European Federation of Corrosion Publications NUMBER 38
Corrosion of reinforcement in concrete Mechanisms, monitoring, inhibitors and rehabilitation techniques Edited by M. Raupach, B. Elsener, R. Polder and J. Mietz
Published for the European Federation of Corrosion by Woodhead Publishing and Maney Publishing on behalf of The Institute of Materials, Minerals & Mining
CRC Press Boca Raton Boston New York Washington, DC
WOODHEAD
PUBLISHING LIMITED
Cambridge England i © 2007, Institute of Materials, Minerals and Mining
Woodhead Publishing Limited and Maney Publishing Limited on behalf of The Institute of Materials, Minerals & Mining Published by Woodhead Publishing Limited, Abington Hall, Abington Cambridge CB21 6AH, England www.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 Mining The 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. Reasonable efforts have been made to publish reliable data and information, but the authors and the publishers cannot assume responsibility for the validity of all materials. Neither the authors nor the publishers, nor anyone else associated with this publication, shall be liable for any loss, damage or liability directly or indirectly caused or alleged to be caused by this book. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming and recording, or by any information storage or retrieval system, without permission in writing from the Woodhead Publishing Limited. The consent of Woodhead Publishing Limited does not extend to copying for general distribution, for promotion, for creating new works, or for resale. Specific permission must be obtained in writing from Woodhead Publishing Limited for such copying. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. Library of Congress Cataloging in Publication Data A 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: 978-1-4200-4401-0 CRC Press ISBN-10: 1-4200-4401-X CRC Press order number: WP4401 The publishers’ policy is to use permanent paper from mills that operate a sustainable forestry policy, and which has been manufactured from pulp which is processed using acid-free and elementary chlorine-free practices. Furthermore, the publishers ensure that the text paper and cover board used have met acceptable environmental accreditation standards. Typeset by Replika Press Pvt Ltd, India Printed by T J International Limited, Padstow, Cornwall, England
ii © 2007, Institute of Materials, Minerals and Mining
Contents
Contributor contact details Series introduction
xvii
Volumes in the EFC series
xix
Preface 1
xi
Corrosion of metals in contact with mineral building materials
xxiii 1
U. NÜRNBERGER, University of Stuttgart, Germany
1.1 1.2 1.3 1.4 1.5 1.6 2
Corrosion behaviour during contact with building materials that contain cement Reaction in case of contact with aqueous cement solutions and alkaline waters Corrosion performance in chloride containing alkaline building materials Corrosion behaviour during contact with building materials containing magnesia cement Corrosion behaviour during contact with gypsum products References Corrosion and electrochemistry of zinc in alkaline solutions and in cement mortar
1 6 7 8 8 9 10
K. VIDEM, University of Oslo, Norway
2.1 2.2 2.3 2.4 2.5 2.6
Introduction Experimental methods Results Discussion Conclusions References
10 11 12 17 24 25 iii
© 2007, Institute of Materials, Minerals and Mining
iv
Contents
3
Corrosion behaviour of galvanized steel rebars in the presence of coating discontinuities
27
T. BELLEZZE, R. FRATESI, and F. TITTARELLI, Università Politecnica delle Marche, Italy
3.1 3.2 3.3 3.4 3.5
Introduction Experimental methods Results and discussion Conclusions References
27 28 29 36 36
4
Influence of scale and rust on steel activation in model concrete pore solution
38
P. NOVÁK, R. MALÁ and M. KOUˇRIL, Institute of Chemical Technology, Prague, Czech Republic
4.1 4.2 4.3 4.4 4.5 4.6
Introduction Experimental methods Results Conclusions Acknowledgements References
38 38 39 43 43 43
5
The surface of iron and Fe10Cr alloys in alkaline media
44
A. ROSSI, G. PUDDU and B. ELSENER, University of Cagliari, Italy
5.1 5.2 5.3 5.4 5.5 5.6 5.7
Introduction Experimental methods Results Discussion Conclusions Acknowledgements References
44 45 46 54 59 60 60
6
Risk of galvanic corrosion induced by CFRP strengthening in reinforced concrete
62
L. BERTOLINI, M. GASTALDI and M. P. PEDEFERRI, Politecnico di Milano, Italy
6.1 6.2 6.3 6.4 6.5 6.6
Introduction Experimental procedure Results and discussion Conclusions Acknowledgements References
© 2007, Institute of Materials, Minerals and Mining
62 63 65 72 73 74
Contents
7
Macrocell corrosion of steel in concrete – experiments and numerical modelling
v
75
S. JÄGGI and H. BÖHNI, ETH Zürich, Switzerland and B. ELSENER, ETH Zürich and University of Cagliari, Italy
7.1 7.2 7.3 7.4 7.5 7.6
Introduction Experimental methods Experimental and modelling results Discussion Conclusions References
75 77 78 83 87 87
8
Modelling of chloride-induced corrosion of reinforcement in cracked high-performance concrete based on laboratory investigations
89
M. RAUPACH and C. DAUBERSCHMIDT, Aachen University, Germany
8.1 8.2 8.3 8.4 8.5 8.6
Background Test programme Results Numerical simulation Conclusions References
89 90 91 100 103 104
9
Influence of stray currents on corrosion of steel in concrete
105
L. BERTOLINI, M. CARSANA and P. PEDEFERRI, Politecnico di Milano, Italy
9.1 9.2 9.3 9.4 9.5
Introduction Experimental tests Results and discussion Conclusions References
105 107 109 117 119
10
Assessment and monitoring of corrosion velocity of rebars and prestressing cables of a bridge
120
D. BINDSCHEDLER, Swiss Society for Corrosion Protection, Switzerland
10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8
Introduction Results of detailed corrosion inspection Repair Monitoring system Results of the monitoring Corrosion velocity and further service of the bridge Conclusions References
© 2007, Institute of Materials, Minerals and Mining
120 120 122 122 124 130 131 132
vi
Contents
11
On-line monitoring of corrosion in reinforced concrete structures
133
Y. SCHIEGG, L. AUDERGON and H. BÖHNI, ETH Zürich, Switzerland and B. ELSENER, ETH Zürich and University of Cagliari, Italy
11.1 11.2 11.3 11.4 11.5 11.6 11.7 11.8
Introduction Instrumentation Field tests Modelling of the temperature dependence of RW and Icorr Results and discussion Conclusions Acknowledgements References
133 133 135 135 138 145 145 145
12
Integrated system for corrosion monitoring of reinforced concrete structures
146
U. SCHNECK, T. WINKLER and S. MUCKE, Concrete Improvement Technologies, Germany
12.1 12.2 12.3 12.4 12.5 12.6
The task The solution The displays Monitoring in use – the results Acknowledgements References
146 146 153 154 157 157
13
Use of portable equipment to determine the corrosion state of concrete structures
159
R. BÄßLER and A. BURKERT, Federal Institute for Materials Research and Testing, Germany and T. FRØLUND and O. KLINGHOFFER, Force Technology, Denmark
13.1 13.2 13.3 13.4 13.5 13.6 13.7 13.8 13.9
Introduction Background Experimental setup Results Discussion Conclusions Outlook Acknowledgements References
© 2007, Institute of Materials, Minerals and Mining
159 159 162 162 167 168 168 169 169
14
Contents
vii
Corrosion inhibitors for reinforced concrete – an EFC state of the art report
170
B. ELSENER, University of Cagliari, Italy and ETH Zürich, Switzerland
14.1 14.2 14.3 14.4
170 172 173
14.5 14.6 14.7 14.8 14.9
Introduction Mode of action of corrosion inhibitors Corrosion inhibitors to prevent or delay corrosion initiation Corrosion inhibitors to reduce the propagation rate of corrosion Field tests with corrosion inhibitors Transport of the inhibitor into mortar or concrete Critical evaluation of corrosion inhibitors Conclusions References
15
Mixed-in inhibitors for concrete structures
185
176 178 179 181 182 182
F. BOLZONI, G. FUMAGALLI, L. LAZZARI, M. ORMELLESE and M. P. PEDEFERRI, Politecnico di Milano, Italy
15.1 15.2 15.3 15.4 15.5 15.6 15.7
Introduction Service life Experimental methods Results Discussion Conclusions References
185 187 188 190 196 200 201
16
Effectiveness of mixed-in organic corrosion inhibitors on extending the service life of reinforced concrete structures
203
R. CIGNA, Consultant, Italy, A. MERCALLI, Autostrade S.p.A, Italy, L. GRISONI, Sika Italia, Italy and U. MÄDER, Sika A. G., Switzerland
16.1 16.2 16.3 16.4
Introduction Experimental methods Discussion and conclusions References
203 204 204 210
17
Migrating inhibitors on corrosion in reinforced concrete 211 F. BOLZONI, G. FUMAGALLI, L. LAZZARI, M. ORMELLESE and M. P. PEDEFERRI, Politecnico di Milano, Italy
17.1 17.2 17.3 17.4 17.5
Introduction Experimental methods Results and discussion Conclusions References
© 2007, Institute of Materials, Minerals and Mining
211 212 214 222 223
viii
18
Contents
Effectiveness of corrosion inhibitors – a field study
226
Y. SCHIEGG, F. HUNKELER and H. UNGRICHT, Swiss Society for Corrosion Protection (SGK) and Technical Research and Consulting on Cement and Concrete (TFB), Switzerland
18.1 18.2 18.3 18.4 18.5 18.6 18.7
Introduction Field study in Naxbergtunnel Investigation Results Conclusions Acknowledgements References
226 226 227 229 237 238 238
19
Corrosion protection of steel rebar in concrete using migrating corrosion inhibitors
239
B. BAVARIAN and L. REINER, California State University, USA
19.1 19.2 19.3 19.4 19.5
Introduction Experimental procedures Results and discussion Conclusions References
239 241 242 244 248
20
Determination of coating permeability on concrete using EIS
250
J. VOGELSANG, G. MEYER and M. BEPOIX, Sika GmbH, Germany
20.1 20.2 20.3 20.4 20.5
Introduction Experimental design Results and discussion Conclusions References
250 252 256 260 261
21
Chloride extraction from reinforced concrete – a new defined way of application
263
U. SCHNECK, T. WINKLER and H. GRÜNzIG, Concrete Improvement Technologies, Germany
21.1 21.2 21.3 21.4 21.5 21.6 21.7 21.8
The task The solution Description of configuration Application to a highway bridge abutment Results of the follow-up survey Conclusions Acknowledgements References and further reading
© 2007, Institute of Materials, Minerals and Mining
263 263 267 268 272 273 275 275
22
Contents
ix
Microscopy study of the interface between concrete and the conductive coating used as an anode for cathodic protection
277
R. B. POLDER and W. H. A. PEELEN, TNO Building and Construction Research, The Netherlands and J. LEGGEDOOR and G. SCHUTEN, Leggedoor Concrete Repair, The Netherlands
22.1 22.2 22.3 22.4 22.5 22.6 22.7 22.8
Introduction Theoretical background Samples and microscopy examination Results Discussion Conclusions Acknowledgements References
277 277 279 280 285 286 287 287
23
Protection of reinforced concrete piles in marine structures with sacrificial anodes
288
L. BERTOLINI, M. GASTALDI, M. PEDEFERRI and E. REDAELLI, Politecnico di Milano, Italy
23.1 23.2 23.3 23.4 23.5
Introduction Experimental procedure Results and discussion Conclusions References
288 289 290 295 298
24
Renovation of the cathodic protection system of a concrete bridge after 12 years of operation
300
G. SCHUTEN and J. LEGGEDOOR, Leggedoor Concrete Repair, The Netherlands and R. B. POLDER and W. H. A. PEELEN, TNO Building and Construction Research, The Netherlands
24.1 24.2 24.3 24.4 24.5 24.6 24.7 24.8 24.9 24.10
History The CP installation of the southern bicycle path (1986) Replacement of the northern bicycle path (1996) CP system behaviour in 1998 System upgrade 1999 Cost aspects Durability aspects Conclusions Acknowledgement References
© 2007, Institute of Materials, Minerals and Mining
300 300 301 301 302 304 304 305 305 306
Contributor contact details
(* = main contact)
Editors Prof Dr-Ing M. Raupach Aachen University Institute of Building Materials Research Schinkelstr. 3 52056 Aachen Germany E-mail:
[email protected]
Professor Dr B. Elsener ETH Zurich Institute for Building Materials ETH Hönggerberg CH-8093 Zürich Switzerland E-mail:
[email protected]
Dr R. Polder TNO Dr Built Environment and Geosciences PO Box 49 2600 AA Delft The Netherlands E-mail:
[email protected]
Dr-Ing J. Mietz Federal Institute for Materials Research and Testing (BAM) Unter den Eichen 87 12205 Berlin Germany E-mail:
[email protected]
Chapter 1 Professor U. Nürnberger Materialprüfungsanstalt Universität Stuttgart Pfaffenwaldring 4 70569 Stuttgart Germany E-mail:
[email protected]
Chapter 2 Dr K. Videm Centre for Materials Science University of Oslo Gaustadalleen 21 N-0349 Oslo Norway E-mail:
[email protected] xi
© 2007, Institute of Materials, Minerals and Mining
xii
Contributor contact details
Chapter 3
Chapter 6
Dr. T. Bellezze*, Dr R. Fratesi and Dr F. Tittarelli Dipartimento di Fisica e Ingegneria dei Materiali e del Territorio Università Politecnica delle Marche Via Brecce Bianche 60131, Ancona Italy
Dr L. Bertolini*, Dr M. Gastaldi and Dr M. P. Pedeferri Dipartimento di Chimica, Materiali Ingegneria Chimica ‘G. Natta’ Politecnico di Milano Via Mancinelli 7 20131 Milano Italy
E-mail:
[email protected]
E-mail:
[email protected]
Chapter 4
Chapter 7
Professor P. Novak*, Dr R. Malá and Dr M. Kouřil Department of Metals and Corrosion Engineering Institute of Chemical Technology Prague Technická 5 CZ-166 28 Prague 6 Czech Republic
Professor B. Elsener*, Dr S. Jäggi and Professor H. Böhni ETH Zürich Institute for Building Materials ETH Hönggerberg CH-8093 Zürich Switzerland E-mail:
[email protected]
E-mail:
[email protected]
Chapter 8 Chapter 5 Professor A. Rossi*, Dr G. Puddu and Professor B. Elsener Department of Inorganic and Analytical Chemistry University of Cagliari I – 09042 Monserrato Cagliari Sardinia Italy E-mail:
[email protected]
© 2007, Institute of Materials, Minerals and Mining
Prof Dr-Ing M. Raupach* and Dr C. Dauberschmidt Aachen University Institute of Building Materials Research Schinkelstr. 3, 52056 Aachen Germany E-mail:
[email protected]
Contributor contact details
Chapter 9
Chapter 12
Dr L. Bertolini*, Dr M. Carsana and Dr P. Pedeferri Dipartimento di Chimica, Materiali Ingegneria Chimica ‘G. Natta’ Politecnico di Milano Via Mancinelli 7 20131 Milano Italy
Dr U. Schneck*, Dipl-Ing T. Winkler and Dipl-Ing (FH) S. Mucke CITec Concrete Improvement Technologies GmbH Dresdner Strasse 42 D-01462 Cossebaude Germany
E-mail:
[email protected]
E-mail:
[email protected]
Chapter 10 Dr D. Bindschedler Swiss Society for Corrosion Protection Technoparkstrasse 1 CH-8005 Zürich Switzerland E-mail:
[email protected]
Chapter 13 Dr Ralph Bäßler* and Dr Andreas Burkert Federal Institute for Materials Research and Testing (BAM) Unter den Eichen 87 D-12205 Berlin Germany
Chapter 11
E-mail:
[email protected]
Dr Y. Schiegg*, Dr L. Audergon, Dr B. Elsener and Professor H. Böhni Schweizerische Gesellschaft für Korrosionsschutz Technoparkstrasse 1 CH-8005 Zürich Switzerland
Thomas Frølund and Oskar Klinghoffer Force Technology DK-2605 Brøndby Denmark
E-mail:
[email protected]
Professor B. Elsener ETH Zürich Institute for Building Materials ETH Hönggerberg CH-8093 Zürich Switzerland
Chapter 14
E-mail:
[email protected]
© 2007, Institute of Materials, Minerals and Mining
xiii
xiv
Contributor contact details
Chapter 15
Chapter 18
Dr F. Bolzoni*, Dr G. Fumagalli, Dr L. Lazzari, Dr M. Ormellese and Dr M. Pedeferri Dipartimento di Chimica, Materiali e Ingegneria Chimica ‘G. Natta’ Politecnico di Milano Via Mancinelli 7 20131 Milano Italy
Dr Y. Schiegg*, Dr F. Hunkeler and Dr H. Ungricht Schweizerische Gesellschaft für Korrosionsschutz Technoparkstrasse 1 CH-8005 Zürich Switzerland
E-mail:
[email protected]
Chapter 19
Chapter 16
Professor B. Bavarian* and Dr L. Reiner Department of MSEM California State University, Northridge 18111 Nordhoff Street Northridge CA 91330-8332 USA
Dr R. Cigna*, Dr A. Mercalli, Dr L. Grisomi and Dr U. Mäder Haward Technology Smith Moore LLP Suite 750 1355 Peachtree Street Atlanta GA 30309-3214 USA E-mail:
[email protected];
[email protected]
Chapter 17 Dr F. Bolzoni*, Dr G. Fumagalli, Dr L. Lazzari, Dr M. Ormellese and Dr M. Pedeferri Dipartimento di Chimica, Materiali e Ingegneria Chimica ‘G. Natta’ Politecnico di Milano Via Mancinelli 7 20131 Milano Italy E-mail:
[email protected]
© 2007, Institute of Materials, Minerals and Mining
E-mail:
[email protected]
E-mail:
[email protected]
Chapter 20 Dr J. Vogelsang*, Dr G. Meyer and Dr M. Bepoix Sika Technology AG Tueffenwies 16 CH 8048 Zürich Switzerland E-mail:
[email protected]
Contributor contact details
Chapter 21
Chapter 24
Dr U. Schneck*, Dipl-Ing T. Winkler and Dipl-Chem H. Grünzig CITec Concrete Improvement Technologies GmbH Dresdner Strasse 42 D-01462 Cossebaude Germany
Dr G. Schuten*, Dr J. Leggedoor, Dr R. Polder and Dr W. Peelen Leggedoor Tuinstraat 58 PO Box 3 9514 ZG Gasselternijveen The Netherlands
E-mail:
[email protected]
Chapter 22 Dr R. Polder*, Dr W. Peelen, Dr J. Leggedoor and Dr G. Schuten TNO Built Environment and Geosciences PO Box 49 2600 AA Delft The Netherlands E-mail:
[email protected]
Chapter 23 Dr L. Bertolini*, Dr M. Gastaldi, Dr M. Pedeferri and Dr E. Redaelli Dipartimento di Chimica, Materiali Ingegneria Chimica ‘G. Natta’ Via Mancinelli 7 20131 Milano Italy E-mail:
[email protected]
© 2007, Institute of Materials, Minerals and Mining
E-mail:
[email protected]
xv
European Federation of Corrosion Publications: Series introduction
The EFC, incorporated in Belgium, was founded in 1955 with the purpose of promoting European co-operation in the fields of research into corrosion and corrosion prevention. Membership is based upon participation by corrosion societies and committees in technical Working Parties. Member societies appoint delegates to Working Parties, whose membership is expanded by personal corresponding membership. The activities of the Working Parties cover corrosion topics associated with inhibition, education, reinforcement in concrete, microbial effects, hot gases and combustion products, environment sensitive fracture, marine environments, surface science, physico–chemical methods of measurement, the nuclear industry, computer based information systems, the oil and gas industry, the petrochemical industry, coatings, automotive engineering and cathodic protection. 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 instructional material, including films. The activities of the Working Parties are co-ordinated, through a Science and Technology Advisory Committee, by the Scientific Secretary. The administration of the EFC is handled by three Secretariats: DECHEMA e. V. in Germany, the Société de Chimie Industrielle in France, and The Institute of Materials, Minerals and Mining in the United Kingdom. These three Secretariats meet at the Board of Administrators of the EFC. There is an annual General Assembly at which delegates from all member societies meet to determine and approve EFC policy. News of EFC activities, forthcoming conferences, courses etc. is published in a range of accredited corrosion and certain other journals throughout Europe. More detailed descriptions of activities are given in a Newsletter prepared by the Scientific Secretary. The output of the EFC takes various forms. Papers on particular topics, for example, reviews or results of experimental work, may be published in scientific and technical journals in one or more countries in Europe. Conference xvii © 2007, Institute of Materials, Minerals and Mining
xviii
EFC series introduction
proceedings are often published by the organisation responsible for the conference. In 1987 the, then, Institute of Metals was appointed as the official EFC publisher. Although the arrangement is non-exclusive and other routes for publication are still available, it is expected that the Working Parties of the EFC will use The Institute of Materials, Minerals and Mining for publication of reports, proceedings etc. wherever possible. The name of The Institute of Metals was changed to The Institute of Materials on 1 January 1992 and to the Institute of Materials, Minerals and Mining with effect from 26 June 2002. The series is now published by Woodhead Publishing and Maney Publishing on behalf of the Institute of Materials, Minerals and Mining. P. McIntyre EFC Series Editor, The Institute of Materials, Minerals and Mining, London, UK EFC Secretariats are located at: Dr B. A. Rickinson European Federation of Corrosion, The Institute of Materials, Minerals and Mining, 1 Carlton House Terrace, London, SW1Y 5DB, UK Dr J. P. Berge Fédération Europénne de la Corrosion, Société de Chimie Industrielle, 28 rue Saint Dominique, F-75007 Paris, FRANCE Professor Dr G. Kreysa Europäische Föderation Korrosion, DECHEMA e. V., Theodor-Heuss-Allee 25, D-60486, Frankfurt, GERMANY
© 2007, Institute of Materials, Minerals and Mining
Volumes in the EFC series
1 Corrosion in the nuclear industry Prepared by the Working Party on Nuclear Corrosion 2 Practical corrosion principles Prepared by the Working Party on Corrosion Education (Out of print) 3 General guidelines for corrosion testing of materials for marine applications Prepared by the Working Party on Marine Corrosion 4 Guidelines on electrochemical corrosion measurements Prepared by the Working Party on Physico-Chemical Methods of Corrosion Testing 5 Illustrated case histories of marine corrosion Prepared by the Working Party on Marine Corrosion 6 Corrosion education manual Prepared by the Working Party on Corrosion Education 7 Corrosion problems related to nuclear waste disposal Prepared by the Working Party on Nuclear Corrosion 8 Microbial corrosion Prepared by the Working Party on Microbial Corrosion 9 Microbiological degradation of materials – and methods of protection Prepared by the Working Party on Microbial Corrosion 10 Marine corrosion of stainless steels: chlorination and microbial effects Prepared by the Working Party on Marine Corrosion 11 Corrosion inhibitors Prepared by the Working Party on Inhibitors (Out of print) xix © 2007, Institute of Materials, Minerals and Mining
xx
Volumes in the EFC series
12 Modifications of passive films Prepared by the Working Party on Surface Science and Mechanisms of Corrosion and Protection 13 Predicting CO2 corrosion in the oil and gas industry Prepared by the Working Party on Corrosion in Oil and Gas Production (Out of print) 14 Guidelines for methods of testing and research in high temperature corrosion Prepared by the Working Party on Corrosion by Hot Gases and Combustion 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 alloy steels for H2S-containing environments in oil and gas production Prepared by the Working Party on Corrosion in Oil and Gas Production 18 Stainless steel in concrete: state of the art report Prepared by the Working Party on Corrosion of Reinforcement in Concrete 19 Sea water corrosion of stainless steels – mechanisms and experiences Prepared by the Working Parties on Marine Corrosion and Microbial Corrosion 20 Organic and inorganic coatings for corrosion prevention – research and experiences Papers from EUROCORR ‘96 21 Corrosion-deformation interactions CDI ‘96 in conjunction with EUROCORR ‘96 22 Aspects of microbially induced corrosion Papers from EUROCORR ‘96 and the EFC Working Party on Microbial Corrosion 23 CO2 Corrosion control in oil and gas production – design considerations Prepared by the Working Party on Corrosion in Oil and Gas Production 24 Electrochemical rehabilitation methods for reinforced concrete structures – a state of the art report Prepared by the Working Party on Corrosion of Reinforcement in Concrete
© 2007, Institute of Materials, Minerals and Mining
Volumes in the EFC series
xxi
25 Corrosion of reinforcement in concrete – monitoring, prevention and rehabilitation Papers from EUROCORR ‘97 26 Advances in corrosion control and materials in oil and gas production Papers from EUROCORR ‘97 and EUROCORR ‘98 27 Cyclic oxidation of high temperature materials Proceedings of an EFC Workshop, Frankfurt/Main, 1999 28 Electrochemical approach to selected corrosion and corrosion control studies Papers from 50th ISE Meeting, Pavia, 1999 29 Microbial corrosion (Proceedings of the 4th international EFC workshop) 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 measures and monitoring Prepared by F. P. Ijsseling and the Working Party on Marine Corrosion 31 Corrosion of reinforcement in concrete: corrosion mechanisms and corrosion protection Papers from EUROCORR ‘99 and the Working Party on Corrosion of Reinforcement in Concrete 32 Guidelines for the compilation of corrosion cost data and for the calculation of the life cycle cost of corrosion – a working party report Prepared by the Working Party on Corrosion in Oil and Gas Production 33 Marine corrosion of stainless steels: testing, selection, experience, protection and monitoring Edited by D. Féron 34 Lifetime modelling of high temperature corrosion processes Proceedings of an EFC Workshop 2001. Edited by M. Schütze, W. J. Quadakkers and J. R. Nicholls 35 Corrosion inhibitors for steel in concrete Prepared by B. Elsener with support from a Task Group of Working Party 11 on Corrosion of Reinforcement in Concrete 36 Prediction of long term corrosion behaviour in nuclear waste systems Edited by D. Féron of Working Party 4 on Nuclear Corrosion
© 2007, Institute of Materials, Minerals and Mining
xxii
Volumes in the EFC series
37 Test methods for assessing the susceptibility of prestressing steels to hydrogen induced stress corrosion cracking by B. Isecke of EFC WP11 on Corrosion of Reinforcement in Concrete 38 Corrosion of reinforcement in concrete: mechanisms, monitoring, inhibitors and rehabilitation techniques Edited by M. Raupach, B. Elsener, R. Polder and J. Mietz 39 The use of corrosion inhibitors in oil and gas production Edited by J. W. Palmer, W. Hedges and J. L. Dawson 40 Control of corrosion in cooling waters Edited by J. D. Harston and F. Ropital 41 Corrosion by carbon and nitrogen: metal dusting, carburisation and nitridation M. Schutze and H. Grabke 42 Corrosion in refineries J. Harston 43 The electrochemistry and characteristics of embeddable reference electrodes for concrete Prepared by R. Myrdal on behalf of Working Party 11 on Corrosion of Steel in Concrete 44 The use of electrochemical scanning tunnelling microscopy (EC– STM) in corrosion analysis: reference material and procedural guidelines Prepared by R. Lindström, V. Maurice, L. H. Klein and P. Marcus on behalf of Working Party 6 on Surface Science
© 2007, Institute of Materials, Minerals and Mining
Preface
In 2001 the European Corrosion Conference, Eurocorr 2001, was organised on behalf of the European Federation of Corrosion (EFC) by the Assoziatione Italiana di Metallurgia (AIM), a member society of the EFC, in Riva del Garda. Each of the 17 working parties of EFC prepared sessions on particular topics in their respective fields of corrosion and corrosion protection. Jürgen Mietz, the chair of Working Party (WP) 11 at that time, organised sessions on corrosion of the reinforcement in concrete. Altogether 27 papers were 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 of these presentations was excellent and that the papers should be published together as an EFC book. Unfortunately, a number of problems meant that publication of the papers was delayed. In 2006 these problems were solved, allowing quick publication of the papers. As the papers still present current knowledge 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 in the field of corrosion of reinforcement in concrete and hope that all readers will receive new information, insights and ideas. Prof Dr-Ing Michael Raupach Chairman of EFC Working Party 11
xxiii © 2007, Institute of Materials, Minerals and Mining
1 Corrosion of metals in contact with mineral building materials U. N Ü R N B E R G E R, University of Stuttgart, Germany
1.1
Corrosion behaviour during contact with building materials that contain cement [1–5]
1.1.1
Iron and steel
Corrosion rate
In sufficiently moist oxygen-containing aqueous media that are nearly neutral to weakly basic, iron is transformed into iron(II) hydroxide with the help of water 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 medium pH-region of the iron–water system. (Fig. 1.1).
Iron
Zinc
Lead Aluminium Copper
2
4
6
8 10 pH-value
12
14
16
1.1 Effect of pH on corrosion rate of metals; reference data from a structural point of view [1].
1 © 2007, Institute of Materials, Minerals and Mining
2
Corrosion of reinforcement in concrete
In Portland cement-based concrete, steel is protected against corrosion because of contact with the highly alkaline concrete pore water. As a result of the strongly alkaline reaction during the hydration of cement, concrete has a large proportion of alkaline ingredients. The pH-value of the aqueous phase of normal concrete is about 12.6 to 13.8 depending on the content of ingredients 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 ingredients that are aggressive to steel and destroy passivity (causing pitting corrosion); there is, therefore, complete inhibition of the anodic partial reaction. The passivating layer of hydrated iron oxide is 2–20 nm thick. The corrosionprotective effect of the concrete for embedded steel is lost if the pH falls short of the above-mentioned value. However, it is not until below pH 9.0 that severe corrosion starts, as, for example, after carbonation of the concrete and the neutralisation of all alkaline ingredients. In addition to this, the cement must contain as much free water in its pores as possible and oxygen from the 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 negative standard electrode potential (–1.66 VH) results from the development of passivating oxide and hydroxide surface films. In the pH range 4 to 9, these films are largely insoluble. Because of this, aluminium materials are distinguished by good resistance to corrosion in nearly neutral to weakly acidic aqueous media and also in humid atmospheric conditions. This explains the wide application of aluminium in constructional engineering. Aluminium is an amphoteric metal. Because of this, the protective effect of the coating is lost as a result of its disintegration in strongly acidic and alkaline media. Aluminium and its alloys are mainly attacked by general corrosion in these pH regions and there is no question of applying these materials in such cases. However, noticeable disintegration can take place even in the more weakly alkaline region above pH 9 (Fig. 1.1). Aluminium materials then become as active as might be expected from their position in the electrochemical series and, even in the absence of oxygen, react to evolve hydrogen 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-based building materials, aluminium and its alloys are attacked through a reaction with the free alkali hydroxides of the cement solution within the pores because of general corrosion. In Fig. 1.2 (on the left-hand side), the corrosion reaction of aluminium is compared with that of other structural metals in wet cement mortar.
© 2007, Institute of Materials, Minerals and Mining
Corrosion of metals in contact with mineral building materials
3
PC-mortar + 1.6% Cl–(CaCl2) Gypsum plaster
PC-mortar 600
2
Mass weight loss (g/m2)
500
1 – Steel 2 – Aluminium 99.3 3 – Copper 4 – Lead 5 – Zinc
5
400
300
Liquid cement mortar in g m–2: 50 95 45 1190 955
Water 30 cm
200 4
4
1
2
5
100 1 1
3
5
3
2
3 4
1.2 Loss of mass of metals after twelve months of storage in moist building materials (the mortar blocks were immersed to a depth of 2 cm in the water during the tests) and in liquid cement mortar (numerical data) (in [1] according to the results of [2]).
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 and spall-off the concrete cover. Account must be taken of the fact that, depending on time spent in formwork and the dry-out conditions, hardening building materials often evolve their excess water very slowly, so that considerable corrosion damage can occur in fresh concrete before a low equilibrium moisture level is achieved. According to the information given in reference 5, the relative extent of corrosion in Portland cement mortar is in the ratio of 1 : 6 : 25 for dry, moist, and wet states, respectively. Figure 1.3 shows that the intensity of attack decreases with time, since the corrosion products that develop impede the transport of alkalis to the corroding surface. Anodically produced anodisation layers (Eloxal) are also attacked by moist alkaline-reacting building materials and under these circumstances do not offer effective protection. To preclude corrosion damage through contact with moist alkaline building materials, aluminium must, in addition, be protected by proper organic coatings. In metals that are susceptible to alkalis, such as aluminium and its alloys,
© 2007, Institute of Materials, Minerals and Mining
4
Corrosion of reinforcement in concrete 400 Al Mg Si 1 Al Mg 3
Loss in thickness (mm)
300 Fresh PC-mortar
200 Wet exposure Hardened PC-concrete 100
Moist exposure Dry exposure
0 0
3 6 9 Exposure period (month)
12
1.3 Corrosion behaviour of aluminium alloys in alkaline building materials 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.
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 of a suitable bonding agent (Fig. 1.4). Concretes, that, depending on their production route are only mildly alkaline, e.g. autoclave-treated pore concrete (gas concrete), are not to be classified as aggressive in respect of structural metals that are susceptible to alkalis, such as aluminium. For aluminium, carbonation of the building material will also have a corrosion-protective effect.
1.1.3
Copper
On one hand, the excellent corrosion resistance of the copper-based materials depends on the ‘noble’ character (the standard electrode potential of copper is +0.34 VH) and on the inability to form protective layers in many normal environments and on contact with building materials. In water and neutral salt solutions, copper-based materials have very good resistance to corrosion over a wide pH range (Fig. 1.1). In diluted (non-oxidising) acids and in the alkaline region, copper, above all, is superior to other non-ferrous metals. Copper and its alloys are unapplicable only if the formation of the protective
© 2007, Institute of Materials, Minerals and Mining
Corrosion of metals in contact with mineral building materials
5
Mass weight loss (g m–2)
250
200
150
100
1 = Portland cement 2 = Hydraulic lime 3 = Trass cement 4 = Portland blast-furnace cement 5 = High-alumina 6 = Gypsum
Lead
Aluminium
50
0 Cement 1 pH 12.7
Zinc
2 12.6
Copper
3 12.3
Iron
4 12.3
5 11.7
6 8.7
1.4 Loss of mass of metals after twelve months of storage in wet mortar with various bonding agents (testing arrangement as shown in Fig. 1.2) (in [1] according to the results of [2]).
layers is hampered and the material is heavily attacked through the formation of complex salts, e.g. in contact with ammonia and ammonia-containing solutions. The layers of slightly soluble copper(I) oxide, that form in air, are virtually insoluble in alkalis. Copper and its alloys therefore undergo negligible uniform corrosion when embedded in moist concrete or cement mortar (Fig. 1.2, lefthand side and Fig. 1.4). When applied in cements with higher alkality (pH value 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 in many media, including building materials, by reaction with its environment. Zinc is also an amphoteric metal and, therefore, is not resistant to corrosion in both acid (< pH 5) and alkaline regions (> pH 12) (Fig. 1.1). In more alkaline solutions, zinc hydroxide, which is transformed into readily soluble and non-protective zincates by reaction with the alkaline compound, develops under the formation of hydrogen. In alkaline concrete with pore solution pH values of between 12.6 and 13.8, for alkali-enriched cements, galvanisation would be expected to be susceptible to corrosion because of the amphoteric reaction. However, in fact it is found that, at least for pH values £ 13.3, the
© 2007, Institute of Materials, Minerals and Mining
6
Corrosion of reinforcement in concrete
dissolution rate of zinc under the formation of hydrogen quickly diminishes because of passivation. If Ca(OH)2 exists together with Zn(OH)2 , a further corrosion product, the slightly soluble calcium hydroxozincate Ca[Zn(OH)3]2 · 2H2O, develops, which is held responsible by some researchers for the passivation of zinc in concrete. Because of this, zinc is uniformly attacked in alkaline 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 restricted with rising alkality, and the corrosion of zinc increases greatly. In carbonated concrete, the corrosion rate of zinc can, indeed, be a little higher than in alkaline concrete, but it is still considerably below that of e.g. steel. Because of that, galvanised reinforcing steels may be applied where premature carbonation is expected. Heightened contents of chromium in the cements have a beneficial effect on the corrosion of zinc in the alkaline region because they promote rapid initial passivation of the zinc.
1.1.5
Lead [11]
Lead is relatively ‘noble’ in regard to corrosion, because of its position in the electrochemical 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 in acids, but also in alkalis (Fig. 1.1). In alkaline electrolytes, lead is heavily attacked above pH 9: Pb + 1/2O2 + Ca(OH)2 Æ CaPbO2 + H2O
(1.2)
Thus, primarily developed lead hydroxides are transformed into readily soluble plumbate, mainly calcium plumbate. Because of that, lead, in comparison to aluminium, is very susceptible to corrosion in moist alkaline building materials (Fig. 1.2, left-hand side). Intensified attack results as the pH 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 time and as the moisture level of the building material falls. In wet concrete or mortar, lead can be protected, e.g. by insulation with thick 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 become moistened by aqueous cement solutions (fresh concrete) or by aqueous extracts © 2007, Institute of Materials, Minerals and Mining
Corrosion of metals in contact with mineral building materials
7
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, and behaves in an alkaline manner. In such media, aluminium materials and lead in 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 rate hardly slows down. In zinc and copper as well, a much stronger attack than in hardened wet building material can be determined. Thus the corrosion reaction in alkaline solutions can not be compared with that in solid phases of a building material. In the case of contact of the metals with alkalinereacting electrolytes, irregular alterations in color can occur even after shortterm attack and, under certain circumstances, this can strongly affect the exterior 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 from salt penetration and of more than 0.5 to 1.0% content by mass (referred to the cement content), will cause extensive pitting corrosion of steels embedded in it and in a passive state (Fig. 1.2). In addition to the chloride content, the intensity of chloride-induced corrosion depends on other parameters of the concrete (pH value and kind of steel, aeration and water content). In structural elements made of zinc or in galvanized items with otherwise almost passive behaviour, a small increase of corrosion in alkaline building materials must be expected if the chloride content exceeds about 1.5% by mass related to the weight of cement. The higher critical content of chloride compared with steel results from the fact that the chloride ions are partially bound as slightly soluble basic zinc chlorides. In the presence of chlorides in the concrete/mortar, serious susceptibility to widespread pitting exists for aluminium, which in any case is sensitive to corrosive attack. The presence of chloride can intensify the corrosion in alkaline building materials by a factor of several times (Fig. 1.2). Even the smallest 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 additional chlorides (Fig. 1.2). This can be explained by the formation of slightly soluble 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.
© 2007, Institute of Materials, Minerals and Mining
8
1.4
Corrosion of reinforcement in concrete
Corrosion behaviour during contact with building materials containing magnesia cement [2,12]
A number of building materials, such as stone wood (for flooring) and light building boards, are made of wood-shavings that contain magnesia cement as a binding agent. Magnesia cement is tempered by mixing lightly burnt magnesite and concentrated magnesium chloride solution and hardened by the formation of a compound of low solubility (probably 3MgO·MgCl2·11H2O). The pores of the building material are filled with a magnesium chloride solution. If the building materials that are bound with magnesia cement come 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 the strong hygroscopic character of magnesium chloride (pores containing magnesium chloride do not dry up at relative humidities >32%) on the one hand, and the corrosion-promoting characteristics of concentrated chloride solutions on the other. Therefore, corrosion is possible even in a comparatively dry environment. In the case of a reaction between the moisture that is always present in stone 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, very aggressive corrosion conditions are created for steel pipes and, therefore, pipes made of unalloyed steels or galvanised steels may not be laid in stone wood floorings without external protection.
1.5
Corrosion behaviour during contact with gypsum products [1–3, 6, 12]
Structural gypsum (e.g. gypsum mortar) that is mixed with water forms a compound of needle-shaped dihydrate crystals CaSO4·2H2O, as soon as the fluid pulp reacts. Because of the unusually high surplus water resulting from fresh gypsum, the porosity of the hardened building material is quite high. If the hardened gypsum products (gypsum plasters, gypsum pasteboards, gypsum wall-construction boards) are kept moist, the pores are filled with a saturated calcium sulphate solution. Since this salt has a corrosion stimulating effect in neutral building materials, gypsum and gypsum mortar attack zinc and iron (or steel) very strongly in combination with humidity (Fig. 1.2 on the right-hand side and Fig. 1.4). Steel pipes and galvanised steel pipes that are in contact with gypsum, which is moistened long-term, are attacked by thick rust products and can be destroyed even after a few years. At relative humidities of <99%, gypsum mortar completely drains with time and no longer causes corrosion of steel and zinc. © 2007, Institute of Materials, Minerals and Mining
Corrosion of metals in contact with mineral building materials
9
Aluminium materials and lead in general are not likely to be attacked by the 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 pitting corrosion. However, aluminium alloys that contain copper sometimes corrode quite intensively in wet gypsum. Copper materials are also largely resistant to gypsum because the surfaces are 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 Nichteisenmetallen gegenüber verschiedenen Zementen und Mörteln. Schweizer Archiv, 1965, 31, 273– 293. 3. W. Wiederholt and J. Sonntag, Korrosion von Metallen im Bauwesen. Berichte aus der 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 aus Nichteisenmetallen bei Berührung mit Baustoffen. Werkstoffe Korrosion, 1986, 37, 318–322. 6. E. Fischer and H. Voßkühler, Verhalten von Aluminiumlegierungen gegenüber Mörtelmischungen. Aluminium, 1957, 33, 602–612. 7. H. Kaesche, Zum Elektrodenverhalten des Zinks und des Eisens in Calciumhydroxidlösung und in Mörtel. Werkstoffe und Korrosion, 1969, 20, 119– 124. 8. C. Andrade, J. D. Holst, U. Nürnberger, J. D. Whiteley and N. Woodman, Protection systems for reinforcement. Bull. ‘Information No 211’, CEB, Lausanne, 1992. 9. K. Menzel, Zur Korrosion von verzinktem Stahl in Kontakt mit Beton. Dissertation Universität Stuttgart, 1992. 10. U. Nürnberger, Besondere Maßnahmen für den Korrosionsschutz und zur Sanierung von 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 Korrosionsschutz metallischer Werkstoffe im Hoch- und Ingenieurbau. Verlag Stahleisen mbH Düsseldorf, 1976, 235–246. 12. U. Nürnberger, Korrosionsverhalten von feuerverzinktem Stahl bei Berührung mit Baustoffen. Werkstoffe und Korrosion, 1986, 37, 302–309.
© 2007, Institute of Materials, Minerals and Mining
2 Corrosion and electrochemistry of zinc in alkaline solutions and in cement mortar K. V I D E M, University of Oslo, Norway
2.1
Introduction
According to the Pourbaix diagram [1], shown in Fig. 2.1, zinc is expected to corrode at pH >10.5 with the formation of the soluble hydrogen zincate ions, HZn O 2– . Above pH 13.1, the dominating soluble product is zincate ions, Zn O 2– 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, the behaviour of zinc-coated steel reinforcement in concrete is usually good. Galvanized steel reinforcement is used successfully in situations where steel will corrode due to chloride or carbonation. However, the corrosion rate in aerated, strongly alkaline solutions is too high for most practical applications. Therefore, this subject has received little attention. Bocris et al. [2] describe cathodic and anodic reactions in the presence of fairly high levels of zincate ions – Tafel slopes, exchange current and the mechanism of dissolution and deposition being the main subjects. Chloride induced passivity breakdown and pitting are treated in detail by Guo et al. [3]. The anodic behaviour at about 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]. The present report describes experiments with zinc in KOH solutions, synthetic 1 Corrosion
0.2
Passivity
E [V(NHE)]
0.6
–0.2 –0.6
Corrosion
–1 Immunity –1.4 –1.8 –2
0
2
4
6
8
10
12
14
pH
2.1 Simplified potential pH diagram for zinc [1].
10 © 2007, Institute of Materials, Minerals and Mining
16
Corrosion and electrochemistry of zinc
11
concrete pore water and in cement mortar, including studies of some selected phenomena by cyclic voltammetry and electrochemical impedance spectroscopy (EIS).
2.2
Experimental methods
Static corrosion experiments were carried out in 0.5 L glass vessels open to air. Test conditions are briefly described together with the results. Page and Vennesland [7] have found that the pore water in new concrete can consist of about 0.2M KOH – 0.1M NaOH saturated with Ca(OH)2. The present test solutions 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. The chemicals were analytical grade. The electrodes used for electrochemical experiments in both solutions and in cement mortar were made from superpure zinc in the form of 6-mm-diameter discs masked off with evacuated epoxy, giving a flat, circular working area of 0.5 cm2. The surface was prepared by abrasion with 1000 grit paper followed by etching in 0.05 wt % HCl. Electrochemical experiments in solutions were carried out in 0.6 L closed bottles with apertures for the entry of the working electrode, reference electrode and AISI 316 counter electrode, and with a stirring magnet and air bubbled though the solutions. Similar zinc electrodes were cast into cement mortar cylinders of about 8 cm diameter and 15 cm high. Sand and Portland cement 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% of the cement weight. The cement mortar slabs were covered with water for the first 10 days and then stored in air with 95% relative humidity in a stainless-steel container at 22 ± 2∞C. The response to different moisture levels was of interest. Therefore, after 130 days exposure, the mortar was saturated with water. The slabs were then allowed to become dryer for a period of 200 days and then wetted again. The zinc electrodes in cement mortar were accompanied by embedded reference electrodes made of copper at a separation of about 3 mm. The cement mortar slabs also had embedded copper wire counter electrodes at the periphery. This significantly improved the electrochemical measurements with applied current as they could be performed without wetting the surface. The potentials of the copper reference electrodes and corrosion potentials of the zinc electrodes were measured at intervals with a calomel electrode contacting the cement mortar surface via a small, moist piece of paper. As the zinc electrodes sometimes had polarization resistance in the MW-range, a normal multimeter would draw far too much current. An electrometer with as low bias as 10 pA was used. Electrochemical impedance spectroscopy was performed with a Gamry 900 EIS instrument. A Gamry CMS 100 computer controlled potentiostat and galvanostat was used for the other measurements.
© 2007, Institute of Materials, Minerals and Mining
12
Corrosion of reinforcement in concrete
2.3
Results
2.3.1
Static corrosion exposures
Specimens of 15 cm2 surface area were exposed in 0.5 L glass vessels open to air under static conditions. The corrosion product dissolved when the hydroxyl concentration was 0.03M and above. A period with a nearly constant corrosion potential and constant corrosion rate occurred shortly after the start of the exposure. The weight loss after 18 h exposure was the same for exposures in 0.03 to 1M hydroxyl (pH 12.5 to 13.9) and corresponded to a corrosion rate of 0.4 mm y–1. Corrosion potential and linear polarization resistance (LPR) were measured by contacting the specimens with a zinc needle of about 1 mm diameter. LPR was about 400 W cm2 after exposure for 3 and 9 h and independent of pH. Gentle stirring of the solutions and even work at the bench with the test vessels leading to vibrations reduced the corrosion potential. Therefore, these corrosion rates are not generally applicable, but apply for the test geometry and conditions in this lab. The solutions were not replaced in these tests and soon contained high levels of hydrogen zincate ions. After 10 days, specimens exposed to hydroxyl concentrations of 0.1M and below had a white corrosion film and gained weight. As the corrosion film did not protect the complete surface, pitting occurred. The corrosion product liberated CO2 when dissolved in acid, proving the presence of zinc carbonate. It is assumed that CO2 from the air had removed the Ca2+ in the solutions, despite solid Ca(OH)2 having been placed at the bottom of the test vessels. These experiments revealed interesting aspects. However, as CO2 is absent 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 continuously bubbled though the test solutions contained in closed glass vessels. Figure 2.2 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 to a 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 the step, LPR varied between 600 and 800 W cm2, showing a change from active to passive corrosion. A small constant anodic current was applied for a short period to determine LPR and the potential shift measured with an electrometer. To avoid disturbance of the electrode, currents that gave potential shifts of only 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. To check this, experiments in 0.1M KOH were carried out. The change from the active to the passive state also took place in this case, but appeared to require
© 2007, Institute of Materials, Minerals and Mining
Corrosion and electrochemistry of zinc
13
–0.400 Passive
E [V(SCE)]
–0.600 –0.800 –1.000 –1.200 Active
–1.400 –1.600 0
200
400 Time (min)
600
800
2.2 Corrosion potential as a function of time at pH 13 in a solution containing KOH and NaOH in the proportion 2:1 and saturated with Ca(OH)2. Experiments in KOH at the same pH gave rather similar results.
slightly longer time. Any effect of calcium on the corrosion potential and LPR in the active, as well as in the passive state, was hardly detectable. It is possible that calcium affects corrosion, but is not the cause of the passivity. These measurements were carried out under agitation. The corrosion rate in the active state was as high as 5 mm y–1. and thus much higher than in the static exposures. The high corrosion rate led to a significant release of HZn O 2– to the test solution. When passivity occurred after 210 minutes in the experiment shown in Fig. 2.2, the HZn O 2– concentration was about 0.03M. The possibility was considered that the shift from active to passive was initiated by accumulation of this corrosion product. To explore this, the experiments were repeated in the used solutions. In this way, the hypothesis that hydrogen zincate ion accumulation caused the passivity was rejected.
2.3.3
Cyclic voltammetry
Figure 2.3 shows current density as a function of potential at pH 13. The effect of hydroxyl concentrations from 0.01 to 1M is illustrated in Fig. 2.4. In the last case, the measurements were taken in deaerated KOH solutions. The sweeps started in the positive direction from a slightly cathodic potential. It is seen from these figures that the corrosion potential decreased with increased alkalinity and that the current density at a given potential increased. The anodic dissolution took place in different regions, as indicated at Fig. 2.4. Up to about 30 mV above the open circuit potential, the current increased sharply with the potential. This region is indicated as ‘Tafel’ in Fig. 2.4. At higher potentials, the current went over a flat maximum in a region with current little affected by potential. As the anodic current density in this region increased with increased agitation, it is assumed that the anodic dissolution was restricted
© 2007, Institute of Materials, Minerals and Mining
14
Corrosion of reinforcement in concrete
CD (mA cm–2)
1
Cathodic peak
0.1
Up
0.01
Reversed
0.001 –1.5
–1.3
–1.1
–0.9 E [V(SCE)]
–0.7
–0.5
–0.3
2.3 Current density as a function of potential at pH 13. Same solution as for Fig. 2.2. Sweep rate 1 mV s–1. Curves obtained with automatic compensation for the potential drop between working and reference electrode. 1000 Diffusion
CD (mA cm–2)
100
1M
Passive
10 Tafel 1
0.1 M
0.1 0.01 M
0.01 0.001 –1.6
–1.4
–1.2
–1
E [V(SCE)]
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.
by the slow escape of dissolved oxidation products by diffusion and forced convection. This region is marked ‘Diffusion’ in Fig. 2.4. At still higher potentials, passivity occurred. However, passivity reduced the dissolution rate 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 was also agitation dependent. An unusual aspect was that the passive oxidation rate 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 corrosion products 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 in alkaline solutions under anodic polarization. A detailed study, including also
© 2007, Institute of Materials, Minerals and Mining
Corrosion and electrochemistry of zinc
15
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 of chloride was detected under conditions where the corrosion product dissolved. In the experiment referred to in Fig. 2.5, local passivity breakdown and pitting occurred at –0.46 V (SCE) leading to a current increase of almost three orders of magnitude. Very interesting aspects were observed in the reverse 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 the current at a given potential was more than an order of magnitude higher than before passivity breakdown occurred. An astonishing aspect is that the cathodic current was even higher. It appears that the breakdown event has created defects with very high anodic as well as cathodic activity. This indicates that so called active zinc is not ‘naked’ in the solution, as the increased cathodic reaction rate would then be difficult to explain. Figure 2.6 shows corrosion potential without applied current as a function of time at pH 13 and with 0.15 M NaCl added to the solution of KOH, NaOH 1000
CD (mA cm–2)
Pitting
Cathodic
100 10
Reverse
1 0.1 Forward 0.01 0.001 –1.5
–1 E [V(SCE)]
–0.5
2.5 Current density as a function of potential at pH 13 and a solution containing 0.15M NaCl. Sweep starting from a slightly cathodic potential and with a rate of 1 mV s–1. –0.400
E [V(SCE)]
–0.600 –0.800 –1.000 –1.200 –1.400
0
2000
4000 Time (min)
6000
8000
2.6 Corrosion potential as a function of time at pH 13 and a solution containing 0.15 M NaCl.
© 2007, Institute of Materials, Minerals and Mining
16
Corrosion of reinforcement in concrete
and Ca(OH)2. It is surprising that chloride did not prevent the set up of passivity. Just at the onset of passivity, the corrosion potential was about the same as without chloride for a short time (see also Fig. 2.2). Thereafter, a sudden drop in the potential took place. It is assumed that the passive potential occuring after 1500s was higher than the pitting potential. Therefore, defects were formed. At the lower potential after the drop some pits must have stifled, 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 2.7 applies for two electrodes cast into mortar of sand and cement, while Fig. 2.8 relates to mortar with a chloride content equivalent to 1% of the cement weight. To study the response to moisture level, the mortar was water saturated in some periods. This is indicated with ‘w’ at the top of Figs. 2.7 and 2.8. It
E [mV(SCE)]
200
WW
W
W
W
–200 –600 –1000 –1400 0
500
1000 Time (days)
1500
2.7 Corrosion potential as a function of time for two electrodes in mortar of sand and cement. Periods with water saturated mortar are indicated with ‘w’ at the top of the figure.
E [mV(SCE)]
200
WW
W
W
W
–200 –600 –1000 –1400
0
500
1000 Time (days)
1500
2.8 Corrosion potential as a function of time for two electrodes in mortar with 1% Cl–. Periods with water saturated mortar are indicated with ‘w’ at the top of the figure.
© 2007, Institute of Materials, Minerals and Mining
Corrosion and electrochemistry of zinc
17
is seen that moisture level is a very important parameter, leading to low corrosion potentials. As will be described later, this is indicative of high corrosion rates. The addition of chloride to the mortar had a large effect and made corrosion more severe. Without chloride, the electrodes seldom had potentials below –0.7 V (SCE). For mortar with 1% Cl– the potentials usually were below –0.7 V (SCE) and many reading below –0.9 V (SCE), as seen from Fig. 2.8. The average of all measurements was reduced by about 150 mV with 1% chloride. In contrast to exposures in alkaline solutions, hardly any well-defined polarization resistance exists for zinc in cement mortar. Different values were obtained with different techniques. Figure 2.9 shows the polarization resistance obtained by a potential step of 20 mV. Polarization resistance in this case is the ratio between this potential shift and the current density. The values obtained in this way increased by about an order of magnitude when the polarization time was increased from 10 to 100 s. As the lines in Fig. 2.9 are 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 corrosion rate 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 polarization curves have no straight parts for experiments in solutions with oxygen. The 100
LPR (MW cm2)
10
–51 mV(SCE)
1
–656 mV(SCE)
0.1 –1133 mV(SCE) 0.01
0.001 1
10
100
1000
Time (s)
2.9 Examples of values for polarization resistance for electrodes in cement mortar obtained with potential steps of 20 mV and various polarization times. Electrodes with corrosion potentials of –1133 mV, 685 mV and –51 mV (SCE).
© 2007, Institute of Materials, Minerals and Mining
18
Corrosion of reinforcement in concrete 1
CD (mA cm–2)
Modelled 0.1 Observed 0.01
0.001 –1.45
–1.43
–1.41 –1.39 E [V(SCE)]
–1.37
–1.35
2.10 Measured current as a function of potential at pH 13 and modelled 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 curve shown in Fig. 2.3.
curvature is due to corrosion rates being so high that the anodic reaction rate takes place at the border of the Tafel region. As shown from Fig. 2.10, good fit with the Tafel region in Fig. 2.3 was obtained by assuming a corrosion current of 0.034 mA cm–2, a cathodic Tafel slope of –110 mV decade–1 and anodic Tafel slope of 27 mV decade–1. The cathodic slope fits with oxygen reduction as the cathodic reaction. The anodic current in the Tafel region was independent of agitation. Increased agitation raised the Tafel region to higher potentials due to reduced restrictions caused by mass transfer. The Tafel slope varied somewhat with pH. Sweep rate of 1 mV s–1 led to slopes of 27 and 34 mV decade–1 at pH 14 and 13, respectively. More accurate measurements with instantaneous potential steps resulted in 22.5, 31.0 and 31.5 mV decade–1 for pH 14, 13.5 and 13, respectively. At pH 12.5 and below, the Tafel region became so small that a determination was hardly possible. Kabakof [8] reports an anodic Tafel slope of 30 mV decade–1, in good agreement with this study, while Bocris et al. [2] obtained a value of 49±13 mV decade–1 in a carefully purified solution. The present Tafel slopes for the active state give a Stern–Geary constant of 10.0 mV for estimation of corrosion rates from LPR, corresponding to a corrosion rate of 0.145 mm y–1 for an LPR of 1 kW cm2. Using this value for the static corrosion experiments in alkaline solutions, gave acceptable agreement with corrosion rates obtained by weight loss. As mentioned, the corrosion rate was independent of pH in these static corrosion tests despite a large pH effect on the anodic kinetics. Corrosion being limited by oxygen supply rate is probably the main phenomenon responsible for this.
2.4.2
Passivity
Figures 2.3 to 2.4 show sudden current reductions occurring at potentials a few hundred mV above the corrosion potential. The existence of passivity © 2007, Institute of Materials, Minerals and Mining
Corrosion and electrochemistry of zinc
19
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 from the 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 the current in the reverse sweep followed the same trace, this phenomenon is real. It should be noted that these curves were obtained with automatic compensation of the resistance potential drop in the gap between the working and the reference electrode. The reduction in anodic rate was about 10 times at pH 14 and 5 times at pH 13. From this it is concluded that the passive film is only slightly protective, compared to typical passive film that often reduces the reaction rate many orders of magnitude. In the classical description of passivity, the current at steady state is independent of potential. From Fig. 2.3 it is seen that the current was raised by 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 film was attacked by hydroxyl ions, and that charge was consumed to maintain the film. This view is supported by the observation that the passive current at a given potential was more or less proportional to the hydroxide concentration. The various observations indicated the presence of a passive film that is porous (at least at the outer part), becomes thicker at increased potential, and has a larger area in contact with the solution due to the pores. The last point is meant to take care of the potential dependence of the passive current. Passivity occurring without polarization in aerated solutions is very surprising. The required passivation current density was rather high under polarization. From the experiments with cyclic voltammetry (see Figs. 2.3 and 2.4), oxygen reduction appears to be far too slow to supply the necessary current. A period in the active state between 5 and 40 ks was necessary to bring the electrodes to the passive state without polarization. An interesting aspect 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 constant until 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 containing Ca2+ is shown in Fig. 2.11. This diagram resembles two arcs. The first arc
© 2007, Institute of Materials, Minerals and Mining
20
Corrosion of reinforcement in concrete
Imag (W)
25 0.01
20 15 10 5 0 –5 –10
1
0.1 10 0.001
Max. 10000 0
0.0001
50
100
150
Real (W)
2.11 Nyquist diagram at the corrosion potential in the active period for exposure at pH 13 with air and CO2 free conditions. Exposure 2 h when EIS was started. Corrosion potential was –1.34 V (SCE). Specimen area 0.5 cm2. Numbers are frequencies in Hz. 20000
Imag (W)
15000
0.2
10000
0.02
5000 0.002 0
Max. 2000
0.0002
–5000 0
20000 Real (W)
40000
2.12 Nyquist diagram at the corrosion potential in the passive state and same solution as in Fig. 2.11. Corrosion potential was –0.430 V (SCE). Specimen area 0.5 cm2. Numbers are frequencies in Hz.
extrapolates to about 100 W. As the Nyquist diagrams are not corrected for the specimen area (being 0.5 cm2), this indicates a LPR value of about 50 W cm2 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 an inductive tail that could indicate an adsorption phenomenon. EIS is difficult to interpret in this case, as the change of modulus and phase angle with frequency is not caused by this parameter alone. Many hours were required for the measurements at the lowest frequencies, leading to changes of the electrodes. However, the theory of an adsorption phenomenon triggering passivity appears reasonable as no other phenomena are identified. Figure 2.12 shows a Nyquist diagram from the passive state and has a shape in harmony with a metal with a passive film. The zigzag path at frequencies below 0.001 Hz is due to variations of the corrosion potential and illustrates the difficulties with EIS at such low frequencies that the measurements take many hours. Two examples of Nyquist diagrams for zinc in cement mortar are shown in Figs. 2.13 and 2.14. Figure 2.13 applies for an electrode with a very low
© 2007, Institute of Materials, Minerals and Mining
Corrosion and electrochemistry of zinc
21
200 0.00019
Imag (kW)
150 100 50 0.0019 Max 30000 0
0
20
40
60
Real (kW)
2.13 Example of Nyquist diagram for zinc in cement mortar. Electrode with a low corrosion potential, –1.009 V (SCE). 425 days exposure. Specimen area 0.5 cm2. 500 0.00019
Imag (kW)
400 300 200
0.0019
100 Max 30000 0 0
100
200 Real (kW)
300
400
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.
corrosion potential, –1.009 V (SCE), while Fig. 2.14 is for one with a much higher potential, –0.613 V (SCE). All Nyquist diagrams for electrodes in mortar consisted of constant phase angle lines at the low frequency side. As the constant-phase-angle elements were not observed in solutions, it is associated with the conditions in the mortar. The phenomena responsible for this have not been identified. Warburg impedance theory encounters difficulties with the slope in the Nyquist diagrams and the frequency response of impedance. The slope in the Nyquist diagram in Fig. 2.13 is about 4 and thus very different from the unity predicted by Warburg theory for homogenous environments. 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 value for Warburg impedance being –0.5. LPR values cannot be found from these diagrams because arcs for extrapolations are missing. However, qualitatively EIS responded to the severity of corrosion. The values for the real and imaginary components at a given
© 2007, Institute of Materials, Minerals and Mining
22
Corrosion of reinforcement in concrete
frequency were much higher for electrodes with high corrosion potentials than those with low, as seen from Figs. 2.13 and 2.14.
2.4.4
Polarization resistance and corrosion rate for zinc in mortar
As already described, active electrodes corroding in solutions had clearly defined values for Tafel slopes and polarization resistance. The corrosion rate can be obtained from electrochemical measurements in these cases. Both the Stern Geary treatment as well as modelling of polarization curves appear to function satisfactorily. The polarization resistance for passive electrodes in solutions varied with the measuring parameters, but to a much lesser extent as was seen for electrodes in cement mortar. In contrast to alkaline solutions, the assessment of the corrosion rate for zinc in cement mortar from electrochemical measurements is subject to severe difficulties. The problem is that the metal lacks a clearly defined polarization resistance, the values varying substantially with the method and parameters of the measurements. Figure 2.9 gives one example, showing polarization resistance values varying by two orders of magnitude depending on the duration of polarization. An earlier report describes polarization resistance obtained by galvanostatic polarization [9]. The apparent values for polarization resistance varied in the same way with the duration of polarization, but the values for the electrodes were nearly ten times higher. As could be expected, polarization resistance obtained by potentiodynamic sweep increased with reduced sweep rate. Therefore, quantitative values for the corrosion rate cannot be obtained from any of these measurements. No theory exists for assessing the severity of corrosion from these electrochemical measurements. Also, for the time being, the engineering database for doing this empirically is lacking. It is tempting to suggest that the lowest potentials in Figs. 2.7 and 2.8 apply for active zinc and the highest for passive metal. This was done in a previous report [9]. However, both the EIS and dc methods indicate that corrosion becomes gradually more severe with reduced corrosion potential without any threshold for change from the passive to active state. The various types of measurements agreed qualitatively in that they indicated that low corrosion 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. This indicates that corrosion rates of zinc in this study exhibit very large variations, possibly two orders of magnitude. However, it is premature to draw any conclusion about the applicability of zinc under conditions giving very low corrosion potentials. It is well established that moisture is a very important variable for the
© 2007, Institute of Materials, Minerals and Mining
Corrosion and electrochemistry of zinc
23
corrosion of galvanized steel reinforcement in concrete. Moreno et al. [10] have observed corrosion potentials in the region –350 to 200 mV (SCE) for galvanized steel in concrete without added chloride stored in air, and potentials between –600 and 400 mV (SCE) under wet conditions. This is in good agreement with the measurements shown in Fig. 2.7. An important difference between zinc and steel is that steel is passive in mortar without chloride and has rather high corrosion potentials independent of moisture level. The cement mortar slabs used in this work contained embedded steel electrodes. Wetting reduced the corrosion potential for steel in mortar with chloride only [11]. As already described, chloride reduced the corrosion potentials of zinc. Figure 2.9 demonstrates that low corrosion potentials are linked to low polarization resistance and hence higher corrosion rates, as confirmed by Moreno et al. [10]. Due to differences in the experimental techniques, their results are not directly comparable. Figure 2.8 shows potentials as low as –1000 mV (SCE) under moist conditions for mortar with 1% Cl– even after more than 1000 days exposure. This is far below the potentials reported by Moreno et al. [10] for concrete without chloride additions. Reliable relationships between corrosion potential and corrosion rate are missing both from laboratory experiments and from real life. Therefore, it is premature to draw any conclusion about the applicability of zinc under conditions giving very low corrosion potentials. Nevertheless, the low potentials in water saturated mortar is an aspect that needs attention. A really puzzling aspect is that zinc alters from a fairly plain polarization behaviour in alkaline solutions to a very complex one in cement mortar. The polarization resistance for zinc, shown in Fig. 2.9, varies in the same manner as for the steel electrodes embedded in the mortar slabs being used for the present work [11]. Potential steps gave currents decreasing with time to the power 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. As iron has oxides with different valence, redox of corrosion products can take place. The charge consumed by this has been suggested as the reason for the variation of current with the duration of polarization as well as the constant phase angle elements with EIS [12]. As zinc is only two-valent, redox of corrosion product cannot occur. The mechanism behind the changes of the performance with duration of the polarization is far from understood. The current change with the slope of –0.5 for a potential step, as well as the constant phase angle elements illustrated in Figs. 2.13 and 2.14 could indicate diffusion limited kinetics. To agree with the experimental results, the reaction rate 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 KOH solutions at the same pH as the present experiments. The same product has © 2007, Institute of Materials, Minerals and Mining
24
Corrosion of reinforcement in concrete
also been reported for the anodic film by Rudd and Breslin [14], who also state 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 in mortar formed corrosion films. Belaid et al. [15] reported that a film of calcium hydroxyzincate, Ca(Zn(OH)3)2 · 2H2O, passivates zinc in fresh concrete. 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 from which it is formed. Zn5(OH)8Cl2 · H2O occupies a 3.62 fold higher volume than that of zinc and has been held responsible for cracking of concrete with a high chloride content [15]. When zinc passivates in KOH solutions under anodic polarization, there are hardly any other possible passive films than zinc oxide or hydroxide. In the absence of polarization, passivation occurred faster when Ca2+ was present. This is consistent with calcium hydroxyzincate as the film material. However, absence of calcium also led to passive films that appeared to be equally protective.
2.5
Conclusions
The corrosion rate was about 0.4 mm y–1 in static corrosion tests conducted in alkaline solutions open to air. The rate was independent of pH in the range 12.5 to 13.9 during the initial period because the corrosion products were soluble and the corrosion rate was controlled by oxygen reduction kinetics. After some days of exposure, carbonate-containing corrosion films formed due to the reaction of CO2 from the air. Zinc exhibited an active/passive transition during cyclic voltammetry in alkaline solutions without CO2. The Tafel slope for active anodic dissolution was about 30 mV decade–1 and for cathodic oxygen reduction –110 mV decade–1. The passive current density was about 0.5 mA cm–2 at pH 13 and increased in proportion with the hydroxyl concentration. Increased agitation also raised the passive current, indicating that the passive film was continuously dissolved or destroyed by attack from hydroxyl ions. Zinc became passive after prolonged exposure without applied current in solutions that are in contact with air from which CO2 had been removed. At pH 13, passivation occurred after a few hours in solutions with Ca2+. Without Ca2+, passivation set in after slightly longer exposures. The polarization resistance in the active state was 20 to 30 W cm2 and between 600 and 800 W cm2 in the passive state. The passive reaction rate is thus much lower than that observed in cyclic voltammetry. The processes taking place in the induction period that suddenly trigger passivity are not understood. Adsorption phenomena are suggested from EIS. Chloride caused passivity breakdown and pitting under polarization at high potentials. The film damage was not fully repaired when the potential was reduced. A surprising observation was that the metal had a higher anodic
© 2007, Institute of Materials, Minerals and Mining
Corrosion and electrochemistry of zinc
25
as well as cathodic reaction rate at low potentials after the pitting event at high 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-saturated mortar contributed towards low corrosion potentials. The corrosion potentials were consistently lower in mortar with 1% Cl– than in mortar without and were reduced by 150 mV on average. It is concluded that the mixed-in chloride made corrosion more severe. In contrast to electrodes in solutions, Nyquist diagrams for electrodes in mortar consisted only of constant phase angle lines. The phenomena responsible for this were not identified. LPR values could not be determined from these diagrams because arcs for extrapolations were missing. The values for the real and imaginary components at a given frequency were much higher for electrodes with high corrosion potentials than those with low. In contrast with exposure in alkaline solutions, a well-defined polarization resistance hardly exists for zinc in cement mortar. Very different values were obtained with different techniques. Therefore, the corrosion rate in mortar could not be obtained from the electrochemical measurements. Polarization resistance achieved by the same measuring technique could be two orders of magnitude lower for electrodes corroding with a low potential than those with a high. Thus, a very large span of corrosion rates occurred in these experiments. It is premature to draw any conclusions about whether the rate was always acceptably low for practical applications in concrete. Caution regarding the use of zinc in very humid concrete is recommended until more knowledge 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 pozzolanic additions. 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.
© 2007, Institute of Materials, Minerals and Mining
26
Corrosion of reinforcement in concrete
12. C. Andrade, F. Bolzoni, A. Collazo, X. R. Novoa and M. C. Perez, Measurement of steel corrosion in concrete by electrochemical technique: influence of the redox processes 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.
© 2007, Institute of Materials, Minerals and Mining
3 Corrosion behaviour of galvanized steel rebars in the presence of coating discontinuities T. B E L L E Z Z E, R. F R AT E S I and F. T I T T A R E L L I Università Politecnica delle Marche, Ancona, Italy
3.1
Introduction
The limited durability of reinforced concrete structures has prompted considerable effort on research, aimed at the establishment of an adequate level of knowledge on the properties of this material. When concrete technology is correctly applied, it is possible to increase the service life of reinforced structures, even if it is difficult to ensure complete protection from aggressive agents, particularly when cracking of concrete or accidental causes of degradation occur. Therefore, under critical conditions for corrosion, only additional protection to the steel reinforcement can guarantee the durability of the structure. Among the possible methods for improving the corrosion resistance of reinforcement in concrete, new consideration has been given to the use of galvanized rebars, because of their relatively low cost relative to other protection systems. It is clear that the galvanized bars increase the initial cost of the concrete structures but, during their whole service life, this is not a major cost considering the rising costs of restoration and maintenance. Although good practical results have been reported in the literature [1–4], the benefits of using galvanized steel in reinforced concrete structures are still uncertain because of some controversial laboratory test results [5–11]. However, Swamy [12] stated that the results of laboratory tests must be viewed with caution due to the fact that the simulated environment does not fully match the actual conditions. Furthermore, it has been observed that zinc coating delays the onset of corrosion of reinforcing steel, as explained by the conceptual model proposed by Yeomans [13]. One of the unanswered questions concerning the use of galvanized reinforcement is the risk of corrosion where discontinuities are present in the zinc coating owing to bending of the bars or welding procedures which leave uncoated spots. In this work, the corrosion behaviour of galvanized reinforcement with discontinuities in the zinc coating was studied. For this purpose, several 27 © 2007, Institute of Materials, Minerals and Mining
28
Corrosion of reinforcement in concrete
reinforced concrete specimens were manufactured. The coating discontinuities were simulated by a small piece of black bar assembled with two lateral galvanized bars, electrically isolated from each other, in order to measure externally the galvanic corrosion current; these specimens were submitted to wet–dry cycles by ponding both with tap water and with a sodium chloride solution. At the end of the tests, all specimens were broken and the embedded reinforcements were examined to assess visually the corrosion attack.
3.2
Experimental methods
Sixteen prismatic specimens (14 ¥ 12 ¥ 44 cm) were manufactured using CEM II/A-L 42.5 R cement with a water–cement ratio of 0.70. The specimens were reinforced in the longitudinal direction with a bar (diameter = 12 mm; cover = 15 mm) obtained by assembling three electrically isolated bars: two lateral galvanized bars (anodic parts) of the same length and a small piece of black bar (cathodic part) in the centre. Four different types of bars were produced with anodic to cathodic surface area ratios, Sa/Sc as shown in Table 3.1. Figure 3.1 shows two assembled bars (Sa/Sc = 20 and 7.5) ready for casting. The assembly and electrical insulation were performed with a PVC insert (not shown in the figure) and epoxy resin between the two lateral galvanized bars and the central black bar. In Figure 3.1, the electrical cables for 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 (Table 3.1): two for each type of exposure condition. All specimens were removed from their moulds three days after casting. The exposure conditions included wet–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 curing and they consisted of five days of drying and two days of wetting. During the experimentation period, there were some longer drying periods. During the whole test period (about 270 days), the galvanic coupling between 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 (cm)
Central black bar (cm)
80 40 20 7.5
16 18 15 15
0.4 0.9 1.5 4.0
+ + + +
16 18 15 15
© 2007, Institute of Materials, Minerals and Mining
Corrosion behaviour of galvanized steel rebars
29
3.1 Two assembled bars with Sa/Sc = 20 (top) and Sa/Sc = 7.5 (bottom), used to study the galvanic protection on discontinuities.
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 resistance ammeter AMEL Mod. 668. Furthermore, potential measurements were performed with a calomel electrode (SCE) as a reference both during the coupling conditions and in free corrosion conditions; in this last case, the external electrical contacts between the bars were removed and the free corrosion potential measurements were performed after 1.5 h at open circuit. Two further specimens of the same type, but without reinforcement, were manufactured and submitted to ponding with 5% NaCl solution, in order to determine periodically the depth of chloride penetration during the wet–dry cycles. In the two different exposure conditions, concrete resistivity was also evaluated to study the possible ohmic control of galvanic corrosion. Four cubic 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 specimens and positioned vertically at a separation of 11 cm. Measurements of conductivity were performed during the whole corrosion test period using a digital conductometer AMEL Mod. 160 where 50 mV sinusoidal peak-to-peak signal and a frequency of 1 kHz is set. The values reported in the following section are the average of the measurements carried out on each type of specimen.
3.3
Results and discussion
Straight after casting the concrete, the galvanized steel assumed an active state with a free corrosion potential of about –1350 to –1400 mV, while the
© 2007, Institute of Materials, Minerals and Mining
30
Corrosion of reinforcement in concrete
black steel reached corrosion potential values of –400 to –520 mV (Fig. 3.2). After galvanic coupling of the anodic and cathodic parts, the galvanized steel cathodically polarises black steel, which assumed potential values in the range –850 to –1050 mV, very close to the thermodynamic immunity conditions for steel. A day after casting, the zinc coating became passive with a relative free corrosion potential value of –750 to –850 mV; the galvanic corrosion 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 after casting and they changed from a minimum value of 300 mA up to a maximum value of 1000 mA as the ratio Sa/Sc decreased, implying that an increase in bare 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 showed anodic macrocell control in the case of Sa/Sc = 20 and of Sa/Sc = 7.5, while for Sa/Sc = 40 the cathodic polarisation was almost equal to the anodic polarisation. For Sa/Sc = 80 there was cathodic macrocell control; this last result is quite clear considering the low value for the black steel surface. As a consequence, the cathodic protection is much more effective for a small defect in the zinc coating. The short circuit currents became very low with time and assumed values lower than 10 mA. The resistivity was low and ranged between 5 and 20 W cm during wetting periods, and between 20 and 75 W cm during normal drying periods. Because of the low values both of the short-circuit current and of the resistivity of concrete, ohmic drop control of galvanic corrosion has 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 and galvanized 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 of sodium chloride solution the galvanic corrosion came under cathodic control for every Sa/Sc value (Fig. 3.4), due to the damage of the zinc coating passive film. In particular, after 3 wet–dry cycles, the galvanized bars exerted ‘cathodic prevention’ [14] toward the chloride attack with respect to the coupled black bars. In fact, considering that at a depth of 15 mm (equal to the cover of the
© 2007, Institute of Materials, Minerals and Mining
900 800
–600
700
–800
600
–1000
500
–1200
400
–1400
300 200
–1600
Sa /Sc = 80
–1800
Sa /Sc = 40
0 1000
–2000 0 –400
800
–600
700
–800
600
–1000
500
–1200
400
–1400
300 200
–1600
Sa /Sc = 20
–1800
Sa /Sc = 7.5
–2000 0
7
14 Time (days)
21
28 0
7
14 Time (days)
21
100
Short circuit current (mA)
900
–200 Potential (mV vs SCE)
100
0 28
© 2007, Institute of Materials, Minerals and Mining
31
3.2 Trend of the free corrosion potential, short circuit potential and short circuit currents, during 28 days of curing in atmosphere: ■ free corrosion potential of black steel; ● free corrosion potential of galvanised steel; æ■æ short circuit potential of black steel; æ●æ short circuit potential of galvanized steel; ···▲··· short circuit current.
Corrosion behaviour of galvanized steel rebars
Potential (mV vs SCE)
–200 –400
Short circuit current (mA)
1000
0
320
–600
280
–800
240
–1000
200
–1200
160
–1400
120
–1600
80
Sa/Sc = 80
–1800
Sa/Sc = 40
–200
360
–400
320
–600
280
–800
240
–1000
200
–1200
160
–1400
120 80
–1600
Sa/Sc = 7.5
Sa/Sc = 20
–1800 –2000 28
56
84
112
140 168 196 Time (days)
224
252
280 28
56
84
112
140 168 196 Time (days)
224
252
Short circuit current (mA)
0 400
–2000 0 Potential (mV vs SCE)
40
Short circuit current (mA)
–400
40 0 280
3.3 Trend of the free corrosion potential, short circuit potential and short circuit currents, during wet-dry cycles with tap water: ■ free corrosion potential of black steel; ● free corrosion potential of galvanised steel; æ■æ short circuit potential of black steel; æ●æ short circuit potential of galvanized steel; ···▲··· short circuit current; ····wetting periods. © 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete
Potential (mV vs SCE)
400 360
32
0 –200
1800 1600
–600
1400
–800
1200
–1000
1000
–1200
800
–1400
600
–1600
Sa/Sc = 40
Sa/Sc = 80
–1800
200
0
2000
–200
1800
–400
1600
–600
1400
–800
1200
–1000
1000
–1200
800 600
–1400 –1600
Sa/Sc = 7.5
Sa/Sc = 20
200
–1800 –2000 28
400
Short circuit current (mA)
0
–2000
Potential (mV vs SCE)
400
56
84
112
140 168 196 Time (days)
224
252
280 28
56
84
112
140 168 196 Time (days)
224
252
0 280
© 2007, Institute of Materials, Minerals and Mining
33
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 circuit potential of black steel; æ●æ short circuit potential of galvanised steel; ···▲··· short circuit current; ···· wetting periods.
Corrosion behaviour of galvanized steel rebars
Potential (mV vs SCE)
–200 –400
Short circuit current (mA)
2000
0
34
Corrosion of reinforcement in concrete
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), which approximately corresponds to the pitting potential of a black bar embedded in 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, which simulates 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 values lower than –600 mV vs SCE, which assures steel cathodic prevention, even if 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 and galvanized steel, respectively. In this case, the low values of free corrosion potential 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 force between the galvanized and bare steel. Furthermore, the short-circuit current increased as the Sa/Sc decreased, implying that an increase in bare steel surface leads to a higher consumption of the adjacent zinc coating. The resistivity, which contributes to the possible ohmic drop control, was very low: 3 to 5 ohm cm during wetting periods and 5 to 15 W cm during drying periods. Therefore, even if the short-circuit currents in this case are significantly high, the ohmic drop control of the galvanic corrosion has still to be considered ineffective.
Chlorides content (% by weight of cement)
6 5 4 3 2 Concrete cover 1 0 0.0
0.5
1.0
1.5 2.0 2.5 3.0 Concrete depth (cm)
3.5
4.0
4.5
3.5 Trends of chlorides content at various depths in the concrete specimens, after 3 wet-dry cycles (—◆—) and 10 (—●—) wet-dry cycles with 5% NaCl solution.
© 2007, Institute of Materials, Minerals and Mining
Corrosion behaviour of galvanized steel rebars
3.3.3
35
Examination of the reinforcement
At the end of the exposure period, all bars were examined in order to compare their visible corrosion conditions with the data obtained by the electrochemical measurements. No red rust was observed on any black bar embedded in the manufactured specimens exposed both to wet–dry cycles with tap water and to wet–dry cycles with NaCl solution. These simulated discontinuities of the zinc coating were protected against corrosion by the cathodic protection offered by the adjacent zinc, even if the potential values did not reach thermodynamic immunity. The galvanized bars embedded in the specimens exposed to wet–dry cycles with tap water were not corroded, with the exception of those bars coupled with the largest black bars (Sa/Sc = 7.5). In this case, near the bar joints, the galvanized bars appeared dark due to the consumption of the external pure zinc layer of the coating (h phase), which permits the underlying Zn–Fe alloy layer to appear on the surface (z phase visible from the cross-section reported in Fig. 3.6a). In the presence of chloride ions, the corrosion attack of the zinc coating close to the pieces of black bar was detected for all Sa/Sc values: very low for samples 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 the joints with black steel was heavily corroded and the localised attack had penetrated into the Zn–Fe alloy layers leading to local damage of the zinc coating (Fig. 3.6b). This shows that the larger the galvanized coating discontinuity to be protected, the higher the corrosion of adjacent zinc. Concerning the specimens manufactured in this work, the experimental results obtained demonstrate the beneficial protective effect against steel corrosion by zinc coating, even when discontinuities as large as 4 cm are
3.6 Metallographic pictures of galvanized bar cross-section in a zone close to the joint with the black steel bar when Sa/Sc = 7.5: (a) the galvanized bar was embedded in the specimen exposed to wet–dry cycles with tap water; (b) the galvanized bar was embedded in the specimen exposed to wet–dry cycles with NaCl solution.
© 2007, Institute of Materials, Minerals and Mining
36
Corrosion of reinforcement in concrete
present and a high chloride level has reached the bars. However, a larger discontinuity produces higher corrosion in the adjacent zinc than a smaller one and the galvanized rebars offer more effective cathodic protection when the discontinuity is smaller.
3.4
Conclusions
Experimental tests were performed to simulate field conditions for reinforcing bars in concrete structures. Pieces of bare steel immersed in chloride-free concrete and in concrete contaminated with chlorides were coupled with galvanized steel with the aim to simulate the defects of galvanized surface due to bending or welding, that might be cathodically protected by the adjacent zinc coating. The short circuit potentials measured indicate that the zinc of galvanized steel exerts cathodic protection on bare steel for all Sa/Sc ratios examined. In particular, it exerts a ‘cathodic prevention’ against chloride attack by increasing the chloride concentration threshold able to induce the localized corrosion on bare steel. The damage to the zinc coating depends on the aggressiveness of the environment surrounding the bars and on the dimensions of the defect. In detail, in the absence of chlorides the ‘macrocell effect’ is very low, independent of the Sa/Sc ratio. However, in more aggressive environments, because of the presence of chlorides, zinc coating always increases the durability of reinforcement by a beneficial protective effect against steel corrosion, even when large discontinuities on the coating are present and a high chlorides level has reached the bars. In these conditions, even though the discontinuities up to 4 cm resulted protected from the adjacent zinc, discontinuities greater than 1 cm can be considered critical for corrosion phenomena of the zinc coating.
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 in Concrete, (Eds. D. E. Tonini and J. M. Gaidis), ASTM STP 713, American Society for Testing and Materials, Philadelphia, 1980, p. 102. 5. I. Cornet and B. Bresler, Galvanized Reinforcement for Concrete – II, International Lead 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.
© 2007, Institute of Materials, Minerals and Mining
Corrosion behaviour of galvanized steel rebars
37
7. C. Andrade and A. Macias, Surface Coating-2 (Eds. A. D. Wilson, J. W. Nicholson and 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 in Concrete (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.
© 2007, Institute of Materials, Minerals and Mining
4 Influence of scale and rust on steel activation in model concrete pore 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
4.1
Introduction
Many authors have used laboratory tests in model solutions to determine critical conditions for steel activation, using steel specimens with bare steel surfaces [1–3]. Nevertheless, in real conditions the steel reinforcement is utilized with a scaled surface that, depending on the duration of atmospheric exposure, is covered with rust to a varied extent. Laboratory tests, as well as long-term exposure tests in concrete, have revealed the negative influence of a rust layer on the corrosion resistance of steel reinforcement in concrete. The corrosion rate of pre-rusted reinforcement in moist non-carbonated concrete was found to be unacceptable even in the case of concrete that was not contaminated with chlorides [4, 5]. The published literature on the critical chloride concentration causing steel activation reports a wide range of values over several orders of magnitude (0.04 to 35 g L–1, pH = 12.5) [6]. The socalled critical concentration ratio of Cl–/OH– is believed to be the determining factor for steel activation [7, 8]. However, certain theories suggest that the laboratory determined dependence of critical chloride concentration on pH cannot, for various reasons, provide any practically usable specifications of limiting conditions for steel activation in concrete. In regard to the acceptable chloride content, the limiting criterion for construction use of reinforcement in non-carbonated concrete is an empirical value of 0.1 to 0.2 wt. % Cl– in the cement. Depending on the degree of water saturation and the quality of the concrete, this value corresponds to chloride concentrations in the pore solution of approx. 2 to 10 g L–1. In carbonated concrete, chlorides do not determine steel activation, although they affect the corrosion rate in the active state.
4.2
Experimental methods
The corrosion behaviour of carbon steel (0.2% C) in a model pore solution of 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 38 © 2007, Institute of Materials, Minerals and Mining
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. The aim was to determine – on the basis of electrochemical measurements of polarization resistance (LPR) – the critical chloride concentration at which the steel surface becomes activated. Measurements were performed for three types of steel surface: bare (as received clear metallic surface after degreasing), scaled (650∞C, 10 minutes) and pre-rusted surface (five months in an outdoor atmosphere with corrosion category C3 according to ISO 9223). The average thickness of the scale was 15 mm, the average thickness of the rust was 22 mm. The testing apparatus allowed for the parallel exposure of 15 specimens and statistical data evaluation. After 4 h of spontaneous passivation in alkaline solution with a particular pH value (without chlorides), the steel surface was exposed for 6 h to a solution containing chlorides. Measurements of polarization resistance were carried out by means of a Gamry measuring system with the CMS100 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 concentrations can be most precisely determined, for the bare surface at both values of pH and with a balanced concentration of oxygen, to be 1 g L–1. The pairs of identical lines in Figs. 4.1–4.4 express the ranges of measured values for 100 Bare Scaled Pre-rusted 10
Rp(W m2)
pH = 12.5 O2 = 1 mg L–1
1
0.1 <0.01
0.1
1 Cl– (g L–1)
10
4.1 Influence of chloride content in model pore solution on polarization resistance data of carbon steel at pH 12.5 (oxygen 1 mg L–1).
© 2007, Institute of Materials, Minerals and Mining
100
40
Corrosion of reinforcement in concrete
100 Bare Scaled Pre-rusted 10
Rp(W m2)
pH = 12.5 O2 = 8 mg L–1
1
0.1 <0.01
0.1
1 Cl– (g L–1)
10
100
4.2 Influence of chloride content in model pore solution on polarization resistance data of carbon steel at pH 12.5 (oxygen 8 mg L–1). 100 Bare Scaled Pre-rusted pH = 13.3 O2 = 1 mg L–1
Rp(W m2)
10
1
0.1 <0.01
0.1
1 Cl– (g L–1)
10
100
4.3 Influence of chloride content in model pore solution on polarization resistance data of carbon steel at pH 13.3 (oxygen 1 mg L–1).
each chloride concentration and each superficial state. In the case of the scaled 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 resistance only in pH 13.3 solution (with chloride concentrations approx. > 2 g L–1),
© 2007, Institute of Materials, Minerals and Mining
Influence of scale and rust on steel activation
41
100 Bare Scaled Pre-rusted pH = 13.3 O2 = 8 mg I–1
Rp(W m2)
10
1
0.1 <0.01
0.1
1 Cl– (g l–1)
10
100
4.4 Influence of chloride content in model pore solution on polarization resistance data of carbon steel at pH 13.3 (oxygen 8 mg L–1).
whereas at pH 12.5 no considerable decrease was observed with increasing chloride concentration. Compared with the bare and scaled surfaces, the corrosion resistance of the pre-rusted surface was the lowest in the region of low chloride concentrations (<1 g L–1). This concentration region corresponds to the unstable passive state of scaled and metallic surfaces during shortterm exposure; in the case of pre-rusted surfaces even the partial activation can 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 the basis of comparison of all the values obtained in the respective chloride concentration interval (see Tables 4.1 and 4.2). The short-term tests in model solution show the value of 2 Wm2 that allows for distinguishing the unstable Table 4.1 Portion of values of polarization resistance £ 2 W m2, model pore solution pH 12.5 [sat. Ca(OH)2], in bold when portion ≥50% Surface
Oxygen (mg L–1)
Portion of Rp £ 2 W m2 (%) <0.01 g L–1Cl–
0.05–1 g L–1Cl–
>1–10 g L–1Cl–
>10 g L–1Cl–
Bare
8 1
5 0
4 0
100 55
100 100
Scaled
8 1
13 0
25 8
50 42
100 100
Pre-rusted
8 1
67 83
67 100
100 100
100 100
© 2007, Institute of Materials, Minerals and Mining
42
Corrosion of reinforcement in concrete
Table 4.2 Portion of values of polarization resistance £ 2 W m2, model pore solution pH 13.3 [KOH + sat. Ca(OH)2], in bold when portion ≥ 50% Surface
Oxygen (mg L–1)
Portion of Rp £ 2 Wm 2 (%) <0.01 g L–1Cl– 0.05–1 g L–1Cl–
>1–10 g L–1Cl–
>10 g L–1Cl–
Bare
8 1
44 28
14 14
100 83
100 100
Scaled
8 1
40 17
29 29
67 67
77 70
Pre-rusted
8 1
83 100
67 100
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 (mg L–1)
Rp (W m 2) <0.01 g L–1Cl– 0.05–1 g L–1Cl–
>1–10 g L–1Cl–
>10 g L–1Cl–
Bare
8 1
20.8 39.7
8.2 54.8
0.2 13.3
0.1 0.5
Scaled
8 1
8.3 29.6
3.7 9.2
2.5 3.1
0.2 0.5
Pre-rusted
8 1
2.3 1.2
2.0 1.0
0.5 0.3
0.2 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 (mg L–1)
Rp (W m 2) <0.01 g L–1Cl– 0.05–1 g L–1Cl–
>1–10 g L–1Cl–
>10 g L–1Cl–
Bare
8 1
4.5 4.2
4.6 3.0
0.5 1.0
0.4 0.5
Scaled
8 1
3.1 4.3
3.0 2.9
2.0 1.2
0.9 1.6
Pre-rusted
8 1
1.4 0.6
2.2 0.4
0.6 0.4
0.3 0.3
passive surface from the partial surface activation. It is in contrast to the long-term tests, where the polarization resistance value corresponding with technically acceptable corrosion rate of carbon steel < 1 to 2 mm year–1 was determined to be >30 Wm2 [9]. The number of measurements in each interval of chloride concentrations was 6 to 30. The average values of polarization resistance for each interval are listed in Tables 4.3 and 4.4.
© 2007, Institute of Materials, Minerals and Mining
Influence of scale and rust on steel activation
4.4
43
Conclusions
On the basis of short-term measurements it was found that a scale layer on the steel surface increases the critical chloride concentration for steel activation compared with that for a bare surface by a factor of 3 to 6 times. The breakdown of the passive state on the bare steel surface proceeds in alkaline solution at a balanced concentration of atmospheric oxygen when the chloride concentration exceeds 1g L–1. Lower oxygen concentration leads to an increase in critical activation concentration of chlorides for both bare and scaled surfaces. No significant shift in critical chloride concentration was observed on increasing the pH of the pore solution from 12.5 to 13.3, corresponding to a six-fold increase in the OH– concentration. The results prove that the Cl–/OH– concentration ratio has no practical significance on the corrosion aggressivity of concrete towards steel reinforcement. In the region of low chloride concentrations, the pre-rusted surface showed the lowest corrosion resistance of all the surfaces studied, and the results confirm the negative effect of the rust layer formed by atmospheric exposure. Increasing the solution alkalinity did not explicitly lead to the spontaneous passivation of steel with a pre-rusted surface even in solutions with very low chloride content.
4.5
Acknowledgements
The authors aknowledge financial support of this research, which was the part of the Czech Grant Agency Project 103/02/0282 and MSM 223100002 project.
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 Corrosion Publication No 31, IOM Communications Ltd, 2000, p. 25. 9. P. Novák and R. Malá, Corrosion of Reinforcement in Concrete, Corrosion Mechanisms and Corrosion Protection, EFC Publication No 31, IOM Communications Ltd, 2000, p. 41.
© 2007, Institute of Materials, Minerals and Mining
5 The 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
5.1
Introduction
An important part of our infrastructure is based on reinforced concrete, with concrete taking the compressive load and the embedded steel the tensile load of the structures. The durability of this composite material is based on the excellent chemical stability of hydrated portland cement and the passivity of steel in the alkaline pore solution of concrete (pH 12.5–13.5) [1, 2]. Despite this decisive fact for the durability of steel-reinforced concrete structures, comparatively little is known about the surface chemistry of iron and iron alloys in model alkaline media [3, 4], in synthetic pore solutions containing sodium, calcium and potassium cations as well as sulphates [5, 6], or in concrete. A recent work of Joiret et al. [7] studied the dissolution and passivation of iron with electrochemical techniques and in-situ Raman spectroscopy, showing that with increasing potential a gradual oxidation of the oxide film is observed. At very low potentials magnetite (Fe3O4) is present, at more anodic potentials a-FeOOH and Fe2O3 are formed. These results are in agreement with short time experiments on sputtered iron immersed in alkaline solutions [3]. From a thermodynamic point of view iron oxides or oxyhydroxides are stable compounds at high pH, leading to the formation of a thin protective oxide 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) and the subsequent drop in pH or by the ingress of chloride ions, leading to localised corrosion attack [2, 8]. It is well known from the literature and practical experience that prolonged exposure of steel to alkaline media (ageing of the passive film) increases the critical chloride content for the initiation of chloride-induced localised corrosion, as has been reported for pore solution, mortar and concrete [9, 10]. Ageing of the passive film at the same time decreases the efficiency of cathodic oxygen reduction [10] and thus may reduce the rate of corrosion [10]. In conclusion, the thickness, composition 44 © 2007, Institute of Materials, Minerals and Mining
The surface of iron and Fe10Cr alloys in alkaline media
45
and electronic configuration of these thin iron oxide films (passive films) in alkaline media are of great importance for the corrosion behaviour of steel in concrete. In this work, the results of a combined electrochemical and XPS surface analytical study of the surface of iron and Fe10Cr alloys in alkaline media are reported. The results obtained are discussed with respect to the iron Fe(II)/Fe(III) ratio in the film and its relation to the open circuit potential and to the influence of the chromium addition on the stability of the passive film.
5.2
Experimental methods
5.2.1
Materials and sample preparation
Iron sheets (99.99%) were supplied by GoodFellow and the Fel0Cr alloys were prepared by the Institute of Physical Chemistry and Electrochemistry of the University of Düsseldorf. The samples – already fixed on the XPS sample holder – were ground in bi-distilled water using 200, 320, 500 and 1000-grit silicon carbide paper and in ethanol to a 1 mm finish with diamond paste (this state is called mechanically polished, m.p.). They were washed with ethanol, dried under a nitrogen stream and transferred under nitrogen to the spectrometer. Samples were analysed by XPS immediately after immersion in deaerated alkaline solutions of pH 13 and after exposure for prolonged time (up to 20 days) to air at 35% relative humidity and subsequent immersion in alkaline solution 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; pH ª 6.5) were used for mechanical polishing. The alkaline solutions of pH 13 were prepared from NaOH of analytical grade (Carlo Erba). Solution pH was monitored using a Metrohm 654 pH meter. Deaereated solutions were prepared by argon gas bubbling for at least four hours.
5.2.3
Electrochemical experiments
The electrochemical experiments have been carried out in a cylindrical electrochemical cell with an opening of diameter 1 cm (surface area 0.78 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 potentiostat under computer control. A saturated calomel electrode (SCE) with a double protection diaphragm was used for all experiments. © 2007, Institute of Materials, Minerals and Mining
46
5.2.4
Corrosion of reinforcement in concrete
X-ray photoelectron spectroscopy
XPS analyses were performed in an ESCALAB 200 spectrometer (Vacuum Generators Ltd., UK). The vacuum system consisted of a turbomolecular pump, fitted with a liquid nitrogen trap, and a titanium sublimation pump. The residual pressure in the spectrometer during the data acquisition was always lower than 5 ¥ 10–7 Pa. The X-ray source was AlKa (1486.6 eV), run at 20 mA and 15 kV. The spectra were obtained in the digital mode (VG Eclipse software on IBM 486). The electron analyser was operated in Fixed Analyser Transmission (FAT) mode with a pass energy of 20 eV Full Width at Half Maximum height (FWHM) Ag 3d5/2 = 1.1 eV. The analysed area was ca. 0.5 cm2. The instrument was calibrated using the inert-gas-ion-sputtercleaned reference materials SCAA90 of Cu, Ag and Au [11]. For calibration purposes the Au 4f7/2 line at 83.98 eV, the Cu 2p3/2 line at 932.67 eV, the Cu LMM signal at 334.94 eV and the Ag3d5/2 at 368.26 eV were taken. The spectra were resolved into their components after background subtraction according to Shirley and Sherwood [12]. The Gaussian /Lorentzian ratio and the FWHM were determined on standards and held constant, the peak energy and height were fitted using a least-squares algorithm.
5.2.5
Quantitative analysis
From the integrated peak intensities the thickness and composition of the surface film were determined with a three-layer model [13] taking into account the attenuation of the photoelectrons by the hydrocarbon contamination layer and the passive film.
5.3
Results
5.3.1
Electrochemical results
Open circuit potential (OCP) measurements have been carried out on iron and Fe10Cr samples in alkaline solutions at pH 13 immediately after mechanical polishing and after mechanical polishing and air oxidation for different time periods in a dessicator. In Fig. 5.1 examples of the OCP versus time curves for mechanically polished samples of Fe and Fe10Cr immersed in NaOH solution at pH 13.0 are shown for solutions saturated with O2 and deaerated with Ar. The OCP of the mechanically polished iron samples (Fig. 5.1a) immediately after immersion is very negative, –0.6 V (SCE), and it increases rapidly 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 mechanically polished samples after 12–14 h is more negative than those of iron both in deaerated and aerated solutions.
© 2007, Institute of Materials, Minerals and Mining
The surface of iron and Fe10Cr alloys in alkaline media
47
–0.1
OCP vs SCE (V)
A
O2
–0.2
In air for 17 days
–0.3
Ar
–0.4 –0.5 –0.6 –0.7 0
1.7 ¥ 104 3.4 ¥ 104 5.1 ¥ 104 6.8 ¥ 104 Time (s)
–0.1 B
OCP vs SCE (V)
–0.2 –0.3 O2
In air for 4 days
–0.4 Ar –0.5 –0.6 –0.7
0
1.7 ¥ 104 3.4 ¥ 104 5.1 ¥ 104 6.8 ¥ 104 Time (s)
5.1 Open circuit potential versus time for mechanically polished (—) and air exposed (- - -) iron (A) and Fe10Cr (B) samples immersed in NaOH solutions of pH 13.
A second series of experiments has been performed using iron and Fe10Cr samples mechanically polished and then stored in a dessicator for times ranging between 1 day and 20 days at 25∞C ± 0.5 ∞C and relative humidity of 35%. These air-exposed samples were then immersed in alkaline solution at pH 13.0 deaerated with argon gas and the OCP was measured for 20 h; the results 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 after immersion in the solutions is much more positive than for the mechanically polished samples. Iron samples exposed to the dry air start with an OCP at –0.4 V (SCE), at longer immersion times the potentials become identical with those for the mechanically polished samples. The air-exposed Fe10Cr samples start at an open circuit potential of ca. –0.3 V (SCE); in the first hours of exposure to the solution the OCP decreases but remains more positive throughout the experiment than the values for mechanically polished samples
© 2007, Institute of Materials, Minerals and Mining
48
Corrosion of reinforcement in concrete
in deaerated solution (Fig. 5.1b). Similar behaviour was found for air exposure times 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 mechanically polished, air oxidised in a controlled atmosphere and immersed in NaOH solution at pH 13. Mechanically polished surface The surface state after mechanically polishing in ethanol was chosen as a reproducible reference and starting state of the two alloys under study. As can 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 of the oxidised iron is found at 710.2 eV, suggesting a major contribution to this signal of Fe(II) in the surface film. Similar results were found for the Fe10Cr alloy (Fig. 5.2b). A more detailed analysis of the iron Fe2p3/2 and the oxygen O1s signal after background subtraction and curve fitting is shown in Fig. 5.3 (example of Fe10Cr alloy). The full widths at half maximum height (FWHM) and the position of the individual peaks in the iron Fe2p3/2 region used for curve fitting were held constant (see Table 5.1), only the peak height was allowed to vary. As can be seen from Fig. 5.3a, the iron Fe2p3/2 signal can be resolved into 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 (adsorbed water). Samples of mechanically polished iron showed similar results, the contribution of the Fe(III) oxyhydroxide is higher than in the mechanically polished Fe10Cr alloy. Quantitative analysis of the oxide film on the Fe10Cr alloy after mechanical polishing indicated a composition of 16% chromium oxide (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: the signal of metallic iron became much less pronounced, indicating a higher film thickness, and the maximum of the iron oxide peak was shifted to
© 2007, Institute of Materials, Minerals and Mining
The surface of iron and Fe10Cr alloys in alkaline media Fe
Intensity (a.u.)
Mech polished
m.p. and 20 h pH 13
740
735
730
725 720 715 710 Binding energy (eV)
705
700
Intensity (a.u.)
Fe10Cr
740
Mech. polished 20 h at pH 13
735
730
725 720 715 710 Binding energy (eV)
705
700
5.2 XPS spectra of iron Fe2p recorded on iron (upper) and Fe10Cr samples (lower) after mechanical polishing and after subsequent immersion 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 (eV)
Binding Energy (eV)
FWHM (eV)
Comments
Iron metal Fe(II) oxide Fe(III) oxide Fe(III) oxyhydroxide
779.9 ± 0.1 777.80 776.30 774.50
706.7 708.8 710.3 712.1
1.3 2.6 2.6 3.0
exp 0.63 G/L* = 0.45 G/L = 0.45 G/L = 0.45
*Gaussian/Lorentzian
© 2007, Institute of Materials, Minerals and Mining
49
50
Corrosion of reinforcement in concrete
Intensity (a.u.)
Fe2p3/2/Fe10Cr mp
716
714
712 710 708 Binding energy (eV)
706
704
Intensity (a.u.)
O1s/Fe10Cr mp
538
536
534 532 530 Binding energy (eV)
528
526
5.3 Fe2p3/2 and O1s spectra of mechanically polished Fe10Cr alloy after background subtraction and curve fitting (for parameters see Table 5.1). Table 5.2 Thickness and composition of the oxide film and the composition of the alloy beneath the film of mechanically polished Fe10Cr alloys immersed for different times in deaerated alkaline solutions of pH 13 Time
Mechanically polished 16 h 25 h 3 days
Thickness (nm)
Oxide film
Alloy beneath the film
% Fe
% Cr
% Fe
% Cr
2.0
84
16
90
10
2.6 2.6 2.7
75 66 63
25 34 37
91 93 94
9 7 6
higher binding energies (Fig. 5.4). The more detailed analysis after curve fitting of the Fe2p3/2 spectrum of the mechanically polished Fe10Cr alloy following 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
© 2007, Institute of Materials, Minerals and Mining
The surface of iron and Fe10Cr alloys in alkaline media
51
Intensity (a.u.)
Fe2p3/2/Fe10Cr mp 20 h pH 13.0
716
714
712 710 708 Binding energy (eV)
706
704
Intensity (a.u.)
O1s/Fe10Cr mp 20 h pH 13.0
538
536
534 532 530 Binding energy (eV)
528
526
5.4 Fe2p3/2 and O1s spectra of Fe10Cr alloy, mechanically polished and immersed for 20 h at pH 13 after background subtraction and curve fitting (for parameters see Table 5.1).
that the major contribution was then due to the component assigned to Fe(III) oxide. The same occurred on mechanically polished pure iron after immersion at pH 13, but the decrease of the Fe(II) component was less pronounced, the Fe(II) component remained at ca. 20%. The oxygen O1s spectra of iron and Fe10Cr 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 film together with the thickness of the oxide layer were evaluated using the peak intensities of the different components and the three-layer model successfully employed in other samples [13]. The results of the quantitative analysis of the Fe10Cr alloy (Table 5.2) indicate an increase in the chromium content in the oxide film from 10 to 25 % after 16 h of immersion. Prolonged immersion of 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 alloy beneath the oxide film (Table 5.2).
© 2007, Institute of Materials, Minerals and Mining
52
Corrosion of reinforcement in concrete
Mechanically polished and exposed to air (35% rh) A series of experiments were performed to study the influence of air exposure on the surface film composition. The Fe2p3/2 and O1s spectra were resolved into their components, as described above, and the relative intensities (percentage) of the oxidised iron signals Fe(II), Fe(III) and FeOOH were determined. As can be seen from Fig. 5.5, the main effect of exposure to dry air is a marked decrease in the Fe(II) component and a corresponding increase in the intensity of the Fe(III) and FeOOH signals (Fig. 5.5a). This film transformation was complete only after one day, and prolonged exposure to dry air for up to 20 days did not change the composition of the surface film. 80
Intensity ratio %
70
Fe(II) % Fe(III) % FeOOH %
60 50 40 30 20 10 0
0
1
80 70
Intensity ratio %
60
5 10 15 Time of air exposure (d) (a)
20
MOH % MO % H2O %
50 40 30 20 10 0 0
1
5 10 15 Time of air exposure (d) (b)
20
5.5 Effect of the exposure time to air on the composition of the air formed film on pure iron. (a) iron compounds, (b) oxygen compounds.
© 2007, Institute of Materials, Minerals and Mining
The surface of iron and Fe10Cr alloys in alkaline media
53
4.5 Air exposed Immersed Air exposed Immersed
Oxide film thickness (nm)
4
Fe10Cr
3.5
3
2.5 Fe 2
1.5
0
5
10 15 Time of air exposure (d)
20
25
5.6 Thickness of the oxide film of mechanically polished iron and Fe10Cr alloys after air exposure and air exposure with subsequent immersion for 20 h in solution of pH 13.
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 times of air exposure (Fig. 5.5b) shows a more gradual change: the percentage of the component at the highest binding energy (ca. 533 eV) decreases with time 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 is worth pointing out that the Fe(II) intensities were lower from the beginning of 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 increased from 2.0 to 4.0 nm after exposure to air for 20 days (Fig. 5.6). The thin oxide film formed after mechanical polishing was found to contain 16% of oxidised chromium on average. After air exposure from 1 to 20 days, it remained constant 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 immersed subsequently for 20 hours in alkaline solution of pH 13. The Fe2p3/2 and Ols spectra were resolved into their components, as described above, and the relative intensities of the oxidised iron signals, Fe(II), Fe(III) and FeOOH, were determined. Upon immersion of the air-exposed Fe10Cr alloys part of the air-formed oxide film dissolved (Fig. 5.6) and a film thickness of 3.0 ± 0.2 nm resulted after 20 h of immersion. Parallel to the decrease of the
© 2007, Institute of Materials, Minerals and Mining
54
Corrosion of reinforcement in concrete
Table 5.3 Thickness and composition of the oxide film and composition of the alloy beneath the film on mechanically polished Fe10Cr alloys air exposed for different times and then immersed for 20 h in deaerated alkaline solutions of pH 13 Time
Thickness (nm)
Air exposed 1 day 5 days 10 days 15 days 20 days
see Fig. 5.6 2.8 3.2 3.3 3.2 3.1
Oxide film
Alloy beneath the film
% Fe
% Cr
% Fe
% Cr
90 ± 1 88 88 88 84 79
10 ± 1 12 12 12 16 21
92 ± 1 92 92 93 92 90
8 ±1 8 8 7 8 10
film thickness the content of oxidised chromium in the film increased from 10 to 20% (Table 5.3). The greatest increase in chromium content was found for the samples with long exposure to air and correspondingly high film thickness (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 the Fe(III) content in the film. This is in agreement with results that have been reported for sputtered iron surfaces exposed to oxygen or air [14]. It is interesting to note that only 24 h of air exposure are sufficient to induce this oxidation reaction, a further air exposure for up to 20 days does not significantly alter the Fe(II)/Fe(III) ratio in the film both for iron (Fig. 5.5) and Fe10Cr alloy. Changes are found instead in the film thickness (Fig. 5.6): the film thickness of the oxide film on pure iron increased from 1.9 ± 0.1 nm of the mechanically polished sample to 2.4 ± 0.1 nm after 20 days. On the Fe10Cr alloy, the increase in film thickness was more pronounced (Fig. 5.6); the oxide film composition was close to the nominal (10 ± 1% Cr) composition of the alloy. Thus, the presence of chromium in the alloy accelerates the formation of an air-formed oxide film. The plot of film thickness vs. log t (t = time of air 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 is mainly determined by the cathodic oxygen reduction reaction and, thus, by © 2007, Institute of Materials, Minerals and Mining
The surface of iron and Fe10Cr alloys in alkaline media
55
the pH of the pore solution and the availability of oxygen; corrosion potentials in the range of –0.15 to 0 V (SCE) are observed for aerated structures [2]. In this laboratory study, working with well-defined mechanically polished surfaces, the influence of the surface state before immersion (mechanically polished, air exposure) and of alloying with 10% chromium has been investigated. As can be noted from Fig. 5.1, in the presence of oxygen more positive potentials are found both for iron and the Fe10Cr alloy, in agreement with the results for steel in concrete [2, 7]. Whereas freshly polished samples exhibit very negative initial corrosion potentials of around –0.6 V (SCE), samples exposed to air for several days show more positive initial corrosion potentials in deaerated solution due to the presence of the air-formed oxide film (Fig. 5.1). This behaviour can be rationalised further by combining the electrochemical data with surface analytical information. As established by Haupt et al. [3] in a laboratory study on sputtered (oxide-free) iron exposed to alkaline solutions at pH 13, the Fe(II)/Fe(III) ratio in the oxide film depends on the electrochemical potential 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 at more positive potentials, around –0.1 V, the films contain only about 10% of Fe(II). In this work, the oxide films on the surface of iron and Fe10Cr alloys were formed naturally at the open circuit potential (without anodic polarisation) in alkaline solutions. By plotting the percentage of Fe(II) and Fe(III) oxides in the surface film, as determined by XPS analysis in this work, versus the final value of the open circuit potential (Fig. 5.1), a clear trend can be observed (Fig. 5.8): with increasing open circuit potential the percentage of
Amount Fe(II) and Fe(III) (%)
100
80
Fe (II) Fe (III)
60
40
20
0 –1
–0.8
–0.6 –0.4 Potential [V (SCE)]
–0.2
0
5.7 Change of Fe(II) and Fe(III) content in the passive film on iron exposed to pH 13 solution with the potential of film formation (according to reference 3).
© 2007, Institute of Materials, Minerals and Mining
56
Corrosion of reinforcement in concrete
Amount Fe(II) and Fe(III) (%)
100
80 Fe (III) 60
40 Fe (II) 20
0 –0.45
–0.4
–0.35
–0.3 –0.25 –0.2 Potential [V (SCE)]
–0.15
–0.1
5.8 Open circuit potential and Fe(II) 䊊䊉 and Fe(III)ox 䊏䊐 content in the surface films after exposure of iron (䊊䊐), Fe 10Cr alloys (䊉䊏) and bulk oxides (▼ ▲) [14] to alkaline solution at pH 13.
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], with results from in situ Raman spectroscopy [7], and with bulk oxides immersed in alkaline solutions at the same pH. The Fe10Cr alloy follows the same trend as pure iron, showing a slightly lower Fe(II) content at the same open circuit potential. As additional confirmation, freshly polished iron samples showed very negative potentials upon immersion in alkaline solution of pH 13 (Fig. 5.1). The percentage of Fe(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 a film formation potential of –0.6 V (SCE). Thus, the initial surface film rich in Fe(II) becomes oxidised in alkaline solutions and the open circuit potential increases in agreement with an increase in the Fe(III) content. The air-formed films 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 that immersion of mechanically polished samples in alkaline solution for 20 h leads to an increase in oxide film thickness of 0.2 nm for pure iron and of 0.5 nm for the Fe10Cr alloy (Fig. 5.6). Both materials are passive and the film growth is associated with the increase in open circuit potential (Fig. 5.1). This has been confirmed by recent experiments combining
© 2007, Institute of Materials, Minerals and Mining
The surface of iron and Fe10Cr alloys in alkaline media
57
potentiostatic passivation and quartz crystal microbalance (QCM) measurements [15] on mechanically polished iron in alkaline solutions of pH 13: in the potential region from –0.4 to 0 V a slight mass increase is observed during 1 h of polarisation. On the Fe25Cr alloy the mass increase was more pronounced [15]. The growth in film thickness observed in this work and the mass increase associated with a shift in the electrode potential to 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 Raman spectroscopy [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 the anodic direction and to film dissolution in the cathodic direction. In the oxidation direction the reaction is accompanied by the uptake of one oxygen atom into the lattice. On the air oxidised surfaces, a reduction of the oxide film thickness was observed 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. During this process the open circuit potential remained constant or increased only slightly (Fig. 5.1), thus the constant film thickness of the pure iron is consistent with the film transformation reaction discussed above. In the case of the Fe10Cr alloy, additional reactions must occur that are not primarily associated with 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 increase in the content of oxidised chromium in the film, thus the iron component of the oxide film is preferentially dissolved. Summarising the results from this XPS surface analytical study, together with results from the literature [3, 7, 21], a quite consistent picture of the surface of iron in alkaline media is obtained (Fig. 5.9): at low potentials both by 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 (either imposed potentiostatically or from natural immersion at the open circuit potential) the oxide film thickens and gradually transforms to a-FeOOH and Fe2O3, 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, electrochemical restoration techniques influence the potential of steel in concrete. During electrochemical chloride removal or electrochemical realkalisation, the steel is polarised to very negative potentials. For example, after switching off the current in a realkalisation experiment potentials of ca. –1.1 to –0.8 V (SCE)
© 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete Amount Fe(II) and Fe(III)(%)
58
100 Fe(III)
80 60
This work
40 20 Fe(II) 0 –0.9 –0.8
–0.7
–0.6 –0.5 –0.4 –0.3 Potential [V(SCE)]
–0.2
–0.1
Fe2O3 a-FeOOH Fe3O4
Initial
24 h
Potential
5.9 Schematic summary of the electrochemical and surface analytical results of iron and Fe10Cr alloy in alkaline solutions. Top: Fe(II) / Fe(III) ratio of this work and [3], middle: iron oxides identified by Raman spectroscopy [21], bottom: polarisation curve with initial and final OCP of mechanically polished samples (this work).
have been measured [16]. Using half-cell potential measurements as a diagnostic technique [17] for the success of electrochemical restoration techniques, these negative potentials could be misinterpreted as a corrosion state, rather than repassivated rebars [18]. The results of the surface analysis undertaken in this study allow this interpretation to be ruled out: the very negative potentials during the electrochemical treatment change the composition of the oxide film to mainly Fe(II) and the long time required (in practice several weeks [16]) to achieve a stable, more positive potential for steel in concrete is due to the reoxidation to Fe(III). The question of pre-rusted steel embedded in mortar or concrete has not been addressed in this work as all the iron and Fe10Cr surfaces were prepared by mechanical polishing or by exposure to dry air. In the case of a localised corrosion occurring on the otherwise passive steel surface, the reduction of the Fe(III) in Fe2O3 to Fe(II) in Fe3O4 might be an additional reduction reaction in parallel with oxygen reduction, as proposed for atmospheric corrosion [19], that might increase the corrosion rate or sustain a corrosion reaction in the absence of oxygen. The amount of cathodic current of this
© 2007, Institute of Materials, Minerals and Mining
The surface of iron and Fe10Cr alloys in alkaline media
59
origin is determined by the quantity of Fe(III) ions present in the oxide film and, thus, by the thickness of the oxide or rust layer. A rough estimation shows 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 be produced (without taking account of possible kinetic hindrance). Further studies are required in this respect. When electrochemical measurements (e.g. polarisation resistance, impedance spectroscopy and pulse techniques) are applied, usually polarising the sample by ± 20 mV at the open circuit potential, the valence change in the iron oxide film (oxidation, reduction) may produce a current flow due to the Fe(II)/Fe(III) redox couple. This redox couple current could lead to a low frequency time constant in impedance spectra (pseudo-capacitance) and influence the corrosion rate values obtained [20, 21], but the magnitude of this charging current is not yet known. In the light of the surface analytical results obtained in this work, a small potential change of ± 20 mV would not be expected to affect the Fe(II)/Fe(III) ratio (Fig. 5.8). Anodic polarisation (polarisation curve, cyclic voltammetry) leads to an additional oxidation of the film, film thickening and the presence of only Fe(III) can be expected. During the reverse cycle of the voltammetry, an additional contribution in the cathodic curves will be measured that can be attributed to the film dissolution and reduction reaction. The corrosion resistance of pure iron and Fe10Cr alloys has not been studied in this work. The results of this surface analytical study nevertheless allow some of the results in the literature to be interpreted. Ageing of the passive film of pure iron or mild steel has been found to improve the resistance to pitting attack by increasing the pitting potential [10] and to decrease the efficiency of the cathodic reduction of oxygen [10]. Ageing has been shown to influence markedly the pitting potential of steel in alkaline solutions [10] and might also influence the ‘critical chloride content’. This may be explained by the film growth asssociated with a decrease in the Fe(II) content in the passive film, leading to an electronically less defective passive film. The improved corrosion resistance of chromium steels in concrete, and thus their much higher critical chloride content for depassivation [22], may be explained by the marked enrichment of the passive film in oxidised chromium found in this work.
5.5
Conclusions
The results have been reported of an electrochemical and XPS surface analytical study on the model systems iron and Fe10Cr alloy in NaOH solution of pH 13, simulating the concrete pore environment. As a starting point, iron and Fe10Cr samples were studied after mechanical polishing in ethanol. The XPS results show that a thin oxyhydroxide film is present, the oxidation state
© 2007, Institute of Materials, Minerals and Mining
60
Corrosion of reinforcement in concrete
of iron being mainly Fe(II). These mechanically polished samples were subsequently subjected to conditions simulating natural exposure. Immediately after immersion in alkaline solutions of pH 13, very negative OCP values were recorded. Continuing immersion resulted in asymptotically increasing, more positive OCP values. The OCP values depend on alloy composition and on the degree of aeration (presence or absence of oxygen) of the solution. XPS surface analysis shows that the surface film after immersion is mainly formed of Fe(III) oxyhydroxide. After prolonged immersion of the Fe10Cr alloy up to 35–40% of oxidised chromium is present in the passive film. Samples exposed for a prolonged time (up to 20 days) to air (35% relative humidity) showed progressive oxidation of the surface film and a gradual change of Fe(II) to Fe(III). The chromium oxide content of the Fe10Cr alloy remained constant at the nominal composition 10 ± 1%. After immersion of these oxidised samples in alkaline solutions, the initial OCP was much more positive than that of mechanically polished samples. Immersion of air oxidised samples in alkaline solutions of pH 13 led to a transformation of the oxide film. Especially for the Fe10Cr alloy, the oxide film thickness decreased and after 20 h of immersion a chromium oxide content of 20% was reached. The results of this study allow the electrochemical behaviour of iron and Fe10Cr alloy in alkaline solutions to be correlated with the surface chemistry of the oxide films. It can be concluded that the value of the open circuit potential (OCP) in alkaline solutions is strongly related to the percentage of Fe(II) and Fe(III) in the film, more positive OCP values corresponding to a higher Fe(III) content in the film.
5.6
Acknowledgements
The authors are pleased to acknowledge the financial contribution of Regione Autonoma della Sardegna (RAS), of the Italian National Research Council (CNR) and of the Italian Ministry of University and Scientific and Technological Research. The Institute of Physical Chemistry and Electrochemistry of the University of Düsseldorf is acknowledged for having supplieding the Fe10Cr alloy.
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.
© 2007, Institute of Materials, Minerals and Mining
The surface of iron and Fe10Cr alloys in alkaline media
61
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, ‘Chloride penetration – Non destructive determination of the free chloride content in mortar and 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-SIMS analyses 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. VIII Symposium Passivity of Metals and Semiconductors, Eds. M. B. Ives, J. L. Luo, and J. R. Rodda, Electrochem. Proc, 99–42, The Electrochemical Society, Pennington NY, 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 of Rebar Corrosion’, Corrosion of Reinforcement in Concrete: Corrosion Mechanisms and 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 in alkaline solutions’, Corrosion of Reinforcement in Concrete: Corrosion Mechanisms and 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 reinforcement in concrete – monitoring, prevention and rehabilitation, Eds. J. Mietz, B. Elsener and 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 in concrete – 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 of oxides 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’, EFC Publication No. 18, The Institute of Materials, London, 1996.
© 2007, Institute of Materials, Minerals and Mining
6 Risk of galvanic corrosion induced by CFRP strengthening in reinforced concrete L. B E R T O L I N I, M. G A S T A L D I and M. P. P E D E F E R R I, Politecnico di Milano, Italy
6.1
Introduction
Galvanic phenomena often occur in reinforced concrete structures when corroding steel bars in chloride-contaminated concrete are connected with passive steel bars embedded in concrete of lower chloride content. The passive reinforcement is the cathode of this macrocouple, while the active reinforcement is anodically polarised and experiences an increase in the corrosion rate.1,2 Galvanic coupling can also be induced when the corroding structure is repaired by replacing damaged concrete with alkaline and chloride free mortar. If the concrete in the area surrounding the patch repair contains a significant amount of chloride, a macrocouple takes place between repassivated bars in contact with the repair mortar and depassivated bars in the original chloride contaminated concrete3. Consequently, early corrosion damage can occur in the area surrounding the patch repair. For this reason, in the rehabilitation of structures damaged by chloride-induced corrosion, it is recommended that even the mechanically sound concrete should be removed, if it contains significant amounts of chlorides.3,4 The risk of galvanic corrosion can also arise when new materials are used for repair. For instance, in the past great concern was expressed with regard to the use of stainless-steel bars. It has now been clearly shown that stainlesssteel bars do not increase the risk of galvanic coupling in reinforced concrete structures. In fact, because of the high overvoltage of the cathodic reaction of oxygen evolution, stainless steel is a poor cathode compared with normal carbon steel.5,6 Even non-metallic materials can provide sites for the cathodic reaction to take place and, thus, can generate macrocouples with corroding carbon steel. This is the case for composite materials with carbon fibres (carbon fibre reinforced plastics, CFRP). It has been shown that reinforcing bars made of CFRP can induce a macrocouple on steel reinforcement in chloridecontaminated concrete.7 For several reasons, mainly related to concerns about their long-term performance and durability, CFRP reinforcing or prestressing 62 © 2007, Institute of Materials, Minerals and Mining
Risk of galvanic corrosion induced by CFRP strengthening
63
bars are rarely used; on the contrary, CFRP is often used for structural strengthening (e.g. for seismic retrofitting) and rehabilitation of deteriorated structures.8,9 High-strength carbon fibres are bonded to the concrete surface in the form of laminates or sheets; epoxy adhesives are normally used for bonding. The low weight of the composite materials and high design flexibility make this technology quite attractive. Nevertheless, the durability of this complex system, consisting of the combination of the composite, the adhesive, and the reinforced concrete, has to be studied. Furthermore, when the composite materials are applied to corrosion damaged structures, it should be clear that they are only aimed at strengthening the structure and a proper repair should be carried out before application of the composite to stop ongoing corrosion of the reinforcing steel. Indeed, if carbon fibre composites are used, even a harmful effect of the composite can be hypothesised, owing to possible galvanic effects. In this work the consequences of using commercial CFRP are investigated with regard to corrosion of reinforcement in structures contaminated by chlorides. The effects of galvanic coupling produced by these materials were studied and compared with those normally produced by the contact between active reinforcement (that are corroding) and passive bars (that are not corroding) made both of common carbon steel and stainless steel.
6.2
Experimental procedure
Tests were carried out using commercial laminates (Sika Carbodur) and sheets (Sikawrap HEX 230C) of unidirectional carbon fibre. To investigate the potential risks of galvanic coupling induced by the composite, a series of tests were carried out on the specimen shown in Fig. 6.1a. In one part of the specimen, a carbon steel bar (10 mm in diameter) was embedded in concrete containing 3% of chlorides by mass of cement; a laminate of CFRP (15 mm wide and 1.4 mm thick) was embedded in chloride-free concrete in the adjacent part. Concrete was mixed with a water-to-cement ratio of 0.55, 350 kg 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 was electrically connected to the laminate. Macrocell current was evaluated through the ohmic drop on a 100 W shunt; potentials of steel and CFRP laminate were measured versus embedded reference electrodes made of mixed metal oxide (MMO) activated titanium. Tests were carried out in a climatic chamber; humidity (95% and 65% relative humidity (rh) and temperature (20, 40 and 60 ∞C) were changed in steps of at least 15 days. Comparison tests were also carried 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 and 6.1c to evaluate the risk of galvanic coupling when the composite is externally
© 2007, Institute of Materials, Minerals and Mining
64
CFRP laminate
Concrete chloride free
20
CFRP sheet
10
20
5
5
20
Carbon steel
R
R
R
10
Carbon steel
R CFRP laminate or graphite, AISI 316L or carbon steel bars (a)
CFRP sheet
(b)
(c)
6.1 Schematic representation of the specimens (dimensions in cm). A carbon steel bar was embedded in concrete with 3% chlorides by cement weight and was coupled with: (a) laminate embedded in chloride free concrete (other specimens had a carbon steel, stainless steel, or graphite bars in place of the CFRP laminate), (b) laminate and sheet bonded to the two opposite faces, (c) CFRP sheet wrapping. © 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete
Concrete with 3% of chloride
Risk of galvanic corrosion induced by CFRP strengthening
65
bonded. In the specimens of Fig. 6.1b a corroding bar of carbon steel, embedded in concrete with 3% chloride by mass of cement, was simultaneously coupled with a laminate and a sheet bonded to the surface of the concrete. Both the laminate and the sheet had a width of 25 mm. Two specific epoxy systems were used to apply the laminate and the sheet, according to the manufacturer’s instructions. Macrocell tests were carried out in the same environment as for specimens with embedded laminate. Subsequently, a 1.5-cm-thick layer of mortar was applied to coat both the laminate and the sheet, and changes in the macrocouple current density were monitored. Galvanic coupling tests were also carried out on the specimen shown in Fig. 6.1c, which was wrapped with the CFRP sheet for a length of 10 cm. The cathodic behaviour of the materials in alkaline environments was studied 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 of 20 mV min–1 from the free corrosion potential to –1.2 V vs. SCE.
6.3
Results and discussion
6.3.1
CFRP laminate embedded in concrete
Figure 6.2 shows the results of galvanic coupling between a corroding steel bar in concrete contaminated with 3% of chloride by mass of cement and different cathodic materials. Before coupling, at 20 ∞C and 95% relative humidity, the corroding bars of carbon steel had free corrosion potentials of around –500 mV vs. MMO and corrosion rates higher than 20 mA m–2 (evaluated with polarisation resistance measurements). Passive carbon steel in chloride-free concrete had a free corrosion potential of about –50 mV vs. MMO; potential measured on the stainless-steel bar, the CFRP laminate and the graphite bar was about –100 mV vs. MMO. Following the electrical connection, a macrocouple current flowed from the corroding carbon steel bar in 3% Cl– concrete (anode) to the passive bar or the CFRP laminate (cathode). Changes in potential and macrocouple current density occurred soon after the coupling, while rather stable values were measured during the following 15 days at rh of 95–98%. The current density decreased from initial values higher than 100 mA m–2 to lower and stable values within a few hours. Changes in potential produced by the macrocouple were 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 coupling between active carbon steel and passive carbon steel reinforcement was about 10 mA m–2 during the period of exposure to 20 ∞C and 95% rh. With a 316L stainless-steel cathode (Fig. 6.2b) the macrocouple current was significantly © 2007, Institute of Materials, Minerals and Mining
66
Corrosion of reinforcement in concrete
lower (2 mA m–2). Conversely the CFRP laminate (Fig. 6.2c) led to a higher current density of 35–40 mA m–2. Therefore, the coupling with the composite material seems to be more dangerous than coupling with passive carbon steel. Comparing results obtained with the laminate with those of a graphite bar of equal surface, a basic similarity can be observed (Fig. 6.2c and d). The currents always maintained similar values throughout the entire time of exposure. This can indicate that the epoxy matrix of the CFRP laminate does not insulate the fibres from the concrete and allows the cathodic process to develop in the same way that an element of graphite of similar surface does.
0
Coupling
–200 –300
10 –400 –500
1
–600 95% R.H. 0.1
65% R.H. –700
0
10
20 30 Time (d) (a)
40
1000
Current density (mA m–2)
Potential (mV vs MMO)
–100 100
0 –100
100
–200 –300
10 –400 –500
1
–600 95% R.H. 0.1
Potential (mV vs MMO)
Current density (mA m–2)
1000
65% R.H. –700
0
10
20 30 Time (d) (b)
40
6.2 Macrocouple tests between corroding bars of carbon steel in concrete with 3% of chlorides and: (a) passive carbon steel in chloride free concrete, (b) 316L stainless steel, (c) CFRP laminate, (d) graphite. Potential of corroding carbon steel (D), potential of cathode material (䊊), and macrocouple current (—).
© 2007, Institute of Materials, Minerals and Mining
Risk of galvanic corrosion induced by CFRP strengthening 0
–200 –300
10 –400 –500
1
–600 95% R.H.
Potential (mV vs MMO)
–100 100
65% R.H. –700
0.1 0
10
20 Time (d) (c)
40
40
1000
0 –100
100
–200 –300
10 –400 –500 1 –600 95% R.H.
Potential (mV vs MMO)
Current density (mA m–2)
1000
Current density (mA m–2)
67
65% R.H.
0.1
–700 0
10
20 40 Time (d) (d)
40
6.2 Continued
Differences in the macrocouple current density observed in Fig. 6.2 on specimens of the same geometry exposed to the same environment can only be explained by a different cathodic behaviour of the materials coupled with the corroding carbon steel. Figure 6.3 shows results of potentiodynamic polarization tests in Ca(OH)2 saturated solution. It can be observed that the cathodic polarization curve of stainless steel is shifted to more negative potentials with respect to the curve of carbon steel, while cathodic polarization curves of CFRP laminate and graphite are shifted to more positive potentials. This confirms that stainless steel has higher overvoltage for the cathodic reaction of oxygen reduction than carbon steel and, thus, when it is cathodically polarized, for a given potential it can supply a lower cathodic current density. However CFRP laminate and graphite have lower overvoltages, and the cathodic current density for a given potential is higher. Indeed, CFRP laminate is a quite effective cathodic material and
© 2007, Institute of Materials, Minerals and Mining
68
Corrosion of reinforcement in concrete 0 Graphite
Potential (mV vs SCE)
–100 –200 –300 CFRP laminate
–400 Carbon steel –500 –600
316L stainless steel
–700 –800 0.01
0.1
1 10 Current density (mA m–2)
100
1000
6.3 Cathodic polarization curves in Ca(OH)2 saturated solution for the materials studied in this work.
can produce large macrocell currents when it is cathodically polarized by coupling with corroding steel. Obviously, the macrocell current is influenced by the resistivity of concrete and, thus, by its humidity content. Figure 6.2 shows that when the relative humidity was reduced from 95 to 65%, the macrocouple current density progressively decreased and potentials increased in time. However, the difference in the macrocouple behaviour for the materials coupled with active carbon 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 these composite materials are utilised to strengthen existing structures damaged by corrosion, they are usually bonded to the surface of the concrete structure. However, even in this case they could come in contact with the reinforcement and galvanic coupling may occur. Figure 6.4 shows the macrocell current densities measured by coupling an active bar of carbon steel in concrete with 3% chlorides with the CFRP laminate 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 macrocouple current density at 20 ∞C had values of about 40 mA m–2; a slight increase was
© 2007, Institute of Materials, Minerals and Mining
Risk of galvanic corrosion induced by CFRP strengthening
69
observed on increasing the temperature to 40 ∞C. At the end of exposure at 40 ∞C, the concrete cover cracked due to corrosion of the carbon steel reinforcement. This crack caused the current density to decrease, even during the 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 a macrocell 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 it was even of the order of 0.1 mA m–2. The externally bonded sheet also generated a macrocouple current of the order of 0.1 mA m–2 (Fig. 6.4c and d).
Current density (mA m–2)
100
10 Embedded laminate 1
0.1 20 ∞C 40 ∞C 60 ∞C
0.01
0.001 0
5
10
15
Time (d) (a) 100
Current density (mA m–2)
Externally bonded laminate 10
1
0.1
0.01
0.001 0
5
10
15
Time (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 of carbon steel in concrete with 3% of chlorides is coupled with: (a) CFRP laminate embedded in chloride free concrete, (b) CFRP laminate externally bonded, (c) CFRP sheet externally bonded, (d) CFRP sheet wrapped around the concrete specimen.
© 2007, Institute of Materials, Minerals and Mining
70
Corrosion of reinforcement in concrete 100
Current density (mA m–2)
Externally bonded sheet 10
1
0.1
0.01
0.001 0
5
10
15
Time (d) (c)
Current density (mA m–2)
100
Externally bonded sheet (wrapped specimen)
10
1
0.1
0.01
0.001 0
5
10
15
Time (d) (d)
6.4 Continued
Increasing in temperature to 40 and 60 ∞C led to ambiguous results in both the laminate and the sheet bonded parallel to the steel bar; in fact, often a decrease in 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 by corrosion of the steel bar in concrete with 3% of chloride by mass of cement during exposure at high temperature. The effect of temperature was clearer in the specimen wrapped with the CFRP sheet, since the geometry of this specimen allowed the macrocouple current to flow even after cracking of the concrete cover. Furthermore, the CFRP wrapping could also limit the crack width. Figure 6.5 plots the average values of the macrocouple current density during different tests as a function of temperature. The laminate embedded
© 2007, Institute of Materials, Minerals and Mining
Risk of galvanic corrosion induced by CFRP strengthening
71
in concrete (Fig. 6.5a) led to a current density higher than 10 mA m–2 even after cracking of the concrete specimens. The current density generated by the CFRP laminate bonded to the surface of the concrete was two orders of magnitude lower, as for the bonded sheet (Fig. 6.5b). In spite of several drops in the macrocouple current density due to cracking of the specimen, it can be observed that, in general, the current increased as the temperature increased from 20 to 60 ∞C. Nevertheless, the current was always negligible for the composite bonded to the surface of the concrete; only at 60 ∞C on the wrapped specimen did the current density reach values of 1 mA m–2.
Current density (mA m–2)
100
10
1
0.1
0.01
0.001 0
20
40 60 Temperature (∞C) (a)
80
0
20
40 60 Temperature (∞C) (b)
80
Current density (mA m–2)
100
10
1
0.1
0.01
0.001
6.5 Effect of temperature on the macrocouple current density (steady values) generated at 95% rh by: (a) CFRP laminate embedded in concrete (∑) and externally bonded (䊊); (b) CFRP sheet bonded on a side ( ) or wrapped around ( ) the concrete specimens. Grey symbols show results obtained after cracking of the concrete cover.
© 2007, Institute of Materials, Minerals and Mining
72
Corrosion of reinforcement in concrete
These results show that when CFRP laminates or sheets are bonded to a concrete surface, even when they accidentally come into contact with corroding steel, they should produce negligible macrocell currents. The epoxy used to bond the composite to the concrete surface greatly reduces electrolytic continuity with the concrete, even though it does not insulate them completely. However the electrolytic continuity can be restored if a coating of cement mortar (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 mortar and, thus, with the underlying concrete. Figure 6.6 shows the effect produced by the application of a layer of 1 cm of mortar on specimens of the type shown in Fig. 6.1b. The mortar layer was first applied on the side of the specimens where the CFRP laminate was bonded (Fig. 6.6a) and, after 5 days, 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 in the macrocell current, especially in the laminate, where stable values of 2–4 mA m–2 were reached (Fig. 6.6a). Nevertheless, values never reached those 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 current density approached values of 0.2–0.3 mA m–2; in this case, the fibres were embedded in epoxy resin and were thus insulated, even from contact with a layer of mortar applied afterwards.
6.4
Conclusions
Composite materials with carbon fibres are potentially able to generate galvanic coupling with corroding steel in concrete. CFRP, like graphite, have very low overvoltages for the cathodic reaction of oxygen reduction. A CFRP laminate embedded in concrete and coupled with a steel bar in chloridecontaminated concrete could induce a macrocouple current several times higher than with passive carbon steel and more than one order of magnitude higher 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 at temperatures higher than 20 ∞C. Therefore, when an existing structure is repaired by applying laminates or sheets on the surface of the concrete with an epoxy adhesive, the effects of galvanic coupling are negligible even if the CFRP is accidentally in contact with the corroding reinforcement. However, when a cement plaster is applied, electrolytic continuity with the concrete may 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
© 2007, Institute of Materials, Minerals and Mining
Risk of galvanic corrosion induced by CFRP strengthening
73
Current density (mA m–2)
10
1
0.1 Application of mortar layer
0.01 0
5
10 15 Temperature (∞C) (a)
20
Current density (mA m–2)
10 Application of mortar layer 1
0.1
0.01 0
5
10 15 Temperature (∞C) (b)
20
6.6 Changes in the macrocouple current density in two replicate specimens, exposed at 95% rh and 20 ∞C, on which a mortar layer was applied on the externally bonded CFRP laminate (a) and sheet (b).
corroding reinforcement. Even in the absence of galvanic coupling, propagation of corrosion of reinforcement in chloride-contaminated concrete can lead to cracking of the concrete cover and to serious consequences on the effectiveness of the strengthening.
6.5
Acknowledgements
The authors are grateful to Sika Italia SpA for supplying the composite materials.
© 2007, Institute of Materials, Minerals and Mining
74
6.6
Corrosion of reinforcement in concrete
References
1. C Andrade, I Rz-Maribona, S Feliu and J A Gonzales, ‘Macrocell versus microcell corrosion 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 the corrosion of steel embedded in concrete’, Corrosion ’94, NACE, Houston, 1994, paper No. 94037. 3. RILEM, Technical Recommendation 124 SRC, ‘Guidelines to repair strategies for concrete 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 on stainless steel behaviour in reinforced concrete’, Int. Conf. Eurocorr ’98, European Federation 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 galvanic coupling between carbon steel and stainless steel reinforcement in concrete’, Int. Conf. Corrosion and Rehabilitation of Reinforced Concrete Structures, Federal Highway Administration, Orlando, 7–10 December 1998. 7. A Torres-Acosta, A Sagués, R Sen, ‘Galvanic interaction between carbon fiber reinforced plastic (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.
© 2007, Institute of Materials, Minerals and Mining
7 Macrocell 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
7.1
Introduction
Reinforcing steel in good quality concrete does not corrode even if sufficient moisture and oxygen are available. This is due to the spontaneous formation of a thin protective oxide film (passive film) on the steel surface in the highly alkaline pore solution of the concrete. When sufficient chloride ions (from deicing salts or from sea water) have penetrated to the reinforcement or when the pH of the pore solution drops to low values due to carbonation, the protective film is destroyed and the reinforcing steel is depassivated. Corrosion in the form of rust formation and/or loss in cross-section of the rebars then occurs in the presence of oxygen and water (humidity) [1–3]. The corrosion of steel in concrete is essentially an electrochemical process, where, at the anode, 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-circuited corrosion cell, with the flow of electrons in the steel and of ions in the pore solution of the concrete [1–3]. According to the different spatial location of anode and cathode, corrosion of 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. Uniform corrosion is generally caused by carbonation of the concrete or by very high chloride content at the rebars. ∑ as macrocells, where a net distinction between corroding areas of the rebar (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 otherwise passive rebar (Fig. 7.1), a typical face to face situation is the corroding upper 75 © 2007, Institute of Materials, Minerals and Mining
76
Corrosion of reinforcement in concrete
7.1 Localised corrosion attack on a 20 mm rebar, loss in cross-section ca. 30%.
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 to the large cathode/anode area ratio [4–6]. Indeed, values of local corrosion rates up to 1 mm per year have been reported for bridge decks, sustaining walls or other chloride contaminated reinforced concrete structures [7–9]. This rapid corrosion attack may lead – if not detected early – to structural safety problems. Macrocell corrosion can be considered like a battery, the total current flowing, IME, is given by the driving voltage (potential difference between uncoupled anode and cathode), DU, divided by the sum of the resistance of the electrolyte, REl, the resistance of the anodic, RA(i), and the cathodic reaction, 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 driving voltage DU [10, 11]. The resistance of the electrolyte, REl, contains the geometry factor anode/cathode (e.g. increases for small anodes) and the mortar or concrete resistivity. The influence of porosity (w/c ratio, hydration...), relative humidity and temperature on the resistivity of cement-based materials is well known [12, 13]. The temperature dependence of the electrolyte resistance can 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 range of 1700 K (synthetic pore solution, pH 13.5) and 3800 K (concrete exposed
© 2007, Institute of Materials, Minerals and Mining
Macrocell corrosion of steel in concrete
77
for a long time at 60% rh) have been reported [12, 13]. The cathodic oxygen reduction and especially its temperature dependence have not yet been studied extensively and only few data on Tafels slope are available [14]. Corresponding data for the anode reaction in mortar or concrete are missing. For the temperature dependence of the two electrochemical reactions and of the total macrocell current an equation similar to eq. (7.2) can be written with the constant in the Arrhenius equation called a: I = I0/exp (a [1/T – 1/T0])
(7.3)
The aim of this work was to investigate the temperature dependence of the cathodic and the anodic reactions in the macrocell and to provide the necessary input data for the numerical modelling. The intensity and the temperature dependence of the macrocell current, IME, and the current distribution on the cathode are evaluated by numerical modelling.
7.2
Experimental methods
7.2.1
Cathodic and anodic reactions
The reaction kinetics were studied with potentiodynamic polarisation curves in a conventional three electrode electrochemical cell in a thermostatic bath in order to vary the temperature. The reference electrode was a saturated calomel electrode. The cathodic reaction was studied on polished mild steel in synthetic pore solution with pH 13.5 [14] open to air or deaerated with argon gas. The anodic reaction was studied in 0.1M HCl. Measurements of the cathodic polarisation curves in mortar were performed with specially designed cylindrical mortar samples with a diameter of 4 cm. In the centre, a degreased rebar sample (Ø 1 cm) was mounted; the counter electrode was a stainless-steel grid (Ø 2.5 cm) and, as the reference electrode, 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 28 days 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 of a 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 for steel in mortar with the experimental setup reported previously [15]. A linear macrocell arrangement was prepared as a mortar block of 30 ¥ 30 mm with a length of 34 cm. It contained a segmented cathode (10 electrically isolated segments of 25 mm length) and, in the centre, an anode of 10 mm length (precorroded, embedded in a chloride-containing mortar). The cover depth © 2007, Institute of Materials, Minerals and Mining
78
Corrosion of reinforcement in concrete
in this case was 1 cm. A more realistic two- and three-dimensional macrocell arrangement was prepared in mortar blocks of 55 ¥ 55 cm with a height of 15 cm, the cover depth was 7 cm on both sides. The total macrocell current and the currents to the individual cathode segments were measured with a zero resistance ammeter, and the switching was performed on a programmable multimeter (Keithley). A special switchboard guaranteed a complete short circuit during the measurements. Data were recorded on a personal computer. The corrosion potentials of the anode and cathodes were measured with a saturated calomel electrode (SCE).
7.2.3
Numerical modelling
The numerical modelling was performed with a commercial boundary element program BEASY Corrosion and Cathodic Protection Design (Computational Mechanics, Ashurst, England). As input parameters the anodic and cathodic polarisation 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 temperature dependence 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 a diffusion-limited current density. The Tafel slope increased with increasing prepassivation time to values higher than 240 mV per decade, leading to a
Current density (mA cm–2)
1000.00 20 min 24 h 1 mo 4 mo
100.00
10.00 Flow: 1 mL s–1 open to air 1.00
0.10 –1400 –1200
–1000
–800 –600 –400 Potential (mVSCE)
–200
0
7.2 Influence of the ageing time of the passive film in synthetic pore solution on cathodic polarisation curves (flow rate 1 mL s–1, open to air) [14].
© 2007, Institute of Materials, Minerals and Mining
Macrocell corrosion of steel in concrete
79
decrease in the reduction current at constant potential. It is interesting to note that the diffusion-limited current density was practically independent of temperature in the range of 5–50 ∞C. The cathodic oxygen reduction in mortar showed lower current densities than in solution [14, 15]. The current densities increased with increasing temperature. 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 – was identical 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 are shown in Fig. 7.4. The Tafel slope is about 75 mV per decade at 20 ∞C and becomes 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
% Cathodic current normalized to 20 ∞C
The macrocell corrosion current measured between anode (rebar of 10 cm2 area embedded in chloride-contaminated mortar) and cathode (rebar in synthetic pore solution, aerated) is shown in Fig. 7.5 together with the temperature variation over time. The macrocell current for the five (in principle identical) individual macrocells differs in intensity (probably due to a different size of the effective anode area) but the temperature dependence is similar; it is
500
400
Cathodic, solution Cathodic, mortar Anodic, solution
300
200
100 0 0
10
20 30 40 Temperature (∞C)
50
7.3 Normalised temperature dependence of the anodic iron dissolution and the cathodic oxygen reduction in solution and in mortar.
© 2007, Institute of Materials, Minerals and Mining
80
Corrosion of reinforcement in concrete
Current density [mA cm–2]
1000
100
10
Temperature increase
3 deg 20 deg 40 deg 47 deg
1
0.1 –550
–500
–450 –400 Potential (mVSCE)
–350
7.4 Anodic polarisation curves of rebar in 0.1M HCI at different temperatures. Scan rate 1 mV s–1. 120
Current (mA)
100 80 60 40 20 0 50
100
150 Time (h) (a)
200
50
100
150 Time (h) (b)
200
250
Temperature (∞C)
50 40 30 20 10 0 250
7.5 (a) Macrocell current of five macrocells in synthetic pore solution and (b) temperature program versus time. © 2007, Institute of Materials, Minerals and Mining
Macrocell corrosion of steel in concrete
81
evident that the macrocell current increases with increasing temperature. The highest values measured were 120 mA at 50 ∞C, demonstrating very high corrosion rates. Further, the macrocell current remains fairly constant over the whole measuring period (e.g. compare the currents at 10 ∞C at the beginning and at 250 h). As for the anodic and cathodic reactions, the macrocell currents were normalised to the value at 20 ∞C; the resulting temperature dependence for the five macrocells is shown in Fig. 7.6. The constant of the Arrhenius equation results in a = 4350 ± 80 K in the same range as for the anodic and cathodic reactions (Fig. 7.3).
7.3.4
Macrocell corrosion in mortar – one dimensional model
The segmented one dimensional model macrocell bar allows the current distribution on the cathode to be determined as a function of the distance from the anode. The cathode current slightly decreases with time. The main effect is the decrease of the cathode currents with increasing distance from the 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 the macrocell bar. Modelling the same geometrical arrangement of the linear segmented macrocell with the boundary element program BEASY, using the mortar resistivity, the cathodic polarisation curve for the cathode segments and the anodic polarisation curve for the anode as input data, the macrocell current 400
IME normalized at 20 ∞C (%)
350 300 250 200 150 100 50 0 –10
0
10 20 30 Temperature (∞C)
40
50
7.6 Normalised temperature dependence of the macrocell current measured in solution.
© 2007, Institute of Materials, Minerals and Mining
82
Corrosion of reinforcement in concrete 3 Model calculation Experimental
2.5
Current (mA)
2
1.5
1
0.5 0
K5
K4
K3
K2
K1 Anode K1
K2
K3
K4
K5
7.7 Distribution of the cathodic currents as a function of the distance from the anode, results from experiments on the one-dimensional macrocell in mortar and from the numerical model.
and its distribution were calculated. A very good agreement between experimental and calculated values of the cathodic current was found (Fig. 7.7).
7.3.5
Macrocell corrosion in mortar – two dimensional model
The macrocell current in the two-dimensional model macrocell was measured as a function of temperature in several experiments. The macrocell current increases with temperature. It is interesting to note that even at temperatures as low as –10 or –20 ∞C a small macrocell current of some pA was still flowing (Fig. 7.8). The temperature dependence of the macrocell current in the mortar block is identical with that found in solution (Fig. 7.6), the coefficient of the Arrhenius equation describing the temperature dependence is 4209 ± 88 K. The two dimensional macrocell was modelled numerically with the boundary element program BEASY, using as input data the mortar resistivity, the polarisation curve of the cathodic reaction for the cathode segments and the anodic polarisation curve for the anode segment. The resulting current distribution is shown in Fig. 7.9. A gradual decrease from the centre (current density at the cathode >2.9 mA cm–2) to the end of the 55 cm-long cathode bars (current density <2.3 mA cm–2) can be seen. The current distribution is symmetrical for all four directions; it can be noted that the current density at the end of the bars is far from being negligible. The total macrocell current calculated with the numerical model as a function of temperature agrees well with the experimental results.
© 2007, Institute of Materials, Minerals and Mining
Macrocell corrosion of steel in concrete
83
Current (mA), temperature (∞C)
50 Temperature Current
40 30 20 10 0 –10 –20 0
50
100 Time (h)
150
200
7.8 Macrocell current and temperature versus time, two-dimensional macrocell in mortar. Anode area 3.1 cm2, cathode area 301 cm2, cover 7 cm.
7.9 Numerical calculations with the boundary element program BEASY CP of the current distribution in a two-dimensional macrocell in a mortar block (cover 35 mm). Temperature 5 ∞C, total macrocell current 45.9 mA, anode area 3.14 cm2.
7.4
Discussion
Macrocell corrosion between actively corroding areas of rebars and large passive areas (either beside the active spot or behind in a second layer of reinforcement) is of great concern because it results in very high local anodic current densities with corrosion rates up to 0.5 to 1 mm per year. The resulting local loss in cross-section has dangerous implications for structural safety if
© 2007, Institute of Materials, Minerals and Mining
84
Corrosion of reinforcement in concrete
the corroded rebars are located in a zone of high tensile or shear stresses. Furthermore, these dangerous attacks very often do not manifest themselves at the concrete surface by cracking or spalling because soluble iron chloride complexes are formed [16]. The implications of this very inhomogeneous corrosion situation on different monitoring techniques that can be used to detect 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 in concrete has been studied for the first time over a wide range of temperatures, and second, numerical modelling with the boundary element program has been based not only on geometry (area ratio cathode/anode, cover depth) and concrete resistivity but also on the actual polarisation curves of the cathodic and 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 anodic and cathodic partial reactions of the corrosion reaction of steel in concrete has been studied in the range 0 to 50 ∞C. The coefficients, a, of the temperature dependence are summarised in Table 7.1. As can be noted, the temperature dependence of the anodic iron dissolution and the cathodic oxygen reduction reaction in alkaline media as well as the total macrocell current in alkaline solution and mortar agree very well, a = 4280 ± 150 K. A comparison with results from the literature shows good agreement (Fig. 7.10), although it has to be noted that the temperature interval may influence the calculated value of the temperature coefficient a. Numerical modelling of the macrocell with the polarisation curve of the anodic and cathodic partial reactions and the mortar resistivity allowed the influence of temperature on the macrocell current to be calculated and a very good agreement with the experimental results was observed (Fig. 7.10). It is very interesting to note that the coefficients of temperature dependence calculated from macrocell currents measured between an anode in instrumented cores and the rebar network of bridge decks [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 and total macrocell current in solutions and mortar experiments Current/Media
Solution
Macrocell Anodic Cathodic
4350 ± 80 4300 4310
© 2007, Institute of Materials, Minerals and Mining
Mortar 1-dim
4250
Mortar 2-dim
Field 3-dim
4210 ± 90
4000 ± 250 [22]
Macrocell corrosion of steel in concrete
85
IME normalized at 20 ∞C (%)
500
400
300
Experimental Arya (11) Schiessl/Raupach (70) Raupach (106) Weyers/Liu (155) Calculated
200
100
0 –20
–10
0
10 20 30 Temperature (∞C)
40
50
7.10 Temperature dependence of the macrocell current of steel in mortar and concrete. Comparison of literature values of Arya [18], Schiessl [19], Raupach [20] and Weyers and Liu [21] with the current experimental (∑) and numerical (䉬) results.
IME normalized at 20 ∞C (%)
500
400
Cathodic Anodic Resistance [Ohm] Macrocell current
300
200
100
0 –20
–10
0
10 20 30 Temperature (∞C)
40
50
7.11 Temperature dependence of the anodic, cathodic and macrocell corrosion currents compared with the temperature dependence of the concrete resistivity (b = 3080 K, rh 85%).
A comparison of the temperature dependence of the anodic and cathodic reactions and the macrocell current of steel in concrete with the temperature dependence of the mortar (at 85% RH) is shown in Fig. 7.11. As can be noted, the temperature dependence is much more pronounced for the electrochemical reactions than for the mortar resistivity. This indicates that
© 2007, Institute of Materials, Minerals and Mining
86
Corrosion of reinforcement in concrete
the proportionality between polarisation resistance and concrete resistivity often reported in the literature [23, 24] cannot be assumed to be valid a priori or – in other words – for chloride-induced corrosion, the corrosion rates 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 electrolyte resistance, Rel, and the two polarisation resistances of the anodic (Ra) and cathodic (Rc) reactions. Which one of these three resistances determines the overall macrocell current is still under discussion. It is often stated that the concrete resistance is the controlling factor for the corrosion rate of steel in concrete. From the results of this work it can be concluded that for large cathode areas and in mortar with a quite low resistivity of about 100 W m (as frequently occurs in chloride-induced corrosion), it is the cathodic oxygen reduction reaction occuring on the passive reinforcement that is controlling the overall reaction (> 90% of the total resistance in eq. 7.1) whereas the part of the electrolyte resistance is only ca. 5–10%. The anode is practically unpolarised. Numerical simulation with a plan-parallel arrangement of the electrodes [10] has also shown that the cathodic reaction controls the overall macrocell corrosion by more than 60%. The difference to this work arises from the different geometrical arrangement: the small anode in a relatively large mortar block (55 ¥ 55 ¥ 15 cm) has a lower resistive control than the plan-parallel arrangement. It can be concluded that the geometry used in laboratory studies with mortar beams or blocks can greatly influence the experimental results; similarly the cathodic polarisation curve (Tafel slope, exchange current density) used as 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 the polarisation curve of the anode and the cathode together with the concrete resistivity as input data has been shown to be a powerful tool in studying macrocell corrosion. The total macrocell current, its distribution on the cathode and its temperature dependence agreed very well with the laboratory experiments with identical parameters. This allows the usual way of doing experiments to be changed: based on a small set of input data, numerical modelling is first performed for different geometries (e.g. long slabs, decks etc.) and for different concrete resistivities. In a second step, laboratory experiments are designed and the results of numerical modelling are verified.
© 2007, Institute of Materials, Minerals and Mining
Macrocell corrosion of steel in concrete
7.5
87
Conclusions
In real structures, localised chloride-induced corrosion macrocells are formed that greatly accelerate the local dissolution rate of the anode. From the current work, performed on active/passive model macrocells in the laboratory combined with numerical modelling, it can be concluded that: 1. The temperature dependence of anodic dissolution of iron, of the cathodic oxygen reduction reaction and of the overall macrocell corrosion is nearly identical (a = 4200 K) and much higher than the temperature dependence of the mortar or concrete resistivity (rh 85%). 2. Numerical modelling with the boundary element program BEASY provides practically the same results for the total macrocell current, current distribution on the cathode and temperature dependence as found in the experiments in mortar for a given geometrical arrangement, concrete resistivity and cathodic polarisation curves of the passive steel in concrete. 3. Chloride-induced macrocell corrosion concrete of low to moderate resistivity is governed by the cathodic oxygen reduction reaction and not by the resistivity of the concrete. 4. This numerical approach allows parameter studies to be performed very rapidly and experiments with macrocells in concrete to be designed in a rational 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 Environmental Degradation, 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 – theoretical background and practical consequences’, Constr. Build. Mater., 1996, 10, 329. 6. B. Elsener, A. Hug, D. Bürchler and H. Böhni, ‘Evaluation of localised corrosion rate on steel in concrete by galvanostatic pulse technique’, Corrosion of Reinforcement in 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 (in German) Schweiz. Ingen. Architekt, 1991, 109, 272. 9. B. Elsener, Corrosion Rate on Reinforced Concrete Structures Determined by Electrochemical Methods, Mater. Sci. Forum, 1995, 192-194, 857.
© 2007, Institute of Materials, Minerals and Mining
88
Corrosion of reinforcement in concrete
10. M. Raupach and J. Gulikers, ‘Investigations on cathodic control of chloride induced reinforcement corrosion’, in Corrosion of Reinforcement in Concrete – Corrosion Mechanism 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 reinforced concrete structures’, in Corrosion of Reinforcement in Concrete, Monitoring, Prevention and 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 Thesis No. 11876, 1996, ETH Zürich (in German). 13. D. Bürchler, B. Elsener and H. Böhni, ‘Electrical resistivity and dielectric properties of hardened cement paste and mortar’, in Corrosion of Reinforcement in Concrete Construction, 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 stainless steel in alkaline solutions’, in Corrosion of Reinforcement in Concrete – Corrosion Mechanism 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 von Stahl 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, ‘Electrochemical behaviour of steel under cathodic protection in medium simulating concrete’, Corrosion and Corrosion Protection of Steel in Concrete, ed. R. N. Swamy, Sheffield Academic Press, Vol. 2, 1382–1391. 17. B. Elsener, ‘Macrocell corrosion of steel in concrete – implications for corrosion monitoring’, 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 Applied Science, 49–58. 20. M. Raupach, ‘Results from laboratory tests and evaluation of literature on the influence of temperature on reinforcement corrosion’, Corrosion of Reinforcement in Concrete, EFC Publication No. 25, J. Mietz, B. Elsener and R. Polder, Eds., IOM Communications London, 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.
© 2007, Institute of Materials, Minerals and Mining
8 Modelling of chloride-induced corrosion of reinforcement in cracked high-performance concrete based on laboratory investigations M. R A U P A C H and C. D A U B E R S C H M I D T, Aachen University, Germany
8.1
Background
High-strength concrete with a compressive strength above 80 N mm–2 has been used for many years. Besides increasing the load-bearing capacity of the concrete one important aspect for the development of high-performance concrete mixtures is the durability, especially the resistance against chloride diffusion. Numerous investigations have been performed with regard to the abrasion 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 the reinforcement in high-performance concrete structures when exposed to aggressive environmental conditions. Cracks in reinforced concrete structures allow aggressive agents like chlorides from de-icing salts or sea-water to penetrate into the concrete. As shown in Fig. 8.1, the corrosion rate of a macrocell is a function of the anodic polarisation resistance, RA, the cathodic polarisation resistance, RC, the resistivity of the electrolyte (concrete), Rel, and the difference between the rest potentials at the anode (ER,A) and the cathode (ER,C). It can be deduced from the equation that the corrosion rate is reduced if only one of the three resistances (RA, RC and Rel), especially the most corrosion rate determining resistance, increases. Due to the low permeability and high electrolytic resistivity of highperformance concrete, it might be expected that the corrosion rates of steel in the area of cracks in concrete are far lower than in normal concrete. Another reason for a possible reduction of corrosion rates in high-performance concrete may be the restricted volume expansion of corrosion products at the anode due to the limited space in small cracks causing the anodic polarisation resistance, RA, to increase with time. In order to evaluate the corrosion behaviour of steel in cracked highperformance concrete, laboratory tests have been performed on cracked concrete beams. 89 © 2007, Institute of Materials, Minerals and Mining
90
Corrosion of reinforcement in concrete Crack
Ie =
Concrete
Rel
E R,C – E R,A R A + R C + R el
ER,C RA
D E c = Rc I e
RC
ER,A Steel
ER,C
RSt ª 0
DE
DE
EC,C
D Eel = Rel Ie D EA = R A I e
Ie Anode
Cathode
EC,A
ER,A Ie
8.1 Simplified electrical circuit model for the corrosion of steel in cracked concrete [ER,C and ER,A: rest potentials at the cathode and anode; RA, RC and Rel: polarisation resistances at the anode, the cathode and resistivity of the electrolyte/concrete; Ie: macrocell current (~ corrosion rate)] [2]. Section B-B 25 100 25
Section A-A 100
50
350 Chloride solution 1%
Uncoated anodic steel area
Mild steel cathodes
Crack
Epoxy coated rebar Activated titanium cathodes
150
100
Titanium cathode 150
700 Top view
䉯B Anode Rebar Ø 12 mm
A
A
50
䉯B 600 700
50
Cathodes: Mild steel Epoxy coated
(Measures in mm)
8.2 Design of test specimens.
8.2
Test programme
The design of the test specimens (Fig. 8.2) allowed the measurement of various corrosion parameters such as macrocell corrosion currents (anodic and cathodic currents), corrosion potentials and electrolytic resistance of the concrete. The main rebar intersecting the crack was designed to be the anode, therefore the active area was restricted by coating the reinforcement in the uncracked area. The cathodes consisted of 24 mild steel rebars, fixed at
© 2007, Institute of Materials, Minerals and Mining
Modelling of chloride-induced corrosion of reinforcement
91
defined distances from the crack (50 to 250 mm) and with two concrete covers (23 and 47 mm). After concreting, the specimens were stored for 7 days in a fog-room and afterwards for another 21 days in a climate at 20 ∞C and 65 % rh. The cracks were formed by fixing the specimens in steel frames and by bending them to the designed crack widths (0.1, 0.25 and 0.5 mm 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 days dry in a laboratory atmosphere at about 65 % rh). In order to evaluate the corrosion behaviour of reinforcing steel bars in the cracked concrete beams, macrocell corrosion currents, corrosion potentials and electrolytic resistances were measured over a period of 64 cycles (at least 15 months). The concrete mixtures used in the tests are specified in Table 8.1.
8.3
Results
8.3.1
Measurement of the electrolytic resistivity of the concrete, Rel
To measure the resistivity over time at different depths an embedded multiring electrode (MRE) was used [6]. The sensor consisted of several stainlesssteel rings maintained at a defined distance from one another by insulating plastic rings (Fig. 8.3). Cable connections through the sensor enabled the resistivity of the concrete to be determined between each pair of neighbouring stainless-steel rings by means of impedance measurements. The ac resistance values (in W) can be converted to resistivity values by a sensor-specific transfer factor determined in aqueous solutions with known conductivity. The standard type of the multi-ring electrode-sensor allowed the measurement of eight resistances between nine rings, down to a distance of 42 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 of concrete are presented. The highest measurable value is 20 kW m. A significant increase in the resistivity with time is noted for all depths for the C 35 concrete mixture (Fig. 8.4). This increase for all depths can be explained because this comparably permeable concrete had fully dried, whereas concrete mixtures C 65 and C 85-0 (Fig. 8.5 and 8.6) had only dried to about 17 mm as 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) drying only occurred to a depth of 12 mm. All curves show a clear profile of decreasing resistivity with increasing depth. There is a tendency for the resistivity of the humid inner area of
© 2007, Institute of Materials, Minerals and Mining
92
Mixture
–
Cement
Cement
type
content
Water
Silica fume
kg m3
–
Super-plasticiser*
Water-to-binder
Compressive strength†
ratio
7d
%
–
28 d 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
†
© 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete
Table 8.1 Concrete mix proportions
Modelling of chloride-induced corrosion of reinforcement
93
Ring (noble metal) Cable
Top view
A
A Section A-A Cable
Concrete surface
2.5 2.5 2.5
Electrolytic resistance in W
7 12 17 22 27 32 37 42
Distance from surface
(mm)
8.3 Schematic presentation of the multi-ring electrode (MRE). Concrete mixture C 35
Resistivity of concrete (W m)
100000
Max. measurable value 7 mm
12 mm
17 mm
22 mm 27 mm
32 mm
10000 37 mm
42 mm
1000
100 0
100
200 300 Concrete age (d)
400
500
8.4 Concrete resistivity of specimens C 35 at various distances from surface.
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 higher than 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). © 2007, Institute of Materials, Minerals and Mining
94
Corrosion of reinforcement in concrete Concrete mixture C 65
Resistivity of concrete (W m)
100000
Max. measurable value 7 mm
12 mm
10000 17 mm 22 mm 1000
27 mm
32 mm
37 mm
42 mm
100 0
100
200 300 Concrete age (d)
400
500
8.5 Concrete resistivity of specimens C 65 at various distances from surface. Concrete mixture C85-0
Resistivity of concrete (W m)
100000
Max. measurable value 7 mm
12 mm 17 mm
10000
22 mm 27 mm 1000 32 mm 37 mm 42 mm 100 0
100
200 300 Concrete age (d)
400
500
8.6 Concrete resistivity of specimens C 85 at various distances from surface.
8.3.2
Results of potential measurements ER,A and ER,C
The average differences in rest potential between the anode and cathodes of the macrocells in each specimen were measured several times. The values differ between 370 mV and 460 mV for all corroding macrocells. No significant influence was observed for the different concrete mixtures or concrete covers (Fig. 8.10). © 2007, Institute of Materials, Minerals and Mining
Modelling of chloride-induced corrosion of reinforcement
95
Concrete mixture C 85-SF
Resistivity of concrete (W m)
100000
Max. measurable value 7 mm
12 mm 17 mm
10000
22 mm 27 mm 32 mm 37 mm 42 mm
1000
100 0
100
200 300 Concrete age (d)
400
500
8.7 Concrete resistivity of specimens C 85-SF at various distances from surface. Concrete mixture C 115-SF
Resistivity of concrete (W m)
100000
Max. measurable value 7 mm
12 mm 17 mm
10000
22 mm 27 mm 32 mm 37 mm 42 mm 1000
100
0
100
200 300 Concrete age (d)
400
500
8.8 Concrete resistivity of specimens C 115-SF at various distances from surface.
8.3.3
Results of corrosion current measurements Ie
The currents for specimens made of the five different concrete mixtures with crack widths of 0.10, 0.25 and 0.50 mm were measured for 64 wet-dry cycles. The current has been recorded separately for three macrocells consisting of the four bottom cathodes (concrete cover: 47 mm) with different distances
© 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete
Depth under concrete surface (mm)
96
Electrolytic resistivity of concrete (W m) 1000 10000
100 0
100000
Concrete age t = 50 d 10 C 65
C 85-S
20
30
C 35 C 85-0
Max. measurable value C 115-S
40
Depth under concrete surface (mm)
50 Electrolytic resistivity of concrete (W m) 1000 10000
100 0
100000
Concrete age t = 400 d C 85-S 10 Max. measurable value
C 65
20
C 115-S 30
40
C 35 C 85-0
50
8.9 Profiles of concrete resistivity for all specimens at concrete age of 50 days (upper) and 400 days (lower).
(50, 150 and 250 mm) from the crack and the anode and for three macrocells consisting of the four top cathodes (concrete cover: 23 mm) with different distances from the crack and the anode. To calculate the mass loss of the reinforcement it was necessary to integrate the current versus time. This cumulative charge is shown as a function of the number of wetting periods in Fig. 8.11. By far the highest corrosion charges from 0.5 mm wide cracks were recorded for the reference concrete mixture C 35 (average: 20369 mA d), the lowest 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 is obvious 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: 9505 mA d). © 2007, Institute of Materials, Minerals and Mining
Modelling of chloride-induced corrosion of reinforcement Top cathodes c = 23 mm Bottom cathodes c = 47 mm
500
Voltage (mV)
97
450
400
350
300 0.10 0.25 0.50 C 35
0.10 0.25 0.50 0.50 0.50 C 65 C 85-0 C 85-SF
Crack width (mm) 0.50 Concrete mixture C 115-SF
Cumulative corrosion current (mA d)
8.10 Differences in rest potential between anode and cathode in mV (age of specimens: 187 to 383 days).
25000 20000 15000 10000 5000 0 0.10 C 35
0.25
0.50
0.10 0.25 C 65
0.50
0.50
C 85-0
0.50
0.50
C 85-SF
Specimens (average value of two macrocells)
C 115-SF
56 40 24 Number of wetting 8 and drying cycles Crack width (mm) Concrete mixture
8.11 Cumulative macrocell corrosion currents as a function of the number of wetting and drying cycles.
A decrease in the corrosion rate over time due to an increasing anodic polarisation resistance was not observed. However, it can not be excluded that this effect might occur in the course of time, which can only be verified by long-term investigations.
8.3.4
Cathodic current distribution
To evaluate possible changes in the current distribution of macrocells in high-performance concrete mixtures, the local currents for the different anode–
© 2007, Institute of Materials, Minerals and Mining
Balance of the current (%)
98
Corrosion of reinforcement in concrete
100% C 35; w = 0.10 mm C 65; w = 0.50 mm C 85-0; w = 0.50 mm C 85-SF; w = 0.50 mm C 115-SF; w = 0.50 mm
75%
50%
25% Top cathodes (c = 23 mm)
250
150
Bottom cathodes (c = 47 mm)
0% 50 50 Distance anode – cathode (mm)
150
250
8.12 Balances of the cathodic currents of the macrocells (w = crack width).
cathode distances have been calculated. Assuming that the total corrosion rate of the anode with the top cathodes is 100 %, the contribution of each anode–cathode cell (three cells with different distances from the crack) was evaluated. Figure 8.12 shows the determined current balances as percentages of the total current for the top cathodes in the left part. The same evaluation was made for the macrocells between anodes and the bottom cathodes (right part of Fig. 8.12). No systematic differences between the corrosion current of the top macrocell and the current of the bottom macrocell could be observed for all tested specimens. Furthermore, no change of the current distribution resulting from the use of high-performance concrete could be evaluated.
8.3.5
Visual examinations
To verify the results of the current measurement the specimens were broken after 64 cycles and the positions of the cracks in the specimens as well as the degree of corrosion were determined. Whereas the cracks in specimens with C 35 and C 65 crossed the anode (main rebar) in the planned uncoated area, in specimens of high-performance concrete with small crack widths (0.10 and 0.25 mm), the cracks were found to cross the anodes in the coated region. It seems that the more brittle the concrete the more the cracks formed at discontinuities in the coating of the reinforcement. The results for specimens where the crack crossed the coated region of the main rebar are neglected in the further evaluation.
© 2007, Institute of Materials, Minerals and Mining
Modelling of chloride-induced corrosion of reinforcement
99
However, the degree of corrosion was clearly related to the measured cumulative corrosion current according to Faraday’s law.
8.3.6
Influence of water-to-binder ratio on resistivity and corrosion current
Figure 8.13 shows the water-to-binder ratio of the concrete mixtures (see Table 8.1) versus the resistivity of the concrete measured with the multi-ring electrode at a depth of 42 mm. As expected, a decrease of the water-to-binder ratio 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 values for each concrete mixture) are related to the water-binder-ratio of the concrete mixtures. The determined average corrosion rate is strongly related to the water-to-binder ratio of the concrete in macrocell corrosion.
8.3.7
Relation between resistivity and corrosion current
Figure 8.15 shows the resistivity against the measured average corrosion rate, as drawn from Fig. 8.13 and 8.14. As can be seen from Fig. 8.15, the decrease of corrosion current is not linearly related to the increase of resistivity. Thus, the corrosion rate is not controlled totally by Rel, but mainly by the polarisation resistances RA and RC . (Multi-ring electrode: depth 42 mm)
Electrolytic resistivity (W m)
10000
C 115-S
C 85-S 1000
C 85-0
100 0.2
C 65-0
0.3 0.4 Water-to-binder ratio
C 35-0
0.5
8.13 Water-to-binder ratio of the concrete mixtures versus measured resistivity of the specimens at a depth of 42 mm.
© 2007, Institute of Materials, Minerals and Mining
100
Corrosion of reinforcement in concrete 50
Average corrosion current (mA)
C 35-0 40 C 85-0
30
20
C 65-0
C 115-S C 85-S
10
0 0.2
0.3 0.4 Water-to-binder ratio
0.5
8.14 Water-to-binder ratio of the concrete mixtures versus measured corrosion current. 50
Average corrosion current (mA)
C 35-0 40
30
C 65-0 C 85-0
20 C 115-S C 85-S
10
0 100
1000 Electrolytic resistivity (Wm) (Multi-ring electrode: depth 42 mm)
10000
8.15 Electrolytic resistivity of the concrete (measured with multiringelectrode) versus average corrosion current.
8.4
Numerical simulation
To verify the results of Fig. 8.15 numerical simulations of the corrosion process have been carried out. Actively corroding steel may exhibit microcell action in which anodic and cathodic sites are randomly distributed over the exposure surface of the steel electrode, giving rise to uniform corrosion attack. In this condition, the polarisation behaviour of corroding steel is described by the following equation relating average current density, i, and potential change (overvoltage), DU:
© 2007, Institute of Materials, Minerals and Mining
Modelling of chloride-induced corrosion of reinforcement
ln(10) DU ˆ – ln(10) DU ˆ ˘ È i = icorr Í expÊ – exp Ê b bc Ë ¯ Ë ¯ ˙˚ a Î
with
ln(10) ba bc icorr
101
(8.1)
= 2.303 = anodic Tafel slope = 90.7 mV dec–1 = cathodic Tafel slope = 176.3 mV dec–1 = self corrosion rate, here 1.0 mA cm –2 [4].
For passive steel it is assumed that anodic reactions can only proceed to a very limited extent. The electrochemical behaviour of passive reinforcing steel under cathodic polarisation is given by: –ln(10) DU ˆ i = 1 – exp Ê bc Ë ¯
with
1 – icorr
exp Ê Ë
– ln(10) DU ˆ bc ¯ ilim
(8.2)
ilim = limiting diffusion current density due to oxygen diffusion based on data obtained from experimental investigations [4].
Based on these equations, the corrosion current of the macrocells with coplanar arrangement of local anode and macro cathode as a function of the distance of the cathodes from the anodes can be calculated. Figure 8.16 shows 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 from the anode. Furthermore, the corrosion current of the specimens can be calculated as a function of the density of reinforcement. In Fig. 8.18, the curves of the
El. current density (mA cm–3)
0.15
I/b/h = 60/15/15 cm Ue = 400 mV 0.12
C 35, Rel = 500 W *m
0.09
0.06
0.03
C 115, Rel = 5000 W *m
0.00 6
12 18 24 Distance from the anode/crack (cm)
8.16 Calculated corrosion current of cracked specimens with C 35 and C 115-SF concrete as a function of the distance between anode and cathode.
© 2007, Institute of Materials, Minerals and Mining
30
102
Corrosion of reinforcement in concrete 0
–50
C 115, Rel = 5000 W *m
Potential (mV)
–100 –150 –200 –250 –300 –350 C 35, Rel = 500 W *m
–400
I/b/h = 60/15/15 cm Ue = 400 mV
–450 6
12 18 Distance from the anode/crack (cm)
24
30
8.17 Calculated distribution of the potential of cracked specimens with C 35 and C 115-SF concrete as a function of the distance between anode and cathode. 250 C 35, Rel = 500 W m
I/b/h = 60/15/15 cm Ue = 400 mV
Macrocell current (mA)
200
150
100 C 115, Rel = 5000 W m 50
0 0
5
10
15 20 25 30 35 Reinforcement density (cm2 cm–1)
40
45
50
8.18 Calculated corrosion current of cracked specimens with C 35 and C 115-SF concrete versus density of reinforcement.
calculation with Ue = 400 V and Ra = 0 W are shown. With increasing density of reinforcement, the corrosion current increases too. For the high-performance concrete C 115-SF (Rel = 5000 W m) the calculated corrosion current is lower than for the concrete C 35 (Rel = 500 W m), following a non-linear relationship. Figure 8.19 shows the ratios of calculated corrosion current of C 115-SF to the corrosion current of C 35. For concretes with a 10-fold electrolytic resistivity ratio, the corrosion current decreased only by a factor of two for
© 2007, Institute of Materials, Minerals and Mining
Modelling of chloride-induced corrosion of reinforcement
103
Rel. macrocell currents C 115/C 35
1
I/b/h = 60/15/15 cm Ue = 400 mV Rel,C115 = 5000 W *m Rel,C35 = 500 W *m
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
10
20 30 Reinforcement density (cm2 cm–1)
40
50
8.19 Ratio of calculated corrosion current of cracked specimens with C 35 and C 115-SF concrete versus content of reinforcement.
low reinforcement densities. For high reinforcement densities this factor is about 4.
8.5
Conclusions
Tests have been carried out to determine the corrosion mechanisms of specimens produced with cracked high-performance concrete beams. As expected, the resistivity of the electrolyte (concrete) increases significantly in high-performance concrete. The resistivity of concrete mixture C 115 in humid conditions is about ten times higher than the resistivity of concrete mixture C 65 in humid conditions. This increase in the electrolytic resistivity leads to a reduction in the corrosion rates in specimens with high-performance concrete. Accordingly, the corrosion currents of macrocells in high-performance concrete mixtures are also reduced, e.g. the average corrosion rate of concrete mixtures C 85SF and C 115-SF is about 1/3 of the current of C 35 (Fig. 8.11) under the conditions investigated. It can be summarised that the mechanisms of corrosion do not change in high-performance concrete: the major type of corrosion is macrocell corrosion, the balances of currents as a function of the distance from the cathode to the anode remain nearly the same for normal and high-strength concrete and there is no sign of a reduction in the current due to insufficient space for corrosion products in specimens of high-performance concrete after 15 months of exposure to aggressive wet and drying cycling. Comparing the ratio of the increasing resistivity and the ratio of reduction of the corrosion rate with regard to the equation of Fig. 8.1, it can be determined
© 2007, Institute of Materials, Minerals and Mining
104
Corrosion of reinforcement in concrete
that the corrosion rate is not controlled totally by Rel, but mainly by the polarisation resistances RA and RC. These results have been confirmed by numerical simulations of the corrosion process. These simulations show that the corrosion rate is mainly dependent on the polarisation resistances of the anode and the cathode. The use of high-performance concrete with cracks exposed to severe chloride attack leads to a reduction in the corrosion rate of the anode in comparison to normal strength concrete. So the expected service lifetime of the structure can be prolonged significantly by using high-performance concrete. To investigate long term effects, like a possible anodic self polarisation, additional tests over increased periods are planned.
8.6
References
1. Guse, U. and Hilsdorf, H. K., Durability Aspects of High Strength Concrete. in HighPerformance 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 – theoretical background 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 and calculations using a transmission line model. Corrosion of Reinforcement in Concrete Construction, 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 – a new approach based on investigations of macrocells, in 8th International Conference on 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 on water distribution in concrete using multi-ring electrodes, in Concrete Bridges in Aggressive Environments, Philip D. Cady International Symposium, Minneapolis, November 9–10, 1993, Weyers, R. E. (Ed.), American Concrete Institute, Detroit ACI SP-151, 1994, 119–133.
© 2007, Institute of Materials, Minerals and Mining
9 Influence 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 and P. P E D E F E R R I, Politecnico di Milano, Italy
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 the source. Stray currents can deviate from their intended path because they find a parallel and alternative route. They may also find a low resistance path by flowing through metallic structures buried in the soil (pipelines, tanks, industrial and marine structures). For instance, underground pipelines can pick up stray current from a railway system at some point remote from the traction power substation and discharge the current to the soil and then back to the rail near to the substation. Stray direct currents are known to be much more dangerous than stray alternating currents. In the case of dc, a cathodic reaction (e.g. oxygen reduction or hydrogen evolution) takes place where the current enters the buried structure, while an anodic reaction (e.g. metal dissolution) occurs where the current returns to the original path, through the soil. Metal loss 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 has been 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 shown that stray ac current can induce corrosion only in those particular circumstances where very high currents are picked up by the buried structure, so that the current density in the points where the current enters or leaves the structure is extremely high (a threshold ranging from 20–100 A m–2 has been proposed4). Stray currents can also flow through reinforced or prestressed concrete and produce an alteration of the electrical field inside the concrete, which can influence the corrosion of embedded steel. Several types of structures may be subjected to stray current, such as bridges and tunnels of the railway networks or structures placed in the neighbourhoods of railways. Here, the 105 © 2007, Institute of Materials, Minerals and Mining
106
Corrosion of reinforcement in concrete
concrete, like the soil in buried structures, is the electrolyte and the reinforcing bars or prestressing wires can pick up the stray current (Fig. 9.1). It has been shown that stray dc currents rarely have corrosive consequences on steel in concrete, in contrast to their effect on metallic structures in the soil.5,6 In fact, steel in alkaline and chloride-free concrete is passive. Passivity, besides being essential for protecting the reinforcement from the environmental aggressiveness, also provides resistance to stray currents. Figure 9.2 depicts the anodic polarisation curve of passive steel in concrete and shows that at
Dc electric substation
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 steel reinforcement in concrete.
Potential (mV vs SCE) Ea
Potential (mV vs SCE) 500
500 DV
0
Ecorr Ec –500
0 Ecorr
I –500
–1000
–1000 Log (current density) (a)
Log (current density) (b)
9.2 Schematic representation of electrochemical conditions in (a) the cathodic and (b) the anodic zones of reinforcement in noncarbonated and chloride-free concrete which is subject to stray current I.
© 2007, Institute of Materials, Minerals and Mining
Influence of stray currents on corrosion of steel in concrete
107
potentials below +600 mV vs SCE no iron dissolution or any other anodic process takes place and, thus, it is impossible for the current to leave the metal. Before the stray current is picked up by the reinforcement, a significant potential difference (DV) has to be produced between the point where the current enters the reinforcement (cathodic site, Fig. 9.2a) and the point where the current returns to the concrete (anodic site, Fig. 9.2b). Experimental tests showed that passive reinforcement in non-carbonated and chloride-free concrete offers a high intrinsic resistance to stray current, since the driving force required to produce the circulation of an appreciable current 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 through the reinforcement, this does not automatically lead to corrosive attack. The anodic 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 current flows for sufficiently long periods of time.7 The initiation of corrosion was ascribed to the depletion of the alkalinity in the vicinity of the anodic areas promoted by the anodic reaction of oxygen evolution (2H2O Æ O2 + 4H+ + 4e–). In the case of structures contaminated by chlorides, even at levels too low to initiate pitting corrosion, stray currents may have more serious consequences. In fact, stray currents stimulate the initiation of pitting corrosion by taking the steel potential to values higher than the pitting potential. Once corrosion has initiated on the reinforcement, for instance due to carbonation or chlorides, the effects of stray currents become similar to that experienced by steel buried in the soil. In this paper, the effects of stray currents on the corrosion of steel embedded in atmospherically exposed concrete are studied. Both the mechanisms of corrosion initiation on initially passive reinforcement and the effects of stray currents on reinforcement that was already corroding have been studied. The influence of several factors has been investigated, such as: the type of current (ac or dc), the current density, the presence of interruptions in the circulation of current, and the chloride content in the concrete. The mechanism of initiation of corrosion has also been investigated by measuring changes in pH and chloride content induced by the stray current in the cement paste in the vicinity of the steel surface. This work deals with consequences of stray currents on reinforced concrete structures; high-strength steels for prestressed concrete, which can also suffer corrosion due to hydrogen embrittlement in cathodic areas, are not taken into consideration.
9.2
Experimental tests
Tests were carried out on specimens of cement paste and concrete with two steel inserts subjected to the circulation of dc and ac currents (Fig. 9.3).
© 2007, Institute of Materials, Minerals and Mining
108
Corrosion of reinforcement in concrete
Specimens of Fig. 9.3a had two parallel plates of carbon steel embedded in cement paste made of portland cement and a water/cement (w/c) ratio of 0.55. Only the two opposite surfaces of the plates were exposed to the cement paste (the rest was masked with epoxy). Fixed reference electrodes, made of a thin wire of mixed metal oxides activated titanium (MMO), were embedded in the vicinity of each plate. Specimens of Fig. 9.3b had two parallel bars of carbon steel embedded in concrete made of 350 kg m–3 of portland cement, w/c ratio 0.55, and 1900 kg m–3 of crushed limestone aggregate (the average strength of the concrete was 50 MPa). Chlorides were added in amounts ranging from 0.1 to 0.8% by weight of cement to some of the specimens of cement paste or concrete. The effects of direct current were studied by applying a constant direct current between the two electrodes (steel plates or rods) in the specimens, so that one was the cathode and the other one was the anode. Specimens were exposed in a climatic chamber at 20 ∞C and 95% rh. Current densities of 1 and 10 A m –2 were applied to specimens made of cement paste and 8.6 A m–2 to specimens in concrete. In order to study the effects of the interruption of current, tests were also carried out with cycles during which current 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 the feeding voltage were monitored. Periodically, depolarisation tests were carried out by interrupting the current for 5 min. and measuring the subsequent changes in the steel potential. The potential of the activated titanium electrodes was regularly calibrated
Steel bar f10 mm
Steel plate (35 ¥ 35 mm) Metal mixed oxide activated titanium electrode (MMO)
70 50
70
50
10
(a)
0
10
0
(b)
9.3 Specimens used for tests: (a) steel plates embedded in cement paste, (b) steel bars embedded in concrete (dimensions are in mm).
© 2007, Institute of Materials, Minerals and Mining
Influence of stray currents on corrosion of steel in concrete
109
against an external SCE. These measurements also allowed estimation of the pH of the pore solution near the surface of the steel, since the potential of activated titanium depends on pH. On specimens of the type shown in Fig. 9.3a, whose geometry favours a uniform distribution of current between the opposing steel plates, further analyses were carried out at the end of tests, i.e. when cracking occurred on the anodic side of the specimen. The steel plates were removed from the cement paste and weight loss was measured on the steel plate that was anodically polarised and underwent corrosion. The pH of the cement paste was analysed by spraying phenolphthalein and other commercial pH indicators on a fracture surface perpendicular to the steel plates; the area close to the steel plates was observed with a stereomicroscope. Chloride content was also measured by cutting the cement paste between the steel plates into slices with a thickness of 10 mm; powder from each slice was digested in nitric acid and chloride content was measured by means of potentiometric titration. Chloride analyses were also carried out on fragments of cement paste collected within 1–2 mm of the steel surface. Tests with alternating current were carried out on specimens of cement paste with 0 and 0.2% of chloride by weight of cement; a current density of 40 A m–2 was applied between the two plates for 2 months and the corrosion rate of steel was monitored, by means of polarisation resistance measurements carried out after the interruption of ac for 4 h. Tests were also performed on three specimens with the geometry of Fig. 9.3b, which had the two bars embedded in concrete with different chloride contaminations: 0–0.4, 0–0.8 and 0.4–0.8% by weight of cement. Alternate current of 50 A m–2 was initially applied to the passive bars for 5 months and the corrosion rate was measured. Afterwards, corrosion was initiated in the bar in concrete with the higher chloride contamination of each specimen (by imposing an anodic direct current of 3 A m–2). Consequently, a macrocouple was generated between the two bars in each specimen. Alternating currents of 20 A m–2 were superimposed for various lengths of time and their influence on the macrocouple current was monitored.
9.3
Results and discussion
9.3.1
Effects of direct current
Corrosion can be induced on passive steel in concrete when it is subjected to the circulation of anodic current for a long period. Afterwards the anodic current 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 with cement paste contaminated with 0.4% of chloride by weight of cement which was subjected to the circulation of a nominal current of 1 A m–2 (actual
© 2007, Institute of Materials, Minerals and Mining
110
Corrosion of reinforcement in concrete
current density was 1.1 A m–2). Both steel plates were initially passive and had a free corrosion potential of around –200 mV against the fixed reference electrode of mixed metal oxide (MMO) activated titanium. When the external current density of 1 A m–2 was applied, the steel plate that was polarised anodically reached a potential around +700 mV, showing that oxygen evolution took place at its surface. Conversely, the steel plate that was polarised cathodically reached a potential of about –1.1 V (Fig. 9.4a). The feeding voltage was 2.2 V (Fig. 9.4b). The initiation of corrosion on the anodic steel plate took place after 78 hours of testing, when a charge of 84 A h m–2 had been circulated; it was detected by a sharp decrease in the potential (and the feeding voltage as well). The initiation of corrosion also led to a significant change 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 had initiated, a potential of about –500 mV vs MMO was quickly reached by switching the current off. After corrosion initiation, the potential of anodic steel also showed remarkable fluctuations. Cracks developed in the cement paste 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 tests carried out on specimens made of cement paste or concrete with different chloride contaminations and applied currents. The onset of corrosion was associated with a decrease in the potential of the anodic steel (both in the
Potential (mV vs MMO)
1000
Anode
500
Corrosion initiation
Cracking
0
Ecorr = –200 mV –500 Cathode
–1000
–1500 0
50
100 Time (h) (a)
150
200
9.4 Results of the test on the specimen with cement paste contaminated with 0.4% of chloride by weight of cement that was subjected to the circulation of 1 A m–2. Potential of anodic and cathodic steel plates (a) and feeding voltage (b) in time. Depolarisation tests carried out after (c) 21 and (d) 93 h of application of the current.
© 2007, Institute of Materials, Minerals and Mining
Influence of stray currents on corrosion of steel in concrete
Feeding voltage (V)
2.4 2.3 2.2 2.1
Corrosion initiation
2 1.9 1.8
0
50
100 Time (h) (b)
150
200
1000
Potential (mV vs MMO)
Anode 500 0
Current switched off
–500 –1000 Cathode –1500 0
1
2
3 4 Time (min) (c)
5
6
5
6
1000
Potential (mV vs MMO)
Anode 500 0 –500
Current switched off
–1000 Cathode –1500 0
1
2
9.4 Continued
© 2007, Institute of Materials, Minerals and Mining
3 4 Time (min) (d)
111
112
Corrosion of reinforcement in concrete
presence of current and during depolarisation tests). Tables 9.1 and 9.2 summarise the results, showing the time required for the onset of corrosion and the charge actually circulated (during some tests, current decreased because of the high resistivity of the concrete; this occurred on specimens without chlorides before the initiation of corrosion and in specimens with chlorides only after the initiation of corrosion when a very high ohmic drop contribution occurred in the vicinity of the anode). The times and charges passed from the beginning of the test up to cracking of the specimen are also reported. For specimens made of cement paste (Table 9.1), the weight loss measured on the steel plate is also shown. In general, the measured weight loss, at least for specimens with chlorides, was in agreement with weight loss estimated by assuming that all the charge circulated after initiation led to iron dissolution. In chloride-free cement paste, the weight loss was only a fraction of the theoretical 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 chloride only after more than 200 h of application of an anodic current density of 10 A m–2 (Table 9.1). A current density of 1 A m–2 could not initiate corrosion even after 14 months (>10 000 h) of continuous application, although the charge that circulated (>10 000 A h m–2) was much higher than the charge that could 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 current density can lead to a significant decrease in the aggressiveness of stray current. Tests on steel embedded in concrete without chloride (Table 9.2) are in agreement with those obtained in cement paste, since a charge of about 5700 A h m–2 was enough to initiate corrosion with an intermediate current density (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 generated near the steel surface, and reached values close to the maximum voltage of the current feeder, i.e. 24 V). These results show that the risk of corrosion induced by stray current on steel in alkaline and chloride-free concrete is extremely low. In fact, only a high current density circulating for a very long time can induce corrosion at anodic sites. Since the reinforcement is not coated (and thus current is not forced to concentrate in small areas of defects of the coating, as occurs on steel in metallic structures), it is seldom that high current density can be induced by stray currents in concrete.
© 2007, Institute of Materials, Minerals and Mining
Nominal current (A m–2) 1*
Chloride (% wt cement)
Cycle
ti (h)
Qi (A h m–2)
tcrack (h)
Qcrack (A h m–2)
Dm (g m–2)
0
continuous 1 on – 1 off continuous continuous 1 on – 1 off continuous 1 on – 1 off continuous 1 on – 1 off continuous 1 on – 1 off continuous 1 on – 1 off continuous
>10000 >10000 3885 2200 >10000 78 2230 231 338 22 231 12 43 6 5.8 15 6
>12000 >5600 3920 2220 >5600 84 1135 2206 1690 220 1155 120 215 60 58 75 30
– – 4070 2325 – 190 2280 263 670 43 262 20 101 22 21 22 27
– – 4180 2396 – 204 1430 2520 3790 437 1760 898 508 225 129 133 130
– – 62 38 – 301† 38 371† 1070 151 139† 247† 209 182 57† 58 63
0.1 0.2 0.4 10
0 0.1 0.2 0.4
1 on – 1 off
* 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 actual current 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 not possible.
113
© 2007, Institute of Materials, Minerals and Mining
Influence of stray currents on corrosion of steel in concrete
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 steel plate; time to cracking (tcrack) and charge to cracking (Qcrack); weight loss (Dm) measured at the end of tests (usually tests were interrupted after cracking)
114
Corrosion of reinforcement in concrete
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) for initiation of corrosion on the anodic steel bar, and time (tcrack) and charge (Qcrack) for cracking of concrete Chloride (% wt cement)
Cycle
icorr (mm year–1)
ti (h)
Qi tcrack (A h m–2) (hours)
Qcrack (A h m–2)
0
continuous
1.2 1.3 0.3
1377 1370 1950
5690 5840 4750
1515 1515 2210
6060 6170 5000
1 on – 1 off 0.4
continuous 1 on – 1 off 1 on – 3 off
0.8
continuous 1 on – 1 off 1 on – 3 off
0.2 0.5 1.6 0.5 0.2 1.2
7.5 12.5 20 21 48 40
65 108 86 90 103 86
46 43 135 105 265 135
390 370 640 460 570 330
3.8 1.5 2.2 3.2 1.5 5.4
0.5 0.6 2.3 0.5 0.3 0.6
4 5 10 5 3 5
69 22 23 97 135 97
600 190 105 492 290 212
Chloride contamination Stray dc currents may have serious consequences in chloride contaminated concrete. In Fig. 9.5, the results are plotted of tests carried out in cement pastes with chloride contents up to 0.4% by weight of cement. The charge required for the onset of corrosion shows a remarkable decrease as the chloride content 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 the absence of stray current). Figure 9.5 also confirms the higher risks connected with higher anodic current densities: the charges required for corrosion initiation with a current density of 10 A m–2 are more than one order of magnitude lower than those due to 1 A m–2, i.e. times for initiation of corrosion are more than 100 times lower. Fig. 9.6 compares results with current density of 8.6–10 A m–2 obtained on steel in cement paste and in concrete with chloride contents up to 0.8% by weight of cement. Results in cement paste and concrete are in good agreement; slightly higher charges for tests in concrete may be a consequence of the slightly lower current density. It can be observed that when the chloride content approaches 0.8%, the charge for initiation of the attack reduces to a few A h m–2. This is not surprising, since 0.8% is a chloride content that may in itself be enough to initiate corrosion in non water-saturated concrete. Nonetheless, stray currents may have an adverse effect even under these
© 2007, Institute of Materials, Minerals and Mining
Influence of stray currents on corrosion of steel in concrete
115
conditions, since they can promote the initiation of incipient pits or extend the corroding area. Table 9.2 shows that, even before the tests began, several specimens had average corrosion rates of 3–5 mm year–1 (measured by polarisation resistance tests); such values show that pitting corrosion had already initiated. Following the application of the external current, the steel potential reached values of about +700 mV and, in the case of passive specimens, and only after several minutes, displayed the typical sharp decrease that has been related to initiation of corrosion (the time for initiation was estimated as 0.3–0.6 h, depending on the specimen, Table 9.2). This result suggests that, although corrosion was initiated in some spots even before testing, the application of the anodic current could extend the attack to most of the exposed surface of steel (this was 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 the effect 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 circulation of 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, since it increased the charge required for the initiation of corrosion. This effect was noticeable in cement pastes with chloride contents lower than 0.4% by 10000
Charge (A h m–2)
1 A m–2 1000
100 10 A m–2 1 on–1 off
Cycle: cont 1 A m–2 10 A m–2 10 0
0.1 0.2 0.3 Chloride (% by weight of cement)
0.4
9.5 Charge required for the initiation of corrosion on steel plates embedded in cement pastes with different chloride contents, that were polarised anodically with current densities of 1 A m–2 or 10 A m–2.
© 2007, Institute of Materials, Minerals and Mining
116
Corrosion of reinforcement in concrete 10000 Cycle: Cont. 1 on–1 off 1 on–3 off Concrete:
Charge (A h m–2)
1000
Cement paste:
100
10
1 0
0.2
0.4 0.6 0.8 Chloride (% by weight of cement)
1
9.6 Charge required for the initiation of corrosion on steel in concretes and cement pastes with different chloride contents (current density was 10 A m–2 in specimens made of cement paste and 8.6 A m–2 in specimens made of concrete).
weight of cement (Fig. 9.5). It could not be observed in tests in concrete with 0.8% chloride (Fig. 9.6), since corrosion always initiated during the first hour of the test (before any interruption could be made). The role of the anodic current in promoting the corrosion of steel in concrete has been explained by the production of acidity at the steel surface due 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 steel surface, due to electrical migration, that further favours corrosion. The gradients of ionic concentration in the pore solution produced by the anodic reaction and the electromigration near the steel surface can be mitigated by the interruption of the current. The increase in hydroxyl ions and decrease in chloride ions near the steel surface, which occurs during periods of rest, may delay the initiation of corrosion. At the end of the tests on the cement paste, analysis of pH by means of phenolphthalein and other pH indicators never showed any detectable change in pH near the steel surface (i.e. in the cement paste within 0.1 mm of the steel plate). Activated titanium electrodes fixed 1 mm from the steel surface did not show any change in potential during the tests, confirming that the pH of the cement paste did not vary macroscopically. Chloride analysis did not show evidence of any appreciable change in the chloride content, even very close to the surface of corroding steel. Therefore, possible changes of composition of the cement paste induced by stray current, which can be responsible for © 2007, Institute of Materials, Minerals and Mining
Influence of stray currents on corrosion of steel in concrete
117
corrosion initiation, would be restricted to within a very narrow distance of the steel surface.
9.3.2
Effects of alternating current
The effects of alternating current on passive steel were studied by applying 40 A m–2 ac (50 Hz) to the specimens of Fig. 9.3a made of cement pastes both free of chloride and with 0.2% of chloride by weight of cement. The corrosion rate was monitored with polarisation resistance tests carried out during interruptions of the ac current. No significant changes with respect to the initial value of 1–2 mm year–1 were observed during 60 days of application of the ac current. At the end of the tests, the steel plates were removed from the cement paste and no corrosion could be observed. Tests were also carried out on bars in concrete with up to 0.8% chloride by weight of cement subjected to 50 A m–2 ac. No effects were observed on passive steel in concrete with up to 0.4% chloride; the corrosion rate was lower than 1 mm year–1 even after 5 months of circulation of ac. The corrosion rate increased for steel in concrete with 0.8% chloride; in that case steel was corroding before the application of ac, with a corrosion rate of 1.5–2 mm year–1 and at the end of the test corrosion rates increased to 7.5–10 mm year–1. These results suggest that high ac currents may have an adverse effect on steel subjected to pitting corrosion in chloride-contaminated concrete. Such an influence was also evidenced by an increase in the macrocouple current between corroding and passive steel. For instance, Fig. 9.7 shows that the superposition of 50 A m–2 ac for 5 min led to a temporary noticeable increase in the macrocouple current between corroding steel in concrete with 0.8% chloride and passive steel in concrete with 0.4% chloride. Changes in the macrocouple current density were always observed after the application of ac pulses (even if they only lasted 20 s), showing that ac actually influences the electrochemical behaviour of corroding steel, even for a certain time after it ceases. However, such influence appeared to be rather complex; sometimes ac also led to a temporary change in the direction of the macrocouple current. Further studies are required in order to clarify the actual role of stray ac currents on corrosion of steel in concrete. Attention should also be dedicated to possible synergistic effects of ac and dc stray currents, not only in stimulating the corrosion rate of depassivated steel but also in promoting corrosion on passive steel.
9.4
Conclusions
During this study, it was found that dc stray currents could induce corrosion on steel in contact with cement paste and concrete, in the areas where the
© 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete –200
Macrocouple current (mA m–2)
10
8
Potential
–250
6
4 Macrocouple current
–300
Potential (mV vs MMO)
118
2 50 A m–2 ac 0
0
5
10 15 Time (min)
20
–350 25
9.7 Effect of 50 A m–2 ac on the macrocouple current density between corroding steel in concrete with 0.8% chloride by weight of cement and passive steel in concrete with 0.4% chloride by weight of cement. Potential of anodic steel is also shown.
anodic reaction took place. Corrosion initiated only after the stray current had circulated for a certain time. The initiation of corrosion appeared not to be simply related to the time, the current density or the circulated charge. The amount of charge leading to the initiation of corrosion showed a noticeable increase if the current density decreased, if the concrete did not contain chloride, or if the current was not continuous (i.e. if it was periodically interrupted). In chloride-free cement paste, corrosion did not initiate even after 14 months of continuous application of 1 A m–2, after a charge in excess of 10,000 A h m–2 had been circulated, while only 10 days and a charge of 2,200 A h m–2 were enough to initiate corrosion with an anodic current density of 10 A m–2. Results for steel in concrete were in agreement with those for steel in cement paste. The presence of small amounts of chlorides led 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 rate on steel already corroding in chloride-contaminated concrete, since it can promote the initiation of incipient pits or extend the corroding area. Analyses of cement paste in the vicinity of steel subjected to stray currentinduced corrosion could not detect any significant change in pH or chloride content related to the initiation of corrosion. The ac current proved to be much less dangerous than the dc; in fact, even current densities up to 50 A m–2 were not able to induce initiation of corrosion on steel. Nevertheless, it was shown that ac current may stimulate macrocouples
© 2007, Institute of Materials, Minerals and Mining
Influence of stray currents on corrosion of steel in concrete
119
that take place in the concrete between passive and corroding steel and can increase 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 in reinforced concrete structures’, in Progress in the Understanding and Prevention of Corrosion’, 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 induced corrosion in reinforced concrete structures’, L’industria italiana del cemento, April 1996, 709, 268–279. 7. L Bertolini, F Bolzoni, M F Brunella, T Pastore and P Pedeferri, ‘Stray current induced corrosion in reinforced concrete structures: resistance of rebars in carbon, galvanized and stainless steels’ (in Italian), La metallurgia italiana, 1996, 88(5) 345–351.
© 2007, Institute of Materials, Minerals and Mining
10 Assessment and monitoring of corrosion velocity of rebars and prestressing cables of a bridge D. B I N D S C H E D L E R, Swiss Society for Corrosion Protection, Zürich, Switzerland
10.1
Introduction
During the inspection of a 1 km-long twin road bridge with longitudinal prestressing, corroded and broken tendon wires were detected by chance. Further investigations showed that the corrosion had been induced by deicing salts, which reached the girder through water-bearing aeration tubes and a leaking drainage system [1].
10.2
Results of detailed corrosion inspection
Potential measurements were used to assess the state of corrosion of the reinforcement of the bridge over a surface area of about 12 000 m2. In order to evaluate the effective state of corrosion fully, 35 inspection fields were selected in areas of increased corrosion risk. In these fields, the concrete was removed to reinforce the first layer of prestressing cables, which had a concrete cover of between 35 mm and more than 100 mm. The degree of corrosion of the ducts containing the cables was assessed using the scheme given in Table 10.1. Nearly all of the inspected prestressing cables of the first layer showed corrosion. In about 35% of the inspection fields the ducts had been perforated. Broken tendon wires were detected in two more cases, but there was no evidence for hydrogen-induced cracking. In the other fields, corrosion was observed generally on a few single wires. As far as could be seen, not more than 8 of the 55 wires of one cable had been corroded at the time. In most cases, the cross sectional loss was up to about 25%. The rebars showed localised corrosion in about 25% of the inspection fields. On the outside of the box girder, the reduction of cross section was between 10 and 40%; inside the box girder it was up to 100%. It was estimated that pitting corrosion on rebars must be expected on 3% of the surface of the eastern bridge and on 1% of the western bridge. In Fig. 10.1, the empirical correlation is shown between potential and the degree of corrosion of the duct found in the 35 inspection fields of the 120 © 2007, Institute of Materials, Minerals and Mining
Assessment and monitoring of corrosion velocity
121
Table 10.1 Degrees of corrosion Degree of corrosion
Condition of duct
1 2 3 4 5
Rust points Rust stains Surface completely corroded Local perforations Extended perforations
6
Degree of corrosion
5
4
3
2
1
0 –600
–500
–400
–300 –200 U (mVCSE)
–100
0
10.1 Correlation between potential and degree of corrosion of the duct of prestressing cables.
investigated bridge. This shows quite a sharp limitation in the extent of attack towards more positive potentials. This allowed the definition of a critical potential with a certain safety margin. In areas with potentials more negative than this critical potential the presence of perforated ducts was possible at the time of the inspection. It must be mentioned that a correlation between potential and corrosion of the duct cannot be expected in any case, because the potential is dominated by the corroding rebars. This also explains the fact that, in some cases, the prestressing cables showed no corrosion or just slight corrosion even at very negative potentials. There was no correlation between the degree of corrosion of the tendon wires and the potential. In general, it must be expected that there are corroded wires when the ducts are perforated. From the correlation in Fig. 10.1 it was concluded that there are more than 200 areas on the bridge with a latent corrosion risk for the prestressing cables. © 2007, Institute of Materials, Minerals and Mining
122
10.3
Corrosion of reinforcement in concrete
Repair
Based on the results of the detailed inspection of the bridge, the possibility of a conventional repair with removal of the contaminated concrete, in some places to behind the first layer of prestressing cables, was studied initially. However, the considerations concerning the statics of the structure showed clearly that this would not be possible without detrimental effects on the load bearing behaviour of the bridge. Furthermore, the application of other methods, e.g. cathodic protection, would have been possible only with restrictions and considerable risks. Therefore, it was a question of what would be the remaining service life of the bridge without an ordinary repair. It was decided that a monitoring system should provide the necessary information about whether and to what extent sealing of the bridge deck and the repair of the drainage system would lead to drying out of the concrete and, therefore, to a reduction of the corrosion velocity. The aim of the test programme was, within three years, to collect information to allow an estimation of the remaining service life and/or to show 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 in corrosion velocity based primarily on the measurement of macrocell currents between isolated rebars and the reinforcement of the bridge as well as on the measurement of the humidity-dependent concrete resistance which is inversely proportional 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 difference between isolated rebars and reinforcement) every 6 months to give information on the corrosion velocity; ∑ Repeated potential measurements in 25 test fields at the end of the survey period to give information on the corrosion behaviour of rebars (active/passive); ∑ Visual inspection of isolated rebars (corroded area) at the end of the survey period to give information on the current density and the corrosion rate; ∑ Measurement of specific concrete resistance on drill cores in the laboratory as a function of relative humidity and chloride content to give information concerning the resistance of concrete during the drying phase.
© 2007, Institute of Materials, Minerals and Mining
Assessment and monitoring of corrosion velocity
123
10.4.2 Installation of test fields and laboratory tests on drill cores
1.20 – 1.71
–120
FO T1 Arv T2
–185
–190
tube
Aeration
For the measurements described above, it was necessary to isolate rebars electrically from the reinforcement of the bridge. The removal of drill cores at the crossing points of the vertical and horizontal reinforcement allowed the isolation of rebars of the first and the second layer with a length of 100 to 150 mm (Fig. 10.2). Electrical contact to the rebars was achieved by locking a cable in a pre-drilled hole by means of a conical pin. The isolated rebars were short-circuited to the general reinforcement during the whole of the investigation period by plugs placed in a measuring box within the box girder. The short-circuit connections were opened only for measuring purposes. The contact areas of the cables and the cores were filled with an isolating epoxy-mortar. To allow temperature correction of the resistance measurements, some of the test fields were equipped with temperature probes for measuring the temperature of the structure.
Bridge deck
–200 Arh
–085 T2
B –195
Pt100
–065
–010
–160
Girder 0.50 B F Pt 100 Arhv T1 T2 +–190
0.50
Drill core (Ø 80 mm) for chloride analyses and laboratory tests Drill hole (Ø 8 mm) for measurement of concrete humidity Temperature sensor Pt 100 (Ø 8 mm) Isolated rebar, h: horizontal, v: vertical Drill core for separation of the reinforcement Ø 80 mm Drill core for separation of the reinforcement Ø 50 mm Potential of the reinforcement
10.2 Equipment for monitoring test fields.
© 2007, Institute of Materials, Minerals and Mining
124
Corrosion of reinforcement in concrete
The laboratory tests were effected on drill cores of diameter 80 mm taken from the test fields. One half of the cores were used for chloride analyses in steps of 15 mm. The other half were equipped with copper pins so that the specific resistance at different depths could be measured by means of a 4electrode measurement. At the end of the test period the porosity and carbonation were also investigated. The specific resistance was measured on cores which had been stored at five different relative humidities between 43 and 95% and at temperatures of 15 to 20 ∞C, as well as after 30 days of storage 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 the measured resistance between isolated rebars and the reinforcement (spread resistance of the isolated rebars). For a first approximation the formula for the spread resistance of earthing rods can be used [3]. Figure 10.3 shows changes of the thus-calculated specific concrete resistance during the investigation period in some selected test fields. After an initial period all the curves show the same behaviour. The absolute values of the concrete resistance vary, as expected, over a wide range. The strong temperature
35
6000 TO1 SW35
TO2 T
SO21 25
4000
15
Sep 00
Mar 00
Sep 99
Mar 99
Sep 98
–25
Mar 98
0
Sep 97
–15
Mar 97
1000
Sep 96
–5
Mar 96
2000
Sep 95
5
Mar 95
3000
T(∞C)
rc (W m)
5000
Date
10.3 Concrete resistance as a function of time and temperature.
© 2007, Institute of Materials, Minerals and Mining
Assessment and monitoring of corrosion velocity
125
dependence of the concrete resistance is evident. Average resistance values measured in autumn (T = 1–8 ∞C) were about double those measured in spring and summer (T = 16–25 ∞C). To allow a better assessment of the changes as a function of time, the specific concrete resistance was temperature corrected according to equation 10.1 [2].
rc = rc,o ◊ e
bÊ 1 – 1 ˆ Ë T To ¯
(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 is a constant (K). The constant b was determined for each test field by linear regression from the measured values (To = 240 K). The calculated values for b are between 2700 and 5000 K. This procedure is an approximation, because the temperature dependence is overlaid by the drying process. Furthermore, the constant 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 measured 3 months (84 days) after the installation of the monitoring system. Generally, the following observations were made: ∑ All spread resistances and, therefore, the specific concrete resistances showed a more or less continuous increase during the survey period. This 6 TO1 TO2 SO21 SW35
5
rc /rc,84
4
3
2
Sep 00
Mar 00
Sep 99
Mar 99
Sep 98
Mar 98
Sep 97
Mar 97
Sep 96
Mar 96
Sep 95
0
Mar 95
1
Date
10.4 Concrete resistance as a function of time (normalised to 20 ∞C).
© 2007, Institute of Materials, Minerals and Mining
126
Corrosion of reinforcement in concrete
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 a further increase of the concrete resistance. ∑ The increase of the concrete resistance shows big variations in the different fields. The average increase was 240% for fields inside the box girder and about 90% outside. In both cases 15% of the test fields showed an increase of less than 40% outside and less than 100% inside the box girder. ∑ The spread resistance of the isolated rebars of the first and the second reinforcement layer showed only small differences. Macrocell currents Figure 10.5 shows the macrocell current as a function of time for different test fields. Anodic (in the diagrams negative) currents led to corrosion on the isolated rebars with a weight loss of 10 mg mA–1 per year. The following observations were made:
10
30
5
15
0
0
–5
–15
–10
–30
–15
–45
–20
–60
–25
TO1 TO3 SO5 SW27 T
–30 –35 –40
–75
Temperature (∞C) ISO27 (mA)
I (mA)
∑ In test fields with anodic currents the chloride content at the depth of the rebar was generally above 1% (average value 1.2%) whereas it was lower than 0.73% in test fields with cathodic currents. ∑ The macrocell currents show a temperature dependence which is much stronger than those for the resistance. At the end of the survey period, current and temperature were logged every 10 min in two test fields for 1 month. With these data a temperature correction analogue of equation
–90 –105
98 98 99 99
00 00 01 01
Mar Sep Mar Sep
Mar Sep Mar Sep
95 96 96 97 97 Sep Mar Sep Mar Sep
Mar 95
–120
Date
10.5 Macrocell currents.
© 2007, Institute of Materials, Minerals and Mining
Assessment and monitoring of corrosion velocity
∑
∑
∑
∑
127
10.1 was applied (Fig. 10.6). There was a good correlation but the ‘bvalues’ calculated were 30 to 90% higher than those calculated for the temperature dependence of the resistance. The highest measured macrocell current was 94 mA (T = 20 ∞C). In the case of homogeneous corrosion of the whole surface of the isolated rebar this would correspond to a corrosion rate of only 0.04 mm year–1. On the other hand, if the corroding area were restricted to a length of 1 cm the corrosion rate would reach 1.1 mm year–1. Within 6 years after the sealing of the bridge deck, all macrocell currents decreased considerably. However, in several test fields no noticeable change in corrosion velocity was observed during the first three years. After two years, in one test field an initially cathodic isolated rebar became anodic for about 1 year (Fig. 10.5). Note that the macrocell currents in Fig. 10.5 are 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 parameters was reached only when the potential was measured directly above the anodic isolated rebar. At the end of the investigation period, changes in macrocell current were caused in about equal amounts by changes in the concrete resistance and by variations of the potential difference between the isolated rebars and the reinforcement (Table 10.2). 45
40
35
I (mA)
30
25
20
15
10 285
290
295
300
T(K)
10.6 Correlation between macrocell current and temperature.
© 2007, Institute of Materials, Minerals and Mining
128
Corrosion of reinforcement in concrete 20
0
I (mA)
–20
–40
–60
–80 1998 2000 –100 –500
–400
–300
–200 –100 U (mVCSE)
0
100
10.7 Correlation between macrocell current and potential. Table 10.2 Changes in concrete resistance, macrocell current and potential difference between isolated rebars and reinforcement in selected time intervals (e end of time interval, b beginning of time interval) Test field
D Ue / D U b
1/(Re /Rb)
(DUe /DUb)/(Re /Rb)
Ie /Ib(measured)
TO1 TO1 TO2 TO2 TO5 SO21 SW50 SW50
0.19 0.36 0.96 0.84 1.03 0.31 0.56 1.40
0.55 0.90 0.86 0.88 0.79 0.59 0.60 1.47
0.11 0.34 0.89 0.74 0.81 0.18 0.38 2.06
0.11 0.33 0.78 0.84 0.80 0.10 0.48 2.64
Potential measurements The potential measurements were repeated in the test fields after 3 and 5 years 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 at the end of the test period in restricted areas. Unfortunately, prestressing cables were located in these areas. ∑ The extent of the potential shift toward more positive values showed large variations between the different test fields. There was a tendency for the potentials in the different test fields to approach each other with time.
© 2007, Institute of Materials, Minerals and Mining
Assessment and monitoring of corrosion velocity
129
200 100
U2000 (mVCSE)
0 –100 –200 –300 –400 –500 –600 –600
Outside Inside –400
–200 U1994 (mVCSE)
0
200
10.8 Potentials at the beginning and at the end of the investigation period.
∑ Statistical analysis of the potential measurements led to the conclusion that, in most of the test fields, the rebars were still corroding. This means that the corrosion velocity decreased but corrosion did not stop. Visual inspection At the end of the investigation period, some of the isolated rebars were removed in order to determine the corroded area. This area was between 2 and 6 cm2, which corresponds to localisation factors (ratio of total surface of the 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 of more than 60% reached more or less constant values during the investigation period, whereas cores stored at relative humidities £ 60% showed a tendency for a further increase of resistance even after 940 days (Fig. 10.9). ∑ The concrete resistance decreases with increasing depth. The resistance of 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 correlate with the chloride content of the concrete. The fact that there is no correlation
© 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete 25.0
20000
20.0
15000
15.0
10000
10.0
5000
5.0
0 0
200
400 600 Time (d)
TO 2,45% SO 21,60% T
800
Temperature (∞C)
25000
rc(W m)
130
0.0 1000
TW 1,45% SO 26,60%
10.9 Specific concrete resistance of drill cores as a function of time.
between chloride contamination and (spread) resistance was also observed in the field tests. Otherwise, it seems that, in carbonated concrete, chlorides influence 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 resistance is observed. ∑ The resistance of the cores from different test fields at a given relative humidity shows a variation of about a factor of 3 (Fig. 10.10). This can be interpreted 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 estimation of the absolute corrosion velocity. On the one hand, there is information gathered in inspection fields and, on the other hand, results from macrocell measurements or even galvanostatic pulse measurements. In the first case, the cross-sectional loss of the rebars is known, but not the beginning of the corrosion process; in the second case, the corrosion current is known, but not the corroding area.
© 2007, Institute of Materials, Minerals and Mining
Assessment and monitoring of corrosion velocity
131
30000 Avg Min Max
25000
rc(W m)
20000
15000
10000
5000
0 40
50
60 70 80 Relative humidity (%)
90
100
10.10 Specific concrete resistance of drill cores as a function of relative humidity.
In the present case, the estimation of the maximum corrosion velocity before 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 combination with localisation factors and the yearly temperature variations). The repeated potential measurements, together with the correlation between macrocell current 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 the investigation. The assessment of the actual corrosion risk and the estimation of the remaining service life of the bridge has to be done based on structural considerations and corrosion models for the progress of corrosion with time, above all in the prestressing cables. A first worst case scenario based on simple corrosion models and considering the actual corrosion velocities led to the conclusion that there is no actual or short time risk for the load-bearing behaviour 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 continuation of the monitoring. It is essential to guarantee that the humidity of the concrete does not increase again (e.g. as a consequence of new leakages).
10.7 ∑
Conclusions
The monitoring system that has been described has allowed estimates of the corrosion velocity in a way that, in combination with structural
© 2007, Institute of Materials, Minerals and Mining
132
∑
∑
∑
∑ ∑
∑
Corrosion of reinforcement in concrete
considerations and corrosion models, provides a final assessment of the remaining service life of the bridge. Within five years, the corrosion velocity decreased as a consequence of the drying out of the concrete in all test fields (with chloride concentrations near the rebar of up to 1.9%) by a minimum of 50%, but, in general, the corroding areas remained active. The drying out of the concrete is a slow process and is not complete after five years. Water infiltration (even over short periods) which leads to an increase of the humidity of the concrete must, therefore, be avoided in future. The corrosion velocity is not determined solely by the specific resistance of the concrete. The different temperature dependencies of resistance and macrocell current, as well as the fact that the potential difference between the reinforcement of the bridge and isolated rebars influences the corrosion velocity, are clear indications that electrochemical factors and, particularly, their change with time, determine the corrosion velocity. The chloride contamination (concentration) has no measurable influence on the resistance of alkaline concrete. The strong temperature dependence of the specific concrete resistance and the corrosion velocity have to be taken into account in the design of monitoring systems. Regular, periodic measurements and, preferably, a continuous data acquisition system are necessary. Isolated single measuring points do not often lead to reliable information. The time dependence of the concrete resistance and the corrosion velocity has shown large variations along the investigated bridge. For reliable, meaningful monitoring sufficient measuring points or test fields are required, 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, VulkanVerlag, Essen, 1987, 254. 4. Y. Schiegg, L. Audergon, B. Elsener and H. Böhni, Online-monitoring of the corrosion in reinforced concrete structures, Eurocorr 2001.
© 2007, Institute of Materials, Minerals and Mining
11 On-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
11.1
Introduction
In the last 10 to 15 years many reinforced concrete constructions, which had suffered damage by corrosion of the reinforcement, were repaired using various procedures (local repair, sprayed concrete, coating, or electrochemical procedures). Until now, much priority has been given to methods of condition assessment for reinforced concrete structures. The monitoring of concrete constructions (repaired or new structures) is a new approach with only limited experimental data available. In the case of unalloyed rebars, the time to initiate local corrosion attack and the stable propagation of the corrosion process is primarily influenced by environmental factors such as chloride concentration, pH, oxygen content of the pore solution or the porosity of the concrete. Since these parameters continuously change as a function of time, the exposure condition of a concrete structure is of decisive importance for the prognosis of the service life and the durability of reinforced concrete constructions. The long-term monitoring of important corrosion variables will give us a more precise understanding of the actual processes occurring during the corrosion of reinforced concrete and will lead to better knowledge for the evaluation of repair methods and for the project engineering of new constructions. The term ‘on-line monitoring’ means the continuous measurement of certain parameters 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 systematic way, 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 function of the concrete depth, up to eight chloride and resistivity sensors [1] are cast 133 © 2007, Institute of Materials, Minerals and Mining
134
Corrosion of reinforcement in concrete
into cylindrical concrete samples (Fig. 11.1). The cell factor of the resistivity sensors is first determined in the laboratory. The samples are drill cores taken from real structures or laboratory samples. A further core contains a reinforcing bar (length 40 mm, Ø 8 mm) for corrosion current measurements, a chloride sensor as well as three PT1000 temperature sensors. The reinforcing bar can be removed at a later point in time for further investigations. The lateral surfaces of the cores with the eight sensor elements are coated with an epoxy resin coating, in order to permit only one-dimensional water uptake perpendicular to the concrete surface. The instrumented cores are then fixed with a low viscosity mortar into boreholes at different locations of the concrete structure. The potentials of the chloride sensors are measured versus a MnO2reference electrode embedded in the surrounding concrete [2]. With the instrumented drill cores and additional sensors for the measurement of climatic influences, the following parameters are obtained: ∑ ∑ ∑ ∑ ∑
(W) [V (MnO2)] (A) (∞C) (%)
Electrical concrete resistance, RW Potential of the reinforcement, chloride sensor, U Corrosion current, Icorr Air and concrete temperature, TA/C Relative humidity, rh R.
5mm
1
12.5
20
15
R2
27.5 35 42.5 50
R3
60
R4
R7
20 20
T2
R5
100
100
Epoxy-coating
57.5
R6
T1
35
2.5 Chloride and resistivity sensors
T3 Rebar probe and temperature sensors
Ø75
Ø75
120 mm 40 mm
2
5
4
6
3
1
1 = Silver wire Ø0.5 mm 4 = Stainless steel tube Ø2.5 mm 2 = AgCl-coating 5 = Epoxy sealing 3 = Teflon tube (insulation) 6 = Insulation
11.1 Instrumented cores for field tests, chloride and resistivity sensor and embedded cores in an edge beam of a bridge.
© 2007, Institute of Materials, Minerals and Mining
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 and an auxiliary module). These measuring modules are protected, by means of a protection module, against overvoltages (e.g. lightning impact). Altogether 32 potentials, 8 currents, 24 resistances and 5 auxiliary variables can be measured. The recording interval can be flexibly selected in minute intervals. There is the possibility of connecting together up to 16 data collection systems by a RS485-network. The control and the current supply (12 V-battery) of the devices take place from a central, well accessible place. The selected data are stored on the pc in a text file and can be further processed with normal statistical software.
11.3
Field tests
In 1998, reinforced concrete structures on the national highway A13 in the canton of Grisons were equipped with instrumented cores and data acquisition systems for on-line monitoring. A number of typical exposure conditions and structures were chosen, where different weathering and corrosion propagation characteristics were to be expected (Table 11.1). In order to be able to examine the influence of porosity, an instrumented core from the structure 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 by temperature (as described in section 11.5). In order to describe the moisture content of the concrete and to be able to detect trends for the evolution of the corrosion 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 Edge beam Pile Arch Underside bridge deck Abutment Gallery Main girder
Beyond an asphalt layer Direct weathering / splash water Partially wet Direct weathering Indirect weathering Partially wet Splash water, indirect weathering Indirect weathering
© 2007, Institute of Materials, Minerals and Mining
136
Corrosion of reinforcement in concrete
of these measured variables. The exponential relation between the electrical concrete resistance and temperature (equation 11.1) was derived from the Arrhenius equation. In the literature values, between approx. 1500 and 5000 K for the temperature coefficient b and between 2000 and 7000 K for the temperature coefficients a have been reported [3, 4].
RT1 = RT0 ◊ e I T1 = I T0 ◊ e
b
–a
Ê 1 1 ˆ – Ë T1 T0 ¯
Ê 1 1 ˆ – Ë T1 T0 ¯
(11.1)
where RT1 and I T1 are concrete resistance and corrosion current at temperature T1 (K); RT0 and I T0 are concrete resistance and corrosion current at temperature T0 [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 humidity does not change or only slightly changes. The analysis of the performed field tests, where a measuring interval of 10 min was selected, indicated that the determination of a b-value is optimal over 24 h (day and night). As Fig. 11.2 and 11.3 show, it is essential that measurements of the temperature and the concrete resistance are performed at the same depth from the surface in order to obtain an exact determination of the b-value. It should be noted that the temperature gradient is not constant in a concrete structure and, because of the good heat conductivity of the concrete, gradient changes take place very fast. The heating and cooling rates of the concrete near the surface are different (different time constants). 9.6
In (RB) (W)
9.5
R (5-12.5 mm), T(calc) R (5-12.5 mm), T(60 mm)
9.4
b = 3250, R2 = 0.98 9.3
b = 4312, R2 = 0.78 9.2 9.1
Ponte Nanin, arch
9 0.0034 0.00342 0.00344 0.00346 0.00348 0.0035 0.00352 0.00354 1/ T (K–1)
11.2 Determination of the b-value for two different temperatures with the Ordinary Least Squares method. With an unfavourable temperature, hysterisis loops and wrong b-values result.
© 2007, Institute of Materials, Minerals and Mining
On-line monitoring of corrosion in reinforced concrete structures
20
15h10
Ponte Nanin, arch
18
T (calc) T (15 mm) T (35 mm) T (60 mm) 9.4
8h50
16
9.2
14
9 8.8
12
In (Ra) (W)
T (∞C)
R(5-12.5 mm)
137
15h10 10
8.6
0h00 8h50
8
Mar/10
Mar/10
Mar/10 Time
Mar/10
23h50 8.4 Mar/11
3000
Malabarba, abutment mortar w/c 0.5
Tcalc 20-27.5 mm
294
2500 280
2000
RB 20-27.5 mm RB T-compensated
1500 1000
266 252
500
238
0 Jan/1
Jan/24 Feb/16 Mar/11 Apr/3 Apr/26 May/19 Jun/12 Date
Concrete temperature (K)
Concrete resistivity (W m)
11.3 Concrete resistance and concrete temperatures over time in case of cooling and heating the concrete. T(calc): calculated temperature for the depth of 8.75 mm.
Jul/5
11.4 Concrete resistivity RB (raw data and with temperature compensation, T0 = 293 K) on a depth of 20–27.5 mm and calculated concrete temperature over time in the abutment of a Swiss highway bridge, January to July 1999.
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 highway bridge in a Swiss alpine region. The daily fluctuations of the temperature could practically be eliminated. Because of the constant concrete resistivity, it can be concluded that in the selected time period, January to July 1999, no deep-going humidity changes as a result of water absorption or drying processes could be measured. This is a meaningful result because this structure is exposed to indirect weathering and, therefore, the transport zone (distance from the surface where humidity changes can clearly be observed) is only small.
© 2007, Institute of Materials, Minerals and Mining
138
11.5
Corrosion of reinforcement in concrete
Results and discussion
11.5.1 Corrosion current and temperature Figure 11.5 shows the corrosion current of a corroding rebar (chloride content 1.2 mass %/c) and the concrete temperature over time in the edge beam of the Nanin bridge between July 1998 and September 2000. There is a pronounced temperature dependence of the corrosion current (daily peaks and seasonal differences). In this example, the corrosion current differs between winter and summer by a factor 3.5 to 4. Within the daily fluctuations and the seasonal fluctuations, the increase of the corrosion current during a rise in temperature reaches a factor of 1.8 to 2.0 per 10 ∞C. The corrosion processes are 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. The concrete resistances measured by means of ac resistance measurement provide information about humidity changes in the zone near the surface of the concrete. Figure 11.6 shows the concrete resistance over time at three depth levels in the edge beam of the Nanin bridge, the concrete temperature and the 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 Ponte Nanin, edge beam 0.3
40
Corrosion current (mA)
0.25
30 20
0.2
10 0
0.15
–10
0.1
Icorr
0.05 0 Jan/6
Temperature (∞C)
T (concrete, 15 mm)
Sep/30 Jan/24 May/19 Sep/12 Jan/6 Date
May/1 Aug/24
11.5 Corrosion current of a corroding rebar in the edge beam of a highway bridge (Nanin bridge) and the concrete temperature over time, July 1998 until September 2000.
© 2007, Institute of Materials, Minerals and Mining
On-line monitoring of corrosion in reinforced concrete structures
139 20 10 0 300 280
Depth 9 mm
260
10000 5–12.5 mm 12.5–20 mm 20–27.5 mm
8000
Concrete resistance (W)
R (mm) Tconcrete (K)
Amount of rain
6000 4000 2000 10000 8000 6000
Ponte Nanin-edge beam mortar w/c 0.6
27.5–36 mm
4000 2000 12/16/98
12/21/98
12/26/98 12/31/99 Date
1/5/99
1/10/99
11.6 Concrete resistances over time (depth: 5 to 35 mm) of a mortar core (w/c = 0.6) in the edge beam of a highway bridge (Nanin bridge), concrete temperature and the amount of rain over time (data from Swissmeteo), Dec. 1998 to Jan. 1999.
decreases with increasing concrete cover. Resistance measurements show that it is possible to differentiate clearly between wet and dry periods.
11.5.3 Determination of b -values and temperature compensation of the concrete resistance The b-values over 24 h (144 measured values) for the temperature compensation of the concrete resistances were extrapolated with an Ordinary Least Squaresmethod. 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 is possible to eliminate false b-values, which can occur e.g. during a period of rain precipitation occurring on the concrete structure. Figure 11.7 shows for a mortar with a w/c-ratio of 0.5 in the edge beam of the Nanin bridge the calculated b-values, the value of R2, the amount of rain and the temperature compensated concrete resistances (reference temperature 20 ∞C). The bvalues are situated between 2000 and 4000 K and are relatively constant. The strongest fluctuations occur within the first depth level (5–12.5 mm), which is the result of the changing concrete humidity due to direct weathering. In
© 2007, Institute of Materials, Minerals and Mining
140
Corrosion of reinforcement in concrete
b value (K)
5000
b -value
5-12.5 20-27.5
4000 3000 2000 1000 1.00
R2
0.99 0.98
5–12.5 20–27.5 Correlation factor
0.97 0.96 0.95
r (mm)
80
Amount of rain
Precipitation
60 40 20
Concrete resistivity (W m)
0 500 Resistivity
5–12.5 12.5–20 20–27.5
400 300 200 100 Ponte Nanin–edge beam 0 1/1/99
7/1/99
1/1/00 Date
7/1/00
1/1/01
7/1/01
11.7 Time development of the b-values, correlation factor R 2, amount of rain and temperature-compensated concrete resistances of a mortar core w/c = 0.5 in the edge beam of the Nanin bridge. To make the graph clear, only 10% of the data are plotted in the upper two diagrams. Data were recorded over 3 years.
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 used for the calculation of the temperature-compensated resistances. It is noticeable that, apart from the initial phase, strong fluctuations of the concrete humidity occur mainly in the proximity of the surface. However, the transport of water and pollutants (e.g. chlorides) to larger depths within the concrete structure takes place about twice a year.
© 2007, Institute of Materials, Minerals and Mining
On-line monitoring of corrosion in reinforced concrete structures
141
For the concrete resistivity, at a depth of 5–12.5 mm, there is a clear difference of more than 1000 K between the b-values before and after a strong precipitation of rain (Fig. 11.8). In the depth range 20–27.5 mm the difference between the b-values is smaller (400 K) because the concrete humidity before the precipitation was higher than near the surface (since drying out of the concrete occurs from the outside of the structure).
11.5.4 Temperature dependence of the corrosion current The modelling of the temperature-dependence of the corrosion current was executed in the same way as for the concrete resistances. The a-value in relation 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 the same range as the b-values. In Fig. 11.10, the a- and b-values of different concrete structures with different exposure conditions are represented. The a-values are situated between 3000 and 5000 K. In many cases, the a-values correspond relatively well with the b-values. In particular, for mortar cores with a w/c = 0.6 ratio, a clear increase in the a-value is to be determined for concrete structures with indirect weathering. This suggests that the a-value, similarly to the b-value, is humidity dependent. With the comparison between
1200 Ponte Nanin, edge beam
Concrete resistivity (W m)
1000
5–12.5 mm before precipitation 20–27.5 mm before precipitation 5–12.5 mm after precipitation 20–27.5 mm after precipitation
b = 3863 K 800
600
b = 3704 K 400
b = 3294 K 200
b = 2774 K 0 275
280
285 290 Temperature (K)
295
300
11.8 Concrete resistivity and b-values vs. concrete temperature before (March 2000) and after (April 2000) a strong precipitation of rain, edge beam of the Nanin bridge.
© 2007, Institute of Materials, Minerals and Mining
142
Corrosion of reinforcement in concrete
Icorr (mA)
0.2
Edge beam, Ponte Nanin
0.15 0.1 0.05 0
Jun/16 Aug/31 Nov/16 Feb/1 Apr/18 Jul/4
a value (K)
5000 4000
Sep/19 Dec/4 Feb/19 May/6 Jul/21 Oct/6 Date
Edge beam, Ponte Nanin
3000 2000 1000 Jun/16
Aug/31 Nov/16 Feb/1 Apr/18 Jul/4 Sep/19 Dec/4 Feb/19 May/6 Jul/21 Oct/6 Date
11.9 Corrosion current over time (Icorr ≥ 10 mA) in the edge beam of the Nanin bridge and related a-values, R 2 ≥ 0.975.
a and b, it must be considered that the corrosion current flow takes place both over the mortar and over the concrete of the structure (macrocell corrosion), while the concrete resistance is measured only within the core because of the coating on the lateral surfaces. This can entail, depending upon the quality of the concrete environment, stronger fluctuations of the a-value. The larger dispersion of the a-value is also probably to be attributed to the fact that the corrosion current is limited not only by the concrete resistance but also by the electrochemical resistances at the anode and at the cathode. Jäggi found, in his laboratory tests with mortar blocks, b-values from approx. 2250 to 3530 K and a-values between 4000 and 4400 K, whereby the stronger temperature dependence of the corrosion current in relation to the concrete resistance became clearly recognisable only at temperatures over 30 ∞C [6]. In these investigations the concrete resistance had only a subordinate influence on the temperature behaviour of the macrocell corrosion. Because of the results in Fig. 11.10, no clear conclusions can be drawn over which resistance controls the corrosion current. In addition, further analysis of the data must be done, where the voltage drop related to the corrosion current and the resistance between the corroding rebar and the cathodic reinforcement is compared with the calculated potential difference between the anode and the cathode (approximately 400 mV). It is to be expected that the proportions of ohmic and cathodic control vary with the humidity of the concrete.
11.5.5 Corrosion rate and propagation To allow prediction of future developments and evaluation of the expected cross-sectional losses at the rebars, it is necessary to obtain information about the corrosion rate from differently exposed concrete structures. For
© 2007, Institute of Materials, Minerals and Mining
On-line monitoring of corrosion in reinforced concrete structures
143
6000
b (mortar w/c = 0.5) Indirect weathering
a (mortar w/c = 0.5) 5000
Pile
Thrust
Pile
Thrust
Abutment
Arch
Edge beam 2
1000
Bridge deck 2
2000
Edge beam 1
3000
Underside bridge deck (first 2 months)
4000
Bridge deck 1
a, b value (K)
Direct weathering
Structure 6000
b (mortar w/c = 0.6)
Indirect weathering
a (mortar w/c = 0.6) 5000
4000
Underside bridge deck
Abutment
Arch
Edge beam 2
1000
Edge beam 1
2000
Bridge deck 2
3000
Bridge deck 1
a, b value (K)
Direct weathering
Structure
11.10 Mean value and standard deviation of the a- and b-values for various concrete structures, exposure conditions and mortar types; measured values over a duration of approx. 2 years were used.
this purpose, corroding rebars were removed from various constructions and the active surface was determined visually after cleaning. The mass loss and loss in cross section were calculated on the basis of the measured corrosion currents using Faraday’s law (equation 11.2). G = S M I Dt t zF
© 2007, Institute of Materials, Minerals and Mining
(11.2)
144
Corrosion of reinforcement in concrete
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 is time (s) Figure 11.11 shows the increase of the cross-sectional loss of differently exposed structures over time. While the cross-sectional losses increase rapidly in the edge beam and in the gallery column, the reinforcing bars in the underside of the bridge deck corrode only slowly (icorr < 0.01 mm year–1). The propagation curves show clearly that the exposure conditions play a decisive role in the increase of the cross-sectional loss because the chloride content in each structure is higher than 1.0 mass %/c. The maximum corrosion rate in the edge beam measured during the summer months (high temperatures and humidity) was approximately 0.4 mm year–1 (0.6 mm year–1 in an edge beam of another bridge). In winter, the corrosion rate was about two times smaller. From experience, these corrosion rates are realistic. In Switzerland, bridges that are 25 to 30 years old often show large areas with corrosion attack and high rates of corrosion have led to a decrease of about 10 mm of the rebar diameters. This means that in many cases the propagation period was much longer than the initiation period and the activation of the corrosion processes must have started only a few years after completion of the concrete construction. 0.8
Loss in cross section (mm)
Underpass, abutment portal zone Bridge, underside bridge deck Bridge, edge beam Gallery, column 0.6 0.4 mm year–1 Summer 0.4
0.2 mm year–1 Winter XD3
0.2
XD4
XD2
XD1
<0.01 mm year–1
0 Nov/19
Jul/5
Feb/18 Date
Oct/3
May/19
11.11 Corrosion propagation over time of corroding rebar probes in variously exposed concrete structures indirect weathering (XD1) spray, partially wet (XD2) splash water (XD3) direct weathering (XD4).
© 2007, Institute of Materials, Minerals and Mining
On-line monitoring of corrosion in reinforced concrete structures
11.6
145
Conclusions
With the development and use of sensor-instrumented cores, as well as a data acquisition system for the continuous measuring of corrosion-relevant variables, a new measuring technique is available for monitoring the status of reinforced concrete constructions. The results of field tests at different concrete structures permit the following conclusions to be drawn: ∑ The monitoring concept of instrumented cores is well suited to practical application and provides meaningful results. ∑ Concrete resistance and corrosion current are considerably influenced by temperature. For high quality modelling of the temperature dependence of these two parameters, characterisation of the temperature (by measurement or 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 mainly determined by the exposure conditions of the concrete structures. ∑ At bridge structures with direct weathering and splash water, corrosion rates up to approximately 0.4 to 0.6 mm year–1 are to be expected, while structures with indirect weathering clearly corrode more slowly (< 0.01 mm year–1). Therefore, the exposure conditions and, to a lesser extent, the chloride content in the concrete are decisive in controlling corrosion propagation.
11.7
Acknowledgements
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-destructive determination 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 Road Research Laboratory, Publ. No. 80, 1995, 1–35. 4. P. Schiessl and M. Raupach, ‘Influence of temperature on the corrosion rate of steel in 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 properties of hardened cement paste and mortar’, Materials Research Society, Electrically Based Microstructural Characterisation, Symposium Proceedings, Vol. 411, Boston, 1995, 407–412. 6. S. Jäggi, ‘Experimentelle und numerische Modellierung der lokalen Korrosion von Stahl in Beton unter besonderer Berücksichtigung der Temperaturabhängigkeit’, PhD Thesis No. 14058, Zürich, 2001. © 2007, Institute of Materials, Minerals and Mining
12 Integrated 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, Concrete Improvement Technologies, Germany
12.1
The task
During a project dealing with electrochemical chloride extraction from reinforced concrete one requirement was to design a generally usable, detailed and precise working system for describing and evaluating the state of corrosion in reinforced concrete structures. This should deliver, reliably, decision criteria about the state of corrosion and contain all of the data needed to configure a chloride extraction treatment. The required hardware should mostly consist of a minimal, commercially available configuration, added only by specific sensors, and the software surface should be able to manage all actions such as data acquisition, data storage, data evaluation and data export. Another requirement was to design both the hardware and software in such a way that a single user would be able to undertake all of the field operations (mainly data acquisition) conveniently and rapidly. In an extended design of the system, contractors (operators), clients and consultants should all be allowed personal access to the data to allow better interaction between the parties.
12.2
The solution
For receiving and storing all data in a common, reproducible way, any structure is to be divided logically into sections (e.g. an abutment, a column or a bridge deck). Above the surface of such a section a grid overlay will be drawn that splits the section into basic cells of 60 ¥ 60 cm. These cells will contain the main data.
*The manuscript of this chapter was submitted originally for the EUROCORR 2001 and represents the state of knowledge and development in early 2001. Scientific and technological progress since then has led to improved solutions.
146 © 2007, Institute of Materials, Minerals and Mining
Integrated system for corrosion monitoring
147
Table 12.1 Data organisation of the monitoring system Data layer Project
Comment File for structure
Globals Data Notes Docu Section 1..x Data Rebar Docu Inspection 1..y
General, non-timedependent data Administrative General remarks Photos, drawings, other documents Logical structural parts (to be named) Used materials, concrete and steel data Scaled image of the top rebar layer Photos, drawings, other documents Time-dependent data of a section Data Date, weather, circumstances Visual Cracks, spalls, concrete strength per cell Potential Potential mapping per cell Humidity Humidity values (weighed, electrically) Chloride Five layers, content of concrete and cement Carbonation Depth Docu Photos, drawings, other documents
12.2.1 Data organisation The appearance of the monitoring system follows the well-known exploration structure: on the left side the main and sub-topics are listed and, on the right side, the data from the selected topics are presented or can be edited. After setting up a project file, which is usually named as the structure, the sections can be generated. There all the important time-independent data are to be recorded. Within the sections, the inspections can be added as needed, where all the time-dependent data will be stored as shown in Table 12.1. A database in the background organises all input values and allows the data content to be extended e.g. for monitoring on-line sensors. The documentation items allow the addition of any files that contain further information such as photos, drawings, other measurement data (e.g. galvanostatic pulse measurements).
© 2007, Institute of Materials, Minerals and Mining
148
Corrosion of reinforcement in concrete
12.2.2 The interfaces Some basic import and export functionality has been implemented which makes the interaction with several external devices possible. For data acquisition, there are: – import of *.csv (table structure) files and *.bmp (bitmap image) files from the HILTI Ferroscan 10 software [1] for creating the scaled rebar image and for getting both the minimum and the average concrete cover of each cell; – on-line potential mapping via PC Card on a notebook – with a connected half-cell, potential readings can be taken directly into the monitoring system (one reading per cell); – on-line humidity readings via PC Card on a notebook – similar to the potential mapping, but with a connected 4-electrode sensor/signal generator (one reading per cell). For continuing the surveillance into an electrochemical repair, all data are to be exported. The following related data currently will be used by a special control 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, some important data, such as potential and humidity readings, can be exported for evaluation in spreadsheet programs.
12.2.3 Data input The ‘Data’ subtopics for the global, sectional and inspectional items have a spreadsheet appearance and should be populated directly as marked. Other special data inputs must be explained in more detail. Before being able to edit any further data, the grid system of a section must be generated. A distinctive point of origin – any certain marker close to the section – can be used to give a reproducible geometric system. The cells are identified by their row and column numbers. The rebar menu Before editing a cell, the related data (*.csv and *.bmp) from the HILTI Ferroscan must have been made accessible to the monitoring system. The imported table of rebars will be simply orthogonal, so some orientational corrections may be necessary. For the purposes of comparison and adjustment, the HILTI scan bitmap can be displayed in the background (Fig. 12.1). © 2007, Institute of Materials, Minerals and Mining
Integrated system for corrosion monitoring
149
12.1 Rebar edit menu with table import file and bitmap display.
Automatically, the rebar area, its relation to the concrete surface, and both minimum and average concrete cover are calculated. Furthermore, the location of the potential mapping and the position of the electrode mounting in the case of an electrochemical repair can be sketched in. If needed, this gives the opportunity to define a certain distance from the half cell to the rebars which makes the potential readings more readily comparable. The measurement positions of the potential readings are kept and displayed in the potential mapping menu for each inspection. This means that the rest potentials can be obtained from exactly the same positions with an accuracy of +/–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.
© 2007, Institute of Materials, Minerals and Mining
150
Corrosion of reinforcement in concrete
12.2 Visual edit menu with input of cracks and spalls (example).
Each cell can contain two different cracks and three different spalls. Automatically, a sum of all cracks and spalls can be found in the inspections main item and gives the total amount for a concrete repair. The value of the concrete strength can be included as well (Fig. 12.2). The damage codes used correspond with a paper by Browne and Pocock presented at the Comett Course ‘The corrosion of steel in concrete’, 1992, which refers to ACI instructions (Table 12.2). An ‘Info’ card gives the opportunity to add specific information that does not fit into the main input mask. 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 the PC card of a notebook. There is no specific type of half cell required, but this can be set up as desired (Fig. 12.3). The humidity menu For a proper interpretation of potential mapping values it is essential to consider the current humidity content of the concrete. The monitoring system allows the insertion of humidity information in three ways:
© 2007, Institute of Materials, Minerals and Mining
Table 12.2 Visual inspection codes – according to Browne and Pocock [2] Feature
Description
Cause
Details to be given
A1
Cracking (general)
Jagged separations of concrete from no gap upwards
Overload, corrosion, shrinkage
Direction, width
A2
Pattern cracking
As cracking but formed as pattern
Differential volume change between internal and external concrete
Surface area, width
B1
Exudation
Viscous gel-like material exuding through a pore
ASR
Severity
B2
Incrustation
A crust (white) on the concrete surface
Leaching of lime from cement
Severity, dampness
Rust stains
Brown stains
Corrosion of rebar
Severity
Dampness
The extent of water on the surface
Leakage, rundown
Severity
C1
Pop-out
Shallow, conical depression
Local internal pressure i.e. expansion of aggregate particle
Area, depth
C2
Spall
A fragment detached from a larger mass
Exertion of internal pressure i.e. by rebar corrosion or external force
Area, depth
C3
Delamination
A sheet spall
Exertion of internal pressure over a larger area
Area, depth
C4
Weathering
Loss of the concrete surface
Environmental action wears away the laitance and paste
Area, depth
D1
Tearing
Similar to cracks
Adhesion to slipform shuttering
Width, depth
D2
Honeycombing
Voidage between coarse aggregates
Lack of vibration
Area, depth, severity
© 2007, Institute of Materials, Minerals and Mining
151
B3 B4
Integrated system for corrosion monitoring
Code
152
Corrosion of reinforcement in concrete
12.3 Potential input menu (example).
– 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 of the electrical surface resistivity a Wenner type electrode is needed. Its diameter is suitable for measurements up to 4–5 cm depth which usually represents the steel environment. With the logged positions of potential mapping, the electrode can be placed in the same position and gives a good correlation between both readings. The chloride menu Chloride contents can be inserted manually in five layers per cell, representing a 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 known cement content and concrete raw density – to be inserted in the section data menu – the mass % related to the cement mass will be calculated and shown automatically (Fig. 12.4).
© 2007, Institute of Materials, Minerals and Mining
Integrated system for corrosion monitoring
153
12.4 Chloride edit and input menu.
The carbonation menu Here the manually determined carbonation depths are to be inserted – one per 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 appearance according to Table 12.1, and can be printed or will be available within special evaluation menus. The basic previews and printouts are orientated on the 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 menus ranking below, such as the cells with maximum and minimum chloride content, the sum of cracks and spalls per section and inspection. The number of cells with 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 relative position of potential measurement and – for later use – the selected mounting point of a repair electrode. Furthermore, a scaled image of the outer rebar layer can be generated. In this way, the true rebar layout can be viewed and compared with the construction plans.
© 2007, Institute of Materials, Minerals and Mining
154
Corrosion of reinforcement in concrete
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 measurement per 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 notice with 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 minimum concrete 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 and cataloguing functions, the next steps for an integrated evaluation of all input values 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 colour borders; – the change of potential, humidity, chloride or carbonation values of a row or 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 two highway bridges in Saxony, Germany, where investigations for selecting a reference 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.
© 2007, Institute of Materials, Minerals and Mining
Integrated system for corrosion monitoring
155
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 cap has been investigated. Despite very low potential readings (about –280 to –350 mV vs. CSE), no corrosion could be found as well as no chlorides. The concrete cover is fairly high – about 60 to 70 mm, but the rebar image looks very 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 at Uhyst, 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 splash zone of the abutments. This seems to contradict the observations made at the Neisse bridge, but obviously the chloride ingress is closely connected to the humidity content in the concrete: a bridge cap, being exposed to the weather stays relatively wet and has a low capillary suction ability; thawing salts will mostly 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.
© 2007, Institute of Materials, Minerals and Mining
156
Corrosion of reinforcement in concrete
A monthly surveillance of both abutments was started which included potential readings, humidity values and chloride measurements. There were some repaired cracks to be seen (Fig. 12.6), and on the western abutment three zones of dampness showed a considerable potential drop of more than 100 mV. As an example, an image of the potential distribution of the western abutment is given (Fig. 12.7). It shows a typical slope from the abutment head (back of the figure) down to the abutment bottom (front of the abutment). This appearance is disturbed by vertical potential breakdowns in the areas of spotted repaired cracks and zones of dampness. The potential mapping of the eastern abutment shows a perfect slope from head to bottom over the whole width. The potential values range from +80 to –240 mV vs. CSE and are fairly high. If being interpreted by the advice given in [3], no corrosion activity should be determined. However, a different evaluation results from a diagram displaying the sum frequency of all measured potential values vs. the potentials according to [4] (Fig. 12.8). Here, the three ranges of corrosion likelihood lie within the potential spectrum and show with the curve form of the western potentials that the vertical lower potentials do not result from chlorides. The investigated chloride profiles show a higher content in the splash zone – up to 2.5 wt % cement – but with the thick concrete covers, there is definitely no corrosion damage yet. When evaluating the critical potentials 100
0 –50 –100 –150 –200 –250
R5
–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
R1
12.7 Rest potentials of the western abutment.
© 2007, Institute of Materials, Minerals and Mining
Rest potential (mV vs CSE)
50
Integrated system for corrosion monitoring
157
100
Sum frequency (%)
Widerlager West Widerlager Ost
10
1
Danger of corrosion Transition stage Corrosion safe stage 0 –250
–200
–150
–100 –50 Rest potential (mV vs CSE)
0
50
100
12.8 Potential evaluation by the sum frequency.
down from –100 mV, a possible corrosion activity is concentrated on the row 0–0.6 m above ground. The structure is perfectly suited for demonstrating electrochemical chloride extraction – no concrete damage yet, different concrete zones to be controlled, a high concrete cover and a rather difficult concrete made with CEM III. It should be pointed out that the monitoring system is not designed to remove responsibility from the operator. It shall assist him in a very convenient and detailed way, but will not give any ultimate conditional evaluations or suggestions about what to do. This is not possible because corrosion continues to be a very complex matter. Nevertheless, by the unification and assisted reproduction of very different measurement data on a single location – the basic cell – the monitoring system offers an extended scale of interpretation without much effort.
12.5
Acknowledgements
This project was carried out with the financial support of the Federal Ministry of Economics of the FRG within the FUTOUR program. Furthermore, the Saxonian Highway Administration provided assistance in the field monitoring of 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.
© 2007, Institute of Materials, Minerals and Mining
158
Corrosion of reinforcement in concrete
3. H. Wojtas, Elektrochemische, zerstörungsfreie Prüfmethoden für Zustandsanalysen und 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 in Concrete.
© 2007, Institute of Materials, Minerals and Mining
13 Use of portable equipment to determine the corrosion state of concrete structures R. B Ä ß L E R and A. B U R K E R T, Federal Institute for Materials Research 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 R Force Technology, Denmark
13.1
Introduction
Much of the infrastructure in Europe has reached an age where capital costs have decreased. However, inspection and maintenance costs have grown so extensively that they constitute the major part of the current costs [1]. During a Brite/Euram Project several European partners have developed and produced an integrated monitoring system to reduce inspection and maintenance costs and disturbances to traffic. Additionally, the system will allow the operators to take preventative actions before damaging processes start. A major part of this project involved determination of the corrosion state of the rebars in new and existing structures depending on the deterioration of the concrete [2]. In addition to the evaluation of different types of sensors, newly developed portable equipment using the galvanostatic pulse technique was tested in laboratory conditions. The objective was to test the suitability of portable monitoring equipment for the unambiguous, non-destructive determination of reinforcement corrosion. Evaluation of the results from destructive testing in the laboratory provides background information of value for on-site situations.
13.2
Background
The galvanostatic pulse technique was introduced for field application in 1988 to overcome problems with interpretation of the corrosion risk to reinforcement 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 been conducted in order to allow quantitative evaluation of ongoing reinforcement corrosion [5–9]. The galvanostatic pulse method is a rapid, non-destructive polarisation technique, which has been used for evaluating the corrosion of reinforcement both in the laboratory and on site. 159 © 2007, Institute of Materials, Minerals and Mining
160
Corrosion of reinforcement in concrete
A short duration anodic current pulse is impressed galvanostatically on the reinforcement from a counter electrode placed on the concrete surface together with a reference electrode. The applied current is normally in the range 5 to 400 mA and the typical pulse duration is up to 10 s. The small anodic current causes a change of the reinforcement potential, which is recorded as a function of the polarisation time. The reinforcement is polarised in the anodic direction compared with its free corrosion potential. A typical potential transient response is shown in Fig. 13.1. When the constant current Iapp is applied to the system, the polarised potential of the reinforcement Et, at a given time t can be expressed as:
È Ê ˘ – t ˆ E t = I app Í RP Á 1 – e RP Cdl ˜ + RW ˙ ¯ ÍÎ Ë ˙˚
(13.1)
where: Rp = polarisation resistance, Cdl = double layer capacitance and RW = 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( E max – E t ) = ln( I app Rp ) –
t RP Cdl
(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
300
Emax
Polarisation (mV)
200 I · RP 100 I · RW
0 –100
Ecorr
–200 1
2
3
4 Time (s)
13.1 Typical polarisation pattern.
© 2007, Institute of Materials, Minerals and Mining
5
6
Use of portable equipment to determine the corrosion state
161
this analysis, the corrosion current, Icorr, can be calculated from the SternGeary equation (3):
I corr = B RP
(13.3)
where B is an empirical constant determined to be 25 mV for actively corroding steel and 50 mV for passive steel. The dc polarisation resistance technique with calculation of the instantaneous corrosion current (Icorr) from the Stern-Geary equation has been applied extensively since 1970. The problem is that in real structures the area of the counter electrode is much smaller than that of the working electrode (reinforcement) and the electrical signal tends to vanish with increasing distance. As a result, the measured effective polarisation resistance cannot be converted 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 the reinforcement (counter electrode diameter) are known, the instantaneous corrosion rate can be calculated. It is important to emphasise that the corrosion rate obtained is an instantaneous average rate for the confined area that strictly applies to the measuring conditions. Exposure conditions, especially temperature and concrete humidity, can alter Icorr by a factor of 10 or more. Experimental data from on-site measurements have shown that the average corrosion rates determined from RP measurements in the case of chlorideinduced localised corrosion underestimates the real corrosion rate by a factor of 5–10 or even more. From an engineering point of view such local reduction of the reinforcement cross-section is dangerous for the safety of structures,
Ag/AgCl Reference electrode
PSION WorkAbout
Counter electrode Guardring
Reinforcement
70 mm
13.2 Conditions on pulse head.
© 2007, Institute of Materials, Minerals and Mining
162
Corrosion of reinforcement in concrete
especially in zones with high tensile or shear forces. It is obvious that the wrong estimation of the amount of reinforcement for parallel or crossing bars makes the average corrosion rate too high but cracks and peeling off are also often the reason for incorrect corrosion rate estimations. For lifetime predictions, more detailed knowledge of the daily and seasonal changes of the corrosion rate is required in order to obtain meaningful values. It is essential to combine the corrosion rate measurements with supplementary mounted corrosion and chloride sensors or a number of other non-destructive evaluation 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) were exposed in the laboratory atmosphere (20–25 ∞C, 70% rh). Both chloridefree samples (KR) and specimens with mixed-in chloride (2% by mass of cement – SC) were used. After curing in the normal way, the corrosion potential and corrosion rate were measured using the galvanostatic pulse device. Reliable verification of the corrosion rate measurements was only possible by gravimetric determination of the weight loss. Therefore, at particular time intervals complete specimens were smashed and rebar weight loss was determined. Directly before the measurement the surface and the contact sponge were wetted. The sponge was cleaned (squeezed in fresh tap water) after every sample (max. 11 readings). The pulse current was 14 mA for all measurements, unless otherwise stated. At different locations, rebars A, B and C were measured separately, then A was connected to B and finally all three rebars together (Fig. 13.3), both before and after NaCl-injection. Potential was measured versus an Ag/AgCl gel electrode (incorporated within the measurement head) having a potential of 207 ± 1 mV versus a standard-Helectrode. In some of the specimens holes were drilled above rebar A. These holes were kept filled with NaCl solution and, after several days, it became apparent that the humidity had distributed itself within the specimen. Moisture marks developed on the specimen, confirming that the aggressive environment had reached the rebar.
13.4
Results
13.4.1 Influence of surface area on partly active reinforcement After stabilization had taken place measurements were performed at the locations shown in Fig. 13.3. These were compared to readings obtained in dry conditions (i. e. before NaCl injection). Both potential and current density readings showed activity on rebar A, where NaCl was injected (Fig. 13.4).
© 2007, Institute of Materials, Minerals and Mining
Use of portable equipment to determine the corrosion state
Bore holes
163
11
1, 4, 3, 5, 2, 6, 8 9 7
C
10
S
S 2% NaCl
A
B
A
B
C Cross section
13.3 Perforated specimen with measurement locations and cross section.
On dry (passive) specimens no significant difference in current density could be observed. The slight increase of potential might be caused by polarisation effects due to the pulse of the previous measurement. On wet specimens both diagrams clearly show activity on rebar A whilst B and C remained passive (separated bars). Connecting active and passive rebars does not show any effects of surface area or position. All values are dominated by the active partner. This leads to the problem that the current density value shows activity, but does not allow calculation of the corrosion rate 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 processes were observed for a longer time period using periodic pulse measurements. Significant differences were detected between samples containing chloride from the beginning (cast-in) and chloride-free samples. So, for instance, directly-chloride-exposed rebar A of specimen SC41 showed, during the first 14 days, a more rapid increase of current density values from below 1 up to 10 mA cm–2. Meanwhile, the potential dropped within a few hours from –275 to –400 mV (vs. AgCl). Subsequently, the current density values returned tentatively and varied around a low level. This return was accompanied by an increase of potential to values in the range around –350 mV. The
© 2007, Institute of Materials, Minerals and Mining
164
Corrosion of reinforcement in concrete
Current density (mA cm–2)
10 KR dry KR A-wet 1
0.1
0.01 1 A
2 B
3 C
Potential [mV (Ag/AgCl)]
200
4
5 A+B
6 (a)
7
8
9 10 A+B+C
11
100 0 KR dry KR A-wet
–100 –200 –300 –400 (b)
13.4 Potential and current readings at different locations and various rebar configurations.
variations were mainly caused by discontinuous ponding of the rebars with NaCl solution (temporary drying out). After 200 h, ponding was reduced, whereby the current density significantly decreased. Only on indirectly affected bar (SC41-B) was the current density increase more time-delayed and only by a small amount of 3 mA m–2. Potential was also less time-delayed and took > 500 h to reach values below –300 mV (Fig. 13.5). On specimen KR41 (Fig. 13.6) on directly exposed rebar A, a quick increase 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. After 1250 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 interruption of the NaCl addition. This drying-out effect correlated to the behaviour, also observed on other specimens. Rebars KR41-B and KR41-C, only indirectly affected, showed only a slight increase of the current density to values of around 0.5 mA cm–2 within the first 500 h. Values remained in that range until 2500 h. On rebar KR-C, an increase to 1 mA cm2 was measured. During exposure, rebars A and B were permanently connected. Pulse measurements in these conditions showed essentially the same behaviour of current density and potential as on directly exposed rebars A. © 2007, Institute of Materials, Minerals and Mining
Use of portable equipment to determine the corrosion state 0 SC 41 – A SC 41 – B
8 7
Potential [mV (Ag/AgCl)]
Current density (mA cm–2)
9
6 5
Drying
4 3 2 1 0
165
0
1000
2000 Time (h) (a)
–100 –150 –200 –250 –300 –350 –400 –450
3000
SC 41 – A SC 41 – B
–50
0
1000 2000 Time (h) (b)
3000
13.5 Values of (a) current density and (b) potential on rebars of 2% chloride–containing specimen SC41 with perforated concrete cover and 2% chloride ponding on rebar A.
0 KR 41 – A KR 41 – B KR 41 – C
8 7 6 5 4
Drying
3 2 1 0 0
1000 2000 Time (h) (a)
3000
Potential [mV (Ag/AgCl)]
Current density (mA cm–2)
9
KR 41 – A KR 41 – B KR 41 – C
–50 –100 –150 –200 –250 –300 –350 –400 –450
0
1000 2000 Time (h) (b)
3000
13.6 Values of (a) current density and (b) potential on rebars of chloride-free specimen KR41 with perforated concrete cover and 2% chloride ponding on rebar A.
Corresponding curves are shown in Fig. 13.7. In addition to the average values, the spread of three measurements is displayed. Also the different behaviours of specimens SC41 and KR41 is obvious. The surface appearance of the specimens was analysed after crushing the concrete cover and pickling off the corrosion products. In both examples, the more severe corrosion on rebar A (where NaCl solution was injected) was visible. Almost uniform attack was apparent which allowed the measured values 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. This means that the measured values cannot be related to the whole rebar surface.
© 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete 0
10 9 8
SC 41 A+B KR 41 A+B
Potential [mV (Ag/AgCl)]
Current density (mA cm–2)
166
7 6 5 4 3 2 1 0
SC 41 A+B KR 41 A+B
–50 –100 –150 –200 –250 –300 –350 –400 –450
0
1000 2000 Time (h) (a)
3000
0
1000 2000 Time (h) (b)
3000
13.7 Values of (a) current density and (b) potential on rebars of specimens SC41 and KR41 with perforated concrete cover and 2% chloride ponding on rebar A, rebar A and B connected.
Therefore, measured values will only be discussed where uniform attack took place. By integration of the current ‘considering the polarised area’ the amount of charge (Q), transferred during the experiment, can be obtained. Dividing that 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 the rebar (A = 31.4 cm2), the average current density, im, is obtained, which would cause the same charge transfer by integration over time. For comparison, block SC was destroyed after 4 months, and the weight loss of the rebars was determined by pickling off the corrosion product. From the weight loss thus obtained and the exposure time, the corrosion current density icorr was calculated by Faraday’s law. Comparison of current densities obtained by pulse measurement and by integrated weight loss is shown in Table 13.1. The values show a relatively good correspondence.
Table 13.1 Current density values calculated from weight loss and obtained from GPM measurements SC 41
Weight loss whole bar
Current density from weight loss
Description
Dm (g)
icorr (mA cm–2 )
Mean current density calculated from GPM over 4 months im (mA cm–2)
bar A (2 cm depth) bar B (3 cm depth) bar A + B (centre)
1.30 1.36 2.66
4.8 5.0 4.9
3.6 1.5 5.0
© 2007, Institute of Materials, Minerals and Mining
Use of portable equipment to determine the corrosion state
13.5
167
Discussion
Investigations in laboratory conditions clearly show that GPM is suitable for evaluating the real extent of corrosion of reinforcement in concrete. Active and passive conditions can be detected exactly. In addition, the influence of corrosion stimulation by ponding with NaCl solution has been proven. The influence of discontinuous wetting (with temporary drying phases) on the corrosion behaviour of reinforcing steel could be detected by periodic GPM measurements. So GPM is an important addition to corrosion potential measurements. Even though separate investigations in laboratory conditions indicated a relatively good correlation between current densities calculated from weight loss and obtained by GPM, it needs to be emphasised that the determination of 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 polarisation resistance measurements, and lifetime estimates based on these values can be 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 an assumed one based on the size of the polarised area, taking into account the field distribution and position of the measurement head rather than the actual corroding 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 provide an instantaneous indication of the actual corrosion situation, which is significantly affected by the condition of the concrete (for instance moisture content, pH-value, and chloride content). Also, on one hand the corrosion rate can be reduced by the formation of corrosion products on the surface (providing a barrier to diffusion). However, on the other hand, promoting the formation of thick corrosion product layers can accelerate the corrosion (by crack formation, the detachment of concrete, and hygroscopic salt effects). These systematically and randomly influencing factors can cause the corrosion current to be miscalculated by up to an order of magnitude, ruling out the possibility of accurate lifetime estimation. However, it should not be overlooked that the influence of these factors will be much lower under on-site conditions, because the separation between active and passive areas will be greater. Therefore, the interaction between different areas that occurred in the laboratory experiments will be significantly smaller. What is more, uniform corrosion attack is observed in practice contrary to what was observed in the laboratory tests. However, in on-site conditions other factors tend to make estimation of corrosion more difficult than under well-defined laboratory conditions.
© 2007, Institute of Materials, Minerals and Mining
168
Corrosion of reinforcement in concrete
Pulse measurements do not provide information about past or future corrosion conditions and their development. These are affected by changing conditions in the surrounding concrete, which are subjected to statistical, random, and real, physical deviations. Only information on the current situation is available. That is why there is a need for long-term observation combined with periodical pulse measurements to provide information about corrosion trends (i.e. whether steady, increasing or decreasing). A semi-quantitative estimate of the actual corrosion rate is all that is possible based on GPM measurements when further considerations like the potential of reinforcement and the concrete humidity, etc., are taken into account.
13.6
Conclusions
A newly developed, easy-to-handle portable instrument using the galvanostatic pulse method has been tested on several materials in different environmental conditions to provide quick information on the actual corrosion behaviour of reinforcement in concrete. Key parameters were concrete composition, rebar conditions, humidity and temperature. Special attention was paid to the comparison of instrument readings with actual behaviour. Various combinations were tested and the response of the instrument under various circumstances was 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 showed very good correlation with the real corrosion state and enabled users to estimate the corrosion behaviour of reinforcement. However, exact lifetime estimations using nothing but GPM results are only partly successful and need much further consideration because the measurement conditions (moisture content, temperature, unknown active area, etc.) strongly affect the values obtained.
13.7
Outlook
The work described here is continuing, including investigations to determine the area influence on lifetime prediction. The observed behaviour in laboratory tests will be compared to results from on-site investigations. The results will be presented in future publications. These results combined with others relating to deformation and vibration probes, obtained during this BRITE/EURAM project will contribute to the development of an integrated corrosion monitoring system so that end-users will be able to optimise their maintenance management systems and, therefore, costs and disruption to traffic can be reduced.
© 2007, Institute of Materials, Minerals and Mining
Use of portable equipment to determine the corrosion state
13.8
169
Acknowledgements
This work was funded by the European Community as a Brite/Euram project ‘Smart Structures’, contract number BRPR-CT98-0751. The authors are grateful for 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 Highway Bridges, HMSO, London, UK, April 1983. 2. Broomfield, J. P., Corrosion of Steel in Concrete, Understanding, Investigating and Repair, 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 in Concrete, 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’, Zerstörungsfreie Prüfung an Stahlbetonbauwerken, SIA Dokumentation D020, Schweizer Ingenieurund Architektenverein, Zürich, 1988. 9. Newton, C. J. and Sykes, J. M., ‘A galvanostatic pulse technique for investigation of steel corrosion in concrete’, Corros. Sci.,1988, 28, 1051–1074.
© 2007, Institute of Materials, Minerals and Mining
14 Corrosion inhibitors for reinforced concrete – an EFC state of the art report B. E L S E N E R, University of Cagliari, Italy and ETH Zürich, Switzerland
14.1
Introduction
In general, reinforced concrete has proved to be successful in terms of both structural performance and durability. However, there are instances of premature failure of reinforced concrete components due to corrosion of the reinforcement. The two principal factors provoking corrosion are the ingress of chloride ions from deicing salts or sea water and the reaction of the alkaline pore solution with carbon dioxide from the atmosphere, a process known as carbonatation. Despite the huge demand, a simple, cheap, and reliable technique which either protects the steel from corrosion or at least lowers its corrosion rate is still lacking. Over the past decade, however, the concrete repair industry has developed novel techniques that are claimed to prevent, or at least to reduce, 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 as a preventative measure for new structures (as an addition to the mixing water) or as surface applied inhibitors for preventive and restorative purposes. Addition to the mixing water does not require any additional working steps and allows a simple handling of the inhibitor, unless it affects the properties of the cement paste adversely. Application from the concrete surface could be a promising technique to protect existing structures from corrosion or to increase the lifetime of structures that already show corrosion attack. The application of inhibitors on the concrete surface requires the migration of the substance to the rebar where it has to reach a sufficiently high concentration to 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 the successful use of corrosion inhibitors, implying that this success is relevant to applications in reinforced concrete. This is a priori not correct because the mechanism of inhibitor action is completely different: ∑ in the oil and gas industry applications (and most others), the steel to be protected is uniformly corroding in slightly acidic or neutral media. Thus, 170 © 2007, Institute of Materials, Minerals and Mining
Corrosion inhibitors for reinforced concrete
171
the inhibitors have to protect the bare metal surface, e.g. as adsorption inhibitors acting specifically on the anodic or on the cathodic partial reaction of the corrosion process or as film forming inhibitors blocking the surface more or less completely [1, 2]. Usually, a reduction in corrosion rate of 95 to 99% is achieved with a very small inhibitor concentration in the order of 10–3 – 10–2 mol L–1. ∑ in contrast, steel in concrete is in a highly alkaline environment; the high concentration of hydroxyl ions acts as a passivation-promoting inhibitor and, indeed, steel in concrete is passive, being protected by a thin oxyhydroxide layer. This is the starting point for any mechanistic action of inhibitors in concrete. ∑ inhibitors for chloride-induced pitting corrosion have received far less study [3]. Inhibitors for pitting corrosion can act by forming a film before the ingress of chlorides, by buffering the pH in the local pit environment, by competitive surface adsorption processes between inhibitor and chloride ions or by competitive migration of inhibitor and chloride ions into the pit. Another point concerning the terminology of ‘inhibitors’ must be clarified: a corrosion inhibitor prolongs the service life due to chemical/electrochemical interaction with the reinforcement. Any other substances that may prevent the 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 be corrosion inhibitors. Most of the results published in the literature and reviewed recently [4, 5] are from laboratory studies involving solution experiments or relatively small mortar samples. Long term performance results are available for admixed inhibitors only, in particular calcium nitrite [6]. Results from well documented field tests involving surface applied inhibitors are rare. There are, however, other difficulties in obtaining unambiguous, conclusive results on the performance of corrosion inhibitors on reinforced concrete structures: ∑ most of the ‘inhibitors’ available under different trade names are blends of essentially unknown composition that could be changed without notice. This makes even laboratory experiments difficult. ∑ sometimes the use of surface-applied inhibitors is recommended only in conjunction with other corrosion protection methods, such as hydrophobation of 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 corrosion inhibitors for steel in concrete [5] and the literature results reviewed therein. In particular, calcium nitrite (DCI), the migrating corrosion inhibitors (SIKA or MCI) and MFP (monofluorophosphate) are addressed. The problem of testing different inhibitors for steel in concrete is addressed and – as far as
© 2007, Institute of Materials, Minerals and Mining
172
Corrosion of reinforcement in concrete
available – results from field tests with inhibitors are presented. Finally, a critical 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 phase corresponds to the initiation time, ta or tb,c taken for chlorides or CO2 to penetrate the concrete cover in sufficient quantities to destroy the passive film (depassivation). The second phase covers the period of active corrosion after ta or tb,c until the point where the serviceability of the structure are affected (loss of load bearing capacity, spalling or delamination) and maintenance or repair is needed. The length of this period is determined by the rate of corrosion (slope a, governed by the oxygen availability, humidity and temperature) and the ability of concrete to withstand internal stress. In view of this general picture of corrosion, admixed (preventative) corrosion inhibitors 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 prevent or 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 initiation period (t < to), the mode of action is in principle identical (given the necessary
End of service life
Degradation
Limit state
a
b c
t0 Initiation
t0b
Time Propagation
14.1 Lifetime of a reinforced concrete structure according to Tuutti [7] adapted on the action of corrosion inhibitors. Time of corrosion initiation (depassivation) is t0, or t0b, time for maintenance is ta, tb, tc.
© 2007, Institute of Materials, Minerals and Mining
Corrosion inhibitors for reinforced concrete
173
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
Samples with active potential (%)
The most frequently used technique is addition of the inhibitors to the mixing water of concrete as admixtures for new structures in order to prevent or at least delay the onset of corrosion. Calcium nitrite is the most extensively tested admixed corrosion inhibitor [7] and has – when applied according to the 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 passivator due to its oxidising properties and stabilises the passive film [8]. All investigations have revealed a critical concentration ratio (threshold value) between inhibitor (nitrite) and chloride of about 0.6 (with some variation from 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 chemical reaction with fresh concrete [10]. Laboratory studies of the preventive inhibitor action against chloride-induced corrosion have shown that by applying several intense flushings before the ingress of chlorides [11] it can prevent the onset of corrosion during a test duration of 90 days, even at chloride concentrations as high as 2% by weight of cement. A critical concentration ratio MFP/ chlorides greater than 1 must be achieved, otherwise the reduction in corrosion
99.9 95.5 90.0 80.0 60.0 40.0
0% Ca(NO2)2 0.5% Ca(NO2)2 2.0% Ca(NO2)2 4.0% Ca(NO2)2
20.0 10.0 5.0
1.0 0.91.0 2.0 3.0 4.0 6.0 8.0 0.0 Time for noble-to-active potential shift (days ¥ 10–2)
14.2 Time to corrosion initiation of steel in mortar samples with admixed Ca(NO2)2 inhibitor exposed to sea water [9].
© 2007, Institute of Materials, Minerals and Mining
174
Corrosion of reinforcement in concrete
rate is not significant [11]. In solutions containing Ca(OH)2, MFP is reported to react with the calcium ion to form insoluble products such as calcium phosphate and calcium fluoride [10, 11], thus the active substance, the PO3F2– ion, disappears from the pore solution. The main problem in using MFP as a surface-applied liquid is the penetration to the reinforcement in order to act as inhibitor. In early field tests in Switzerland, insufficient penetration of MFP 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 MFP applications on the surface. In more recent field applications [13], e.g. on the Peney Bridge near Geneva [13], concrete buildings and balconies, MFP was applied onto cleaned, dry concrete surfaces in up to 10 passes and the concrete was 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 on the 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 inhibitor blends of usually complex formulations [14]. These blends are often not sufficiently well described so most of the published work has been undertaken with commercially available systems. A comparative test with different corrosion inhibitors [15] showed very good corrosion inhibition of the commercial inhibitor blend at a high concentration, pure dimethylethanolamine instead being practically ineffective. Recent research work at ETH Zürich investigating a commercial migrating inhibitor blend has shown that the blend can be fractionated into a volatile (dimethylethanolamine) and a nonvolatile (benzoate) component [16]. For the complete prevention of corrosion initiation in saturated Ca(OH)2 solution with 1M NaCl added, the presence of both components at the steel surface in a concentration ratio of inhibitor/ chloride of ca. 1 was necessary: neither component of the inhibitor when present alone in solution could prevent initiation of corrosion (Fig. 14.3). Modern surface analytical techniques, such as XPS [17], have shown that for the formation of a significantly thicker organic film on iron in alkaline solutions both components of the commercial inhibitor blend have to be present. This might be significant for the mechanism of the inhibitor action. The inhibitor blend in the recommended dosage could delay the average time 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 were tested using different admixed dosages (Fig. 14.5): calcium nitrite (DO), an organic corrosion inhibitor (ORG1), an inhibitor based on alkanolamine (ORG2) and another calcium nitrite product (CN2). All inhibitors could delay the onset of corrosion but only the commercial calcium nitrite with the highest dosage gave a significant improvement. The same alkanolaminebased commercial inhibitor blend was tested as an admixture in mortar and
© 2007, Institute of Materials, Minerals and Mining
Corrosion inhibitors for reinforced concrete
175
Potential (mV SCE)
0 1
–100
2 No inhibitor Non-volatile constituent Volatile constituent
–200 –300 –400
Opening of the cell
–500 –600 0
10
20
30
40 t (d)
50
60
70
14.3 Corrosion potentials of rebar samples in solutions containing the two components of the inhibitor: 䊏 no inhibitor, ∑ volatile constituent, D non-volatile constituent 1: prepassivation in saturated Ca(OH)2; 2: immersion in saturated Ca(OH)2 + 1M NaCl [16].
Corroding samples (%)
100 80 60 40 Blank 0.35 kg m–3 1.75 kg m–3 8.75 kg m–3
20 0 0
50
100
150
200 t (d)
250
300
350
14.4 Percentage of corroding rebars in mortar vs. time of cyclic chloride treatment [16].
concrete samples exposed to chlorides [19]. After one year of testing, corrosion had started in specimens with w/c = 0.6, the chloride threshold values for the inhibitor-containing samples are in all cases higher (4–6% Cl– by weight of cement) compared with the control samples (1–3% Cl–). Prevention or at least prolongation of the onset of corrosion has been reported also for an organic corrosion-inhibiting admixture (OCI) proposed in 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 aliphatic carboxylic acid with a mono-, di- or tri-hydric alcohol and the water phase contains a saturated fatty acid, an amphoteric compound, a glycol and a soap. The admixture is added to concrete before placement. Upon contact with the high pH environment of concrete the waterproofing ester component
© 2007, Institute of Materials, Minerals and Mining
176
Corrosion of reinforcement in concrete Control
Control CN 10
Calcium nitrite DCl
CN 20
Inhibitor (L m–3)
CN 30 ‘Pore blocker’
ORG1 5 ORG1 7 ORG1 9 ORG2 20 ORG2 25 ORG2 30 CN2 3 CN2 4 CN2 5 0
100
200
300 400 Time (d)
500
600
700
14.5 Time to corrosion initiation of four steel bars in mortar blocks exposed to cyclic ponding with chloride solutions [18] for different inhibitors admixed to the mortar in three dosages.
becomes hydrolysed, forming carboxylic anions that are precipitated in the presence of calcium ions as a hydrophobic coating within the pore system, reducing ingress of water and chlorides into the concrete [21, 22]. This poreblocking effect is not a true corrosion inhibition. In summary, to prevent or strongly delay the onset of pitting corrosion on passive steel in alkaline solutions or mortar, all investigations – independent of 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 inhibitor concentrations have to be present in the pore water of concrete in order to act against chlorides penetrating from the concrete surface. To avoid chloride ingress and, thus, the use of excessively high inhibitor concentrations, the use of admixed inhibitors is recommended only together with high quality concrete [6]. Too low a concentration of certain inhibitors may cause an increased localised corrosion rate, as has been found in laboratory studies with 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 treatment with subsequent transport of the inhibitor to the corroding steel with the
© 2007, Institute of Materials, Minerals and Mining
Corrosion inhibitors for reinforced concrete
177
5 4
Before After
3 2 1 0
0% 0.6% 1.2% 2.4% Chloride concentration (w%/cem) (a)
Current density (mA cm–2)
Current density (mA cm–2)
effect of stopping or at least reducing ongoing corrosion. Several laboratory and field tests have been performed to investigate this particular situation. For monofluorophosphate, repeated drying and MFP-immersion cycles have been found a suitable method to allow the penetration of the inhibitor to the steel, but high concetrations and long treatments are needed to significantly reduce active corrosion due to carbonation [24]. In recent research [5] at Aston University, 15% by weight solutions of MFP were applied repeatedly to reinforced concrete specimens (water to cement ratio, w/c, 0.65, cover 12 mm) with various levels of chloride contamination. The embedded bars, precorroded under cyclic wetting and drying conditions for about 6 months before the MFP treatment, did not exhibit marked reductions in corrosion rate [2] (Fig. 14.6). Experiments with a commercial migrating inhibitor blend [16] have shown that the polarisation resistance measured after the onset of corrosion in solution increases with the inhibitor concentration (Table 14.1); both the volatile and the non-volatile fraction could reduce the corrosion rate slightly compared with the non-inhibited solution [4, 16]. In mortar experiments with cyclic ponding in 6% chloride solution, however, the polarisation resistance after the onset of corrosion did not change with inhibitor concentration and was
5 4
Before After
3 2 1 0
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 after treatment with inhibitor: (a) MFP, (b) proprietary alkanolamine based inhibitor, after Page et al. [4]. Table 14.1 Average polarisation resistance of three rebar samples after the addition of 1M NaCl to sat. Ca(OH)2 solution with inhibitor [15] Inhibitor (wt %)
Rp (kW cm2)
10 1 0.1 0
490 ± 80 11 ± 3 2±1 3±1
© 2007, Institute of Materials, Minerals and Mining
178
Corrosion of reinforcement in concrete Table 14.2 Average polarisation resistance of rebar samples in mortar with different inhibitor concentration after 343 days cyclic treatment in 6% chloride solution Series
Inhibitor (wt%)
Rp (kW cm2)
1 2 3 4
0 0.015 0.075 0.375
5.2 6.0 5.6 6.2
± ± ± ±
2.2 1.4 1.6 0.9
similar to the mortar without inhibitor (Table 14.2) [16]. Thus, a delay in the onset of corrosion is obtained (Fig. 14.4) but no reduction in the corrosion propagation rate. Alkanolamine-based inhibitors have been tested in similar conditions. For ongoing chloride-induced corrosion with a chloride level of ca. 1–2% a reduction 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 of another proprietary migrating inhibitor blend for surface application was tested in solution. After the addition of the inhibitor an increase of the polarisation 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, cover 25 mm) did not show any increase in polarisation resistance after inhibitor treatment despite low cover and porous mortar [26]. A comparative study performed with four admixed inhibitors [18] found a similar result: all the inhibitors were found to have little detectable effect on the corrosion rate of the embedded steel once active corrosion had been initiated. For ‘penetrating inhibitors’ the favourable effects found in solution do not occur when applied to 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 inhibitors were found to be virtually ineffective at chloride concentrations of 1–2% by weight of cement (Fig. 14.7) [25]. Other field tests with proprietary vapour-phase inhibitors [27] in a parking garage with chloride-contaminated precast slabs did not show encouraging results. Corrosion rate measurements showed a reduction of 60% in areas with initially intense corrosion but also an increase in areas with low corrosion rates. On structures dating from 1960 with admixed chloride content >1%, already featuring patch repairs, a three-year corrosion rate survey showed lower corrosion rates in the treated areas compared with the untreated ones but cracking and spalling nevertheless increased in the treated areas [28].
© 2007, Institute of Materials, Minerals and Mining
Corrosion inhibitors for reinforced concrete 99.99
99.99 99.9
99.9
FG 97 FG 99
MFP 97 MFP 99
99
Frequency distribution (%)
99
Frequency distribution (%)
179
95 90 80 70 50 30 20 10 5
95 90 80 70 50 30 20 10 5
1
1
.1
.1
.01 –600 –500–400–300–200–100 0 Potential [mV (CSE)]
100
.01 –600 –500–400 –300–200 –100 0 100 Potential [mV (CSE)]
14.7 Cumulative frequency distribution of half-cell potentials measured on the chloride contaminated field tests for surface applied inhibitors before and two years after application: (a) SIKA Ferrogard, (b) MFP [25].
14.6
Transport of the inhibitor into mortar or concrete
It is claimed for several inorganic and organic inhibitor blends that these inhibitors can be applied to existing reinforced concrete structures and the corrosion inhibitor will be carried by water or by vapour-phase migration into the proximity of the reinforcing steel [9, 13]. Several diffusion experiments showed that alkanolamine-based inhibitors in particular can diffuse through the concrete although great discrepancies in the measured diffusion rates exist. This might be due partially to the different experimental setups (humidity) and measuring techniques used. In addition, it is difficult to determine the diffusion rate of an inhibitor blend of unknown composition. A detailed study on the transport of a proprietary amino-alcohol-based inhibitor (FG903) into cement paste and mortar has been reported by Tritthart [29]. The results showed that both the amount and the rate of inhibitor ingress into the alkaline cement paste was higher for the pure amino alcohol compared with the inhibitor blend also containing phosphates (Fig. 14.8a). This discrepancy could be explained by a reaction of the inorganic phosphate component with the calcium ions in the fresh cement paste blocking further ingress of the inhibitor. To avoid a reaction with calcium ions, the transport of the inhibitor was studied on cores taken from a 100 year old, fully carbonated concrete structure, varying the dosage and the way of inhibitor application (Fig. 14.8b). The recommended dosage (500 g m–2) and way of application
© 2007, Institute of Materials, Minerals and Mining
180
Corrosion of reinforcement in concrete 8 ¥ 104 Amino 28 days Amino 50 days FG 903 14 days FG 903 50 days
Concentration (ppm)
7 ¥ 104 6 ¥ 104 5 ¥ 104 4 ¥ 104 3 ¥ 104 2 ¥ 104 1 ¥ 104 0
0.5
2
3.5
5 6.5 Depth (cm) (a)
8
9.5
8 ¥ 104
Concentration (ppm)
7 ¥ 104 6 ¥ 104
Phosphorus 0–1 cm Phosphorus 1.5–2.5 cm Aminoalcohol 0–1 cm Aminoalcohol 1.5–2.5 cm
5 ¥ 104 4 ¥ 104 3 ¥ 104 2 ¥ 104 1 ¥ 104 0 500 g (m–2)
1000 g (m–2)
1500 g (m–2) (b)
28 d
50 d
14.8 Transport of a proprietary aminoalcohol-based inhibitor (SIKA Ferrogard 903) into (a) alkaline cement paste and (b) fully carbonated concrete cores, from Tritthart [29].
(several brushings) showed only a moderate concentration of the amino alcohol in the first 15 mm. An increase in the dosage to 1500 g m–2 increased the 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 inhibitor blend can be analysed as in the case of a proprietary migrating corrosion inhibitor [16, 30]. Using the amine electrode, the diffusion of the volatile part of the inhibitor through a mortar disk could be measured [16, 30], no information of the diffusion of the non-volatile part could be obtained, and
© 2007, Institute of Materials, Minerals and Mining
Corrosion inhibitors for reinforced concrete
181
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 steel surface to get an inhibiting effect (Fig. 14.3), but only one component easily diffuses through the porous concrete, may explain the discrepancy between solution experiments and mortar or field tests [4, 16, 26]. A high diffusion rate – logically – does not depend on the diffusion direction; so it has been found 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 been established, there remain two critical points for successful and reliable application on reinforced concrete structures: ∑ The inhibitor has to be present at the reinforcing steel in sufficiently high concentration with respect to the aggressive (chloride) ions over a long period 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 new structures, the inhibitor dosage thus has to be specified with respect to the expected chloride level for the design life of the structure. Surface-applied inhibitors on existing structures may present even more difficulties in achieving the necessary concentration at the rebar level. Firstly, because chloride contamination 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 crucial to specify the critical concentration to be achieved at the rebar level and not – as in the application notes of surface applied inhibitors – an average weight of inhibitor solution to be applied per m2 concrete. This is usually omitted, in part due to the lack of analytical methods to measure the inhibitor concentration. Regarding long term durability, it has to be taken into account that 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 to assess the inhibitor action on rebar corrosion ‘on site’. The interpretation of half-cell potential measurements may present difficulties due to changes in
© 2007, Institute of Materials, Minerals and Mining
182
Corrosion of reinforcement in concrete
the concrete resistivity. Further, a reduction of corrosion rate due to an inhibitor action may not be reflected straightforwardly in the half-cell potential: potentials may become more negative or more positive after inhibitor application, depending on the mechanism of the inhibitor action. Shifts in the half-cell potential may also occur due to the wetting and drying of the concrete [31, 32]. LPR measurements are considered suitable for on-site testing [28], but results of corrosion rate measurements on site depend on the type of device used for the measurements and can be interpreted so far only by specialists. The main problems are the daily and seasonal changes of the corrosion rate with temperature and concrete humidity making it difficult to evaluate inhibitor action. Macrocell current measurements between isolated anodes (located and instrumented before inhibitor application) and the surrounding cathode may give the most indicative results [25] but can be installed only on test sites.
14.8
Conclusions
The use of corrosion inhibitors could be a promising technique in restoring reinforced concrete structures, offering benefits such as reduced costs and inconvenience of repairs. It has, however, to be taken into account that the use of corrosion inhibitors in repair systems is far less well-established than their applications as admixtures in new structures. This paper presents the results and conclusions based on the available literature. Briefly, admixed inhibitors with the correct dosage can strongly delay the onset of chloride-induced corrosion. Once corrosion started no significant reduction in corrosion rate has been found. The overall performance of surface-applied organic and inorganic corrosion inhibitors intended to reduce ongoing chloride-induced corrosion cannot be considered positive, for the case of corrosion due to carbonation there remain at least some doubts. Engineers and contractors working in the area of concrete maintenance should be aware of the fact that the performance of proprietary corrosion inhibitors in repair systems marketed under different trade names is not yet documented by independent research work, especially when considering field 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.
© 2007, Institute of Materials, Minerals and Mining
Corrosion inhibitors for reinforced concrete
183
4. C. L. Page and V. T. Ngala and M. M. Page, ‘Corrosion inhibitors in concrete repair systems’, Mag. Concrete Res, 2000, 52, 25–37. 5. B. Elsener, Corrosion Inhibitors for Steel in Concrete – State of the Art Report, EFC Series 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 och Betonginstituet, Stockholm. 8. B. El-Jazairi and N. Berke, ‘The use of calcium nitrite as a corrosion inhibiting admixture to steel reinforcment in concrete’, in Corrosion of Reinforcement in Concrete Construction, 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 of reinforcing 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 Protection of 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 of organic amines in simulated concrete environment’, in Proc. Int. Conf. on Understanding Corrosion Mechanisms of Metals in Concrete – a Key to Improving Infrastructure Durability. Massachusetts Institute of Technology MIT, Cambridge, 1997, section 6. 16. B. Elsener B, M. Büchler, F. Stalder and H. Böhni, ‘A migrating corrosion inhibitor blend 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-SIMS analyses 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 corrosion rates of steel in mortar’, Cement Concrete Res. 2001, 31, 713. 19. P. H. Laamanen and K. Byfors, ‘Corrosion inhibitors in concrete – alkanolamine based inhibitors’, Nordic Concrete Res No. 19, 2/1996, Norsk Betongforengingk Oslo, 1996. 20. G. S. Bobrowski, M. A. Bury, S. A. Farrington and C. K Nmai, ‘Admixtures for inhibiting 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.
© 2007, Institute of Materials, Minerals and Mining
184
Corrosion of reinforcement in concrete
22. C. K. Nmai and D. McDonald, ‘Long term effectiveness of corrosion inhibiting admixture and implications for the design of durable reinforced concrete structures: a laboratory investigation’ RILEM Int. Symp. on the Role of Admixtures in High Performance 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 Urban Transportation Challenges’, Lucerne 18–21. Sept. 2000, (on CD). (See also this volume Ch. 18.) 26. B. Elsener, M. Bürchler, F. Stalder and H Böhni, ‘A migrating corrosion inhibitor blend 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 to corroding reinforced concrete structures, CORROSION 2000, paper 0791, NACE International 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 Workshop Corrosion 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 reinforced concrete’, ConChem J., 1993, 1, 38–42. 31. B. Elsener and H. Böhni, ‘Half cell potential measurements – from theory to condition assessement of RC structures’, Proc. Int. Conference ‘Understanding Corrosion Mechanisms 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.
© 2007, Institute of Materials, Minerals and Mining
15 Mixed-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
15.1
Introduction
The corrosion of reinforcement in concrete is the most important cause of premature failure of reinforced concrete structures world-wide and became of great interest in the late 1980s and early 1990s when its huge economic and social impact was pointed out [1]. Steel reinforcements in concrete structures are in passive conditions, that is protected by a thin oxide layer, promoted by the concrete alkalinity. In these conditions, carbon steel in concrete at pH 13 behaves as stainless steels in contact with fresh water, hence, like stainless steel, reinforcement can suffer corrosion attack when the passivity is destroyed. This can occur in two ways: either due to carbonation of the concrete, that is reaction of cement paste with carbon dioxide present in the atmosphere, which lowers the pH and causes general corrosion; or due to the presence of chlorides at the steel surface in concentrations higher than the critical one, which is in the range 0.4–1% by cement weight. Chlorides can be added erroneously to concrete in the mix water or in aggregates (nowadays this is strictly forbidden), or can penetrate by diffusion, for example in highway viaducts where de-icing salts are employed, or in marine structures. Local passivity destruction by chlorides causes pitting corrosion, with a mechanism similar to that observed for stainless steels [1]. The prevention of reinforcement corrosion is primarily achieved in the design phase by using high quality concrete and adequate cover. Additional prevention methods are adopted when severe environmental conditions occur on structures requiring a very long service life, as well as during repair and rehabilitation [1]. Among available methods, corrosion inhibitors can offer a simple and cost effective technique [2]. Inhibitors can be divided into two groups: mixedin inhibitors directly added to fresh concrete for new structures and migrating inhibitors, which can penetrate into the hardened concrete and are usually adopted in rehabilitation. Mixed-in inhibitors have been studied since the 185 © 2007, Institute of Materials, Minerals and Mining
186
Corrosion of reinforcement in concrete
1950s, and have been commercially available since the 1970s [1–5], while migrating inhibitors for concrete structures were proposed in the last 20 years, 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]; organic compounds based on mixtures of alkanolamines, amines or aminoacids [8– 12], proposed both as mixed-in and migrating inhibitors; and emulsions of an unsaturated fatty acid ester of an aliphatic carboxylic acid and a saturated fatty acid [13], proposed as a mixed-in inhibitor. Nitrite-based inhibitors are considered to be the most effective products available on the market for the protection and prevention of chloride-induced corrosion: they have been studied since the 1960s both in the laboratory and in field tests and several applications have confirmed their effectiveness. Nitrite acts as a passivator, due to its oxidising properties, and its inhibitive effectiveness is related to the nitrite/chloride molar ratio, which should be at least between 0.7 and 1 to prevent corrosion [2–6]. Concerns relate to their toxicity and solubility, and the possibility that they may cause an increase of corrosion rate if underdosed. For these reasons, in the last 20 years more interest has been given to new organic-based products. Organic corrosion inhibitors act by adsorption on the metal surface, forming a thin organic layer that may inhibit both anodic and cathodic processes; for this reason they are considered to be mixed inhibitors [10, 13–15]. Laboratory tests showed conflicting results about the efficiency of these products, both in solution and in concrete; frequently, the test conditions and minimum effective inhibitor concentration are not well defined. Moreover, because of their recent introduction and the very few field applications, there is not sufficient reliable data on their long term efficiency. What is more, the lack of a standard procedure to evaluate the effectiveness of these products makes it very hard to compare the results from the various experiments [12, 16]. Solution tests show a positive effect on corrosion initiation time and a reduction of corrosion rate only in the presence of a high concentration of corrosion inhibitor (about 100 g L–1) and with chloride contents of up to 1 mol L–1 [17, 18]. However, such a high inhibitor dosage may negatively influence the properties of the concrete (workability, setting time or compressive strength). Tests on samples of concrete containing chlorides in dosages higher than 1% with respect to the weight of cement, and on samples subjected to ponding cycles with a 3.5% NaCl solution, show that commercial organic corrosion inhibitors, when added with the recommended dosages, increase the initiation time [2, 11–13, 20], but the effect on critical chlorides threshold is not clear [2, 20, 21]; most of the works presented in the literature report that organic inhibitors do not significantly affect the corrosion rate of either mortar or concrete [2, 11, 19, 20].
© 2007, Institute of Materials, Minerals and Mining
Mixed-in inhibitors for concrete structures
187
This paper deals with the effectiveness of commercial mixed-in inhibitors on the corrosion of rebars embedded both in chloride-contaminated concrete and in carbonated concrete. Four different commercial inhibitors have been considered: three organic corrosion inhibitors, two being amine- and alkanolamine-based (A, B) and a fatty acid emulsion (C) and, for comparison, a commercial nitrite-based product (N) were added to the concrete mixture in the concentration suggested by the manufacturers. Control samples were also cast for comparison. The inhibitive effectiveness has been evaluated by means of two electrochemical parameters: the free corrosion potential and the 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]. During the initiation period, the corrosion rate of steel is negligible but meanwhile the characteristics of the concrete are changing, promoting the breakdown of steel passivity. Once corrosion has started (the propagation period), two main consequences occur: corrosion of the reinforcement and spalling of the 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 takes to reach the external rebars. The depth of the carbonated layer, x, increases with time following a parabolic law: x = k÷ t
(15.1)
where x, t and k are, respectively, layer thickness, time and the carbonation coefficient, which depends on concrete porosity (i.e. water/cement ratio) and on environmental conditions (i.e. relative humidity and temperature). Service life can be increased both by increasing the carbonation coefficient k or, once carbonation reaches the rebar level, by decreasing the corrosion rate. Corrosion Initiation
Propagation
Maximum penetration
Time
15.1 Concrete structure service life: Tuutti’s model [22].
© 2007, Institute of Materials, Minerals and Mining
188
Corrosion of reinforcement in concrete
15.2.2 Chlorides In the case of chloride-induced corrosion, the universally accepted approach considers only the corrosion initiation period, that can be calculated by applying Fick’s second law for non-stationary diffusion. If we suppose that the chloride concentration at the concrete surface (Cs) is constant with time, and that the ‘effective’ chloride diffusion coefficient (Dce) does not vary with time and space, i.e. concrete is homogeneous, Fick’s second law presents an analytical solution: Ê ˆ x C x = Cs – ( Cs – C0 ) Á 1 – erf 2 Dce t ˜¯ Ë
(15.2)
where C0 and Cx correspond to the total chloride content (by cement or concrete weight) at time t = 0 and at general time t, at depth x from the concrete surface; this equation allows calculation of chloride profiles with time and then estimation of the time that chlorides take to reach the critical concentration at the reinforcement level. To hinder chloride-induced corrosion, since it is not possible to slow down chloride diffusion, it is necessary to increase the critical chloride concentration. This may be done by adding substances directly into the fresh concrete that inhibit chloride-induced corrosion, thus increasing the service life.
15.3
Experimental methods
15.3.1 Samples and materials Three different series (Table 15.1) of concrete samples were cast in order to simulate both carbonation and chloride-induced corrosion. In particular, for chloride-induced corrosion, two possibilities have been considered: chlorides directly 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), 0.6 w/c ratio and 1770 kg m–3 of limestone aggregate of 12 mm maximum diameter. Chlorides were added to the mixing water in dosages of 0.8% and 1.2% by weight of cement. After two days in the mould, the concrete was
Table 15.1 Tested series and corrosion conditions Series
Conditions
1 2 3
Chlorides in the mix: 0/0.8/1.2% by weight of cement Chlorides from outside: ponding cycles Carbonation
© 2007, Institute of Materials, Minerals and Mining
Mixed-in inhibitors for concrete structures
189
cured until 28 days had elapsed in a wet environment (95% rh). Samples for carbonation were moved to the carbonation chamber (65% rh) after 3 days of exposure at 95% rh. Full penetration of carbonation was detected by means of phenolphthalein tests on 10 cm cubic samples. For each condition, 10 cm cubic samples were cast in order to check the compressive strength: the mean value of three samples ranged from 40 MPa to 43 MPa, both for concrete with and without chlorides; the addition of inhibitors did not adversely influence the compressive strength. Inhibitors were added following the manufacturers’ recommendations (Table 15.2). Two carbon steel rebars were placed in each specimen, 10 mm in diameter and 29 cm length. The ends of each rebar were coated with a heat shrinkable sleeve, so that only a length of 21 cm was exposed to the concrete. The net rebar surface area exposed to concrete was 66 cm2. The cover was 20 mm. A thin wire of mixed metal oxide (MMO) activated titanium, placed near each rebar, was used as a reference electrode [23, 24) and 3 AISI 304 stainless steel wires (2 mm in diameter) were embedded in each specimen as a counterelectrode for polarisation resistance measurements (Fig. 15.2 and 15.3). Ponding samples were equipped on the top with an appropriate container for NaCl solution. Table 15.2 Inhibitor descriptions and dosages Inhibitor
Kind of inhibitor
Dosage (kg m–3)
A B C N
Amines and alkanolamines (liquid) Amines and alkanolamines (liquid) Fatty acid emulsion (liquid) Nitrite based (solution 30%)
10 1.6 5 10
Carbon steel rebar Ti electrode
Stainless steel wire
20 10
50
20 50
50
50
50
200
15.2 Design of concrete specimen: frontal view (dimensions in mm).
© 2007, Institute of Materials, Minerals and Mining
190
Corrosion of reinforcement in concrete Heat shrinkable sleeve
Reference electrode
250
40
10
15.3 Design of concrete specimen: lateral view (dimensions in mm).
15.3.2 Exposures Chloride-containing and carbonated samples (series 1 and 3) were exposed out of doors in Milan. Ponding, i.e. accelerated chloride penetration (series 2), has been carried out using three-week cycles: one week of wetting the concrete 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 rebar were monitored. Corrosion potential was measured versus a saturated calomel electrode (SCE) put in contact with the concrete surface by means of a wet sponge. Rp was measured using the linear polarisation technique [25], by applying a potential scan rate of 10 mV per minute in the range ±10 mV with respect to the free corrosion potential. The mean corrosion rate (mm per year) was calculated by means of the Stern–Geary relationship: icorr = 1.17 ¥ C/Rp where Rp is the measured polarisation resistance (evaluated from the slope of the potential/current density curve) and the constant C is assumed to be equal to 26 mV [26].
15.3.4 Chloride concentration Concrete cores, taken from the samples, have been ground and dissolved in nitric acid. The chloride concentration was determined by potentiometric titration 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 potential corrosion (vs SCE) and polarisation resistance as a function of time. The © 2007, Institute of Materials, Minerals and Mining
Mixed-in inhibitors for concrete structures Summer
Potential [mV (SCE)]
100
Winter
191
Summer
0 –100 –200 –300 –400 1W0
1A0
1B0
1C0
400 Time (d)
600
1N0
–500 0
200
800
15.4 Free corrosion potential vs SCE: chloride-free specimens. Summer
Winter
Summer
Polarisation resistance (W m2)
1000
100
10 1W0 1
0
100
1A0 200
1B0 300 400 Time (d)
1C0 500
1N0 600
700
15.5 Polarisation resistance: chloride-free specimens.
keys show the kind of inhibitor, A, B, C and N (W indicates concrete samples without inhibitor), and chloride concentration (0, 0.8% and 1.2%). During the period of exposure to the atmosphere, in concrete samples without 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. Potential fluctuations were mainly due to seasonal variations, i.e. change in temperature and relative humidity: the highest values probably correspond to dry periods. Polarisation resistance showed small fluctuations during the whole test period for all chloride concentrations. The Rp values were calculated considering
© 2007, Institute of Materials, Minerals and Mining
192
Corrosion of reinforcement in concrete Summer
Winter
Summer
100
Potential [mV (SCE)]
1W08
1A08
1B08
1C08
1N08
0 –100 –200 –300 –400 –500
0
200
400 Time (d)
600
800
15.6 Free corrosion potential vs. SCE: 0.8% chloride by weight of cement. Summer
Winter
Summer
Polarisation resistance (W m2)
1000 1W08
1A08
1B08
1C08
1N08
100
10
1 0
100
200
300 400 Time (d)
500
600
700
15.7 Polarisation resistance: 0.8% chloride by weight of cement.
the whole rebar surface area exposed to concrete (66 cm2), although it is known that, in the presence of chlorides, localised corrosion occurs. It may be assumed that the corrosion rate is negligible if its average value is lower than a threshold value (1–2 mm year–1), according to [27]; that is, the polarisation resistance is higher than 10–20 W m2. The average values are different for the three chloride concentrations; in samples 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 lower polarisation resistance values than chloride-free concrete, even though the measured values are higher than 10 W m2; except for one specimen; it must
© 2007, Institute of Materials, Minerals and Mining
Mixed-in inhibitors for concrete structures Summer
Winter
193
Summer
100
Potential [mV (SCE)]
1W12
1A12
1B12
1C12
400 Time (d)
600
1N12
0 –100 –200 –300 –400 –500 0
200
800
15.8 Free corrosion potential vs. SCE: 1.2% chloride by weight of cement.
Summer
Winter
Summer
1000
Polarisation resistance (W m2)
1W12
1A12
1B12
1C12
1N12
100
10
1 0
100
200
300 400 Time (d)
500
600
700
15.9 Polarisation resistance: 1.2% chloride by weight of cement.
be emphasised that 0.8% chloride by weight of cement is in the range generally considered to be the critical chloride content, i.e. 0.4–1%; so corrosion attack may or may not initiate. Finally, samples containing 1.2% chloride show very low polarisation resistance values of 2–5 W m2, i.e. significant corrosion rates. Only in the presence of inhibitor A (amine- and alkanolamine-based) and inhibitor N (nitrite-based) do rebars display Rp values higher than 10 W m2 for the whole test period.
© 2007, Institute of Materials, Minerals and Mining
194
Corrosion of reinforcement in concrete
15.4.2 Ponding tests Figures 15.10–15.11 show the variation in time of free corrosion potential and 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). The polarisation resistance is about 100 W m2 to start with, then falls to a value close to 80 W m2. Only in the presence of inhibitor A does it reach values 100 2W
2A
2B
2C
2N
Potential [mV (SCE)]
0 –100 –200 –300 –400 –500 0
100
200
300 400 Time (d)
500
600
700
15.10 Free corrosion potential vs. SCE: specimens subjected to ponding cycles.
Polarisation resistance (W m2)
1000 2W
2A
2B
2C
2N
100
10
1 0
100
200
300 400 Time (d)
500
600
700
15.11 Polarisation resistance: specimens subjected to ponding cycles.
© 2007, Institute of Materials, Minerals and Mining
Mixed-in inhibitors for concrete structures
195
lower than 10 W m2, but, after 3 cycles, the polarisation resistance returns to high values. After 14 ponding cycles a core was taken for each sample and the chloride profile of the sample was determined. The chloride content at a depth of 1.5 cm was lower in concrete containing organic inhibitors, especially in the presence 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 polarisation resistance for samples subjected to accelerated carbonation (series 3). 100
Carbonation
Outside exposure
2W
Potential [mV (SCE)]
0
3A
–100
3B
–200
3C 3N
–300 –400 –500 –600 –700 0
200
400 Time (d)
600
800
15.12 Free corrosion potential vs. SCE: carbonation.
Polarisation resistance (W m2)
1000
Carbonation Outside exposure
2W 3A 3B
100
3C 3N
10
1 0
200
400 Time (d)
600
15.13 Polarisation resistance: carbonation.
© 2007, Institute of Materials, Minerals and Mining
800
196
Corrosion of reinforcement in concrete
Full carbonation was verified on concrete cubes with phenolphthalein tests in the first 200 days. Just after being exposed outside (Milan atmosphere), the free corrosion potential decreased and became stable at a constant value of between –400 mV and –500 mV (SCE). Fluctuations in free corrosion potential depend upon the weather conditions, i.e. relative humidity and temperature. Polarisation resistance shows similar behaviour, decreasing to below 10 W m2 once the samples were exposed outside.
15.5
Discussion
With respect to the service life of concrete structures, inhibitors can act in different ways: they may delay the initiation of corrosion, by increasing the critical chloride content or slowing down penetration by chlorides or carbonation (this is not an electrochemical effect), or, even once corrosion has started, they may reduce the corrosion rate. The following discussion will 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 presence of inhibitor A and inhibitor N, and is low in the other cases. This is probably due to both the positive effect of the inhibitors and because a chloride concentration of 0.8% by weight of cement is in the range of 0.4–1%, usually considered as the critical chloride content for corrosion initiation in alkaline concrete exposed to the atmosphere. In samples with 1.2% chloride by weight of cement, both potential and polarisation resistance measurements confirmed that corrosion occurs in samples with inhibitors B and C, while those with inhibitors A and N showed a low to negligible corrosion rate. It must be pointed out that the presence of chlorides causes a localised attack. So, it is necessary to verify the actual area affected by the corrosion attack; in fact, the lower the area, the higher the 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 concrete sample. In concrete samples subjected to accelerated chloride penetration, a chloride concentration of 0.5 to 1% by weight of cement was reached after 14 cycles at 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 the commercial inhibitors. From the results, it may be concluded that only in the presence 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 © 2007, Institute of Materials, Minerals and Mining
Mixed-in inhibitors for concrete structures
197
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 1.5 2.6 3.7 4.7
0.44 0.25 0.06 0.02 0.00
0.36 0.16 0.07 0.07 0.00
0.44 0.17 0.04 0.07 0.00
0.5 0.13 0.04 0.05 0.00
0.47 0.31 0.06 0.04 0.00
Table 15.4 Chloride diffusion coefficients evaluated by non-linear least squares regression of chloride profiles (Table 15.3) Inhibitor
Diffusion coefficient (108 cm2 s–1)
W A B C N
7.2 6.9 4.5 3.2 6.1
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 chloride contents of 0.8 and 1.2% by weight of cement, that display negligible and low 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 inhibitors effective 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 chloride diffusion [13, 20], although this is not an electrochemical effect. Diffusion coefficients were determined on the basis of a non-linear least squares regression analysis of the chloride profiles by means of equation (15.2). Diffusion coefficients are lower in samples with inhibitors than in those without (Table 15.4). The maximum reduction in the diffusion coefficient occurs in samples containing inhibitor C (fatty acid emulsion), by 50% with respect to the sample without inhibitor; similar results can be found in references [13, 20]. This effect is probably due to the formation of a hydrophobic layer within
© 2007, Institute of Materials, Minerals and Mining
198
Corrosion of reinforcement in concrete
the pores [2]. If this trend is confirmed by further measurements, under the same 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 of mean corrosion rate (mA m–2) for various chloride concentrations are compared with those without inhibitor. Corrosion rate has been calculated by polarisation resistance 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 mm year–1, corrosion is negligible [27]. In samples with chlorides of 0.8% by cement weight there are no significant differences in corrosion rate with or without commercial inhibitors, except with inhibitor C that presents corrosion rate 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 about 30% with respect to those in samples without inhibitor. The other commercial inhibitors do not show a significant reduction. A slight decrease in corrosion rate in concrete containing inhibitors can be related to an increase in the electrical resistivity of the concrete, probably due to a reduction in water content, as found for example in reference [21]. On the other hand, most of the published literature reports that organic inhibitors do not significantly affect the corrosion rate in mortar or in concrete [2, 11, 19, 20]; as previously mentioned, different behaviour was observed in solution [2, 11].
Corrosion rate (mA m–2)
10
1
1.2 0.8
0.1 W
A
B
C
N
0 Chlorides (%)
15.14 Corrosion rate in specimen containing inhibitors A, B, C and N and without inhibitor (W).
© 2007, Institute of Materials, Minerals and Mining
Mixed-in inhibitors for concrete structures
199
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, about 5% lower than the one determined in concrete without inhibitor. Carbonation coefficients were calculated by interpolating equation (15.1). Few data about the influence of organic inhibitors on carbonation penetration have been reported; amine-based inhibitors have not been found to affect carbonation penetration [29]. Corrosion rate In Fig. 15.15 rebar corrosion rates in carbonated concrete samples with inhibitors are compared with the average value obtained in a sample without inhibitor. A small inhibiting effect may be observed, and this has been evaluated in terms of inhibition efficiency in Fig. 15.16. The inhibition efficiency is defined as follows: Table 15.5 Carbonation coefficients calculated after three months of accelerated test Inhibitor
Carbonation coefficient (mm year–1/2)
W A B C N
38.0 36.2 38.2 41.4 39.2
8
Corrosion rate (mA m–2)
7 6 5 4 3 2 1 0 A
B
C
N
W
15.15 Mean corrosion rate in carbonated specimens containing inhibitors (A, B, C and N) and without inhibitors (W).
© 2007, Institute of Materials, Minerals and Mining
200
Corrosion of reinforcement in concrete 80 70
Efficiency (%)
60 50 40 30 20 10 0
A
B
C
N
15.16 Mean efficiency of inhibitors in carbonated specimens.
h=
* icorr – icorr * icorr
(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 suggested by producers, corresponding in this case (inhibitor N) to 1% by weight of cement, was not sufficient to reduce the corrosion rate significantly in carbonated concrete. This is in accordance with data in the literature that report a threshold value of 2–3% [30].
15.6
Conclusions
The commercial corrosion inhibitors studied in this work display a limited effectiveness both in chloride-contaminated and carbonated concrete. All of the commercial organic inhibitors seem to slow down chloride penetration. These results must be confirmed by further investigations. The critical chloride content at which carbon steel in concrete exposed to the atmosphere loses its passivity is between 0.4 and 1% by weight of cement. The commercial inhibitors increase the minimum threshold value of the critical concentration range to 0.8%, but not as far as 1.2%. Only in the presence of inhibitor A (amine- and alkanolamine-based) does the corrosion rate remain negligible even with 1.2% chlorides by weight of cement. In carbonated concrete no significant effect of the commercial inhibitors has been observed on carbonation penetration, i.e. corrosion initiation period. Only inhibitor B (amine- and alkanolamine-based) reduces the corrosion rate significantly (by 50%), increasing the propagation time.
© 2007, Institute of Materials, Minerals and Mining
Mixed-in inhibitors for concrete structures
201
The results obtained with a nitrite-based commercial inhibitor are in agreement with the literature, which states that they are effective against chloride attack if the molar ratio of NO 2– /Cl– exceeds 0.7 and against carbonation if the nitrite concentration exceeds 2–3%.
15.7
References
1. P. Pedeferri and L. Bertolini, La durabilità del calcestruzzo armato, McGraw-Hill Libri Italia, Milano, 2000. 2. B. Elsener, Corrosion inhibitors for steel in concrete – an EFC state of the art report, EFC, Number 35, 2001. 3. B. El-Jazairi and N. S. Berke, Eds., ‘The use of C.N. as a corrosion inhibiting admixture 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 to improve 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 the inhibition effect of sodium nitrite on reinforcement corrosion’, Cement, concrete aggregates, 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 Infrastructure Durability, Cambridge, 1997. 7. C. Andrade, C. Alonso, M. Acha and B. Malric, ‘Preliminary tests of Na2PO4F as a curative corrosion inhibitor for steel reinforcements in concrete’, Cement concrete res., 1992, 22, 869. 8. D. Bjegovic, L. Sipos and V. Uckrainczyk, ‘Diffusion of the MCI 2020 and 2000 corrosion inhibitors into concrete’, Int. Conference Corrosion and Corrosion Protection of 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 in concrete’, 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 inhibiting admixtures for reinforced concrete’, Concrete Int., 1992, 4, 45. 14. A. Phanasgaonkar, B. Cherry and M. Forsyth, ‘Corrosion inhibition properties of organic amines in simulated concrete environment: mechanism’, Int. Conference: Understanding Corrosion Mechanism in Concrete; a Key to Improve Infrastructure Durability, Cambridge, 1997. 15. A. Welle, J. D. Liao, M. Grunze, K. Kaiser, U. Maeder and N. Blank, ‘Interactions of 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 admixtures for structural concrete – assessment method and predictive modelling’, NACE, San Diego, California, Paper 655, 2000. © 2007, Institute of Materials, Minerals and Mining
202
Corrosion of reinforcement in concrete
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 pore solution’, Cement Concrete Res., 1999, 29, 1583. 18. V. Nobel-Pujol, T. Chaussadent and C. Fiaud, ‘Effects of organic and mineral inhibitors on the corrosion of reinforcements in hardened concrete’, 9th European Symposium on Corrosion Inhibitors, Ferrara, 2000, 313. 19. E. Pazini, S. Leao and C. Estefani, ‘Corrosion inhibitors. Behaviour of NaNO2 and Amine-based Products in the Prevention and Control of Corrosion in Reinforced Concrete’, NACE, Cancun, 1998. 20. M. Berra, F. Bolzoni, T. Pastore and P. Pedeferri, ‘Inibitori di Corrosione per Strutture in 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 steel corrosion’, 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 iridium dioxide electrodes’, J. Electroanal Chem., 1981, 126, 287. 24. K. Kinoshita and M. J. Madou, ‘Electrochemical measurements on Pt, Ir, and Ti oxides as pH probes’, J. Electrochem. Soc., 1984, 131, 1089. 25. M. Stern and A. L. Geary, ‘Electrochemical polarisation I: a theoretical analysis of the slope of polarisation curves’, J. Electrochem. Soc. 1957, 104, 56. 26. J. A. González, A. Molina, M. L. Escudero and C. Andrade, ‘Errors in the electrochemical evaluation of very small corrosion rates. I. Polarization resistance method 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 the effectiveness of corrosion inhibitors’ COST 521 Workshop, Belfast, 2000, 259. 29. B. Elsener, M. Büchler, F. Stalder and H. Böhni, ‘Migrating corrosion inhibitor blend 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 contaminated and chloride-free carbonated mortars’, ACI – Mat. J., 1995, 3–4, 130.
© 2007, Institute of Materials, Minerals and Mining
16 Effectiveness of mixed-in organic corrosion inhibitors on extending the service life of reinforced 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
16.1
Introduction
Corrosion inhibitors are chemical substances which, when added to the corrosive environment at a suitable concentration, prevent, reduce, or eventually stop corrosion occurring in various types of metals and alloys: they are usually used in aqueous solutions, but can also be used as volatile compounds to prevent the corrosion of objects exposed to the atmosphere [1]. Inhibitors are classified into anodic, cathodic, and mixed types, according to which reaction is more influenced. The use of corrosion inhibitors in concrete structures may concern both new and existing structures. In the first case, the inhibitors are admixed to the fresh concrete and are intended to delay the initiation of corrosion resulting from both carbonation and chloride ingress. In the second case, they may be applied on the surface of existing structures in which corrosion has already initiated and in this case they must penetrate in order to decrease the corrosion rate of the reinforcement. The inhibitors, or better admixtures containing the inhibitors to be used for reinforced concrete should not obviously adversely affect the prescribed concrete properties, e.g. mechanical properties and setting time. With regard to the use of corrosion inhibitors as a preventative, or supplementary measure for new reinforced concrete structures, one of the major problems is how to test their effectiveness and predict their influence in delaying the initiation of corrosion, and extending the service life. Tests in solution and in mortar specimens should not be considered suitable for the evaluation of the effectiveness of inhibitors since these environments are rather far from the real situation of the structures. Besides, experiments with concrete specimens must take into account the difficulty of accelerating the corrosion, necessary in order to obtain significant results in rather a short time.
203 © 2007, Institute of Materials, Minerals and Mining
204
16.2
Corrosion of reinforcement in concrete
Experimental methods
Two reinforced concrete slabs with dimensions of 1500 ¥ 750 ¥ 300 mm were 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, and the cement dosage was 350 kg m–3. One slab was prepared with an addition of a corrosion-inhibiting admixture containing amino alcohols and its corrosion behaviour was compared with that of the other slab, without inhibitor [2]. Six measuring probes were embedded in each slab to monitor the state of corrosion of the reinforcement. The probes consisted of three electrodes, made of short pieces of bar, each having an area of 20 cm2. Two electrodes were used for corrosion rate measurements (the auxiliary electrode being the rebar itself) and were positioned at the level of the upper rebar, closer to the surface onto which the salt solution was poured; the third electrode was positioned at the level of the lower rebar, further from the surface exposed to the aggressive solution, and was used to measure the macrocell emf in connection through a voltmeter with one of the upper electrodes. The position of one of the probes was inverted, in order to monitor the corrosion rate of a bar in the passive state. Moreover, the corrosion potential of the rebar in approximately the same positions as the internal probes was measured by means of a saturated calomel electrode (SCE) positioned over the upper surface of the slab. After four months of ageing, electrochemically forced chloride ingress was initiated, with a current density of 0.01 mA cm–2. Three months later the current was stopped, two cores were taken from each slab and the chloride profile was determined: the chloride concentration at the upper rebar level was approximately 0.06% w/w of concrete for both slabs. The test then continued with accelerated conditions achieved by cyclic ponding of the upper surface of the slabs with a saturated NaCl solution; ponding and normal exposure conditions were maintained for one and two weeks, respectively. The results of the measurements carried out over approximately three years 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 following conclusions can be drawn. ∑ in the presence of the corrosion-inhibiting admixture both the corrosion rate and the macrocell emfs were very low, thus showing an excellent passivation state even in the presence of growing amounts of chlorides; ∑ without inhibitor, noticeable corrosion attack started immediately after, or even during, the forced ingress of chlorides; ∑ the rather wide fluctuations of the corrosion rate values measured in the
© 2007, Institute of Materials, Minerals and Mining
Table 16.1 The emf and corrosion rate (CR) values for the slab without inhibitor; probe no. 1 is inverted
0 3 13 18 24 31 52 67 110 200 257 375 431 452 501 543 613 676 907 1076 1139 1487
Probe no.1
Probe no. 2
Probe no. 3
Probe no.4
Probe no. 5
emf (mV)
CR (mm y–1)
emf (mV)
CR (mm y–1)
emf (mV)
CR (mm y–1)
emf (mV)
CR (mm y–1)
emf (mV)
CR (mm y–1)
emf (mV)
–179 –183 –303 –245 –270 –254 –372 –294 –296 –165 –62 –406 –442 –387 –320 –360 –136 –132 –117 –104 –119 –39
2.7 2.7 2.6 2.3 2.3 2.4 2.6 3.1 2.9 2.7 2 2.1 2.9 3.9 2.9 4 2.8 1.6 4.7 3.7 2 1.4
141 163 262 257 259 255 362 281 170 191 390 406 403 354 340 330 130 122 161 176 198 88
4.1 4.9 9.7 6.6 3.3 5.3 9.3 8.1 13 16 14 12 21 19 19 20 8 5.1 43 12 9.1 7.3
142 164 301 337 327 314 391 246 133 148 379 267 322 307 290 266 150 138 138 156 181 58
2.7 2.6 6.1 3.4 3.9 6.1 21 9.9 9.3 7.6 9 6.6 11 7 8.9 11 5.4 3.9 37
259 277 378 228 232 201 349 281 205 135 382 329 342 214 220 247 110 120 78 62 103 26
5.4 5.4 9.6 6.1 5.6 6.3 10 11 13 12 12 10 17 17 16 17 5.6 4.4 34 50 6.3 4.9
120 174 253 153 148 130 225 112 55 154 401 405 442 289 156 190 78 100 100 44 95 36
6.3 6.3 11 7.3 6.4 6.6 12 10 9.7 7.9 9 11 12 11 7.9 8.3 3.1 2.9 26 37 3.6 3.3
213 229 429 251 264 263 431 155 272 208 573 470 448 299 266 287
6.9 6.9
Probe no. 6
92 115 126 68
CR (mm y–1) 8.1 7.9 9.1 8.3 9.6 24 13 17 16 16 13 24 23 20 19 6.9 3.6 54 57 8.1 8.3
Effectiveness of mixed-in organic corrosion inhibitors
Time (days)
205
© 2007, Institute of Materials, Minerals and Mining
206
Time days
0 7 22 28 77 84 98 140 196 217 266 308 378 441 672 841 904 1252
Probe no.1 emf (mV) 21 12 –6 –14 –8 10 –2 7.1 33 30 42 38 –16 74 29 10
CR (mm y–1)
Probe no. 2
Probe no. 3
Probe no.4
Probe no. 5
Probe no. 6
emf (mV)
CR (mm y–1)
emf (mV)
CR (mm y–1)
emf (mV)
CR (mm y–1)
emf (mV)
CR (mm y–1)
emf (mV)
CR (mm y–1)
0.6 0.6 0.7 0.6 0.6 0.6 0.6 0.6 0.6 1.1 0.9 0.8 0.7 0.1 0.7 0.3 0.4 0.4
27 18 –2 –4 –5 –6 –7 –11 –4 0 40 37 19 16 –20 –43 –23 –18
0.6 0.5 0.5 0.5 0.5 0.6 0.5 0.5 0.5 0.8 0.8 0.7 0.5 0.3 0.6 0.4 0.4 0.4
15 8 0 1 3 3 2 3 –2 –30 –33 –37 –1
0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.5 0.4 0.7 0.6 0.6 0.4 0.3 0.4 0.2 0.2 0.3
5 3 –1 –1 –2 0 –1 –7 6 –10 –27 –25 –2 –3 –5 0 –1 3
0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.6 0.8 0.4 0.3 0.2 0.3 0.1 0.2 0.2
–20 –12 –1 –5 –8 –5 –7 6 –6 –8 –39 –32 –3 –4 0 3 8 6
0.4 0.4 0.4 0.4 0.3 0.4 0.4 0.4 0.3 0.5 0.4 0.4 0.3 0.2 0.3 0.1 0.2 0.2
0.7 0.7 0.7 0.6 0.6 0.6 0.6 0.6 0.5 0.9 0.5 0.7 0.5 0.3 0.7 0.4 0.6 0.5
© 2007, Institute of Materials, Minerals and Mining
11 13 15 12 13 15 14 –5 –3 –1.8 –1 –1 –1 –1 3 4 5 2
–5 0 –3 0
Corrosion of reinforcement in concrete
Table 16.2 The emf and corrosion rate (CR) values for the slab with inhibitor; probe no. 1 is inverted
Effectiveness of mixed-in organic corrosion inhibitors
207
0
Corrosion potential (mV)
–50 –100 –150
Cl = 0.06%
–200
Cl = 0.14%
Cl = 0.14%
–250 –300 –350 –400 –450 –500 0
200
400
600 Days (a)
800
1000
1200
0 Cl = 0.13%
Corrosion potential (mV)
–50 –100 –150 Cl = 0.08%
–200 –250
Cl = 0.10%
–300 –350 –400 –450
Cl = 0.06%
–500 0
100
200
300
400
500 Days (b)
600
700
800
900
1000
16.1 Reinforcement corrosion potential measured over the upper concrete surface vs. SCE (a) without and (b) with inhibitor. The six curves refer to the rebar close to the positions of the six probes embedded in the slabs. The chloride content refers to the weight of concrete.
slab without inhibitor, especially in comparison with the values determined for the probe in the passive state, are certainly due to variations in the environmental conditions (temperature and humidity) and also to the fact that, for a certain time, the ponding cycles were interrupted (at 600–700 days, and since the day 1100; between 700 and 1100 days regular ponding with saline solution was done). In any case, the corrosion rates for the five probes situated at the level of the upper rebar are much higher than those of the control probe, thus indicating that the upper rebar is in corrosive conditions;
© 2007, Institute of Materials, Minerals and Mining
208
Corrosion of reinforcement in concrete
Corrosion (mm yr–1)
15
10
5
0 0
200
400
600
800 Days (a)
1000
1200
1400
1600
Corrosion (mm yr–1)
15
10
5
0 0
200
400
600
800
1000
1200
1400
Days (b)
16.2 Internal probe corrosion rate values measured on slabs (a) without and (b) with inhibitor; the thick line refers to the inverted probe.
∑ similarly, the fluctuations of the macrocell emfs for all six probes (the control probe gives inverted values of course), depend on the same variables and 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 the control probe (inverted, so that the upper electrode, close to the upper rebar, is corroding) are very high in comparison with those measured for the 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 inhibiting mixture may delay the onset of corrosion of the reinforcement in the presence of chlorides as aggressive agents. However, a rough preliminary calculation,
© 2007, Institute of Materials, Minerals and Mining
Effectiveness of mixed-in organic corrosion inhibitors
209
600
emf (mV)
400
200
0 0
200
400
600
800
1000
1200
1400
1600
1200
1400
–200
–400 Days (a) 600
emf (mV)
400
200
0 0
200
400
600
800
1000
–200
–400 Days (b)
16.3 Internal probe macrocell emf measured on slabs (a) without and (b) with inhibitor; the thick line refers to the inverted probe.
based on the use of Fick’s second law of diffusion for a semi-infinite slab, may be done as follows: È ˘ c( t , x ) = c 0 Í1 – erf x ˙ 2 Dt ˚ Î
(16.1)
where c(t, x) is the chloride concentration at depth x and time t, co is the chloride surface concentration, erf is the error function and D is the diffusion coefficient. Assuming that the time to initiate corrosion for structures of similar concrete mixes without inhibitor is 22 years for the slag cement concretes (a hypothesis suggested by Autostrade, Italy [3]) and that the diffusion coefficient, calculated for the slab without the inhibitor, is the same for the other slab, it may be
© 2007, Institute of Materials, Minerals and Mining
210
Corrosion of reinforcement in concrete Table 16.3 Calculation of the corrosion initiation time (c is the chloride concentration at the rebar level)
Co (weight %) C (weight %) D (cm2 s–1) Corrosion initiation time
Without inhibitor time 0
With inhibitor 30 months
0.3 0.06 3.8 ¥ 10–9 22 years
0.3 0.13 3.8 ¥ 10–9 >62 years
concluded (Table 16.3) that the presence of the tested inhibitors delays the time to corrosion initiation by a factor of at least three. This calculation is based on the use of the chloride content values determined in time in cores taken from the slabs; it must also be taken into account that the penetration of the chlorides in the slab containing the corrosion inhibiting admixture appears to be much lower than in the control slab; this means that the admixture used has a positive synergic action in facilitating the passivation of the steel in the presence of chlorides while at the same time slowing their ingress into the concrete.
16.4
References
1. B. Elsener, M. Büchler and H. Böhni, Corrosion of Reinforcement in Concrete, European Federation 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. On Corrosion 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.
© 2007, Institute of Materials, Minerals and Mining
17 Migrating 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 Politecnico di Milano, Italy
17.1
Introduction
Reinforced concrete buildings have a limited operating life because of corrosion of their steel reinforcement. The phenomenon involves not only bridges or highway and marine infrastructures, still the most seriously affected, but also public and private structures, churches, stadiums, monuments and others. Corrosion begins after concrete loses its protective properties, with respect to the steel, by means of two phenomena: ingress of chlorides (usually from deicing salt or a marine environment) and reduction of the pH of pore solution promoted by the reaction with carbon dioxide from the atmosphere (carbonation process). The prevention of reinforcement corrosion is primarily achieved at the design stage by using high quality concrete and adequate cover. Additional prevention methods are adopted when severe environmental conditions occur or on structures requiring a very long service life, as well as in rehabilitation. Among the available methods, corrosion inhibitors are attractive because of their low cost and easy handling, compared with other preventative methods. Inhibitors can be divided into two groups (1, 2): admixed inhibitors, directly added to fresh concrete for new structures, and migrating inhibitors, which can penetrate into hardened concrete and are usually adopted in rehabilitation. While the first admixed corrosion inhibitors, inorganic nitrites, have been commercially available since the 1970s [1–5], migrating corrosion inhibitors for concrete structures have been studied in the last 20 years. The proposed inhibitors include inorganic compounds (sodium monofluorophosphate) [6– 8], organic mixtures (primarily amine-based compounds) [9–20], and organicinorganic mixtures. The effectiveness of migrating corrosion inhibitors is related to their ability to reach the rebars [2, 12]. Inhibitor migration in concrete occurs by different mechanisms: 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 211 © 2007, Institute of Materials, Minerals and Mining
212
Corrosion of reinforcement in concrete
high percentage of volatile compounds can increase the migration rate in the concrete cover, but may also cause retro-migration towards the atmosphere. Low concrete cover can favour the penetration of inhibitors, but also of aggressive agents (carbon dioxide and/or chlorides). The real effectiveness of corrosion inhibitors in different situations, i.e. of chloride content, exposure, etc., is not well defined. The published results on migrating corrosion inhibitors are controversial, with respect to both the inhibitor penetration and to the effectiveness on the corrosion rate. Summarising very briefly, there are doubts on migration ability [10, 11, 19], or the inhibitors may penetrate only in conditions (low concrete cover, less compact concrete) that also favour penetration of aggressive agents [6, 7]; probably not surprisingly, only tests carried out by the manufacturers show the good penetration of migrating corrosion inhibitors [14–16]. Some tests carried out by the inhibitor producers show a reduction of corrosion rate using migrating corrosion inhibitors in chloride-contaminated concrete [13, 14], while other conflicting test results show their ineffectiveness [2, 12, 17–20]. Besides, test conditions and minimum effective inhibitor dosage are not well defined in the literature. There are concerns about the effectiveness of these products, since their homologous counterparts for fresh concrete (amine and alkanolamine based admixed inhibitors) show questionable inhibiting effect [21–23]. The aim of the present work was to verify the performance of two commercial organic migrating corrosion inhibitors, amine-and alkanolamine-based, named FM and DM; their ability to slow or to stop corrosion attack and to prevent corrosion initiation have also been studied.
17.2
Experimental methods
17.2.1 Materials and specimens The concrete mixture was prepared with cement CEM II A/L 42,5R and limestone aggregate of 12 mm maximum diameter (Table 17.1). The concrete was cured for 28 days at 20 ∞C and 95% relative humidity (rh), except for the specimens for carbonation, which were cured for only 3 days at 95% relative humidity and then placed in a carbonation chamber (65% rh). Table 17.1 Concrete mix design
Cement CEM II A/L 42.5R Distilled water Water/cement (w/c) ratio Aggregates Plasticiser – 0.6% by cement weight Compressive strength
© 2007, Institute of Materials, Minerals and Mining
(kg m–3) (L m–3 ) (kg m–3) (kg m–3) (MPa)
Series 1, 3, 4
Series 2
367 220 0.6 1770 2.2 43
338 220 0.65 1790 – 30
Migrating inhibitors on corrosion in reinforced concrete
213
Two carbon steel rebars (Feb44k), 10 mm diameter, were placed in each concrete specimen (Fig. 17.1). Net rebar surface area exposed to concrete was 66 cm2. Activated titanium reference electrodes [24, 25] were placed near to each rebar, and three stainless-steel wires (2 mm in diameter) were placed in the specimens for corrosion rate measurements (linear polarisation resistance method, LPR). The tested migrating inhibitors were amine- and alkanolamine-based. The dosages were those recommended by the manufacturers, i.e. 400 g m–2 of concrete surface for inhibitor FM, and 250 g m–2 for inhibitor DM, increased 30% approximately for the expected losses. The inhibitors were applied twice, with 4 months between the two applications. Four series of concrete specimens (200 ¥ 200 ¥ 50 mm) were cast. Series 1, with chloride in the mix (0.8 and 1.2% by cement weight), on which migrating 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 migrating inhibitors 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. 2 months and 6 months after concrete carbonation); and finally, series 4, concrete subjected to chloride ponding cycles on which migrating inhibitors were applied after 7 months and after 1 year, before corrosion initiation. For comparison, for series 1, one specimen was cast for each chloride concentration and corrosion inhibitor; instead, for series 2, 3 and 4, two specimens were prepared for each corrosion inhibitor. Specimens with mixed-in chloride were exposed out of doors in Milan. Specimens subjected to chloride penetration were subjected to alternating ponding cycles of 1 week wetting with 3.5% NaCl solution, followed by 2 weeks of drying in the laboratory. Specimens for carbonation, after curing, Carbon steel rebar
Stainless steel wire
Ti electrode
20 10 20 50
50
50
50
200
17.1 Reinforced concrete specimen (dimensions in mm).
© 2007, Institute of Materials, Minerals and Mining
50
214
Corrosion of reinforcement in concrete
were exposed in a chamber at 65% rh and 20 ∞C, in which for one hour each day 95% CO2 was introduced. After carbonation, specimens were wetted and exposed out of doors in Milan.
17.2.2 Corrosion monitoring Free corrosion potential was measured with respect to a SCE reference electrode placed in contact with the concrete surface by means of a wet sponge. Corrosion rate was evaluated through linear polarisation resistance measurements [1, 26, 27]. Stainless-steel wires were used as counter electrodes, and the activated titanium electrode, placed near to the carbon steel bar, was used as the reference electrode (Fig. 17.1). Measurements were carried out by a potentiodynamic technique, scanning the potential from –10 mV to +10 mV with respect to the free corrosion potential (Ecorr) with a scan rate of 10 mV min–1. Polarisation resistance (Rp, in W m2) was evaluated from the slope of the linear zone of the potential/current density curve [27]. The corrosion rate (icorr) was calculated by means of the Stern–Geary formula [26]:
icorr = C Rp
(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 electrode placed on the concrete surface [28]. Carbonation was also confirmed by a phenolphthalein test on cubic control specimens placed in the same carbonation chamber. For chloride analysis, cores were extracted from concrete specimens; slices were cut from the cores and, after milling, the concrete powder was dissolved with nitric acid. The chloride content of the solutions was evaluated by potentiometric titration with AgNO3 (0.01N).
17.3
Results and discussion
17.3.1 Mixed-in chlorides In concrete specimens exposed to the atmosphere (series 1), the corrosion behaviour depends on the chloride concentration. In specimens with chlorides 0.8% by cement weight, the free corrosion potential of the rebar was higher than –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 than 10 W m2 (Fig. 17.3a), i.e. the corrosion rate was higher than 2 mA m–2, so was
© 2007, Institute of Materials, Minerals and Mining
Migrating inhibitors on corrosion in reinforced concrete Summer
Winter
215
Summer
100
Potential [mV (SCE)]
0 –100 –200 –300 –400 –500 0
100
200
Summer
300 400 Time (d) (a) Winter
500
600
Summer
100 Inhibitor FM Inhibitor DM
Potential [mV (SCE)]
0 –100 –200 –300 –400 –500 0
100
200
300 400 Time (d) (b)
500
600
17.2 Free corrosion potential in series 1 specimens with chloride contents of (a) 0.8% and (b) 1.2% by cement weight. Vertical dotted lines indicate the migrating inhibitor applications (only for 1.2% chlorides specimens).
not negligible [29]. It must be pointed out that a chloride content of 0.8% by cement weight is within the critical chloride concentration range, which is between 0.4–1%, so corrosion may or may not be initiated [1]. Migrating inhibitors were not applied on these specimens. In specimens with a chloride content 1.2% by cement weight, a corrosion rate higher than 2 mm y–1 was measured immediately after atmospheric exposure (see Rp values lower than 10 W m2 in Fig. 17.3b). The potential values were very scattered: the value was lower in 1.2% Cl– specimens in the first 300 day period (Fig. 17.2b). Migrating corrosion inhibitors were applied to 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
© 2007, Institute of Materials, Minerals and Mining
216
Corrosion of reinforcement in concrete Summer
Winter
Summer
Polarisation resistance (W m2)
1000
100
10
1 0
100
200
Summer
300 Time (d) (a)
400
Winter
500
600
Summer
Polarisation resistance (W m2)
1000 Inhibitor FM Inhibitor DM 100
10
1 0
100
200
300 Time (d) (b)
400
500
600
17.3 Polarisation resistance (Rp) in series 1 specimens with chloride contents of (a) 0.8% and (b) 1.2% by cement weight. Vertical dotted lines indicate the migrating inhibitor applications (only for 1.2% chlorides specimens).
by the presence of the inhibitors. The mean values of corrosion rate before any treatment and after the first and second applications were not very different (Fig. 17.4); the standard deviation decreased after the first application and increased after the second one. Under these conditions migrating corrosion inhibitors are not effective in reducing the corrosion rate. Measurements of the penetration of inhibitors into the concrete specimens are not available. As already mentioned, inhibitor penetration depends strongly on 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,
© 2007, Institute of Materials, Minerals and Mining
Migrating inhibitors on corrosion in reinforced concrete
217
Corrosion rate (mA m–2)
20 Mean value Standard deviation 15
10
5
0
Before application
After first application (a)
After second application
Corrosion rate (mA m–2)
20 Mean value Standard deviation 15
10
5
0
Before application
After first application (b)
After second application
17.4 Corrosion rate in specimens with a chloride content of 1.2% by cement weight (series 1) with (a) inhibitor FM and (b) inhibitor DM. Mean, maximum and minimum values are shown, with standard deviation.
been carried out with inhibitor FM applied to concrete specimens cast with the same mixture proportions [30]. It was observed that the most effective penetration was due to capillary sorption (oven-dried samples). Nevertheless, in all experiments, the penetration of inhibitor was less than 20 mm. On the other hand, as previously reported, even if such inhibitors are able to reach the rebars there is concern about the real effectiveness of amine and alkanolamine 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) subjected to chloride penetration cycles (1 ponding cycle consists of 1 week wetting with 3.5% NaCl solution followed by 2 weeks drying in the laboratory) is
© 2007, Institute of Materials, Minerals and Mining
218
Corrosion of reinforcement in concrete
shown in Fig. 17.5. After four ponding cycles, i.e. ~80 days, the free corrosion potential and polarisation resistance decreased and the corrosion rate was higher than 5 mA m–2 for all specimens (Fig. 17.6). The application of migrating inhibitors was carried out after 3 months and 7 months. Again, the results were not good; the corrosion rate increased after each corrosion inhibitor application (Fig. 17.6) and the standard deviation of the corrosion rate increased greatly. It must be remembered that the ponding cycles were continued 3 weeks after the migrating inhibitor application and that the chloride concentration at the level of the rebars reached high values, approximately 2% by cement weight after 10 ponding cycles (250 days). In the case of ponding, it can be confirmed that migrating corrosion inhibitors are not effective. The results are in accordance with the experiments described in 0 Inhibitor FM Inhibitor DM
Potential [mV (SCE)]
–100 –200 –300 –400 –500 –600 0
100
200 Time (d) (a)
300
400
Polarisation resistance (W m2)
1000 Inhibitor FM Inhibitor DM 100
10
1
0.1 0
100
200 Time (d) (b)
300
400
17.5 Measurements of (a) free corrosion potential and (b) polarisation resistance in series 2 specimens subjected to ponding cycles (w/c = 0.65). Vertical dotted lines indicate the migrating inhibitor applications, after corrosion initiation.
© 2007, Institute of Materials, Minerals and Mining
Migrating inhibitors on corrosion in reinforced concrete
219
80
Corrosion rate (mA m2)
70
Mean value Standard deviation
60 50 40 30 20 10 0 Passive rebars
Before any application
After first application
After second application
After first application
After second application
(a) 80
Corrosion rate (mA m2)
70
Mean value Standard deviation
60 50 40 30 20 10 0 Passive rebars
Before any application (b)
17.6 Corrosion rate (mean, maximum and minimum values with standard deviation are shown) for (a) inhibitor FM and (b) inhibitor DM in concrete subjected to chloride ponding (series 2, w/c = 0.65).
[20]: no reduction of corrosion rate after corrosion initiation was noticed with amine- and alkanolamine-based migrating inhibitors. According to the literature [11] and to the results obtained in the parallel experiments described previously [30], low inhibitor penetration can be expected under the present experimental conditions.
17.3.3 Carbonated concrete After six months of accelerated carbonation, all specimens of series 3 were exposed to the atmosphere. The steel rebar exhibited a low corrosion potential
© 2007, Institute of Materials, Minerals and Mining
220
Corrosion of reinforcement in concrete
of about –500 mV (SCE) and a low polarisation resistance value, i.e. a corrosion rate higher than 5 mA m–2 (Fig. 17.7 and 17.8). Migrating corrosion inhibitors were applied after 8 months and 1 year (i.e. 2 months and 6 months after concrete carbonation). The mean corrosion rate was reduced (see Fig. 17.8 and 17.9); the mean inhibition efficiency approached 50% for inhibitor DM after the second application. Nevertheless, scattering of the results is evident, especially in the presence of inhibitor FM. Moreover, the residual corrosion rate is not negligible; after the 2nd application (Fig. 17.7) Carbonation
External exposure
100 Inhibitor FM Inhibitor DM
Potential [mV (SCE)]
0 –100 –200 –300 –400 –500 –600 –700 0
200
Carbonation
400 Time (d) (a)
600
800
External exposure
Polarisation resistance (W m2)
1000 Inhibitor FM Inhibitor DM 100
10
1 0
200
400 Time (d) (b)
600
800
17.7 Measurements of (a) free corrosion potential and (b) polarisation resistance in series 3 specimens subjected to carbonation (first period) and then exposed to the atmosphere. Vertical dotted lines indicate the migrating inhibitor applications.
© 2007, Institute of Materials, Minerals and Mining
Migrating inhibitors on corrosion in reinforced concrete
221
Corrosion rate (mA m2)
20 Mean value Standard deviation
10
0
Alkaline concrete
Carbonated After first After second concrete application application (a)
Corrosion rate (mA m2)
20 Mean value Standard deviation
10
0 Alkaline concrete
Carbonated concrete
After first application
After second application
(b)
17.8 Corrosion rate (mean, maximum and minimum values with standard deviation are shown) for (a) inhibitor FM and (b) inhibitor DM in carbonated concrete (series 3) after migrating inhibitor application.
a corrosion rate higher than the threshold value, 1–2 mA m–2 [29], was measured. These results are in agreement with the few available literature data [20]; according to this work, no effect was noticed on the corrosion rate after exposure for one year. In the case of carbonation, according to the literature, inhibitor penetration is effective only after several weeks in dried specimens [11]. It is probable that under the present experimental conditions, inhibitors did not reach the rebar level and the effect on corrosion rate may be due to some change in concrete resistivity.
© 2007, Institute of Materials, Minerals and Mining
222
Corrosion of reinforcement in concrete 100
Mean efficiency (%)
Inhibitor FM Inhibitor DM
50
0 After first application
After second application
17.9 Carbonated concrete (series 3): mean inhibition efficiency.
17.3.4 Chloride penetration – application of inhibitor before corrosion initiation Results obtained on applying migrating inhibitors before corrosion initiation (series 4) are reported in Fig. 17.10. Twenty-seven ponding cycles were carried out. For purposes of comparison, measurements carried out on reference specimens 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 and 21st 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 than 10 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 ponding cycles, that is 650 days, was between 0.7–1.4% by cement weight. The application of these migrating corrosion inhibitors seems to delay corrosion initiation; nevertheless, it is worth noticing that the corrosion rate is still negligible on rebars with low potential values. For these reasons further investigations are needed.
17.4
Conclusions
Commercial amine- and alkanolamine-based migrating inhibitors, applied according to the manufacturer’s instructions, show no appreciable reduction of corrosion rate in chloride-contaminated concrete, either on adding chlorides to the mixture or in conditions of chloride penetration from the outside. In the case of carbonation, migrating inhibitors are able to reduce the mean value of the corrosion rate, nevertheless the scatter of results is high and the residual corrosion rate is not negligible.
© 2007, Institute of Materials, Minerals and Mining
Migrating inhibitors on corrosion in reinforced concrete
223
Potential [mV] (SCE)
0
–100
–200
–300
No inhibitor Inhibitor FM Inhibitor DM
–400 0
100
200
300 400 Time (d) (a)
500
600
700
300 400 Time (d) (b)
500
600
700
Polarisation resistance (W m2)
1000
100
10 No inhibitor Inhibitor FM Inhibitor DM 1 0
100
200
17.10 Measurements of (a) free corrosion potential and (b) polarisation resistance in series 4: specimens subjected to chloride ponding. Vertical dotted lines indicate the migrating inhibitor applications.
Concerning the delay in corrosion initiation in specimens subjected to chloride penetration after migrating inhibitor treatment, the available results are inconclusive and further investigations are needed.
17.5
References
1. P. Pedeferri and L. Bertolini, La durabilità del calcestruzzo armato, McGraw-Hill Italia, Milan 2000. 2. B. Elsener, Corrosion inhibitors for steel in concrete – an EFC state of the art report, EFC, Number 35, 2001. 3. C. Andrade, C. Alonso and J. A. Gonzalez, ‘Some laboratory experiments on the
© 2007, Institute of Materials, Minerals and Mining
224
4. 5.
6.
7. 8.
9.
10. 11. 12.
13. 14.
15. 16. 17. 18. 19. 20.
21. 22.
23.
Corrosion of reinforcement in concrete inhibition effect of sodium nitrite on reinforcement corrosion’, Cement, concrete aggregates, 1986, 8(2), 110. N. S. Berke and T. G. Weil, ‘World-wide review of corrosion inhibitors in concrete’, Advances in Concrete Technology, Athens, CANMET, 1992, 899. B. El-Jazairi, N. Berke and W. R. Grace, ‘The use of C.N. as a corrosion inhibiting admixture to steel reinforcement in concrete’, in Corrosion of reinforcement in concrete, Elsevier Applied Science, London, 1990, 571. C. Alonso, C. Andrade, C. Argiz and B. Malric, ‘Preliminary tests of Na2PO4F as a curative corrosion inhibitor for steel reinforcements in concrete,’ Cement Concrete Res, 1992, 22, 869. C. Andrade, C. Alonso, M. Acha and B. Malric, ‘Na2PO4F as inhibitor of corroding reinforcement in carbonated concrete’, Cement Concrete Res., 1996, 26, 405. A. Raharinaivo and B. Malric, ‘Performance of MFP for inhibiting corrosion of steel in reinforced concrete structures’, International Conference on Corrosion and Rehabilitation of Reinforced Concrete Structures, Orlando, 1998. V. Nobel-Pujol, T. Chaussadent and C. Fiaud, ‘Effects of organic and mineral inhibitors on the corrosion of reinforcements in hardened concrete’, 9∞ SEIC (European Symposium on Corrosion Inhibitors), Ferrara, 2000, 313. J. Tritthart, ‘Transport of corrosion inhibitors in cement paste and concrete’, COST 521 Workshop, Belfast, 2000, 203. J. Tritthart, ‘Transport of the corrosion inhibitor SIKA Ferrogard 903 in cement paste and concrete’, COST 521 Workshop, Tampere, 2001, 191. C. L. Page, ‘Aspects of the performance of corrosion inhibitors applied to reinforced concrete’, 9∞ SEIC (European Symposium on Corrosion Inhibitors), Ferrara, 2000, 261. 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. D. Bjegovic, L. Sipos, V. Uckrainczyk and B. Micksic, ‘Diffusion of the MCI 2020 and 2000 corrosion inhibitors into concrete’, Int. Conf. on ‘Corrosion and corrosion protection of steel in concrete’, Sheffield, 1994, 865. D. Bjegovic and B. Miksic, ‘Migrating corrosion inhibitor protection of concrete’, Mater. Perf., 11, 1999, 52. D. Bjegovic, ‘Accelerating testing of migrating corrosion inhibitors effectiveness’, COST 521 Workshop, Belfast, 2000, 235. B. Elsener, M. Büchler and H. Böhni, ‘Organic corrosion inhibitors for steel in concrete’, Eurocorr ’ 99, Aachen, 1999. B. Elsener, M. Büchler and H. Böhni, ‘Corrosion inhibitors for steel in concrete’, Eurocorr ’97, Trondheim 1997, 469. B. Elsener, ‘A review of the performance of corrosion inhibitors for steel in concrete’, COST 521 Workshop, Belfast, 2000, 271. B. Elsener, M. Büchler, F. Stalder and H. Böhni, ‘Migrating corrosion inhibitor blend for reinforced concrete : Part 1 – prevention of corrosion’, Corrosion, 1999, 55(12), 1155. C. Alonso, C. Andrade, J. Fullea and J. Sanchez, ‘Accelerating test to ascertain the effectiveness of corrosion inhibitors’, COST 521 Workshop, Belfast, 2000, 259. E. Pazini, S. Leao and G. Estefani, ‘Corrosion inhibitors. Behaviour of NaNO2 and amine-based products in the prevention and control of corrosion in reinforced concrete’, NACE Latin-American Congress, Cancun, 1998. M. Berra, F. Bolzoni, T. Pastore and P. Pedeferri, ‘Inibitori di corrosione per strutture
© 2007, Institute of Materials, Minerals and Mining
Migrating inhibitors on corrosion in reinforced concrete
24. 25. 26. 27.
28. 29. 30.
31.
225
in c.a.’, Giornate Nazionali sulla corrosione e protezione, 4∞ edizione, AIM, Genova, 1999, 293. S. Ardizzone, A. Carugati and S. Trasatti, ‘Properties of thermally prepared iridium dioxide electrodes’, J. Electroanal. Chem., 1981, 126, 287. K. Kinoshita and M. J. Madou, ‘Electrochemical measurements on Pt, Ir, and Ti oxides as pH probes’, J. Electrochem. Soc., 1984, 131(5), 1089. M. Stern and A. L. Geary, ‘Electrochemical polarisation I: a theoretical analysis of the slope of polarisation curves’, J. Electrochem. Soc., 1957, 104, 56. J. A. González, A. Molina, M. L. Escudero and C. Andrade, ‘Errors in the electrochemical evaluation of very small corrosion rates. I. Polarization resistance method applied to corrosion of steel in concrete’, Corrosion Sci., 1985, 25, 917. L. Bertolini, F. Bolzoni, P. Pedeferri and T. Pastore, ‘Cathodic protection of reinforcement in carbonated concrete’, Corrosion 98, NACE, Paper 639. C. Andrade, ‘Determination of chloride threshold in concrete’, COST 521 Workshop, Luxembourg, 2002, 108. F. Malservigi, ‘Inibitori per la prevenzione della corrosione delle armature e per il ripristino delle strutture in c.a.’, Tesi di Laurea (Degree Thesis), Politecnico di Milano, A.A. 2000-01. F. Bolzoni, G. Fumagalli, L. Lazzari, M. Ormellese and P. Pedeferri, ‘Mixed-in inhibitors for concrete structures’, Eurocorr 2001, Riva del Garda, 2001, 10 (see Chapter 15).
© 2007, Institute of Materials, Minerals and Mining
18 Effectiveness 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 and Concrete (TFB), Switzerland
18.1
Introduction
The number of reinforced concrete structures showing signs of deterioration or damage due to corrosion of rebars has increased dramatically over the last 20 years. There is, therefore, an obvious and urgent need by the owners of reinforced concrete structures for simple, quick, durable and cost-efficient repair techniques. The application of corrosion inhibitors might be such a rehabilitation method since the chloride-contaminated or carbonated concrete does not have to be removed. Thus, this repair method seems to be very promising. Although inhibitors are used in practice [1–2] and some field trials are underway [3] there is still a lack of results from well monitored long-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 the effectiveness of two inhibitors, sodium monofluorophosphate (MFP) and FerroGard-903 (FG), in the case of chloride-induced rebar corrosion and to verify the results of laboratory experiments [4–5] as well as to evaluate an appropriate monitoring system for such a repair technique. The study was started 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 A2 from Lucerne through the Gotthardtunnel to Italy. It has two normal lines and 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 side walls of the tunnel are covered with prefabricated, 2.10 m wide elements 226 © 2007, Institute of Materials, Minerals and Mining
Effectiveness of corrosion inhibitors – a field study
227
(panels) with a thickness of only 40 to 50 mm. The reinforcement of the elements 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 chloride measurements) of the whole tunnel, 16 elements, situated approximately in the middle part of the tunnel, were chosen for this investigation. A more detailed condition survey of these 16 elements was undertaken in 1997. The following 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 corrosion potential. At the lower part of the elements (<2.0 m), the chloride content is very high (1.5–2.5% by mass with respect to the mass of cement). At a height of about 2.0 to 3.0 m, the chloride content near the rebars (cover of the rebars: mean value 19 mm) is lower than 0.5% by mass with respect to the mass of cement. Obviously, the corrosion potentials decrease and the intensity of 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 treated with FerroGard-903, ∑ 2 elements treated with FerroGard-903 and Sikagard-701 W (hydrophobic impregnation), ∑ 2 elements treated with Sikagard-701 W. In the autumn of 1997 these test fields were instrumented with the following monitoring 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 general information on the monitoring of concrete structures after a repair, are given © 2007, Institute of Materials, Minerals and Mining
228
Corrosion of reinforcement in concrete Autumn 1997 2.5
Chloride content (mass %/c)
0–1 m 2
1–2 m
1.5
2–3 m
1
0.5
0 0
10
20 30 Depth (mm) (a)
Autumn 1997
40
50
–100
0
2.5
Chloride content (mass %/c)
KG3 2 KG4 1.5
1 KG2 0.5
0 –500
–400
–300 –200 Potential [mV (CSE)] (b)
18.1 (a) Chloride profiles at different heights above ground and (b) relation between chloride content and potential and corrosion state of the rebar (KG: grade of corrosion) in openings: KG 0: blank, no corrosion; KG 1: slight signs of corrosion; KG 2: small areas with corrosion; KG 3: corrosion on the whole surface; KG 4: pitting corrosion.
elsewhere [6]. At the same time cores were taken out of the elements to determine the ohmic resistivity of the concrete as a function of the relative humidity under controlled laboratory atmospheres.
18.3.2 Application of the inhibitors The application of the inhibitors was carried out in June 1998, about eight months after the instrumentation of the test fields. This procedure should © 2007, Institute of Materials, Minerals and Mining
Effectiveness of corrosion inhibitors – a field study
229
have allowed the concrete and the new mortar to regain the equilibrium moisture content. Before the application of the inhibitors the surface of the concrete was washed and cleaned with water. The inhibitors were then applied in several steps. This work was carried out by the providers of the inhibitors under the supervision of the project leader. The amount of FerroGard-903 applied during the treatment was more than 500 g per m2 of concrete surface and thus higher than the recommended target value of 300–500 g m–2. The applied amount of MFP was about 2.4 l m–2. The concentrations of the inhibitors were analysed on concrete cores and concrete dust samples. The first analyses were made directly after the application and the second in the autumn of 1999.
18.3.3 Measurements A full set of measurements was executed after the instrumentation of the test fields, then just before and after the application of the inhibitors (June 1998) and thereafter approximately every 6 months. It included the following measurements: ∑ 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 by data 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 varied from –12.3 to 28.5 ∞C. During rainfall outside of the tunnel, the rh was comparatively 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 potentials measured in the different test fields are shown. The following conclusions can be drawn:
© 2007, Institute of Materials, Minerals and Mining
230
99
99
95 90 80 70
95 90 80 70
Percentage (%)
Percentage (%)
99.9
50 30 20 10 5 1
99.99 MFP 97 MFP 99 MFP 00
99.9 99
50 30 20 10 5 1
.1 0
100
.01 –600 –500 –400 –300 –200 –100 Potential [mV (CSE)] (b)
95 90 80 70 50 30 20 10 5 1
.1
.01 –600 –500 –400 –300 –200 –100 Potential [mV (CSE)] (a)
FG 97 FG 99 FG 00
Corrosion of reinforcement in concrete
99.9
99.99 R 97 R 99 R 00
Percentage (%)
99.99
.1 0
100
.01 –600 –500 –400 –300 –200 –100 Potential [mV (CSE)] (c)
0
18.2 Statistical analysis of the corrosion potentials measured 1997, 1999 and 2000 in the reference fields (R), the fields treated 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 half a year after the elements were placed outside of the tunnel (August).
© 2007, Institute of Materials, Minerals and Mining
100
Effectiveness of corrosion inhibitors – a field study
231
∑ There were only minor changes of the corrosion potentials of the test fields between 1997 and 1999. ∑ In the reference- and MFP-fields a slightly positive shift is apparent in the potential range above –50 mV (CSE). This might be due to a slight reduction in the moisture content of the concrete. ∑ The fields with FerroGard-903 show a slight increase of the potentials in the upper part and slight decrease in the lower part of the curve. This might be caused by the increase of the concrete conductivity of the concrete cover due to the application of the inhibitor (alkaline solutions with salts). ∑ In 2000, the corrosion potentials were generally more negative. The conditions (temperature, humidity) in the elements were different from the 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 profiles of the MFP-elements showed a slight decrease of the potential in the first 10–20 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 isolated rebars are given in the Fig. 18.3 and 18.4. The cleaning of the surface and the 0.12 Cleaning + MFP application Cleaning + FerroGard application
0.1
Macrocell current (mA)
E87, E87, E82, E82,
2.3 0.3 0.5 2.2
mFG mFG mMFP mMFP
June 2./3. 0.08
0.06
0.04
0.02
0 May/30
Jun/2
Jun/4
Jun/6 Jun/9 Date
Jun/11
Jun/13
Jun/16
18.3 Macrocell currents of some electrically isolated rebars during the time of the cleaning process of the surface and the application of the inhibitors. MFP: MFP fields, FG: FerroGard-903 fields.
© 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete 0.12 0.10
20
0.08
0
10
0.06 0.04 0.02 0.00 0.12 0.10
Concrete temp. [∞C]
Macrocell current (mA)
232
–10 E82 MFP, 0.5 m E82 MFP, 2.23 m Concrete temp. E87 FG, 2.32 m E87 FG, 0.33 m
Treatment
0.08 0.06 0.04 0.02 0.00 Jul/1/1998
Jan/1/1999 Date (a)
Jan/1/2000
R (2.09 m) R (0.97 m) MFP (0.96 m) MFP (1.45 m) FG (0.33 m) FG (1.44 m)
0.05
Macrocell current (mA)
Jul/1/1999
0.04 0.03 0.02 0.01 0.00
–0.01 Jul/1/1998
Jan/1/1999
Jul/1/1999 Date (b)
Jan/1/2000
Jul/1/2000
18.4 Concrete temperature and macrocell currents of some electrically isolated rebars of the differently treated test fields located at different heights above ground [(a) data logger; (b): single measurements]. R: reference field, MFP: MFP fields, FG: FerroGard903 fields.
application of the inhibitors led to a sharp increase of the currents by a factor of 1.5 to 3 (increased conductivity of the concrete). This effect is more pronounced in the lower part of the elements, where higher macrocell currents were measured and probably a larger amount of water reached the surface during the cleaning. The short term transients (peaks) are caused by some single steps of the whole process (cleaning, prewetting, application).
© 2007, Institute of Materials, Minerals and Mining
Effectiveness of corrosion inhibitors – a field study
233
The more or less regular variations of the currents within hours or days corresponds to temperature changes. This is more pronounced at higher currents. The highest currents are measured during the summer season. The corrosion process does not stop at temperature below 0 oC. The macrocell currents of the isolated rebars in the lower part of the elements are generally higher than those in the upper part due to the higher chloride content and, thus, more active anodic areas. A clear decrease of the currents 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 the elements (Fig. 18.6). Significant differences between the test fields can not be seen. The variation over time corresponds to the seasonal changes of the temperature (winter/summer). The concrete resistances over the depth (resistance profile) measured with the instrumented concrete cores are shown in Fig. 18.7. The measurements were taken in summer 1998 (after the treatment) and in summer 1999 at similar air temperatures. In most cases, the resistances are higher at the exposed side than in the middle or backside of the elements because of the carbonation of the concrete (increases the resistances). Compared with other structures the profiles are rather flat. The highest resistances are measured in the field with the hydrophobic impregnation, which probably reduced the moisture content of the concrete. Apart from the field with the hydrophobic
Normalized resistance
2.5
R (m) MFP (m) FG (m)
2.0
1.5
1.0
0.5 Jul/1/1998
Jan/1/1999
Jul/1/1999 Date
Jan/1/2000
Jul/1/2000
18.5 Normalized concrete resistances (averages) over time measured with the electrically isolated rebars. R: reference field, MFP: MFP fields, FG: FerroGard-903 fields.
© 2007, Institute of Materials, Minerals and Mining
234
Corrosion of reinforcement in concrete
Normalized resistance
2.5
2.0
R MFP FG
5–12.5 mm
1.5
1.0
0.5 Jul/1/1998
Jan/1/1999
Jul/1/1999 Date (a)
Jan/1/2000
Jul/1/2000
Normalized resistance
2.5
2.0
R MFP FG
20–27.5 mm
1.5
1.0
0.5 Jul/1/1998
Jan/1/1999
Jul/1/1999 Date (b)
Jan/1/2000
Jul/1/2000
18.6 Normalized concrete resistances over time measured with the instrumented cores at different depths: (a) 5–12.5 mm; (b) 20–27.5 mm. R: reference field, MFP: MFP fields, FG: FerroGard-903 fields.
treatment, the test fields do not show significant changes of the concrete resistances. The values are similar in the lower and upper parts of the elements. Figure 18.8 shows the correlation between the macrocell current, measured with the electrically isolated rebars, and the concrete resistance, measured with the instrumented cores (depth 12.5–20 mm), in a logarithmic scale. The linear correlation between macrocell current and concrete resistance indicates an ohmic control of the corrosion process (slope ª –1). The different intensity of the macrocell currents is mainly related to the various anodic areas on the rebars. The application of the inhibitors did not result in a clear decrease of the corrosion activities.
© 2007, Institute of Materials, Minerals and Mining
Effectiveness of corrosion inhibitors – a field study Height over ground >100 cm
2.5 ¥ 104
R MFP FG SG
R MFP FG SG
2 ¥ 104
28.7.99
30.6.98
Temperature approx. 15 ∞C
1.5 ¥ 104
Road side
Resistance (W)
3 ¥ 104
1 ¥ 104
235
5000 0 0
5
10
15
20 25 Depth (mm)
30
35
40
18.7 Concrete resistance over the depth, measured with the instrumented cores. R: reference fields; MFP: MFP fields; FG: FerroGard-903 fields; SG: Sikagard 701-W fields.
Macrocell current (mA)
0.1 E82 2.11 m
7 6 5 4 3 2
E87 2.32 m
0.01 7 6 5 4
MFP before application MFP after application FG before application FG after application
3 2
0.001 3
4
5
6 7 8 Resistance [W]
9 ¥104
18.8 Correlation of the macrocell current, measured with the electrically isolated rebars, and the concrete resistance, measured with the instrumented cores (depth 12.5 to 20 mm). MFP: MFP fields, FG: FerroGard-903 fields.
18.4.4 Galvanic pulse measurements The galvanic pulse measurements were carried out with equipment which was developed by IBWK, ETH Zürich [7]. Figure 18.9 shows the variation of the polarisation resistances measured over the isolated rebars (in this case connected
© 2007, Institute of Materials, Minerals and Mining
236
Corrosion of reinforcement in concrete 800 E78 E80 E82 E85 E86 E87 E92 E93
Polarisation resistance [W]
700 600 500 400
R MFP MFP R FG FG FG/SG FG/SG
300 200 100
0 Apr/15/1998
Nov/2/1998
Apr/7/1999
Jul/28/1999
Date
18.9 Polarisation resistance over time of some electrically isolated rebars of the differently treated test fields: R: reference field, MFP: MFP fields, FG: FerroGard-903 fields, SG: Sikagard 701-W fields.
to the rebar mat) in differently treated test fields over time. The higher resistances of E78R, E85R and E93FG/SG were measured at a height of 2.0 to 3.0 m above ground, where the chloride content is lower than 0.5 mass %/c. There is almost no change in the polarisation resistances of the different elements at the end of the field study. No effect of the treatment is recognised. As with the macrocell currents and the ohmic resistance, the polarisation resistance depends strongly on the temperature too. The ohmic resistances, determined from the pulse measurements, 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 analysis from 1998 and 1999. The analysis methods of the two promoters and TFB are the same. Therefore, the results can be compared directly. The analysis gave, in both cases, inhibitor concentrations near the rebars which are higher Table 18.1 Inhibitor content, analysed from concrete cores [FerroGard-903: organic component (ppm)] Depth (mm) 0–7 10–17 20–27 30–37 40–47
1998 (Analysis by the promoter) 5088 ± 146 168 ± 19 16 ± 1 < 13 < 13
5780 413 41 16
© 2007, Institute of Materials, Minerals and Mining
± 194 ± 28 ±1 ±1 < 13
6709 556 80 26 15
± 342 ± 68 ± 10 ±2 ±2
1999 (Analysis by TFB) 1180 340
760 110
Effectiveness of corrosion inhibitors – a field study
237
Table 18.2 Inhibitor content, analysed from concrete cores and concrete dust samples [MFP: in PO 2– (m.%/concrete)] 4 Depth (mm)
0–10 10–20 20–33 33–44 50–60
1998 (Analysis by the promoter, dust samples) 0.406 0.214 0.237 0.056
0.920 1.093 0.150 0.117
0.229 0.240 0.171 0.090
0.203 0.097 0.055 0.044
1999 (concrete cores) promoter 0.047 0.034 0.055 0.058
TFB 0.250 0.119
0.232 0.081
0.099
0.079
than the target values (MFP: 0.05 mass % in respect to the mass of concrete, FerroGard: approximately 13 ppm organic component). The profiles are steep, especially in the elements treated with FerroGard. The content of FerroGard decreased from 1998 to 1999 (possibly because of evaporation). The mentioned target value near the rebars of 13 ppm corresponds to the limit of the chemical analysis to detect FerroGard. There is no reason why the promoter did not find any MFP in the concrete core in 1999. The mentioned target value near the rebars of 0.05 mass % in respect to the mass of concrete corresponds to the natural ground level of the phosphate content of the concrete.
18.5
Conclusions
The extensively instrumented and monitored field study on the effectiveness of corrosion inhibitors was started in 1997. The side elements (panels) of the walls of the Naxbergtunnel were chosen as test fields. The following conclusions might be drawn: ∑ In both cases were the inhibitor concentrations near the rebars higher than the target values. ∑ No significant effects of the inhibitors MFP and FerroGard-903 on the corrosion of the rebars could be detected either with potential mapping or with the measurements on the electrically isolated rebars or with the instrumented concrete cores. ∑ The hydrophobic impregnation led to an increase in concrete resistance due to the reduced moisture content of the concrete. ∑ Potential measurements are generally possible and give useful results. But they are not sufficient to evaluate the effectiveness of inhibitors. Measurements of the corrosion currents of isolated rebars are a good method to get information about the changes of the corrosion rate. ∑ The instrumentation as well as the monitoring (combination of manual measurements with data logging) has proven to be appropriate.
© 2007, Institute of Materials, Minerals and Mining
238
18.6
Corrosion of reinforcement in concrete
Acknowledgements
The authors gratefully acknowledge the Construction Department of the Canton 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 of the 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 steel corrosion’, 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 techniques for monitoring the conditions of concrete bridge structures’, Int. Conf. Repair of Concrete Structures’, Solvear, May 1997, 213–222. 7. B. Elsener, D. Flückiger, H. Wojtas, H. Böhni, ‘Methoden zur Erfassung der Korrosion von Stahl in Beton’, VSS-Ber., 521, 1996.
© 2007, Institute of Materials, Minerals and Mining
19 Corrosion 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
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 been focused on the design of new structures to reduce or eliminate corrosion through increased concrete coverage using reduced permeability concrete or replacing the steel reinforcement with alternative materials. However, little effort has been made in establishing reliable techniques for the repair of existing structures. Since many of the structures built after WWII are reaching the end of their design life and there are no plans to replace them, a rehabilitation programme is necessary. It was cited in a 1993 survey by the Strategic Highway Research Program in the United States that the cost of repairing bridge decks that had suffered chloride-induced deterioration was $20 billion and was increasing at a rate of $500 million annually. Reinforcing steel embedded in concrete shows a high amount of resistance to corrosion. The cement paste in the concrete provides an alkaline environment that protects the steel from corrosion. This corrosion resistance stems from a passivating or protective ferric oxide film that forms on the steel when it is embedded in concrete. This film is stable in the highly alkaline concrete (pH approx. 11–13). The corrosion rate of steel in this state is negligible. Factors influencing the ability of the rebar to remain passivated are the water-tocement ratio, permeability and electrical resistance of the concrete. These factors determine whether corrosive species like carbonation and chloride ions 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 ions become 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 protect the embedded steel rebar/concrete structure. These inhibitors can be organic 239 © 2007, Institute of Materials, Minerals and Mining
240
Corrosion of reinforcement in concrete
or inorganic compounds; however organic compounds seem to be more effective (for neutralising and film forming). Recent MCIs are based on amino-carboxylate chemistry [1–3]. Normally, the most effective type of inhibitor lessens corrosion at the anodes and cathodes simultaneously. Organic inhibitors are a subgroup of the combined inhibitors. They utilise compounds that work by forming a monomolecular film between the metal and the water. These compounds are polar and have a strong affinity for the surfaces onto 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. These molecules will arrange themselves parallel to one another and perpendicular to the reinforcement such that a continuous barrier is formed. The presence of this film on samples of reinforcement encased in concrete with an organic inhibiting admixture has been shown by methods of ultraviolet spectroscopy and gas chromatography [6]. These types of inhibitors are known as migrating corrosion inhibitors if they are able to penetrate into existing concrete to protect the steel in the presence of chloride [7]. The means by which the inhibitor migrates is first by diffusion through the moisture that is normally available in concrete, then by its high vapor pressure and finally by following hairlines and microcracks. This mechanism allows a greater amount to be applied where it is most needed. The diffusion process requires time to migrate through the concrete pores to reach the rebar’s surface and form a protective layer. This suggests that the migratory inhibitors are physically adsorbed onto the metal surfaces [1]. MCIs can be incorporated as an admixture or can be used by surface impregnation of existing concrete structures. With surface impregnation, diffusion transports the MCIs into the deeper concrete layers. They will delay and inhibit the onset of corrosion on steel rebar. Bjegovic and Miksic recently demonstrated the effectiveness of MCIs over five years of continuous testing [1–3]. They also showed that a migrating amine-based corrosioninhibiting admixture can be effective when it is incorporated in the repair process of concrete structures [2]. Furthermore, laboratory tests have proven that MCI corrosion inhibitors migrate through the concrete pores to protect the rebar against corrosion even in the presence of chlorides [3–4]. However, the amount of additive inhibitor should be calculated based on the concrete chloride 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 corresponds to 0.10% soluble chloride ion by weight of cement [7–8]. McGovern [11] reports work by the United States Federal Highway Administration Laboratories which suggests that the threshold value for steel corrosion is 0.20% acidsoluble chlorides by weight of cement. This is equivalent to between 0.6 and 0.8 kg of chlorides per cubic metre of concrete. The chloride threshold concentration is generally within 0.9 to 1.1 kg of chlorides per cubic metre of concrete [5]. © 2007, Institute of Materials, Minerals and Mining
Corrosion protection of steel rebar in concrete
241
The objective of this investigation was to study the corrosion inhibition of commercially available migrating corrosion inhibitors on steel rebar in three concrete densities. Theoretically, high-density concrete impedes corrosive species from reaching the surface of the rebar. It may also prevent the inhibitor from reaching the surface of the concrete. Electrochemical monitoring techniques were applied while samples were immersed in 3.5% NaCl at ambient temperature. Because of the low conductivity of concrete, the corrosion behavior of steel rebar had to be monitored using AC electrochemical impedance spectroscopy (EIS). During this investigation, changes in the polarisation resistance and the corrosion potential of the rebar were monitored to ascertain the degree of effectiveness of these MCI products. The results were compared with previous investigations conducted on several admixtures and stainless steel rebar. X-ray photoelectron spectroscopy (XPS) and depth profiling were 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 porous solution that can be modelled by a Randles electrical circuit. EIS tests performed on a circuit containing a capacitor and two resistors indicate that this model is an accurate representation of an actual corroding specimen. EIS testing allows for the determination of fundamental parameters relating to the electrochemical kinetics of the corroding system. It involves the application of a small-amplitude alternating-potential signal of varying frequency to the corroding system. Because processes at the surface absorb electrical energy at 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 measure of the polarisation resistance or the resistance of the surface of the material to corrosion. RW is a measure of the solution resistance to the flow of the corrosion current. By monitoring the Rp value over time, the relative effectiveness of the sample against corrosion can be determined. If the specimen maintains 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 displays a decreasing Rp value over time it is corroding and the inhibitor is not providing corrosion resistance. The EG&G Instruments Potentiostat/Galvanostat Model 273A and EG&G M398 Electrochemical Impedance Software were used to conduct these experiments and to record the results. Bode and Nyquist plots were produced from the data obtained using the single sine technique. Potential values were recorded and plotted with respect to time. By comparing the Bode plots, changes in the slopes of the curves were monitored as a means of establishing a trend in the Rp value over time. To verify this analysis, the Rp values were also estimated by using a curve fit algorithm on the Nyquist plots (available in the software). © 2007, Institute of Materials, Minerals and Mining
242
Corrosion of reinforcement in concrete
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 low frequency range of the Bode plot and the RW value can be seen in the high frequency range of the Bode plot. The diameter of the Nyquist plot is a measure of the Rp value. Concrete samples with dimensions 8≤ ¥ 4≤ ¥ 4≤ (approx. 200 mm ¥ 100 mm ¥ 100 mm) were prepared, and their densities were adjusted to achieve 130, 140, and 150 lb ft–3 (2.08, 2.24 and 2.40 g cm–3). Each sample consisted of one 8≤ (20 cm) steel (class 60) rebar 1/2≤ (12.7 mm) in diameter and one 8≤ (20 cm) Inconel metal strip (counter electrode). The rebars, before being placed in concrete, were exposed to 100% rh (relative humidity) to initiate corrosion. The coverage layer was maintained at one inch (25.4 mm) of concrete 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 strength than the high density samples. The concrete blocks were sandblasted to remove loose particles, leaving the concrete with a marginally smoother surface. Two coats of MCI 2022 and MCI 2021 were applied with a paint brush to all but two of the concrete samples (used as a control). The samples were then immersed in 3.5% NaCl solution [roughly 7≤ (17.8 cm) of each sample was immersed in the solution continuously]. EIS (electrochemical ac impedance spectroscopy) testing started 24 h after immersing the samples. A Cu/CuSO4 electrode was used as the reference and each sample was tested weekly. XPS analyses were performed on steel rebar that was in concrete treated with MCI, immersed for 400 days, using a KRATOS AXIS ultra Xray photoelectron spectrometer, and for the depth profiling, sputtering was conducted using a 2 kV Ar+ ion gun. The thickness of the deposited film was estimated from the rate of removal of Ta2O5.
19.3
Results and discussion
The corrosion inhibition of two commercially available migrating corrosion inhibitors (Cortec MCI 2022 and 2021) for three concrete densities was investigated over a period of 400 days using ac electrochemical impedance spectroscopy (EIS). Throughout this investigation, changes in the polarisation resistance and the corrosion potential of the rebar were monitored to determine the degree of effectiveness for Cortec MCI 2021 & 2022 products. According to the ASTM (C876) standard, if the open circuit potential (corrosion potential) is –200 mV or higher, this indicates a 90% probability that no reinforcing steel has corroded. Corrosion potentials more negative than –350 mV are assumed to have a greater than 90% likelihood of corrosion. In Fig. 19.1 the corrosion potentials for the untreated control samples dropped from –200 mV to –545 mV, which indicates a 90% probability of corrosion attack on
© 2007, Institute of Materials, Minerals and Mining
0
Potential (mV)
–200
L2022–1 L2022–2 L2021–1 L untreated L2021–2 H2022–1 H2022–2 H untreated H2021–1 H2021–2
–300
–400 –500
–600 0
50
100
150
200 250 300 Time of submersion (d)
350
400
450
19.1 Corrosion potential of steel rebar vs time, ASTM C876-91, protected by Cortec MCI 2022 & 2021 compared with unprotected concrete (various concrete densities). L = low density, H = high density, 1 = sample 1, 2 = sample 2.
Corrosion protection of steel rebar in concrete
–100
243
© 2007, Institute of Materials, Minerals and Mining
244
Corrosion of reinforcement in concrete
the reinforcing steel. The corrosion potentials for MCI treated concrete samples were –120 to –150 mV. The past six years of data analysis for immersed samples show that the ASTM C876 criteria can be used to validate the probability of corrosion attack. In Fig. 19.2 the resistance polarisation for MCI treated concrete samples gradually increased from 10 000 to 100 000 ohms. The Rp value for a nontreated sample (control) was 10 000 at the beginning of the experiment and ended at less than 700 W. Changes in the Rp value were observed after about 90 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) concrete samples. Preliminarily data show that MCI treated concrete samples display an increase in their Rp values compared with the control samples that show a decreasing trend. XPS analysis demonstrated the presence of inhibitor on the steel rebar surface; MCI was able to penetrate through the concrete coverage layer and reach the rebar to retard corrosion. Figure 19.4 shows the XPS spectrum for the rebar removed from the MCI treated sample after 400 days. Figure 19.5 shows depth profiling results using 2 kV Ar+ ions for a steel rebar removed from MCI treated concrete, showing that a 100 nm layer of amine-rich compound is present on the rebar surface. The high resolution XPS analysis of carbon and oxygen showed the organic compound to have carboxylate chemistry. The inhibitor is an amine-based carboxylate organic compound, therefore it was concluded that the film on the rebar surface was the inhibitor molecules. Chloride was detected at about 0.10 atomic % and up to 50 nm deep on the top surface of the rebar. The XPS results demonstrate that both MCI and corrosive species migrated into the concrete samples, but MCI managed to protect the steel rebar. The lower density samples coated with 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 by diffusion through moisture that is normally available in concrete, then by its high vapour pressure and finally by following hairline cracks and microcracks. Therefore, lower density concrete samples provide an easier path for the inward diffusion of MCI, and faster corrosion retardation. These results are extremely promising for the MCI product in its ability to protect steel rebar in concrete in aggressive environments.
19.4
Conclusions
Corrosion inhibition by two commercially available migrating corrosion inhibitors (Cortec MCI 2022 and 2021) on steel rebar in concrete was investigated while the concrete was immersed in 3.5% NaCl at ambient
© 2007, Institute of Materials, Minerals and Mining
1.00E+05 2022–day 3 2022–day 375 Control day 4 Control day 378
|Z | (W)
1.00E+03
1.00E+02
1.00E+01 1.00E–04
1.00E–03
1.00E–02
1.00E–01
1.00E+00 Frequency (Hz)
1.00E+01
1.00E+02
1.00E+03
1.00E+04
Corrosion protection of steel rebar in concrete
1.00E+04
19.2 EIS results, Bode plot for MCI 2022 (concrete density 14 lb ft–3). Comparison with control samples.
245
© 2007, Institute of Materials, Minerals and Mining
246
1.00E+05 2L-Day 238 L-Day 233 2S-Day 236
2L-Day 325 L-Day 332 2S-Day 327
|Z | (W)
1.00E+04
1.00E+03
1.00E+02
1.00E+01 1.00E–04
1.00E–03
1.00E–02
1.00E–01
1.00E+00 1.00E+01 Frequency (Hz)
1.00E+02
1.00E+03
1.00E+04
1.00E+05
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.
© 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete
2L-Day 1 L-Day 1 2S-Day 1
Corrosion protection of steel rebar in concrete
247
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 ¥103 80
O 1s
Cu 2p
Peak Position BE (eV) Fe 2p 710.4 O 1s 531.2 N 1s 398.5 C 1s 285.0 Si 2p 101.8 Cu 2p 935.2 Cl 2p 196.2
Intensity (cps)
60
FWHM Raw area (eV) (CPS) 4.1 5520.2 3.0 10963.2 2.1 416.3 2.5 5401.2 2.5 1054.3 1.4 656.1 1.6 76.1
Fe 2p
RSF 2.957 0.780 0.477 0.278 0.328 5.321 0.622
Atomic mass 55.8 15.9 14.0 12.0 28.0 63.5 35.5
Atomic conc% 4.39 34.54 2.24 50.34 7.42 0.27 0.09
C
40
20
N 1s
1000
800
600 400 Binding energy (eV)
Si 2s Si 2p
Cl 2p
200
0
19.4 XPS on MCI 2022 treated concrete after 378 days. Large area (1000 ¥ 800 mm) survey scan from corroded surface. 60
Concentration (%)
50 O C Fe Si N Cl
40
30
20
10
0 0
200
400
600
800 1000 1200 Etch time (s)
1400
1600
1800
2000
19.5 XPS depth profile on steel rebar removed from MCI treated concrete sample after 378 days of testing (etched using 2kV Ar+ ions). © 2007, Institute of Materials, Minerals and Mining
248
Corrosion of reinforcement in concrete
temperatures using electrochemical monitoring techniques. The MCI products have successfully inhibited corrosion of the rebar in a 3.5% NaCl solution for 400 days. Steel rebar corrosion potentials were maintained at approximately –150 mV, and rebar polarisation resistance showed a gradual increase reaching as high as 100 000 W. However, the low density concrete demonstrated better protection than the other samples which is in agreement with the migration mechanism of these inhibitors. XPS analysis verified the presence of the inhibitor on the steel rebar surface indicating MCI migration through the concrete layer. Depth profiling showed a 100 nm layer of amine-rich carboxylate compound on the rebar surface, which assures satisfactory corrosion resistance even in the presence of chloride ions. In summary, the experimental results demonstrate that the MCI products offer an inhibiting system for protecting reinforced concrete in an aggressive 3.5% NaCl solution. These results are extremely promising for the protection of steel rebar and concrete in aggressive environments.
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 migrating corrosion 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 in Natural and Industrial Environments: Problems and Solutions, Grado, Italy, May 23–25, 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-inhibiting admixture for reinforced concrete’, Concrete Int., 1992, 14(4). 6. P. H. Emmons and V. M. Alexander, ‘Corrosion protection in concrete repair myth and reality’, Concrete Int, 1997, 19(3), 47–56. 7. D. Stark, ‘Influence of design and materials on corrosion resistance of steel in concrete’, Res. Dev. Bull., RD098.01T, Portland Cement Association, Skokie, Illinois, 1989. 8. W. Hime and B. Erlin, ‘Some chemical and physical aspects of phenomena associated with chloride-induced corrosion’, Corrosion, Concrete and Chlorides: Steel Corrosion in Concrete: Causes and Restraints, Frances W. Gibson (Ed.), American Concrete Institute, Detroit, Michigan, 1987. 9. W. J. Jang and I. Iwasaki, ‘Rebar corrosion under simulated concrete conditions using galvanic current measurements’, Corrosion, 1991, 47(11), 875–884. 10. T. Liu and R. W. Weyers, ‘Modeling the dynamic corrosion process in chloride contaminated concrete structures’, Cement Concrete Res., 1998, 28(3), 365–379. 11. M. S. McGovern, ‘A new weapon against corrosion’, Concrete Repair Dig., June, 1994.
© 2007, Institute of Materials, Minerals and Mining
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 the 14th NCUR, Missoula, Montana, April 2000.
© 2007, Institute of Materials, Minerals and Mining
20 Determination 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
20.1
Introduction
20.1.1 Permeability of coatings on concrete Permeation of water (H2O) and carbon dioxide (CO2) through coatings is important for coated concrete structures and has a big influence on the durability of the entire structure. Such coatings have to show a low permeability for CO2 and liquid H2O, but a high permeability for water vapour. The concrete has to allow water to evaporate, especially when structures suffer from ascending water, leakage or humidity inside a building. A barrier against CO2 is necessary to avoid carbonation, which leads to severe corrosion problems of the embedded rebars (reinforcing steel) and is probably the most severe durability problem of concrete structures. Surface protection systems are applied to assure the durability and lifetime of concrete structures. Surface protection systems fulfil special requirements, such as CO2–resistance, H2O vapour permeability and water resistance. Besides these three major properties, others have to be achieved, depending on the purpose and environmental conditions of the structure, for example crack bridging, chemical resistance to industrial atmospheres, resistance to deicing salts, and colour resistance against UV irradiation. In order to quantify the diffusion of CO2 and H2O, a European Standard CEN 1062 was created. Therein, the diffusion resistance of a coating is expressed as the equivalent thickness of an air layer (sD), having the same resistance against CO2 diffusion as the coating. The sD–value is obtained from permeability measurements using two different atmospheres with known partial pressures of CO2. From the measurements described below a diffusion resistance value (m) is obtained by using the formula given in Fig. 20.1. The diffusion equivalent air layer thickness sD is obtained by multiplying m by the coating thickness (d) of the investigated samples (sD = md). The sD value for CO2 should be high and the sD value for water vapour should be low. If the partial pressures are not sufficiently known the value of m can also be obtained 250 © 2007, Institute of Materials, Minerals and Mining
Determination of coating permeability on concrete using EIS
251
Free film or film on substrate
Cup Absorbance
Ê ˆ p – p2 m = 1 Á dL A 1 – SL ˜ ¯ s Ë I WDD =
m=
24 m2 – m1 ; t 2 – t1 A
Ê g ˆ Á 2 ˜ Ëm d ¯
(1)
(2)
k WDD S
20.1 Design of the cup test. Equation 1 describes the water diffusion value based on partial pressure differences and equation 2 is used when weighing is carried out for the determination of the amount of diffused gas.
14 12
SD = 0
Depth (mm)
10 8
SD = 1.0 m
6 4
SD = 50 m
2
SD = •
0 1
4 Years
9 –1 (a 2
16
25
)
20.2 Carbonation depth as a function of time under the influence of different equivalent air layer thickness (principle plot, based on calculations).
by determination of the amount of diffused H2O via the WDD value (water vapour transmission rate; see also Fig. 20.1 and 20.2) For details please refer to the CEN standard [1].
© 2007, Institute of Materials, Minerals and Mining
252
Corrosion of reinforcement in concrete
20.1.2 Discussion of known techniques for the determination of coating permeability For permeability measurements, the traditional cup test is used. A H2O absorbant or a CO2 absorbant is placed in a cup sealed carefully with the film under investigation (see Fig. 20.1). Then the sealed cup is weighed and placed in an atmosphere containing CO2 or water vapour. After a certain time the weight increase is measured and this can be directly related to the permeability of the coating in the chosen atmosphere. The cup method involves the use of free films, with all the related problems of their preparation, especially when the substrate (mostly plastic or aluminum foils) has to be removed mechanically. A second way of producing a film for permeation testing is by using porous glass filters as a substrate. In this method, the film stays on its substrate and does not have to be removed before testing. However, it is hardly possible to say that a glass substrate has exactly the same properties as mortar or concrete. Moreover, it is likely that the properties of the coating show some influences which are caused by using 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 exposed to a CO2 atmosphere and, after a certain exposure time, the samples were cut. At the cut, the mortar is treated with an alkalinity indicator such as phenolphthalein and those areas which have been neutralised by the diffusing CO2 will be visible because the indicator remains colourless (pH < 9). This method cannot be applied for H2O diffusion measurements, because no sensitive and reliable indicator for water or humidity exists for slices of concrete or mortar, 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 of coatings on concrete. The method is based on conductivity measurements of the mortar or concrete, using two stainless-steel screws embedded in mortar prisms as conductivity sensors. The conductivity of the mortar is measured using electrochemical impedance spectroscopy (EIS), with which additional information is obtained. The conductivity of mortar and concrete depends on the humidity of the specimens. Any change in the humidity level becomes visible by a changed conductivity. 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 the concrete formulation, including cement type, water-to-cement (w/c) ratio and soluble contaminants of the sand and aggregates. The degree of hydration
© 2007, Institute of Materials, Minerals and Mining
Determination of coating permeability on concrete using EIS
253
certainly has a significant influence on the conductivity. Therefore, it is necessary to compare only specimens which are prepared identically and are formulated, manufactured and conditioned simultaneously. The main experimental approach was to investigate the conductivity of dried and coated concrete before and after exposure to 100% relative humidity. The measurements were repeated over a certain period, in order to monitor the change in the mortar conductivity due to humidity uptake (the detailed procedure is shown in Fig. 20.3). Sample preparation and conditioning were done as follows: the mortar was made with a water/cement ratio of 0.4, using a cement/sand ratio of 2:3 with a grain size <0.6 mm (higher grain sizes could cause inhomogenity over the cross section of the prism, which might result in non-uniform diffusion of water). The dimensions of the prisms were 4 ¥ 4 ¥ 16 cm, with removal from the mould after 24 h and an additional 3 days at 100% rh for sufficient curing.
Preparation of the mortar, proper fixing of the screws, curing and demoulding, final curing for 3 days at 100% rh, 23 ∞C and 3 days at 23 ∞C, 50% rh
7 days at 100% rh
7 days at 60 ∞C ambient humidity
6 prisms uncoated 6 coated twice with acrylic 6 coated twice with epoxy
6 prisms uncoated 6 coated twice with acrylic 6 coated twice with epoxy
All coated samples cured at 23 ∞C and 50% rh for 7 days
All samples conditioned for 4 weeks at 98% rh
All samples conditioned for 4 weeks at 60% ambient humidity
Samples transferred from dry to humid (98% rh) Measurements started here
Samples transferred from humid to dry. Measurements started here
Note: 100% rh is achieved by spraying water (fog room); 98% rh is achieved in a cabinet with saturation in water vapour.
20.3 Experimental procedure.
© 2007, Institute of Materials, Minerals and Mining
254
Corrosion of reinforcement in concrete
(a)
(b)
20.4 (a) Principle design of specimen and (b) photographs of a prism, showing one of the embedded screws.
These electrodes are screws in the 4 ¥ 4 cm2 faces (Fig. 20.4). Six prisms were prepared for each system: uncoated, coating system 1 and coating system 2. The coating materials were applied in two layers each. System 1 had a water-based acrylic binder and system 2 consisted of a solvent-based epoxy. Impedance measurements were performed for monitoring the conductivity of the mortar. A Zahner IM6 impedance spectrometer was used. To measure impedance spectra, a sinusoidal potential with a given frequency and amplitude was applied and the resulting current was measured, whereas the modulus (ratio of U0 and I0) and the phase shift are plotted over the frequency in double logarithmic scale (Bode plot). In Fig. 20.5a the principle of EIS measurements is shown in the upper diagram. The spectra were recorded
© 2007, Institute of Materials, Minerals and Mining
Determination of coating permeability on concrete using EIS
255
1.2
1.2
IZ (w )I = U0 /i 0
Voltage (arbitrary units)
0.4
0.4
U0
i0
0
0
–0.4
–0.4 Phase shift
–0.8
0
2
–0.8
U(t ) = U0 sin (w t ) i (t ) = i0 sin (w t + q )
tan q = Im (Z )/Re(Z )
–1.2
Current (arbitrary units)
0.8
0.8
–1.2
4 6 Time (arbitrary units) (a)
8
10
9
6
N
5
4 8
N
2 1 (b)
20.5 (a) Principle relation between current and voltage in the case of EIS and (b) the equivalent circuit, suitable for modelling of mortar samples containing conductivity sensors made of stainless steel.
over a frequency range of 0.1 to 100 kHz, five points per decade with five samples for each point and an amplitude of 5 mV (10 mV peak to peak). For data presentation the Bode plot is preferred, because of the strong change of spectral 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
© 2007, Institute of Materials, Minerals and Mining
256
Corrosion of reinforcement in concrete
electrochemical system by curve fitting. Such a model is assembled with capacitors, resistors and other electronic circuit elements, each describing a certain property of the measured system. A relatively complicated example is given in Fig. 20.5b. It is suitable for modelling the EIS measurements of all related mortar prisms, independent of their state. The resistor 4 represents the electrolyte resistance in the mortar, which is directly linked with the conductivity of the mortar. All the other elements are required for a proper modelling of the complicated signals generated from the two stainless steel screws, the wiring and the mortar itself. Although it has been proven that this model is suitable for all the spectra obtained from the mortar prisms, the results from the EIS measurements are clear enough to show evidence for the usefulness of this new method, even without modelling of the EIS data. In order to demonstrate the different water vapour permeability of the water-based acrylic and the solvent-based epoxy, two experimental paths were followed. For the first path, the samples were kept humid before coating and then, after sufficient curing over 3 days at 23 ∞C and 50% rh, they were transferred into a dry environment. Here, the rate of evaporation of humidity from the mortar samples was measured via an increase of the mortar resistance with drying time. Using the second path, the samples were first kept under dry condition and then transferred into 100% humidity. Directly at the beginning and after each day, impedance spectra were recorded. The complete sample preparation procedure is shown in Fig. 20.3.
20.3
Results and discussion
In Fig. 20.6, 20.7 and 20.8 the resulting spectra of the uncoated and coated specimens are shown. The consecutive spectra of one specimen are plotted together 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 the coated samples it is evident that the uncoated samples show a much faster change 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 is not at all astonishing and is to be expected, but its fulfilment is highly necessary to provide confidence in the method. Going into more detail: in Fig. 20.6 the first recorded spectrum for the humid condition has the lowest impedance and with longer drying time the impedance increases from day to day. The equivalent behaviour is obtained when recording of the series of spectra is started in the dry state: The first spectrum shows the highest impedance and the phase angle is in the range of –90∞. The mortar resistance could be extracted in the frequency range between 10 to 10000 Hz. The plateau of the modulus of most spectra is directly linked with the mortar resistance.
© 2007, Institute of Materials, Minerals and Mining
Determination of coating permeability on concrete using EIS
257
90
75
3M
60
1M
45
300K
30
100K
15
30K 100m
1
10 100 1K Frequency (Hz) (a)
10K
Phase temperature (∞C)
Impedance (W)
10M
1tro000 1tro001 1tro002 1tro003 1tro004 1tro005 1tro006 1tro007
0 100K
90 10M
60 100M 45
30 1M
Phase temperature (∞C)
Impedance (W)
75 akli000 akli001 akli002 akli003 akli004 akli005 akli006 akli007
15
100m
1
10
100 1K Frequency (Hz) (b)
10K
0 100K
20.6 Results from the uncoated samples. Last number is related to the number of days in the respective environment: (a) change from humidity to dryness; (b) change from dryness to humidity.
Only the first impedance curve in the initial dry state shows no plateau because of its very high impedance values due to poor conductivity. But already the second and third curves begin to show this plateau at low frequencies. The arrows indicate the sequence of the increase or decrease of impedance with time. Comparing Fig. 20.7 and 20.8 with Fig. 20.6 it emerges that the uncoated samples showed the strongest change of impedance, followed by the less
© 2007, Institute of Materials, Minerals and Mining
258
Corrosion of reinforcement in concrete 90
Impedance (W)
1M
75 4tro000 4tro001 4tro002 4tro003 4tro004 4tro005 4tro006 4tro007
60 300K 45 100K
30
15 30K 100m
1
10 100 1K Frequency (Hz) (a)
10K
0 100K
90 10G 75 60 100M 45 10M 30 1M 15
Phase temperature (∞C)
Impedance (W)
1G dkli000 dkli001 dkli002 dkli003 dkli004 dkli005 dkli006 dkli007
100K 100m
1
10 100 1K Frequency (Hz) (b)
10K
0 100K
20.7 EIS measurements of the samples with the waterbased acrylic coating. Last number is related to the number of days in the respective environment: (a) change from humidity to dryness; (b) change from dryness to humidity.
pronounced change of those coated with the water-based acrylic paint and the least change was obtained with the solvent-based epoxy coated species. Actually, the change of impedance from ‘dry’ to ‘wet’ for the coated samples is extremely small and slow (Fig. 20.7 and 20.8). It clearly demonstrates the ability 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, independent of their subsequent coating, the impedance values after 7 days (recorded at 1
© 2007, Institute of Materials, Minerals and Mining
Determination of coating permeability on concrete using EIS
259
90 500
Impedance (W)
75
60 200 45 100 30 50
100m
7tro000 7tro001 7tro002 7tro003 7tro004 7tro005 7tro006 7tro007
15
1
10 100 1K Frequency (Hz) (a)
10K
0 100K
90 10G 75
Impedance (W)
60 100M 45 30 1M
gkli000 gkli001 gkli002 gkli003 gkli004 gkli005 gkli006 gkli007
15
100m
1
10 100 1K Frequency (Hz) (b)
10K
0 100K
20.8 EIS measurements of the samples with the solvent-based epoxy coating. Last number is related to the number of days in the respective environment: (a) change from humidity to dryness; (b) change from dryness to humidity.
Hz) are quite easy to understand. These values are given in Table 20.1, together with those of the dry state before coating. It should also be mentioned that the corresponding uniform impedance value was observed for the samples of the dry 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 between the different coating materials, although the differences are not too pronounced between the two types of coatings. The reason for the relatively small difference
© 2007, Institute of Materials, Minerals and Mining
260
Corrosion of reinforcement in concrete
20.9 Cross-section through typical mortar specimen. Two honeycombs become visible. Reduced thickness of the coating, but coating 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) Uncoated Water-based acrylic Solvent-based epoxy
0.055 20 2 0.6
5000 0.4 1000 3000
might be seen in the imperfect surface preparation before the coating was applied. In Fig. 20.9 a cross section is shown in which some traces of the imperfect preparation are visible. Honeycombs persist on the surface, resulting in pores or insufficient coating thickness. This balances the different coating resistance against water vapour because of weak points allowing easy diffusion, similar to uncoated mortar. Nevertheless, even though these weak points were found, the method revealed significant differences between the coatings. Further improvement of the surface pretreatment, examples of which are shown in Fig. 20.10, will certainly allow the sensitivity of the method to be increased with respect to smaller differences between coating materials.
20.4
Conclusions
Traditional testing techniques for the water vapour permeability of coatings require free films or films on porous glass plates. The new method using EIS allows measurements on a more realistic substrate (mortar). © 2007, Institute of Materials, Minerals and Mining
Determination of coating permeability on concrete using EIS
261
Original surface
A1 Blast cleaning with blasting agents A2 High-water pressure blasting
(a)
Flexible filler
Coating
(b)
20.10 Two possible techniques are shown to overcome the honeycomb problem: (a) abrasive blasting removes loosely adherent mortar and opens the honeycomb for better paintability; (b) a filler may be applied in order to seal the honeycombs and to act as a smooth surface for easier paint application.
EIS allows uncoated, one-component acrylic or two-component epoxy coatings to be distinguished. So far, the qualitative comparison of coatings is possible. Absolute and quantitative 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 of surface preparation could be recommended, but no results are available so far.
20.5
References
1. CEN 1062, Paints and Varnishes – Coating Materials and Coating Systems for Exterior Masonry and Concrete: Part 2: Determination and classification of water vapour
© 2007, Institute of Materials, Minerals and Mining
262
Corrosion of reinforcement in concrete
permeability; Part 3: Determination and classification of liquid–water transmission rate (permeability); Part 6: Determination of carbon dioxide permeability; Part 11: Methods of conditioning before testing. ISO 7783-2, Ausgabe:1999-03, Lacke und Anstrichstoffe Beschichtungsstoffe und Beschichtungssysteme für mineralische Untergründe und Beton im Außenbereich - Teil 2: Bestimmung und Einteilung der Wasserdampf-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 mit der Mörtelmethode’ Farbe Lack, 1994, 100, 100. 4. J. Tritthart and H. Gehmayer, Zement Beton, 1985, 1, 74–79.
© 2007, Institute of Materials, Minerals and Mining
21 Chloride 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
21.1
The task
In 1999, a research project was launched to improve and to extend the wellknown merits of electrochemical chloride extraction (CE) from reinforced concrete. 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 able to be scaled up, being replaceable and supported by the manufacturers over a long period of time.
21.2
The solution
First of all, the problem of the inhomogeneity of the concrete surface had to be solved. For this purpose, therefore, a grid overlay with cell dimensions of 60 ¥ 60 cm was applied to the surface, where the cells could be investigated *The manuscript of this chapter was submitted originally for the EUROCORR 2001 and represents the state of knowledge and development in early 2001. Scientific and technological progress since then has led to improved solutions.
263 © 2007, Institute of Materials, Minerals and Mining
264
Corrosion of reinforcement in concrete
on a scale that was small enough to take all special conditions into account. A more 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 must be related to the area of the rebar surface instead of the concrete surface. With the division of the concrete surface into small areas, different conditions of concrete cover, permeability or related rebar surface could be considered in such a way that every cell is treated separately in order to achieve the optimal result – to undertake the CE as quickly as possible and without overheating. Having characterised the concrete surface, neighbouring cells with similar characteristics of concrete cover, rebar surface, permeability and chloride content can be combined again and controlled by a representative cell in order to save expensive hardware. In order to prevent the evolution of chlorine gas, an ion exchanger capable of regeneration almost without producing waste, was included within the CE process. Earlier investigations about the possible amount of chlorine gas evolution [3] drove the decision to include such an item.
21.2.2 Hardware components A combination electrode (60 ¥ 60 cm) was designed that contains everything needed on the concrete surface for running CE: a dimensionally stable anode and an electrolyte reservoir (fibreglass). Apart from this, there is an ion exchange layer and an outer stabilising plate that also provides an evaporation protection. This sandwich electrode is mounted using a centrally positioned nylon 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 via Ethernet to a central computer. The extraction voltage is also supplied centrally by high current power supplies (Fig. 21.1). For the supply of electrolyte, a water treatment system de-ionises tap water, adds some NaOH for raising the conductivity and distributes the electrolyte between the electrodes via a system of hoses.
21.2.3 Control components The chloride extraction process works by measurement and control within the cells. This means that cells or groups of cells are measured and controlled individually, being ruled by centrally stored, individually defined start values. The CE system is scalable and has no limitations on the number of cells or groups of cells (Fig. 21.2).
© 2007, Institute of Materials, Minerals and Mining
Public internet/ GSM connection/ telephone
Gateway
Local PC-based control and supervision
Ethernet
High current power supply Voltage bus
Data acquisition
Actor and sensor control for – switch – reference cell – voltage and current Combination electrode active/passive Grid layout on a bridge structure
Chloride extraction from reinforced concrete
Laptop Remote PC-based control and supervision
21.1 Schematic layout of the CE operating system.
265
© 2007, Institute of Materials, Minerals and Mining
266
Corrosion of reinforcement in concrete
21.2 Control menu for the CE process (graph bar, physical configuration, logging).
All electrodes are supplied by a common voltage source – if possible at 40 V DC. According to their predicted current consumption, the electrode groups are divided into two groups and are switched alternately. When limiting criteria (e.g. reference potential or current) are exceeded, the related groups will be switched off until their next on-period. For safety reasons, several internal checks will shut down the process if signals are missing or values are out of range or an alarm state is reached. The CE program will run locally at the site, but can be remotely controlled and provides the opportunity for alarm 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 charge that was needed during laboratory tests with various concrete compositions, which can be compared with the concrete found on-site. The recorded amount of charge per group of the electrodes provides the first on-line indication of the possible success of the treatment. Tools can be added to signal the end of © 2007, Institute of Materials, Minerals and Mining
Chloride extraction from reinforced concrete
267
treatment or saturation of the ion-exchange module. An evaluation tool can be used to monitor and analyze the whole CE process and to provide a detailed documentation 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 the structure has to be done (Fig. 21.3). In addition to other information, the following physical data are directly transferred into the CE control system: ∑ the rebar surface area for each cell (equals the curved surface of the outer rebar layer); ∑ the minimum concrete cover; ∑ the average concrete cover. Other data – mainly rest potentials and chloride contents – cannot be taken into account automatically because there are not enough mathematical dependencies on which to base valid formulae. Thus, the operator has to evaluate those data manually. Usually, this evaluation will be supported by the trial desalination of some core samples taken from the structure in advance and representing the actual concrete constitution.
21.3.2 Design of a chloride extraction application With the imported grid layout and the data it contains, groups of cells must be configured. Those cells that have a similar rebar surface area, concrete cover, permeability and chloride content can be combined into a group and Project
C5 C6 C7 C8 C9 C10 2,4…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]
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
R2 0,6…1,2 [m] 0,2 64,0 45,0 260/300
0,2 60,3 46,0 180/300
0,4 59,1 43,0 330/300
0,4 29,9 0,0 220/300
0,4 52,5 43,0 190/300
0,3 54,7 43,0 270/300
R3 1,2…1,8 [m] 0,2 60,6 43,0 260,300
0,2 66,6 53,0 180/300
0,4 59,8 43,0 380/300
0,3 32,9 0,0 220/300
0,4 51,0 30,0 190/300
0,2 56,1 43,0 270/300
R4 1,8,…2,4 [m] 0,2 60,3 53,0 260/300
0,2 64,8 53,0 180/300
0,4 58,8 43,0 330/300
0,3 31,1 0,0 220/300
0,4 51,8 30,0 190/300
0,2 56,6 48,0 270/300
R5 2,4…3,0 [m] 0,2 65,5 55,0 260/300
0,3 63,8 56,0 180/300
0,4 59,6 43,0 330/300
0,3 51,5 30,0 220/300
0,4 51,8 30,0 190/300
0,3 58,0 43,0 270/300
R6 3,0…3,6 [m] 0,2 62,0 53,0 260/300
0,2 66,4 52,0 180/300
0,4 60,5 53,0 330/300
0,3 35,8 0,0 220/300
0,4 52,3 42,0 190/300
0,3 62,3 49,0 270/300
21.3 Appearance of the rebar layout being the cathode.
© 2007, Institute of Materials, Minerals and Mining
268
Corrosion of reinforcement in concrete
will be represented by the selected active cell/electrode. This electrode can then control the chloride extraction process of the group. According to previous experiments [4], chloride extraction can be designed to run with interuptions. A pause of some hours can be imposed which raises the efficiency and reduces the current consumption. The power supplies can 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 the monitoring files gives the exact positions for drilling the holes for the electrode fixings and reference cells. This is essential because the electrodes have to fit like a tile layout. After cross checking all cables and appliances, wetting of the electrodes and the concrete will start for a duration of about 3 days – depending on the permeability of the concrete and the concrete cover thickness. The progress of wetting can be monitored by reading the rest potentials, which fall when the 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 was completed in 1995. Although the abutments show no corrosion, a considerable amount of chloride (up to 2.5% by weight of cement) was detected in the concrete of the splash zone. Other details of the surveillance – especially rest potentials – have been reported [2]. The road crossing the highway below has a slight descent so that the abutments are shaped like a rhombus. Because the highest chloride concentrations are found close to the curb, the orthogonal electrodes have to follow this line, and the grid layout was rotated so as to be perpendicular 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 the bottom 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 inspections revealed that, despite the start of electrochemical changes in the rebar vicinity, no signs of active corrosion were apparent. [2]. Therefore, the first aim of chloride extraction was not to reduce the chloride concentration in the cathode area but to extract chloride from the outer concrete cover, documenting the whole process and verifying the equipment functionality while checking and verifying the data models given in [3] and in [5]. © 2007, Institute of Materials, Minerals and Mining
t) Cl (mass % cemen
t) Cl (mass % cemen
Chloride extraction from reinforced concrete
269
2.5
2.5 2 1.5 1
0.5 0 Dep 1 2 th fr 3 om surf ace 4 5 (cm )
C7R8 C7R7 C7R6 C7R5 Cells
2 1.5 1 0.5 0
Dep
C15R8 C15R7 1 2 C15R6 th fr 3 om 4 surf C15R5 Cells ace 5 (cm )
21.4 Initial chloride profiles of the eastern abutment. Table 21.1 Arrangement of the groups according to the structural situation of the abutment; numbers representing the positions of the active cells; grey cells left empty
0
7
8
4
5
1
9
6
2
3
21.4.2 Setup and results of the application According to the survey, the configuration data of the groups were as shown in Table 21.1. In principle, the even and the odd groups were switched alternately in a 12-hour-cycle. These 12-hour periods were reported to be much more effective for chloride removal than a constant voltage application [3]. For avoiding excessively negative cathodic polarisation, minimum reference potential values could be set to switch off the related group during the operating interval. In total, the installation was left on the concrete surface for 50 days. In order to maintain wetness, the electrodes were supplied with de-ionised water once a day by a ‘microdrip’ water hose system. By use of the control program CITec CeControl, the following data were permanently measured and saved according to specified differential values: groups (switch, voltage, current, reference potential), power supply (total current, voltage) and temperature. From the measured values, the charges were 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 considerable chloride content or corrosion activity, it was treated as well in order to obtain a concrete with higher alkalinity. © 2007, Institute of Materials, Minerals and Mining
270
Corrosion of reinforcement in concrete
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) Operation hours (cumulative) Ratio operation/installation
530:39:00 152:06:00 0.29
694:59:00 262:32:00 0.38
934:13:00 360:46:00 0.39
1199:26:00 537:49:00 0.45
Group 0 Operation hours (cumulative) Charge [A h m–2] (cumulative)
24:44:00 13.20
82:15:00 27.50
137:49:00 39.72
231:00:00 68.46
Group 1 Operation hours (cumulative) Charge [A h m–2] (cumulative)
80:16:00 26.95
133:11:00 39.46
175:51:00 48.63
235:46:00 59.97
Group 2 Operation hours (cumulative) Charge [A h m–2] (cumulative)
71:50:00 26.01
129:21:00 40.66
184:55:00 54.80
278:06:00 73.60
Group 3 Operation hours (cumulative) Charge [A h m–2] (cumulative)
65:40:00 30.64
118:35:00 49.74
161:15:00 66.05
245:07:00 90.90
Reference potential, not I-R corrected (mV vs MnO2)
0 –1000 –2000 –3000 –4000 –5000 Group 8 Group 9 Group 7
–6000 –7000 14.9
19.9
24.9
29.9 Date
4.10
9.10
14.10
21.5 Development of reference potentials – not I-R-corrected – of groups 7 to 9 during times of switching on and off.
The reference potentials obtained during the chloride extraction process were very negative during the switch-on phases of the groups. Since no I-Rcorrection could be made these values cannot be interpreted. However, during the switch-off phases, the values increased by some 100 mV and give an orientation about the polarisation stage of the related rebars. Figure 21.5 shows how the potential of the groups 7 to 9 returned to the range of –500 to –600 mV vs. MnO2 during the off-times from active potentials between –1500 to –7000 mV. Furthermore, the effects of process interruptions due to hardware tests can be seen. © 2007, Institute of Materials, Minerals and Mining
Chloride extraction from reinforced concrete
271
The measured current densities (Fig. 21.6) remained quite low compared with the high voltage of 40 V. This might be due to the vertical layout where moistening cannot be conducted as effectively as in a horizontal layout, and the CEM III concrete has, for a given concrete cover, a higher electrolytic resistivity. The analysis of the chloride content showed a very reasonable removal efficiency. In Fig. 21.7 the chloride content in rows 7 and 8 is shown. The samples were taken from the grid cells without specific reference to the
Current density (A m–2)
1.00 max max max max
0.80
G0 G1 G2 G3
avg avg avg avg
G0 G1 G2 G3
0.60
0.40
0.20 0.00 16.9
21.9
26.9
1.10
6.10
11.10
16.10
21.10
Date
21.6 Development of the maximum and average current densities – related to the rebar surface of the groups 0 to 3. 2.00
1.60 1.40 1.20 1.00 0.80 0.60 0.40 1 2 3 4 5
Start Day 21 1 2 Day 28 3 4 Day 38 5 4 cm 5 cm Day 50 1 cm 2 cm 3 cm
0.20
Free chloride (mass % cement)
1.80
0.00 4 cm 5 cm 1 cm 2 cm 3 cm
21.7 Progress of chloride removal in zones located 0 to 60 cm from the bottom (right) and 60 to 120 cm from the bottom (left).
© 2007, Institute of Materials, Minerals and Mining
272
Corrosion of reinforcement in concrete
rebars and, with the permanent removal progress, an even chloride distribution within the cells can be assumed. The values shown are average results from two grid cells. No corrections have been made so that natural deviations are part of the graphic. After finishing the treatment, an average chloride content of 0.29% remained in the concrete. This corresponds to a total removal of ca. 70%. For the future, the corrosion-inducing chloride content should have been raised according to the development of hydroxyl ions on the rebar surface. On removing the electrodes from the concrete, no visible changes to the visual appearance of the concrete surface could be seen. Because of the trial stage of the project no 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 carried out. This showed (Fig. 21.8) very low potentials and obviously high remnant cathodic polarisation. For the next measurement, four months later, a much more positive result and less value deviation was found (Fig. 21.9). Another important finding comes from the humidity measurements. Table 21.3 compares surface resistance values and shows that, during the time between finishing the CE and February 2002, no change in humidity could be measured. This means that the change of potentials is simply due to the process of losing polarisation. Furthermore, it shows the evidence of improved humidity content during the CE treatment. Chloride sampling from 13/02/2002 (Table 21.4) shows an interesting result. The outer 2 cm contains (in quite an even distribution) much more chloride due to the winter season. For the inner 3 cm the results are almost identical. The evaluation of the potentials from 13/02/2002 showed no indication of corrosion activity, and the raised chloride content in the outer zones does not appear to influence the rebars.
Rows
0 1 2 3 4 5 6 7 8 0
2
4
6
8
10
12 14 Columns
16
18
20
–600.0 –400.0 –300.0 –200.0 –100.0
22 0
24
26
100.0
21.8 Result of the potential mapping from 02/11/2001 (mV vs. SCE).
© 2007, Institute of Materials, Minerals and Mining
Chloride extraction from reinforced concrete
273
0 1
Row
2 3 4 5 6 7 8 0
2
4
6
8
10 –600.0
12 14 Columns
16
18
20
–400.0 –300.0 –200.0 –100.0
22 0
24
26
100.0
21.9 Rest potentials four months after finishing the CE (13/02/2002) (mV vs. SCE). Table 21.3 Statistical evaluation of the surface resistivity in rows 5 to 8 (0 to 2.4 m from bottom) Surface resistance [kW m]
Before start of CE 21/06/01
After finishing CE 02/11/01
Four months later 13/02/02
Average Standard deviation Minimum Maximum
14.4 10.2 2.0 73.0
4.1 1.6 1.0 10.0
3.7 1.2 2.0 6.0
21.6
Conclusions
Electrochemical chloride extraction has been demonstrated successfully on the substructure of the highway bridge at exit 34 on the A4 highway. Despite some testing-related interruptions, within 7 weeks ca. 70% of the chlorides was removed at quite a low power consumption. The technical concept allows a free and mostly appropriate electrode configuration, does not interfere with traffic and provides detailed documentation. The combination electrodes do not cause any loss or change in the appearance of the concrete surface. During the follow-up survey it could be ensured that the corrosion activity was removed with the chloride, but the need for an appropriate protection for the treated surface becomes clear as well, if the cause of the damage is still present. Although the new chloride extraction system requires much effort in time of preparation and in connection with the electrode material that has to be produced in advance, it is able deliver more detailed and more uniform results at a higher efficiency than usual, because: ∑ the CE system is open and scalable and can be fitted under any circumstances to the structure;
© 2007, Institute of Materials, Minerals and Mining
274
C7 1 cm
2 cm
C15
3 cm
4 cm
5 cm
1 cm
2 cm
3 cm
4 cm
5 cm
R7
content change
0.91 0.59
0.74 0.36
0.23 0.04
0.18 –0.02
0.14 –0.08
0.85 0.57
0.77 0.39
0.19 –0.06
0.20 0.04
0.16 0.00
R8
content change
0.98 0.43
0.46 0.10
0.20 –0.03
0.18 0.00
0.18 –0.03
0.84 0.34
0.48 0.13
0.19 –0.05
0.40 0.06
0.36 0.02
© 2007, Institute of Materials, Minerals and Mining
Corrosion of reinforcement in concrete
Table 21.4 Free chloride content [wt-% cement] at 13/02/2002 and changes compared with 17/10/2001
Chloride extraction from reinforced concrete
275
∑ the success of treatment can be monitored automatically by means of different measurements (e.g. amount of charge, development of offpotentials); ∑ inhomogeneous concrete zones will be considered automatically and be treated in an optimal way. They have to be known with their coordinates, and the process limitations will decline or accelerate the treatment (e.g. at re-profiled zones with a different permeability); ∑ all important basic physical data of a structure can be taken into account automatically, allowing the chloride extraction process to be related to the true rebar surface area; ∑ with the logging protocol there is a very detailed report, containing all sensor signals, actor actions and calculations e.g. of charge vs. time, to be generated for filing and as a process description; ∑ the logged data provide a basis to verify and to expand mathematical models that can be improved with every new application; ∑ there is no change in the visual appearance, because the fibreglass has high chemical stability; and ∑ no waste will be produced, and the electrodes are entirely re-usable. For evaluating and implementing mathematical models, much more data have to be collected in order to obtain a reliable and statistically based database.
21.7
Acknowledgements
We gratefully acknowledge that this project has been carried out with the financial support of the Federal Ministry of Economics of the FRG within the FUTOUR program. Furthermore, we thank the Saxonian Highway Administration for its assistance in the reference application of the CE system.
21.8
References and further reading
1. U. Schneck, ‘Zu Mechanismen der Stahlkorrosion in Beton bei der elektrochemischen Entsalzung’, Diss. TU Dresden, Dresden (1994). 2. U. Schneck, T. Winkler and S. Mucke, ‘Integrated system for corrosion monitoring at reinforced concrete structures’, Proc. Eurocorr, 2001. 3. U. Schneck, ‘Investigations on the chloride transformation during the electrochemical chloride extraction process’, Mater. Corros., 2000, 51, 91–96. 4. U. Schneck, H. Grünzig and S. Mucke, ‘Pulse width modulation – investigations for raising 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 for Electrochemical Removal of Chloride from Concrete Structures’, Corrosion, 1998, 54(4), 323–332.
© 2007, Institute of Materials, Minerals and Mining
276
Corrosion of reinforcement in concrete
6. prEN 14038-1, ‘Electrochemical re-alkalisation and chloride extraction treatments for reinforced concrete – Part 1: Re-alkalisation’, DIN e.V., 2000. 7. D. Whitmore, SHRP Product 2033: Guideline For Performing Electrochemical Chloride Extraction To Concrete Structures, AASHTO, Washington DC, USA.
© 2007, Institute of Materials, Minerals and Mining
22 Microscopy study of the interface between concrete and the conductive coating used as an anode for cathodic protection R. B. P O L D E R and W. H. A. P E E L E N, TNO Building and 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
22.1
Introduction
Cathodic protection (CP) of reinforcing steel in concrete structures has become a well-established and widely used technique for stopping corrosion in cases where chloride contamination has caused corrosion and concrete damage. Practical cases have been described [1, 2], a European standard was published recently [3] and further technical and economic information was given in [4]. In particular, the introduction of conductive coatings has stimulated the application of CP to concrete structures, partly due to their lower price in comparison with activated titanium systems. However the shorter service life of conductive coatings may be a disadvantage. Usually, their service life is assumed to be equal to that of ‘normal’ coatings for concrete. In practical cases, a 10-year guarantee of absence of corrosion is given, suggesting a service life of slightly more than 10 years. Hard evidence to support or deny this is practically non-existent. Within the framework of the European Concerted Action COST 521 ‘Corrosion of steel in reinforced concrete structures’, project NL-1 ‘Cost effective cathodic protection’, a study was carried out into the degradation of a particular conductive coating system, with which good experience existed in Norway and The Netherlands.
22.2
Theoretical background
Cathodic protection involves current flow through the concrete from the anode to the steel (cathode). The dominant reaction at the steel surface is reduction of oxygen, suppressing corrosion and producing hydroxide. The main reaction at the anode is oxidation of hydroxide, producing oxygen. Hydroxide consumption is equivalent to acid production, which may dissolve calcium hydroxide and other alkaline components of the concrete (cement paste). The theoretical amount of acid formation can be calculated using Faraday’s law, from the amount of electrical charge (current ¥ time) that has passed. 277 © 2007, Institute of Materials, Minerals and Mining
278
Corrosion of reinforcement in concrete
For each mole (equivalent) of electrons flowing, a mole of hydroxide is produced (at the cathode) and consumed (at the anode), equivalent to a mole of acid being produced. If the current density is 1 mA m–2 (assuming anode surface equals concrete surface), 0.33 mole of hydroxide ions are consumed per m2 of anode/concrete interface in one year. If all these hydroxide ions are provided by solid calcium hydroxide in the hardened cement paste, this would correspond to 12 g m–2 y–1 of calcium hydroxide dissolved or about 24 g of cement paste dissolved per m2 per year. With a density of 2000 kg m–3 that is about 12 mm ‘dissolution depth’ per year for the theoretical maximum amount of solid material dissolved by acid production at 1 mA m–2. It should be realised that the effective amount of hydroxide consumed will be mitigated due to the current circulation: the current causes migration of hydroxide ions from the cathode to the anode. Ion transport in concrete due to current flow may be described using transport numbers, as for aqueous solutions. Assuming that sodium and hydroxide ions are the only mobile species, 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 and consequently 80% of the theoretical maximum acid production will be neutralised by hydroxide migrating from the cathode to the anode, as shown in 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 our understanding of electrical transport in concrete, this amount of hydroxide consumption and cement paste dissolution seems more realistic than the maximum amount of dissolution calculated above. A current density of 1 mA m–2 would produce an effective amount of dissolution of 2.4 mm in the coating/concrete interface per year. In the study reported here, UV/visible light microscopy and scanning electron microscopy were used to assess the amount of dissolution that had occurred in samples subjected to CP in the field. In principle, oxidation processes at the anode could also involve carbon particles in the conductive layer (producing carbon dioxide or monoxide) or chloride ions from the hardened cement paste (producing chlorine gas). At high rates, oxidation of carbon would increase the porosity of the conductive Cathode; hydroxide production = Q
Hydroxide 0.8 Q
Sodium 0.2 Q
2H2O + O2 + 4e– Æ 4 OH–
Total charge passed = Q 4 OH– Æ 2H2O + O2 + 4e–
Anode; hydroxide consumption = Q
22.1 Mass transport and electrode reactions due to current flow under CP.
© 2007, Institute of Materials, Minerals and Mining
Microscopy study of the concrete–coating interface
279
layer. An example of similar processes has been described previously; see reference [5]. However, due to subsequent reactions this process would produce the same amount of acid as oxidation of hydroxide. Oxidation of chloride ions would decrease the consumption of hydroxide. This reaction would, however, not be visible using microscopy, but at high rates it would produce the smell of chlorine gas. At the current densities relevant to this study, it is improbable 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: a Norwegian building and five Dutch apartment houses. Aassiden is a highrise building in Oslo, Norway, with cast in chlorides. It was provided with CP with a conductive coating (primer) of the type AHEAD in 1990–1991. The coating system was applied directly onto the concrete surface, which had been prepared by sand blasting. A non-conductive top coat was applied over the conductive coating. Two cores were taken in 1999 from this bridge and sent to TNO. Five apartment buildings in Groningen, in the northern part of The Netherlands, were provided with a CP system based on the same coating in the years 1993 to 1997, one building each year. Further details of these CP systems are given in [1]. For aesthetic reasons, the surface was grit blasted and subsequently levelled with a filling mortar (fine quartz sand and Portland cement) before applying the coating. A non-conductive top coat was applied over the conductive coating. In 1999, one core was taken from each of these buildings. Both types of CP system involved a current source which reverses the polarity of the current every minute for just one second. In all cases, the CP systems complied with the usual criterion of sufficient steel protection for atmospheric concrete structures (>100 mV depolarisation). The driving voltages were between 1.5 and 2.0 V. Accurate (local) current densities were not available, but average current densities were about 0.5 to 1 mA m–2 of concrete surface. Cores were taken from positions close to the primary anodes (silver wire mesh), where the current density may be supposed to 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 for microscopy by careful drying and impregnating 10 mm thick slices taken from the cores with epoxy resin containing fluorescent dye and polishing down to ‘thin sections’ of about 25 mm thickness. This allows examination using transmitted visible and ultraviolet light. The procedure is similar to ASTM Standard C856-88 ‘Standard Practice for Petrographic Examination © 2007, Institute of Materials, Minerals and Mining
280
Corrosion of reinforcement in concrete
of Hardened Concrete’. The magnification used was 100 to 400 times (field of 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 carbon for the necessary conduction. The magnification applied was 50 to 4000 times. The highest resolution obtained in the SEM experiments was 1 mm.
22.4
Results
Light microscopy examination showed various characteristics. The two-layer coating system had a total thickness of 0.2 to 0.3 mm. This is the upper dark band in Fig. 22.2, showing sample NO1 in visible light. The two layers can be 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 fluorescent yellow/green colour in the upper part. The bond between the two layers (conductive and non-conductive) was good. In both figures, some air bubbles are visible inside the conductive coating layer. The lower part in the figures is the concrete, consisting mainly of hardened cement paste. The features in the 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, showing aggregate (white particles), in cement paste (darkish), with a vertical crack; field of view is 1.4 mm ¥ 0.8 mm.
© 2007, Institute of Materials, Minerals and Mining
Microscopy study of the concrete–coating interface
281
22.3 The same interface of NO1 as in Fig. 22.2, fluorescent in ultraviolet light with aggregate particles (dark), conductive coating layer (middle, black) and non-conductive coating layer (upper, light); field of view 1.4 mm ¥ 0.8 mm.
feature passing from the conductive coating/cement paste interface along the middle sand grain is a microcrack. It measures about 50 mm at the widest point (lower part) and about 10 mm at the narrowest point (curved part touching the sand grain). The interface between the coating system and the concrete substrate was studied in detail. Looking again at Fig. 22.2 (sample NO1), it is clearly visible that the black coating layer tightly adheres to the dark brown cement paste, without any sign of a gap or dissolved zone between them. From comparison with the microcrack described above, it can be seen that any gap or dissolved zone that might be present has a width of less than 10 mm. Acid dissolution would have been even more visible as a zone of increased transmission in Fig. 22.3. The absence of any detectable lighter zone in the interface confirms the absence of such dissolution on the scale of observation of these figures. Similar results have been obtained from sample GL1. This is shown in Fig. 22.4 in visible light at lower magnification than the previous figures. In the lower part, a grey aggregate particle is visible; the middle zone is the fine filling mortar paste, the black band is the conductive coating plus top coat and 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 are visible. On the right hand side, an irregularly shaped light zone is present between the coating and the substrate. This is a region where adhesion is
© 2007, Institute of Materials, Minerals and Mining
282
Corrosion of reinforcement in concrete
22.4 Concrete/coating interface of sample GL1 in plane polarised light with large aggregate below (grey), filling mortar (darkish) with fine aggregate (white), spherical air bubble (left), conductive layer and top coat above (black), showing non-adhering parts (right); field of view 3 mm ¥ 2 mm.
poor. Any apparent deformation or cracking of the coating is absent. The gap does not contain solid remnants of paste dissolved by acid; the void is completely empty. It is concluded that this void has been present since the application of the coating. It appears that the filling mortar contains a high amount of air bubbles and shows high overall porosity, which in part is due to 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 between the coating and the concrete. In some Dutch samples (codes GL and GB) various levels of poor bonding were found, which appeared to have been present from the beginning. Air bubbles in the surface and in particular in the filling mortar prohibited intimate contact between the coating and the substrate over considerable parts of the interface. Signs of dissolution or chemical deterioration were absent. None of the samples showed deterioration in the contacting parts of the interface where the protection current would have passed, as is illustrated in Fig. 22.2, 22.3 and 22.4. Such dissolution would have been visible as narrow zones (filled with resin), appearing light in the fluorescent photos. Increased porosity of the conductive layer was not observed. Local carbonation occurred in the paste of all samples, which is recognised as calcium carbonate, showing a change of colour in the cement paste under crossed-polarised light. Local frost attack was found in two of the Dutch
© 2007, Institute of Materials, Minerals and Mining
Microscopy study of the concrete–coating interface
283
Table 22.1 Results of microscopy of conductive coating/concrete interfaces; NO1, NO2 samples from Norway, GL1-4, GB5 samples from The Netherlands Sample code
Condition of coating
Coating–concrete bond
Description and general remarks
NO1 1991 Intact, no evidence of damage
Excellent; no evidence of de-bonding
Coating is continuous over the entire surface; coating–concrete bond is very good; no visual damage to coating
NO2 1991 Intact; identical to NO1
Very good
Coating is continuous and bond is very good; no damage to coating
GL1 1993
Intact, no evidence of damage
Moderate
Coating is continuous over the entire surface but locally bond is absent
GL2 1994
Intact, no evidence of damage
Moderate to poor
Identical to GL1
GL3 1995
Local tearing (due to sample removal or preparation)
Moderate to poor
Coating is not continuous; there is clear evidence of tearing and local absence of bonding over the entire surface
GL4 1996
Intact, no evidence of damage
Moderate to poor
Identical to GL1 and GL2
GB5 1999
Intact, no evidence of damage
Reasonably good
Coating is continuous; coating– concrete bond is quite good; no visual damage to coating
samples, shown by some microcracking well below the interface, which is obviously not related to the passing of protection current. It may be assumed that 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 hardened cement paste (lower part). Although the difference may be difficult to see in this photo, both phases can be identified readily at lower magnifications (see also Fig. 22.6). Even at this high magnification (the bar indicates 10 mm), no clear gap is visible between the paste and the coating. If there are local fissures between paste and coating (upper arrow), they are less than 1 mm wide. They are not wider than what seem to be microcracks (lower arrow) in the paste itself. SEM photos of GL1 are shown in Fig. 22.6 and 22.7. Figure 22.6 shows the 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 (dark grey, middle band) and the mortar (lower part, light grey) is clear. Figure 22.7 confirms this at a tenfold higher magnification. © 2007, Institute of Materials, Minerals and Mining
284
Corrosion of reinforcement in concrete
Microcrack Interfacial fissure? in paste
Conductive coating
Cement paste
22.5 Scanning electron microscopy photo of sample NO1.
Top coat Conductive coating
Filling mortar paste
22.6 Scanning electron microscopy photo of sample GL1 at moderate magnification (roughly same area as Fig. 22.4)
None of the samples studied by SEM showed evidence of chemical dissolution in the interface in the range of magnifications applied, giving resolutions 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 the interface and the bulk, with no obvious gradients. © 2007, Institute of Materials, Minerals and Mining
Microscopy study of the concrete–coating interface
285
Conductive coating
Cement paste
Acc.V Spot Magn Det WD 50 mm 10.0 kV 3.0 500x SE 11.4 GL1 boundary 01060a04
22.7 Scanning electron microscopy photo of sample GL1 at higher magnification.
22.5
Discussion
In Table 22.2, the period for which structures that had samples taken from them received CP and the amount of charge passed are given expressed as mole of hydrogen ions per square metre of concrete surface that have been formed. Because the samples were taken near the current distributors, it is reasonable to assume that the samples are representative of the structures in terms of current density. The following columns in the table give the maximum and the effective dissolved thickness resulting from the theoretical calculations given above. Finally, the observed material loss using light microscopy and scanning electron microscopy (for two samples) are given. It follows that the investigated samples NO1, NO2 (9 years) and GL1 (6 years), GL2 (5 years) and GL3 (4 years) have certainly experienced more than the amount of charge that would have caused 10 mm dissolution. This is the poorest resolution of the light microscopy examination technique. The two samples investigated using scanning electron microscopy show that even less than 1 mm of dissolution is present. The results show that: ∑ the amount of dissolution is very much less than the theoretically calculated maximum; ∑ the amount of dissolution is less than the amount calculated, considering the mitigation by transport of hydroxide by ionic migration; ∑ there may be another mechanism, other than migration, that mitigates the acid dissolution of cement paste at the anode/concrete interface; and
© 2007, Institute of Materials, Minerals and Mining
286
Corrosion of reinforcement in concrete
Table 22.2 Duration of cathodic protection (CP), amount of electrical charge passed (as moles of hydroxide consumed) and calculated amount of dissolution for seven samples of conductive coating/concrete interfaces assuming a protection current density of 1 mA m–2 (concrete interface) and observed amount of dissolution by light microscopy and scanning electron microscopy (SEM) Sample Duration Charge of CP passed (year) (mole of hydroxide)
Calculated maximum dissolved (mm)
Calculated effective dissolved mm)
Dissolved as observed by light microscopy (mm)
Dissolved as observed by SEM (mm)
NO1, NO2 GL1 GL2 GL3 GL4 GB5
108
22
<10
<1
14 12 10 7 2.4
<10 <10 <10 <10 <10
<1 – – – –
9
3
6 5 4 3 ca. 1
2 1.7 1.3 1 0.33
72 60 48 36 12
– not determined
∑ speculating, this could be diffusion of NaOH to the interface from elsewhere in the paste. The absence of acid dissolution suggests that degradation in these systems is substantially less than calculated from the amount of current circulation and the expected anode reactions. Deterioration due to acid dissolution has not occurred on a detectable scale. It appears that the service life is not determined by acid production at the coating/concrete interface. The service life of these conductive coating CP systems may be much longer than the oldest 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 light microscopy and scanning electron microscopy. Chemical dissolution in the coating/concrete interface was absent down to the level of the resolution of both 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 occurred in the samples is relatively large (and would certainly have been visible). It is obvious that such an amount of acid attack has not occurred. Hydroxide ion migration due to current flow would reduce the total amount of acid production. Considering the transport numbers of hydroxide and other ions, the ‘effective’ amount of acid production was calculated. However, such a level of dissolution was again not observed in samples having received CP
© 2007, Institute of Materials, Minerals and Mining
Microscopy study of the concrete–coating interface
287
current for more than four years. It seems probable that other mechanisms reduce the amount of acid attack on the cement paste under the current densities studied (1 mA m–2 of concrete surface). With regard to the service life of conductive coating CP systems, the results suggest that service lives of well over ten years may be obtained. In the samples studied, no degradation of the interface was present after six to nine years, although they originated from structures fully exposed to NorthWest European field conditions, including frost and snowfall. The bond between the coating and the concrete was relatively poor in the samples from The Netherlands, due to non-optimal application of the filling mortar used to even out the surface of the structures for aesthetic reasons. Despite that, no bond deterioration had occurred on these structures and the CP system functioned properly. It appears that the service life of this type of conductive coating CP system is well above the ten-year period that is normally guaranteed by the supplier or the contractor.
22.7
Acknowledgements
Mr Jan Eri of Protector, Drammen (Norway) is gratefully acknowledged for providing the samples, information on Aassiden and his permission to publish the information. The contribution of the members of COST 521 Working Group C-2 ‘Electrochemical Maintenance Methods’ in the discussion of preliminary results is thankfully acknowledged.
22.8
References
1. R. B. Polder, 1998, ‘Cathodic protection of reinforced concrete structures in The Netherlands – 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 Publication number 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 in Switzerland’, 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 Publication number 25, The Institute of Materials, 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, European Commission, 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 concrete structures – results from 15 years’ experience’, Mater. Perf., December, 2001, 22–26.
© 2007, Institute of Materials, Minerals and Mining
23 Protection of reinforced concrete piles in marine structures with sacrificial anodes L. B E R T O L I N I, M. G A S T A L D I, M. P. P E D E F E R R I and E. R E D A E L L I, Politecnico di Milano, Italy
23.1
Introduction
In reinforced concrete structures exposed to sea-water, pitting corrosion usually occurs in the emerged part (tidal and splash zones), where wetting and drying cycles favour the presence of both oxygen and chlorides. Corrosion hardly ever initiates in the reinforced concrete parts that are permanently immersed in seawater, where the relative lack of oxygen leads to very negative values of potential, and the chloride threshold for the onset of pitting corrosion is high.1,2 Furthermore, even when corrosion initiates, the corrosion rate is negligible owing to the small amount of oxygen that can reach the steel surface. 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 to the high resistivity of concrete and to the complexity of the reinforcement geometry, current is usually applied by means of a distributed anode placed on the surface of the concrete (e.g. an activated titanium mesh embedded in a cementitious overlay) and a current feeder. If the structure is buried or immersed in seawater, external sacrificial anodes can also be used. There have been studies of the use of local sacrificial anodes to provide cathodic protection to corroding steel in the non-submerged part of marine piles.5–7 These have shown that, unless the concrete resistivity is very low, the protection provided by submerged anodes is of limited effectiveness above the waterline, so that above-surface extended anodes are required. For several reasons, sacrificial anodes can be expected to be more effective when applied to new structures in order to achieve cathodic prevention. This technique relies on a cathodic current applied to the passive reinforcement in chloride-free concrete and it is aimed at delaying the initiation of pitting corrosion. It leads to an increase in the critical chloride content, since it lowers 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 field experiences8–10 have shown that even impressed current densities lower than 288 © 2007, Institute of Materials, Minerals and Mining
Protection of reinforced concrete piles in marine structures
289
2 mA m–2 can maintain steel at potential values where pitting corrosion cannot initiate when the chloride content exceeds 3% by weight of cement. Such current densities would be insufficient for the cathodic protection of already corroding steel, which requires current densities on the order of 10–20 mA m–2. Cathodic prevention also differs from cathodic protection with regard to its throwing power.11–12 It has been shown that, in spite of the high resistivity of concrete, the beneficial effects of cathodic prevention can extend to reinforcing bars at significant distances from the anode, owing to the higher cathodic polarisability of passive steel compared with that of corroding steel. Conversely, the effects of cathodic protection are usually limited to distances of a few tens of centimetres from the bars nearest to the anode.12 The enhanced throwing power of cathodic prevention suggests that the above-mentioned limitations of submerged sacrificial anodes applied for cathodic protection of 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 submerged sacrificial anodes to hinder corrosion initiation in the non-submerged part of marine piles and discusses the results of laboratory research aimed at studying the height above sea level to which protection and prevention can be achieved. Only a summary of the main findings is presented here; a more detailed account 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 in diameter were embedded in the concrete at 8 cm intervals. In order to simulate operative conditions of cathodic prevention, a specimen was made with chloride-free concrete; another one was made with concrete contaminated with 3% chloride by weight of cement, to promote corrosion of the embedded steel and, thus, reproduce conditions of cathodic protection. The steel bars were electrically connected outside the concrete; shunt resistances were used to measure the current circulating in each bar. A reference electrode made of a thin wire of mixed metal oxide (MMO) activated titanium was 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 and 1900 kg m–3 of limestone aggregate. Chlorides were added as CaCl2 to the mixing water. After curing, the columns were partially immersed in a shallow aqueous solution with 3.5% by weight of sodium chloride, so that only the steel bar at the bottom was submerged. Tests were carried out at room temperature, and the columns were regularly wetted with the test solution in
© 2007, Institute of Materials, Minerals and Mining
290
Corrosion of reinforcement in concrete 15 cm
120 cm
15 cm
16 cm
8 cm
Sacrificial Al-Zn-In anodes 3.5% NaCl (a)
(b)
23.1 Schematic representation of columns 15 ¥ 15 ¥ 120 cm for tests of cathodic prevention and cathodic protection with sacrificial anodes: (a) initial configuration with 15 connected bars and (b) configuration with eight connected bars.
order to simulate splashes of seawater. The 15 steel bars were connected to sacrificial anodes, made of a commercial Al–Zn–In alloy for cathodic protection, which were immersed in the test solution. After 16 months of testing, half of the bars were disconnected in both specimens to achieve a new configuration with only eight protected bars, with an interaxial spacing of 16 cm (Fig. 23.1b). Steel potential and current density were monitored in each bar. Depolarisation tests were regularly carried out by disconnecting the steel bars from the anodes for 24 h during which the potential of each bar was monitored to detect 4-h and 24-h decay. Normally rebars were not disconnected from each other during these tests; however, some tests were also carried out by individually disconnecting every bar. During the depolarisation tests, the potential of the activated titanium reference electrodes was calibrated against an external calomel reference electrode (SCE).
23.3
Results and discussion
Before the application of cathodic protection, the free corrosion potential of the steel bars was of the order of –200 to –300 mV vs SCE in chloride-free concrete and about –450 mV vs SCE in chloride-contaminated concrete. The
© 2007, Institute of Materials, Minerals and Mining
Protection of reinforced concrete piles in marine structures
291
potential of the anode was about –1050 mV vs SCE. Figures 23.2a and 23.2b show changes in time of the potential of the rebars, after connection to the sacrificial anodes. The cathodic polarisation of the steel bars following their connection to the anodes depended on the height above the level of the solution. In the specimen with passive steel in chloride-free concrete (Fig. 23.2a), the immersed bar reached a potential of about –1000 mV vs MMO (–1050 mV vs SCE), i.e. roughly the same potential as the sacrificial anodes. The second bar, 8 cm above the level of the solution, had a potential of around –800 mV vs MMO (–850 mV vs SCE). More positive values of potential were measured in the bars placed at greater heights. Nevertheless, even the bar at the top of the column (112 cm above the level of the solution) reached a potential value about 100 mV lower than the initial free corrosion potential. In the column with chloride-contaminated concrete (Fig. 23.2b), the electrical connection to sacrificial anodes led to a significant lowering of the potential only on the lowest four corroding bars. The immersed bar reached a potential value of about –1 V vs MMO, similar to the bar in the specimen without chlorides; 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 the specimen without chlorides. After 16 months of testing, half of the bars were disconnected in order to increase their spacing from 8 to 16 cm and reduce the steel surface to be protected. The reduction in the protected area led to a small decrease in the steel potential both in chloride-free concrete and in chloride-contaminated concrete. Figures 23.3a and 23.3b show the variation in time of the current density circulating in each bar. For a given height, the current density was much higher for the active bars in chloride-contaminated concrete compared to the passive 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 chloridecontaminated concrete, while only about 50 mA m–2 circulated in the bar in chloride-free concrete. Indeed, the total current fed by the anode was about 30–50 mA in the chloride-contaminated specimen and only 5–15 mA in the chloride-free concrete. The current which was received by bars at 8 and 16 cm 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 cm in both columns. Negative values of the current (i.e. anodic currents) were measured 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 tests were regularly carried out on the two specimens by disconnecting the steel bars from the sacrificial anodes for 24 h. Four-hour and 24-h decays were
© 2007, Institute of Materials, Minerals and Mining
292
Corrosion of reinforcement in concrete –100 15 bars connected
8 bars connected
Potential (mV vs MMO)
–200 Height: 72–112 cm
–300 –400
40–64 cm
–500 –600
16–32 cm
–700 8 cm
–800 –900
Immersed
–1000 –1100 0
–100
6
12 18 24 Time (months) (a)
15 bars connected
30
36
8 bars connected
Potential (mV vs MMO)
–200 Height: 56–112 cm
–300 –400
40–48 cm
–500 –600
32 cm
–700
16 cm
24 cm –800
8 cm
Immersed
–900 –1000 –1100 0
6
12 18 24 Time (months) (b)
30
36
23.2 Potential of steel bars at different heights above the level of the solution in the specimen with (a) chloride-free concrete and (b) 3% chloride-contaminated concrete, as a function of time.
then calculated as the differences between potentials measured after 4 h and 24 h, respectively, and the instant off potential value, measured within 1 s after the disconnection of the anodes. Figure 23.4a and 23.4b report the 4-h decay as a function of time, in chloride-free concrete and chloride-contaminated concrete, respectively. In each column, the four-hour decay decreased with height, showing very high values for the immersed bar (around 500 mV in
© 2007, Institute of Materials, Minerals and Mining
Protection of reinforced concrete piles in marine structures
293
1000
Current density (mA m–2)
15 bars connected 8 bars connected
100
Height: 16 cm
10
32 cm
1
48–112 cm
0.1
Current density (mA m–2)
1000
0
6
12 18 24 Time (months) (a)
30
36
15 bars connected 8 bars connected
100
Height: 16 cm 32 cm
10 48 cm 56–112 cm
1
0.1
0
6
12 18 24 Time (months) (b)
30
36
23.3 Current density circulating in steel bars at different heights above the level of the solution in the specimen with (a) chloride-free concrete and (b) 3% chloride-contaminated concrete, as a function of time. The black filled symbols (®) refer to the immersed bars.
chloride-free concrete and 300 mV in chloride-contaminated concrete). Depolarisation values were always higher for the rebars in chloride-free concrete if compared with the corresponding rebars in chloride-contaminated concrete. A decay of 100 mV can be considered sufficient to provide both protection to corroding steel as well as prevention of the onset of pitting corrosion on passive steel (cathodic prevention).1,4,14 According to this criterion, the result
© 2007, Institute of Materials, Minerals and Mining
294
Corrosion of reinforcement in concrete 700 15 bars connected
8 bars connected
Four-hour decay (mV)
600 Height:
500 400
Immersed
8 cm
300 16–40 cm 200 48–56 cm 100 64–112 cm 0 0
700
6
12
18 24 Time (months) (a)
15 bars connected
30
36
8 bars connected
Four-hour decay (mV)
600 500
Height:
400 Immersed
300
8–16 cm 200 32–40 cm 100
48–112 cm
0 0
6
12
18 24 Time (months) (b)
30
36
23.4 Four-hour decay of steel bars at different heights above the level of the solution in the specimen with (a) chloride-free concrete and (b) 3% chloride-contaminated concrete, as a function of time.
of Fig. 23.4 would indicate that protection was fulfilled up to a height of about 60 cm in the specimen with chloride-free concrete and about 40 cm in the specimen with chloride-contaminated concrete. The high values of potential decay (Fig. 23.4) measured on the immersed bar were not consistent with its low current density (Fig. 23.3). For this reason, the current circulating in each bar during the depolarisation tests was
© 2007, Institute of Materials, Minerals and Mining
Protection of reinforced concrete piles in marine structures
295
also monitored; current densities measured on the lowest four bars after the disconnection from the anodes are reported in Fig. 23.5, where cathodic currents are considered positive and anodic currents negative. These measurements showed that in both specimens, as soon as the sacrificial anodes were disconnected, the immersed bar, which was previously receiving a cathodic current, generated an anodic current. Therefore, the bars in the non-submerged part of the column continued to be cathodically polarised owing to the electrical connection with the immersed bar, even after disconnection from the anodes. As a consequence, the results obtained from depolarisation tests carried out by simply disconnecting the anodes were not reliable for the evaluation of the height at which protection conditions can be reached, since they were affected by the presence of this macrocouple. Further depolarisation tests were then carried out by avoiding the onset of macrocouples. Each single bar was disconnected from the others, so that no current could circulate. The actual potential decay was then measured. The 4-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 potential after 4 h and 24 h of depolarisation was observed in the rebars in the nonsubmerged part; hence their potential decay increased. Figures 23.6a and 23.6b show the profile of 4-h decay along the height of each column, respectively, in chloride-free concrete and chloride-contaminated concrete. Comparison between results obtained from the tests with connected and disconnected bars shows that the actual throwing power is higher than that 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 of oxygen in permanently saturated conditions (the instant off potential is lower than –1 V vs SCE). The trend of 4-h decay along the height of the column suggests that 4-h decay in passive steel maintains values higher than 100 mV even at heights greater than 1.2 m. On the contrary, sacrificial anodes did not provide sufficient protection to corroding steel reinforcement placed above 80 cm. The difference in the throwing power between the chloride-free specimens and the chloride-contaminated ones were also confirmed by 24-h depolarisation tests. In fact, profiles of 24-h decay shown in Fig. 23.7 do not differ much from those obtained after 4 h.
23.4
Conclusions
The effects of localised sacrificial anodes on passive and corroding rebars in concrete piles partially immersed in seawater were studied. Depolarisation tests carried out by simply disconnecting the anode from the rebars were not reliable for evaluating the actual throwing power of cathodic protection and prevention, owing to a macrocouple that was generated between bars at
© 2007, Institute of Materials, Minerals and Mining
296
Corrosion of reinforcement in concrete 50 Immersed 32 cm
Current density (mA m–2)
40
16 cm 48 cm
30 20 10 0 –10 –20 –30 –40 –50 –4
0
4
8 12 Time (h) (a)
16
20
24
200 Immersed 32 cm
Current density (mA m–2)
150
16 cm 48 cm
100 50 0 –50 –100 –150 –200 –4
0
4
8 12 Time (h) (b)
16
20
24
23.5 Current density on steel bars at different heights from the level of the solution measured during a depolarisation test with connected bars in the specimen with (a) chloride-free concrete and (b) 3% chloride-contaminated concrete, as a function of time.
different heights. The immersed bar, which had a very negative potential, supplied an anodic current that cathodically polarised the rebars at higher heights and, thus, altered their potential decay. When depolarisation tests were carried out by disconnecting each bar from the others, so that the macrocouple could not generate, an effective 4-h decay of about 150 mV was obtained at a height of more than 1 m if the
© 2007, Institute of Materials, Minerals and Mining
Protection of reinforced concrete piles in marine structures
297
120 Connected bars Disconnected bars
Height (cm)
100
80
60
40
20
0 0
100
200 300 400 500 Four-hour decay (mV) (a)
600
700
120 Connected bars Disconnected bars
Height (cm)
100
80
60
40
20
0 0
100
200 300 400 500 Four-hour decay (mV) (b)
600
700
23.6 Comparison between the profile of 4-h decay obtained with connected or disconnected bars in (a) chloride-free concrete and (b) chloride-contaminated concrete.
steel was passive, whereas, if the steel was already corroding, the decay at the same height was negligible. The measurements of the actual decay of rebars suggest that, while sacrificial anodes have poor effectiveness in protecting corroding bars that are located above sea level, they can provide cathodic polarisation to passive steel sufficient to avoid the onset of corrosion even at significant heights above sea level.
© 2007, Institute of Materials, Minerals and Mining
298
Corrosion of reinforcement in concrete 120 Connected bars Disconnected bars
Height (cm)
100 80 60
40
20 0 0
100
200 300 400 500 600 Twenty-four-hour decay (mV) (a)
700
120 Connected bars Disconnected bars
Height (cm)
100
80 60 40
20 0 0
100
200 300 400 500 600 Twenty-four-hour decay (mV) (b)
700
23.7 Comparison between the profile of 24-h decay obtained with connected 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 in concrete’, 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 Marine Environment, Svolvear, Norway, 28–30 May 1997, p. 123.
© 2007, Institute of Materials, Minerals and Mining
Protection of reinforced concrete piles in marine structures
299
4. EN 12696-1 Standard, ‘Cathodic protection of steel in atmospherically exposed concrete’, March 2000. 5. S C Kranc and A A Sagues, ‘Computation of reinforcing steel corrosion distribution in 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, ‘Performance of 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 experimental investigation 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 cathodic prevention 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 protection of steel in concrete and cathodic prevention’, European Community, COST 521 Workshop, Annecy, 21–24 September 1999. 10. L Bertolini, ‘Cathodic prevention’, Proc. European Workshop Corrosion of Steel in Reinforced Concrete Structures, T D Sloan and P A M Basheer (Eds.), Queens University, Belfast, 28–31 August 2000, p. 107. 11. L Bertolini, F Bolzoni, A Cigada, T Pastore and P Pedeferri, ‘Cathodic protection of new 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 in the 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 corrosion in 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 concrete in marine exposure zones’, J. Mater. Civil Eng., 2000, 12(2), 164.
© 2007, Institute of Materials, Minerals and Mining
24 Renovation of the cathodic protection system of a concrete bridge after 12 years of operation G. S C H U T E N and J. L E G G E D O O R, Leggedoor Concrete Repair, The Netherlands, and R. B. P O L D E R and W. H. A. P E E L E N, TNO Building and Construction Research, The Netherlands
24.1
History
The ‘Stadionviaduct’ bridge in Rotterdam, which was built in 1937, connects a 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 to normal motor vehicle traffic and bicycles, it is used by large numbers of pedestrians before and after football matches and other events. The bridge consists of a steel support framework with three separate concrete decks, one for the central driveway and two combined bicycle/ pedestrian paths along the sides. The driveway has an asphalt overlay, while the concrete of the bicycle/pedestrian paths is finished with a thin wearing course 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 to chloride penetration from de-icing salts was found on the two bicycle paths. At that time, the only option available was conventional concrete repair. This was carried out on the northern bicycle path in 1986. By the end of 1986, the southern path was provided with a cathodic protection system, a technique which had then just been introduced in the Netherlands. The installation was the first CP system for protection of a reinforced concrete structure in the Netherlands.
24.2
The CP installation on the southern bicycle path (1986)
The installation was designed to protect approximately 350 m2 of concrete in two zones of about 2.4 m ¥ 70 m surface area. One zone was situated east and one zone was situated west of the expansion joint in the middle of the bridge deck. Before applying CP, cracked and spalled concrete was removed, exposing the steel over about 50% of the surface. Additional reinforcement was provided where necessary and a new concrete cover layer was cast to provide a substrate for the anode system. The anode was a Ferex 100S cable, 300 © 2007, Institute of Materials, Minerals and Mining
Renovation of the cathodic protection system
301
consisting of a twisted copper wire with niobium coating, surrounded by a graphite-filled polymer mantle forming a cable of 8 mm diameter. The cables were 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 installed in this way. The anode cables were embedded in a cast layer of fine aggregate concrete made with fly ash cement. Two reference electrodes of the Ag/AgCl type were located near the two ends of each zone. The power supplies were designed 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 standards of that time. This included removing cracked and spalled parts, cleaning the steel by grit blasting and applying new concrete cover using shotcrete. The surface was finished with an epoxy wearing course. However, the repair was found to be ineffective in stopping corrosion in the long term. In 1996, advancing corrosion and concrete damage in the form of cracking and spalling had seriously compromised the safe operation of the northern bicycle path. It was decided to remove the complete deck and the supporting steelwork and 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 corrosion and no damage to the concrete. However the operation of the CP system raised some concerns. Since 1992, some of the reference electrodes were not indicating normal depolarisation behaviour. The current delivered to one of
24.1 Ferex 100S anode system before casting the concrete overlay (1986).
© 2007, Institute of Materials, Minerals and Mining
302
Corrosion of reinforcement in concrete
the zones had become increasingly erratic over the years. TNO was asked to asses the technical condition of the installation, to study the feasibility of renovating the CP system and, if possible, to identify the measures needed to do so. Electrical measurements were carried out and cores were taken for visual and microscopic analysis. This assessment showed that the problems were limited to improper functioning of the power supplies, poor electrical connections and poor functioning of the reference electrodes [1]. Most importantly, however, the anode system showed no visible degradation. A micrograph of the anode/concrete interface is shown in Fig. 24.2. No deterioration of the carbon/polymer anode surface or carbonation of the interface with the concrete, which might indicate oxidation of the graphite in the anode, was detected using ultraviolet and visible light. The conducting polymer appeared to be intact. The overlay showed some microcracking, probably due to drying shrinkage in the early stages; one such microcrack can be seen in Fig. 24.2. There was no evidence of deterioration of the overlay 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 least another ten years.
24.5
System upgrade 1999
Based on the assessment, a programme was outlined to renovate the CP system by providing new power supplies, new reference electrodes and multiple new connections to both the anode cables and the reinforcement. The contract
24.2 Micrograph of the Ferex 100S anode/concrete interface (KSA is the anode, M cement paste, A aggregate and Sc is a microcrack); field of view c. 2 mm ¥ 3 mm.
© 2007, Institute of Materials, Minerals and Mining
Renovation of the cathodic protection system
303
to carry out the work was commissioned to Leggedoor bv and TNO was asked to supervise it. Four slots of approximately 2.0 m ¥ 0.2 m were excavated in the concrete surrounding the anode cables at the ends of the bicycle paths for making the new connections. A schematic is given in Fig. 24.3 and a picture of the work being carried out is given in Fig. 24.4. The Ferex polymer mantle was removed locally from all 16 cables in the slots and connections were soldered to the copper wires, which were then insulated using heat shrinking plastic sleeves.
Expansion West
joint
East
Driveway Bicycle
CP Zone 1
CP Zone 2
Pedestrian
24.3 Layout of deck, CP system and locations of the four slots for making new connections.
24.4 Making new electrical connections to the Ferex 100S anode and reinforcement.
© 2007, Institute of Materials, Minerals and Mining
304
Corrosion of reinforcement in concrete
In the slots, additional reinforcement connections were made. One manganese dioxide reference electrode was installed in each of the slots, which were then backfilled using a polymer-modified cementitious repair mortar. Modern power supplies and sockets for testing were installed in the control cabinet. Existing cables in the concrete were reused for connection of the system with the new cabinet. Subsequently, the system was reactivated in December 1999. Before switching on the current, cable resistances in the anode and reinforcement circuits were checked and found to be satisfactory (<1 W). A voltage of about 2 V was applied, resulting in a current of about 0.5 A per zone. Short term polarisation was between 125 and 300 mV in the negative direction. The current 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 of the Dutch recommendation CUR 45 ‘Cathodic Protection of Reinforcement in Concrete Structures’ [3].
24.6
Cost aspects
Financial records of repair and maintenance actions in the past were not available. Consequently, the costs of repair and maintenance can only be estimated. The cost of installing the CP system on the southern bicycle path in 1986/87 is estimated at 150 000 7. The cost for routine control of the CP system between 1986 and 1998 is estimated at 15 000 7. The costs involved in the assessment and renovation of the CP system from 1998 to 2000 were about 55 000 7. Control of the system between 2000 and 2010 may cost 25 000 7. The total cost from 1987 to 2010 is about 250 000 7. The total cost of replacing the deck of the northern bicycle path was about 1.3 million 7. The cost of the repairs in 1986 is unknown. It clear that the total costs for replacing the deck exceed by far the costs for the original CP system, its maintenance and control, including the renovation of the system and ten more years of operation; according to the estimates, the difference is by about a factor of five.
24.7
Durability aspects
In the present installation, it was found that the anode system was durable and was able to provide sufficient corrosion protection to the reinforcement for about 12 years, despite malfunctioning of the electrical system after some time. Moreover, the anode was expected to be able to provide protection for at least another ten years. The durability of the anode system reported here 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
© 2007, Institute of Materials, Minerals and Mining
Renovation of the cathodic protection system
305
until 6 to 8 years of service. Detailed examination after 15 years showed that the carbon had dissolved from the outer layers of the anode cable and the polymer had become brittle. This caused high resistance build-up in the circuit and decreasing current density, until sufficient depolarisation could no longer be achieved. There is no obvious explanation for the difference between the two systems. Most probably the current density in the Berlin system was significantly higher compared with that in the Rotterdam ‘Stadionviaduct’ system. An observation is that the epoxy wearing course applied to the bicycle decks (in 1986/87) on its own does not appear to prevent corrosion of the reinforcement. This may be due to chloride left in the concrete when repairs were carried out. Possibly incipient anode effects have played a role. However, it may be questioned whether such an epoxy wearing course could effectively prevent the ingress of water and new chlorides. In a study of a concrete deck with a similar wearing course, it was concluded that degradation of the wearing 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 in Rotterdam based on the Ferex 100S Anode cable was upgraded successfully after 12 years of service to meet contemporary CP requirements at the relatively low cost of 55 000 7. This was mainly due to the good condition of the anode. Despite problematic (electrical) operation of the system during the second half of the 12-year period, reinforcement corrosion had not taken place to the extent that damage to the concrete occurred. Although repaired conventionally in 1986, the northern bicycle path had to be replaced in 1996 at the considerable cost of 1.3 million 7. Due to the CP 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 least 25 years (starting 1987) and for a much lower cost, which was estimated at 250 000 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 Ingenieursbureau Gemeentewerken (public works) Rotterdam for her permission to publish this information.
© 2007, Institute of Materials, Minerals and Mining
306
Corrosion of reinforcement in concrete
24.10 References 1. R. B. Polder, ‘Cathodic protection Stadionviaduct, Phase 1’, TNO Building and Construction Research Report 99-BT-MK-R0026, 1999 (in Dutch). 2. R. B. Polder and W.H.A. Peelen, ‘Cathodic protection Stadionviaduct, Phase 2, Revision of existing installation’, TNO Building and Construction Research Report 2000-BTMK-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 concrete structures – 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 concrete from a 30 year old bridge’, HERON, 45, (2), 109–124.
© 2007, Institute of Materials, Minerals and Mining
