 1970. The problem is that in real structures the area of thecounter electrode is much smaller than that of the working electrode(reinforcement) and the electrical signal tends to vanish with increasingdistance. As a result, the measured effective polarisation resistance cannot beconverted to a corrosion rate.

To overcome this problem, a second concentric counter electrode (Guardring)is used to confine the current to the area of the central counter electrode (Fig.13.2). When the diameter of the reinforcement and the confined length of thereinforcement (counter electrode diameter) are known, the instantaneouscorrosion rate can be calculated. It is important to emphasise that the corrosionrate obtained is an instantaneous average rate for the confined area thatstrictly applies to the measuring conditions. Exposure conditions, especiallytemperature and concrete humidity, can alter Icorr by a factor of 10 or more.Experimental data from on-site measurements have shown that the averagecorrosion rates determined from RP measurements in the case of chloride-induced localised corrosion underestimates the real corrosion rate by a factorof 5–10 or even more. From an engineering point of view such local reductionof the reinforcement cross-section is dangerous for the safety of structures,

Ag/AgClReferenceelectrode

Counterelectrode

Guardring

Reinforcement

PSIONWorkAbout

70 mm

13.2 Conditions on pulse head.



especially in zones with high tensile or shear forces. It is obvious that thewrong estimation of the amount of reinforcement for parallel or crossingbars makes the average corrosion rate too high but cracks and peeling off arealso often the reason for incorrect corrosion rate estimations.

For lifetime predictions, more detailed knowledge of the daily and seasonalchanges of the corrosion rate is required in order to obtain meaningful values.It is essential to combine the corrosion rate measurements with supplementarymounted corrosion and chloride sensors or a number of other non-destructiveevaluation methods to determine the concrete integrity and penetration rates.

13.3 Experimental setup

Various types of specimens made of poor concrete (w/c ratio = 0.6) wereexposed in the laboratory atmosphere (20–25 ∞C, 70% rh). Both chloride-free samples (KR) and specimens with mixed-in chloride (2% by mass ofcement – SC) were used. After curing in the normal way, the corrosion potentialand corrosion rate were measured using the galvanostatic pulse device.

Reliable verification of the corrosion rate measurements was only possibleby gravimetric determination of the weight loss. Therefore, at particular timeintervals complete specimens were smashed and rebar weight loss wasdetermined. Directly before the measurement the surface and the contactsponge were wetted. The sponge was cleaned (squeezed in fresh tap water)after every sample (max. 11 readings). The pulse current was 14 mA for allmeasurements, unless otherwise stated. At different locations, rebars A, Band C were measured separately, then A was connected to B and finally allthree rebars together (Fig. 13.3), both before and after NaCl-injection. Potentialwas measured versus an Ag/AgCl gel electrode (incorporated within themeasurement head) having a potential of 207 ± 1 mV versus a standard-H-electrode. In some of the specimens holes were drilled above rebar A. Theseholes were kept filled with NaCl solution and, after several days, it becameapparent that the humidity had distributed itself within the specimen. Moisturemarks developed on the specimen, confirming that the aggressive environmenthad reached the rebar.

13.4 Results

13.4.1 Influence of surface area on partly activereinforcement

After stabilization had taken place measurements were performed at thelocations shown in Fig. 13.3. These were compared to readings obtained indry conditions (i. e. before NaCl injection). Both potential and current densityreadings showed activity on rebar A, where NaCl was injected (Fig. 13.4).



On dry (passive) specimens no significant difference in current densitycould be observed. The slight increase of potential might be caused bypolarisation effects due to the pulse of the previous measurement.

On wet specimens both diagrams clearly show activity on rebar A whilstB and C remained passive (separated bars). Connecting active and passiverebars does not show any effects of surface area or position. All values aredominated by the active partner. This leads to the problem that the currentdensity value shows activity, but does not allow calculation of the corrosionrate since the surface area is not known.

13.4.2 Long term exposure of perforated specimens

Seven specimens that had been perforated to accelerate the corrosion processeswere observed for a longer time period using periodic pulse measurements.Significant differences were detected between samples containing chloridefrom the beginning (cast-in) and chloride-free samples. So, for instance,directly-chloride-exposed rebar A of specimen SC41 showed, during thefirst 14 days, a more rapid increase of current density values from below1 up to 10 mA cm–2. Meanwhile, the potential dropped within a few hoursfrom –275 to –400 mV (vs. AgCl). Subsequently, the current density valuesreturned tentatively and varied around a low level. This return was accompaniedby an increase of potential to values in the range around –350 mV. The

Boreholes

C

S

C

2% NaCl

S

BA

Cross section

BA

11

10

1, 4,7

3, 5,8

2, 6,9

13.3 Perforated specimen with measurement locations and crosssection.



variations were mainly caused by discontinuous ponding of the rebars withNaCl solution (temporary drying out). After 200 h, ponding was reduced,whereby the current density significantly decreased. Only on indirectly affectedbar (SC41-B) was the current density increase more time-delayed and onlyby a small amount of 3 mA m–2. Potential was also less time-delayed andtook > 500 h to reach values below –300 mV (Fig. 13.5).

On specimen KR41 (Fig. 13.6) on directly exposed rebar A, a quickincrease of current density to values up to 3 mA cm–2 was also observed.However, current densities returned to 1.5 mA cm–2 after only 400 h. After1250 h a slow increase of the current density to 4 mA cm–2 was observed,which ended after 2000 h due to temporary drying out caused by interruptionof the NaCl addition. This drying-out effect correlated to the behaviour, alsoobserved on other specimens. Rebars KR41-B and KR41-C, only indirectlyaffected, showed only a slight increase of the current density to values ofaround 0.5 mA cm–2 within the first 500 h. Values remained in that range until2500 h. On rebar KR-C, an increase to 1 mA cm2 was measured. Duringexposure, rebars A and B were permanently connected. Pulse measurementsin these conditions showed essentially the same behaviour of current densityand potential as on directly exposed rebars A.

KR dryKR A-wet

KR dryKR A-wet

Po

ten

tial

[m

V (

Ag

/Ag

Cl)

]

200

100

0

–100

–200

–300

–400

(b)

1 2 3 4 5 6 7 8 9 10 11A B C A+B A+B+C

(a)

Cu

rren

t d

ensi

ty (mA

cm

–2)

10

1

0.1

0.01

13.4 Potential and current readings at different locations and variousrebar configurations.



Corresponding curves are shown in Fig. 13.7. In addition to the averagevalues, the spread of three measurements is displayed. Also the differentbehaviours of specimens SC41 and KR41 is obvious.

The surface appearance of the specimens was analysed after crushing theconcrete cover and pickling off the corrosion products. In both examples, themore severe corrosion on rebar A (where NaCl solution was injected) wasvisible. Almost uniform attack was apparent which allowed the measuredvalues to be related to the whole surface. For rebar C of the KR-specimen,local corrosion attack in the area of holes at rebar A was clearly visible. Thismeans that the measured values cannot be related to the whole rebar surface.

Cu

rren

t d

ensi

ty (mA

cm

–2)

9

8

7

6

5

4

3

2

1

0

SC 41 – ASC 41 – B

Drying

0 1000 2000 3000Time (h)

(a)P

ote

nti

al [

mV

(A

g/A

gC

l)]

0

–50

–100

–150

–200

–250

–300

–350

–400

–450

SC 41 – ASC 41 – B

0 1000 2000 3000Time (h)

(b)

13.5 Values of (a) current density and (b) potential on rebars of 2%chloride–containing specimen SC41 with perforated concrete coverand 2% chloride ponding on rebar A.

Cu

rren

t d

ensi

ty (mA

cm

–2)

9

8

7

6

5

4

3

2

1

0

KR 41 – AKR 41 – BKR 41 – C

Drying

0 1000 2000 3000Time (h)

(a)

Po

ten

tial

[m

V (

Ag

/Ag

Cl)

]

0

–50

–100

–150

–200

–250

–300

–350

–400

–450

KR 41 – AKR 41 – BKR 41 – C

0 1000 2000 3000Time (h)

(b)

13.6 Values of (a) current density and (b) potential on rebars ofchloride-free specimen KR41 with perforated concrete cover and 2%chloride ponding on rebar A.



Therefore, measured values will only be discussed where uniform attacktook place.

By integration of the current ‘considering the polarised area’ the amountof charge (Q), transferred during the experiment, can be obtained. Dividingthat amount of charge by the exposure time, tA, an average corrosion current,Im, can be calculated. By dividing that quantity by the surface area of therebar (A = 31.4 cm2), the average current density, im, is obtained, whichwould cause the same charge transfer by integration over time.

For comparison, block SC was destroyed after 4 months, and the weightloss of the rebars was determined by pickling off the corrosion product.From the weight loss thus obtained and the exposure time, the corrosioncurrent density icorr was calculated by Faraday’s law.

Comparison of current densities obtained by pulse measurement and byintegrated weight loss is shown in Table 13.1. The values show a relativelygood correspondence.

Cu

rren

t d

ensi

ty (mA

cm

–2)

1098

76

5

432

10

SC 41 A+BKR 41 A+B

0 1000 2000 3000Time (h)

(a)

Po

ten

tial

[m

V (

Ag

/Ag

Cl)

]

0

–50

–100

–150

–200

–250

–300

–350

–400

–4500 1000 2000 3000

Time (h)(b)

SC 41 A+BKR 41 A+B

13.7 Values of (a) current density and (b) potential on rebars ofspecimens SC41 and KR41 with perforated concrete cover and 2%chloride ponding on rebar A, rebar A and B connected.

Table 13.1 Current density values calculated from weight loss and obtained fromGPM measurements

SC 41 Weight loss Current density Mean current densitywhole bar from weight loss calculated from GPM

over 4 monthsDescription Dm (g) icorr (mA cm–2 ) im (mA cm–2)

bar A (2 cm depth) 1.30 4.8 3.6bar B (3 cm depth) 1.36 5.0 1.5bar A + B (centre) 2.66 4.9 5.0



13.5 Discussion

Investigations in laboratory conditions clearly show that GPM is suitable forevaluating the real extent of corrosion of reinforcement in concrete. Activeand passive conditions can be detected exactly. In addition, the influenceof corrosion stimulation by ponding with NaCl solution has been proven.The influence of discontinuous wetting (with temporary drying phases) onthe corrosion behaviour of reinforcing steel could be detected by periodicGPM measurements. So GPM is an important addition to corrosion potentialmeasurements.

Even though separate investigations in laboratory conditions indicated arelatively good correlation between current densities calculated from weightloss and obtained by GPM, it needs to be emphasised that the determinationof the corrosion current density by GPM is only a semi-quantitative method.

Arguably, it is possible to distinguish between areas of strong, medium,and low or no corrosion. However, corrosion currents, derived from polarisationresistance measurements, and lifetime estimates based on these values canbe affected by many influences, leading to incorrect interpretations. In particular,it must be pointed out that the area to which all values are referred is anassumed one based on the size of the polarised area, taking into account thefield distribution and position of the measurement head rather than the actualcorroding area (which may involve localised or partial corrosion attack).Furthermore, an interaction of active and passive areas needs to be assumed,which cannot be simulated in laboratory measurements in different conditions.Additionally, it must be appreciated that GPM measurements only providean instantaneous indication of the actual corrosion situation, which issignificantly affected by the condition of the concrete (for instance moisturecontent, pH-value, and chloride content).

Also, on one hand the corrosion rate can be reduced by the formation ofcorrosion products on the surface (providing a barrier to diffusion). However,on the other hand, promoting the formation of thick corrosion product layerscan accelerate the corrosion (by crack formation, the detachment of concrete,and hygroscopic salt effects). These systematically and randomly influencingfactors can cause the corrosion current to be miscalculated by up to an orderof magnitude, ruling out the possibility of accurate lifetime estimation.However, it should not be overlooked that the influence of these factors willbe much lower under on-site conditions, because the separation betweenactive and passive areas will be greater. Therefore, the interaction betweendifferent areas that occurred in the laboratory experiments will be significantlysmaller. What is more, uniform corrosion attack is observed in practicecontrary to what was observed in the laboratory tests. However, in on-siteconditions other factors tend to make estimation of corrosion more difficultthan under well-defined laboratory conditions.



Pulse measurements do not provide information about past or futurecorrosion conditions and their development. These are affected by changingconditions in the surrounding concrete, which are subjected to statistical,random, and real, physical deviations. Only information on the current situationis available. That is why there is a need for long-term observation combinedwith periodical pulse measurements to provide information about corrosiontrends (i.e. whether steady, increasing or decreasing).

A semi-quantitative estimate of the actual corrosion rate is all that ispossible based on GPM measurements when further considerations like thepotential of reinforcement and the concrete humidity, etc., are taken intoaccount.

13.6 Conclusions

A newly developed, easy-to-handle portable instrument using the galvanostaticpulse method has been tested on several materials in different environmentalconditions to provide quick information on the actual corrosion behaviour ofreinforcement in concrete. Key parameters were concrete composition, rebarconditions, humidity and temperature. Special attention was paid to thecomparison of instrument readings with actual behaviour. Various combinationswere tested and the response of the instrument under various circumstanceswas compared with actual material losses for the evaluated rebars.

In laboratory conditions the actual corrosion state could be determined.Evaluation of the results obtained during long-term investigations showedvery good correlation with the real corrosion state and enabled users toestimate the corrosion behaviour of reinforcement. However, exact lifetimeestimations using nothing but GPM results are only partly successful andneed much further consideration because the measurement conditions (moisturecontent, temperature, unknown active area, etc.) strongly affect the valuesobtained.

13.7 Outlook

The work described here is continuing, including investigations to determinethe area influence on lifetime prediction. The observed behaviour in laboratorytests will be compared to results from on-site investigations. The results willbe presented in future publications.

These results combined with others relating to deformation and vibrationprobes, obtained during this BRITE/EURAM project will contribute to thedevelopment of an integrated corrosion monitoring system so that end-userswill be able to optimise their maintenance management systems and, therefore,costs and disruption to traffic can be reduced.




This work was funded by the European Community as a Brite/Euram project‘Smart Structures’, contract number BRPR-CT98-0751. The authors are gratefulfor this support. Furthermore, the contributions of all partners in this project,including Autostrade, the Danish Road Institute, OSMOS-Dehacom, DLR,Rambøll and S+R Sensortec, are gratefully acknowledged.

13.9 References

1. Wallbank, E. J., The Performance of Concrete Bridges, A Survey of 200 HighwayBridges, HMSO, London, UK, April 1983.

2. Broomfield, J. P., Corrosion of Steel in Concrete, Understanding, Investigating andRepair, E & FN SPON, 1997.

3. Klinghoffer, O., Rislund, E., Frølund, T., Elsener, B., Schiegg, Y. and Böhni, H.,‘Assessment of Reinforcement Corrosion by Galvanostatic Pulse Technique’, Proc.Int. Conf. on Repair of Concrete Structures, Svolvaer, Norway, 1997, 391–400.

4. Danish Patent 171925B1, 1997.5. Andrade, C., Alonso, C., Gonzalez, J. A. and Rodriguez, J., ‘Remaining service life of

corroding structures’, IABSE Report 57/1, Durability of Structures, 1989, 359–364.6. Andrade, C., Alonso, C. and Gonzalez, J. A., ‘An initial effort to use the corrosion rate

measurements to estimating rebar durability’, ASTM STP Corrosion Rate of Steel inConcrete, 1990, 29–37.

7. Clear, K., ‘Measuring rate of corrosion of steel in field concrete structures’, paper No.88-0324, 68th Annual Transportation Research Meeting, Washington DC, 1989.

8. Elsener, B., ‘Elektrochemische Methoden zur Bauwerksüberwachung’, ZerstörungsfreiePrüfung an Stahlbetonbauwerken, SIA Dokumentation D020, Schweizer Ingenieur-und Architektenverein, Zürich, 1988.

9. Newton, C. J. and Sykes, J. M., ‘A galvanostatic pulse technique for investigation ofsteel corrosion in concrete’, Corros. Sci.,1988, 28, 1051–1074.


170

14.1 Introduction

In general, reinforced concrete has proved to be successful in terms of bothstructural performance and durability. However, there are instances of prematurefailure of reinforced concrete components due to corrosion of the reinforcement.The two principal factors provoking corrosion are the ingress of chlorideions from deicing salts or sea water and the reaction of the alkaline poresolution with carbon dioxide from the atmosphere, a process known ascarbonatation. Despite the huge demand, a simple, cheap, and reliable techniquewhich either protects the steel from corrosion or at least lowers its corrosionrate is still lacking. Over the past decade, however, the concrete repair industryhas developed novel techniques that are claimed to prevent, or at least toreduce, the corrosion of steel in concrete. The use of these ‘corrosion inhibitors’is of increasing interest as they can be used in reinforced concrete either asa preventative measure for new structures (as an addition to the mixingwater) or as surface applied inhibitors for preventive and restorative purposes.Addition to the mixing water does not require any additional working stepsand allows a simple handling of the inhibitor, unless it affects the propertiesof the cement paste adversely. Application from the concrete surface couldbe a promising technique to protect existing structures from corrosion or toincrease the lifetime of structures that already show corrosion attack. Theapplication of inhibitors on the concrete surface requires the migration of thesubstance to the rebar where it has to reach a sufficiently high concentrationto protect steel against corrosion or reduce the rate of ongoing corrosion.

Usually the long experience with chemicals operating as corrosion inhibitors,e.g. in the oil-field, gas or petroleum industry, is taken as evidence of thesuccessful use of corrosion inhibitors, implying that this success is relevantto applications in reinforced concrete. This is a priori not correct because themechanism of inhibitor action is completely different:

∑ in the oil and gas industry applications (and most others), the steel to beprotected is uniformly corroding in slightly acidic or neutral media. Thus,

14Corrosion inhibitors for reinforced concrete

– an EFC state of the art report

B. E L S E N E R, University of Cagliari, Italy andETH Zürich, Switzerland


Corrosion inhibitors for reinforced concrete 171

the inhibitors have to protect the bare metal surface, e.g. as adsorptioninhibitors acting specifically on the anodic or on the cathodic partialreaction of the corrosion process or as film forming inhibitors blockingthe surface more or less completely [1, 2]. Usually, a reduction in corrosionrate of 95 to 99% is achieved with a very small inhibitor concentration inthe order of 10–3 – 10–2 mol L–1.

∑ in contrast, steel in concrete is in a highly alkaline environment; the highconcentration of hydroxyl ions acts as a passivation-promoting inhibitorand, indeed, steel in concrete is passive, being protected by a thinoxyhydroxide layer. This is the starting point for any mechanistic actionof inhibitors in concrete.

∑ inhibitors for chloride-induced pitting corrosion have received far lessstudy [3]. Inhibitors for pitting corrosion can act by forming a film beforethe ingress of chlorides, by buffering the pH in the local pit environment,by competitive surface adsorption processes between inhibitor and chlorideions or by competitive migration of inhibitor and chloride ions into thepit.

Another point concerning the terminology of ‘inhibitors’ must be clarified:a corrosion inhibitor prolongs the service life due to chemical/electrochemicalinteraction with the reinforcement. Any other substances that may preventthe onset of corrosion or reduce ongoing corrosion by surface treatment (e.g.hydrophobation) or by admixtures that reduce porosity of the concrete (e.g.fly ash, silica fume, waterproofing admixtures etc.) are not considered to becorrosion inhibitors.

Most of the results published in the literature and reviewed recently [4, 5]are from laboratory studies involving solution experiments or relatively smallmortar samples. Long term performance results are available for admixedinhibitors only, in particular calcium nitrite [6]. Results from well documentedfield tests involving surface applied inhibitors are rare. There are, however,other difficulties in obtaining unambiguous, conclusive results on theperformance of corrosion inhibitors on reinforced concrete structures:

∑ most of the ‘inhibitors’ available under different trade names are blends ofessentially unknown composition that could be changed without notice.This makes even laboratory experiments difficult.

∑ sometimes the use of surface-applied inhibitors is recommended only inconjunction with other corrosion protection methods, such as hydrophobationof the surface, and it is then difficult to isolate the inhibitor performance.

This paper is based on a state of the art report by the author on corrosioninhibitors for steel in concrete [5] and the literature results reviewed therein.In particular, calcium nitrite (DCI), the migrating corrosion inhibitors (SIKAor MCI) and MFP (monofluorophosphate) are addressed. The problem oftesting different inhibitors for steel in concrete is addressed and – as far as



available – results from field tests with inhibitors are presented. Finally, acritical evaluation of corrosion inhibitors for steel in concrete is given.

14.2 Mode of action of corrosion inhibitors

The service life of a concrete structure with respect to reinforcement corrosion,as described by Tuutti [7] (Fig. 14.1), consists of two phases: the first phasecorresponds to the initiation time, ta or tb,c taken for chlorides or CO2 topenetrate the concrete cover in sufficient quantities to destroy the passivefilm (depassivation). The second phase covers the period of active corrosionafter ta or tb,c until the point where the serviceability of the structure areaffected (loss of load bearing capacity, spalling or delamination) andmaintenance or repair is needed. The length of this period is determined bythe rate of corrosion (slope a, governed by the oxygen availability, humidityand temperature) and the ability of concrete to withstand internal stress.

In view of this general picture of corrosion, admixed (preventative) corrosioninhibitors can act in two ways:

∑ the inhibitor extends the initiation time from t0 to a later moment, t0b,following the corrosion process according to slope b (inhibitors that preventor delay corrosion initiation) and

∑ the inhibitor reduces the corrosion rate after depassivation has occured(slope c) and the service life is extended until t.

From the point of view of the design process and the durability of a structure,the first mode of action, extending the initiation time, is much more reliable.

When inhibitors are applied on the surface of concrete during the initiationperiod (t < to), the mode of action is in principle identical (given the necessary

Limit stateEnd of

service life

a

b

c

Deg

rad

atio

n

t0 t0b Time

Initiation Propagation

14.1 Lifetime of a reinforced concrete structure according to Tuutti [7]adapted on the action of corrosion inhibitors. Time of corrosioninitiation (depassivation) is t0, or t0b, time for maintenance is ta, tb, tc.



concentration at the rebar is reached). If corrosion has already started (t > t0),the only possible mode of action is to lower the corrosion rate.

14.3 Corrosion inhibitors to prevent or delay

corrosion initiation

The most frequently used technique is addition of the inhibitors to the mixingwater of concrete as admixtures for new structures in order to prevent or atleast delay the onset of corrosion. Calcium nitrite is the most extensivelytested admixed corrosion inhibitor [7] and has – when applied according tothe specifications together with high quality concrete and sufficient cover –a long and proven track record in the USA, Japan and the Middle East [7].It is used in parking, marine and highway structures. Nitrite acts as a passivatordue to its oxidising properties and stabilises the passive film [8]. Allinvestigations have revealed a critical concentration ratio (threshold value)between inhibitor (nitrite) and chloride of about 0.6 (with some variationfrom 0.5 to 1) in order to prevent the onset of corrosion.

The action of calcium nitrite inhibitor has to be treated statistically (Fig.14.2), thus the delay in corrosion initiation may vary considerably [9].

Another inorganic inhibitor, sodium monofluorophosphate (Na2PO3F; MFP),can be used only as a surface-applied inhibitor due to its adverse chemicalreaction with fresh concrete [10]. Laboratory studies of the preventive inhibitoraction against chloride-induced corrosion have shown that by applying severalintense flushings before the ingress of chlorides [11] it can prevent the onsetof corrosion during a test duration of 90 days, even at chloride concentrationsas high as 2% by weight of cement. A critical concentration ratio MFP/chlorides greater than 1 must be achieved, otherwise the reduction in corrosion

0% Ca(NO2)2

0.5% Ca(NO2)2

2.0% Ca(NO2)2

4.0% Ca(NO2)2

0.91.0 2.0 3.0 4.0 6.0 8.0 0.0Time for noble-to-active potential shift (days ¥ 10–2)

99.9

95.590.080.060.040.0

20.0

10.0

5.0

1.0Sam

ple

s w

ith

act

ive

po

ten

tial

(%

)

14.2 Time to corrosion initiation of steel in mortar samples withadmixed Ca(NO2)2 inhibitor exposed to sea water [9].



rate is not significant [11]. In solutions containing Ca(OH)2, MFP is reportedto react with the calcium ion to form insoluble products such as calciumphosphate and calcium fluoride [10, 11], thus the active substance, the PO3F

2–

ion, disappears from the pore solution. The main problem in using MFP asa surface-applied liquid is the penetration to the reinforcement in order to actas inhibitor. In early field tests in Switzerland, insufficient penetration ofMFP was found [12]. This was partly due to concrete of too high a density,to a cover depth greater than 45 mm or to an insufficient number of MFPapplications on the surface. In more recent field applications [13], e.g. on thePeney Bridge near Geneva [13], concrete buildings and balconies, MFP wasapplied onto cleaned, dry concrete surfaces in up to 10 passes and the concretewas impregnated to the reinforcement level in a few days or weeks [13].More recently, it has been found that the use of an MFP-containing gel onthe concrete surface could improve the penetration of MFP.

Organic inhibitors, especially alkanolamines and amines and their salts,with organic and inorganic acids are used as components in corrosion inhibitorblends of usually complex formulations [14]. These blends are often notsufficiently well described so most of the published work has been undertakenwith commercially available systems. A comparative test with differentcorrosion inhibitors [15] showed very good corrosion inhibition of thecommercial inhibitor blend at a high concentration, pure dimethylethanolamineinstead being practically ineffective. Recent research work at ETH Zürichinvestigating a commercial migrating inhibitor blend has shown that theblend can be fractionated into a volatile (dimethylethanolamine) and a non-volatile (benzoate) component [16]. For the complete prevention of corrosioninitiation in saturated Ca(OH)2 solution with 1M NaCl added, the presence ofboth components at the steel surface in a concentration ratio of inhibitor/chloride of ca. 1 was necessary: neither component of the inhibitor whenpresent alone in solution could prevent initiation of corrosion (Fig. 14.3).Modern surface analytical techniques, such as XPS [17], have shown that forthe formation of a significantly thicker organic film on iron in alkalinesolutions both components of the commercial inhibitor blend have to bepresent. This might be significant for the mechanism of the inhibitor action.The inhibitor blend in the recommended dosage could delay the averagetime to corrosion initiation of passive steel in mortar by a factor of 2–3 (Fig.14.4).

In a comparative study [18], four commercially available inhibitors weretested using different admixed dosages (Fig. 14.5): calcium nitrite (DO), anorganic corrosion inhibitor (ORG1), an inhibitor based on alkanolamine(ORG2) and another calcium nitrite product (CN2). All inhibitors coulddelay the onset of corrosion but only the commercial calcium nitrite with thehighest dosage gave a significant improvement. The same alkanolamine-based commercial inhibitor blend was tested as an admixture in mortar and



concrete samples exposed to chlorides [19]. After one year of testing, corrosionhad started in specimens with w/c = 0.6, the chloride threshold values for theinhibitor-containing samples are in all cases higher (4–6% Cl– by weight ofcement) compared with the control samples (1–3% Cl–).

Prevention or at least prolongation of the onset of corrosion has beenreported also for an organic corrosion-inhibiting admixture (OCI) proposedin a United States Patent [20]. The admixture is an oil/water emulsion,wherein the oil phase consists of an unsaturated fatty acid ester of an aliphaticcarboxylic acid with a mono-, di- or tri-hydric alcohol and the water phasecontains a saturated fatty acid, an amphoteric compound, a glycol and asoap. The admixture is added to concrete before placement. Upon contactwith the high pH environment of concrete the waterproofing ester component

Po

ten

tial

(m

V S

CE

)

0

–100

–200

–300

–400

–500

–600

No inhibitorNon-volatile constituentVolatile constituent

1 2

Opening ofthe cell

0 10 20 30 40 50 60 70t (d)

14.3 Corrosion potentials of rebar samples in solutions containingthe two components of the inhibitor: � no inhibitor, ∑ volatileconstituent, D non-volatile constituent 1: prepassivation in saturatedCa(OH)2; 2: immersion in saturated Ca(OH)2 + 1M NaCl [16].

Blank0.35 kg m–3

1.75 kg m–3

8.75 kg m–3

0 50 100 150 200 250 300 350t (d)

Co

rro

din

g s

amp

les

(%)

100

80

60

40

20

0

14.4 Percentage of corroding rebars in mortar vs. time of cyclicchloride treatment [16].



becomes hydrolysed, forming carboxylic anions that are precipitated in thepresence of calcium ions as a hydrophobic coating within the pore system,reducing ingress of water and chlorides into the concrete [21, 22]. This pore-blocking effect is not a true corrosion inhibition.

In summary, to prevent or strongly delay the onset of pitting corrosion onpassive steel in alkaline solutions or mortar, all investigations – independentof the type of inhibitor – seem to indicate that a critical ratio of inhibitor/chloride of about 1 has to be exceeded. This implies that quite high inhibitorconcentrations have to be present in the pore water of concrete in order to actagainst chlorides penetrating from the concrete surface. To avoid chlorideingress and, thus, the use of excessively high inhibitor concentrations, theuse of admixed inhibitors is recommended only together with high qualityconcrete [6]. Too low a concentration of certain inhibitors may cause anincreased localised corrosion rate, as has been found in laboratory studieswith nitrites in cracked reinforcing beams [23].

14.4 Corrosion inhibitors to reduce the propagation

rate of corrosion

The most interesting application of inhibitors would be a surface treatmentwith subsequent transport of the inhibitor to the corroding steel with the

‘Pore blocker’

Calcium nitrite DCl

Control

0 100 200 300 400 500 600 700Time (d)

Inh

ibit

or

(L m

–3)

Control

CN 10

CN 20

CN 30

ORG1 5

ORG1 7

ORG1 9

ORG2 20ORG2 25

ORG2 30

CN2 3

CN2 4

CN2 5

14.5 Time to corrosion initiation of four steel bars in mortar blocksexposed to cyclic ponding with chloride solutions [18] for differentinhibitors admixed to the mortar in three dosages.



effect of stopping or at least reducing ongoing corrosion. Several laboratoryand field tests have been performed to investigate this particular situation.

For monofluorophosphate, repeated drying and MFP-immersion cycleshave been found a suitable method to allow the penetration of the inhibitorto the steel, but high concetrations and long treatments are needed tosignificantly reduce active corrosion due to carbonation [24]. In recent research[5] at Aston University, 15% by weight solutions of MFP were appliedrepeatedly to reinforced concrete specimens (water to cement ratio,w/c, 0.65, cover 12 mm) with various levels of chloride contamination. Theembedded bars, precorroded under cyclic wetting and drying conditions forabout 6 months before the MFP treatment, did not exhibit marked reductionsin corrosion rate [2] (Fig. 14.6).

Experiments with a commercial migrating inhibitor blend [16] have shownthat the polarisation resistance measured after the onset of corrosion in solutionincreases with the inhibitor concentration (Table 14.1); both the volatile andthe non-volatile fraction could reduce the corrosion rate slightly comparedwith the non-inhibited solution [4, 16]. In mortar experiments with cyclicponding in 6% chloride solution, however, the polarisation resistance afterthe onset of corrosion did not change with inhibitor concentration and was

Cu

rren

t d

ensi

ty (mA

cm

–2)

5

4

3

2

1

0

BeforeAfter

0% 0.6% 1.2% 2.4%Chloride concentration (w%/cem)

(a)

Cu

rren

t d

ensi

ty (mA

cm

–2)

5

4

3

2

1

0

BeforeAfter

0% 0.6% 1.2% 2.4%Chloride concentration (w%/cem)

(b)

14.6 Corrosion rate of rebars in mortar (w/c 0.65) before and aftertreatment with inhibitor: (a) MFP, (b) proprietary alkanolamine basedinhibitor, after Page et al. [4].

Table 14.1 Average polarisation resistance of threerebar samples after the addition of 1M NaCl to sat.Ca(OH)2 solution with inhibitor [15]

Inhibitor (wt %) Rp (kW cm2)

10 490 ± 801 11 ± 30.1 2 ± 10 3 ± 1



similar to the mortar without inhibitor (Table 14.2) [16]. Thus, a delay in theonset of corrosion is obtained (Fig. 14.4) but no reduction in the corrosionpropagation rate.

Alkanolamine-based inhibitors have been tested in similar conditions. Forongoing chloride-induced corrosion with a chloride level of ca. 1–2% areduction in corrosion rate was not found (Fig. 14.6) either in the laboratory[4] or in the field [25], except at low chloride concentrations. The effect ofanother proprietary migrating inhibitor blend for surface application wastested in solution. After the addition of the inhibitor an increase of thepolarisation resistance by a factor 3 to values of ca. 4 ± 1 kW cm2 was found[26]. In contrast to this result, precorroded rebars in mortar (w/c 0.75, cover25 mm) did not show any increase in polarisation resistance after inhibitortreatment despite low cover and porous mortar [26]. A comparative studyperformed with four admixed inhibitors [18] found a similar result: all theinhibitors were found to have little detectable effect on the corrosion rate ofthe embedded steel once active corrosion had been initiated. For ‘penetratinginhibitors’ the favourable effects found in solution do not occur when appliedto hardened mortar on concrete laboratory specimens.

14.5 Field tests with corrosion inhibitors

In comparative field tests on chloride-contaminated side walls in a tunnel,MFP and a proprietary alkanolamine inhibitor were tested [25]. Both inhibitorswere found to be virtually ineffective at chloride concentrations of 1–2% byweight of cement (Fig. 14.7) [25].

Other field tests with proprietary vapour-phase inhibitors [27] in a parkinggarage with chloride-contaminated precast slabs did not show encouragingresults. Corrosion rate measurements showed a reduction of 60% in areaswith initially intense corrosion but also an increase in areas with low corrosionrates. On structures dating from 1960 with admixed chloride content >1%,already featuring patch repairs, a three-year corrosion rate survey showedlower corrosion rates in the treated areas compared with the untreated onesbut cracking and spalling nevertheless increased in the treated areas [28].

Table 14.2 Average polarisation resistance of rebarsamples in mortar with different inhibitor concentrationafter 343 days cyclic treatment in 6% chloride solution

Series Inhibitor (wt%) Rp (kW cm2)

1 0 5.2 ± 2.22 0.015 6.0 ± 1.43 0.075 5.6 ± 1.64 0.375 6.2 ± 0.9



14.6 Transport of the inhibitor into mortar or

concrete

It is claimed for several inorganic and organic inhibitor blends that theseinhibitors can be applied to existing reinforced concrete structures and thecorrosion inhibitor will be carried by water or by vapour-phase migrationinto the proximity of the reinforcing steel [9, 13]. Several diffusion experimentsshowed that alkanolamine-based inhibitors in particular can diffuse throughthe concrete although great discrepancies in the measured diffusion ratesexist. This might be due partially to the different experimental setups (humidity)and measuring techniques used. In addition, it is difficult to determine thediffusion rate of an inhibitor blend of unknown composition.

A detailed study on the transport of a proprietary amino-alcohol-basedinhibitor (FG903) into cement paste and mortar has been reported by Tritthart[29]. The results showed that both the amount and the rate of inhibitoringress into the alkaline cement paste was higher for the pure amino alcoholcompared with the inhibitor blend also containing phosphates (Fig. 14.8a).This discrepancy could be explained by a reaction of the inorganic phosphatecomponent with the calcium ions in the fresh cement paste blocking furtheringress of the inhibitor. To avoid a reaction with calcium ions, the transportof the inhibitor was studied on cores taken from a 100 year old, fully carbonatedconcrete structure, varying the dosage and the way of inhibitor application(Fig. 14.8b). The recommended dosage (500 g m–2) and way of application

FG 97FG 99

Freq

uen

cy d

istr

ibu

tio

n (

%)

99.99

99.9

99

9590807050

302010

5

1

.1

.01–600 –500–400–300–200–100 0 100

Potential [mV (CSE)]

MFP 97MFP 99

Freq

uen

cy d

istr

ibu

tio

n (

%)

99.99

99.9

99

95908070

50

302010

5

1

.1

.01–600 –500–400 –300–200 –100 0 100

Potential [mV (CSE)]

14.7 Cumulative frequency distribution of half-cell potentialsmeasured on the chloride contaminated field tests for surfaceapplied inhibitors before and two years after application: (a) SIKAFerrogard, (b) MFP [25].



(several brushings) showed only a moderate concentration of the aminoalcohol in the first 15 mm. An increase in the dosage to 1500 g m–2 increasedthe amino alcohol concentration, but the penetration depth remained low.Only ponding for 28 or 50 days resulted in a significant inhibitor concentration(both amino alcohol and phosphate) at depths higher then 30 mm [29].

More often only one – the most volatile – component of the inhibitorblend can be analysed as in the case of a proprietary migrating corrosioninhibitor [16, 30]. Using the amine electrode, the diffusion of the volatilepart of the inhibitor through a mortar disk could be measured [16, 30], noinformation of the diffusion of the non-volatile part could be obtained, and

14.8 Transport of a proprietary aminoalcohol-based inhibitor (SIKAFerrogard 903) into (a) alkaline cement paste and (b) fully carbonatedconcrete cores, from Tritthart [29].

Amino 28 daysAmino 50 daysFG 903 14 daysFG 903 50 days

Phosphorus 0–1 cmPhosphorus 1.5–2.5 cmAminoalcohol 0–1 cmAminoalcohol 1.5–2.5 cm

0.5 2 3.5 5 6.5 8 9.5Depth (cm)

(a)

Co

nce

ntr

atio

n (

pp

m)

8 ¥ 104

7 ¥ 104

6 ¥ 104

5 ¥ 104

4 ¥ 104

3 ¥ 104

2 ¥ 104

1 ¥ 104

0

Co

nce

ntr

atio

n (

pp

m)

8 ¥ 104

7 ¥ 104

6 ¥ 104

5 ¥ 104

4 ¥ 104

3 ¥ 104

2 ¥ 104

1 ¥ 104

0500 1000 1500 28 d 50 d

g (m–2) g (m–2) g (m–2)(b)



it can be reasonably assumed that only the volatile compound is diffusing.The fact that both components of an inhibitor blend are needed at the steelsurface to get an inhibiting effect (Fig. 14.3), but only one component easilydiffuses through the porous concrete, may explain the discrepancy betweensolution experiments and mortar or field tests [4, 16, 26]. A high diffusionrate – logically – does not depend on the diffusion direction; so it has beenfound that the volatile component of organic inhibitor blends evaporates [16,26].

14.7 Critical evaluation of corrosion inhibitors

Assuming that the inhibitor action in laboratory experiments has beenestablished, there remain two critical points for successful and reliableapplication on reinforced concrete structures:

∑ The inhibitor has to be present at the reinforcing steel in sufficiently highconcentration with respect to the aggressive (chloride) ions over a longperiod of time.

∑ The inhibitor action on corrosion of steel in concrete should be measurable.

14.7.1 Concentration dependence

The available literature reports a concentration dependent effect of inhibitors,a critical inhibitor/chloride ratio has to be exceeded (see above). For newstructures, the inhibitor dosage thus has to be specified with respect to theexpected chloride level for the design life of the structure. Surface-appliedinhibitors on existing structures may present even more difficulties in achievingthe necessary concentration at the rebar level. Firstly, because chloridecontamination or carbonation may vary strongly along the surface, secondly,because the cover and permeability of the concrete may also vary and, thirdly,because the inhibitor may react with pore solution components. It is crucialto specify the critical concentration to be achieved at the rebar level and not– as in the application notes of surface applied inhibitors – an average weightof inhibitor solution to be applied per m2 concrete. This is usually omitted,in part due to the lack of analytical methods to measure the inhibitorconcentration. Regarding long term durability, it has to be taken into accountthat inhibitors may be washed out from the concrete or evaporate.

14.7.2 Measurement and control of inhibitor action

One of the main difficulties in evaluating the performance of inhibitors is toassess the inhibitor action on rebar corrosion ‘on site’. The interpretation ofhalf-cell potential measurements may present difficulties due to changes in



the concrete resistivity. Further, a reduction of corrosion rate due to aninhibitor action may not be reflected straightforwardly in the half-cell potential:potentials may become more negative or more positive after inhibitorapplication, depending on the mechanism of the inhibitor action. Shifts inthe half-cell potential may also occur due to the wetting and drying of theconcrete [31, 32]. LPR measurements are considered suitable for on-sitetesting [28], but results of corrosion rate measurements on site depend on thetype of device used for the measurements and can be interpreted so far onlyby specialists. The main problems are the daily and seasonal changes of thecorrosion rate with temperature and concrete humidity making it difficult toevaluate inhibitor action. Macrocell current measurements between isolatedanodes (located and instrumented before inhibitor application) and thesurrounding cathode may give the most indicative results [25] but can beinstalled only on test sites.

14.8 Conclusions

The use of corrosion inhibitors could be a promising technique in restoringreinforced concrete structures, offering benefits such as reduced costs andinconvenience of repairs. It has, however, to be taken into account that theuse of corrosion inhibitors in repair systems is far less well-established thantheir applications as admixtures in new structures.

This paper presents the results and conclusions based on the availableliterature. Briefly, admixed inhibitors with the correct dosage can stronglydelay the onset of chloride-induced corrosion. Once corrosion started nosignificant reduction in corrosion rate has been found. The overall performanceof surface-applied organic and inorganic corrosion inhibitors intended toreduce ongoing chloride-induced corrosion cannot be considered positive,for the case of corrosion due to carbonation there remain at least somedoubts.

Engineers and contractors working in the area of concrete maintenanceshould be aware of the fact that the performance of proprietary corrosioninhibitors in repair systems marketed under different trade names is not yetdocumented by independent research work, especially when consideringfield tests.

14.9 References

1. G. Trabanelli, ‘Corrosion inhibitors’, in Corrosion Mechanisms, F. Mansfeld (ed.),Marcel Dekker NY, 1986, chapter 3.

2. U. Nürnberger, Corrosion Inhibitors for Steel in Concrete, Otto Graf J. 1996, 7, 128.3. D. W. DeBerry, ‘Organic inhibitors for pitting corrosion’, in Review on Corrosion

Inhibitor Science and Technology, A. Raman and P. Labine (eds), NACE, Houston,1993.



4. C. L. Page and V. T. Ngala and M. M. Page, ‘Corrosion inhibitors in concrete repairsystems’, Mag. Concrete Res, 2000, 52, 25–37.

5. B. Elsener, Corrosion Inhibitors for Steel in Concrete – State of the Art Report, EFCSeries Number 35, IOM Communcations, Institute of Materials, London, 2001.

6. N. S. Berke and T. G. Weil, ‘World Wide Review of Corrosion Inhibitors in Concrete’,Advances in Concrete Technology, V. M. Malhotra (ed.), CANMET Ottawa, 1994,899–1022.

7. K. Tuutti, Corrosion of Steel in Concrete, CBI forskniong 4/82, Cement ochBetonginstituet, Stockholm.

8. B. El-Jazairi and N. Berke, ‘The use of calcium nitrite as a corrosion inhibitingadmixture to steel reinforcment in concrete’, in Corrosion of Reinforcement in ConcreteConstruction, C. L. Page, K. W. J. Treadaway and P. B. Bamforth (eds.), Elsevier,London, 1990, 571.

9. W. H. Hartt and A. M. Rosenberg, ‘Influence of Ca(NO2)2 on sea water corrosion ofreinforcing steel in concrete’, American Concrete Institute, Detroit, SP 65-33, 1989,609–622.

10. M. Hynes and B. Malric, ‘Use of Migratory Corrosion Inhibitors’, Constr. Repair,1997, 11(4), 10.

11. C. Alonso, C. Andrade, C. Argiz and B. Malric, Cement Concrete Res. 1992, 22, 869.12. P. Schmalz and B. Malric, ‘Korrosionsbekämpfung in Stahlbeton durch Inhibitoren

auf MFP Basis’, Erhaltung von Brücken, SIA Dokumentation D099, 1993, p. 65,Publ. Schweiz. Ingenieur and Arcchitektenverein, Zürich.

13. P. Annen and B. Malric, ‘Surface applied inhibitor in rehabilitation of Peney Bridge,Geneva (CH). Bridge Management 3, E. Harding, G. A. R. Parke and M. J. Ryall(Eds.), E&FN Spon, London, 1996, p. 437.

14. U. Mäder, ‘A new class of corrosion inhibitors’, in Corrosion and Corrosion Protectionof Steel in Concrete, N. Swamy (Ed.), Sheffield Academic Press, 1994, Vol. 2, p.851.

15. A. Phanasgaonkar, B. Cherry and M. Forsyth, ‘Corrosion inhibition properties oforganic amines in simulated concrete environment’, in Proc. Int. Conf. onUnderstanding Corrosion Mechanisms of Metals in Concrete – a Key to ImprovingInfrastructure Durability. Massachusetts Institute of Technology MIT, Cambridge,1997, section 6.

16. B. Elsener B, M. Büchler, F. Stalder and H. Böhni, ‘A migrating corrosion inhibitorblend for reinforced concrete – Part 1: Prevention of corrosion’, Corrosion, 1999,55, 1155–1163.

17. A. Rossi, B. Elsener, M. Textor and N. D. Spencer, ‘Combined XPS and ToF-SIMSanalyses in the study of inhibitor function – organic films on iron’, Analusis, 1997,25, (5), M30.

18. S. M. Trépanier, B. B. Hope and C. M. Hansson, ‘Corrosion inhibitors in concrete.Part III: Effect on time to chloride-induced corrosion initiation and subsequent corrosionrates of steel in mortar’, Cement Concrete Res. 2001, 31, 713.

19. P. H. Laamanen and K. Byfors, ‘Corrosion inhibitors in concrete – alkanolaminebased inhibitors’, Nordic Concrete Res No. 19, 2/1996, Norsk BetongforengingkOslo, 1996.

20. G. S. Bobrowski, M. A. Bury, S. A. Farrington and C. K Nmai, ‘Admixtures forinhibiting corrosion of steel in concrete’, United States Patent, Patent No. 5.262.089,16.11.1993.

21. C. K. Nmai, S. A. Farrington and G. S. Bobrowski, Concrete Int., 1992, 14, 45.



22. C. K. Nmai and D. McDonald, ‘Long term effectiveness of corrosion inhibitingadmixture and implications for the design of durable reinforced concrete structures:a laboratory investigation’ RILEM Int. Symp. on the Role of Admixtures in HighPerformance Concrete, March 1999.

23. U. Nürnberger and W. Beul, Mater. Corrosion, 1991, 42, 537–546.24. C. Andrade, C. Alonso, M. Acha and B. Malric, Cem. Concr. Res., 1996, 26, 405.25. Y. Schiegg, F. Hunkeler and H. Ungricht, ‘The effectiveness of corrosion inhibitors

– a field study’, Proc. IABSE Congress ‘Structural Engineering for Meeeting UrbanTransportation Challenges’, Lucerne 18–21. Sept. 2000, (on CD). (See also thisvolume Ch. 18.)

26. B. Elsener, M. Bürchler, F. Stalder and H Böhni, ‘A migrating corrosion inhibitorblend for reinforced concrete – Part 2: inhibitor as repair strategy’, Corrosion, 2000,56, 727.

27. J. P. Broomfield, ‘Results of long term monitoring of corrosion inhibitors applied tocorroding reinforced concrete structures, CORROSION 2000, paper 0791, NACEInternational Houston (TX) USA.

28. J. P. Broomfield, ‘The pros and cons of corrosion inhibitors’, Constr. Repair, July/August 1997, 16.

29. J. Tritthart, ‘Transport of corrosion inhibitors in concrete’, Proc. COST 521 WorkshopCorrosion of Steel in Reinforced Concrete Structures, 28–31 August 2000, ed. T. D.Sloan and P. A. M Basheer (eds.), The Queens University Belfast, 2000, 289–300.

30. A Eydelnant, B. Miksik and L. Gelner, ‘Migrating corrosion inhibitors for reinforcedconcrete’, ConChem J., 1993, 1, 38–42.

31. B. Elsener and H. Böhni, ‘Half cell potential measurements – from theory to conditionassessement of RC structures’, Proc. Int. Conference ‘Understanding CorrosionMechanisms of Metals in Concrete – A Key to Improving Infrastructure Durability’,MIT, Cambridge, USA, 27–31 July 1997, paper No. 3.

32. B. Elsener, ‘Half-cell potential mapping to assess repair work on RC structures’,Constr. Build. Mater. 2001, 15, 133–139.


185

15.1 Introduction

The corrosion of reinforcement in concrete is the most important cause ofpremature failure of reinforced concrete structures world-wide and becameof great interest in the late 1980s and early 1990s when its huge economicand social impact was pointed out [1].

Steel reinforcements in concrete structures are in passive conditions, thatis protected by a thin oxide layer, promoted by the concrete alkalinity. Inthese conditions, carbon steel in concrete at pH 13 behaves as stainless steelsin contact with fresh water, hence, like stainless steel, reinforcement cansuffer corrosion attack when the passivity is destroyed. This can occur in twoways: either due to carbonation of the concrete, that is reaction of cementpaste with carbon dioxide present in the atmosphere, which lowers the pHand causes general corrosion; or due to the presence of chlorides at the steelsurface in concentrations higher than the critical one, which is in the range0.4–1% by cement weight. Chlorides can be added erroneously to concretein the mix water or in aggregates (nowadays this is strictly forbidden), or canpenetrate by diffusion, for example in highway viaducts where de-icing saltsare employed, or in marine structures. Local passivity destruction by chloridescauses pitting corrosion, with a mechanism similar to that observed forstainless steels [1].

The prevention of reinforcement corrosion is primarily achieved in thedesign phase by using high quality concrete and adequate cover. Additionalprevention methods are adopted when severe environmental conditions occuron structures requiring a very long service life, as well as during repair andrehabilitation [1].

Among available methods, corrosion inhibitors can offer a simple andcost effective technique [2]. Inhibitors can be divided into two groups: mixed-in inhibitors directly added to fresh concrete for new structures and migratinginhibitors, which can penetrate into the hardened concrete and are usuallyadopted in rehabilitation. Mixed-in inhibitors have been studied since the

15Mixed-in inhibitors for concrete structures

F. B O L Z O N I, G. F U M A G A L L I, L. L A Z Z A R I,M. O R M E L L E S E and M. P. P E D E F E R R I

Politecnico di Milano, Italy



1950s, and have been commercially available since the 1970s [1–5], whilemigrating inhibitors for concrete structures were proposed in the last 20years, in connection with the increasing interest in rehabilitation and repair.

Nowadays, there are several admixtures available on the market [2]:inorganic compounds based on nitrites, especially used as additives [3–6],and sodium monofluorophosphate used as a migrating inhibitor [7]; organiccompounds based on mixtures of alkanolamines, amines or aminoacids [8–12], proposed both as mixed-in and migrating inhibitors; and emulsions ofan unsaturated fatty acid ester of an aliphatic carboxylic acid and a saturatedfatty acid [13], proposed as a mixed-in inhibitor.

Nitrite-based inhibitors are considered to be the most effective productsavailable on the market for the protection and prevention of chloride-inducedcorrosion: they have been studied since the 1960s both in the laboratory andin field tests and several applications have confirmed their effectiveness.Nitrite acts as a passivator, due to its oxidising properties, and its inhibitiveeffectiveness is related to the nitrite/chloride molar ratio, which should be atleast between 0.7 and 1 to prevent corrosion [2–6]. Concerns relate to theirtoxicity and solubility, and the possibility that they may cause an increase ofcorrosion rate if underdosed. For these reasons, in the last 20 years moreinterest has been given to new organic-based products.

Organic corrosion inhibitors act by adsorption on the metal surface, forminga thin organic layer that may inhibit both anodic and cathodic processes; forthis reason they are considered to be mixed inhibitors [10, 13–15]. Laboratorytests showed conflicting results about the efficiency of these products, bothin solution and in concrete; frequently, the test conditions and minimumeffective inhibitor concentration are not well defined. Moreover, because oftheir recent introduction and the very few field applications, there is notsufficient reliable data on their long term efficiency. What is more, the lackof a standard procedure to evaluate the effectiveness of these products makesit very hard to compare the results from the various experiments [12, 16].

Solution tests show a positive effect on corrosion initiation time and areduction of corrosion rate only in the presence of a high concentration ofcorrosion inhibitor (about 100 g L–1) and with chloride contents of up to1 mol L–1 [17, 18]. However, such a high inhibitor dosage may negativelyinfluence the properties of the concrete (workability, setting time or compressivestrength). Tests on samples of concrete containing chlorides in dosages higherthan 1% with respect to the weight of cement, and on samples subjected toponding cycles with a 3.5% NaCl solution, show that commercial organiccorrosion inhibitors, when added with the recommended dosages, increasethe initiation time [2, 11–13, 20], but the effect on critical chlorides thresholdis not clear [2, 20, 21]; most of the works presented in the literature reportthat organic inhibitors do not significantly affect the corrosion rate of eithermortar or concrete [2, 11, 19, 20].


Mixed-in inhibitors for concrete structures 187

This paper deals with the effectiveness of commercial mixed-in inhibitorson the corrosion of rebars embedded both in chloride-contaminated concreteand in carbonated concrete. Four different commercial inhibitors have beenconsidered: three organic corrosion inhibitors, two being amine- andalkanolamine-based (A, B) and a fatty acid emulsion (C) and, for comparison,a commercial nitrite-based product (N) were added to the concrete mixturein the concentration suggested by the manufacturers. Control samples werealso cast for comparison. The inhibitive effectiveness has been evaluated bymeans of two electrochemical parameters: the free corrosion potential andthe corrosion rate (determined by the polarisation resistance method).

15.2 Service life

The service life of a concrete structure can be divided into two periods:initiation and propagation, according to Tuutti’s classical model [22]. Duringthe initiation period, the corrosion rate of steel is negligible but meanwhilethe characteristics of the concrete are changing, promoting the breakdown ofsteel passivity. Once corrosion has started (the propagation period), twomain consequences occur: corrosion of the reinforcement and spalling ofthe concrete cover, once a maximum penetration depth has been obtained(Fig. 15.1).

15.2.1 Carbonation

Corrosion initiation corresponds to the time that the carbonation front takesto reach the external rebars. The depth of the carbonated layer, x, increaseswith time following a parabolic law:

x = k÷ t (15.1)

where x, t and k are, respectively, layer thickness, time and the carbonationcoefficient, which depends on concrete porosity (i.e. water/cement ratio) andon environmental conditions (i.e. relative humidity and temperature). Servicelife can be increased both by increasing the carbonation coefficient k or, oncecarbonation reaches the rebar level, by decreasing the corrosion rate.

Corrosion

Maximumpenetration

Initiation Propagation

Time

15.1 Concrete structure service life: Tuutti’s model [22].



15.2.2 Chlorides

In the case of chloride-induced corrosion, the universally accepted approachconsiders only the corrosion initiation period, that can be calculated byapplying Fick’s second law for non-stationary diffusion. If we suppose thatthe chloride concentration at the concrete surface (Cs) is constant with time,and that the ‘effective’ chloride diffusion coefficient (Dce) does not vary withtime and space, i.e. concrete is homogeneous, Fick’s second law presents ananalytical solution:

C C C C xD t

x = – ( – ) 1 – erf2

s s 0ce

Ê

ËÁˆ

¯ (15.2)

where C0 and Cx correspond to the total chloride content (by cement orconcrete weight) at time t = 0 and at general time t, at depth x from theconcrete surface; this equation allows calculation of chloride profiles withtime and then estimation of the time that chlorides take to reach the criticalconcentration at the reinforcement level.

To hinder chloride-induced corrosion, since it is not possible to slowdown chloride diffusion, it is necessary to increase the critical chlorideconcentration. This may be done by adding substances directly into the freshconcrete that inhibit chloride-induced corrosion, thus increasing the servicelife.


15.3.1 Samples and materials

Three different series (Table 15.1) of concrete samples were cast in order tosimulate both carbonation and chloride-induced corrosion. In particular, forchloride-induced corrosion, two possibilities have been considered: chloridesdirectly added to the mixture and chlorides diffusing from outside.

Concrete was mixed with 367 kg m–3 of cement (CEM II A/L 42.5R), 0.6w/c ratio and 1770 kg m–3 of limestone aggregate of 12 mm maximumdiameter. Chlorides were added to the mixing water in dosages of 0.8% and1.2% by weight of cement. After two days in the mould, the concrete was

Table 15.1 Tested series and corrosion conditions

Series Conditions

1 Chlorides in the mix: 0/0.8/1.2% by weight of cement2 Chlorides from outside: ponding cycles3 Carbonation



cured until 28 days had elapsed in a wet environment (95% rh). Samples forcarbonation were moved to the carbonation chamber (65% rh) after 3 days ofexposure at 95% rh. Full penetration of carbonation was detected by meansof phenolphthalein tests on 10 cm cubic samples. For each condition, 10 cmcubic samples were cast in order to check the compressive strength: themean value of three samples ranged from 40 MPa to 43 MPa, both forconcrete with and without chlorides; the addition of inhibitors did not adverselyinfluence the compressive strength.

Inhibitors were added following the manufacturers’ recommendations (Table15.2).

Two carbon steel rebars were placed in each specimen, 10 mm in diameterand 29 cm length. The ends of each rebar were coated with a heat shrinkablesleeve, so that only a length of 21 cm was exposed to the concrete. The netrebar surface area exposed to concrete was 66 cm2. The cover was 20 mm.A thin wire of mixed metal oxide (MMO) activated titanium, placed neareach rebar, was used as a reference electrode [23, 24) and 3 AISI 304 stainlesssteel wires (2 mm in diameter) were embedded in each specimen as a counter-electrode for polarisation resistance measurements (Fig. 15.2 and 15.3). Pondingsamples were equipped on the top with an appropriate container for NaClsolution.

Table 15.2 Inhibitor descriptions and dosages

Inhibitor Kind of inhibitor Dosage(kg m–3)

A Amines and alkanolamines (liquid) 10B Amines and alkanolamines (liquid) 1.6C Fatty acid emulsion (liquid) 5N Nitrite based (solution 30%) 10

Ti electrode

Carbonsteel rebar

Stainlesssteel wire

20

10

20

50

50 50 50 50

200

15.2 Design of concrete specimen: frontal view (dimensions in mm).



15.3.2 Exposures

Chloride-containing and carbonated samples (series 1 and 3) were exposedout of doors in Milan. Ponding, i.e. accelerated chloride penetration (series2), has been carried out using three-week cycles: one week of wetting theconcrete surface with 1 L of 3.5% NaCl solution, and two weeks of drying.

15.3.3 Corrosion tests

The free corrosion potential and polarisation resistance (Rp) of each rebarwere monitored. Corrosion potential was measured versus a saturated calomelelectrode (SCE) put in contact with the concrete surface by means of a wetsponge. Rp was measured using the linear polarisation technique [25], byapplying a potential scan rate of 10 mV per minute in the range ±10 mV withrespect to the free corrosion potential. The mean corrosion rate (mm peryear) was calculated by means of the Stern–Geary relationship: icorr = 1.17 ¥C/Rp where Rp is the measured polarisation resistance (evaluated from theslope of the potential/current density curve) and the constant C is assumed tobe equal to 26 mV [26].

15.3.4 Chloride concentration

Concrete cores, taken from the samples, have been ground and dissolved innitric acid. The chloride concentration was determined by potentiometrictitration with AgNO3 (0.01M). The accuracy of the measurements was ±0.01%by weight of cement.

15.4 Results

15.4.1 Chloride in the mix

Figures 15.4–15.9 present the results of 600 days of tests of free potentialcorrosion (vs SCE) and polarisation resistance as a function of time. The

Heat shrinkablesleeve Reference

electrode

25040 10

15.3 Design of concrete specimen: lateral view (dimensions in mm).



keys show the kind of inhibitor, A, B, C and N (W indicates concrete sampleswithout inhibitor), and chloride concentration (0, 0.8% and 1.2%).

During the period of exposure to the atmosphere, in concrete sampleswithout chlorides, the free corrosion potential ranged from –200 to 0 mV(SCE) for all types of inhibitors while, in samples with 0.8% and 1.2%chloride by weight of cement, potentials were between –400 mV and–50 mV (SCE), and –450 mV and –100 mV (SCE), respectively. Potentialfluctuations were mainly due to seasonal variations, i.e. change in temperatureand relative humidity: the highest values probably correspond to dry periods.

Polarisation resistance showed small fluctuations during the whole testperiod for all chloride concentrations. The Rp values were calculated considering

Po

ten

tial

[m

V (

SC

E)]

100

0

–100

–200

–300

–400

–500

Summer Winter Summer

1W0 1A0 1B0 1C0 1N0

0 200 400 600 800Time (d)

15.4 Free corrosion potential vs SCE: chloride-free specimens.

Po

lari

sati

on

res

ista

nce

(W m

2 )

1000

100

10

1


1W0 1A0 1B0 1C0 1N0

0 100 200 300 400 500 600 700Time (d)

15.5 Polarisation resistance: chloride-free specimens.



the whole rebar surface area exposed to concrete (66 cm2), although it isknown that, in the presence of chlorides, localised corrosion occurs. It maybe assumed that the corrosion rate is negligible if its average value is lowerthan a threshold value (1–2 mm year–1), according to [27]; that is, the polarisationresistance is higher than 10–20 W m2.

The average values are different for the three chloride concentrations; insamples without chloride Rp is 100 W m2, i.e. the corrosion rate is negligible,much less than 1 mm year–1. Samples containing 0.8% chloride display lowerpolarisation resistance values than chloride-free concrete, even though themeasured values are higher than 10 W m2; except for one specimen; it must

Po

ten

tial

[m

V (

SC

E)]

100

0

–100

–200

–300

–400

–500


1W08 1A08 1B08 1C08 1N08

0 200 400 600 800Time (d)

Po

lari

sati

on

res

ista

nce

(W

m2 )

1000

100

10

1


0 100 200 300 400 500 600 700Time (d)

1W08 1A08 1B08 1C08 1N08

15.6 Free corrosion potential vs. SCE: 0.8% chloride by weight ofcement.

15.7 Polarisation resistance: 0.8% chloride by weight of cement.



be emphasised that 0.8% chloride by weight of cement is in the range generallyconsidered to be the critical chloride content, i.e. 0.4–1%; so corrosion attackmay or may not initiate.

Finally, samples containing 1.2% chloride show very low polarisationresistance values of 2–5 W m2, i.e. significant corrosion rates. Only in thepresence of inhibitor A (amine- and alkanolamine-based) and inhibitor N(nitrite-based) do rebars display Rp values higher than 10 W m2 for the wholetest period.


1W12 1A12 1B12 1C12 1N12

0 200 400 600 800Time (d)

Po

ten

tial

[m

V (

SC

E)]

100

0

–100

–200

–300

–400

–500

15.8 Free corrosion potential vs. SCE: 1.2% chloride by weight ofcement.


1W12 1A12 1B12 1C12 1N12

0 100 200 300 400 500 600 700Time (d)

Po

lari

sati

on

res

ista

nce

(W

m2 )

1000

100

10

1

15.9 Polarisation resistance: 1.2% chloride by weight of cement.



15.4.2 Ponding tests

Figures 15.10–15.11 show the variation in time of free corrosion potentialand polarisation resistance of rebars in samples subjected to ponding cycles.Up to 30 ponding cycles, each of 3 weeks duration, were carried out.

The initial free corrosion potential is about –100 mV (SCE) in all cases,and it remains almost constant until falling to a low value of –400 mV(SCE). This occurred during the 11th cycle for the sample without inhibitor(2W), compared with the 7th cycle in the presence of inhibitor A, the 11th

cycle for inhibitor B, and the 17th cycle for inhibitor N (Fig. 15.10). Thepolarisation resistance is about 100 W m2 to start with, then falls to a valueclose to 80 W m2. Only in the presence of inhibitor A does it reach values

2W 2A 2B 2C 2N

Po

ten

tial

[m

V (

SC

E)]

100

0

–100

–200

–300

–400

–5000 100 200 300 400 500 600 700

Time (d)

15.10 Free corrosion potential vs. SCE: specimens subjected toponding cycles.

Po

lari

sati

on

res

ista

nce

(W

m2 )

1000

100

10

1

2W 2A 2B 2C 2N

0 100 200 300 400 500 600 700Time (d)

15.11 Polarisation resistance: specimens subjected to ponding cycles.



lower than 10 W m2, but, after 3 cycles, the polarisation resistance returns tohigh values.

After 14 ponding cycles a core was taken for each sample and the chlorideprofile of the sample was determined. The chloride content at a depth of1.5 cm was lower in concrete containing organic inhibitors, especially in thepresence of the fatty acid emulsion (inhibitor C). At the rebar level (2 cm)the chloride concentration was 0.5–1% (± 0.01%) by weight of cement.

15.4.3 Carbonated concrete tests

Figures 15.12 and 15.13 show trends in free corrosion potential and polarisationresistance for samples subjected to accelerated carbonation (series 3).

Carbonation Outside exposure 2W

3A

3B

3C3N

0 200 400 600 800Time (d)

Po

ten

tial

[m

V (

SC

E)]

100

0

–100

–200

–300

–400

–500

–600

–700

Carbonation Outside exposure 2W

3A

3B

3C3N

0 200 400 600 800Time (d)

Po

lari

sati

on

res

ista

nce

(W

m2 )

1000

100

10

1

15.12 Free corrosion potential vs. SCE: carbonation.

15.13 Polarisation resistance: carbonation.



Full carbonation was verified on concrete cubes with phenolphthaleintests in the first 200 days. Just after being exposed outside (Milan atmosphere),the free corrosion potential decreased and became stable at a constant valueof between –400 mV and –500 mV (SCE). Fluctuations in free corrosionpotential depend upon the weather conditions, i.e. relative humidity andtemperature. Polarisation resistance shows similar behaviour, decreasing tobelow 10 W m2 once the samples were exposed outside.

15.5 Discussion

With respect to the service life of concrete structures, inhibitors can act indifferent ways: they may delay the initiation of corrosion, by increasing thecritical chloride content or slowing down penetration by chlorides orcarbonation (this is not an electrochemical effect), or, even once corrosionhas started, they may reduce the corrosion rate. The following discussionwill take into account all of these different aspects.

15.5.1 Chloride-induced corrosion

Critical chloride content

Free corrosion potential and polarisation resistance measurements on 0.8%chloride samples showed that corrosion attack is negligible in the presenceof inhibitor A and inhibitor N, and is low in the other cases. This is probablydue to both the positive effect of the inhibitors and because a chlorideconcentration of 0.8% by weight of cement is in the range of 0.4–1%, usuallyconsidered as the critical chloride content for corrosion initiation in alkalineconcrete exposed to the atmosphere.

In samples with 1.2% chloride by weight of cement, both potential andpolarisation resistance measurements confirmed that corrosion occurs insamples with inhibitors B and C, while those with inhibitors A and N showeda low to negligible corrosion rate. It must be pointed out that the presence ofchlorides causes a localised attack. So, it is necessary to verify the actualarea affected by the corrosion attack; in fact, the lower the area, the higherthe penetration. But this survey can be done only at the end of the experiment,by visual inspection of the steel rebar surface, after destroying the concretesample.

In concrete samples subjected to accelerated chloride penetration, a chlorideconcentration of 0.5 to 1% by weight of cement was reached after 14 cyclesat the rebar level (chloride profiles are reported in Table 15.3).

These tests do not allow a critical chloride content to be defined for all thecommercial inhibitors. From the results, it may be concluded that only in thepresence of inhibitor A is the critical content equal to or higher than 1.2%,while in the other cases it ranges from 0.8 to 1.2%. In the literature, no clear



effect on the critical chloride threshold has been reported for organic inhibitors[2, 19, 20, 28].

Nitrite-based inhibitors must be treated separately. In samples with chloridecontents of 0.8 and 1.2% by weight of cement, that display negligible andlow corrosion rates, the molar ratio [NO ]2

– /[Cl–] is 0.67 and 0.45, respectively.These results are in accordance with references [2–3]; nitrite-based inhibitorseffective if the molar ratio is 0.7–1, while if the ratio is lower than 0.7,corrosion attack can occur.

Chloride diffusion

Commercial corrosion inhibitors may be effective also with respect to chloridediffusion [13, 20], although this is not an electrochemical effect. Diffusioncoefficients were determined on the basis of a non-linear least squares regressionanalysis of the chloride profiles by means of equation (15.2). Diffusioncoefficients are lower in samples with inhibitors than in those without (Table15.4). The maximum reduction in the diffusion coefficient occurs in samplescontaining inhibitor C (fatty acid emulsion), by 50% with respect to thesample without inhibitor; similar results can be found in references [13, 20].This effect is probably due to the formation of a hydrophobic layer within

Table 15.3 Chloride profile inside specimens after 14 ponding cycles(accuracy ± 0.01%)

Chloride concentration (% by weight of concrete)

Depth (cm) W A B C N

0.5 0.44 0.36 0.44 0.5 0.471.5 0.25 0.16 0.17 0.13 0.312.6 0.06 0.07 0.04 0.04 0.063.7 0.02 0.07 0.07 0.05 0.044.7 0.00 0.00 0.00 0.00 0.00

Table 15.4 Chloride diffusion coefficients evaluated bynon-linear least squares regression of chloride profiles(Table 15.3)

Inhibitor Diffusion coefficient(108 cm2 s–1)

W 7.2A 6.9B 4.5C 3.2N 6.1



the pores [2]. If this trend is confirmed by further measurements, under thesame conditions of chloride concentration, concrete cover and exposure,inhibitor C will double the service life of the concrete structure.

Corrosion rate

In Fig. 15.14 the performance of the four commercial inhibitors in terms ofmean corrosion rate (mA m–2) for various chloride concentrations are comparedwith those without inhibitor. Corrosion rate has been calculated by polarisationresistance data, considering all the rebar surface area exposed to concrete(66 cm2). If corrosion rate values are lower than 1 mA m2, i.e. 1 mmyear–1, corrosion is negligible [27]. In samples with chlorides of 0.8% bycement weight there are no significant differences in corrosion rate with orwithout commercial inhibitors, except with inhibitor C that presents corrosionrate higher than 1 mA m–2.

With a chloride content of 1.2%, rebars in samples containing inhibitor A(amine- and alkanolamine-based) show a reduction in corrosion rate of about30% with respect to those in samples without inhibitor. The other commercialinhibitors do not show a significant reduction.

A slight decrease in corrosion rate in concrete containing inhibitors canbe related to an increase in the electrical resistivity of the concrete, probablydue to a reduction in water content, as found for example in reference [21].On the other hand, most of the published literature reports that organicinhibitors do not significantly affect the corrosion rate in mortar or in concrete[2, 11, 19, 20]; as previously mentioned, different behaviour was observed insolution [2, 11].

15.14 Corrosion rate in specimen containing inhibitors A, B, C and Nand without inhibitor (W).

Co

rro

sio

n r

ate

(mA

m–2

)

10

1

0.1W A B

C N

1.20.8

0Chlorides (%)



15.5.2 Carbonation-induced corrosion

Carbonation penetration

Commercial inhibitors do not significantly reduce carbonation penetration(Table 15.5). Only inhibitor A shows a reduced carbonation coefficient, about5% lower than the one determined in concrete without inhibitor. Carbonationcoefficients were calculated by interpolating equation (15.1). Few data aboutthe influence of organic inhibitors on carbonation penetration have beenreported; amine-based inhibitors have not been found to affect carbonationpenetration [29].

Corrosion rate

In Fig. 15.15 rebar corrosion rates in carbonated concrete samples withinhibitors are compared with the average value obtained in a sample withoutinhibitor. A small inhibiting effect may be observed, and this has been evaluatedin terms of inhibition efficiency in Fig. 15.16.

The inhibition efficiency is defined as follows:

Table 15.5 Carbonation coefficients calculated afterthree months of accelerated test

Inhibitor Carbonation coefficient (mm year–1/2)

W 38.0A 36.2B 38.2C 41.4N 39.2

15.15 Mean corrosion rate in carbonated specimens containinginhibitors (A, B, C and N) and without inhibitors (W).

Co

rro

sio

n r

ate

(mA

m–2

)

8

7

6

5

4

3

2

1

0A B C N W



h = – corr

*corr

corr*

i ii

(15.3)

where icorr* and icorr are the average corrosion rates without and with inhibitor,

respectively.It is worth noting that only inhibitor B (amine- and alkanolamine-based)

displays an efficiency higher than 50%. The nitrite concentration suggestedby producers, corresponding in this case (inhibitor N) to 1% by weight ofcement, was not sufficient to reduce the corrosion rate significantly incarbonated concrete. This is in accordance with data in the literature thatreport a threshold value of 2–3% [30].

15.6 Conclusions

The commercial corrosion inhibitors studied in this work display a limitedeffectiveness both in chloride-contaminated and carbonated concrete.

All of the commercial organic inhibitors seem to slow down chloridepenetration. These results must be confirmed by further investigations.

The critical chloride content at which carbon steel in concrete exposed tothe atmosphere loses its passivity is between 0.4 and 1% by weight of cement.The commercial inhibitors increase the minimum threshold value of thecritical concentration range to 0.8%, but not as far as 1.2%. Only in thepresence of inhibitor A (amine- and alkanolamine-based) does the corrosionrate remain negligible even with 1.2% chlorides by weight of cement.

In carbonated concrete no significant effect of the commercial inhibitorshas been observed on carbonation penetration, i.e. corrosion initiation period.Only inhibitor B (amine- and alkanolamine-based) reduces the corrosionrate significantly (by 50%), increasing the propagation time.

Eff

icie

ncy

(%

)

80

70

60

50

40

30

20

10

0A B C N

15.16 Mean efficiency of inhibitors in carbonated specimens.



The results obtained with a nitrite-based commercial inhibitor are inagreement with the literature, which states that they are effective againstchloride attack if the molar ratio of NO2

–/Cl– exceeds 0.7 and againstcarbonation if the nitrite concentration exceeds 2–3%.

15.7 References

1. P. Pedeferri and L. Bertolini, La durabilità del calcestruzzo armato, McGraw-HillLibri Italia, Milano, 2000.

2. B. Elsener, Corrosion inhibitors for steel in concrete – an EFC state of the artreport, EFC, Number 35, 2001.

3. B. El-Jazairi and N. S. Berke, Eds., ‘The use of C.N. as a corrosion inhibitingadmixture to steel reinforcement in concrete’ Corrosion of reinforcement in concrete,Elsevier Applied Science, London, 1990, 571.

4. I. A. Callander and F. Gianetti, ‘A review on the use of C.N. corrosion inhibitor toimprove the durability of reinforced concrete’, The 2nd Annual Middle East Protection& Rehabilitation of Reinforced Concrete Conference, Dubai, 1996.

5. C. Andrade, C. Alonso and J. A. Gonzalez, ‘Some laboratory experiments on theinhibition effect of sodium nitrite on reinforcement corrosion’, Cement, concreteaggregates, 1986, 8(2), 110.

6. N. S. Berke and M. C. Hicks, ‘Protection mechanism of calcium nitrite’, Int. Conference:Understanding Corrosion Mechanism in Concrete; a Key to Improve InfrastructureDurability, Cambridge, 1997.

7. C. Andrade, C. Alonso, M. Acha and B. Malric, ‘Preliminary tests of Na2PO4F as acurative corrosion inhibitor for steel reinforcements in concrete’, Cement concreteres., 1992, 22, 869.

8. D. Bjegovic, L. Sipos and V. Uckrainczyk, ‘Diffusion of the MCI 2020 and 2000corrosion inhibitors into concrete’, Int. Conference Corrosion and Corrosion Protectionof Steel in Concrete, Sheffield, 1994, 865.

9. U. Mäder, ‘A new class of corrosion inhibitors for reinforced concrete’, Concrete,1999, 9, 215.

10. B. Elsener, M. Büchler and H. Böhni, ‘Corrosion inhibitors for steel in concrete’,EUROCORR, 1997, Trondheim, 469.

11. B. Elsener, M. Büchler and H. Böhni, ‘Organic corrosion inhibitors for steel inconcrete’, EUROCORR, 1999, Aachen.

12. B. Elsener, ‘A review of the performance of corrosion inhibitors for steel in concrete’,COST 521 Workshop, Belfast, 2000.

13. C. K. Nmai, S. A. Farrington and G. S. Bobrowsky, ‘Organic based corrosion inhibitingadmixtures for reinforced concrete’, Concrete Int., 1992, 4, 45.

14. A. Phanasgaonkar, B. Cherry and M. Forsyth, ‘Corrosion inhibition properties oforganic amines in simulated concrete environment: mechanism’, Int. Conference:Understanding Corrosion Mechanism in Concrete; a Key to Improve InfrastructureDurability, Cambridge, 1997.

15. A. Welle, J. D. Liao, M. Grunze, K. Kaiser, U. Maeder and N. Blank, ‘Interactionsof N, N-dimethylaminoethanol with rebar surfaces in alkaline and chlorine solutions’,Appl. Surf. Sci., 1997, 119, 185.

16. M. Yunovich and N. G. Thompson, ‘Performance of corrosion inhibiting admixturesfor structural concrete – assessment method and predictive modelling’, NACE, SanDiego, California, Paper 655, 2000.



17. C. M. Hansson, L. Mammoliti and B. B. Hope, ‘Corrosion inhibitors in concrete –Part II: Effect of chloride thresold values for corrosion of steel in synthetic poresolution’, Cement Concrete Res., 1999, 29, 1583.

18. V. Nobel-Pujol, T. Chaussadent and C. Fiaud, ‘Effects of organic and mineral inhibitorson the corrosion of reinforcements in hardened concrete’, 9th European Symposiumon Corrosion Inhibitors, Ferrara, 2000, 313.

19. E. Pazini, S. Leao and C. Estefani, ‘Corrosion inhibitors. Behaviour of NaNO2 andAmine-based Products in the Prevention and Control of Corrosion in ReinforcedConcrete’, NACE, Cancun, 1998.

20. M. Berra, F. Bolzoni, T. Pastore and P. Pedeferri, ‘Inibitori di Corrosione per Strutturein c.a’ Giornate Nazionali sulla corrosione e protezione, 4∞ed. AIM, Genova, 1999,293.

21. M. Salta, E. Pereira and P. Melo, ‘Influence of organic inhibitors on reinforcing steelcorrosion’, COST 521 Workshop, Belfast, 2000.

22. K. Tuutti, Corrosion of steel in concrete, Swedish foundation for concrete research,1982.

23. S. Ardizzone, A. Carugati and S. Trasatti, ‘Properties of thermally prepared iridiumdioxide electrodes’, J. Electroanal Chem., 1981, 126, 287.

24. K. Kinoshita and M. J. Madou, ‘Electrochemical measurements on Pt, Ir, and Tioxides as pH probes’, J. Electrochem. Soc., 1984, 131, 1089.

25. M. Stern and A. L. Geary, ‘Electrochemical polarisation I: a theoretical analysis ofthe slope of polarisation curves’, J. Electrochem. Soc. 1957, 104, 56.

26. J. A. González, A. Molina, M. L. Escudero and C. Andrade, ‘Errors in theelectrochemical evaluation of very small corrosion rates. I. Polarization resistancemethod applied to corrosion of steel in concrete’, Corrosion Sci., 1985, 25, 917.

27. C. Andrade, ‘Determination of chloride threshold in concrete’, COST 521 Workshop,Luxembourg, 2002, 108.

28. C. Alonso, C. Andrade, J. Fullea and J. Sanchez, ‘Accelerating test to ascertain theeffectiveness of corrosion inhibitors’ COST 521 Workshop, Belfast, 2000, 259.

29. B. Elsener, M. Büchler, F. Stalder and H. Böhni, ‘Migrating corrosion inhibitorblend for reinforced concrete : Part 1 – prevention of corrosion’, Corrosion, 1999,55, 1155.

30. C. Alonso and C. Andrade, ‘Effect of nitrite as a corrosion inhibitor in contaminatedand chloride-free carbonated mortars’, ACI – Mat. J., 1995, 3–4, 130.


203

16.1 Introduction

Corrosion inhibitors are chemical substances which, when added to thecorrosive environment at a suitable concentration, prevent, reduce, or eventuallystop corrosion occurring in various types of metals and alloys: they areusually used in aqueous solutions, but can also be used as volatile compoundsto prevent the corrosion of objects exposed to the atmosphere [1]. Inhibitorsare classified into anodic, cathodic, and mixed types, according to whichreaction is more influenced.

The use of corrosion inhibitors in concrete structures may concern bothnew and existing structures. In the first case, the inhibitors are admixed tothe fresh concrete and are intended to delay the initiation of corrosion resultingfrom both carbonation and chloride ingress. In the second case, they may beapplied on the surface of existing structures in which corrosion has alreadyinitiated and in this case they must penetrate in order to decrease the corrosionrate of the reinforcement. The inhibitors, or better admixtures containing theinhibitors to be used for reinforced concrete should not obviously adverselyaffect the prescribed concrete properties, e.g. mechanical properties andsetting time.

With regard to the use of corrosion inhibitors as a preventative, orsupplementary measure for new reinforced concrete structures, one of themajor problems is how to test their effectiveness and predict their influencein delaying the initiation of corrosion, and extending the service life. Tests insolution and in mortar specimens should not be considered suitable for theevaluation of the effectiveness of inhibitors since these environments arerather far from the real situation of the structures. Besides, experiments withconcrete specimens must take into account the difficulty of accelerating thecorrosion, necessary in order to obtain significant results in rather a shorttime.

16Effectiveness of mixed-in organic corrosion

inhibitors on extending the service life ofreinforced concrete structures

R. C I G N A, Consultant, Italy, A. M E R C A L L I,Autostrade S.p.A., Italy, L . G R I S O N I, Sika Italia, Italy,

and U. M Ä D E R, Sika A.G., Switzerland




Two reinforced concrete slabs with dimensions of 1500 ¥ 750 ¥ 300 mmwere produced making use of slag cement type CEM III/A. For each slab,the concrete cover was 30 mm, the water/cement ratio was equal to 0.5, andthe cement dosage was 350 kg m–3. One slab was prepared with an additionof a corrosion-inhibiting admixture containing amino alcohols and its corrosionbehaviour was compared with that of the other slab, without inhibitor [2].

Six measuring probes were embedded in each slab to monitor the state ofcorrosion of the reinforcement. The probes consisted of three electrodes,made of short pieces of bar, each having an area of 20 cm2. Two electrodeswere used for corrosion rate measurements (the auxiliary electrode being therebar itself) and were positioned at the level of the upper rebar, closer to thesurface onto which the salt solution was poured; the third electrode waspositioned at the level of the lower rebar, further from the surface exposed tothe aggressive solution, and was used to measure the macrocell emf inconnection through a voltmeter with one of the upper electrodes. The positionof one of the probes was inverted, in order to monitor the corrosion rate ofa bar in the passive state. Moreover, the corrosion potential of the rebar inapproximately the same positions as the internal probes was measured bymeans of a saturated calomel electrode (SCE) positioned over the uppersurface of the slab.

After four months of ageing, electrochemically forced chloride ingresswas initiated, with a current density of 0.01 mA cm–2. Three months later thecurrent was stopped, two cores were taken from each slab and the chlorideprofile was determined: the chloride concentration at the upper rebar levelwas approximately 0.06% w/w of concrete for both slabs. The test thencontinued with accelerated conditions achieved by cyclic ponding of theupper surface of the slabs with a saturated NaCl solution; ponding and normalexposure conditions were maintained for one and two weeks, respectively.

The results of the measurements carried out over approximately threeyears are reported in Tables 16.1 and 16.2, and shown in Figures 16.1–16.3.

16.3 Discussion and conclusions

From the examination of the results related to the slabs, the followingconclusions can be drawn.

∑ in the presence of the corrosion-inhibiting admixture both the corrosionrate and the macrocell emfs were very low, thus showing an excellentpassivation state even in the presence of growing amounts of chlorides;

∑ without inhibitor, noticeable corrosion attack started immediately after, oreven during, the forced ingress of chlorides;

∑ the rather wide fluctuations of the corrosion rate values measured in the


Effectiveness of m

ixed-in organic corrosion inhibitors205

Table 16.1 The emf and corrosion rate (CR) values for the slab without inhibitor; probe no. 1 is inverted

Time Probe no.1 Probe no. 2 Probe no. 3 Probe no.4 Probe no. 5 Probe no. 6(days)

emf CR emf CR emf CR emf CR emf CR emf CR(mV) (mm y–1) (mV) (mm y–1) (mV) (mm y–1) (mV) (mm y–1) (mV) (mm y–1) (mV) (mm y–1)

0 –179 2.7 141 4.1 142 2.7 259 5.4 120 6.3 213 8.13 –183 2.7 163 4.9 164 2.6 277 5.4 174 6.3 229 7.9

13 –303 2.6 262 9.7 301 6.1 378 9.6 253 11 42918 –245 2.3 257 6.6 337 3.4 228 6.1 153 7.3 251 9.124 –270 2.3 259 3.3 327 3.9 232 5.6 148 6.4 264 8.331 –254 2.4 255 5.3 314 6.1 201 6.3 130 6.6 263 9.652 –372 2.6 362 9.3 391 21 349 10 225 12 431 2467 –294 3.1 281 8.1 246 9.9 281 11 112 10 155 13

110 –296 2.9 170 13 133 9.3 205 13 55 9.7 272 17200 –165 2.7 191 16 148 7.6 135 12 154 7.9 208 16257 –62 2 390 14 379 9 382 12 401 9 573 16375 –406 2.1 406 12 267 6.6 329 10 405 11 470 13431 –442 2.9 403 21 322 11 342 17 442 12 448 24452 –387 3.9 354 19 307 7 214 17 289 11 299 23501 –320 2.9 340 19 290 8.9 220 16 156 7.9 266 20543 –360 4 330 20 266 11 247 17 190 8.3 287 19613 –136 2.8 130 8 150 5.4 110 5.6 78 3.1 6.9676 –132 1.6 122 5.1 138 3.9 120 4.4 100 2.9 3.6907 –117 4.7 161 43 138 37 78 34 100 26 92 54

1076 –104 3.7 176 12 156 62 50 44 37 115 571139 –119 2 198 9.1 181 6.9 103 6.3 95 3.6 126 8.11487 –39 1.4 88 7.3 58 6.9 26 4.9 36 3.3 68 8.3



ent in concrete206

Table 16.2 The emf and corrosion rate (CR) values for the slab with inhibitor; probe no. 1 is inverted

Time Probe no.1 Probe no. 2 Probe no. 3 Probe no.4 Probe no. 5 Probe no. 6days

emf CR emf CR emf CR emf CR emf CR emf CR(mV) (mm y–1) (mV) (mm y–1) (mV) (mm y–1) (mV) (mm y–1) (mV) (mm y–1) (mV) (mm y–1)

0 21 0.7 11 0.6 27 0.6 15 0.5 5 0.5 –20 0.47 12 0.7 13 0.6 18 0.5 8 0.5 3 0.5 –12 0.4

22 –6 0.7 15 0.7 –2 0.5 0 0.4 –1 0.4 –1 0.428 0.6 12 0.6 –4 0.5 1 0.4 –1 0.4 –5 0.477 –14 0.6 13 0.6 –5 0.5 3 0.4 –2 0.4 –8 0.384 0.6 15 0.6 –6 0.6 3 0.4 0 0.4 –5 0.498 –8 0.6 14 0.6 –7 0.5 2 0.4 –1 0.4 –7 0.4

140 10 0.6 –5 0.6 –11 0.5 3 0.5 –7 0.4 6 0.4196 –2 0.5 –3 0.6 –4 0.5 –2 0.4 6 0.4 –6 0.3217 7.1 0.9 –1.8 1.1 0 0.8 –30 0.7 –10 0.6 –8 0.5266 33 0.5 –1 0.9 40 0.8 –33 0.6 –27 0.8 –39 0.4308 30 0.7 –1 0.8 37 0.7 –37 0.6 –25 0.4 –32 0.4378 42 0.5 –1 0.7 19 0.5 –1 0.4 –2 0.3 –3 0.3441 38 0.3 –1 0.1 16 0.3 0.3 –3 0.2 –4 0.2672 –16 0.7 3 0.7 –20 0.6 –5 0.4 –5 0.3 0 0.3841 74 0.4 4 0.3 –43 0.4 0 0.2 0 0.1 3 0.1904 29 0.6 5 0.4 –23 0.4 –3 0.2 –1 0.2 8 0.2

1252 10 0.5 2 0.4 –18 0.4 0 0.3 3 0.2 6 0.2


Effectiveness of mixed-in organic corrosion inhibitors 207

slab without inhibitor, especially in comparison with the values determinedfor the probe in the passive state, are certainly due to variations in theenvironmental conditions (temperature and humidity) and also to the factthat, for a certain time, the ponding cycles were interrupted (at 600–700days, and since the day 1100; between 700 and 1100 days regular pondingwith saline solution was done). In any case, the corrosion rates for the fiveprobes situated at the level of the upper rebar are much higher than thoseof the control probe, thus indicating that the upper rebar is in corrosiveconditions;

Co

rro

sio

n p

ote

nti

al (

mV

)

0

–50

–100

–150

–200

–250

–300

–350

–400

–450

–500

Cl = 0.06%

Cl = 0.14% Cl = 0.14%

0 200 400 600 800 1000 1200Days

(a)

Co

rro

sio

n p

ote

nti

al (

mV

)

0

–50

–100

–150

–200

–250

–300

–350

–400

–450

–500

Cl = 0.08%

Cl = 0.10%

0 100 200 300 400 500 600 700 800 900 1000Days(b)

Cl = 0.13%

Cl = 0.06%

16.1 Reinforcement corrosion potential measured over the upperconcrete surface vs. SCE (a) without and (b) with inhibitor. The sixcurves refer to the rebar close to the positions of the six probesembedded in the slabs. The chloride content refers to the weight ofconcrete.



∑ similarly, the fluctuations of the macrocell emfs for all six probes (thecontrol probe gives inverted values of course), depend on the same variablesand finally show that there is corrosion in action for the upper rebar,nearest to the surface of the concrete ponded with the chloride solution:the values of +50 mV (average for the 5 test probes) and –50 mV for thecontrol probe (inverted, so that the upper electrode, close to the upperrebar, is corroding) are very high in comparison with those measured forthe six probes of the slab containing the inhibitor (average value 0 mV).

It is now difficult to predict by which factor the addition of the inhibitingmixture may delay the onset of corrosion of the reinforcement in the presenceof chlorides as aggressive agents. However, a rough preliminary calculation,

Co

rro

sio

n (mm

yr–1

)

0 200 400 600 800 1000 1200 1400 1600Days

(a)

15

10

5

0

Co

rro

sio

n (mm

yr–1

)

0 200 400 600 800 1000 1200 1400Days(b)

15

10

5

0

16.2 Internal probe corrosion rate values measured on slabs (a)without and (b) with inhibitor; the thick line refers to the invertedprobe.


Effectiveness of mixed-in organic corrosion inhibitors 209

based on the use of Fick’s second law of diffusion for a semi-infinite slab,may be done as follows:

c t x c erf xDt

( , ) = 1 – 2

0

È

ÎÍ

˘

˚˙ (16.1)

where c(t, x) is the chloride concentration at depth x and time t, co is thechloride surface concentration, erf is the error function and D is the diffusioncoefficient.

Assuming that the time to initiate corrosion for structures of similar concretemixes without inhibitor is 22 years for the slag cement concretes (a hypothesissuggested by Autostrade, Italy [3]) and that the diffusion coefficient, calculatedfor the slab without the inhibitor, is the same for the other slab, it may be

16.3 Internal probe macrocell emf measured on slabs (a) without and(b) with inhibitor; the thick line refers to the inverted probe.

emf

(mV

)

0 200 400 600 800 1000 1200 1400 1600

600

400

200

0

–200

–400

0 200 400 600 800 1000 1200 1400

Days(a)

Days(b)

emf

(mV

)

600

400

200

0

–200

–400



concluded (Table 16.3) that the presence of the tested inhibitors delays thetime to corrosion initiation by a factor of at least three. This calculation isbased on the use of the chloride content values determined in time in corestaken from the slabs; it must also be taken into account that the penetrationof the chlorides in the slab containing the corrosion inhibiting admixtureappears to be much lower than in the control slab; this means that the admixtureused has a positive synergic action in facilitating the passivation of the steelin the presence of chlorides while at the same time slowing their ingress intothe concrete.

16.4 References

1. B. Elsener, M. Büchler and H. Böhni, Corrosion of Reinforcement in Concrete, EuropeanFederation of Corrosion Publ. No. 25, Institute of Materials, London, 1998.

2. R. Cigna, A. Mercalli, G. Peroni, L. Grisoni and U. Mäder, Proc. Int. Conf. OnCorrosion and Rehabilitation of Reinforced Concrete Structures, Orlando, 1998, Publ.N. FHWA-SA-99-014.

3. R. Cigna, G. Familiari, F. Gianetti and E. Proverbio, Ind. Ital. Cemento, 1995, 703(10),577–582.

Table 16.3 Calculation of the corrosion initiation time (c is the chlorideconcentration at the rebar level)

Without inhibitor With inhibitortime 0 30 months

Co (weight %) 0.3 0.3C (weight %) 0.06 0.13D (cm2 s–1) 3.8 ¥ 10–9 3.8 ¥ 10–9

Corrosion initiation time 22 years >62 years


211

17.1 Introduction

Reinforced concrete buildings have a limited operating life because of corrosionof their steel reinforcement. The phenomenon involves not only bridges orhighway and marine infrastructures, still the most seriously affected, but alsopublic and private structures, churches, stadiums, monuments and others.Corrosion begins after concrete loses its protective properties, with respectto the steel, by means of two phenomena: ingress of chlorides (usually fromdeicing salt or a marine environment) and reduction of the pH of pore solutionpromoted by the reaction with carbon dioxide from the atmosphere (carbonationprocess).

The prevention of reinforcement corrosion is primarily achieved at thedesign stage by using high quality concrete and adequate cover. Additionalprevention methods are adopted when severe environmental conditions occuror on structures requiring a very long service life, as well as in rehabilitation.Among the available methods, corrosion inhibitors are attractive because oftheir low cost and easy handling, compared with other preventative methods.Inhibitors can be divided into two groups (1, 2): admixed inhibitors, directlyadded to fresh concrete for new structures, and migrating inhibitors, whichcan penetrate into hardened concrete and are usually adopted in rehabilitation.

While the first admixed corrosion inhibitors, inorganic nitrites, have beencommercially available since the 1970s [1–5], migrating corrosion inhibitorsfor concrete structures have been studied in the last 20 years. The proposedinhibitors include inorganic compounds (sodium monofluorophosphate) [6–8], organic mixtures (primarily amine-based compounds) [9–20], and organic-inorganic mixtures.

The effectiveness of migrating corrosion inhibitors is related to their abilityto reach the rebars [2, 12]. Inhibitor migration in concrete occurs by differentmechanisms: capillary suction, diffusion, and vapour phase transport [10,11]. The effectiveness of these mechanisms depends on concrete cover, porosity,water content, solubility and the volatility of the inhibitor. For example, a

17Migrating inhibitors on corrosion in

reinforced concrete

F. B O L Z O N I, G. F U M A G A L L I, L. L A Z Z A R I,M. O R M E L L E S E and M. P. P E D E F E R R I




high percentage of volatile compounds can increase the migration rate in theconcrete cover, but may also cause retro-migration towards the atmosphere.Low concrete cover can favour the penetration of inhibitors, but also ofaggressive agents (carbon dioxide and/or chlorides).

The real effectiveness of corrosion inhibitors in different situations, i.e. ofchloride content, exposure, etc., is not well defined. The published results onmigrating corrosion inhibitors are controversial, with respect to both the inhibitorpenetration and to the effectiveness on the corrosion rate. Summarising verybriefly, there are doubts on migration ability [10, 11, 19], or the inhibitors maypenetrate only in conditions (low concrete cover, less compact concrete) thatalso favour penetration of aggressive agents [6, 7]; probably not surprisingly,only tests carried out by the manufacturers show the good penetration ofmigrating corrosion inhibitors [14–16]. Some tests carried out by the inhibitorproducers show a reduction of corrosion rate using migrating corrosion inhibitorsin chloride-contaminated concrete [13, 14], while other conflicting test resultsshow their ineffectiveness [2, 12, 17–20]. Besides, test conditions and minimumeffective inhibitor dosage are not well defined in the literature.

There are concerns about the effectiveness of these products, since theirhomologous counterparts for fresh concrete (amine and alkanolamine basedadmixed inhibitors) show questionable inhibiting effect [21–23].

The aim of the present work was to verify the performance of two commercialorganic migrating corrosion inhibitors, amine-and alkanolamine-based, namedFM and DM; their ability to slow or to stop corrosion attack and to preventcorrosion initiation have also been studied.


17.2.1 Materials and specimens

The concrete mixture was prepared with cement CEM II A/L 42,5R andlimestone aggregate of 12 mm maximum diameter (Table 17.1). The concretewas cured for 28 days at 20 ∞C and 95% relative humidity (rh), except for thespecimens for carbonation, which were cured for only 3 days at 95% relativehumidity and then placed in a carbonation chamber (65% rh).

Table 17.1 Concrete mix design

Series 1, 3, 4 Series 2

Cement CEM II A/L 42.5R (kg m–3) 367 338Distilled water (L m–3 ) 220 220Water/cement (w/c) ratio 0.6 0.65Aggregates (kg m–3) 1770 1790Plasticiser – 0.6% by cement weight (kg m–3) 2.2 –Compressive strength (MPa) 43 30


Migrating inhibitors on corrosion in reinforced concrete 213

Two carbon steel rebars (Feb44k), 10 mm diameter, were placed in eachconcrete specimen (Fig. 17.1). Net rebar surface area exposed to concretewas 66 cm2. Activated titanium reference electrodes [24, 25] were placednear to each rebar, and three stainless-steel wires (2 mm in diameter) wereplaced in the specimens for corrosion rate measurements (linear polarisationresistance method, LPR).

The tested migrating inhibitors were amine- and alkanolamine-based. Thedosages were those recommended by the manufacturers, i.e. 400 g m–2 ofconcrete surface for inhibitor FM, and 250 g m–2 for inhibitor DM, increased30% approximately for the expected losses. The inhibitors were appliedtwice, with 4 months between the two applications.

Four series of concrete specimens (200 ¥ 200 ¥ 50 mm) were cast. Series1, with chloride in the mix (0.8 and 1.2% by cement weight), on whichmigrating inhibitors were applied only on specimens with 1.2% chloride,after 8 months and after 1 year; series 2, specimens without chlorides (w/c= 0.65), subjected to chloride ponding cycles and application of migratinginhibitors at the third and at the seventh month; series 3, carbonated concrete,on which inhibitors were applied after 8 months and after 1 year (i.e. 2months and 6 months after concrete carbonation); and finally, series 4, concretesubjected to chloride ponding cycles on which migrating inhibitors wereapplied after 7 months and after 1 year, before corrosion initiation. Forcomparison, for series 1, one specimen was cast for each chloride concentrationand corrosion inhibitor; instead, for series 2, 3 and 4, two specimens wereprepared for each corrosion inhibitor.

Specimens with mixed-in chloride were exposed out of doors in Milan.Specimens subjected to chloride penetration were subjected to alternatingponding cycles of 1 week wetting with 3.5% NaCl solution, followed by 2weeks of drying in the laboratory. Specimens for carbonation, after curing,

Ti electrode

Carbon steelrebar

Stainlesssteel wire

50 50 50 50

200

20

10

20

50

17.1 Reinforced concrete specimen (dimensions in mm).



were exposed in a chamber at 65% rh and 20 ∞C, in which for one hour eachday 95% CO2 was introduced. After carbonation, specimens were wettedand exposed out of doors in Milan.

17.2.2 Corrosion monitoring

Free corrosion potential was measured with respect to a SCE reference electrodeplaced in contact with the concrete surface by means of a wet sponge. Corrosionrate was evaluated through linear polarisation resistance measurements [1,26, 27]. Stainless-steel wires were used as counter electrodes, and the activatedtitanium electrode, placed near to the carbon steel bar, was used as thereference electrode (Fig. 17.1). Measurements were carried out by apotentiodynamic technique, scanning the potential from –10 mV to +10 mVwith respect to the free corrosion potential (Ecorr) with a scan rate of 10 mVmin–1. Polarisation resistance (Rp, in W m2) was evaluated from the slope ofthe linear zone of the potential/current density curve [27].

The corrosion rate (icorr) was calculated by means of the Stern–Gearyformula [26]:

i CRcorr

p = (17.1)

where the constant C is assumed to equal 26 mV.Carbonation was checked by measuring the potential of an embedded

activated titanium reference electrode with respect to a SCE reference electrodeplaced on the concrete surface [28]. Carbonation was also confirmed by aphenolphthalein test on cubic control specimens placed in the same carbonationchamber.

For chloride analysis, cores were extracted from concrete specimens; sliceswere cut from the cores and, after milling, the concrete powder was dissolvedwith nitric acid. The chloride content of the solutions was evaluated bypotentiometric titration with AgNO3 (0.01N).


17.3.1 Mixed-in chlorides

In concrete specimens exposed to the atmosphere (series 1), the corrosionbehaviour depends on the chloride concentration. In specimens with chlorides0.8% by cement weight, the free corrosion potential of the rebar was higherthan –300 mV vs SCE, although great seasonal variations were observed:the highest potential values were probably associated with dry periods(Fig. 17.2a). After 400 days, the polarisation resistance (Rp) was lower than10 W m2 (Fig. 17.3a), i.e. the corrosion rate was higher than 2 mA m–2, so was



not negligible [29]. It must be pointed out that a chloride content of 0.8% bycement weight is within the critical chloride concentration range, which isbetween 0.4–1%, so corrosion may or may not be initiated [1]. Migratinginhibitors were not applied on these specimens.

In specimens with a chloride content 1.2% by cement weight, a corrosionrate higher than 2 mm y–1 was measured immediately after atmosphericexposure (see Rp values lower than 10 W m2 in Fig. 17.3b). The potentialvalues were very scattered: the value was lower in 1.2% Cl– specimens in thefirst 300 day period (Fig. 17.2b). Migrating corrosion inhibitors were appliedto these specimens twice, after 8 months and 1 year of atmospheric exposure.The results were not so promising: the corrosion rate seemed to be unaffected


Po

ten

tial

[m

V (

SC

E)]

100

0

–100

–200

–300

–400

–5000 100 200 300 400 500 600

Time (d)(a)


Po

ten

tial

[m

V (

SC

E)]

100

0

–100

–200

–300

–400

–5000 100 200 300 400 500 600

Time (d)(b)

Inhibitor FMInhibitor DM

17.2 Free corrosion potential in series 1 specimens with chloridecontents of (a) 0.8% and (b) 1.2% by cement weight. Vertical dottedlines indicate the migrating inhibitor applications (only for 1.2%chlorides specimens).



by the presence of the inhibitors. The mean values of corrosion rate beforeany treatment and after the first and second applications were not very different(Fig. 17.4); the standard deviation decreased after the first application andincreased after the second one. Under these conditions migrating corrosioninhibitors are not effective in reducing the corrosion rate.

Measurements of the penetration of inhibitors into the concrete specimensare not available. As already mentioned, inhibitor penetration depends stronglyon the concrete properties and environment. In this case, the concrete is good(compressive strength 43 MPa) and the cover is 20 mm (not too thin). So,poor penetration of inhibitors may be expected. Parallel tests have, however,


Po

lari

sati

on

res

ista

nce

(W

m2 )

1000

100

10

10 100 200 300 400 500 600

Time (d)(a)


0 100 200 300 400 500 600Time (d)

(b)


Po

lari

sati

on

res

ista

nce

(W

m2 )

1000

100

10

1

17.3 Polarisation resistance (Rp) in series 1 specimens with chloridecontents of (a) 0.8% and (b) 1.2% by cement weight. Vertical dottedlines indicate the migrating inhibitor applications (only for 1.2%chlorides specimens).



been carried out with inhibitor FM applied to concrete specimens cast withthe same mixture proportions [30]. It was observed that the most effectivepenetration was due to capillary sorption (oven-dried samples). Nevertheless,in all experiments, the penetration of inhibitor was less than 20 mm. On theother hand, as previously reported, even if such inhibitors are able to reachthe rebars there is concern about the real effectiveness of amine andalkanolamine based products when used as admixed inhibitors [21–23, 31].

17.3.2 Accelerated chloride penetration

The corrosion behaviour of chloride-free specimens (w/c = 0.65) subjectedto chloride penetration cycles (1 ponding cycle consists of 1 week wettingwith 3.5% NaCl solution followed by 2 weeks drying in the laboratory) is

Mean valueStandard deviation

Before application After first After secondapplication application

(a)

Co

rro

sio

n r

ate

(mA

m–2

)

20

15

10

5

0


Before application After first After secondapplication application

(b)

Co

rro

sio

n r

ate

(mA

m–2

)

20

15

10

5

0

17.4 Corrosion rate in specimens with a chloride content of 1.2% bycement weight (series 1) with (a) inhibitor FM and (b) inhibitor DM.Mean, maximum and minimum values are shown, with standarddeviation.



shown in Fig. 17.5. After four ponding cycles, i.e. ~80 days, the free corrosionpotential and polarisation resistance decreased and the corrosion rate washigher than 5 mA m–2 for all specimens (Fig. 17.6). The application ofmigrating inhibitors was carried out after 3 months and 7 months. Again, theresults were not good; the corrosion rate increased after each corrosion inhibitorapplication (Fig. 17.6) and the standard deviation of the corrosion rate increasedgreatly. It must be remembered that the ponding cycles were continued 3weeks after the migrating inhibitor application and that the chlorideconcentration at the level of the rebars reached high values, approximately2% by cement weight after 10 ponding cycles (250 days). In the case ofponding, it can be confirmed that migrating corrosion inhibitors are noteffective. The results are in accordance with the experiments described in

0 100 200 300 400Time (d)

(b)


Po

lari

sati

on

res

ista

nce

(W

m2 )

1000

100

10

1

0.1

Po

ten

tial

[m

V (

SC

E)]

0

–100

–200

–300

–400

–500

–600


0 100 200 300 400Time (d)

(a)

17.5 Measurements of (a) free corrosion potential and(b) polarisation resistance in series 2 specimens subjected toponding cycles (w/c = 0.65). Vertical dotted lines indicate themigrating inhibitor applications, after corrosion initiation.



[20]: no reduction of corrosion rate after corrosion initiation was noticedwith amine- and alkanolamine-based migrating inhibitors.

According to the literature [11] and to the results obtained in the parallelexperiments described previously [30], low inhibitor penetration can beexpected under the present experimental conditions.

17.3.3 Carbonated concrete

After six months of accelerated carbonation, all specimens of series 3 wereexposed to the atmosphere. The steel rebar exhibited a low corrosion potential

Co

rro

sio

n r

ate

(mA

m2 )

80

70

60

50

40

30

20

10

0


Passive Before any After first After secondrebars application application application

(a)

Co

rro

sio

n r

ate

(mA

m2 )

80

70

60

50

40

30

20

10

0


Passive Before any After first After secondrebars application application application

(b)

17.6 Corrosion rate (mean, maximum and minimum values withstandard deviation are shown) for (a) inhibitor FM and (b) inhibitorDM in concrete subjected to chloride ponding (series 2, w/c = 0.65).



of about –500 mV (SCE) and a low polarisation resistance value, i.e. acorrosion rate higher than 5 mA m–2 (Fig. 17.7 and 17.8). Migrating corrosioninhibitors were applied after 8 months and 1 year (i.e. 2 months and 6months after concrete carbonation). The mean corrosion rate was reduced(see Fig. 17.8 and 17.9); the mean inhibition efficiency approached 50% forinhibitor DM after the second application. Nevertheless, scattering of theresults is evident, especially in the presence of inhibitor FM. Moreover, theresidual corrosion rate is not negligible; after the 2nd application (Fig. 17.7)

0 200 400 600 800Time (d)

(b)


Po

lari

sati

on

res

ista

nce

(W

m2 )

1000

100

10

1

Po

ten

tial

[m

V (

SC

E)]

100

0

–100

–200

–300

–400

–500

–600

–700


0 200 400 600 800Time (d)

(a)

Carbonation External exposure

Carbonation External exposure

17.7 Measurements of (a) free corrosion potential and (b) polarisationresistance in series 3 specimens subjected to carbonation (firstperiod) and then exposed to the atmosphere. Vertical dotted linesindicate the migrating inhibitor applications.



a corrosion rate higher than the threshold value, 1–2 mA m–2 [29], wasmeasured. These results are in agreement with the few available literaturedata [20]; according to this work, no effect was noticed on the corrosion rateafter exposure for one year.

In the case of carbonation, according to the literature, inhibitor penetrationis effective only after several weeks in dried specimens [11]. It is probablethat under the present experimental conditions, inhibitors did not reach therebar level and the effect on corrosion rate may be due to some change inconcrete resistivity.

Mean valueStandarddeviation

Alkaline Carbonated After first After secondconcrete concrete application application

(a)

Co

rro

sio

n r

ate

(mA

m2 )

20

10

0

Mean valueStandarddeviation

Alkaline Carbonated After first After secondconcrete concrete application application

(b)

Co

rro

sio

n r

ate

(mA

m2 )

20

10

0

17.8 Corrosion rate (mean, maximum and minimum values withstandard deviation are shown) for (a) inhibitor FM and (b) inhibitorDM in carbonated concrete (series 3) after migrating inhibitorapplication.



17.3.4 Chloride penetration – application of inhibitorbefore corrosion initiation

Results obtained on applying migrating inhibitors before corrosion initiation(series 4) are reported in Fig. 17.10. Twenty-seven ponding cycles werecarried out. For purposes of comparison, measurements carried out on referencespecimens without corrosion inhibitor treatment are also shown.

Lowering of the potential was observed for some rebars (Fig. 17.10a):including three of the four specimens without inhibitors (at the 10th, 18th and21st cycle), two of the four specimens with inhibitor FM (at the 10th and 24th

cycle) and one of the four specimens with inhibitor DM (at the 23rd cycle).Nevertheless, the polarisation resistance was in all cases higher than10 W m2, i.e. the corrosion rate was negligible for all the rebars (Fig. 17.10b).The chloride concentration at the rebar level, measured after 27 pondingcycles, that is 650 days, was between 0.7–1.4% by cement weight. Theapplication of these migrating corrosion inhibitors seems to delay corrosioninitiation; nevertheless, it is worth noticing that the corrosion rate is stillnegligible on rebars with low potential values. For these reasons furtherinvestigations are needed.

17.4 Conclusions

Commercial amine- and alkanolamine-based migrating inhibitors, appliedaccording to the manufacturer’s instructions, show no appreciable reductionof corrosion rate in chloride-contaminated concrete, either on adding chloridesto the mixture or in conditions of chloride penetration from the outside.

In the case of carbonation, migrating inhibitors are able to reduce themean value of the corrosion rate, nevertheless the scatter of results is highand the residual corrosion rate is not negligible.


Mea

n e

ffic

ien

cy (

%)

100

50

0After first

applicationAfter secondapplication

17.9 Carbonated concrete (series 3): mean inhibition efficiency.



Concerning the delay in corrosion initiation in specimens subjected tochloride penetration after migrating inhibitor treatment, the available resultsare inconclusive and further investigations are needed.

17.5 References

1. P. Pedeferri and L. Bertolini, La durabilità del calcestruzzo armato, McGraw-HillItalia, Milan 2000.

2. B. Elsener, Corrosion inhibitors for steel in concrete – an EFC state of the artreport, EFC, Number 35, 2001.

3. C. Andrade, C. Alonso and J. A. Gonzalez, ‘Some laboratory experiments on the

17.10 Measurements of (a) free corrosion potential and (b)polarisation resistance in series 4: specimens subjected to chlorideponding. Vertical dotted lines indicate the migrating inhibitorapplications.

Po

ten

tial

[m

V]

(SC

E)

0

–100

–200

–300

–400

No inhibitorInhibitor FMInhibitor DM

0 100 200 300 400 500 600 700Time (d)

(a)

No inhibitorInhibitor FMInhibitor DM

0 100 200 300 400 500 600 700Time (d)

(b)

Po

lari

sati

on

res

ista

nce

(W

m2 )

1000

100

10

1



inhibition effect of sodium nitrite on reinforcement corrosion’, Cement, concreteaggregates, 1986, 8(2), 110.

4. N. S. Berke and T. G. Weil, ‘World-wide review of corrosion inhibitors in concrete’,Advances in Concrete Technology, Athens, CANMET, 1992, 899.

5. B. El-Jazairi, N. Berke and W. R. Grace, ‘The use of C.N. as a corrosion inhibitingadmixture to steel reinforcement in concrete’, in Corrosion of reinforcement inconcrete, Elsevier Applied Science, London, 1990, 571.

6. C. Alonso, C. Andrade, C. Argiz and B. Malric, ‘Preliminary tests of Na2PO4F as acurative corrosion inhibitor for steel reinforcements in concrete,’ Cement ConcreteRes, 1992, 22, 869.

7. C. Andrade, C. Alonso, M. Acha and B. Malric, ‘Na2PO4F as inhibitor of corrodingreinforcement in carbonated concrete’, Cement Concrete Res., 1996, 26, 405.

8. A. Raharinaivo and B. Malric, ‘Performance of MFP for inhibiting corrosion of steelin reinforced concrete structures’, International Conference on Corrosion andRehabilitation of Reinforced Concrete Structures, Orlando, 1998.

9. V. Nobel-Pujol, T. Chaussadent and C. Fiaud, ‘Effects of organic and mineral inhibitorson the corrosion of reinforcements in hardened concrete’, 9∞ SEIC (EuropeanSymposium on Corrosion Inhibitors), Ferrara, 2000, 313.

10. J. Tritthart, ‘Transport of corrosion inhibitors in cement paste and concrete’, COST521 Workshop, Belfast, 2000, 203.

11. J. Tritthart, ‘Transport of the corrosion inhibitor SIKA Ferrogard 903 in cementpaste and concrete’, COST 521 Workshop, Tampere, 2001, 191.

12. C. L. Page, ‘Aspects of the performance of corrosion inhibitors applied to reinforcedconcrete’, 9∞ SEIC (European Symposium on Corrosion Inhibitors), Ferrara, 2000,261.

13. U. Maeder, ‘A new class of corrosion inhibitors for reinforced concrete’, Int. Conf.on Corrosion and Corrosion Protection of Steel in Concrete’, Sheffield, 1994, 851.

14. D. Bjegovic, L. Sipos, V. Uckrainczyk and B. Micksic, ‘Diffusion of the MCI 2020and 2000 corrosion inhibitors into concrete’, Int. Conf. on ‘Corrosion and corrosionprotection of steel in concrete’, Sheffield, 1994, 865.

15. D. Bjegovic and B. Miksic, ‘Migrating corrosion inhibitor protection of concrete’,Mater. Perf., 11, 1999, 52.

16. D. Bjegovic, ‘Accelerating testing of migrating corrosion inhibitors effectiveness’,COST 521 Workshop, Belfast, 2000, 235.

17. B. Elsener, M. Büchler and H. Böhni, ‘Organic corrosion inhibitors for steel inconcrete’, Eurocorr ’ 99, Aachen, 1999.

18. B. Elsener, M. Büchler and H. Böhni, ‘Corrosion inhibitors for steel in concrete’,Eurocorr ’97, Trondheim 1997, 469.

19. B. Elsener, ‘A review of the performance of corrosion inhibitors for steel in concrete’,COST 521 Workshop, Belfast, 2000, 271.

20. B. Elsener, M. Büchler, F. Stalder and H. Böhni, ‘Migrating corrosion inhibitorblend for reinforced concrete : Part 1 – prevention of corrosion’, Corrosion, 1999,55(12), 1155.

21. C. Alonso, C. Andrade, J. Fullea and J. Sanchez, ‘Accelerating test to ascertain theeffectiveness of corrosion inhibitors’, COST 521 Workshop, Belfast, 2000, 259.

22. E. Pazini, S. Leao and G. Estefani, ‘Corrosion inhibitors. Behaviour of NaNO2 andamine-based products in the prevention and control of corrosion in reinforced concrete’,NACE Latin-American Congress, Cancun, 1998.

23. M. Berra, F. Bolzoni, T. Pastore and P. Pedeferri, ‘Inibitori di corrosione per strutture



in c.a.’, Giornate Nazionali sulla corrosione e protezione, 4∞ edizione, AIM, Genova,1999, 293.

24. S. Ardizzone, A. Carugati and S. Trasatti, ‘Properties of thermally prepared iridiumdioxide electrodes’, J. Electroanal. Chem., 1981, 126, 287.

25. K. Kinoshita and M. J. Madou, ‘Electrochemical measurements on Pt, Ir, and Tioxides as pH probes’, J. Electrochem. Soc., 1984, 131(5), 1089.

26. M. Stern and A. L. Geary, ‘Electrochemical polarisation I: a theoretical analysis ofthe slope of polarisation curves’, J. Electrochem. Soc., 1957, 104, 56.

27. J. A. González, A. Molina, M. L. Escudero and C. Andrade, ‘Errors in theelectrochemical evaluation of very small corrosion rates. I. Polarization resistancemethod applied to corrosion of steel in concrete’, Corrosion Sci., 1985, 25, 917.

28. L. Bertolini, F. Bolzoni, P. Pedeferri and T. Pastore, ‘Cathodic protection ofreinforcement in carbonated concrete’, Corrosion 98, NACE, Paper 639.

29. C. Andrade, ‘Determination of chloride threshold in concrete’, COST 521 Workshop,Luxembourg, 2002, 108.

30. F. Malservigi, ‘Inibitori per la prevenzione della corrosione delle armature e per ilripristino delle strutture in c.a.’, Tesi di Laurea (Degree Thesis), Politecnico di Milano,A.A. 2000-01.

31. F. Bolzoni, G. Fumagalli, L. Lazzari, M. Ormellese and P. Pedeferri, ‘Mixed-ininhibitors for concrete structures’, Eurocorr 2001, Riva del Garda, 2001, 10 (seeChapter 15).


226

18.1 Introduction

The number of reinforced concrete structures showing signs of deteriorationor damage due to corrosion of rebars has increased dramatically over the last20 years. There is, therefore, an obvious and urgent need by the owners ofreinforced concrete structures for simple, quick, durable and cost-efficientrepair techniques. The application of corrosion inhibitors might be such arehabilitation method since the chloride-contaminated or carbonated concretedoes not have to be removed. Thus, this repair method seems to be verypromising. Although inhibitors are used in practice [1–2] and some fieldtrials are underway [3] there is still a lack of results from well monitoredlong-term field studies as well as established practical experience.

18.2 Field study in the Naxbergtunnel

18.2.1 Goals of the study

The goal of this three-year field study was to determine and to compare theeffectiveness of two inhibitors, sodium monofluorophosphate (MFP) andFerroGard-903 (FG), in the case of chloride-induced rebar corrosion and toverify the results of laboratory experiments [4–5] as well as to evaluate anappropriate monitoring system for such a repair technique. The study wasstarted in 1997 with the condition survey.

18.2.2 Description of the Naxbergtunnel

The 550 m-long Naxbergtunnel, built 1972–79, is a part of the highway A2from Lucerne through the Gotthardtunnel to Italy. It has two normal linesand one emergency line and is located about 1000 m above sea level, thus,in an area where much deicing salt is used during the winter season. The sidewalls of the tunnel are covered with prefabricated, 2.10 m wide elements

18Effectiveness of corrosion inhibitors –

a field study

Y. S C H I E G G, F. H U N K E L E R and H. U N G R I C H T,Swiss Society for Corrosion Protection (SGK), Switzerland and

Technical Research and Consulting on Cement andConcrete (TFB), Switzerland


Effectiveness of corrosion inhibitors – a field study 227

(panels) with a thickness of only 40 to 50 mm. The reinforcement of theelements consists of one mat of rebars (∆ 4 mm) in the centre.

18.2.3 Condition survey

Based on results of a previous condition survey (potential and chloridemeasurements) of the whole tunnel, 16 elements, situated approximately inthe middle part of the tunnel, were chosen for this investigation. A moredetailed condition survey of these 16 elements was undertaken in 1997. Thefollowing steps were carried out:

(a) Potential mapping (measurements grid: 0.15 ¥ 0.15 m),(b) Chloride analysis,(c) Removal of the concrete in small areas (openings) and visual inspection

of the corrosion state of the rebars.

Figure 18.1 shows the chloride profiles at different heights above the ground(0.45–3.0 m) and the relation between the chloride content and the corrosionpotential. At the lower part of the elements (<2.0 m), the chloride content isvery high (1.5–2.5% by mass with respect to the mass of cement). At aheight of about 2.0 to 3.0 m, the chloride content near the rebars (cover of therebars: mean value 19 mm) is lower than 0.5% by mass with respect to themass of cement. Obviously, the corrosion potentials decrease and the intensityof the corrosion process increases with increasing chloride contents.

18.3 Investigation

18.3.1 Test fields and instrumentation

The 16 elements used as test fields were as follows:

∑ 4 elements as reference/4 elements treated with MFP/4 elements treatedwith FerroGard-903,

∑ 2 elements treated with FerroGard-903 and Sikagard-701 W (hydrophobicimpregnation),

∑ 2 elements treated with Sikagard-701 W.

In the autumn of 1997 these test fields were instrumented with the followingmonitoring components:

∑ Electrically isolated rebars,∑ Instrumented cores for resistivity measurements,∑ Sensors for humidity and temperature of the air and concrete,∑ Installation of the data loggers and cabling.

Details of these monitoring components, as well as additional and generalinformation on the monitoring of concrete structures after a repair, are given



elsewhere [6]. At the same time cores were taken out of the elements todetermine the ohmic resistivity of the concrete as a function of the relativehumidity under controlled laboratory atmospheres.

18.3.2 Application of the inhibitors

The application of the inhibitors was carried out in June 1998, about eightmonths after the instrumentation of the test fields. This procedure should

Ch

lori

de

con

ten

t (m

ass

%/c

)

Autumn 19972.5

2

1.5

1

0.5

00 10 20 30 40 50

Depth (mm)(a)

0–1 m

1–2 m

2–3 m

Ch

lori

de

con

ten

t (m

ass

%/c

)

Autumn 19972.5

2

1.5

1

0.5

0–500 –400 –300 –200 –100 0

Potential [mV (CSE)](b)

KG3

KG4

KG2

18.1 (a) Chloride profiles at different heights above ground and (b)relation between chloride content and potential and corrosion stateof the rebar (KG: grade of corrosion) in openings: KG 0: blank, nocorrosion; KG 1: slight signs of corrosion; KG 2: small areas withcorrosion; KG 3: corrosion on the whole surface; KG 4: pittingcorrosion.



have allowed the concrete and the new mortar to regain the equilibriummoisture content.

Before the application of the inhibitors the surface of the concrete waswashed and cleaned with water. The inhibitors were then applied in severalsteps. This work was carried out by the providers of the inhibitors under thesupervision of the project leader. The amount of FerroGard-903 appliedduring the treatment was more than 500 g per m2 of concrete surface andthus higher than the recommended target value of 300–500 g m–2. The appliedamount of MFP was about 2.4 l m–2.

The concentrations of the inhibitors were analysed on concrete cores andconcrete dust samples. The first analyses were made directly after theapplication and the second in the autumn of 1999.

18.3.3 Measurements

A full set of measurements was executed after the instrumentation of the testfields, then just before and after the application of the inhibitors (June 1998)and thereafter approximately every 6 months. It included the followingmeasurements:

∑ Potential mapping of all elements,∑ Potential as a function of depth of concrete (potential profiles),∑ Macrocell current, potential difference, polarisation and ohmic resistance

of the isolated rebars,∑ Macrocell current, potential difference and ohmic resistance of the embedded

stainless steel bars,∑ Ohmic resistance of the instrumented cores (embedded wires),∑ Temperature and relative humidity.

Some of the above mentioned parameters were continuously registered bydata loggers.

18.4 Results

The continuous recording of the temperature and the relative humidity (rh)in the tunnel gave mean values of 9 ∞C and 71%. The temperature variedfrom –12.3 to 28.5 ∞C. During rainfall outside of the tunnel, the rh wascomparatively high; there were often values between 90 and 100%.

18.4.1 Potential mapping

In Fig. 18.2 the results of the statistical analysis of the corrosion potentialsmeasured in the different test fields are shown. The following conclusionscan be drawn:



ent in concrete230

FG 97FG 99FG 00

MFP 97MFP 99MFP 00

R 97R 99R 00

Per

cen

tag

e (%

)99.99

99.9

99

95908070

50

302010

5

1

.1

.01–600 –500 –400 –300 –200 –100 0 100

Potential [mV (CSE)](a)

Per

cen

tag

e (%

)

99.99

99.9

99

95908070

50

302010

5

1

.1

.01–600 –500 –400 –300 –200 –100 0 100

Potential [mV (CSE)](b)

Per

cen

tag

e (%

)

99.99

99.9

99

95908070

50

302010

5

1

.1

.01–600 –500 –400 –300 –200 –100 0 100

Potential [mV (CSE)](c)

18.2 Statistical analysis of the corrosion potentials measured 1997, 1999 and 2000 in the reference fields (R), the fieldstreated with MFP (MFP) and with FerroGard-903 (FG): (a) 1997: Measurements before the application of the inhibitors(September); (b) 1999: approximately one year after the application of the inhibitors (July); (c) 2000: approximately halfa year after the elements were placed outside of the tunnel (August).



∑ There were only minor changes of the corrosion potentials of the testfields between 1997 and 1999.

∑ In the reference- and MFP-fields a slightly positive shift is apparent in thepotential range above –50 mV (CSE). This might be due to a slight reductionin the moisture content of the concrete.

∑ The fields with FerroGard-903 show a slight increase of the potentials inthe upper part and slight decrease in the lower part of the curve. Thismight be caused by the increase of the concrete conductivity of theconcrete cover due to the application of the inhibitor (alkaline solutionswith salts).

∑ In 2000, the corrosion potentials were generally more negative. Theconditions (temperature, humidity) in the elements were different fromthe conditions during the measurements for 1997 and 1999 in the tunnel.Therefore, they can not be compared directly.

∑ The potential mapping provides reproducible results. The potential profilesof the MFP-elements showed a slight decrease of the potential in the first10–20 mm. This effect was not recognised in case of FerroGard-903.

18.4.2 Macrocell currents

The course of the macrocell currents over time of some electrically isolatedrebars are given in the Fig. 18.3 and 18.4. The cleaning of the surface and the

Cleaning + MFP application

Cleaning + FerroGardapplication

June 2./3.

E87, 2.3 mFGE87, 0.3 mFGE82, 0.5 mMFPE82, 2.2 mMFP

May/30 Jun/2 Jun/4 Jun/6 Jun/9 Jun/11 Jun/13 Jun/16Date

Mac

roce

ll cu

rren

t (m

A)

0.12

0.1

0.08

0.06

0.04

0.02

0

18.3 Macrocell currents of some electrically isolated rebars duringthe time of the cleaning process of the surface and the application ofthe inhibitors. MFP: MFP fields, FG: FerroGard-903 fields.



application of the inhibitors led to a sharp increase of the currents by a factorof 1.5 to 3 (increased conductivity of the concrete). This effect is morepronounced in the lower part of the elements, where higher macrocell currentswere measured and probably a larger amount of water reached the surfaceduring the cleaning. The short term transients (peaks) are caused by somesingle steps of the whole process (cleaning, prewetting, application).

Mac

roce

ll cu

rren

t (m

A)

0.120.10

0.08

0.060.040.02

0.000.120.10

0.08

0.06

0.04

0.02

0.00

Treatment

Jul/1/1998 Jan/1/1999 Jul/1/1999 Jan/1/2000Date(a)

E82 MFP, 0.5 mE82 MFP, 2.23 mConcrete temp.E87 FG, 2.32 mE87 FG, 0.33 m

20

10

0

–10

Co

ncrete

temp

. [∞C]

Mac

roce

ll cu

rren

t (m

A)

0.05

0.04

0.03

0.02

0.01

0.00

–0.01

Jul/1/1998 Jan/1/1999 Jul/1/1999 Jan/1/2000 Jul/1/2000Date(b)

R (2.09 m)R (0.97 m)MFP (0.96 m)MFP (1.45 m)FG (0.33 m)FG (1.44 m)

18.4 Concrete temperature and macrocell currents of someelectrically isolated rebars of the differently treated test fields locatedat different heights above ground [(a) data logger; (b): singlemeasurements]. R: reference field, MFP: MFP fields, FG: FerroGard-903 fields.



The more or less regular variations of the currents within hours or dayscorresponds to temperature changes. This is more pronounced at highercurrents. The highest currents are measured during the summer season. Thecorrosion process does not stop at temperature below 0 oC.

The macrocell currents of the isolated rebars in the lower part of theelements are generally higher than those in the upper part due to the higherchloride content and, thus, more active anodic areas. A clear decrease of thecurrents after the application of the inhibitor could not be detected.

18.4.3 Concrete resistances

Cleaning of the concrete surface with water and the application of the inhibitors(June 1998) led to a decrease of the ohmic resistances (Fig. 18.5 and 18.6),which is more pronounced in the concrete cover than in the middle of theelements (Fig. 18.6). Significant differences between the test fields can notbe seen. The variation over time corresponds to the seasonal changes of thetemperature (winter/summer).

The concrete resistances over the depth (resistance profile) measured withthe instrumented concrete cores are shown in Fig. 18.7. The measurementswere taken in summer 1998 (after the treatment) and in summer 1999 atsimilar air temperatures. In most cases, the resistances are higher at theexposed side than in the middle or backside of the elements because of thecarbonation of the concrete (increases the resistances). Compared with otherstructures the profiles are rather flat. The highest resistances are measured inthe field with the hydrophobic impregnation, which probably reduced themoisture content of the concrete. Apart from the field with the hydrophobic

R (m)MFP (m)FG (m)

No

rmal

ized

res

ista

nce

2.5

2.0

1.5

1.0

0.5

Jul/1/1998 Jan/1/1999 Jul/1/1999 Jan/1/2000 Jul/1/2000Date

18.5 Normalized concrete resistances (averages) over time measuredwith the electrically isolated rebars. R: reference field, MFP: MFPfields, FG: FerroGard-903 fields.



treatment, the test fields do not show significant changes of the concreteresistances. The values are similar in the lower and upper parts of the elements.

Figure 18.8 shows the correlation between the macrocell current, measuredwith the electrically isolated rebars, and the concrete resistance, measuredwith the instrumented cores (depth 12.5–20 mm), in a logarithmic scale. Thelinear correlation between macrocell current and concrete resistance indicatesan ohmic control of the corrosion process (slope ª –1). The different intensityof the macrocell currents is mainly related to the various anodic areas on therebars. The application of the inhibitors did not result in a clear decrease ofthe corrosion activities.

RMFPFG 5–12.5 mm

No

rmal

ized

res

ista

nce

2.5

2.0

1.5

1.0

0.5Jul/1/1998 Jan/1/1999 Jul/1/1999 Jan/1/2000 Jul/1/2000

Date(a)

RMFPFG 20–27.5 mm

No

rmal

ized

res

ista

nce

2.5

2.0

1.5

1.0

0.5Jul/1/1998 Jan/1/1999 Jul/1/1999 Jan/1/2000 Jul/1/2000

Date(b)

18.6 Normalized concrete resistances over time measured with theinstrumented cores at different depths: (a) 5–12.5 mm; (b) 20–27.5mm. R: reference field, MFP: MFP fields, FG: FerroGard-903 fields.



18.4.4 Galvanic pulse measurements

The galvanic pulse measurements were carried out with equipment which wasdeveloped by IBWK, ETH Zürich [7]. Figure 18.9 shows the variation of thepolarisation resistances measured over the isolated rebars (in this case connected

Height over ground >100 cm

RMFPFGSG

RMFPFGSG

28.7.99 30.6.98

Temperature approx. 15 ∞CR

oad

sid

e

Res

ista

nce

(W

)3 ¥ 104

2.5 ¥ 104

2 ¥ 104

1.5 ¥ 104

1 ¥ 104

5000

00 5 10 15 20 25 30 35 40

Depth (mm)

18.7 Concrete resistance over the depth, measured with theinstrumented cores. R: reference fields; MFP: MFP fields;FG: FerroGard-903 fields; SG: Sikagard 701-W fields.

E82 2.11 m

E87 2.32 m

MFP before applicationMFP after applicationFG before applicationFG after application

3 4 5 6 7 8 9 ¥104

Resistance [W]

Mac

roce

ll cu

rren

t (m

A)

0.176543

2

0.017654

3

2

0.001

18.8 Correlation of the macrocell current, measured with theelectrically isolated rebars, and the concrete resistance, measuredwith the instrumented cores (depth 12.5 to 20 mm). MFP: MFPfields, FG: FerroGard-903 fields.



to the rebar mat) in differently treated test fields over time. The higher resistancesof E78R, E85R and E93FG/SG were measured at a height of 2.0 to 3.0 m aboveground, where the chloride content is lower than 0.5 mass %/c. There is almostno change in the polarisation resistances of the different elements at the endof the field study. No effect of the treatment is recognised. As with the macrocellcurrents and the ohmic resistance, the polarisation resistance depends stronglyon the temperature too. The ohmic resistances, determined from the pulsemeasurements, showed the same behaviour.

18.4.5 Chemical analysis of the inhibitors

Tables 18.1 (FerroGard) and 18.2 (MFP) contain the results of the analysisfrom 1998 and 1999. The analysis methods of the two promoters and TFBare the same. Therefore, the results can be compared directly. The analysisgave, in both cases, inhibitor concentrations near the rebars which are higher

18.9 Polarisation resistance over time of some electrically isolatedrebars of the differently treated test fields: R: reference field, MFP:MFP fields, FG: FerroGard-903 fields, SG: Sikagard 701-W fields.

Po

lari

sati

on

res

ista

nce

[W

]

800

700

600

500

400

300

200

100

0 Apr/15/1998 Nov/2/1998 Apr/7/1999 Jul/28/1999

Date

E78 RE80 MFPE82 MFPE85 RE86 FGE87 FGE92 FG/SGE93 FG/SG

Table 18.1 Inhibitor content, analysed from concrete cores [FerroGard-903: organiccomponent (ppm)]

Depth (mm) 1998 (Analysis by the promoter) 1999 (Analysis by TFB)

0–7 5088 ± 146 5780 ± 194 6709 ± 342 1180 76010–17 168 ± 19 413 ± 28 556 ± 68 340 11020–27 16 ± 1 41 ± 1 80 ± 1030–37 < 13 16 ± 1 26 ± 240–47 < 13 < 13 15 ± 2



than the target values (MFP: 0.05 mass % in respect to the mass of concrete,FerroGard: approximately 13 ppm organic component). The profiles aresteep, especially in the elements treated with FerroGard. The content ofFerroGard decreased from 1998 to 1999 (possibly because of evaporation).The mentioned target value near the rebars of 13 ppm corresponds to thelimit of the chemical analysis to detect FerroGard.

There is no reason why the promoter did not find any MFP in the concretecore in 1999. The mentioned target value near the rebars of 0.05 mass % inrespect to the mass of concrete corresponds to the natural ground level of thephosphate content of the concrete.

18.5 Conclusions

The extensively instrumented and monitored field study on the effectivenessof corrosion inhibitors was started in 1997. The side elements (panels) of thewalls of the Naxbergtunnel were chosen as test fields. The following conclusionsmight be drawn:

∑ In both cases were the inhibitor concentrations near the rebars higher thanthe target values.

∑ No significant effects of the inhibitors MFP and FerroGard-903 on thecorrosion of the rebars could be detected either with potential mapping orwith the measurements on the electrically isolated rebars or with theinstrumented concrete cores.

∑ The hydrophobic impregnation led to an increase in concrete resistancedue to the reduced moisture content of the concrete.

∑ Potential measurements are generally possible and give useful results. Butthey are not sufficient to evaluate the effectiveness of inhibitors.Measurements of the corrosion currents of isolated rebars are a goodmethod to get information about the changes of the corrosion rate.

∑ The instrumentation as well as the monitoring (combination of manualmeasurements with data logging) has proven to be appropriate.

Table 18.2 Inhibitor content, analysed from concrete cores and concrete dustsamples [MFP: in PO 4

2– (m.%/concrete)]

Depth 1998 1999 (concrete cores)(mm) (Analysis by the promoter, promoter TFB

dust samples)

0–10 0.406 0.920 0.229 0.203 0.047 0.250 0.23210–20 0.214 1.093 0.240 0.097 0.034 0.119 0.08120–33 0.237 0.150 0.171 0.055 0.05533–44 0.056 0.117 0.090 0.044 0.05850–60 0.099 0.079




The authors gratefully acknowledge the Construction Department of theCanton Uri, Altdorf, and the Swiss Federal Highway Administration, Bern,for the financial support of this field study as well as SIKA AG, Zürich,Switzerland and MFP SA, Divonne les Bains, France, for the application ofthe inhibitors and the chemical analysis.

18.7 References

1. M. Haynes and B. Malric, ‘Use of migratory corrosion inhibitors’, Constr. Repair,July/August 1997, 10–15.

2. E. Brühwiler and P. Plancherel, ‘Instandsetzung von Sichtbetonfasaden mit Inhibitoren’,Schweiz. Ing. Architekt, 1999, 26, 583–586.

3. J. Broomfield, ‘The pros and cons of corrosion inhibitors’, Constr. Repair, July/August 1997, 16–18.

4. B. Elsener, M. Büchler and H. Böhni, ‘Organic corrosion inhibitors for steel in concrete’,Eur. Fed. Corrosion Publ., 2000, 31, 61–71.

5. M. Salta, E. Pereira and P. Melo, ‘Influence of organic inhibitors on reinforcing steelcorrosion’, COST 521, Proc. of a European Workshop and Annual Progress Reports,Belfast, 2000, 247–254.

6. Y. Schiegg, L. Zimmermann, B. Elsener and H. Böhni, ‘Electrochemical techniquesfor monitoring the conditions of concrete bridge structures’, Int. Conf. Repair ofConcrete Structures’, Solvear, May 1997, 213–222.

7. B. Elsener, D. Flückiger, H. Wojtas, H. Böhni, ‘Methoden zur Erfassung der Korrosionvon Stahl in Beton’, VSS-Ber., 521, 1996.


239

19.1 Introduction

Corrosion of reinforcing steel in concrete structures, when exposed to chlorides,is a common occurrence. It is a complex phenomenon related to structural,physical, chemical and environmental considerations. Much effort has beenfocused on the design of new structures to reduce or eliminate corrosionthrough increased concrete coverage using reduced permeability concrete orreplacing the steel reinforcement with alternative materials. However, littleeffort has been made in establishing reliable techniques for the repair ofexisting structures. Since many of the structures built after WWII are reachingthe end of their design life and there are no plans to replace them, a rehabilitationprogramme is necessary. It was cited in a 1993 survey by the StrategicHighway Research Program in the United States that the cost of repairingbridge decks that had suffered chloride-induced deterioration was $20 billionand was increasing at a rate of $500 million annually.

Reinforcing steel embedded in concrete shows a high amount of resistanceto corrosion. The cement paste in the concrete provides an alkaline environmentthat protects the steel from corrosion. This corrosion resistance stems froma passivating or protective ferric oxide film that forms on the steel when it isembedded in concrete. This film is stable in the highly alkaline concrete (pHapprox. 11–13). The corrosion rate of steel in this state is negligible. Factorsinfluencing the ability of the rebar to remain passivated are the water-to-cement ratio, permeability and electrical resistance of the concrete. Thesefactors determine whether corrosive species like carbonation and chlorideions can penetrate through the concrete pores to the oxide layer on the rebar,then break down the passive layer, leaving the rebar vulnerable. Typically,concrete is cast without the inclusion of corrosive species. Chloride ionsbecome available when the concrete is exposed to environmental factors,such as deicing salts applied to roads or seawater in marine environments.

Migrating Corrosion Inhibitor (MCI) technology was developed to protectthe embedded steel rebar/concrete structure. These inhibitors can be organic

19Corrosion protection of steel rebar in concrete

using migrating corrosion inhibitors

B. B AVA R I A N and L. R E I N E R,California State University, USA



or inorganic compounds; however organic compounds seem to be moreeffective (for neutralising and film forming). Recent MCIs are based onamino-carboxylate chemistry [1–3]. Normally, the most effective type ofinhibitor lessens corrosion at the anodes and cathodes simultaneously. Organicinhibitors are a subgroup of the combined inhibitors. They utilise compoundsthat work by forming a monomolecular film between the metal and thewater. These compounds are polar and have a strong affinity for the surfacesonto which they may be adsorbed [4, 5]. In the case of film-forming amines,one end of the molecule is hydrophilic and the other hydrophobic. Thesemolecules will arrange themselves parallel to one another and perpendicularto the reinforcement such that a continuous barrier is formed. The presenceof this film on samples of reinforcement encased in concrete with an organicinhibiting admixture has been shown by methods of ultraviolet spectroscopyand gas chromatography [6]. These types of inhibitors are known as migratingcorrosion inhibitors if they are able to penetrate into existing concrete toprotect the steel in the presence of chloride [7]. The means by which theinhibitor migrates is first by diffusion through the moisture that is normallyavailable in concrete, then by its high vapor pressure and finally by followinghairlines and microcracks. This mechanism allows a greater amount to beapplied where it is most needed. The diffusion process requires time tomigrate through the concrete pores to reach the rebar’s surface and form aprotective layer. This suggests that the migratory inhibitors are physicallyadsorbed onto the metal surfaces [1].

MCIs can be incorporated as an admixture or can be used by surfaceimpregnation of existing concrete structures. With surface impregnation,diffusion transports the MCIs into the deeper concrete layers. They willdelay and inhibit the onset of corrosion on steel rebar. Bjegovic and Miksicrecently demonstrated the effectiveness of MCIs over five years of continuoustesting [1–3]. They also showed that a migrating amine-based corrosion-inhibiting admixture can be effective when it is incorporated in the repairprocess of concrete structures [2]. Furthermore, laboratory tests have proventhat MCI corrosion inhibitors migrate through the concrete pores to protectthe rebar against corrosion even in the presence of chlorides [3–4]. However,the amount of additive inhibitor should be calculated based on the concretechloride content. Chloride increases the level of conductivity of concrete[8–10]; it also breaks down the passive film from the steel reinforcement.The level of chloride ions required to initiate corrosion in concrete correspondsto 0.10% soluble chloride ion by weight of cement [7–8]. McGovern [11]reports work by the United States Federal Highway Administration Laboratorieswhich suggests that the threshold value for steel corrosion is 0.20% acid-soluble chlorides by weight of cement. This is equivalent to between 0.6 and0.8 kg of chlorides per cubic metre of concrete. The chloride thresholdconcentration is generally within 0.9 to 1.1 kg of chlorides per cubic metreof concrete [5].


Corrosion protection of steel rebar in concrete 241

The objective of this investigation was to study the corrosion inhibition ofcommercially available migrating corrosion inhibitors on steel rebar in threeconcrete densities. Theoretically, high-density concrete impedes corrosivespecies from reaching the surface of the rebar. It may also prevent the inhibitorfrom reaching the surface of the concrete. Electrochemical monitoringtechniques were applied while samples were immersed in 3.5% NaCl atambient temperature. Because of the low conductivity of concrete, the corrosionbehavior of steel rebar had to be monitored using AC electrochemical impedancespectroscopy (EIS). During this investigation, changes in the polarisationresistance and the corrosion potential of the rebar were monitored to ascertainthe degree of effectiveness of these MCI products. The results were comparedwith previous investigations conducted on several admixtures and stainlesssteel rebar. X-ray photoelectron spectroscopy (XPS) and depth profilingwere used to check if the inhibitors reacted with the rebar surfaces.

19.2 Experimental procedures

In theory, the steel rebar/concrete combination can be treated as a poroussolution that can be modelled by a Randles electrical circuit. EIS tests performedon a circuit containing a capacitor and two resistors indicate that this modelis an accurate representation of an actual corroding specimen. EIS testingallows for the determination of fundamental parameters relating to theelectrochemical kinetics of the corroding system. It involves the applicationof a small-amplitude alternating-potential signal of varying frequency to thecorroding system. Because processes at the surface absorb electrical energyat discrete frequencies, the time lag and phase angle, theta, can be measured.The values of concern in this study are Rp and RW. The Rp value is a measureof the polarisation resistance or the resistance of the surface of the materialto corrosion. RW is a measure of the solution resistance to the flow of thecorrosion current. By monitoring the Rp

value over time, the relativeeffectiveness of the sample against corrosion can be determined. If the specimenmaintains a high Rp value in the presence of chloride it is considered to be‘passivated’ or immune to the effects of corrosion. If the specimen displaysa decreasing Rp value over time it is corroding and the inhibitor is notproviding corrosion resistance.

The EG&G Instruments Potentiostat/Galvanostat Model 273A and EG&GM398 Electrochemical Impedance Software were used to conduct theseexperiments and to record the results. Bode and Nyquist plots were producedfrom the data obtained using the single sine technique. Potential values wererecorded and plotted with respect to time. By comparing the Bode plots,changes in the slopes of the curves were monitored as a means of establishinga trend in the Rp value over time. To verify this analysis, the Rp values werealso estimated by using a curve fit algorithm on the Nyquist plots (availablein the software).



Results from the EIS tests were organised into Bode and Nyquist plots.Based on these plots, the Rp and RW combined values are displayed in the lowfrequency range of the Bode plot and the RW value can be seen in the highfrequency range of the Bode plot. The diameter of the Nyquist plot is ameasure of the Rp value.

Concrete samples with dimensions 8≤ ¥ 4≤ ¥ 4≤ (approx. 200 mm ¥ 100mm ¥ 100 mm) were prepared, and their densities were adjusted to achieve130, 140, and 150 lb ft–3 (2.08, 2.24 and 2.40 g cm–3). Each sample consistedof one 8≤ (20 cm) steel (class 60) rebar 1/2≤ (12.7 mm) in diameter and one8≤ (20 cm) Inconel metal strip (counter electrode). The rebars, before beingplaced in concrete, were exposed to 100% rh (relative humidity) to initiatecorrosion. The coverage layer was maintained at one inch (25.4 mm) ofconcrete for all these samples. The samples were cured for 28 days,after which their compressive strengths ranged between 2700–3000 psi(18.61–20.68 MPa). The low density samples had higher compressive strengththan the high density samples. The concrete blocks were sandblasted toremove loose particles, leaving the concrete with a marginally smoothersurface. Two coats of MCI 2022 and MCI 2021 were applied with a paintbrush to all but two of the concrete samples (used as a control). The sampleswere then immersed in 3.5% NaCl solution [roughly 7≤ (17.8 cm) of eachsample was immersed in the solution continuously]. EIS (electrochemical acimpedance spectroscopy) testing started 24 h after immersing the samples. ACu/CuSO4 electrode was used as the reference and each sample was testedweekly. XPS analyses were performed on steel rebar that was in concretetreated with MCI, immersed for 400 days, using a KRATOS AXIS ultra X-ray photoelectron spectrometer, and for the depth profiling, sputtering wasconducted using a 2 kV Ar+ ion gun. The thickness of the deposited film wasestimated from the rate of removal of Ta2O5.


The corrosion inhibition of two commercially available migrating corrosioninhibitors (Cortec MCI 2022 and 2021) for three concrete densities wasinvestigated over a period of 400 days using ac electrochemical impedancespectroscopy (EIS). Throughout this investigation, changes in the polarisationresistance and the corrosion potential of the rebar were monitored to determinethe degree of effectiveness for Cortec MCI 2021 & 2022 products. Accordingto the ASTM (C876) standard, if the open circuit potential (corrosion potential)is –200 mV or higher, this indicates a 90% probability that no reinforcingsteel has corroded. Corrosion potentials more negative than –350 mV areassumed to have a greater than 90% likelihood of corrosion. In Fig. 19.1 thecorrosion potentials for the untreated control samples dropped from –200mV to –545 mV, which indicates a 90% probability of corrosion attack on


Corrosion protection of steel rebar in concrete

243

Po

ten

tial

(m

V)

0

–100

–200

–300

–400

–500

–6000 50 100 150 200 250 300 350 400 450

Time of submersion (d)

L2022–1L2022–2L2021–1L untreatedL2021–2H2022–1H2022–2H untreatedH2021–1H2021–2

19.1 Corrosion potential of steel rebar vs time, ASTM C876-91, protected by Cortec MCI 2022 & 2021 compared withunprotected concrete (various concrete densities). L = low density, H = high density, 1 = sample 1, 2 = sample 2.



the reinforcing steel. The corrosion potentials for MCI treated concrete sampleswere –120 to –150 mV. The past six years of data analysis for immersedsamples show that the ASTM C876 criteria can be used to validate theprobability of corrosion attack.

In Fig. 19.2 the resistance polarisation for MCI treated concrete samplesgradually increased from 10 000 to 100 000 ohms. The Rp value for a non-treated sample (control) was 10 000 at the beginning of the experiment andended at less than 700 W. Changes in the Rp value were observed after about90 days, indicating that corrosive species or Migrating Corrosion Inhibitors(MCIs) require an induction period for diffusion into the concrete.Figure 19.3 shows the current experimental results for low density(130 lb ft3, 2.08 g cm–3) and high density (150 lb ft–3, 2.4 g cm–3) concretesamples. Preliminarily data show that MCI treated concrete samples displayan increase in their Rp values compared with the control samples that showa decreasing trend. XPS analysis demonstrated the presence of inhibitor onthe steel rebar surface; MCI was able to penetrate through the concretecoverage layer and reach the rebar to retard corrosion. Figure 19.4 shows theXPS spectrum for the rebar removed from the MCI treated sample after 400days. Figure 19.5 shows depth profiling results using 2 kV Ar+ ions for asteel rebar removed from MCI treated concrete, showing that a 100 nm layerof amine-rich compound is present on the rebar surface. The high resolutionXPS analysis of carbon and oxygen showed the organic compound to havecarboxylate chemistry. The inhibitor is an amine-based carboxylate organiccompound, therefore it was concluded that the film on the rebar surface wasthe inhibitor molecules. Chloride was detected at about 0.10 atomic % andup to 50 nm deep on the top surface of the rebar. The XPS results demonstratethat both MCI and corrosive species migrated into the concrete samples, butMCI managed to protect the steel rebar. The lower density samples coatedwith the MCI inhibitor showed the greatest amount of corrosion resistance;their corrosion behaviour was similar to that of stainless steel rebar [13–14].The means by which the MCI inhibitor migrates into the concrete is first bydiffusion through moisture that is normally available in concrete, then by itshigh vapour pressure and finally by following hairline cracks and microcracks.Therefore, lower density concrete samples provide an easier path for theinward diffusion of MCI, and faster corrosion retardation. These results areextremely promising for the MCI product in its ability to protect steel rebarin concrete in aggressive environments.

19.4 Conclusions

Corrosion inhibition by two commercially available migrating corrosioninhibitors (Cortec MCI 2022 and 2021) on steel rebar in concrete wasinvestigated while the concrete was immersed in 3.5% NaCl at ambient


Corrosion protection of steel rebar in concrete

245

|Z|

(W)

1.00E+05

1.00E+04

1.00E+03

1.00E+02

1.00E+011.00E–04 1.00E–03 1.00E–02 1.00E–01 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04

Frequency (Hz)

2022–day 32022–day 375Control day 4Control day 378

19.2 EIS results, Bode plot for MCI 2022 (concrete density 14 lb ft–3). Comparison with control samples.



ent in concrete246

|Z|

(W)

1.00E+05

1.00E+04

1.00E+03

1.00E+02

1.00E+011.00E–04 1.00E–03 1.00E–02 1.00E–01 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05

Frequency (Hz)

2L-Day 1 2L-Day 238 2L-Day 325L-Day 1 L-Day 233 L-Day 3322S-Day 1 2S-Day 236 2S-Day 327

19.3 EIS results Bode plots. LD = untreated low density concrete, 2S = MCI 2022/high density, 2L = MCI 2022/low density;Concrete densities: low = 130 lbs ft–3, high = 150 lb ft–3.


Corrosion protection of steel rebar in concrete 247In

ten

sity

(cp

s)

80

60

40

20

Cu2p

O1s

CFe2p

N1s Cl

2p

Si2s

Si2p

1000 800 600 400 200 0Binding energy (eV)

¥103

Survey neut:2 (steel_corrosion)Lens Mode: Electrostatic Resolution: Pass energy 160 Anode: Mono (Al) (375 W)Step(meV): 1000.0 Dwell(ms): 163 Sweeps: 1 Acquisition Time(s): 180

Peak Position FWHM Raw area RSF Atomic AtomicBE (eV) (eV) (CPS) mass conc%

Fe 2p 710.4 4.1 5520.2 2.957 55.8 4.39O 1s 531.2 3.0 10963.2 0.780 15.9 34.54N 1s 398.5 2.1 416.3 0.477 14.0 2.24C 1s 285.0 2.5 5401.2 0.278 12.0 50.34Si 2p 101.8 2.5 1054.3 0.328 28.0 7.42Cu 2p 935.2 1.4 656.1 5.321 63.5 0.27Cl 2p 196.2 1.6 76.1 0.622 35.5 0.09

19.4 XPS on MCI 2022 treated concrete after 378 days. Large area(1000 ¥ 800 mm) survey scan from corroded surface.

OCFeSiNCl

0 200 400 600 800 1000 1200 1400 1600 1800 2000Etch time (s)

Co

nce

ntr

atio

n (

%)

60

50

40

30

20

10

0

19.5 XPS depth profile on steel rebar removed from MCI treatedconcrete sample after 378 days of testing (etched using 2kV Ar+ ions).



temperatures using electrochemical monitoring techniques. The MCI productshave successfully inhibited corrosion of the rebar in a 3.5% NaCl solutionfor 400 days. Steel rebar corrosion potentials were maintained at approximately–150 mV, and rebar polarisation resistance showed a gradual increase reachingas high as 100 000 W. However, the low density concrete demonstrated betterprotection than the other samples which is in agreement with the migrationmechanism of these inhibitors. XPS analysis verified the presence of theinhibitor on the steel rebar surface indicating MCI migration through theconcrete layer. Depth profiling showed a 100 nm layer of amine-rich carboxylatecompound on the rebar surface, which assures satisfactory corrosion resistanceeven in the presence of chloride ions. In summary, the experimental resultsdemonstrate that the MCI products offer an inhibiting system for protectingreinforced concrete in an aggressive 3.5% NaCl solution. These results areextremely promising for the protection of steel rebar and concrete in aggressiveenvironments.

19.5 References

1. D. Bjegovic and B. Miksic, ‘Migrating corrosion inhibitor protection of concrete’,Mater. Perf., Nov 1999, 52–56.

2. D. Bjegovic and V. Ukrainczyk, ‘Compatability of repair mortar with migratingcorrosion inhibiting admixtures’, CORROSION/97, paper no. 183, Houston, TX,NACE, 1997.

3. D. Rosignoli, L. Gelner, and D. Bjegovic, ‘Anticorrosion systems in the maintenance,repair and restoration of structures in reinforced concrete’, Int. Conf. Corrosion inNatural and Industrial Environments: Problems and Solutions, Grado, Italy, May23–25, 1995.

4. D. Darling and R. Ram, ‘Green chemistry applied to corrosion and scale inhibitors’,Mat. Perf., 1998, 37(12), 42–45.

5. C. K Nmai, S. A. Farrington and G. S. Bobrowski ‘Organic-based corrosion-inhibitingadmixture for reinforced concrete’, Concrete Int., 1992, 14(4).

6. P. H. Emmons and V. M. Alexander, ‘Corrosion protection in concrete repair mythand reality’, Concrete Int, 1997, 19(3), 47–56.

7. D. Stark, ‘Influence of design and materials on corrosion resistance of steel inconcrete’, Res. Dev. Bull., RD098.01T, Portland Cement Association, Skokie, Illinois,1989.

8. W. Hime and B. Erlin, ‘Some chemical and physical aspects of phenomena associatedwith chloride-induced corrosion’, Corrosion, Concrete and Chlorides: Steel Corrosionin Concrete: Causes and Restraints, Frances W. Gibson (Ed.), American ConcreteInstitute, Detroit, Michigan, 1987.

9. W. J. Jang and I. Iwasaki, ‘Rebar corrosion under simulated concrete conditionsusing galvanic current measurements’, Corrosion, 1991, 47(11), 875–884.

10. T. Liu and R. W. Weyers, ‘Modeling the dynamic corrosion process in chloridecontaminated concrete structures’, Cement Concrete Res., 1998, 28(3), 365–379.

11. M. S. McGovern, ‘A new weapon against corrosion’, Concrete Repair Dig., June,1994.


Corrosion protection of steel rebar in concrete 249

12. R. Montani, ‘Concrete repair and protection with corrosion inhibitor’, Water Eng.Manage., 1997, 144(11), 16–21.

13. R. Martinez, A. Petrossian and B. Bavarian, ‘Corrosion of steel rebar in concrete’,presented at the 12th NCUR, April 1998.

14. L. Reiner and B. Bavarian, ‘Corrosion of steel rebar in concrete’, presented at the14th NCUR, Missoula, Montana, April 2000.


250

20.1 Introduction

20.1.1 Permeability of coatings on concrete

Permeation of water (H2O) and carbon dioxide (CO2) through coatings isimportant for coated concrete structures and has a big influence on the durabilityof the entire structure. Such coatings have to show a low permeability forCO2 and liquid H2O, but a high permeability for water vapour. The concretehas to allow water to evaporate, especially when structures suffer fromascending water, leakage or humidity inside a building. A barrier againstCO2 is necessary to avoid carbonation, which leads to severe corrosionproblems of the embedded rebars (reinforcing steel) and is probably themost severe durability problem of concrete structures.

Surface protection systems are applied to assure the durability and lifetimeof concrete structures. Surface protection systems fulfil special requirements,such as CO2–resistance, H2O vapour permeability and water resistance. Besidesthese three major properties, others have to be achieved, depending on thepurpose and environmental conditions of the structure, for example crackbridging, chemical resistance to industrial atmospheres, resistance to deicingsalts, and colour resistance against UV irradiation.

In order to quantify the diffusion of CO2 and H2O, a European StandardCEN 1062 was created. Therein, the diffusion resistance of a coating isexpressed as the equivalent thickness of an air layer (sD), having the sameresistance against CO2 diffusion as the coating. The sD–value is obtainedfrom permeability measurements using two different atmospheres with knownpartial pressures of CO2. From the measurements described below a diffusionresistance value (m) is obtained by using the formula given in Fig. 20.1. Thediffusion equivalent air layer thickness sD is obtained by multiplying m by thecoating thickness (d) of the investigated samples (sD = md). The sD value forCO2 should be high and the sD value for water vapour should be low. If thepartial pressures are not sufficiently known the value of m can also be obtained

20Determination of coating permeability on

concrete using EIS

J. V O G E L S A N G, G. M E Y E R, and M. B E P O I X,Sika GmbH, Germany


Determination of coating permeability on concrete using EIS 251

by determination of the amount of diffused H2O via the WDD value (watervapour transmission rate; see also Fig. 20.1 and 20.2) For details please referto the CEN standard [1].

Free film or film on substrate

AbsorbanceCup

m d = 1 –

– L1 2

LsA

p pI

SÊËÁ

ˆ¯ (1)

WDD = 24

– m – m

; 2 1

2 12t t A

gm d

ÊËÁ

ˆ¯ (2)

m =

WDD k

S

20.1 Design of the cup test. Equation 1 describes the water diffusionvalue based on partial pressure differences and equation 2 is usedwhen weighing is carried out for the determination of the amount ofdiffused gas.

Dep

th (

mm

)

14

12

10

8

6

4

2

0

SD = 0

SD = 50 m

SD = 1.0 m

SD = •

1 4 9 16 25

Years (a )–1

2

20.2 Carbonation depth as a function of time under the influence ofdifferent equivalent air layer thickness (principle plot, based oncalculations).



20.1.2 Discussion of known techniques for thedetermination of coating permeability

For permeability measurements, the traditional cup test is used. A H2O absorbantor a CO2 absorbant is placed in a cup sealed carefully with the film underinvestigation (see Fig. 20.1). Then the sealed cup is weighed and placed inan atmosphere containing CO2 or water vapour. After a certain time theweight increase is measured and this can be directly related to the permeabilityof the coating in the chosen atmosphere.

The cup method involves the use of free films, with all the related problemsof their preparation, especially when the substrate (mostly plastic or aluminumfoils) has to be removed mechanically. A second way of producing a film forpermeation testing is by using porous glass filters as a substrate. In thismethod, the film stays on its substrate and does not have to be removedbefore testing. However, it is hardly possible to say that a glass substrate hasexactly the same properties as mortar or concrete. Moreover, it is likely thatthe properties of the coating show some influences which are caused byusing a porous glass plate.

A simple method for the measurement of coating permeability for CO2

has been described by Bagda [2, 3]. Coated mortar specimens were exposedto a CO2 atmosphere and, after a certain exposure time, the samples werecut. At the cut, the mortar is treated with an alkalinity indicator such asphenolphthalein and those areas which have been neutralised by the diffusingCO2 will be visible because the indicator remains colourless (pH < 9). Thismethod cannot be applied for H2O diffusion measurements, because no sensitiveand reliable indicator for water or humidity exists for slices of concrete ormortar, which usually have a certain but mostly unknown humidity.

20.2 Experimental design

This paper deals with a method for testing the water vapour permeability ofcoatings on concrete. The method is based on conductivity measurements ofthe mortar or concrete, using two stainless-steel screws embedded in mortarprisms as conductivity sensors. The conductivity of the mortar is measuredusing electrochemical impedance spectroscopy (EIS), with which additionalinformation is obtained.

The conductivity of mortar and concrete depends on the humidity of thespecimens. Any change in the humidity level becomes visible by a changedconductivity. This was already used in the past by e.g. Tritthart and Gehmayer[4]; they studied wetting and drying of concrete by monitoring the resistance.The relationship between humidity and conductivity is dependent on theconcrete formulation, including cement type, water-to-cement (w/c) ratioand soluble contaminants of the sand and aggregates. The degree of hydration



certainly has a significant influence on the conductivity. Therefore, it isnecessary to compare only specimens which are prepared identically and areformulated, manufactured and conditioned simultaneously.

The main experimental approach was to investigate the conductivity ofdried and coated concrete before and after exposure to 100% relative humidity.The measurements were repeated over a certain period, in order to monitorthe change in the mortar conductivity due to humidity uptake (the detailedprocedure is shown in Fig. 20.3).

Sample preparation and conditioning were done as follows: the mortarwas made with a water/cement ratio of 0.4, using a cement/sand ratio of 2:3with a grain size <0.6 mm (higher grain sizes could cause inhomogenity overthe cross section of the prism, which might result in non-uniform diffusionof water).

The dimensions of the prisms were 4 ¥ 4 ¥ 16 cm, with removal from themould after 24 h and an additional 3 days at 100% rh for sufficient curing.

Preparation of the mortar, proper fixing of the screws, curing anddemoulding, final curing for 3 days at 100% rh, 23 ∞C and 3 days

at 23 ∞C, 50% rh

7 days at 100% rh7 days at 60 ∞C ambient

humidity

6 prisms uncoated6 coated twice with acrylic6 coated twice with epoxy

6 prisms uncoated6 coated twice with acrylic6 coated twice with epoxy

All coated samples cured at 23 ∞C and 50% rh for 7 days

All samples conditioned for4 weeks at 98% rh

All samples conditioned for4 weeks at 60% ambient

humidity

Samples transferred fromdry to humid (98% rh)

Measurements started here

Samples transferred fromhumid to dry.

Measurements started here

Note: 100% rh is achieved by spraying water (fog room); 98% rh isachieved in a cabinet with saturation in water vapour.

20.3 Experimental procedure.



These electrodes are screws in the 4 ¥ 4 cm2 faces (Fig. 20.4). Six prismswere prepared for each system: uncoated, coating system 1 and coatingsystem 2. The coating materials were applied in two layers each. System 1had a water-based acrylic binder and system 2 consisted of a solvent-basedepoxy.

Impedance measurements were performed for monitoring the conductivityof the mortar. A Zahner IM6 impedance spectrometer was used. To measureimpedance spectra, a sinusoidal potential with a given frequency and amplitudewas applied and the resulting current was measured, whereas the modulus(ratio of U0 and I0) and the phase shift are plotted over the frequency indouble logarithmic scale (Bode plot). In Fig. 20.5a the principle of EISmeasurements is shown in the upper diagram. The spectra were recorded

(a)

(b)

20.4 (a) Principle design of specimen and (b) photographs of a prism,showing one of the embedded screws.



over a frequency range of 0.1 to 100 kHz, five points per decade with fivesamples for each point and an amplitude of 5 mV (10 mV peak to peak). Fordata presentation the Bode plot is preferred, because of the strong change ofspectral signals from the measured samples.

For the interpretation of the spectra a suitable equivalent circuit is required,which allows the modelling of the physical reality of the measured

IZ (w)I = U0/i0V

olt

age

(arb

itra

ry u

nit

s)

1.2

0.8

0.4

0

–0.4

–0.8

–1.2

i0 U0

Phase shift

tan q = Im(Z)/Re(Z )U(t) = U0 sin (w t)i(t) = i0 sin (w t + q )

0 2 4 6 8 10Time (arbitrary units)

(a)

Cu

rren

t (a

rbit

rary

un

its)

1.2

0.8

0.4

0

–0.4

–0.8

–1.2

9

6

5N

N

4

8

21

(b)

20.5 (a) Principle relation between current and voltage in the case ofEIS and (b) the equivalent circuit, suitable for modelling of mortarsamples containing conductivity sensors made of stainless steel.



electrochemical system by curve fitting. Such a model is assembled withcapacitors, resistors and other electronic circuit elements, each describing acertain property of the measured system. A relatively complicated exampleis given in Fig. 20.5b. It is suitable for modelling the EIS measurements ofall related mortar prisms, independent of their state. The resistor 4 representsthe electrolyte resistance in the mortar, which is directly linked with theconductivity of the mortar. All the other elements are required for a propermodelling of the complicated signals generated from the two stainless steelscrews, the wiring and the mortar itself.

Although it has been proven that this model is suitable for all the spectraobtained from the mortar prisms, the results from the EIS measurements areclear enough to show evidence for the usefulness of this new method, evenwithout modelling of the EIS data.

In order to demonstrate the different water vapour permeability of thewater-based acrylic and the solvent-based epoxy, two experimental pathswere followed. For the first path, the samples were kept humid before coatingand then, after sufficient curing over 3 days at 23 ∞C and 50% rh, they weretransferred into a dry environment. Here, the rate of evaporation of humidityfrom the mortar samples was measured via an increase of the mortar resistancewith drying time. Using the second path, the samples were first kept underdry condition and then transferred into 100% humidity. Directly at the beginningand after each day, impedance spectra were recorded. The complete samplepreparation procedure is shown in Fig. 20.3.


In Fig. 20.6, 20.7 and 20.8 the resulting spectra of the uncoated and coatedspecimens are shown. The consecutive spectra of one specimen are plottedtogether in order to allow judgement of the extent of the changing conductivity.Comparing the results for uncoated samples in Fig. 20.6 with those of thecoated samples it is evident that the uncoated samples show a much fasterchange in conductivity than the coated samples. This is found in both directions,from humid to dry and vice versa. It must be emphasized that this result isnot at all astonishing and is to be expected, but its fulfilment is highlynecessary to provide confidence in the method.

Going into more detail: in Fig. 20.6 the first recorded spectrum for thehumid condition has the lowest impedance and with longer drying time theimpedance increases from day to day. The equivalent behaviour is obtainedwhen recording of the series of spectra is started in the dry state: The firstspectrum shows the highest impedance and the phase angle is in the range of–90∞. The mortar resistance could be extracted in the frequency range between10 to 10000 Hz. The plateau of the modulus of most spectra is directly linkedwith the mortar resistance.



Only the first impedance curve in the initial dry state shows no plateaubecause of its very high impedance values due to poor conductivity. Butalready the second and third curves begin to show this plateau at lowfrequencies. The arrows indicate the sequence of the increase or decrease ofimpedance with time.

Comparing Fig. 20.7 and 20.8 with Fig. 20.6 it emerges that the uncoatedsamples showed the strongest change of impedance, followed by the less

Imp

edan

ce (W

)

10M

3M

1M

300K

100K

30K

100m 1 10 100 1K 10K 100KFrequency (Hz)

(a)

90

75

60

45

30

15

0

Ph

ase temp

erature (∞C

)

1tro0001tro0011tro0021tro0031tro0041tro0051tro0061tro007

Imp

edan

ce (W

)

10M

100M

1M


(b)

90

75

60

45

30

15

0

Ph

ase temp

erature (∞C

)

akli000akli001akli002akli003akli004akli005akli006akli007

20.6 Results from the uncoated samples. Last number is related tothe number of days in the respective environment: (a) change fromhumidity to dryness; (b) change from dryness to humidity.



pronounced change of those coated with the water-based acrylic paint andthe least change was obtained with the solvent-based epoxy coated species.Actually, the change of impedance from ‘dry’ to ‘wet’ for the coated samplesis extremely small and slow (Fig. 20.7 and 20.8). It clearly demonstrates theability of both coatings to prevent water vapour ingress to a large degree.

Taking into account the fact that the initial impedance in the humid state(before coating) was more or less the same for all kinds of samples, independentof their subsequent coating, the impedance values after 7 days (recorded at 1

1M

300K

100K

30K


(a)

90

75

60

45

30

15

0


Imp

edan

ce (W

)

10G

1G

100M

10M

1M

100K


(b)

90

75

60

45

30

15

0

Ph

ase temp

erature (∞C

)

dkli000dkli001dkli002dkli003dkli004dkli005dkli006dkli007

20.7 EIS measurements of the samples with the waterbased acryliccoating. Last number is related to the number of days in therespective environment: (a) change from humidity to dryness; (b)change from dryness to humidity.

Imp

edan

ce (W

)



Hz) are quite easy to understand. These values are given in Table 20.1, togetherwith those of the dry state before coating. It should also be mentioned that thecorresponding uniform impedance value was observed for the samples of thedry state before their transfer to humidity. This value is also reported in Table 20.1.

With this simple method, it was clearly possible to distinguish betweenthe different coating materials, although the differences are not too pronouncedbetween the two types of coatings. The reason for the relatively small difference

500

200

100

50


(a)

90

75

60

45

30

15

0


10G

100M

1M


(b)

90

75

60

45

30

15

0

gkli000gkli001gkli002gkli003gkli004gkli005gkli006gkli007

20.8 EIS measurements of the samples with the solvent-based epoxycoating. Last number is related to the number of days in therespective environment: (a) change from humidity to dryness;(b) change from dryness to humidity.

Imp

edan

ce (W

)Im

ped

ance

(W

)



might be seen in the imperfect surface preparation before the coating wasapplied. In Fig. 20.9 a cross section is shown in which some traces of theimperfect preparation are visible. Honeycombs persist on the surface, resultingin pores or insufficient coating thickness. This balances the different coatingresistance against water vapour because of weak points allowing easy diffusion,similar to uncoated mortar. Nevertheless, even though these weak pointswere found, the method revealed significant differences between the coatings.

Further improvement of the surface pretreatment, examples of which areshown in Fig. 20.10, will certainly allow the sensitivity of the method to beincreased with respect to smaller differences between coating materials.

20.4 Conclusions

Traditional testing techniques for the water vapour permeability of coatingsrequire free films or films on porous glass plates. The new method using EISallows measurements on a more realistic substrate (mortar).

20.9 Cross-section through typical mortar specimen. Twohoneycombs become visible. Reduced thickness of the coating, butcoating still covers the honeycomb.

Table 20.1 Impedance at 1 Hz after seven days, values in M Ohm

Type of sample Humid to dry Dry to humid

Initial value (0 day) 0.055 5000Uncoated 20 0.4Water-based acrylic 2 1000Solvent-based epoxy 0.6 3000



EIS allows uncoated, one-component acrylic or two-component epoxycoatings to be distinguished.

So far, the qualitative comparison of coatings is possible. Absolute andquantitative values for diffusion rates require further investigation.

Honeycombs are a problem if more similar materials have to be compared.To overcome this disadvantage the application of a filler or other kinds ofsurface preparation could be recommended, but no results are available sofar.

20.5 References

1. CEN 1062, Paints and Varnishes – Coating Materials and Coating Systems for ExteriorMasonry and Concrete: Part 2: Determination and classification of water vapour

20.10 Two possible techniques are shown to overcome thehoneycomb problem: (a) abrasive blasting removes loosely adherentmortar and opens the honeycomb for better paintability; (b) a fillermay be applied in order to seal the honeycombs and to act as asmooth surface for easier paint application.

(a)

Flexible filler

Coating

(b)

Original surface

A1 Blast cleaning with blasting agentsA2 High-water pressure blasting



permeability; Part 3: Determination and classification of liquid–water transmissionrate (permeability); Part 6: Determination of carbon dioxide permeability; Part 11:Methods of conditioning before testing. ISO 7783-2, Ausgabe:1999-03, Lacke undAnstrichstoffe Beschichtungsstoffe und Beschichtungssysteme für mineralischeUntergründe und Beton im Außenbereich - Teil 2: Bestimmung und Einteilung derWasserdampf-Diffusionsstromdichte (Permeabilität).

2. E. Bagda and R. Michel, ‘Zur Beurteilung des Feuchtehaushaltes von Beschichtungen’,Farbe Lack, 1995, 101, 603.

3. E. Bagda, ‘Zur Bestimmung der CO2 – Durchlässigkeit von Beschichtungsstoffen mitder Mörtelmethode’ Farbe Lack, 1994, 100, 100.

4. J. Tritthart and H. Gehmayer, Zement Beton, 1985, 1, 74–79.


263

21.1 The task

In 1999, a research project was launched to improve and to extend the well-known merits of electrochemical chloride extraction (CE) from reinforcedconcrete. Some results of previous work by the author gave the basic ideas[1]. A solution was sought in which the following features were to be included:

∑ Improved efficiency – rehabilitating with less energy in less time,∑ exclusion of possible chlorine gas evolution,∑ remote control of the entire CE process,∑ detailed logging of all events and sensor readings,∑ automatic use of detailed surveillance data of the structure,∑ strict avoidance of waste,∑ consumption of as low as possible a volume on a structure,∑ consideration of the rebar surface area for controlling the current density,∑ an easy-to-mount and re-usable electrode system with dimensionally stable

anodes,∑ freedom from optical imperfections on the concrete surface,∑ use of an energy supply that is commonly available.

In addition, all hardware was to be suitable for rough outdoor use and ableto be scaled up, being replaceable and supported by the manufacturers overa long period of time.

21.2 The solution

First of all, the problem of the inhomogeneity of the concrete surface had tobe solved. For this purpose, therefore, a grid overlay with cell dimensions of60 ¥ 60 cm was applied to the surface, where the cells could be investigated

21Chloride extraction from reinforced concrete

– a new defined way of application*

U. S C H N E C K, T. W I N K L E R and H. G R Ü N Z I G,Concrete Improvement Technologies, Germany

*The manuscript of this chapter was submitted originally for the EUROCORR 2001 andrepresents the state of knowledge and development in early 2001. Scientific and technologicalprogress since then has led to improved solutions.



on a scale that was small enough to take all special conditions into account. Amore detailed description of the approach can be found in Schneck et al. [2].

21.2.1 Basic settings

In order to maximise the efficiency of the CE process, the applied current mustbe related to the area of the rebar surface instead of the concrete surface. Withthe division of the concrete surface into small areas, different conditions ofconcrete cover, permeability or related rebar surface could be considered insuch a way that every cell is treated separately in order to achieve the optimalresult – to undertake the CE as quickly as possible and without overheating.

Having characterised the concrete surface, neighbouring cells with similarcharacteristics of concrete cover, rebar surface, permeability and chloridecontent can be combined again and controlled by a representative cell inorder to save expensive hardware.

In order to prevent the evolution of chlorine gas, an ion exchanger capableof regeneration almost without producing waste, was included within the CEprocess. Earlier investigations about the possible amount of chlorine gasevolution [3] drove the decision to include such an item.

21.2.2 Hardware components

A combination electrode (60 ¥ 60 cm) was designed that contains everythingneeded on the concrete surface for running CE: a dimensionally stable anodeand an electrolyte reservoir (fibreglass). Apart from this, there is an ionexchange layer and an outer stabilising plate that also provides an evaporationprotection. This sandwich electrode is mounted using a centrally positionednylon rod and is re-usable. Electrodes that work actively (controlling other,passive electrodes) also include a reference cell and an electronic switch.

Data acquisition equipment for outdoor use connects the active cells viaEthernet to a central computer. The extraction voltage is also supplied centrallyby high current power supplies (Fig. 21.1). For the supply of electrolyte, awater treatment system de-ionises tap water, adds some NaOH for raisingthe conductivity and distributes the electrolyte between the electrodes via asystem of hoses.

21.2.3 Control components

The chloride extraction process works by measurement and control withinthe cells. This means that cells or groups of cells are measured and controlledindividually, being ruled by centrally stored, individually defined start values.The CE system is scalable and has no limitations on the number of cells orgroups of cells (Fig. 21.2).


Chloride extraction from

reinforced concrete265

21.1 Schematic layout of the CE operating system.

Laptop

Remote PC-basedcontrol andsupervision

Public internet/GSM connection/

telephone

Gateway

Grid layout on a bridge structure

Voltage bus

Eth

ern

et

Local PC-based controland supervision

High current power supply

Data acquisition

Actor and sensor control for– switch– reference cell– voltage and currentCombination electrode

active/passive



All electrodes are supplied by a common voltage source – if possible at40 V DC. According to their predicted current consumption, the electrodegroups are divided into two groups and are switched alternately. When limitingcriteria (e.g. reference potential or current) are exceeded, the related groupswill be switched off until their next on-period. For safety reasons, severalinternal checks will shut down the process if signals are missing or valuesare out of range or an alarm state is reached. The CE program will run locallyat the site, but can be remotely controlled and provides the opportunity foralarm events to be sent to other connected computers, mobile phones, etc.When maintenance of the electrodes or data acquisition modules is needed,the related electrode groups can be paused manually.

The required treatment time is being estimated from the amount of chargethat was needed during laboratory tests with various concrete compositions,which can be compared with the concrete found on-site. The recorded amountof charge per group of the electrodes provides the first on-line indication ofthe possible success of the treatment. Tools can be added to signal the end of

21.2 Control menu for the CE process (graph bar, physicalconfiguration, logging).


Chloride extraction from reinforced concrete 267

treatment or saturation of the ion-exchange module. An evaluation tool can beused to monitor and analyze the whole CE process and to provide a detaileddocumentation of all sensor value changes, events and calculated results.

21.3 Description of configuration

21.3.1 Condition survey

As described in [2], before rehabilitation a detailed investigation of thestructure has to be done (Fig. 21.3). In addition to other information, thefollowing physical data are directly transferred into the CE control system:

∑ the rebar surface area for each cell (equals the curved surface of the outerrebar layer);

∑ the minimum concrete cover;∑ the average concrete cover.

Other data – mainly rest potentials and chloride contents – cannot be takeninto account automatically because there are not enough mathematicaldependencies on which to base valid formulae. Thus, the operator has toevaluate those data manually. Usually, this evaluation will be supported bythe trial desalination of some core samples taken from the structure in advanceand representing the actual concrete constitution.

21.3.2 Design of a chloride extraction applicationWith the imported grid layout and the data it contains, groups of cells mustbe configured. Those cells that have a similar rebar surface area, concretecover, permeability and chloride content can be combined into a group and

21.3 Appearance of the rebar layout being the cathode.

C5 C6 C7 C8 C9 C102,4…3,0 [m] 3,0…3,6 [m] 3,6…4,2 [m] 4,2…4,8 [m] 4,8…5,4 [m] 5,4…6,0 [m]

R20,6…1,2 [m] 0,2 0,2 0,4 0,4 0,4 0,3

64,0 60,3 59,1 29,9 52,5 54,745,0 46,0 43,0 0,0 43,0 43,0260/300 180/300 330/300 220/300 190/300 270/300

R31,2…1,8 [m] 0,2 0,2 0,4 0,3 0,4 0,2

60,6 66,6 59,8 32,9 51,0 56,143,0 53,0 43,0 0,0 30,0 43,0260,300 180/300 380/300 220/300 190/300 270/300

R41,8,…2,4 [m] 0,2 0,2 0,4 0,3 0,4 0,2

60,3 64,8 58,8 31,1 51,8 56,653,0 53,0 43,0 0,0 30,0 48,0260/300 180/300 330/300 220/300 190/300 270/300

R52,4…3,0 [m] 0,2 0,3 0,4 0,3 0,4 0,3

65,5 63,8 59,6 51,5 51,8 58,055,0 56,0 43,0 30,0 30,0 43,0260/300 180/300 330/300 220/300 190/300 270/300

R63,0…3,6 [m] 0,2 0,2 0,4 0,3 0,4 0,3

62,0 66,4 60,5 35,8 52,3 62,353,0 52,0 53,0 0,0 42,0 49,0260/300 180/300 330/300 220/300 190/300 270/300

Project

Globals

Widerlager Ost, RiFB DD

Data

Rebar

Docu

Erst-Monit, Okt 00

Wiederh.-Monit. I Dez 00

Wiederh.-Monit. II Jan 01

Wiederh.-Monit. III Feb 01

Wiederh.-Monit. IV April 01

Widerlager West, RiFB DD

Data

Rebar

Docu

Erst-Monit, Okt 00

Wiederh.-Monit. I Dez 00

Wiederh.-Monit. II Jan 01

Wiederh.-Monit. III Feb 01

Wiederh.-Monit. IV Apr 01



will be represented by the selected active cell/electrode. This electrode canthen control the chloride extraction process of the group.

According to previous experiments [4], chloride extraction can be designedto run with interuptions. A pause of some hours can be imposed whichraises the efficiency and reduces the current consumption. The power suppliescan share their power between two phases of groups that are switched alternately(see 21.2.3).

21.3.3 Setup on the concrete surface

The combination electrodes have to be attached to the concrete surface,being kept in position by nylon rods. A mounting plan generated with themonitoring files gives the exact positions for drilling the holes for the electrodefixings and reference cells. This is essential because the electrodes have to fitlike a tile layout.

After cross checking all cables and appliances, wetting of the electrodesand the concrete will start for a duration of about 3 days – depending on thepermeability of the concrete and the concrete cover thickness. The progressof wetting can be monitored by reading the rest potentials, which fall whenthe humidity front reaches the rebar layer.

21.4 Application to a highway bridge abutment

21.4.1 Description of the structure

The bridge is situated on the A4 highway, 50 km east of Dresden. It wascompleted in 1995. Although the abutments show no corrosion, a considerableamount of chloride (up to 2.5% by weight of cement) was detected in theconcrete of the splash zone. Other details of the surveillance – especially restpotentials – have been reported [2]. The road crossing the highway belowhas a slight descent so that the abutments are shaped like a rhombus. Becausethe highest chloride concentrations are found close to the curb, the orthogonalelectrodes have to follow this line, and the grid layout was rotated so as to beperpendicular to the curb/bottom line.

The initial chloride profiles for both abutments are shown in Fig. 21.4,each representing two columns of cells, ranging in steps of 0.6 m from thebottom of the abutment (R8) to a height of 2.4 m (R5).

The rebar layer is located at a depth of 5 to 6 cm. Physical inspectionsrevealed that, despite the start of electrochemical changes in the rebar vicinity,no signs of active corrosion were apparent. [2]. Therefore, the first aim ofchloride extraction was not to reduce the chloride concentration in the cathodearea but to extract chloride from the outer concrete cover, documenting thewhole process and verifying the equipment functionality while checking andverifying the data models given in [3] and in [5].



21.4.2 Setup and results of the application

According to the survey, the configuration data of the groups were as shownin Table 21.1.

In principle, the even and the odd groups were switched alternately in a12-hour-cycle. These 12-hour periods were reported to be much more effectivefor chloride removal than a constant voltage application [3]. For avoidingexcessively negative cathodic polarisation, minimum reference potential valuescould be set to switch off the related group during the operating interval. Intotal, the installation was left on the concrete surface for 50 days. In order tomaintain wetness, the electrodes were supplied with de-ionised water once aday by a ‘microdrip’ water hose system.

By use of the control program CITec CeControl, the following data werepermanently measured and saved according to specified differential values:groups (switch, voltage, current, reference potential), power supply (totalcurrent, voltage) and temperature. From the measured values, the chargeswere calculated (Table 21.2). Groups 0 to 3, in particular, shall be discussed,and the electrical values correlated with the measured chloride content.Although the concrete covered by groups 4 to 9 did not have a considerablechloride content or corrosion activity, it was treated as well in order to obtaina concrete with higher alkalinity.

2.5

2

1.5

1

0.50C

l (m

ass

% c

emen

t)

1 2 34 5

Depth from surface (cm)

C7R8C7R7

C7R6C7R5 Cells

2.5

2

1.5

1

0.50C

l (m

ass

% c

emen

t)

1 23

45

Depth from surface (cm)

C15R8C15R7

C15R6C15R5 Cells

21.4 Initial chloride profiles of the eastern abutment.

Table 21.1 Arrangement of the groups according to the structural situation of theabutment; numbers representing the positions of the active cells; grey cells leftempty

7

4

0

8 9

5 6

2 31



The reference potentials obtained during the chloride extraction processwere very negative during the switch-on phases of the groups. Since no I-R-correction could be made these values cannot be interpreted. However, duringthe switch-off phases, the values increased by some 100 mV and give anorientation about the polarisation stage of the related rebars. Figure 21.5shows how the potential of the groups 7 to 9 returned to the range of –500to –600 mV vs. MnO2 during the off-times from active potentials between–1500 to –7000 mV. Furthermore, the effects of process interruptions due tohardware tests can be seen.

Table 21.2 Cumulative operation hours and charges for groups 0 to 3

Installation days 21 days 28 days 38 days 50 days

Installation hours (cumulative) 530:39:00 694:59:00 934:13:00 1199:26:00Operation hours (cumulative) 152:06:00 262:32:00 360:46:00 537:49:00Ratio operation/installation 0.29 0.38 0.39 0.45

Group 0Operation hours (cumulative) 24:44:00 82:15:00 137:49:00 231:00:00Charge [A h m–2] (cumulative) 13.20 27.50 39.72 68.46




Ref

eren

ce p

ote

nti

al,

no

t I-

R c

orr

ecte

d(m

V v

s M

nO

2)

0

–1000

–2000

–3000

–4000

–5000

–6000

–700014.9 19.9 24.9 29.9 4.10 9.10 14.10

Date

Group 8Group 9Group 7

21.5 Development of reference potentials – not I-R-corrected – ofgroups 7 to 9 during times of switching on and off.



The measured current densities (Fig. 21.6) remained quite low comparedwith the high voltage of 40 V. This might be due to the vertical layout wheremoistening cannot be conducted as effectively as in a horizontal layout, andthe CEM III concrete has, for a given concrete cover, a higher electrolyticresistivity.

The analysis of the chloride content showed a very reasonable removalefficiency. In Fig. 21.7 the chloride content in rows 7 and 8 is shown. Thesamples were taken from the grid cells without specific reference to the

max G0 avg G0max G1 avg G1max G2 avg G2max G3 avg G3

16.9 21.9 26.9 1.10 6.10 11.10 16.10 21.10Date

1.00

0.80

0.60

0.40

0.20

0.00

Cu

rren

t d

ensi

ty (

A m

–2)

21.6 Development of the maximum and average current densities– related to the rebar surface of the groups 0 to 3.

12345

1 Start2 Day 213 Day 284 Day 385 Day 50 1 cm 2 cm 3 cm 4 cm 5 cm 1 cm 2 cm 3 cm 4 cm 5 cm

Free

ch

lori

de

(mas

s %

cem

ent)

2.00

1.80

1.60

1.40

1.20

1.00

0.80

0.60

0.40

0.20

0.00

21.7 Progress of chloride removal in zones located 0 to 60 cm fromthe bottom (right) and 60 to 120 cm from the bottom (left).



rebars and, with the permanent removal progress, an even chloride distributionwithin the cells can be assumed.

The values shown are average results from two grid cells. No correctionshave been made so that natural deviations are part of the graphic. Afterfinishing the treatment, an average chloride content of 0.29% remained inthe concrete. This corresponds to a total removal of ca. 70%. For the future,the corrosion-inducing chloride content should have been raised accordingto the development of hydroxyl ions on the rebar surface. On removing theelectrodes from the concrete, no visible changes to the visual appearance ofthe concrete surface could be seen. Because of the trial stage of the projectno actions were taken to protect the surface against new chloride.

21.5 Results of the follow-up survey

Immediately after finishing the treatment, potential mapping was carriedout. This showed (Fig. 21.8) very low potentials and obviously high remnantcathodic polarisation. For the next measurement, four months later, a muchmore positive result and less value deviation was found (Fig. 21.9).

Another important finding comes from the humidity measurements. Table21.3 compares surface resistance values and shows that, during the timebetween finishing the CE and February 2002, no change in humidity couldbe measured. This means that the change of potentials is simply due to theprocess of losing polarisation. Furthermore, it shows the evidence of improvedhumidity content during the CE treatment.

Chloride sampling from 13/02/2002 (Table 21.4) shows an interestingresult. The outer 2 cm contains (in quite an even distribution) much morechloride due to the winter season. For the inner 3 cm the results are almostidentical. The evaluation of the potentials from 13/02/2002 showed no indicationof corrosion activity, and the raised chloride content in the outer zones doesnot appear to influence the rebars.

Ro

ws

012

34

5

67

80 2 4 6 8 10 12 14 16 18 20 22 24 26

Columns

–600.0 –400.0 –300.0 –200.0 –100.0 0 100.0

21.8 Result of the potential mapping from 02/11/2001 (mV vs. SCE).



21.6 Conclusions

Electrochemical chloride extraction has been demonstrated successfully onthe substructure of the highway bridge at exit 34 on the A4 highway. Despitesome testing-related interruptions, within 7 weeks ca. 70% of the chlorideswas removed at quite a low power consumption.

The technical concept allows a free and mostly appropriate electrodeconfiguration, does not interfere with traffic and provides detaileddocumentation. The combination electrodes do not cause any loss or changein the appearance of the concrete surface. During the follow-up survey itcould be ensured that the corrosion activity was removed with the chloride,but the need for an appropriate protection for the treated surface becomesclear as well, if the cause of the damage is still present.

Although the new chloride extraction system requires much effort in timeof preparation and in connection with the electrode material that has to beproduced in advance, it is able deliver more detailed and more uniformresults at a higher efficiency than usual, because:

∑ the CE system is open and scalable and can be fitted under any circumstancesto the structure;

21.9 Rest potentials four months after finishing the CE (13/02/2002)(mV vs. SCE).

Ro

w

0

1

2

3

4

5

6

7

80 2 4 6 8 10 12 14 16 18 20 22 24 26

Columns

–600.0 –400.0 –300.0 –200.0 –100.0 0 100.0

Table 21.3 Statistical evaluation of the surface resistivity in rows 5 to 8 (0 to 2.4 mfrom bottom)

Surface resistance Before start of CE After finishing CE Four months later[kW m] 21/06/01 02/11/01 13/02/02

Average 14.4 4.1 3.7Standard deviation 10.2 1.6 1.2Minimum 2.0 1.0 2.0Maximum 73.0 10.0 6.0



ent in concrete274

Table 21.4 Free chloride content [wt-% cement] at 13/02/2002 and changes compared with 17/10/2001

C7 C15

1 cm 2 cm 3 cm 4 cm 5 cm 1 cm 2 cm 3 cm 4 cm 5 cm

R7 content 0.91 0.74 0.23 0.18 0.14 0.85 0.77 0.19 0.20 0.16change 0.59 0.36 0.04 –0.02 –0.08 0.57 0.39 –0.06 0.04 0.00

R8 content 0.98 0.46 0.20 0.18 0.18 0.84 0.48 0.19 0.40 0.36change 0.43 0.10 –0.03 0.00 –0.03 0.34 0.13 –0.05 0.06 0.02



∑ the success of treatment can be monitored automatically by means ofdifferent measurements (e.g. amount of charge, development of off-potentials);

∑ inhomogeneous concrete zones will be considered automatically and betreated in an optimal way. They have to be known with their coordinates,and the process limitations will decline or accelerate the treatment (e.g. atre-profiled zones with a different permeability);

∑ all important basic physical data of a structure can be taken into accountautomatically, allowing the chloride extraction process to be related to thetrue rebar surface area;

∑ with the logging protocol there is a very detailed report, containing allsensor signals, actor actions and calculations e.g. of charge vs. time, to begenerated for filing and as a process description;

∑ the logged data provide a basis to verify and to expand mathematicalmodels that can be improved with every new application;

∑ there is no change in the visual appearance, because the fibreglass hashigh chemical stability; and

∑ no waste will be produced, and the electrodes are entirely re-usable.

For evaluating and implementing mathematical models, much more datahave to be collected in order to obtain a reliable and statistically baseddatabase.


We gratefully acknowledge that this project has been carried out with thefinancial support of the Federal Ministry of Economics of the FRG withinthe FUTOUR program. Furthermore, we thank the Saxonian HighwayAdministration for its assistance in the reference application of the CEsystem.

21.8 References and further reading

1. U. Schneck, ‘Zu Mechanismen der Stahlkorrosion in Beton bei der elektrochemischenEntsalzung’, Diss. TU Dresden, Dresden (1994).

2. U. Schneck, T. Winkler and S. Mucke, ‘Integrated system for corrosion monitoring atreinforced concrete structures’, Proc. Eurocorr, 2001.

3. U. Schneck, ‘Investigations on the chloride transformation during the electrochemicalchloride extraction process’, Mater. Corros., 2000, 51, 91–96.

4. U. Schneck, H. Grünzig and S. Mucke, ‘Pulse width modulation – investigations forraising the efficiency of an electrochemical chloride extraction from reinforced concrete’,Proc. Eurocorr, 2001.

5. A. M. Hassanein, G. K. Glass and N. R. Buenfeld, ‘A Mathematical Model forElectrochemical Removal of Chloride from Concrete Structures’, Corrosion, 1998,54(4), 323–332.



6. prEN 14038-1, ‘Electrochemical re-alkalisation and chloride extraction treatments forreinforced concrete – Part 1: Re-alkalisation’, DIN e.V., 2000.

7. D. Whitmore, SHRP Product 2033: Guideline For Performing Electrochemical ChlorideExtraction To Concrete Structures, AASHTO, Washington DC, USA.


277

22.1 Introduction

Cathodic protection (CP) of reinforcing steel in concrete structures has becomea well-established and widely used technique for stopping corrosion in caseswhere chloride contamination has caused corrosion and concrete damage.Practical cases have been described [1, 2], a European standard was publishedrecently [3] and further technical and economic information was given in[4]. In particular, the introduction of conductive coatings has stimulated theapplication of CP to concrete structures, partly due to their lower price incomparison with activated titanium systems. However the shorter servicelife of conductive coatings may be a disadvantage. Usually, their service lifeis assumed to be equal to that of ‘normal’ coatings for concrete. In practicalcases, a 10-year guarantee of absence of corrosion is given, suggesting aservice life of slightly more than 10 years. Hard evidence to support or denythis is practically non-existent. Within the framework of the European ConcertedAction COST 521 ‘Corrosion of steel in reinforced concrete structures’,project NL-1 ‘Cost effective cathodic protection’, a study was carried outinto the degradation of a particular conductive coating system, with whichgood experience existed in Norway and The Netherlands.

22.2 Theoretical background

Cathodic protection involves current flow through the concrete from theanode to the steel (cathode). The dominant reaction at the steel surface isreduction of oxygen, suppressing corrosion and producing hydroxide. Themain reaction at the anode is oxidation of hydroxide, producing oxygen.Hydroxide consumption is equivalent to acid production, which may dissolvecalcium hydroxide and other alkaline components of the concrete (cementpaste). The theoretical amount of acid formation can be calculated usingFaraday’s law, from the amount of electrical charge (current ¥ time) that haspassed.

22Microscopy study of the interface between

concrete and the conductive coating usedas an anode for cathodic protection

R. B. P O L D E R and W. H. A. P E E L E N, TNO Buildingand Construction Research, The Netherlands, and

J. L E G G E D O O R and G. S C H U T E N,Leggedoor Concrete Repair, The Netherlands



For each mole (equivalent) of electrons flowing, a mole of hydroxide isproduced (at the cathode) and consumed (at the anode), equivalent to a moleof acid being produced. If the current density is 1 mA m–2 (assuming anodesurface equals concrete surface), 0.33 mole of hydroxide ions are consumedper m2 of anode/concrete interface in one year. If all these hydroxide ions areprovided by solid calcium hydroxide in the hardened cement paste, this wouldcorrespond to 12 g m–2 y–1 of calcium hydroxide dissolved or about 24 g ofcement paste dissolved per m2 per year. With a density of 2000 kg m–3 thatis about 12 mm ‘dissolution depth’ per year for the theoretical maximumamount of solid material dissolved by acid production at 1 mA m–2.

It should be realised that the effective amount of hydroxide consumedwill be mitigated due to the current circulation: the current causes migrationof hydroxide ions from the cathode to the anode. Ion transport in concretedue to current flow may be described using transport numbers, as for aqueoussolutions. Assuming that sodium and hydroxide ions are the only mobilespecies, they will have transport numbers of about 0.20 and 0.80, respectively.That means that 80% of the total current is carried by hydroxide andconsequently 80% of the theoretical maximum acid production will beneutralised by hydroxide migrating from the cathode to the anode, as shownin Fig. 22.1. The amount of dissolution calculated in this way (‘effective’amount) is about 0.20 of the theoretical maximum amount. In view of ourunderstanding of electrical transport in concrete, this amount of hydroxideconsumption and cement paste dissolution seems more realistic than themaximum amount of dissolution calculated above. A current density of1 mA m–2 would produce an effective amount of dissolution of 2.4 mm in thecoating/concrete interface per year.

In the study reported here, UV/visible light microscopy and scanningelectron microscopy were used to assess the amount of dissolution that hadoccurred in samples subjected to CP in the field.

In principle, oxidation processes at the anode could also involve carbonparticles in the conductive layer (producing carbon dioxide or monoxide) orchloride ions from the hardened cement paste (producing chlorine gas). Athigh rates, oxidation of carbon would increase the porosity of the conductive

Cathode; hydroxideproduction = Q 2H2O + O2 + 4e– Æ 4 OH–

Hydroxide 0.8 Q Sodium0.2 Q Total charge passed = Q

4 OH– Æ 2H2O + O2 + 4e–

22.1 Mass transport and electrode reactions due to current flowunder CP.

Anode; hydroxide consumption = Q


Microscopy study of the concrete–coating interface 279

layer. An example of similar processes has been described previously; seereference [5]. However, due to subsequent reactions this process would producethe same amount of acid as oxidation of hydroxide. Oxidation of chlorideions would decrease the consumption of hydroxide. This reaction would,however, not be visible using microscopy, but at high rates it would producethe smell of chlorine gas. At the current densities relevant to this study, it isimprobable that this would be detectable without special measures.

22.3 Samples and microscopy examination

22.3.1 Sample origins

Samples were obtained from two types of structures with CP installed: aNorwegian building and five Dutch apartment houses. Aassiden is a high-rise building in Oslo, Norway, with cast in chlorides. It was provided withCP with a conductive coating (primer) of the type AHEAD in 1990–1991.The coating system was applied directly onto the concrete surface, whichhad been prepared by sand blasting. A non-conductive top coat was appliedover the conductive coating. Two cores were taken in 1999 from this bridgeand sent to TNO. Five apartment buildings in Groningen, in the northern partof The Netherlands, were provided with a CP system based on the samecoating in the years 1993 to 1997, one building each year. Further details ofthese CP systems are given in [1]. For aesthetic reasons, the surface was gritblasted and subsequently levelled with a filling mortar (fine quartz sand andPortland cement) before applying the coating. A non-conductive top coatwas applied over the conductive coating. In 1999, one core was taken fromeach of these buildings. Both types of CP system involved a current sourcewhich reverses the polarity of the current every minute for just one second.In all cases, the CP systems complied with the usual criterion of sufficientsteel protection for atmospheric concrete structures (>100 mV depolarisation).The driving voltages were between 1.5 and 2.0 V. Accurate (local) currentdensities were not available, but average current densities were about 0.5 to1 mA m–2 of concrete surface. Cores were taken from positions close to theprimary anodes (silver wire mesh), where the current density may be supposedto be on the higher side of this range.

22.3.2 Sample preparation

Samples of 50 mm along the surface and 30 mm depth were prepared formicroscopy by careful drying and impregnating 10 mm thick slices takenfrom the cores with epoxy resin containing fluorescent dye and polishingdown to ‘thin sections’ of about 25 mm thickness. This allows examinationusing transmitted visible and ultraviolet light. The procedure is similar toASTM Standard C856-88 ‘Standard Practice for Petrographic Examination



of Hardened Concrete’. The magnification used was 100 to 400 times (fieldof view of the order of 1 mm ¥ 1 mm) and the resolution was 10 mm or better.Two samples (NO1 and GL1) were prepared for scanning electron microscopy(SEM), using the thin sections described above, by coating them with carbonfor the necessary conduction. The magnification applied was 50 to 4000times. The highest resolution obtained in the SEM experiments was 1 mm.

22.4 Results

Light microscopy examination showed various characteristics. The two-layercoating system had a total thickness of 0.2 to 0.3 mm. This is the upper darkband in Fig. 22.2, showing sample NO1 in visible light. The two layers canbe distinguished in the same sample using ultraviolet light, as shown in Fig.22.3. Here, only the conductive part of the coating is visible as a dark band;the top coat transmits ultraviolet light and appeared with the typical fluorescentyellow/green colour in the upper part. The bond between the two layers(conductive and non-conductive) was good. In both figures, some air bubblesare visible inside the conductive coating layer. The lower part in the figuresis the concrete, consisting mainly of hardened cement paste. The features inthe lower middle and right which are whitish in Fig. 22.2 and black in Fig.22.3 are aggregate particles (actually fine sand grains). The roughly vertical

22.2 Interface of concrete (below) and conductive coating/top coat(dark zone above) of sample NO1 in plane polarised light, showingaggregate (white particles), in cement paste (darkish), with a verticalcrack; field of view is 1.4 mm ¥ 0.8 mm.



feature passing from the conductive coating/cement paste interface along themiddle sand grain is a microcrack. It measures about 50 mm at the widestpoint (lower part) and about 10 mm at the narrowest point (curved parttouching the sand grain).

The interface between the coating system and the concrete substrate wasstudied in detail. Looking again at Fig. 22.2 (sample NO1), it is clearlyvisible that the black coating layer tightly adheres to the dark brown cementpaste, without any sign of a gap or dissolved zone between them. Fromcomparison with the microcrack described above, it can be seen that any gapor dissolved zone that might be present has a width of less than 10 mm. Aciddissolution would have been even more visible as a zone of increasedtransmission in Fig. 22.3. The absence of any detectable lighter zone in theinterface confirms the absence of such dissolution on the scale of observationof these figures.

Similar results have been obtained from sample GL1. This is shown inFig. 22.4 in visible light at lower magnification than the previous figures. Inthe lower part, a grey aggregate particle is visible; the middle zone is the finefilling mortar paste, the black band is the conductive coating plus top coatand the light upper band is the epoxy resin used to prepare the thin section.Inside the paste, an air bubble (circular, left) and whitish fine aggregates arevisible. On the right hand side, an irregularly shaped light zone is presentbetween the coating and the substrate. This is a region where adhesion is

22.3 The same interface of NO1 as in Fig. 22.2, fluorescent inultraviolet light with aggregate particles (dark), conductive coatinglayer (middle, black) and non-conductive coating layer (upper, light);field of view 1.4 mm ¥ 0.8 mm.



poor. Any apparent deformation or cracking of the coating is absent. The gapdoes not contain solid remnants of paste dissolved by acid; the void iscompletely empty. It is concluded that this void has been present since theapplication of the coating. It appears that the filling mortar contains a highamount of air bubbles and shows high overall porosity, which in part is dueto its composition or mixing, part introduced during the application.

The overall results of the light microscopy are described in Table 22.1.The Norwegian samples NO1 and NO2 showed a very good bond betweenthe coating and the concrete. In some Dutch samples (codes GL and GB)various levels of poor bonding were found, which appeared to have beenpresent from the beginning. Air bubbles in the surface and in particular in thefilling mortar prohibited intimate contact between the coating and the substrateover considerable parts of the interface. Signs of dissolution or chemicaldeterioration were absent. None of the samples showed deterioration in thecontacting parts of the interface where the protection current would havepassed, as is illustrated in Fig. 22.2, 22.3 and 22.4. Such dissolution wouldhave been visible as narrow zones (filled with resin), appearing light in thefluorescent photos. Increased porosity of the conductive layer was not observed.Local carbonation occurred in the paste of all samples, which is recognisedas calcium carbonate, showing a change of colour in the cement paste undercrossed-polarised light. Local frost attack was found in two of the Dutch

22.4 Concrete/coating interface of sample GL1 in plane polarisedlight with large aggregate below (grey), filling mortar (darkish) withfine aggregate (white), spherical air bubble (left), conductive layerand top coat above (black), showing non-adhering parts (right); fieldof view 3 mm ¥ 2 mm.



samples, shown by some microcracking well below the interface, which isobviously not related to the passing of protection current. It may be assumedthat this local frost attack was present before the CP coating was applied.

A scanning electron microscopy photo of sample NO1 is given in Fig.22.5, showing the interface between the coating (upper part) and the hardenedcement paste (lower part). Although the difference may be difficult to see inthis photo, both phases can be identified readily at lower magnifications (seealso Fig. 22.6). Even at this high magnification (the bar indicates 10 mm), noclear gap is visible between the paste and the coating. If there are localfissures between paste and coating (upper arrow), they are less than 1 mmwide. They are not wider than what seem to be microcracks (lower arrow) inthe paste itself.

SEM photos of GL1 are shown in Fig. 22.6 and 22.7. Figure 22.6 showsthe same area as in Fig. 22.4 (see air bubble). At this relatively low magnification(bar indicates 500 mm), the overall good adhesion between the coating (darkgrey, middle band) and the mortar (lower part, light grey) is clear. Figure22.7 confirms this at a tenfold higher magnification.

Table 22.1 Results of microscopy of conductive coating/concrete interfaces; NO1,NO2 samples from Norway, GL1-4, GB5 samples from The Netherlands

Sample Condition of Coating–concrete Description and general remarkscode coating bond

NO1 1991 Intact, no Excellent; no Coating is continuous over theevidence evidence of entire surface; coating–concreteof damage de-bonding bond is very good; no visual

damage to coating

NO2 1991 Intact; identical Very good Coating is continuous and bondto NO1 is very good; no damage to

coating

GL1 1993 Intact, no Moderate Coating is continuous over theevidence of entire surface but locally bond isdamage absent

GL2 1994 Intact, no Moderate to poor Identical to GL1evidence ofdamage

GL3 1995 Local tearing Moderate to poor Coating is not continuous; there(due to sample is clear evidence of tearing andremoval or local absence of bonding overpreparation) the entire surface

GL4 1996 Intact, no Moderate to poor Identical to GL1 and GL2evidence ofdamage

GB5 1999 Intact, no Reasonably good Coating is continuous; coating–evidence of concrete bond is quite good;damage no visual damage to coating



None of the samples studied by SEM showed evidence of chemicaldissolution in the interface in the range of magnifications applied, givingresolutions down to 1 mm. These findings support the light microscopy results.Chemical analysis by EDAX showed the usual composition of cement paste,with oxides of Ca, Si, Al, Mg, Na, K (and some Fe and S), both in theinterface and the bulk, with no obvious gradients.

22.5 Scanning electron microscopy photo of sample NO1.

22.6 Scanning electron microscopy photo of sample GL1 at moderatemagnification (roughly same area as Fig. 22.4)

Conductivecoating

Cementpaste

Inte

rfac

ial

fiss

ure

?M

icro

crac

kin

pas

te

Top coat

Conductivecoating

Filling mortarpaste



22.5 Discussion

In Table 22.2, the period for which structures that had samples taken fromthem received CP and the amount of charge passed are given expressed asmole of hydrogen ions per square metre of concrete surface that have beenformed. Because the samples were taken near the current distributors, it isreasonable to assume that the samples are representative of the structures interms of current density. The following columns in the table give the maximumand the effective dissolved thickness resulting from the theoretical calculationsgiven above. Finally, the observed material loss using light microscopy andscanning electron microscopy (for two samples) are given. It follows that theinvestigated samples NO1, NO2 (9 years) and GL1 (6 years), GL2 (5 years)and GL3 (4 years) have certainly experienced more than the amount ofcharge that would have caused 10 mm dissolution. This is the poorest resolutionof the light microscopy examination technique. The two samples investigatedusing scanning electron microscopy show that even less than 1 mm of dissolutionis present.

The results show that:

∑ the amount of dissolution is very much less than the theoretically calculatedmaximum;

∑ the amount of dissolution is less than the amount calculated, consideringthe mitigation by transport of hydroxide by ionic migration;

∑ there may be another mechanism, other than migration, that mitigates theacid dissolution of cement paste at the anode/concrete interface; and

22.7 Scanning electron microscopy photo of sample GL1 at highermagnification.

Conductivecoating

Cementpaste

Acc.V Spot Magn10.0 kV 3.0 500x

Det WD 50 mm SE 11.4 GL1 boundary 01060a04



∑ speculating, this could be diffusion of NaOH to the interface from elsewherein the paste.

The absence of acid dissolution suggests that degradation in these systemsis substantially less than calculated from the amount of current circulationand the expected anode reactions. Deterioration due to acid dissolution hasnot occurred on a detectable scale. It appears that the service life is notdetermined by acid production at the coating/concrete interface. The servicelife of these conductive coating CP systems may be much longer than theoldest installation investigated here, which was nine years.

22.6 Conclusions

Samples taken from concrete structures protected by cathodic protection(CP) with a conductive coating of the type AHEAD were studied using lightmicroscopy and scanning electron microscopy. Chemical dissolution in thecoating/concrete interface was absent down to the level of the resolution ofboth types of microscopy. On theoretical grounds, hydroxide consumption,equivalent to acid production, should have taken place to a certain extent.The maximum amount of acid production that theoretically could have occurredin the samples is relatively large (and would certainly have been visible). Itis obvious that such an amount of acid attack has not occurred. Hydroxideion migration due to current flow would reduce the total amount of acidproduction. Considering the transport numbers of hydroxide and other ions,the ‘effective’ amount of acid production was calculated. However, such alevel of dissolution was again not observed in samples having received CP

Table 22.2 Duration of cathodic protection (CP), amount of electrical charge passed(as moles of hydroxide consumed) and calculated amount of dissolution for sevensamples of conductive coating/concrete interfaces assuming a protection currentdensity of 1 mA m–2 (concrete interface) and observed amount of dissolution bylight microscopy and scanning electron microscopy (SEM)

Sample Duration Charge Calculated Calculated Dissolved Dissolvedof CP passed maximum effective as observed as observed(year) (mole of dissolved dissolved by light by SEM

hydroxide) (mm) mm) microscopy (mm)(mm)

NO1, 9 3 108 22 <10 <1NO2GL1 6 2 72 14 <10 <1GL2 5 1.7 60 12 <10 –GL3 4 1.3 48 10 <10 –GL4 3 1 36 7 <10 –GB5 ca. 1 0.33 12 2.4 <10 –

– not determined



current for more than four years. It seems probable that other mechanismsreduce the amount of acid attack on the cement paste under the currentdensities studied (1 mA m–2 of concrete surface).

With regard to the service life of conductive coating CP systems, theresults suggest that service lives of well over ten years may be obtained. Inthe samples studied, no degradation of the interface was present after six tonine years, although they originated from structures fully exposed to North-West European field conditions, including frost and snowfall. The bondbetween the coating and the concrete was relatively poor in the samples fromThe Netherlands, due to non-optimal application of the filling mortar used toeven out the surface of the structures for aesthetic reasons. Despite that, nobond deterioration had occurred on these structures and the CP systemfunctioned properly. It appears that the service life of this type of conductivecoating CP system is well above the ten-year period that is normally guaranteedby the supplier or the contractor.


Mr Jan Eri of Protector, Drammen (Norway) is gratefully acknowledged forproviding the samples, information on Aassiden and his permission to publishthe information. The contribution of the members of COST 521 WorkingGroup C-2 ‘Electrochemical Maintenance Methods’ in the discussion ofpreliminary results is thankfully acknowledged.

22.8 References

1. R. B. Polder, 1998, ‘Cathodic protection of reinforced concrete structures in TheNetherlands – experience and developments’, in Corrosion of Reinforcement in Concrete– Monitoring, Prevention and Rehabilitation, Papers from Eurocorr’97, Mietz, J.,Elsener, B. and Polder, R. (Eds.), The European Federation of Corrosion Publicationnumber 25, The Institute of Materials, London, ISBN 1-86125-083-5, 172–184.

2. C. Haldemann and A. Schreyer, 1998, ‘Ten years of cathodic protection in concrete inSwitzerland’, in Corrosion of Reinforcement in Concrete – Monitoring, Preventionand Rehabilitation, Papers from Eurocorr’97, Mietz, J., Elsener, B. and Polder, R.(Eds.), The European Federation of Corrosion Publication number 25, The Institute ofMaterials, London, ISBN 1-86125-083-5, 184–197.

3. CEN, 2000, Cathodic protection of steel in concrete, EN 12696:2000.4. R. Cigna, C. Andrade, U. Nürnberger, R. Polder, R. Weydert, E. Seitz (Eds.), ‘Corrosion

of steel in reinforced concrete structures, Final Report’, COST 521, 2003, EuropeanCommission, Directorate-General for Research, EUR 20599, ISBN 92-894-4827-X,238 pp.

5. J. Mietz, J. Fischer and B. Isecke, ‘Cathodic protection of steel-reinforced concretestructures – results from 15 years’ experience’, Mater. Perf., December, 2001, 22–26.


288

23.1 Introduction

In reinforced concrete structures exposed to sea-water, pitting corrosion usuallyoccurs in the emerged part (tidal and splash zones), where wetting and dryingcycles favour the presence of both oxygen and chlorides. Corrosion hardlyever initiates in the reinforced concrete parts that are permanently immersedin seawater, where the relative lack of oxygen leads to very negative valuesof potential, and the chloride threshold for the onset of pitting corrosion ishigh.1,2 Furthermore, even when corrosion initiates, the corrosion rate isnegligible owing to the small amount of oxygen that can reach the steelsurface.

Once pitting corrosion has initiated in the emerged part of a structure,cathodic protection can be used to control the corrosion rate.1,3,4 Owing tothe high resistivity of concrete and to the complexity of the reinforcementgeometry, current is usually applied by means of a distributed anode placedon the surface of the concrete (e.g. an activated titanium mesh embedded ina cementitious overlay) and a current feeder. If the structure is buried orimmersed in seawater, external sacrificial anodes can also be used.

There have been studies of the use of local sacrificial anodes to providecathodic protection to corroding steel in the non-submerged part of marinepiles.5–7 These have shown that, unless the concrete resistivity is very low,the protection provided by submerged anodes is of limited effectivenessabove the waterline, so that above-surface extended anodes are required.

For several reasons, sacrificial anodes can be expected to be more effectivewhen applied to new structures in order to achieve cathodic prevention. Thistechnique relies on a cathodic current applied to the passive reinforcement inchloride-free concrete and it is aimed at delaying the initiation of pittingcorrosion. It leads to an increase in the critical chloride content, since itlowers the steel potential, and increases the pH at the steel/concrete interface,as the cathodic reaction takes place at the steel surface.1 Laboratory and fieldexperiences8–10 have shown that even impressed current densities lower than

23Protection of reinforced concrete piles inmarine structures with sacrificial anodes

L. B E R T O L I N I, M. G A S TA L D I,M. P. P E D E F E R R I and E. R E D A E L L I,



Protection of reinforced concrete piles in marine structures 289

2 mA m–2 can maintain steel at potential values where pitting corrosioncannot initiate when the chloride content exceeds 3% by weight of cement.Such current densities would be insufficient for the cathodic protection ofalready corroding steel, which requires current densities on the order of10–20 mA m–2.

Cathodic prevention also differs from cathodic protection with regard toits throwing power.11–12 It has been shown that, in spite of the high resistivityof concrete, the beneficial effects of cathodic prevention can extend toreinforcing bars at significant distances from the anode, owing to the highercathodic polarisability of passive steel compared with that of corroding steel.Conversely, the effects of cathodic protection are usually limited to distancesof a few tens of centimetres from the bars nearest to the anode.12 The enhancedthrowing power of cathodic prevention suggests that the above-mentionedlimitations of submerged sacrificial anodes applied for cathodic protectionof marine piles could be overcome if the anodes were applied to new structures,i.e. before corrosion initiates.

This paper reports the results of a study on the applicability of submergedsacrificial anodes to hinder corrosion initiation in the non-submerged part ofmarine piles and discusses the results of laboratory research aimed at studyingthe height above sea level to which protection and prevention can be achieved.Only a summary of the main findings is presented here; a more detailedaccount of the results has been published elsewhere.13

23.2 Experimental procedure

Tests were carried out on reinforced concrete columns with a base of 15 ¥15 cm and a height of 120 cm (Fig. 23.1a). Fifteen horizontal bars 10 mm indiameter were embedded in the concrete at 8 cm intervals. In order to simulateoperative conditions of cathodic prevention, a specimen was made withchloride-free concrete; another one was made with concrete contaminatedwith 3% chloride by weight of cement, to promote corrosion of the embeddedsteel and, thus, reproduce conditions of cathodic protection.

The steel bars were electrically connected outside the concrete; shuntresistances were used to measure the current circulating in each bar. A referenceelectrode made of a thin wire of mixed metal oxide (MMO) activated titaniumwas fixed near the middle of each bar for potential measurements.

Concrete was mixed with 350 kg m–3 of cement CEM II/A-L 42.5R(according to ENV 197-1 Standard), a 0.55 water/cement ratio and1900 kg m–3 of limestone aggregate. Chlorides were added as CaCl2 to themixing water. After curing, the columns were partially immersed in a shallowaqueous solution with 3.5% by weight of sodium chloride, so that only thesteel bar at the bottom was submerged. Tests were carried out at roomtemperature, and the columns were regularly wetted with the test solution in



order to simulate splashes of seawater. The 15 steel bars were connected tosacrificial anodes, made of a commercial Al–Zn–In alloy for cathodicprotection, which were immersed in the test solution. After 16 months oftesting, half of the bars were disconnected in both specimens to achieve anew configuration with only eight protected bars, with an interaxial spacingof 16 cm (Fig. 23.1b).

Steel potential and current density were monitored in each bar. Depolarisationtests were regularly carried out by disconnecting the steel bars from theanodes for 24 h during which the potential of each bar was monitored todetect 4-h and 24-h decay. Normally rebars were not disconnected from eachother during these tests; however, some tests were also carried out byindividually disconnecting every bar. During the depolarisation tests, thepotential of the activated titanium reference electrodes was calibrated againstan external calomel reference electrode (SCE).


Before the application of cathodic protection, the free corrosion potential ofthe steel bars was of the order of –200 to –300 mV vs SCE in chloride-freeconcrete and about –450 mV vs SCE in chloride-contaminated concrete. The

15 cm 15 cm

120

cm

8 cm

16 c

m

SacrificialAl-Zn-Inanodes

3.5%NaCl

(a) (b)

23.1 Schematic representation of columns 15 ¥ 15 ¥ 120 cm for testsof cathodic prevention and cathodic protection with sacrificialanodes: (a) initial configuration with 15 connected bars and (b)configuration with eight connected bars.



potential of the anode was about –1050 mV vs SCE. Figures 23.2a and 23.2bshow changes in time of the potential of the rebars, after connection to thesacrificial anodes. The cathodic polarisation of the steel bars following theirconnection to the anodes depended on the height above the level of thesolution.

In the specimen with passive steel in chloride-free concrete (Fig. 23.2a), theimmersed bar reached a potential of about –1000 mV vs MMO (–1050 mVvs SCE), i.e. roughly the same potential as the sacrificial anodes. The secondbar, 8 cm above the level of the solution, had a potential of around –800 mVvs MMO (–850 mV vs SCE). More positive values of potential were measuredin the bars placed at greater heights. Nevertheless, even the bar at the top ofthe column (112 cm above the level of the solution) reached a potential valueabout 100 mV lower than the initial free corrosion potential.

In the column with chloride-contaminated concrete (Fig. 23.2b), the electricalconnection to sacrificial anodes led to a significant lowering of the potentialonly on the lowest four corroding bars. The immersed bar reached a potentialvalue of about –1 V vs MMO, similar to the bar in the specimen withoutchlorides; bars at a height of 8 and 16 cm showed potential values between–900 and –800 mV vs MMO, lower than in the corresponding bars in thespecimen without chlorides.

After 16 months of testing, half of the bars were disconnected in order toincrease their spacing from 8 to 16 cm and reduce the steel surface to beprotected. The reduction in the protected area led to a small decrease in thesteel potential both in chloride-free concrete and in chloride-contaminatedconcrete.

Figures 23.3a and 23.3b show the variation in time of the current densitycirculating in each bar. For a given height, the current density was muchhigher for the active bars in chloride-contaminated concrete compared to thepassive bars in chloride-free concrete. For instance, at a height of 8–16 cm,a current density of about 200 mA m–2 circulated in the bar in chloride-contaminated concrete, while only about 50 mA m–2 circulated in the bar inchloride-free concrete. Indeed, the total current fed by the anode was about30–50 mA in the chloride-contaminated specimen and only 5–15 mA in thechloride-free concrete. The current which was received by bars at 8 and 16cm above the level of the solution was 70 and 85% of the total current,respectively, in chloride-free concrete and in chloride-contaminated concrete.Less than 10% of the total current reached the bars at heights above 40 cmin both columns. Negative values of the current (i.e. anodic currents) weremeasured on some of the bars in the higher part of the specimen with 3%chlorides, showing that macrocouples formed between corroding bars.

In order to establish if the steel bars were protected, depolarisation testswere regularly carried out on the two specimens by disconnecting the steelbars from the sacrificial anodes for 24 h. Four-hour and 24-h decays were



then calculated as the differences between potentials measured after 4 h and24 h, respectively, and the instant off potential value, measured within 1 safter the disconnection of the anodes. Figure 23.4a and 23.4b report the 4-hdecay as a function of time, in chloride-free concrete and chloride-contaminatedconcrete, respectively. In each column, the four-hour decay decreased withheight, showing very high values for the immersed bar (around 500 mV in

23.2 Potential of steel bars at different heights above the level of thesolution in the specimen with (a) chloride-free concrete and (b) 3%chloride-contaminated concrete, as a function of time.

15 bars connected 8 bars connected

Height:72–112

cm

40–64cm

16–32cm

8 cm

0 6 12 18 24 30 36Time (months)

(a)

Po

ten

tial

(m

V v

s M

MO

)

–100

–200

–300

–400

–500

–600

–700

–800

–900

–1000

–1100


Height:56–112

cm

40–48cm

32 cm

16 cm

0 6 12 18 24 30 36Time (months)

(b)

Po

ten

tial

(m

V v

s M

MO

)

–100

–200

–300

–400

–500

–600

–700

–800

–900

–1000

–1100

24 cm

8 cmImmersed

Immersed



chloride-free concrete and 300 mV in chloride-contaminated concrete).Depolarisation values were always higher for the rebars in chloride-freeconcrete if compared with the corresponding rebars in chloride-contaminatedconcrete.

A decay of 100 mV can be considered sufficient to provide both protectionto corroding steel as well as prevention of the onset of pitting corrosion onpassive steel (cathodic prevention).1,4,14 According to this criterion, the result

Cu

rren

t d

ensi

ty (

mA

m–2

)

1000

100

10

1

0.1


Height:16 cm

32 cm

48–112cm

0 6 12 18 24 30 36Time (months)

(a)

Cu

rren

t d

ensi

ty (

mA

m–2

)

1000

100

10

1

0.1


Height:16 cm

32 cm

48 cm

56–112cm

0 6 12 18 24 30 36Time (months)

(b)

23.3 Current density circulating in steel bars at different heightsabove the level of the solution in the specimen with (a) chloride-freeconcrete and (b) 3% chloride-contaminated concrete, as a function oftime. The black filled symbols (®) refer to the immersed bars.



of Fig. 23.4 would indicate that protection was fulfilled up to a height ofabout 60 cm in the specimen with chloride-free concrete and about 40 cm inthe specimen with chloride-contaminated concrete.

The high values of potential decay (Fig. 23.4) measured on the immersedbar were not consistent with its low current density (Fig. 23.3). For thisreason, the current circulating in each bar during the depolarisation tests was

23.4 Four-hour decay of steel bars at different heights above the levelof the solution in the specimen with (a) chloride-free concrete and (b)3% chloride-contaminated concrete, as a function of time.

Fou

r-h

ou

r d

ecay

(m

V)

700

600

500

400

300

200

100

0


Height:

Immersed

16–40 cm

48–56 cm

64–112 cm

0 6 12 18 24 30 36Time (months)

(a)

Height:

Immersed

8–16 cm

32–40 cm

48–112 cm


0 6 12 18 24 30 36Time (months)

(b)

Fou

r-h

ou

r d

ecay

(m

V)

700

600

500

400

300

200

100

0

8 cm



also monitored; current densities measured on the lowest four bars after thedisconnection from the anodes are reported in Fig. 23.5, where cathodiccurrents are considered positive and anodic currents negative. Thesemeasurements showed that in both specimens, as soon as the sacrificialanodes were disconnected, the immersed bar, which was previously receivinga cathodic current, generated an anodic current. Therefore, the bars in thenon-submerged part of the column continued to be cathodically polarisedowing to the electrical connection with the immersed bar, even afterdisconnection from the anodes. As a consequence, the results obtained fromdepolarisation tests carried out by simply disconnecting the anodes were notreliable for the evaluation of the height at which protection conditions can bereached, since they were affected by the presence of this macrocouple.

Further depolarisation tests were then carried out by avoiding the onset ofmacrocouples. Each single bar was disconnected from the others, so that nocurrent could circulate. The actual potential decay was then measured. The4-h decay of the immersed bar was negligible; steel potential was around–1 V vs SCE even after 24 h. On the contrary, an increase in the potentialafter 4 h and 24 h of depolarisation was observed in the rebars in the non-submerged part; hence their potential decay increased.

Figures 23.6a and 23.6b show the profile of 4-h decay along the height ofeach column, respectively, in chloride-free concrete and chloride-contaminatedconcrete. Comparison between results obtained from the tests with connectedand disconnected bars shows that the actual throwing power is higher thanthat evaluated in the presence of macrocouple currents.

Even though the immersed bar had a potential decay lower than 100 mV,its corrosion rate can be considered negligible in both cases owing to lack ofoxygen in permanently saturated conditions (the instant off potential is lowerthan –1 V vs SCE). The trend of 4-h decay along the height of the columnsuggests that 4-h decay in passive steel maintains values higher than 100 mVeven at heights greater than 1.2 m. On the contrary, sacrificial anodes did notprovide sufficient protection to corroding steel reinforcement placed above80 cm. The difference in the throwing power between the chloride-freespecimens and the chloride-contaminated ones were also confirmed by 24-hdepolarisation tests. In fact, profiles of 24-h decay shown in Fig. 23.7 do notdiffer much from those obtained after 4 h.

23.4 Conclusions

The effects of localised sacrificial anodes on passive and corroding rebars inconcrete piles partially immersed in seawater were studied. Depolarisationtests carried out by simply disconnecting the anode from the rebars were notreliable for evaluating the actual throwing power of cathodic protection andprevention, owing to a macrocouple that was generated between bars at



different heights. The immersed bar, which had a very negative potential,supplied an anodic current that cathodically polarised the rebars at higherheights and, thus, altered their potential decay.

When depolarisation tests were carried out by disconnecting each barfrom the others, so that the macrocouple could not generate, an effective4-h decay of about 150 mV was obtained at a height of more than 1 m if the

Immersed 16 cm32 cm 48 cm

Cu

rren

t d

ensi

ty (

mA

m–2

)

50

40

30

20

10

0

–10

–20

–30

–40

–50

Immersed 16 cm32 cm 48 cm

–4 0 4 8 12 16 20 24Time (h)

(a)

Cu

rren

t d

ensi

ty (

mA

m–2

)

200

150

100

50

0

–50

–100

–150

–200–4 0 4 8 12 16 20 24

Time (h)(b)

23.5 Current density on steel bars at different heights from the levelof the solution measured during a depolarisation test with connectedbars in the specimen with (a) chloride-free concrete and (b) 3%chloride-contaminated concrete, as a function of time.



steel was passive, whereas, if the steel was already corroding, the decay atthe same height was negligible.

The measurements of the actual decay of rebars suggest that, while sacrificialanodes have poor effectiveness in protecting corroding bars that are locatedabove sea level, they can provide cathodic polarisation to passive steel sufficientto avoid the onset of corrosion even at significant heights above sea level.

Hei

gh

t (c

m)

120

100

80

60

40

20

0

Connected barsDisconnected bars

0 100 200 300 400 500 600 700Four-hour decay (mV)

(a)

Hei

gh

t (c

m)

120

100

80

60

40

20

0


0 100 200 300 400 500 600 700Four-hour decay (mV)

(b)

23.6 Comparison between the profile of 4-h decay obtained withconnected or disconnected bars in (a) chloride-free concrete and(b) chloride-contaminated concrete.



23.5 References

1. P Pedeferri, ‘Cathodic protection and cathodic prevention’, Constr. Build. Mater.,1996, 10, 391.

2. G K Glass and N R Buenfeld, ‘Chloride threshold level for corrosion of steel inconcrete’, Corrosion Sci., 1997, 39, 1001.

3. C L Page, ‘Cathodic protection of reinforced concrete – principles and applications’,Proc. Int. Conf. Repair of Concrete Structures, in Theory to Practice in a MarineEnvironment, Svolvear, Norway, 28–30 May 1997, p. 123.

23.7 Comparison between the profile of 24-h decay obtained withconnected or disconnected bars in (a) chloride-free concrete and(b) chloride-contaminated concrete

Hei

gh

t (c

m)

120

100

80

60

40

20

0


0 100 200 300 400 500 600 700Twenty-four-hour decay (mV)

(a)

Hei

gh

t (c

m)

120

100

80

60

40

20

0


0 100 200 300 400 500 600 700Twenty-four-hour decay (mV)

(b)



4. EN 12696-1 Standard, ‘Cathodic protection of steel in atmospherically exposedconcrete’, March 2000.

5. S C Kranc and A A Sagues, ‘Computation of reinforcing steel corrosion distributionin concrete marine bridge substructures’, Corrosion, 1994, 50(1) 50.

6. O T de Rincón, M F de Romero, A R de Carruyo, M Sánchez and J Bravo, ‘Performanceof sacrificial anodes to protect the splash zone of concrete piles’, Mater. Struct.,1997, 30, 556.

7. S C Kranc, A A Sagues and F J Presuel-Moreno, ‘Computational and experimentalinvestigation of cathodic protection distribution in reinforced concrete marine piling’,Corrosion, 97, paper No. 231, NACE, Houston, 1997.

8. L Bertolini, F Bolzoni, T Pastore and P Pedeferri, ‘New experiences in cathodicprevention of reinforced concrete structures’, in Corrosion of Reinforcement in Concrete,C L Page et al. (Eds.), Society of Chemical Industry, London, 1996, p. 389.

9. L Bertolini, M Gastaldi, T Pastore, M P Pedeferri and P Pedeferri, ‘Cathodic protectionof steel in concrete and cathodic prevention’, European Community, COST 521Workshop, Annecy, 21–24 September 1999.

10. L Bertolini, ‘Cathodic prevention’, Proc. European Workshop Corrosion of Steel inReinforced Concrete Structures, T D Sloan and P A M Basheer (Eds.), QueensUniversity, Belfast, 28–31 August 2000, p. 107.

11. L Bertolini, F Bolzoni, A Cigada, T Pastore and P Pedeferri, ‘Cathodic protection ofnew and old reinforced concrete structures’, Corrosion Sci., 1993, 35(5–8), 1633.

12. T Pastore, P Pedeferri, L Bertolini and F Bolzoni, ‘Current distribution problems inthe cathodic protection of reinforced concrete structures’, Proc. Int. RILEM/CSIRO/ACRA Conf. on Rehabilitation of Concrete Structures, D W S Ho and F Collins(Eds.), Melbourne, 31 August–2 September 1992, p. 189.

13. L Bertolini, M Gastaldi, M P Pedeferri and E Redaelli, ‘Prevention of steel corrosionin concrete exposed to seawater with submerged sacrificial anodes’, Corrosion Sci.,2002, 27(6), 1497.

14. G K Glass, A M Hassanein and N R Buenfeld, ‘CP criteria for reinforced concretein marine exposure zones’, J. Mater. Civil Eng., 2000, 12(2), 164.


300

24.1 History

The ‘Stadionviaduct’ bridge in Rotterdam, which was built in 1937, connectsa residential area with an industrial area and crosses a large railway system.It is a main road towards the Feyenoord Football stadium so, in addition tonormal motor vehicle traffic and bicycles, it is used by large numbers ofpedestrians before and after football matches and other events.

The bridge consists of a steel support framework with three separateconcrete decks, one for the central driveway and two combined bicycle/pedestrian paths along the sides. The driveway has an asphalt overlay, whilethe concrete of the bicycle/pedestrian paths is finished with a thin wearingcourse of a few mm thickness, typical of light-traffic bridges in the Netherlands.In the middle of the 1980s, heavy corrosion and concrete damage due tochloride penetration from de-icing salts was found on the two bicycle paths.At that time, the only option available was conventional concrete repair. Thiswas carried out on the northern bicycle path in 1986. By the end of 1986, thesouthern path was provided with a cathodic protection system, a techniquewhich had then just been introduced in the Netherlands. The installation wasthe first CP system for protection of a reinforced concrete structure in theNetherlands.

24.2 The CP installation on the southern bicycle

path (1986)

The installation was designed to protect approximately 350 m2 of concrete intwo zones of about 2.4 m ¥ 70 m surface area. One zone was situated eastand one zone was situated west of the expansion joint in the middle of thebridge deck. Before applying CP, cracked and spalled concrete was removed,exposing the steel over about 50% of the surface. Additional reinforcementwas provided where necessary and a new concrete cover layer was cast toprovide a substrate for the anode system. The anode was a Ferex 100S cable,

24Renovation of the cathodic protection

system of a concrete bridge after12 years of operation

G. S C H U T E N and J. L E G G E D O O R, LeggedoorConcrete Repair, The Netherlands, and R. B. P O L D E R

and W. H. A. P E E L E N, TNO Building andConstruction Research, The Netherlands


Renovation of the cathodic protection system 301

consisting of a twisted copper wire with niobium coating, surrounded by agraphite-filled polymer mantle forming a cable of 8 mm diameter. The cableswere laid on the deck in loops and fixed with plastic fixings as shown in Fig.24.1. A total length of 2 ¥ 1700 running metres of anode cable was installedin this way. The anode cables were embedded in a cast layer of fine aggregateconcrete made with fly ash cement. Two reference electrodes of the Ag/AgCltype were located near the two ends of each zone. The power supplies weredesigned to deliver 15 V dc at a total current capacity of 4 A for each zone.The system was finalised in December 1986 and energised in January 1987.An epoxy coating wearing course was applied later that year.

24.3 Replacement of the northern bicycle

path (1996)

The 1986 repair of the northern bicycle path was carried out by the standardsof that time. This included removing cracked and spalled parts, cleaning thesteel by grit blasting and applying new concrete cover using shotcrete. Thesurface was finished with an epoxy wearing course. However, the repair wasfound to be ineffective in stopping corrosion in the long term. In 1996,advancing corrosion and concrete damage in the form of cracking and spallinghad seriously compromised the safe operation of the northern bicycle path.It was decided to remove the complete deck and the supporting steelworkand to replace it with a new steel deck and supports.

24.4 CP system behaviour in 1998

The cathodically protected southern bicycle path showed no signs of corrosionand no damage to the concrete. However the operation of the CP systemraised some concerns. Since 1992, some of the reference electrodes were notindicating normal depolarisation behaviour. The current delivered to one of

24.1 Ferex 100S anode system before casting the concrete overlay(1986).



the zones had become increasingly erratic over the years. TNO was asked toasses the technical condition of the installation, to study the feasibility ofrenovating the CP system and, if possible, to identify the measures needed todo so.

Electrical measurements were carried out and cores were taken for visualand microscopic analysis. This assessment showed that the problems werelimited to improper functioning of the power supplies, poor electricalconnections and poor functioning of the reference electrodes [1]. Mostimportantly, however, the anode system showed no visible degradation. Amicrograph of the anode/concrete interface is shown in Fig. 24.2. Nodeterioration of the carbon/polymer anode surface or carbonation of theinterface with the concrete, which might indicate oxidation of the graphite inthe anode, was detected using ultraviolet and visible light. The conductingpolymer appeared to be intact. The overlay showed some microcracking,probably due to drying shrinkage in the early stages; one such microcrackcan be seen in Fig. 24.2. There was no evidence of deterioration of theoverlay connected to the circulation of the CP current. It was concluded that,with minor upgrading, the system could be made to operate well for at leastanother ten years.

24.5 System upgrade 1999

Based on the assessment, a programme was outlined to renovate the CPsystem by providing new power supplies, new reference electrodes and multiplenew connections to both the anode cables and the reinforcement. The contract

24.2 Micrograph of the Ferex 100S anode/concrete interface (KSA isthe anode, M cement paste, A aggregate and Sc is a microcrack);field of view c. 2 mm ¥ 3 mm.



to carry out the work was commissioned to Leggedoor bv and TNO wasasked to supervise it.

Four slots of approximately 2.0 m ¥ 0.2 m were excavated in the concretesurrounding the anode cables at the ends of the bicycle paths for making thenew connections. A schematic is given in Fig. 24.3 and a picture of the workbeing carried out is given in Fig. 24.4. The Ferex polymer mantle was removedlocally from all 16 cables in the slots and connections were soldered to thecopper wires, which were then insulated using heat shrinking plastic sleeves.

WestExpansion joint East

Driveway

Bicycle

Pedestrian

CP Zone 1 CP Zone 2

24.3 Layout of deck, CP system and locations of the four slots formaking new connections.

24.4 Making new electrical connections to the Ferex 100S anode andreinforcement.



In the slots, additional reinforcement connections were made. One manganesedioxide reference electrode was installed in each of the slots, which werethen backfilled using a polymer-modified cementitious repair mortar. Modernpower supplies and sockets for testing were installed in the control cabinet.Existing cables in the concrete were reused for connection of the systemwith the new cabinet.

Subsequently, the system was reactivated in December 1999. Beforeswitching on the current, cable resistances in the anode and reinforcementcircuits were checked and found to be satisfactory (<1 W). A voltage of about2 V was applied, resulting in a current of about 0.5 A per zone. Short termpolarisation was between 125 and 300 mV in the negative direction. Thecurrent decreased in the following months to lower values of about 0.1 A.After three months, depolarisation values in 24 h were well over 100 mV [2].It was concluded that the installation was complying to the requirements ofthe Dutch recommendation CUR 45 ‘Cathodic Protection of Reinforcementin Concrete Structures’ [3].

24.6 Cost aspects

Financial records of repair and maintenance actions in the past were notavailable. Consequently, the costs of repair and maintenance can only beestimated. The cost of installing the CP system on the southern bicycle pathin 1986/87 is estimated at 150 000 7. The cost for routine control of the CPsystem between 1986 and 1998 is estimated at 15 000 7. The costs involvedin the assessment and renovation of the CP system from 1998 to 2000 wereabout 55 000 7. Control of the system between 2000 and 2010 may cost25 000 7. The total cost from 1987 to 2010 is about 250 000 7. The totalcost of replacing the deck of the northern bicycle path was about 1.3 million7. The cost of the repairs in 1986 is unknown.

It clear that the total costs for replacing the deck exceed by far the costsfor the original CP system, its maintenance and control, including the renovationof the system and ten more years of operation; according to the estimates, thedifference is by about a factor of five.

24.7 Durability aspects

In the present installation, it was found that the anode system was durableand was able to provide sufficient corrosion protection to the reinforcementfor about 12 years, despite malfunctioning of the electrical system aftersome time. Moreover, the anode was expected to be able to provide protectionfor at least another ten years. The durability of the anode system reportedhere is in contrast with a CP system in Berlin based on the same anode,described by Mietz and Isecke [4]. The latter system functioned properly



until 6 to 8 years of service. Detailed examination after 15 years showed thatthe carbon had dissolved from the outer layers of the anode cable and thepolymer had become brittle. This caused high resistance build-up in thecircuit and decreasing current density, until sufficient depolarisation couldno longer be achieved. There is no obvious explanation for the differencebetween the two systems. Most probably the current density in the Berlinsystem was significantly higher compared with that in the Rotterdam‘Stadionviaduct’ system.

An observation is that the epoxy wearing course applied to the bicycledecks (in 1986/87) on its own does not appear to prevent corrosion of thereinforcement. This may be due to chloride left in the concrete when repairswere carried out. Possibly incipient anode effects have played a role. However,it may be questioned whether such an epoxy wearing course could effectivelyprevent the ingress of water and new chlorides. In a study of a concrete deckwith a similar wearing course, it was concluded that degradation of thewearing course allowed chloride penetration after 10 years of service [5].

24.8 Conclusions

The CP system of the southern bicycle lane of the ‘Stadionviaduct’ bridge inRotterdam based on the Ferex 100S Anode cable was upgraded successfullyafter 12 years of service to meet contemporary CP requirements at the relativelylow cost of 55 000 7. This was mainly due to the good condition of theanode. Despite problematic (electrical) operation of the system during thesecond half of the 12-year period, reinforcement corrosion had not takenplace to the extent that damage to the concrete occurred.

Although repaired conventionally in 1986, the northern bicycle path hadto be replaced in 1996 at the considerable cost of 1.3 million 7. Due to theCP system installed in 1986 and the upgrade of this system in 1999,the southern bicycle path has an expected service life after repair of at least25 years (starting 1987) and for a much lower cost, which was estimated at250 000 7.

The comparison between the northern and southern bicycle path shows:

∑ the cost effectiveness of CP.∑ the much better durability of CP than conventional repair.

24.9 Acknowledgement

We would like to thank Mrs Carolien Nieuwland of IngenieursbureauGemeentewerken (public works) Rotterdam for her permission to publishthis information.



24.10 References

1. R. B. Polder, ‘Cathodic protection Stadionviaduct, Phase 1’, TNO Building andConstruction Research Report 99-BT-MK-R0026, 1999 (in Dutch).

2. R. B. Polder and W.H.A. Peelen, ‘Cathodic protection Stadionviaduct, Phase 2, Revisionof existing installation’, TNO Building and Construction Research Report 2000-BT-MK-R0079/02, 2000 (in Dutch).

3. CUR 45, ‘Technical Recommendation for cathodic protection of reinforced concrete’,Kathodische bescherming van wapening in betonconstructies, CUR Aanbeveling 45,1996.

4. J. Mietz, J. Fischer and B. Isecke, ‘Cathodic protection of steel-reinforced concretestructures – results from 15 years’ experience’, Mater. Perf., December 2001, 22–26.

5. R.B. Polder and A. Hug, 2000, ‘Penetration of chloride from de-icing salt into concretefrom a 30 year old bridge’, HERON, 45, (2), 109–124.


European Federation of Corrosion Publications NUMBER 38...

Documents

Transcript of European Federation of Corrosion Publications NUMBER 38...