Sulphur-Assisted Corrosion in Nuclear Disposal Systems (EUROPEAN FEDERATION OF CORROSION SERIES)

European Federation of Corrosion Publications NUMBER 59 EFC 59 Sulphur-assisted corrosion in nuclear disposal systems ...

Author: Frank Druyts | Bruno Kursten

243 downloads 712 Views 4MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

European Federation of Corrosion Publications NUMBER 59

EFC 59

Sulphur-assisted corrosion in nuclear disposal systems Edited by Damien Féron, Bruno Kursten & Frank Druyts

Published for the European Federation of Corrosion by Maney Publishing on behalf of The Institute of Materials, Minerals & Mining

Published by Maney Publishing on behalf of the European Federation of Corrosion and The Institute of Materials, Minerals & Mining Maney Publishing is the trading name of W.S. Maney & Son Ltd. Maney Publishing, Suite 1C, Joseph’s Well, Hanover Walk, Leeds LS3 1AB, UK First published 2011 by Maney Publishing © 2011, European Federation of Corrosion The author has asserted his 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 editors, authors and the publishers cannot assume responsibility for the validity of all materials. Neither the editors, 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 Maney Publishing. The consent of Maney Publishing 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 Maney Publishing 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. Maney Publishing ISBN-13: 978-1-907975-17-2 (book) Maney Publishing stock code: B815 ISSN 1354-5116 Cover: SEM (Scanning Electron Micrograph) picture taken from a carbon steel specimen that has been exposed to Boom clay at 170°C. The picture shows the typical pitting corrosion morphology, due to the action of thiosulphate present in the clay environment. Typeset and printed by the Charlesworth Group, Wakeﬁeld, UK.

Contents

Series introduction

vii

Volumes in the EFC series

ix

Editorial EFCN° 59 D. Féron, B. Kursten and F. Druyts

xv

Section 1 - Disposal concepts of nuclear waste and the role of corrosion 1 The Belgian Supercontainer concept – corrosion issues B. Kursten, F. Druyts and R. Gens Section 2 - Sulphur-induced corrosion processes from within and outside the nuclear disposal field 2 Sulphur chemistry of the near-ﬁeld Boom Clay environment O. Azizi and D. D. Macdonald 3

4

5

6

7

1

19

Corrosion mechanisms and material performance in environments containing hydrogen sulﬁde and elemental sulfur L. Smith and B. Craig

46

Lifetime prediction of metallic barriers in nuclear waste disposal systems: overview and open issues related to sulphur-assisted corrosion D. Féron

66

The anaerobic corrosion of carbon steel and the potential inﬂuence of sulphur species N. R. Smart

81

The inﬂuence of chloride on the corrosion of copper in aqueous sulphide solutions J. M. Smith, Z. Qin, F. King and D. W. Shoesmith

109

Interactions between sulphide species and components of rust Ph. Refait, J. A. Bourdoiseau, M. Jeannin, R. Sabot, C. Rémazeilles and J. A. Bourdoiseau

124

Section 3 - Role of microbial processes in sulphur-assisted corrosion 8 Experimental investigation of the impact of microbial activity on the corrosion resistance of candidate container materials V. Madina, I. Azkarate, L. Sánchez and M. Á. Cuñado

137 v

vi

Contents

Section 4 - Modelling of corrosion 9 Reactive-transport modelling of the sulphide-assisted corrosion of copper nuclear waste canisters F. King, M. Kolar and M. Vähänen

152

Section 5 - Panel discussion 10 Sulphur-related issues in deep underground nuclear waste disposal systems P. De Cannière, B. Kursten, F. Druyts, H. Moors and R. Gens

165

Index

171

European Federation of Corrosion (EFC) publications: Series introduction

The European Federation of Corrosion (EFC), incorporated in Belgium, was founded in 1955 with the purpose of promoting European cooperation in the ﬁelds of research into corrosion and corrosion prevention. Membership of the EFC 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, cathodic protection, education, reinforcement in concrete, microbial effects, hot gases and combustion products, environment-sensitive fracture, marine environments, reﬁneries, surface science, physico-chemical methods of measurement, the nuclear industry, the automotive industry, the water industry, coatings, polymer materials, tribo-corrosion, archaeological objects, and the oil and gas industry. Working Parties and Task Forces 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 Working Parties are coordinated, 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 Fédération Française pour les sciences de la Chimie (formely Société de Chimie Industrielle) in France, and The Institute of Materials, Minerals and Mining in the UK. 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, e.g. reviews or results of experimental work, may be published in scientiﬁc and technical journals in one or more countries in Europe. Conference 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 (IoM) 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 Maney Publishing on behalf of The Institute of Materials, Minerals and Mining. vii

viii

Series introduction

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 Mr M. Roche Fédération Européenne de la Corrosion, Fédération Française pour les sciences de la Chimie, 28 rue Saint-Dominique, F-75007 Paris, France Dr W. Meier Europäische Föderation Korrosion, DECHEMA e.V., Theodor-Heuss-Allee 25, D-60486 Frankfurt-am-Main, Germany

Volumes in the EFC series * indicates volume out of print

1

Corrosion in the nuclear industry Prepared by Working Party 4 on Nuclear Corrosion*

2

Practical corrosion principles Prepared by Working Party 7 on Corrosion Education*

3

General guidelines for corrosion testing of materials for marine applications Prepared by Working Party 9 on Marine Corrosion*

4

Guidelines on electrochemical corrosion measurements Prepared by Working Party 8 on Physico-Chemical Methods of Corrosion Testing

5

Illustrated case histories of marine corrosion Prepared by Working Party 9 on Marine Corrosion

6

Corrosion education manual Prepared by Working Party 7 on Corrosion Education

7

Corrosion problems related to nuclear waste disposal Prepared by Working Party 4 on Nuclear Corrosion

8

Microbial corrosion Prepared by Working Party 10 on Microbial Corrosion*

9

Microbiological degradation of materials and methods of protection Prepared by Working Party 10 on Microbial Corrosion

10

Marine corrosion of stainless steels: chlorination and microbial effects Prepared by Working Party 9 on Marine Corrosion

11

Corrosion inhibitors Prepared by the Working Party on Inhibitors*

12

Modiﬁcations of passive ﬁlms Prepared by Working Party 6 on Surface Science*

13

Predicting CO2 corrosion in the oil and gas industry Prepared by Working Party 13 on Corrosion in Oil and Gas Production*

14

Guidelines for methods of testing and research in high temperature corrosion Prepared by Working Party 3 on Corrosion by Hot Gases and Combustion Products ix

x 15

Volumes in the EFC series Microbial corrosion: Proceedings of the 3rd International EFC Workshop Prepared by Working Party 10 on Microbial Corrosion

16 Guidelines on materials requirements for carbon and low alloy steels for H2Scontaining environments in oil and gas production (3rd Edition) Prepared by Working Party 13 on Corrosion in Oil and Gas Production 17 Corrosion resistant alloys for oil and gas production: guidance on general requirements and test methods for H2S service (2nd Edition) Prepared by Working Party 13 on Corrosion in Oil and Gas Production 18

Stainless steel in concrete: state of the art report Prepared by Working Party 11 on Corrosion of Steel in Concrete

19

Sea water corrosion of stainless steels: mechanisms and experiences Prepared by Working Party 9 on Marine Corrosion and Working Party 10 on 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 EFC Working Party 10 on Microbial Corrosion

23

CO2 corrosion control in oil and gas production: design considerations Prepared by Working Party 13 on Corrosion in Oil and Gas Production

24 Electrochemical rehabilitation methods for reinforced concrete structures: a state of the art report Prepared by Working Party 11 on Corrosion of Steel in Concrete 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 Papers from the 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 Working Party 9 on Marine Corrosion

Volumes in the EFC series

xi

31 Corrosion of reinforcement in concrete: corrosion mechanisms and corrosion protection Papers from EUROCORR ‘99 and Working Party 11 on Corrosion of Steel 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 Working Party 13 on Corrosion in Oil and Gas Production

33 Marine corrosion of stainless steels: testing, selection, experience, protection and monitoring Edited by D. Féron on behalf of Working Party 9 on Marine Corrosion 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 Steel in Concrete

36

Prediction of long term corrosion behaviour in nuclear waste systems Edited by D. Féron on behalf of Working Party 4 on Nuclear Corrosion

37

Test methods for assessing the susceptibility of prestressing steels to hydrogen induced stress corrosion cracking By B. Isecke on behalf of Working Party 11 on Corrosion of Steel 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 on behalf of Working Party 11 on Corrosion of Steel in Concrete

39

The use of corrosion inhibitors in oil and gas production Edited by J. W. Palmer, W. Hedges and J. L. Dawson on behalf of Working Party 13 on Corrosion in Oil and Gas Production

40

Control of corrosion in cooling waters Edited by J. D. Harston and F. Ropital on behalf of Working Party 15 on Corrosion in the Reﬁnery Industry

41

Metal dusting, carburisation and nitridation Edited by H. Grabke and M. Schütze on behalf of Working Party 3 on Corrosion by Hot Gases and Combustion Products

42

Corrosion in reﬁneries Edited by J. D. Harston and F. Ropital on behalf of Working Party 15 on Corrosion in the Reﬁnery Industry

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

xii

Volumes in the EFC series

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. Klein and P. Marcus on behalf of Working Party 6 on Surface Science

45

Local probe techniques for corrosion research Edited by R. Oltra on behalf of Working Party 8 on Physico-Chemical Methods of Corrosion Testing

46

Amine unit corrosion survey Edited by J. D. Harston and F. Ropital on behalf of Working Party 15 on Corrosion in the Reﬁnery Industry

47

Novel approaches to the improvement of high temperature corrosion resistance Edited by M. Schütze and W. Quadakkers on behalf of Working Party 3 on Corrosion by Hot Gases and Combustion Products

48 Corrosion of metallic heritage artefacts: investigation, conservation and prediction of long term behaviour Edited by P. Dillmann, G. Béranger, P. Piccardo and H. Matthiesen on behalf of Working Party 4 on Nuclear Corrosion 49 Electrochemistry in light water reactors: reference electrodes, measurement, corrosion and tribocorrosion Edited by R.-W. Bosch, D. Féron and J.-P. Celis on behalf of Working Party 4 on Nuclear Corrosion 50

Corrosion behaviour and protection of copper and aluminium alloys in seawater Edited by D. Féron on behalf of Working Party 9 on Marine Corrosion

51

Corrosion issues in light water reactors: stress corrosion cracking Edited by D. Féron and J-M. Olive on behalf of Working Party 4 on Nuclear Corrosion

52

Progress in Corrosion – The ﬁrst 50 years of the EFC Edited by P. McIntyre and J. Vogelsang on behalf of the EFC Science and Technology Advisory Committee

53

Standardisation of thermal cycling exposure testing Edited by M. Schütze and M. Malessa on behalf of Working Party 3 on Corrosion by Hot Gases and Combustion Products

54

Innovative pre-treatment techniques to prevent corrosion of metallic surfaces Edited by L. Fedrizzi, H. Terryn and A. Simões on behalf of Working Party 14 on Coatings

55

Corrosion-under-insulation (CUI) guidelines Prepared by S. Winnik on behalf of Working Party 13 on Corrosion in Oil and Gas Production and Working Party 15 on Corrosion in the Reﬁnery Industry

56

Corrosion monitoring in nuclear systems Edited by S. Ritter and A. Molander

Volumes in the EFC series

xiii

57

Protective systems for high temperature applications: from theory to industrial implementation Edited by M. Schütze

58

Self-healing properties of new surface treatments Edited by L. Fedrizzi, W. Fürbeth and F. Montemor

60

Methodology of crevice corrosion testing for stainless steels in natural and treated seawaters Edited by U. Kivisäkk, B. Espelid and D. Féron

61

Inter-laboratory study on electrochemical methods for the characterisation of CoCrWo biomedical alloys in simulated body ﬂuids Edited by A. Igual Munoz and S. Mischler All volumes are available from Maney Publishing or its North American distributor. See http://maney.co.uk/index.php/series/efc_series/

Editorial EFCN° 59

In the different disposal concepts for high-level nuclear waste, corrosion of the metallic barriers and, in particular, the overpack/container is a major issue. It is imperative for performance assessment to predict the lifetime of these containers. In the lifetime prediction of metallic barriers for the disposal of high-level nuclear waste (HLW) or of spent fuel, the presence of (reduced) sulphur species is an issue of growing importance, as the sulphur species are involved in localised corrosion phenomena. The international workshop on ‘Sulphur-Assisted Corrosion in Nuclear Waste Disposal Systems’ was held on 21–23 October 2008 in Brussels (Belgium) at the Hotel Metropole. The SACNUC2008 workshop aimed to provide an exchange of information on the inﬂuence of sulphur species on the corrosion of metallic barriers and to gather scientists and engineers to • • •

present the state-of-the-art on sulphur-assisted corrosion receive input from outside the nuclear waste ﬁeld discuss the importance of sulphur-assisted corrosion for the respective disposal concepts.

The workshop consisted of a series of lectures focusing on diverse aspects of sulphur-assisted corrosion. In the introductory session, an overview is given of European concepts for the disposal of high-level nuclear waste, with a particular focus on the Belgian Supercontainer concept. During these opening lectures, the role of corrosion knowledge in the preparation of safety cases, and the possible impact of sulphur-related corrosion are discussed. Evidence of sulphur-related corrosion phenomena is presented, both from within and outside the nuclear disposal ﬁeld. This is complemented by discussion of the fundamental aspects of sulphur-induced corrosion. To conclude the introductory session, the role of sulphur-assisted corrosion in lifetime prediction is discussed. A central session in the workshop and in the book is devoted to discussing the inﬂuence of sulphur species on the corrosion of several materials that may be present in disposal concepts, i.e. carbon steel, copper, and components of rust. The role of microbial processes in sulphur-assisted corrosion is also addressed. An overview is given of microbial processes that may occur in deep repository environments and how they may affect corrosion. This is followed by presentations of investigations of microbially inﬂuenced corrosion of both container materials and concrete. Two papers discuss the modelling of corrosion in disposal conditions and the inﬂuence of sulphur species in particular. To conclude the workshop, a discussion panel tried to identify open issues in the investigation of sulphur-assisted corrosion phenomena and how to incorporate these in robust lifetime predictions of metallic barriers. xv

xvi

Sulphur-assisted corrosion in nuclear disposal systems

This volume includes these ﬁve sections with nine chapters in total, comprising discussions at a high technical and scientiﬁc level which occurred during the 3-day international workshop. The International Workshop on Sulphur-Assisted Corrosion in Nuclear Waste Disposal Systems was co-organised by SCK•CEN and ONDRAF/NIRAS under the auspices of the European Federation of Corrosion (EFC event N°311). The editors and the chairman of the WP4 ‘Nuclear Corrosion’ would like to thank the authors who presented and wrote papers of outstanding scientiﬁc and technical content and who responded enthusiastically to the questions and comments raised by the reviewers. They would also like to thank the members of the Working Party 4 ‘Nuclear Corrosion’ (EFC WP4) of the European Federation of Corrosion who reviewed and commented on these papers. They hope that the readers will enjoy the papers, and that this book will be a useful tool for scientists and engineers to enhance the understanding of the corrosion phenomena that we have to face in the development of safe nuclear waste storage systems. Bruno Kursten,

Frank Druyts

Damien Féron

Project Leader Corrosion Studies R&D Waste Packages Unit SCK•CEN Guest Editor of this volume

Head R&D Waste Packages Unit SCK•CEN Guest Editor of this volume

Chairman of the EFC WP4 ‘Nuclear Corrosion’

1 The Belgian Supercontainer concept – corrosion issues B. Kursten and F. Druyts SCK•CEN, The Belgian Nuclear Research Centre (SCK•CEN), R&D Waste Packages Unit, Boeretang 200, B-2400 Mol, Belgium

R. Gens ONDRAF/NIRAS, The Belgian Agency for Radioactive Waste and Enriched Fissile Materials (NIRAS/ONDRAF), Avenue des Arts 14, B-1210 Brussels, Belgium

1.1

Introduction

Among the options considered for dealing with long-lived radioactive waste (vitrified high-level waste – VHLW – and spent fuel – SF), geological disposal (after a period of storage on the surface to allow gradual decay of both the temperature and radioactivity of the wastes) is the one most widely recommended at the international level and one which is scientifically and technically feasible for maximising the protection of people and their environment, for both the current and future generations. The Belgian radioactive waste management organisation, NIRAS/ONDRAF, is committed to the challenge of developing a concept and design of a disposal facility, and to developing the evidence and arguments to prove that such a facility can be constructed in a safe, technically feasible and economically achievable manner, without neglecting the societal aspects. In Belgium, the Supercontainer (SC) is currently being studied as the reference design for the final disposal of VHLW and SF. The SC Design was developed based on the Contained Environment Concept (CEC), the aim of which is to establish and preserve a favourable chemical environment in the immediate vicinity of the metallic overpack, so that it will be exposed to essentially unchanged, benign conditions for a long time, at least for the duration of the thermal phase. The thermal phase is defined as the timeframe during which the temperature of the host formation is expected to lie above the range of temperatures within which nominal migration properties can be relied upon. The SC is a cylindrical container (L ≈ 4 m; Ø ≈ 2 m) made of a 6 mm thick stainless steel casing (the envelope). This casing comprises a 30 mm thick carbon steel overpack, containing either two VHLW canisters or four SF assemblies, surrounded by a thick concrete buffer (≈ 700 mm thick). Carbon steel has been chosen for the overpack because it is a material for which a broad experience and knowledge of its physico-chemical properties already exists and in particular its corrosion behaviour in a concrete environment is well understood and favourable to meeting the required overpack longevity. A Portland cement-based (PC) concrete has been chosen for the buffer because it will provide a highly alkaline chemical environment, in which the external surface of the overpack will be passivated and it is expected only to be prone to uniform corrosion (passive dissolution). A schematic diagram of a SC for VHLW 1

2

Sulphur-assisted corrosion in nuclear disposal systems

1.1 Schematic diagram of (a) the cross-section and (b) the longitudinal section of a Supercontainer for VHLW with indication of the different components [1]

emplaced within disposal galleries excavated in the Boom Clay is presented in Fig. 1.1 [1]. Methodical R&D studies have focused on the Boom Clay, which is located in the northeast of Belgium at a depth between 180 and 280 m, and have been ongoing for more than 30 years. The scientific and technical advancement has been reported on a regular basis, with a periodicity of approximately 10 years. This has resulted in the publication of several safety and environmental assessment reports, such as

The Belgian Supercontainer concept – corrosion issues

3

SAFIR-1 [2] and SAFIR-2 [3], which were published in 1989 and 2001, respectively. These reports were subjected to international reviews [4] whose outcome confirmed the suitability of Boom Clay as a host rock for geological disposal. To date, however, the current reference disposal option in Belgium lacks formal political and societal acceptance. Neither an institutional nor a political decision exists confirming geological disposal for the long-term management of VHLW and SF. Therefore, NIRAS/ ONDRAF has developed a stepwise process to validate the key decisions. A first step consists of submitting a Waste Plan to the national authorities by 2010. In the Waste Plan, the various options for the long-term management of the radioactive waste will be compared and the Plan will serve as a basis for discussions on a political, social and economic level. The Waste Plan aims at achieving a decision in-principle confirming the acceptability of geological disposal. Under the assumption that geological disposal is the preferred reference solution in Belgium, which needs to be agreed through adoption of the Waste Plan, the next scientific and technological milestone will be the development and submission of the first Safety and Feasibility Case, the SFC1, by the end of 2013 [4,5]. The Safety and Feasibility Case is an integration of scientific, technological and regulatory arguments and evidence that describe, substantiate and, if possible, quantify the safety and feasibility of, and the level of confidence in, the proposed long-term management solution for VHLW and SF (i.e. geological disposal) at any given stage of development. The SFC consists of a series of documents supporting the statements that the proposed disposal system provides long-term safety if implemented according to design specifications and that the proposed repository can be constructed, operated and closed according to these specifications. The purpose of the SFC1 is to illustrate the safety and feasibility of a geological repository in clay, with scientifically wellfounded arguments, with the ultimate goal of obtaining a decision to proceed with site selection. It will also discuss the significance of any remaining uncertainty or open issues in the context of the decision at hand in the process of repository development and provides guidance for work to resolve these issues in future development stages. To ensure the long-term safety of the disposal concept during the successive phases of the disposal period, NIRAS/ONDRAF has defined several safety functions that need to be fulfilled, such as: • • •

containment during the thermal phase, the ‘C’ function. This is accomplished by the Engineered Barrier System (EBS) delay and attenuation of the releases during the repository lifetime, the ‘R’ function. This is accomplished by the host formation (i.e. the Boom Clay layer) isolation, the ‘I’ function. This is accomplished by the geological environment (i.e. the host formation together with its overlying sedimentary formations).

The corrosion studies are focused on the engineered barrier system, whose integrity has to be ensured at least for the duration of the thermal phase, which according to NIRAS/ONDRAF’s specifications, is assumed to last for several hundred years for VHLW and, possibly up to a few thousand years for SF, after emplacement of the wastes in the repository. This paper describes the state of current knowledge of the corrosion behaviour of the carbon steel overpack together with the remaining uncertainties and future key issues from a corrosion point of view.

4

Sulphur-assisted corrosion in nuclear disposal systems

1.2

Integrated approach

In view of the Safety and Feasibility Case I study (SFCI), an integrated R&D methodology was developed to demonstrate and defend the concept that the integrity of the carbon steel overpack can be ensured at least during the thermal phase. This methodology has already been explained in detail elsewhere [6]. The main aspects are briefly summarised below. The environmental conditions surrounding the SC will change with time because: • • •

the oxidising conditions will gradually change to reducing conditions following repository closure the temperature will decrease as heat production from the radioactive waste decreases the geochemistry of the environment surrounding the carbon steel overpack, which is initially governed by the PC-based buffer chemistry, will gradually be modified as Boom Clay porewater penetrates the concrete lining of the disposal galleries and the concrete buffer material of the SC.

As a result of the changing environmental conditions that the overpack will be subjected to during its lifetime, the disposal period can be divided into different phases. A key factor of the R&D strategy consists of determining the ‘best estimate’ of the uniform corrosion rate for each of these separate phases. Predicting accurate uniform corrosion rate values (or ranges) forms the basis of the scientific approach because, under the predicted conditions within the Supercontainer (i.e. a highly alkaline concrete buffer), the carbon steel overpack is expected to undergo uniform corrosion through the mechanism of passive dissolution. However, when proper operational conditions are met, metals whose corrosion resistance depends on maintaining a ‘stable’ passive film, such as plain carbon and low-alloy steels in high pH environments, can show signs of an increased susceptibility to localised forms of corrosion if the protective film is locally disrupted. This could result in locally very high corrosion rates that could lead to very rapid penetration of the carbon steel overpack (i.e. well before the end of the thermal phase). This situation needs to be avoided at all times because of its detrimental nature on the ‘containment function’ of the metallic overpack. Therefore, an additional requirement to increase the confidence in the developed integrated approach is to prove the validity of the so-called ‘exclusion principle’, in which it has to be demonstrated that each corrosion mechanism (e.g. pitting corrosion, crevice corrosion, stress corrosion cracking), other than uniform corrosion, cannot take place under the circumstances described in the evolutionary path. This is accomplished by proving that the predicted concentrations of aggressive species that can be expected within the SC are situated well below the threshold concentrations (above which local breakdown of the passive film will occur). This will be achieved through a limited set of specific laboratory tests and expert judgment. 1.3 1.3.1

Current status Chemical evolution

Figure 1.2 schematically illustrates, from right to left, the time evolution of the nearfield environment surrounding a disposal gallery. Excavation of the disposal galleries

The Belgian Supercontainer concept – corrosion issues

5

1.2 Schematic of the time evolution of the near-field around a disposal gallery [8]

will lead to fracturing of the Boom Clay in a 1 m zone around the galleries. Oxygen, coming from the atmosphere in the repository, will cause some oxidation of the initially anoxic Boom Clay. In particular, pyrite and organic matter, two important compounds in Boom Clay, will interact with this oxygen. Oxidation of pyrite leads to the production of higher concentrations of dissolved sulphate (SO42−) and thiosulphate (S2O32−) in the porewaters within, and close to, the fractures and excavations. Because of the plastic nature of the Boom Clay, the excavation-induced fractures are known to seal within a relatively short period. Water will continuously drain towards the gallery, but oxygen will dissolve in the porewater and continue to diffuse into the host formation. Neglecting the reactivity of the oxygen with the remaining pyrite and organic matter, the in-diffusion will not exceed a depth of about 2 m, even after 20 years of ventilation [7]. Consequently, the extent of the oxidised zone will remain limited to the first few metres, while the degree of oxidation of the host formation will increase. During the early closure phase, the heat-emitting waste will cause a temperature increase, lasting for at least several hundreds to thousands of years. Meanwhile, reactions leading to an equilibrium between the high pH concrete and the surrounding host rock will commence (alkaline plume). As this is a very slow process, these reactions will mainly continue throughout the late closure phase. The extent of this alkaline plume within the Boom Clay is also limited to about 2.5 m after 100 000 years [8]. The initial pH of the concrete pore fluid of about 13.5, will be controlled by the dissolution of the alkali metal hydroxides (K+ and Na+), and decrease to a value of 12.5, where it will be regulated by Portlandite solubility, after about 1000 years (these calculations are very conservative because they do not take into account the porosity decrease in time due to carbonation). The pH value of 12.5 is predicted to be constant for at least 80 000 years, after which it will slowly start to drop. The temperature increase (to ~80°C) during the thermal phase of the repository operation will reduce the pH to about 12, owing to the effect of temperature on the hydrolysis properties of the system [9]. Bouniol [10] conducted radiolysis simulations examining the evolution of radiolysis species within the Supercontainer assuming:

6

Sulphur-assisted corrosion in nuclear disposal systems

1.3 Evolution of the concentration of oxygen-derivative species at the interface between the concrete buffer and the steel overpack taking into account a gamma dose rate of 25 Gy/h and a temperature varying from 90 to 16°C [10]

• • •

an initial gamma irradiation dose rate of 25 Gy/h at the overpack surface a variable temperature (90°C → 16°C) resulting from the thermal evolution of the Supercontainer a closed unsaturated system.

These calculations seem to predict that the oxygen concentration at the overpack/ concrete buffer interface will remain fairly constant (3.5×10−4 mol/L) over a 300 year period (see Fig. 1.3). As a consequence, radiolysis could prolong the duration of the aerobic phase within the Supercontainer, which in turn could lead to an accelerated loss of integrity of the overpack due to localised corrosion. However, no attempt has yet been made to couple these radiolysis calculations with corrosion reactions, which would be expected to consume radiolytically produced oxygen. 1.3.2

Mechanical evolution

The main gas generation mechanisms within the repository near-field are likely to be anaerobic corrosion of the carbon steel overpack and radiolysis of the liquid phase within the concrete buffer. The solubility of a gas and its diffusivity in the buffer pore solution determine how quickly any gas generated can be removed from solution. Weetjens et al. [11] analysed gas transport within the concrete buffer, focusing on the production of hydrogen. The hydrogen production rate due to anaerobic corrosion of the carbon steel overpack was estimated assuming a corrosion rate of 1 μm/year for the first 100 years, followed by 0.1 μm/year thereafter. The highest calculated pressure was 3.4 MPa at a point close to the overpack after ~100 years and at this time, the gas occupied 24% of the porosity. This pressure may be compared with the lithostatic pressure in the Boom Clay at the expected repository depth (~4.5 MPa), and the expected tensile strength of the concrete buffer (~2 MPa).

The Belgian Supercontainer concept – corrosion issues

7

1.4 Evolution of the partial pressures within the porosity of the concrete in the Supercontainer at variable temperature [10]

Bouniol [10] studied the influence of irradiation (dose rate of 25 Gy/h) on the total pressure developed within the Supercontainer (see Fig. 1.4). This study indicated that the calculated maximum total pressure (0.25 MPa) will remain low enough not to exert an influence on the mechanical properties of the concrete buffer. 1.3.3

Thermo-hydraulic evolution

Weetjens et al. [11] performed coupled thermo-hydraulic simulations to investigate the effect of elevated temperature on the saturation of the concrete buffer. Calculations were performed for a 1D radial geometry using the PORFLOW code and assuming an initial saturation state of 70% for all cementitious EBS materials. The calculations suggest that, in the absence of the stainless steel envelope, saturation of the concrete buffer, right up to the overpack surface, would be complete within a few years after tunnel closure. Poyet [12,13] carried out numerical simulations to investigate the thermohydraulic behaviour of the concrete buffer for a closed system with a sealed envelope. A model describing the coupled transfers of heat and water in an incompressible porous medium, and implemented in the CEA Finite Elements code CAST3M, was used. Two sets of calculations were performed corresponding to two different initial degrees of saturation of the concrete buffer: (i) curing at ambient temperature (preventing the concrete from drying) and (ii) preliminary partial drying at 60°C (leading to a reduced moisture content). The results of the calculations show that heating should not adversely affect its properties. The maximum temperature level reached is too low (less than 100°C): •

to induce high pressures (the maximum values are small, about 0.060 MPa and 0.016 MPa for the two calculation cases, because the low temperature levels do not lead to significant vaporisation)

8 • •

Sulphur-assisted corrosion in nuclear disposal systems to create massive dehydration (the amount of water released varied from 0.4 to 4.5 kg/m3 depending on the location, i.e. close to or far from the waste source) or to cause complete desaturation (the saturation rapidly increases during the heating phase, followed by a slower decrease leading to homogenisation of the saturation within the buffer).

Another important output, from a corrosion point of view, is that heating is not expected to generate a dry zone near the overpack. 1.3.4 Corrosion issues General corrosion The oxygen initially present in the repository will be consumed by various processes such as corrosion of the engineered barrier (carbon steel overpack), microbial activity and reaction with minerals. Consequently, the environmental conditions surrounding the SC will eventually become anoxic. The corrosion rate of steel under such conditions is expected to be very low due to the formation of a protective passive film in high pH media. Although the timescale for the repository to become oxygenfree is, at present, still somewhat uncertain, evidence indicates that the overpack will be exposed to anoxic conditions for most of its service life. These long timescales make accurate predictions of the anaerobic uniform corrosion rates (or ranges) imperative to guarantee, because of: • •

the long-term integrity of the overpack (corrosion could ultimately lead to penetration of the waste package containment, i.e. the carbon steel overpack) the long-term stability of the disposal system (e.g. possible pressure build-up due to hydrogen gas generated during the anaerobic corrosion of the carbon steel overpack could disrupt the SC).

Kursten [14] has reviewed uniform corrosion rate data relevant to carbon steel and mild steel in highly alkaline environments. Various approaches have been used to derive uniform corrosion rates, all of which are based on either one of the following three principles: • •

•

the measurement of the weight loss of test coupons electrochemical measurement techniques. The electrochemical methods found in the literature include linear polarisation resistance (LPR), Tafel slope extrapolation, the galvanostatic pulse technique, electrochemical impedance spectroscopy (EIS), and passive current density measurements (potentiodynamically or potentiostatically at steady-state) the measurement of hydrogen gas evolution. The quantification of hydrogen gas evolution due to the anaerobic corrosion of steel is determined in two ways, viz. using a manometric gas cell technique or a gas chromatograph/mass spectrometer. In the manometric gas cell technique, the uniform corrosion rate is correlated with the volume of gas generated, which is measured through the displacement of a column of liquid due to a pressure increase in the test vessel. In the latter technique, the gas evolving from the corrosion processes is passed through a gas chromatograph or a mass spectrometer.

The Belgian Supercontainer concept – corrosion issues

9

The most reliable anaerobic uniform corrosion rate values are believed to be those generated by either hydrogen gas evolution measurements over long periods or by passive current density measurements at steady-state. Almost all anaerobic uniform corrosion rata data reported in the literature today, originate from studies performed in the scope of the national nuclear waste management programmes in the UK (Nirex1 Safety Assessment Research Programme, NSARP), Switzerland (Nagra’s crystalline rock programme), or Japan (feasibility studies for the disposal of lowlevel radioactive waste, RWMC). The few published studies of corrosion rates under anaerobic conditions for non-nuclear industrial applications were conducted under conditions (e.g. temperature, pH, partial aeration) that are not relevant for the highly alkaline concrete buffer environment within the Supercontainer. Table 1.1 shows a compilation of anaerobic uniform corrosion rate data determined from hydrogen gas evolution measurements carried out in saturated Ca(OH)2, diluted alkali hydroxides (KOH, NaOH), and artificial cement pore solutions representative of potential repositories in the UK [15–20], Switzerland [21–25] and Japan [26–29]. Smart and co-workers [15–20], measured the hydrogen evolution rate for mild steel for a wide range of conditions. Table 1.1 summarises only those results that are most relevant for the Supercontainer conditions. Wire specimens were used. The test solutions were put in Teflon or zirconia crucibles. Long-term anaerobic corrosion rate values below 0.01 μm/year were measured after 10 years of testing. In the Swiss programme [21–25], experiments were conducted with pure iron wire. Anaerobic Table 1.1 Compilation of uniform corrosion rate data (generated by hydrogen gas evolution measurements) for mild steel in anoxic, alkaline solutions (in the absence of aggressive anions) Environment

pH

T (°C)

vCORR (μm/year) NSARP (UK)

Ca(OH)2 sat.

Alkali hydroxides (KOH, NaOH)

Artificial cement pore solution

1

12.8 12.8 12.8 12.8 12.5–13.0 13.5 14.0 12.8 13.0 13.0 13.0 12.8–13.0 12.5–13.0 12.5–13.0 13.5 14.0

50 21 15 30 35 35 35 45 30 50 80 21 21 35 35 35

NAGRA (Swiss)

RWMC (Japan)

0.003 0.007–0.01 0.004 0.02 0.05–0.06 0.03 0.5 0.2 0.010 0.011 0.002 0.0035–0.01 0.0035–0.03

Nirex is the former designation of NDA (Nuclear Decommissioning Authority)

0.005–0.01 0.005–0.1 0.01–0.2

10

Sulphur-assisted corrosion in nuclear disposal systems

corrosion rates as low as 0.0035 μm/year have been reported after 2 years of testing. Japanese workers [26–29] recorded long-term anaerobic corrosion rates of the order of 0.004–0.2 μm/year at temperatures up to 45°C. The corrosion rate increased with increasing temperature. The corrosion rate was also found to increase at pH 14, due to the formation of soluble HFeO2−. It is interesting to note that the experiments were conducted in glass cells, which could have caused an inhibitive effect from the dissolved glass (due to the high pH of the test solutions) [17,30]. In the UK and Swiss experimental studies, the specimens were pickled in 10% HCl to remove the airformed film before testing, whereas in the Japanese studies un-pickled samples were tested. Pickling causes an initial peak in the corrosion rate, which then deceases with time to a low value as a layer of corrosion products, predominantly magnetite, builds up. For un-pickled specimens, an incubation period was observed before gas generation started. The presence of an existing corrosion product layer delays the onset of gas production. Table 1.2 shows a compilation of anaerobic uniform corrosion rate data determined from passive current density measurements carried out in saturated Ca(OH)2 and saturated Ca(OH)2+NaOH solutions representative for the Supercontainer buffer environment [31–34]. The corrosion rates determined from passive current density measurements were higher than those measured by the hydrogen gas evolution measurements. The passive current density measurements showed the tendency of the uniform corrosion rate to increase with increasing temperature. In this respect, however, it has to be mentioned that hydrogen gas evolution experiments, performed by Smart and co-workers [15–20], seemed to indicate that the uniform corrosion rates tend to converge towards the same very low value in the long term, regardless of the temperature. A general trend of uniform corrosion rate decreasing with increasing exposure time has been observed by many researchers [18,19,21,23–25,35,36] when carbon steel has been exposed to alkaline media representative of the environment surrounding the carbon steel overpack within the Supercontainer. The surface of the carbon steel overpack will be subjected to a gamma radiation field that could initially be as high as 25 Gy/h and is known to decrease with time as a result of radioactive decay. An experimental study [37] to investigate the possible effects of radiation on the anaerobic corrosion of the carbon steel overpack is ongoing. Hydrogen gas evolution experiments are carried out in an artificial cement pore

Table 1.2 Compilation of uniform corrosion rate data (generated by passive current density measurements under steady-state) for carbon steel in anoxic, alkaline solutions (in the absence of aggressive anions) Environment

pH

T (°C)

vCORR (μm/year)

Ca(OH)2 sat.

12.4

Ca(OH)2 sat. + NaOH

13–13.5

22 40 60 23 40 60 80

0.23 0.33 0.80 0.137 0.426 0.735 1.170

The Belgian Supercontainer concept – corrosion issues

11

solution (136 g/L 1 M Na OH + 370 g/L 1 M KOH + 0.284 g/L Na2SO4) that simulates the cementitious buffer material selected for use in the Supercontainer. To take account of the possible release of gas due to radiolytic breakdown of the cell materials or due to radiolysis of the test solution, the recorded data under irradiation are adjusted against a control cell (control cells are identical to the test cells, but do not contain any steel wires or solution). The results from the gas cell measurements are shown in Fig. 1.5. No significant difference in the rate of gas production in the cells irradiated at 25 Gy/h was found compared to the unirradiated cells. Figure 1.6 presents the results from the electrochemical measurements monitoring the corrosion potential of steel in artificial cement pore solution under unirradiated and irradiated (25 Gy/h) conditions at 80°C. This figure shows that for both unirradiated and irradiated conditions, the corrosion potential of steel rapidly falls to values close to the hydrogen evolution potential at pH 13.4 (~–790 mVSHE). In irradiated conditions, however, a sudden large rise (~700 mV) of the potential was observed after ~200 h. A sound explanation for this positive change in potential is not given. It is not known yet whether this potential rise is due to an effect of radiation on the oxidising behaviour of the solution (e.g. formation of H2O2) or on the stability of the reference electrode. Stress corrosion cracking The Slow Strain Rate Testing (SSRT) technique was used to provide a first screening of the Stress Corrosion Cracking (SCC) susceptibility of plain carbon steel (i.e. without welds) in artificial cement pore solutions [38]. This technique involves pulling a specimen to failure in uniaxial tension at a constant controlled slow strain rate while the test specimen is exposed to the test environment. The outcome of an SSRT experiment is often represented in a so-called stress–strain curve, which is a graphical representation of the relationship between the stress (plotted in the ordinate) and the strain (plotted in the abscissa). The stress is derived from measuring the load applied on the sample (= load/surface area) and the strain is derived from measuring the deformation of the sample, i.e. the elongation (= Lf – L0/L0, where L0 is the original gauge length and Lf is the gauge length after fracture). The SSRT experiments were performed under two different potential regimes, viz. (i) at the free corrosion potential and (ii) under potentiostatic control at potentials situated in the transitional potential region between passive and pitting behaviour (the applied potentials were determined from cyclic potentiodynamic polarisation curves). The recorded stress– strain curves were compared to a reference, which was determined for the same alloy pulled to failure in air at 80°C. After the test, the specimens were removed from the facility and the failure mode was determined by scanning electron microscopy (SEM). It was found that the ductility of carbon steel was independent of the applied potential (within the potential range investigated so far: +415 and +515 mVSHE). The investigated potentials were determined from cyclic potentiodynamic polarisation curves. The potentials +415 and +515 mVSHE were located in the potential region between passive and pitting behaviour. All specimens, including those tested under freely corroding conditions, failed by a completely ductile fracture at elongations of about 46% to 47.5% and the fracture surface showed identical features of a ductile fracture compared to the reference sample tested in air, viz. a large necking region (macroscopic observation) and the presence of many microvoids and dimples (microscopic examination), as can be seen from Fig. 1.7.

12 Sulphur-assisted corrosion in nuclear disposal systems

1.5 Comparison of the anaerobic corrosion rate (from hydrogen gas evolution measurements) for carbon steel in artificial cement pore solution, under unirradiated and irradiated conditions (25 Gy/h) [37]

The Belgian Supercontainer concept – corrosion issues

13

1.6 Electrochemical potential measurements in artificial cement pore solution at 80°C without chlorides under (a) unirradiated conditions and (b) irradiated conditions (25 Gy/h) [37]

1.4

Conclusions

The Belgian radioactive waste management organisation, NIRAS/ONDRAF, is committed to the challenge of developing a concept and design for a disposal facility, and developing the evidence and arguments to prove that such a facility can be constructed in a safe, technically feasible and economically achievable manner, without neglecting the societal aspects. In this respect, the next scientific and technological milestone will be the development and submission of the first Safety and Feasibility Case, the SFC1, by the end of 2013. The Supercontainer (SC) is currently being studied as the reference design for the final disposal of vitrified high-level waste

14

Sulphur-assisted corrosion in nuclear disposal systems

1.7 SEM fractograph images of carbon steel in artificial cement pore solution containing 100 mg/L Cl–, 2560 mg/L SO42–, 60 mg/L S2O32– and 500 mg/L S2– at 80°C after SSRT recorded at a strain rate of 1×10–6 s–1 and various applied potentials (top images: fracture surface, magnification: 50×; bottom images: fracture detail, magnification: 500×) [38]

The Belgian Supercontainer concept – corrosion issues

15

(VHLW) and spent fuel (SF) in deep underground clay layers. The SC comprises a carbon steel overpack, containing two VHLW canisters or four SF assemblies, surrounded by a Portland cement-based (PC) buffer, which, in turn, is entirely encased in a stainless steel envelope. An integrated R&D strategy has been developed to demonstrate and defend the view that the integrity of the carbon steel overpack can be ensured, at least during the thermal phase. The environmental conditions that the overpack will be subjected to during its lifetime will change during the disposal period (e.g. from oxic to anoxic conditions), which makes it possible to divide the disposal period into different phases. A key factor of the R&D strategy consists of determining the ‘best estimate’ of the uniform corrosion rate of each of these separate phases because, under the predicted conditions within the Supercontainer (i.e. a highly alkaline concrete buffer), the carbon steel overpack is expected to undergo uniform corrosion. However, it has also to be demonstrated that each corrosion mechanism, other than uniform corrosion, cannot take place under the circumstances described in the evolutionary path (i.e. the ‘exclusion principle’). Scoping calculations based on a local equilibrium-diffusion transport model indicate that the near-field will probably remain alkaline (pH>12.5) for a timescale of 100 000 years. Radiolysis simulations suggest that the oxygen concentration will remain fairly constant (3.5×10−4 mol/L) over a 300 year period, which could have a significant influence on the duration of the aerobic phase. However, the current calculations do not take into account the role of the corrosion reactions on the oxygen level. Gas generation calculations indicate that hydrogen production due to gamma radiolysis of water is not expected to pose a threat to the integrity of a sealed stainless steel envelope (i.e. no pressure build-up is predicted). Numerical simulations investigating the thermo-hydraulic behaviour of the concrete buffer show that the consequences of heating are expected to be small and not deleterious for the service of the Supercontainer. The maximum temperature level reached is too low (less than 100°C) to induce high pressures, massive dehydration or complete desaturation. From the data, it is clear that the long-term anaerobic uniform corrosion rate of carbon steel under disposal conditions (high pH cementitious environment) will reach a very low constant value of less than 0.1 μm/year. It is also observed that the corrosion rate may even fall as low as 0.0035 μm/year. It was found that radiation at a dose rate of 25 Gy/h has a negligible effect on gas generation rates compared to the unirradiated situation. The ductility of carbon steel was independent of the applied potential in artificial porewaters (within the potential range investigated so far: +415 and +515 mVSHE). All specimens failed by a completely ductile fracture at elongations of about 46% to 47.5%. Future research efforts with the aim of eliminating some of the remaining uncertainties, with respect to the overpack’s integrity under conditions relevant for the Supercontainer (i.e. highly alkaline), will be focused on: •

Estimating more accurately the anaerobic corrosion rate of carbon steel in concrete (including the influence of irradiation on the corrosion rate). Anaerobic corrosion of iron generates hydrogen gas and therefore contributes significantly

16

•

•

• •

Sulphur-assisted corrosion in nuclear disposal systems to the gas source term. Gas transport behaviour within the repository near-field environment constitutes a major aspect of guaranteeing the safety of the repository system (however, gas phase formation within the EBS does not necessarily pose a problem if the absence of gas induced contaminated water transport can be guaranteed). Investigating the SCC behaviour of the carbon steel overpack. This includes (1) performing SSRT tests under potentiostatic control at potentials situated in the potential region wherein lies the transition between active and passive behaviour (determined from cyclic potentiodynamic polarisation curves) and in the potential range where cracking is predicted to occur by hydrogen embrittlement, (2) investigating the susceptibility of welds to SCC, (3) determining the lowest stress at which cracking can occur. Evaluating the role of sulphur species on the corrosion behaviour of the carbon steel overpack. Sulphur species are expected to originate from SRB (sulphate reducing bacteria) activity at the interface between the concrete tunnel lining and the Boom Clay host rock formation (oxidation of pyrite produces sulphate which is used as nutrients by SRB). A remaining uncertainty is predicting to what extent the sulphur species will be transported through the thick concrete buffer to finally reach the carbon steel overpack. Evaluating the impact of pouring a cement-based filler material onto a hot steel surface on the corrosion behaviour of the carbon steel overpack. Modelling the evolution of the properties of the oxide film formed on the carbon steel overpack.

References 1. S. Wickham, Evolution of the Near-Field of the ONDRAF/NIRAS Repository Concept for Category C Wastes. First Full Draft Report, Report NIROND-TR 2007-07E, 2008. 2. ONDRAF/NIRAS, SAFIR-1 – Safety Assessment and Feasibility Interim Report, Report NIROND, 1989. 3. ONDRAF/NIRAS, Safety Assessment and Feasibility Interim Report 2, Report NIROND 2001-06 E, 2001. 4. ONDRAF/NIRAS, The ONDRAF/NIRAS Long-Term Safety Strategy for the Disposal of High Level Waste – SFC 1 Level 4 Report: First Full Draft, Report NIROND-TR 2006-04 E, 2007. 5. A. Dierckx, Roadmap Towards the SFC 1, ONDRAF/NIRAS note 2006-0021 (rev. 0), 2006. 6. B. Kursten and F. Druyts, J. Nucl. Mater., 379(1–3) (2008), 91–96. 7. M. Van Geet, M. De Craen, E. Weetjens and X. Sillen, Extent of Oxidising Conditions in the Host Formation – Experimental Data and Scoping Calculations, SCK•CEN External Report SCK•CEN-ER-05, 2006. 8. M. Van Geet, personal communication, 2007. 9. L. Wang, Near-Field Chemistry of a HLW/SF Repository in Boom Clay – Scoping Calculations Relevant to the Supercontainer Design, SCK•CEN External Report SCK•CENER-17, 2006. 10. P. Bouniol, Radiolysis within the Concrete of a Supercontainer Including Two Primary Waste Forms – Simulation at the Concrete/steel Interface at Variable Temperatures, CEA Report RT DPC/SCCME 07-742-A, 2007. 11. E. Weetjens, X. Sillen, and M. Van Geet, Mass and Energy Balance Calculations for the VHLW/Iron/(concrete)/Clay Reference Concept, NF-PRO Deliverable 5.1.2, 2006.

The Belgian Supercontainer concept – corrosion issues

17

12. S. Poyet, Conception du Supercontaineur ONDRAF-NIRAS Phase 2: Simulation du comportement thermo-hydrique du tampon en béton en service, CEA Report RT DPC/ SCCME 04-690-A, 2005. 13. S. Poyet, Design of the ONDRAF Supercontainer Concept for Vitrified HLW Disposal in Belgium: Study of the Thermo-Hydric Behaviour of the Concrete Buffer, CEA Report RT DPC/SCCME/07-741-A, 2007. 14. B. Kursten, Uniform Corrosion Rate Data of Carbon Steel in Cementitious Environments Relevant to the Supercontainer Design – ‘Best Estimate’ from Available Literature Data. Status on December 2008, SCK•CEN External Report SCK•CEN-ER-94, 2009. 15. C. C. Naish, ‘Corrosion aspects of the proposed Sellafield waste repository’, presented at UK Corrosion ’93, 1993. 16. C. C. Naish, D. J. Blackwood, K. J. Taylor and M. I. Thomas, The Anaerobic Corrosion of Stainless Steels in Simulated Repository Backfill Environments, AEA Technology Report, NSS/R307, 1995. 17. C. C. Naish, D. J. Blackwood, M. I. Thomas and A. P. Rance, The Anaerobic Corrosion of Carbon Steel and Stainless Steel, AEA Technology Report, AEAT/R/ENV/0224, 2001. 18. N. R. Smart, D. J. Blackwood, G. P. Marsh, C. C. Naish, T. M. O’Brien, A. P. Rance and M. I. Thomas, The Anaerobic Corrosion of Carbon and Stainless Steels in Simulated Cementitious Repository Environments: A Summary Review of Nirex Research, AEA Technology Report, AEAT/ERRA-0313, 2004. 19. N. R. Smart, A Survey of Steel Corrosion Data for Use in the GAMMON Computer Program, Serco Assurance Report, SERCO/ERRA-0484, 2002. 20. N. R. Smart, ‘The corrosion behavior of carbon steel radioactive waste packages: A summary review of Swedish and U.K. research’, presented at CORROSION2008, 2008. 21. R. Grauer, B. Knecht, P. Kreis and J. P. Simpson, Werkst. Korros., 42 (1991), 637–642. 22. R. Grauer, B. Knecht, P. Kreis and J. P. Simpson, Mater. Res. Soc. Symp. Proc., 212 (1991), 295–302. 23. P. Kreis, Hydrogen Evolution from Corrosion of Iron and Steel in Low/Intermediate Level Waste Repositories, NAGRA Technical Report 91-21, 1991. 24. P. Kreis and J. P. Simpson, ‘Hydrogen gas generation from the corrosion of iron in cementitious environments’, in Corrosion Problems Related to Nuclear Waste Disposal, European Federation of Corrosion Publication No. 7. Institute of Materials, London, UK, 1992. 25. P. Kreis, Wasserstoffentwicklung durch Korrosion von Eisen und Stahl in anaeroben, alkalischen Medien im Hinblick auf ein SMA-Endlager, NAGRA Technical Report 93-27 (in German), 1993. 26. R. Fujisawa, T. Cho, K. Sugahara, Y. Takizawa, Y. Horikawa, T. Shiomi and M. Hironaga, Mater. Res. Soc. Symp. Proc., 465 (1997), 675–682. 27. R. Fujisawa, T. Kurashige, Y. Inagaki and M. Senoo, Mater. Soc. Res. Symp. Proc., 556 (1999), 1199–1206. 28. A. Fujiwara, I. Yasutomi, K. Fukudome, T. Tateishi and K. Fujiwara, Mater. Res. Soc. Symp. Proc., 663 (2001), 497–505. 29. M. Kaneko, N. Miura, A. Fujiwara and M. Yamamoto, Evaluation of Gas Generation Rate by Metal Corrosion in the Reducing Environment, RWMC Engineering Report, RWMC-TRE-03003, 2004. 30. F. A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, 4th edition, John Wiley & Sons, New York, 1980, 321. 31. D. D. Macdonald, M. Urquidi-Macdonald and G. R. Engelhardt, Simulation of Hydrogen Production in the Annulus of a Supercontainer for the Disposal of High Level Nuclear Waste in a Belgian Boom Clay Repository, Report submitted to NIRAS/ONDRAF, 2006. 32. O. Azizi, Y. Chen and D. Macdonald, Characterization of the Passive Film on Carbon Steel in Saturated Ca(OH)2+NaOH Solution, report submitted to NIRAS/ONDRAF on January 30, 2008.

18

Sulphur-assisted corrosion in nuclear disposal systems

33. O. Azizi, Y. Chen and D. Macdonald, Characterization of the Passive Film on Carbon Steel in Saturated Ca(OH)2+NaOH Solution, report submitted to NIRAS/ONDRAF on June 30, 2008. 34. D. D. Macdonald, O. Azizi and A. Saleh, ‘Characterization of the passive film on carbon steel in saturated Ca(OH)2+NaOH solution as a function of temperature’, presented at NUCPERF2009. 35. C. Andrade and J. A. González, Werkst. Korros., 29 (1978), 515–519. 36. J. A. González, S. Algaba and C. Andrade, Br. Corros. J., 15(3) (1980), 135–139. 37. N. R. Smart, A. P. Rance, R. J. Winsley, P. A. H. Fennell, B. Reddy and B. Kursten, ‘The effect of irradiation on the corrosion of carbon steel in alkaline media’, presented at NUCPERF2009. 38. B. Kursten, SCC Susceptibility Studies of C-steel in Artificial Concrete Pore Solutions, SCK•CEN External report SCK•CEN-ER-83, 2009.

2 Sulphur chemistry of the near-field Boom Clay environment Orchideh Azizi and Digby D. Macdonald Centre for Electrochemical Science and Technology, Department of Materials Science and Engineering, Pennsylvania State University, University Park, PA 16802, USA

2.1

Introduction

The disposal of Belgium’s High Level Nuclear Waste (HLNW) in the proposed Boom Clay repository requires careful consideration of the nature of the near-field environment, because of the presence of potentially corrosive species. Thus, Boom Clay contains significant amounts of pyrite, FeS2, which represents a potential source of the disulphide ion, S22− and higher oxidation products, such as the polysulphides Sx2−, x = 3 to ~6, elemental sulphur, S8, and the polythionic acids and anions, which can be generally represented as HxSyOz and SyOzx−, respectively. The clay also contains significant amounts of sulphate ion, SO42−. While the sulphate ion is generally recognized as being chemically stable, and non-corrosive towards carbon steel and stainless steels, it is known to undergo reduction in the presence of reactive metals (including iron) at elevated temperatures. Furthermore, ‘stability’ is often judged upon the basis of experiments performed over laboratory observation times, which generally do not exceed a few months in the most extended studies. It is also well known that sulphate-reducing bacteria (SRB) effectively reduce sulphate ion in groundwater environments, producing metabolic products ranging from sulphide ion and elemental sulphur to various polysulphides and polythionic acids, many of which are highly corrosive towards iron (carbon steel), nickel, copper, and sensitized Type 304 stainless steel, as noted above. Accordingly, any rational consideration of the chemistry of the near-field environment must address the chemistry of sulphur. The near-field environment is expected to be in contact with both carbon steel and highly alloyed iron-based alloys, such as stainless steels, although it is recognized that the canister material has yet to be specified in the Belgian programme. Nevertheless, the metallic elements that are likely to be in contact with the environment include iron, nickel, and chromium. The latter element is generally regarded as being immune to sulphur-induced corrosion and hence will not be considered further in this paper. On the other hand, iron [1] and nickel [2] are severely attacked by reduced sulphur species, as are sensitized Fe–Cr–Ni alloys [3]. Accordingly, any comprehensive analysis of the chemistry of the near-field environment must include an analysis of the interaction of these metals with various sulphur species in the system. The present paper summarizes the chemistry of sulphur in the form of potential/pH and volt-equivalent diagrams calculated as a function of pH and temperature [3–7]. While volt-equivalent diagrams [7] are a little known tool for representing the redox chemistry of an element, they provide a powerful method for summarizing the chemistry of a complex system, such as the sulphur/water system, which contains 19

20

Sulphur-assisted corrosion in nuclear disposal systems

numerous oxidation states. The chemistry of iron and nickel in contact with the nearfield environment is then summarized in the form of potential–pH diagrams [4–9]. These diagrams, and the volt-equivalent diagrams mentioned above, are calculated for much greater ranges of temperature and pH than will be experienced in the repository. This was done, so as to illustrate the dependence of the chemistry upon these two important independent variables. Finally, we discuss the impact that sulphur exerts upon the kinetics of metal corrosion processes, including active dissolution, inhibition of repassivation phenomena, and hydrogen evolution and promotion of the entry of atomic hydrogen into the metal. The latter phenomenon results in corrosion-related damage, such as hydrogen damage due to the formation of methane (CH4) at carbide precipitates, a phenomenon that occurs principally at elevated temperatures (T > 500°C). However, the entry of hydrogen into the matrix also induces hydrogen embrittlement, one mechanism of which is the recombination of hydrogen atoms in voids, such as those that form ahead of a growing crack due to strain-induced separation of the matrix from precipitates and inclusions, with subsequent formation of dihydrogen gas at high pressures within the cavities. This pressure, which increases with time as hydrogen segregates into the void, adds to the mechanical hydrostatic stress until the fracture stress of the ligament between the crack tip and the void is exceeded. At that point, catastrophic failure of the ligament occurs, resulting in a microfracture event, and hence in advancement of the crack. This sequence of events is also postulated to be the fundamental mechanism of ‘sulphide stress corrosion cracking (SSCC)’ that occurs in high strength, low alloy steels when exposed to H2S-containing environments [10–12]. In this case, adsorbed sulphur on the metal surface (including that at a crack tip) inhibits the recombination of hydrogen atoms on the surface to form H2, which ordinarily escapes into the environment, but, instead, promotes the entry of hydrogen into the metal. Other hydrogen-induced fracture mechanisms have been formulated and demonstrated to account for the deleterious effects of hydrogen on the mechanical properties of metals and alloys, including the decohesion mechanism, in which hydrogen reacts with segregated metalloids at the grain boundaries, resulting in grain boundary decohesion, brittle metal hydride formation, and hydrogen enhanced local plasticity (HELP) [10–12]. 2.2 2.2.1

Potential–pH diagrams General

In the investigation of corrosion phenomena, the thermodynamic properties of the system are of great importance. They not only determine whether or not a given reaction is spontaneous under the prevailing conditions, but they also define the conditions that must be achieved to minimize the effects of corrosion in practical systems. A potential/pH diagram is a compact and useful map of thermodynamic properties, summarizing thermodynamic information in terms of the electrochemical potential and the pH of the corrosion system. Thus, cathodic protection requires that the potential of the structure be displaced into the thermodynamically immune region of potential/pH space, where continued corrosion is thermodynamically impossible. On the other hand, anodic protection requires that the potential be maintained in the ‘passive’ region of potential/pH space, where maintenance of a metastable corrosion product film is thermodynamically possible [8,9,13–20], especially in the potential– pH region where the dissolved metal species dominate.

Sulphur chemistry of the near-ﬁeld Boom Clay environment

21

We now discuss a little known feature of potential–pH diagrams; that of the formation of thermodynamically metastable oxide or sulphide phases on a metal surface. Seemingly, much of the following discussion will have little to do with the thermodynamics of the S/H2O or even Fe/S/H2O systems, but it is included to illustrate the importance of metastable phase formation in the interpretation of potential–pH diagrams. We do this within the context of the resolution of ‘Faraday’s paradox’ resulting from his famous iron-in-nitric acid experiment that was reported in 1836. In Faraday’s iron-in-nitric acid experiment [21], he noted that iron freely corrodes in dilute nitric acid with the evolution of a gas (hydrogen). However, in concentrated nitric acid, no reaction apparently occurred, in spite of the greater acidity of the medium (‘pH’ was an unknown concept in the 1800s, and ‘acidity’ apparently was determined by taste and by reaction of an ‘acid’ with substances, such as limestone). If the surface of the iron in contact with the concentrated nitric acid was scratched in situ, a burst of corrosion activity (gas bubble generation) occurred along the scratch, but then quickly died away. Faraday correctly surmised that the surface had become ‘oxidized’, yet the oxide film was too thin (order of a few nanometres thick) to be detected by the naked eye (thicker films would have produced interference patterns, the physics of which were more-or-less known since the time of Newton and Huygens). The intriguing question arose, then, as to why the surface became ‘passive’ (a word apparently coined by Faraday) in the more aggressive environment, contrary to the expectations at the time of the experiment. In contemporary times, this is known as Faraday’s paradox, the resolution of which was not to evolve for more than 130 years [22–24]. In the 1960s, electrochemistry and corrosion science underwent a profound transformation with the introduction by Marcel Pourbaix [25] of his ‘potential–pH diagrams’. In his Atlas of Electrochemical Equilibria, Pourbaix summarized the electrochemical thermodynamic behaviours of most of the elements in the periodic table. Subsequently, Pourbaix diagrams have been derived for many metals in complex environments (e.g. Cu/NH3/CO2/H2O) and in environments under extreme conditions (e.g. Fe/H2O at supercritical temperatures of T > 374.15°C). For our purposes, only a simple, schematic diagram for the iron–water system (Fig. 2.1) is needed to illustrate the formation of metastable phases and the solution to Faraday’s paradox. With reference to Fig. 2.1, and noting that for spontaneity, the Second Law of Thermodynamics requires that (E–Ee)I ≥ 0 e

[2.1]

where E is the potential, E is the equilibrium potential, and I is the current for the cell, SHE|R/O, which is a hypothetical cell comprising a half cell O + yH+ + ze− ↔ R + cH2O together with a standard hydrogen electrode (H+ + e− ↔ ½H2, where the activity of H+ is 1 and the fugacity of H2 also is 1). We see that the corrosion potential must satisfy the relationship Eae < Ecorr < Ece for the corrosion process to be spontaneous (i.e. for the partial anodic and cathodic currents to be positive and negative, respectively). Thus, the term E–Ee defines the Gibbs energy change for the cell SHE| R/O, where R/O is the indicated half-cell reaction, under the prevailing conditions, such that for E > Ee, ΔG is negative and I is defined as being positive for a reaction that occurs spontaneously in the oxidation sense and for E < Ee, ΔG is negative and I is defined as being negative for a reaction that occurs spontaneously in the reduction sense. The potentials Eae and Ece refer to the equilibrium potentials for the partial anodic and cathodic reactions, respectively, in the corrosion process. Thus, iron in

22

Sulphur-assisted corrosion in nuclear disposal systems

2.1 Pourbaix diagram for iron at 60°C showing the oxides as the stable phases. Extensions of Line 13 (heavy dashed) for Fe/Fe3O4 and Line 17 (heavy dash–dot) for Fe3O4/Fe2O3 into the stability region for Fe2+ define the conditions for the formation of Fe3O4 and Fe2O3, respectively, as metastable phases [20]. The heavy solid line is the equilibrium potential for the NO2–/NO3–couple for equal nitrite and nitrate activities [Reaction 2.1 in the text]

deaerated acid solution, in which the partial anodic and cathodic reactions are Fe → Fe2+ + 2e− (Line 23, Fig. 2.1) and H+ + e− → 1/2 H2 (Line a, Fig. 2.1), respectively, will adopt a corrosion potential that lies between Lines 23 and a, with the value of Ecorr being determined by the relative values of the kinetic parameters (exchange current densities and Tafel constants) of the two partial processes. In oxygenated (aerated) solutions, Ecorr may lie between Lines 23 and b, because the reduction of oxygen is a possible (likely) cathodic reaction. Faraday, as far as we know, did not have a reference electrode or a high impedance voltmeter, so that he could not have known where on Fig. 2.1 (which did not exist at that time, anyway) the corrosion potentials for iron in dilute nitric acid and in concentrated nitric acid lay.1 However, from 174 years of collective experience in electrochemistry, since the time of Faraday, we may speculate on the E/pH conditions that existed in his experiments, as shown in Fig. 2.1. Dilute HNO3 is only a weak oxidizing agent, so that the principal cathodic reaction in the absence of oxygen 1

Faraday’s experiments have been repeated by one of the authors (DDM) and his observations are in agreement with the account given below.

Sulphur chemistry of the near-ﬁeld Boom Clay environment

23

was most likely hydrogen evolution, and hence the corrosion potential is expected to lie between Lines 23 and a at relatively high pH (say, pH 3–5). On the other hand, concentrated HNO3 is a strong oxidizing agent due to the reaction NO3−+ 2H+ + 2e− ' NO2− + H2O

[2.2]

which yields the potential–pH relationship as E e = 0.821 -

Ê aNO - ˆ 2.303RT 2.303RT 2 log Á pH ˜ÁË aNO - ˜¯ F F 3

[2.3]

where aNO - and aNO - are the activities of the nitrite and nitrate ions, respectively, and 2 3 R is the ideal gas constant (R = 8.3142 J/K.mol), F is Faraday’s constant (96 487 C/ equiv.), and T is the Kelvin temperature. Equation 2.2 is plotted in Fig. 2.1 as the heavy solid line, assuming equal values for the activities of aNO - and aNO - . Thus, in 3 2 the presence of concentrated nitric acid, the pH will be low (–1 to 0) and the corrosion potential is expected to lie just below the heavy line for NO3−/NO2−, because the kinetics of Reaction 2.2 are expected to be relatively fast (high exchange current density) compared with those for Fe2+/Fe (Line 23, Fig. 2.1). Accordingly, in the absence of the NO2−/NO3− redox couple, the partial cathodic reaction in the corrosion cell is the hydrogen electrode reaction and the corrosion potential can lie anywhere between Lines 23 and the light broken line, a. Now, the Fe/Fe2+ reaction is likely to be relatively fast compared with the H+/H2 reaction on iron. This results (from Mixed Potential Theory) in the corrosion potential lying below the extension of Line 13 (Fe/Fe3O4) into the Fe2+ stability field, but, of course, above Line 23, with Ecorr most likely lying closer to the latter. Under these conditions (dilute HNO3), Fe3O4 cannot form on the surface, even as a metastable phase and no passivity is possible. The region bounded by Line 23 and the extension of Line 13 is the only unequivocally ‘corrosion’ region in acidic solutions, contrary to the original labelling by Pourbaix, who specified the entire stability fields for Fe2+ and Fe3+ as being ‘corrosion domains’. Thus, iron is active (freely corrodes) only in this region. However, in the presence of concentrated nitric acid, other cathodic partial reactions exist, as noted above, including that provided by the NO2−/NO3−couple, the equilibrium line for which is plotted as the heavy solid line in Fig. 2.1. Again, depending upon the kinetics of the partial anodic and partial cathodic reactions, the corrosion potential can lie anywhere between Line 23 and the heavy solid line. Recognizing that the equilibrium potential for Reaction 2.2 is quite positive, and that Reaction 2.2 is likely to be fast (if for no other reason than the high concentration of NO3− relative to NO2−), the corrosion potential will be high and certainly will be more positive than the extension of Line 13 into the stability region for Fe2+ and to the low pH of concentrated HNO3. Accordingly, Fe3O4 can form as a metastable phase, thus giving rise to passivity and hence to the Faraday’s observed kinetic inactivity of iron in this medium. Removal of the film by scratching would cause the local potential to drop, due to the sudden dissolution of iron, thereby rendering hydrogen evolution a viable cathodic reaction. However, depletion of H+ at the scratch would quickly cause the potential to shift in the positive direction and lead to the reformation of Fe3O4 as a metastable, passivating phase. If the potential becomes sufficiently positive, it may even lie above the extension of Line 17 into the Fe2+ region. In this case, Fe2O3 may form on top of Fe3O4 as an additional metastable phase, resulting in the bilayer structure that is commonly

24

Sulphur-assisted corrosion in nuclear disposal systems

observed on corroded iron [24]. The above is essentially the resolution offered by Gronboy and Shrier [22] and later independently by Macdonald and Cragnolino [23] for Faraday’s paradox. Extensive potential/pH diagrams for the S/H2O, Fe/S/H2O, and Ni/S/H2O systems were developed using a chemical reaction and equilibrium software package, Outokumpu HSC-5 Chemistry for Windows. This is a chemical reaction and equilibrium software package with an extensive thermochemical database, which makes conventional thermodynamic calculations, such as the derivation of potential–pH diagrams, fast and effective. It was found that the chemical, physical, and thermodynamic data stored in the database are in good agreement with (and were frequently taken from) those in the open literature, such as those in the compilations of Naumov et al. [26], the US Bureau of Mines monographs [27], and publications of the National Bureau of Standards [28]. One of the important calculation options of the HSC-5 program is the derivation of E–pH diagrams. All of the E–pH diagrams reported in the present work, with the exception of Fig. 2.1, were developed using the HSC-5 software. Potential/pH diagrams for the S/H2O, Fe/H2O/S, Ni/H2O/S, systems at temperatures up to 300°C were calculated using the HSC-5 program. The diagrams at temperatures of 25°C and 250°C, and for various species activities as specified in the captions, are shown in Figs. 2.1 to 2.8. The dotted lines in these diagrams represent the thermodynamic equilibrium limits for the stability of liquid water. Thus, at voltages above the upper dotted line, oxygen spontaneously evolves via the oxidation of water, whereas at voltages more negative than those given by the lower dotted line, hydrogen evolution occurs from the reduction of water. Figures 2.1 and 2.2 show the potential/pH diagrams for the S/H2O system ([S] = 1.0×10−6 M, where ‘S’ is any dissolved sulphur species) at 25°C and 250°C, respectively. It was found that temperature significantly affects the thermodynamic behaviour of the system. For example, the second dissociation of sulphuric acid

2.2 Potential/pH diagram for S/H2O system at 25°C showing only dissolved species

Sulphur chemistry of the near-ﬁeld Boom Clay environment

25

2.3 Potential/pH diagram for S/H2O system at 250°C showing only dissolved species

HSO4- ¤ SO42 - + H +

[2.4]

occurs at a more acidic (lower) value of pH at 25°C than it does at 250°C. These diagrams do not include the various sulphur species with fractional oxidation states, such as the polysulphides and polythionates, but such diagrams have been developed [2,3,8–11,13–17]. The equilibrium lines for these species appear at potentials above those for the H2S/SO42−, HS−/SO42−, and S2−/SO42− equilibria indicated in Figs. 2.2 and 2.3. Figures 2.4 and 2.5 show the potential/pH diagrams for the Fe/H2O/S system at 25°C and 250°C, respectively. In the case of 25°C, elemental iron is stable over the whole range of pH at sufficiently negative potentials (E < −0.8 VSHE). As the potential is increased, oxidation of iron is predicted to occur over the entire pH range, forming Fe2+, iron disulphide (pyrite, FeS2), pyrrhotite (Fe0.877S), and magnetite (Fe3O4), depending on the pH. As the potential is increased further, these phases are no longer stable and are predicted to be oxidized into Fe3+ and iron oxide, together with a sulphate species, depending on the pH value. Figure 2.5 shows that no species containing sulphur with an activity of 1.0×10−6 M is predominant at 250°C. However, species such as pyrrhotite, iron disulphide, and iron sulphate are stable in potential/ pH regions defined by Fig. 2.6, where the activity is set to 1.0 M. Figures 2.7 and 2.8 show potential/pH diagrams for the Ni/H2O/S system at 25°C and 250°C, respectively. At 25°C, nickel is predicted to be stable over the entire pH range at potentials lower than ca. −0.6 VSHE, but at 250°C, Ni is not thermodynamically stable until much more negative voltages (i.e. increasing temperature has a significant activating effect on this element in the presence of sulphur species). The same trends are displayed by iron in the absence and presence of sulphur species (Figs. 2.4–2.6). By comparing the diagrams for each metal with those for the respective metals in the absence of sulphur, it is evident that this activating effect (reaction product formation at increasingly negative potentials) is associated with the reaction

26

Sulphur-assisted corrosion in nuclear disposal systems

2.4 E–pH diagram for Fe/H2O/S system at 25°C (all species activities = 1.0×10–6)

2.5 Potential/pH diagram for Fe/H2O/S system at 250°C (all species activities = 1.0×10–6)

of the sulphur species with the metal. Thus, in the case of Ni, for example, as the potential is increased in the positive direction from the immune region, the oxidation of nickel is predicted to occur, progressively forming Ni2+, Ni3S2, NiS0.84, NiS2, and NiO, depending on potential and acidity (pH) of the system. Note that the sulphide phases form at lower potentials than does the oxide. A similar behaviour is displayed by iron (Figs. 2.4 to 2.6).

Sulphur chemistry of the near-ﬁeld Boom Clay environment

27

2.6 Potential/pH diagram for Fe/H2O/S system at 250°C (all species activities = 1.0)

2.7 Potential/pH diagram for Ni/H2O/S system at 25°C (all species activities = 1.0×10–6)

Thus, of particular interest, in the present context, in the diagrams for both iron and nickel in sulphur-containing environments, is the formation of the metal sulphides at low potentials. A great deal is known about the formation and reactivity of the sulphides. The diagrams developed in this work indicate that the metal sulphides are predicted to exist at potentials that are significantly more negative than the equilibrium line for the hydrogen electrode reaction, indicating that hydrogen evolution becomes even more viable, thermodynamically, as a partial cathodic reaction in a

28

Sulphur-assisted corrosion in nuclear disposal systems

2.8 Potential/pH diagram for Ni/H2O/S system at 250°C (all species activities = 1.0×10–6)

corrosion process involving sulphur species than in the absence of sulphur. Furthermore, in the presence of sulphide, FeS2 is predicted to form at potentials that are significantly above the hydrogen equilibrium line, at least at ambient temperature, and hence is predicted to form and exist in moderately oxidizing environments, while the lower sulphides form at significantly more negative potentials. This prediction is supported by the fact that pyrite (FeS2) is found to be present in the natural Boom Clay. On increasing the potential into the ‘mildly oxidizing region’ (200+ mV above the hydrogen line), FeS2 and NiS2 are predicted to oxidize to iron oxides (the identity of which depends on pH) and nickel oxide and polysulphide and polythionic species (see later for a discussion of the redox chemistry of sulphur). In any event, the polysulphides and the polythionic acids are extraordinarily corrosive, inducing general corrosion in Fe and Ni in regions of low potential, and stress corrosion cracking in sensitized stainless steels and nickel-base alloys [3,5,6]. Furthermore, partially reduced sulphur species are known to induce pitting corrosion and possibly intergranular attack in sensitized stainless steels and nickel-base alloys, such as Alloy 600. The formation of partially reduced or oxidized sulphur species is therefore a critical factor in the selection of materials for service in HLNW repositories that contain FeS2, sulphate ion, and oxygen, as is discussed in the next section of this report. The corrosion scenario that is suggested by the analyses reported here is that the FeS2, sulphate ion, and oxygen that are initially present in the repository environment react to form partially oxidized and reduced sulphur species, which subsequently react with any iron or nickel in the system to form the disulphides, FeS2 and NiS2, in regions of reduced oxygen fugacity. These species subsequently react (either directly or by contributing S22− to the system) with SO42− or O2 to reproduce the partially reduced and/or oxidized sulphur species. This cycle is probably driven by a gradient in oxygen fugacity that is produced by the direct reaction of oxygen with the metals. In a sense, the sulphur cycle acts as a ‘transporter’ of oxygen to the metal surface,

Sulphur chemistry of the near-ﬁeld Boom Clay environment

29

thereby enhancing the corrosion rate and increasing the corrosion potential to the extent that pitting and intergranular stress corrosion cracking will then occur. Potential–pH diagrams for very complex systems under extreme conditions have been reported, including diagrams for iron and nickel in high salinity geochemical brine containing small amounts of sulphate and sulphide species at 250°C [10]. The diagrams also show that the chloro complex, FeCl3 is the predominant Fe(III) under these conditions. This species forms as a result of the drop in the dielectric constant of water as the temperature is increased. Thus, by forming the neutral chloro complex, the charge of +3 that was originally present on Fe3+ is shielded from the environment and the Gibbs energy of the system is lowered and hence complexing becomes spontaneous. 2.3

Volt equivalent diagrams for the sulphur/water system

2.3.1 The volt-equivalent concept Sulphur can exist in its compounds in at least 14 different oxidation states from −2 to +8, including fractional states. This situation brings about a richness in the chemistry of sulphur that is unmatched by any other element in the periodic table with the possible exception of carbon. Volt-equivalent diagrams [7] are possibly the most effective tool for pictorially displaying the thermodynamic relationships that exist between the various sulphur species. The volt equivalent (VE) of a compound or ion is the reduction potential of the species relative to the element in its standard state multiplied by the oxidation state of the element in the compound. A volt-equivalent diagram shows the volt equivalent for each species containing the element as a function of the average sulphur oxidation state (SOS) [7]. Volt-equivalent diagrams can be used to predict the behaviour of chemical systems and this tool is frequently employed in probing the redox chemistry of an element. While volt-equivalent diagrams for sulphur in its standard state are available [7], diagrams for other conditions and higher temperatures are generally not. To the authors’ knowledge, the diagrams reported here for non-standard conditions of temperatures other than 25°C are the only diagrams of the type in existence. 2.3.2

Volt equivalent diagrams for sulphur/water systems

The reduction reactions for the sulphur species relative to elemental sulphur that were considered in the present work are included in Table 2.1. A worksheet comprising thermodynamic properties of sulphur species for temperatures up to 300°C (using the HSC-5 database and the Naumov compilation) was developed and used to calculate the reduction potential of each sulphur species relative to elemental sulphur, S8) at temperatures up to 300°C. Consider the couple for the reduction of thiosulphate, i.e. S2O32 - + 6 H + + 4 e - = 1 / 3S8 + 3H 2O

[2.5]

E0 for Reaction 2.5 is 0.5 V and the average oxidation state of thiosulphate is +2. Thus, the volt equivalent for thiosulphate is 0.50 V × 2 = 1.00 V. It is worth noting that VE values have units of volts. Accordingly, given the available thermodynamic data (HSC-5 database and the Naumov compilation) for the appropriate reduction reactions at any specific temperature and for any activities of the species involved

30

Sulphur-assisted corrosion in nuclear disposal systems

Table 2.1

Reduction reactions for sulphur species

Species*

Reduction reaction

H2S(a) HS(–a) S(–2a) S2(–2a) S3(–2a) S4(–2a) S5(–2a) S8 S2O3(–2a) S4O6(–2a) H2S2O4(a) HS2O4(–a) S2O4(–2a) H2SO3(a) HSO3(–a) SO3(–2a) H2SO4(a) HSO4(–a) SO4(–2a)

1/8S8(s) + 2H+ + 2e– = H2S(a) 1/4S8(s) + H+ + 2e– = 2HS(–a) 1/8S8(s)+2e– = S(–2a) 1/4S8(s)+2e– = S2(–2a) 3/8S8(s)+2e– = S3(–2a) 1/2S8(s)+2e– = S4(–2a) 5/8S8(s)+2e– = S5(–2a) – S2O3(–2a) + 6H(+a) + 4e– = 1/4S8(s)+3H2O S4O6(–2a) + 12H(+a) + 10e– = 1/2S8(s)+6H2O H2S2O4(a) + 6H(+a) + 6e– = 1/4S8(s)+4H2O HS2O4(–a) + 7H(+a) + 6e– = 1/4S8(s)+4H2O S2O4(–2a) + 8H(+a) + 6e– = 1/4S8(s)+4H2O H2SO3(a) + 4H(+a) + 4e– = 1/8S8(s)+3H2O HSO3(–a) + H(+a) + 4–e– =1/8S8(s) + 3H2O SO3(–2a) + 6H(+a) + 4e– = 1/8S8(s)+3H2O H2SO4(a) + 6H(+a) + 6e– = 1/8S8(s) + 4H2O HSO4(–a) + 7H(+a) + 6e– = 1/8S8(s) + 4H2O SO4(–2a) + 8H(+a) + 6e– = 1/8S8(s)+4H2O

*Species designation as in HSC-5. ‘(–2a)’ designates an anion of charge –2 in aqueous solution.

in the reaction, the VE values for a whole range of sulphur species were determined. A plot of these VE values versus average sulphur oxidation state generated the volt-equivalent diagrams for the S/H2O system with the pH values ranging from 0 to 14 and at temperatures up to 300°C. Figures 2.9 to 2.13 show the volt-equivalent diagrams for the sulphur/H2O system at pH 0 and 10.5, and at temperatures of 25°C, 150°C and 275°C. The diagrams presented herein involve sulphur species at unit activity. The first feature to note is that the slope of a line joining any two species corresponds to the reduction potential of the associated couple. Consider, for example, the SO42−/S2O32− couple in Fig. 2.9. The standard reduction potential E0 for this couple can be calculated, i.e. E0 =

VESO 4(2 - ) - VES 2O 3(2 - ) SOSSO 4 (2 - ) - SOSS 2O 3(2 - )

=

2.12 - 1.00 = 0.280(V ) 6-2

[2.6]

where VE and SOS represent the volt equivalent and sulphur oxidation state of the species, respectively. In this way, the reduction potential for any conceivable redox couple can be readily calculated. It is known that the reduction potential is highly pH-dependent. Figure 2.10 shows a volt-equivalent diagram for the S/H2O system at pH 10.5. It is clear that the shape of the diagram is much different from that shown in Fig. 2.9, with most of the slopes now being negative. The relative positions of some species have also been changed. The observation makes it clear that knowledge of the pH of a system is vital for defining and studying its redox chemistry.

Sulphur chemistry of the near-ﬁeld Boom Clay environment

31

2.9 Volt-equivalent diagram for the S/H2O system at 25°C, pH 0

2.10 Volt-equivalent diagram for the S/H2O system at 25°C, pH 10.5

Temperature is another important parameter to consider. Figure 2.11 shows a volt-equivalent diagram for the S/H2O system at pH 0 and at 150°C. Comparison of Fig. 2.11 and Fig. 2.9 suggests that, although the shapes of the diagrams are similar, the relative positions of some species are noticeably different. This implies that increasing or decreasing the temperature will change the reduction potential for the relevant species and hence will change the redox chemistry of the system.

32

Sulphur-assisted corrosion in nuclear disposal systems

2.11 Volt-equivalent diagram for the S/H2O system at 150°C, pH 0

The volt-equivalent diagrams are also useful in studying a chemical system in the following three aspects. First, if a species lies above a line joining any two other compounds, this species will tend to disproportionate into the other two compounds. See Fig. 2.12 and consider, for example, the decomposition of S2O32− in acidic solution. It

2.12 Volt-equivalent diagram for the S/H2O system at 275°C, pH 0

Sulphur chemistry of the near-ﬁeld Boom Clay environment

33

is clear that S2O32− lies above the line joining S and S4O62−. This suggests that, although the decomposition of S2O32− is a complex reaction forming many products, one reaction that may occur is 5S2O3 2 - + 6 H + Æ 2S + 2S4O6 2 - + 3H 2O

[2.7] 2−

This shows that it is thermodynamically possible for thiosulphate, i.e. S2O3 , to undergo a redox self-disproportionation reaction into a reduction product, S, and an oxidation product, S4O62−. It is reactions of this type that have led to the partially oxidized or reduced sulphur species being termed ‘highly labile’. Second, if a species lies below a line joining any two other compounds, the latter will tend to react to produce the former. A good example is found in the reduction of sulphuric acid by hydrogen sulphide. Figure 2.12 suggests that, at pH 0 and at 275°C, the following reaction is thermodynamically favourable 3H 2 S + SO4 2 - + 2H + ¤ 4S + 4 H 2O

[2.8]

2−

because S lies below the line joining H2S and SO4 . Finally, if a species lies between two or more others on the same straight line (or very close to it), it will tend to only partially disproportionate into these species, forming an equilibrium mixture containing substantial amounts of each. For example, it can be seen from Fig. 2.13 that S2O32−, SO42−, and S are very nearly on the same straight line. Thus, the disproportionation of S2O32− will occur as described by the following reaction 3S2O3 2 - + 2H + ¤ 4S + 2SO4 2 - + H 2O 2−

[2.9] 2−

Starting with 1 M S2O3 , a fixed pH of 10.5, and without S and SO4 initially in the system, the above reaction will reach an equilibrium position in which all three

2.13 Volt-equivalent diagram for the S/H2O system at 275°C, pH 10.5

34

Sulphur-assisted corrosion in nuclear disposal systems

2.14 Structure of the common allotrope of sulphur, S8. Taken from http://en.wikipedia. org/wiki/Sulfur

species are present simultaneously. Similarly, reaction of two species on or near the same line as a third will not go to completion, but will form a mixture containing significant amounts of each species. 2.3.3

Reactivity of sulphur species

At this point, it is worth inquiring into why these partially oxidized/reduced sulphur species are so reactive towards metals and alloys. Elemental sulphur in its most stable state at normal pressure and temperature exists primarily as S8 ring molecules, as shown in Fig. 2.14. Sulphur is a complex substance and forms more than 30 solid allotropes, which is more than any other element in the periodic table. In addition to S8, several other rings are known, including S7, which is more deeply yellow than is S8. Chromatographic analysis of ‘elemental sulphur’ reveals an equilibrium mixture of mainly S8, but also S7 and small amounts of S6. Even larger rings, including S12 and S18, have been prepared. On the other hand, sulphur’s lighter neighbour, oxygen, exists in only two allotropic forms of any significance: O2 and O3 and does not form cyclic structures. However, selenium, the heavier analogue of sulphur, can form rings, but is more often found as linear polymer chains. The crystallography of sulphur is also very complex and, depending on the specific conditions, the sulphur allotropes form several distinct crystal structures, with rhombic and monoclinic S8 being the best known. A noteworthy property of sulphur is that the viscosity in its molten state, unlike most other liquids, increases with increasing temperature above 200°C (392°F), due to the formation of linear polymers This requires opening of the S8 rings and polymerization of those units. Molten sulphur assumes a dark red colour above this temperature. At higher temperatures, however, the viscosity is decreased as depolymerization occurs. The increase in viscosity can be suppressed by dissolving small amounts of H2S into the melt. This occurs, because the ends of the polymer chains become ‘capped’ with SH groups, thereby preventing further polymerization. Upon cooling, the H2S is released as the sulphur reverts back to the S8 rings, with sometimes tragic results in the sulphur handling industry. Amorphous or ‘plastic’ sulphur can be produced through the rapid cooling of sulphur from the molten state. X-ray crystallography studies show that the amorphous form may have a helical structure with eight atoms per turn. This form is metastable

Sulphur chemistry of the near-ﬁeld Boom Clay environment

35

at room temperature and gradually reverts back to the crystalline form. This process happens within a matter of hours to days, but can be rapidly catalyzed. Because the bonding is fully satisfied in the S8 ring, elemental sulphur in the dry state is not particularly corrosive towards metals and alloys as few highly reactive ‘free radical’ entities, of the form −S−S−S•, exist in the solid. As the temperature is raised, the S8 rings open and polymerize to produce linear polymers of the type, •S–(S)n–S•, which contains two reactive, free radical centres per chain. (The free radical centres, designated here as ‘•’, are unpaired electrons, which represents a higher energy state than the molecule would be in if the electrons were paired.) Accordingly, the free radical species are reducing agents and hence can accept electrons from the oxidation of a metal, as M + •S–(S)n–S• Æ M2+ + −S–(S)n–S−

[2.10]

i.e. resulting in the formation of a metal polysulphide and hence corrosion. In the case of the polysulphides themselves, and the polythionates, as well, they are capable of delivering highly reactive mono-atomic sulphur (a diradical) to a reaction centre on a metal, for example, as Sx• Æ Sx–1• + •S•

[2.11]

S2O32− Æ SO32− + •S•,

[2.12]

and

respectively. Of course, these radical species may polymerize to form •S–(S)n–S•, which, if n is sufficiently large, may result in the precipitation of elemental sulphur, •S–(S)n–S• Æ •S–(S)n–8–S• + S8. This is seen upon increasing the pH of a polythionate or polysulphide solution; the solution eventually becomes cloudy due to the formation a suspension of colloidal, elemental sulphur. In the presence of water, sulphur may undergo the following redox disproportionation reaction 1/2S8 + 4H2O = 3H2S + H2SO4

[2.13]

The H2S so formed may then react with elemental sulphur to form the polysulphides, (n/8)S8 + H2S Æ H2Sn+1, thereby producing the reactive free radicals and accounting for the extraordinary reactivity of wet elemental sulphur towards iron [1]. The corrosion product of the reaction between iron and wet elemental sulphur is mackinawite, as noted elsewhere in this review. It forms as a particularly pyrophoric product in the presence of chloride ion, possibly because Cl− causes the mackinawite to be produced in a more finely divided form. 2.4 Potential/pH diagram of two-dimensional phases of elements (sulphur and oxygen) adsorbed on metal surfaces (Fe, Ni) The principles behind, and the use of, classical potential–pH diagrams, which are defined in terms of bulk oxide, hydroxide, oxyhydroxide, and sulphide phases, are well known. But these classic diagrams generally do not predict the formation of two-dimensional phases of adsorbed species on the metal surface, which usually are more stable than the bulk compounds. The formation of two-dimensional, adsorbed phases must not be neglected, because the presence of an adsorbed monolayer can induce marked changes in the reactivity of a metal and are generally the precursors to the formation of the bulk phases, which we recognize as ‘corrosion products’.

36

Sulphur-assisted corrosion in nuclear disposal systems

2.15 Equilibrium potential–pH diagram for the system S–Fe–H2O at 25°C. Activities of dissolved sulphur and iron species are, on the molal scale: (S) = 10–4; (Fe) = 10–6. The domains are limited by the lines: short dashed line, water stability; S–H2O system; Fe–S–H2O system; and Sods(Fe)–S–H2O system, Ɓ is the relative surface coverage of adsorbed sulphur [15]

Figure 2.15 is (apparently) the first reported potential–pH diagram for the Fe–S– H2O system taking into account the existence of adsorbed sulphur on iron [18]. This diagram was obtained for ambient temperature (25°C) and reveals that the monolayer of adsorbed sulphur is stable over wide ranges of potential and pH. Also the monolayer of adsorbed sulphur on iron is stable over a much larger domain than is iron sulphide. It is clear that the adsorbed sulphur overlaps the domains of stability of Fe, Fe2+, Fe2O3, and Fe3O4. This shows that, in the absence of formation of bulk iron sulphide, a monolayer of adsorbed sulphur can be formed on the surface that activates the steel towards corrosion by providing for an oxidation process at more negative potentials than is the case in the absence of sulphur. Similar potential/pH diagrams for Fe/S/H2O in the presence of adsorbed sulphur at higher temperatures, such as at 300°C, have been developed and, again, the importance of adsorbed sulphur in activating the metal is illustrated [18]. A direct manifestation of the activating effect of adsorbed sulphur is the extreme susceptibility of iron towards corrosion when in contact with wet elemental sulphur, particularly when the aqueous phase is

Sulphur chemistry of the near-ﬁeld Boom Clay environment

37

brine [1]. In this case, one of the reaction products is mackinawite, Fe1+xS, 0 < x < 0.1, which provides no protection to the underlying steel. Furthermore, mackinawite is pyrophoric, spontaneously igniting, apparently, when the moisture content drops below a certain level, but is above a certain minimum [1]. The combustion reaction can be written as: 2 Fe1+ x S +

7 + 3x O2 = (1+ x )a - Fe2O3 + 2SO2 2

[2.14]

thereby producing hematite (a - Fe2O3, which has a colour (orange) that is not too different from that of elemental sulphur (yellow). One of the authors (DDM) has personal experience with the problems that may arise in transporting elemental sulphur in steel vessels, which amply illustrates the corrosion chemistry alluded to above. Thus, a bulk sulphur carrier (ship) took on a cargo of elemental sulphur in Vancouver, BC, Canada bound for Mt. Maunganui in New Zealand, where the sulphur was to be converted into sulphuric acid for the production of superphosphate fertilizer. While crossing the Pacific, the cargo was allegedly sprayed with seawater to suppress the formation of explosive sulphur dust. Upon arrival in New Zealand, the sulphur was removed from the hold and it was noted that the last amounts to be removed were contaminated by a black material, which turned out to be mackinawite. The hold was also found to be heavily corroded, including the existence of very large pits (inches across). As time went on, the black contaminant in the sulphur pile on the wharf ‘disappeared’ (corresponding to the oxidation of the surfaces of the mackinawite particles to hematite, resulting in a colour change that caused the mackinawite to ‘blend in’ visually with the sulphur), but then the sulphur pile ignited. The local fire department responded and sprayed water on the pile to quench the fire and then left. Sometime later, the sulphur pile ignited again, corresponding to the moisture content of the mackinawite dropping into the susceptible, ignition range. This sequence continued until, apparently, no more pyrophoric mackinawite was available for re-ignition. The potential/pH diagram for two-dimensional phases of elements (sulphur and oxygen) adsorbed on Ni is presented in Fig. 2.16. This diagram allows one to predict the E–pH conditions in the presence of the monolayer of adsorbed sulphur on nickel [19] under which the metal first activates and, as in the case of iron above, accounts for the activating effect of adsorbed sulphur on the corrosion behaviour of nickel. Thus, when the potential is increased anodically from the immune region, adsorbed water is replaced by adsorbed sulphur, and then the adsorbed sulphur is replaced by solid nickel sulphides, then by adsorbed oxygen and, eventually, by the solid oxides and dissolution products. Because only adsorbed sulphur is required to activate iron and nickel, the amount of sulphur required to activate the metal and induce corrosion can be very small. At ambient temperature, the stability domain of sulphur adsorbed on nickel is much larger than the domains of the nickel sulphides. Also the stability domain of adsorbed sulphur overlaps the domains of Ni, Ni2+, Ni(OH)2(s), and Ni(OH)3−. In other words, adsorbed sulphur tends to destroy the passivity afforded by the oxides. The thermodynamic prediction of large domains of stability of sulphur monolayer’s adsorbed on nickel is therefore of great interest in evaluating corrosion risk. From the potential/pH diagrams for Fe and Ni, it is clear that the stability domain of sulphur adsorption on Ni and Fe is not limited to the Ni and Fe immunity domain. The diagrams presented in Figs. 2.15 and 2.16 predict that sulphur can adsorb in

38

Sulphur-assisted corrosion in nuclear disposal systems

2.16 Equilibrium potential–pH diagrams for the system Sads(Ni)–Oads(Ni)–S–Ni–H2O at (a) 25°C and (b) 300°C. The activities of dissolved sulphur and nickel species are, on the molal scale; (S) =10–4; (Ni)=10–6. The stability domains are limited by the lines (– –) H2O system;(----) S-H2O system (⎯) S–Ni–H2O system (Ƚ) Sads(Ni)– Oads(Ni)–S–H2O. θ is the relative surface coverage of adsorbed sulphur [19]

the domain of anodic dissolution of iron and nickel, which means that, even though the metal is not thermodynamically stable and dissolves, a monolayer of sulphur may adsorb on the fresh surface that is being continuously produced and aid in the extraction of metal ions from the surface. Accordingly, this process most likely increases the rate of dissolution of Fe and Ni to form Ni2+ and Fe2+ (see below). The diagrams also indicate that the stability domain of adsorbed sulphur overlaps the smaller regions of stability of the bulk oxides and hydroxides, namely, Ni(OH)2, NiO, Fe2O3 and Fe3O4. It is well known that the E–pH diagrams are constructed on a thermodynamic basis and do not indicate which species actually forms on a bare Ni or Fe electrode polarized in this domain: the two-dimensional (surface) species Sads or the three-dimensional (bulk) oxide Ni(OH)2, Fe2O3 and Fe3O4, but the fact that adsorbed sulphur is known to activate both nickel and iron indicates that adsorbed sulphur forms preferentially. Furthermore, if the kinetics of adsorption of sulphur on bare Ni or Fe is more rapid than the kinetics of formation of oxides, a sulphur monolayer will form on the bare metal and prevent or delay passivation by adsorbed oxygen or by the formation of the bulk oxides, whereas, in the opposite case, the formation of the passive film would tend to block surface sites available for S adsorption. 2.4.1 Influence of sulphur on the dissolution and passivation of Ni and Fe by electrochemical measurements The best electrochemical techniques to study the electrochemical behaviour of various metals in the presence of adsorbed sulphur are potentiodynamic experiments and electrochemical impedance spectroscopy (EIS). The anodic polarization curve of Ni in 0.1 N H2SO4 shows that, in the presence of adsorbed sulphur, the rate of

Sulphur chemistry of the near-ﬁeld Boom Clay environment

39

2.17 Anodic polarization curves of (a) sulphur-free and (b) sulphur-covered Ni [17]

dissolution of nickel increases in the active region and that the active region is extended significantly towards both more positive and more negative potentials, as shown in Fig. 2.17 [17]. This observation is consistent with adsorbed sulphur being both a thermodynamically activating species and a kinetically activating entity, as postulated above. Figure 2.18 shows the anodic polarization curve of Ni–25Fe in 0.05 M H2SO4 [2]. The most important result obtained from these experiments is that adsorbed sulphur inhibits the growth of the passive layer on the alloy, as long as a complete layer of adsorbed sulphur remains on the surface. This may be due to the complete poisoning of the sites usually available for adsorption of oxygenated species from the solution (such as OH− ions or O atoms). It is also worth noting that only partial desorption of the sulphur is required to allow the passive layer to be formed again. After passivation, the current density has the same magnitude with or without sulphur being present on the surface. Figure 2.19 displays the potentiodynamic polarization curves for carbon steel in [Ca(OH)2+ NaOH] with pH 13 [30]. The potentiodynamic curves displayed in this figure reveal that the anodic branch of the polarization curve moved towards higher current density values by increasing the concentration of the S2− ion in the solution. This indicates that sulphide ion accelerates the anodic dissolution of carbon steel in both the active and passive regions, although it is possible that a redox current due to the oxidation of S2− may contribute to the observed total current flowing across the interface. Sulphide ion is seen to accelerate the hydrogen evolution reaction within

40

Sulphur-assisted corrosion in nuclear disposal systems

2.18 Anodic polarization curves of sulphur-free and sulphur-covered Ni–25Fe. Ɓ represents the sulphur coverage as measured using the radio-tracer (35S) [2]

2.19 Potentiodynamic curves on carbon steel in [Ca(OH)2+ NaOH] with pH 13 in the presence of S2– at different concentrations [30]

the Tafel region at potentials that are not too negative of the zero current potential, a finding that has been noted previously [3]. Finally, the properties of the passive film formed anodically on carbon steel in Ca(OH)2+ NaOH with pH 13 in the presence of sulphide ion have been studied by using electrochemical impedance spectroscopy (EIS), and Nyquist plots of the impedance are plotted in Fig. 2.20. These data show that the impedance of the system decreases with increasing S2− concentration, which is consistent with the results obtained from the potentiodynamic and potentiostatic experiments showing a loss of passivity. Determination of the exact mechanism by which loss of passivity occurs is

Sulphur chemistry of the near-ﬁeld Boom Clay environment

41

2.20 Nyquist plots of impedance data on carbon steel in [Ca(OH)2+ NaOH] with pH 13 in the presence of S2– at different concentrations [29]

currently unknown and must await thorough analysis of the impedance data in terms of the Point Defect Model [24]. It is also clear from Fig. 2.21 that the passive current density measured potentiostatically under steady-state conditions increases with increasing S2− concentration. Thus, according to the potential/pH diagram for iron in the presence of sulphide adsorption, we conclude that sulphide: (a) is a catalyst for the anodic dissolution and (b) inhibits the formation of the passive film. 2.4.2

Stress corrosion cracking

As noted above, environments containing sulphur species are known to promote stress corrosion cracking in metals and alloys. Classical examples include the

2.21 Steady-state passive current transient on carbon steel in [Ca(OH)2+ NaOH] with pH 13 in the presence of S2– at different concentrations [29]

42

Sulphur-assisted corrosion in nuclear disposal systems

‘sulphide stress corrosion cracking’ of high strength, low alloy steel tubulars in the oil industry and the intergranular stress corrosion cracking of sensitized Type 304 SS in the petroleum refining industry and in the pulp and paper industry. Chromiumcontaining alloys that contain sufficient chromium to form a defective chromic oxide, Cr2+xO3–y, barrier layer in the passive film are immune to wet elemental sulphur attack and attack by species that can donate elemental sulphur to a metal surface. However, if the alloy is sensitized, as in the case of the stainless steels, the emergent grain boundaries are denuded of chromium by reaction of chromium with carbon and the subsequent precipitation of chromium carbides. In these cases, the chromium content in the emergent grain boundaries is too low (<11 at.%) to lead to the formation and maintenance of a chromic oxide barrier layer on the alloy surface and severe grain boundary attack and intergranular stress corrosion cracking (IGSCC) occur. Recent work in our laboratory on cracking in sensitized Type 304 SS in thiosulphate solution [5,6] has demonstrated that the enhanced IGSCC is due to hydrogeninduced fracture, whereby the thiosulphate promotes the entry of hydrogen into the chromium-depleted grain boundary matrix ahead of the crack tip. In the case of carbon steels, the heat affected zones of welds are particularly susceptible to sulphide stress corrosion cracking. This is because of the formation of martensite in the heat affected zone a millimetre or so from the weld fusion line. Martensite forms via a diffusionless shear transformation of austenite upon sudden quenching and is an exceptionally hard matrix (Rc > 40 at T < 300°C) that is susceptible to brittle fracture. The fracture commonly occurs intergranularly along the prior austenite grain boundaries, with the exact mechanism still being somewhat controversial. That fracture is induced by hydrogen is not in dispute, nor is the fact that sulphide promotes the entry of hydrogen into the matrix. However, the exact mechanism by which hydrogen entry is promoted is in dispute with one possibility being that H2S poisons the hydrogen atom recombination reaction, thereby leading to an increase in the hydrogen atom concentration on the surface and hence to an increase in the rate of hydrogen entry. On the other hand, it may be postulated that H2S promotes the reduction of H+ to form H on the surface, again leading to an increase in the concentration of hydrogen atoms at the interface and hence to a higher rate of penetration of H into the substrate. The fate of the hydrogen thereafter is also not a settled issue and any one of the HIC mechanisms identified earlier in this review could account for the observed embrittlement. The observed fact, however, is that a combination of high hardness of the metal (Rc > 22) and H2S in the environment leads to stress corrosion cracking, a phenomenon that has plagued the oil production industry for many decades. 2.5

Implications for the Boom Clay repository

The findings of this study have important implications for the choice of canister material for service in Belgium’s Boom Clay repository for the isolation of high-level nuclear waste (HLNW). The principal implications are as follows: 1. The presence of both SO42− and S22− (the latter in the form of pyrite, FeS2) in the clay simultaneously suggests that redox reactions will lead to a variety of polysulphide and polythionic acids, along with elemental sulphur, given sufficient time and particularly under initial oxic conditions.

Sulphur chemistry of the near-ﬁeld Boom Clay environment

43

2. Reaction of iron and nickel with partially reduced (polysulphides) or partially oxidized (polythionic acids and anions) is predicted to produce FeS2 and NiS2, respectively. Reaction of these compounds with sulphate is expected to reproduce the partially reduced and oxidized sulphur species, as described above. Thus, all of the components are present for the operation of a cyclic process in which iron and nickel are corroded and the corrosive agents (polysulphides and polythionic acids and anions) are continually reproduced via the reduction of sulphate. The cycle is envisioned to be driven by a gradient in oxygen fugacity, which is generated by the direct reaction of oxygen with the metals and alloys in the system. 3. Even though the cyclic process described above were proven to be acceptably slow, parallel reactions are affected by sulphate reducing bacteria, which are ubiquitous in groundwater systems. These reactions are known to occur within laboratory observation times under conditions that are prototypical of the Boom Clay repository. 4. Noting the well-recognized aggressiveness of partially reduced and partially oxidized sulphur species (including elemental sulphur) towards iron, nickel, and sensitized stainless steels and nickel alloys (e.g. Alloy 600), often resulting in autocatalytic attack, pitting corrosion, and stress corrosion cracking, considerable care must be exercised in selecting canister material for Boom Clay repository service. 5. The use of carbon steel as an overpack material in the super-container concept of Belgium’s HLNW disposal technology needs to be carefully analysed, because of autocatalytic attack by wet elemental sulphur and by polysulphides and possibly polythionic acid species. Likewise, sensitized austenitic stainless steels and nickelbase alloys should be avoided for the liner or overpack, if they might be in the sensitized condition, because of their susceptibilities to pitting attack and intergranular stress corrosion cracking in environments containing partially reduced and oxidized sulphur species. 6. In the presence of sulphur compounds in the near-field environment, the formation of elemental sulphur and formation of a two-dimensional phase of adsorbed sulphur on the metal surface, which is usually more stable than the bulk metal sulphide, is possible. This phenomenon must not be neglected, because the presence of an adsorbed monolayer can induce marked changes in the reactivity of the metal, leading to both thermodynamic and kinetic activation. 2.6

Summary and conclusion

According to the data and concepts discussed above, it is evident that sulphur and sulphur-containing species, when in contact with iron and nickel in aqueous solution, can have a profound impact on the corrosion behaviours of these metals and their alloys. Specific issues related to the presence of sulphur and sulphur-containing species are as follows: •

Thermodynamic analyses, in the form of potential–pH diagrams, indicate that sulphur species that are capable of donating elemental sulphur to a reaction centre (e.g. a metal surface) activate iron and nickel by allowing for the formation of non-protective metal sulphides (FeS, NiS) and metal disulphides (FeS2, NiS2) at potentials that are significantly more negative than those for the formation of the protective oxides.

44 •

•

•

•

• •

2.7

Sulphur-assisted corrosion in nuclear disposal systems The redox chemistry of the sulphur–water system, as described by volt-equivalent diagrams, is highly complex, because of the existence of 14 oxidation states (some of which are fractional) between −2 and +8 in this system. The species in the S–H2O system, with the exception of SO42− and sulphite, SO32−, but including the polysulphides, Sx2−, and the polythionic acids, HxSyOz and their oxyanions, are highly labile and readily convert into other species as the redox conditions change. Many of these species readily react or disproportionate to yield elemental sulphur. Wet elemental sulphur, particularly when the aqueous phase is a brine, is found to be a powerful corrodent of iron and probably also of nickel and their alloys, due to the formation of non-protective, pyrophoric mackinawite, Fe1+xS. The elemental sulphur may be formed by chemical transformations of the labile sulphur species or by sulphate-reducing bacteria that use sulphate ion in their metabolic process. Adsorbed sulphur on iron and nickel increases the rate of dissolution of the metal in the active region of the polarization curve. This important effect is due to the strong S–metal bonding to form absorbed sulphur and the weakening of the metal–metal bonds of surface atoms, resulting in a lowering of the activation energy for the dissolution of surface metal atoms. Another important effect of adsorbed sulphur is that S prevents, or delays the formation of passivating adsorbed atomic oxygen or the formation of a threedimensional oxide passive film on the metal. The sulphur-covered surface of the metal cannot be passivated as long as a complete monolayer of adsorbed sulphur remains on the metal surface (Ɓ = 1). However, only partial dissolution of the adsorbed sulphur is necessary to allow the passive oxide layer to form again. This effect is due to a complete blocking of the adsorbed sites usually available for OH groups, which are the initial resource in the growth of the passive oxide layer. Adsorbed sulphur may also block the recombination of hydrogen on the surface to form dihydrogen (H2) molecules; this site blocking decreases the rate of H2 evolution and promotes the entry of H into the metal. Chromium-containing alloys that contain sufficient chromium to form a defective chromic oxide, Cr2+xO3–y, barrier layer in the passive are immune to wet elemental sulphur attack and attack by species that can donate elemental sulphur to a metal surface. However, if the alloy is sensitized, as in the case of the stainless steels, the emergent grain boundaries are sufficiently denuded of chromium by reaction of chromium with carbon and the subsequent precipitation of chromium carbides to allow attack. In these cases, the chromium content in the emergent grain boundaries is too low (<11 at.%) to lead to the formation, and to maintain the existence, of a chromic oxide barrier layer on the alloy surface and severe grain boundary attack and intergranular stress corrosion cracking (IGSCC) may occur. Recent work in our laboratory on cracking in sensitized Type 304 SS in thiosulphate solution has demonstrated that the enhanced IGSCC is due to hydrogen-induced fracture, whereby the thiosulphate promotes the entry of hydrogen into the chromium-depleted grain boundary matrix ahead of the crack tip. Acknowledgments

The authors gratefully acknowledge the support of this work by ONDRAF/NIRAS of Belgium via a contract to the Pennsylvania State University.

Sulphur chemistry of the near-ﬁeld Boom Clay environment

45

References 1. D. D. Macdonald, B. Roberts and J. B. Hyne, Corros. Sci., 18 (1978), 499–501. 2. P. Marcus, A. Teissier and J. Oudar, Corros. Sci., 24 (1984), 259. 3. D. D. Macdonald, ‘Critical issues in the use of metals and alloys in sulphur-containing aqueous systems’, in Proc. 31st Annu. Inst. Met. Conf.: Materials Performance, Sulphur and Energy, Edmonton, Canada, 71–89, 1992. 4. D. D. Macdonald and G. Cragnolino, ‘The critical potential for the IGSCC of sensitized type 304 SS in high temperature aqueous systems’, in Proc. 2nd Int. Symp. Env. Deg. Mater. Nucl. Power Syst. – Water Reactors (9–12 September 1985, Monterey, CA, ANS). 5. M. Gomez-Duran and D. D. Macdonald, Corros. Sci., 45(7) (2003), 1455. 6. M. Gomez-Duran and D. D. Macdonald, Corros. Sci., 48(7) (2003), 1608. 7. B. Wassink, P. D. Clark and J. B. Hyne, Alberta Sulphur Research Ltd. Quarterly Bulletin, Vol. XXV, No. 4, 1, January–March 1989. 8. D. D. Macdonald and J. B. Hyne, Atomic Energy of Canada Limited Research, Report AECL-5811, November 1979. 9. D. D. Macdonald, ‘Thermodynamics of corrosion for geothermal systems’, in Proc. ACS Symp. on Corrosion in Technological Environment. Special Technical Publications 717. Philadelphia, PA, 1980. 10. H. P. van Leeuwen, Rev. Coatings Corros. (Corros. Rev.), 4(1) (1979), 5. 11. D. Delafosse and, T. Magnin, Eng. Fract. Mech., 68(6) (2001), 693. 12. S. P. Lynch, ‘Mechanisms of hydrogen assisted cracking – a review’, in International Conference on Hydrogen Effects on Material Behaviour and Corrosion Deformation Interactions, 2003, 449. 13. D. D. Macdonald and B. C. Syrett, Corrosion, 35 (1979), 471. 14. D. D. Macdonald, B. C. Syrett and S. S. Wing, Corrosion, 35(8) (1979), 367. 15. B. C. Syrett and D. D. Macdonald, Corrosion, 35(11) (1979), 505. 16. B. C. Syrett, D. D. Macdonald and S. S. Wing, Corrosion, 35(9) (1979), 409. 17. P. Marcus and E. Protopopoff, J. Electrochem. Soc., 137 (1990), 2709. 18. P. Marcus and E. Protopopoff, J. Electrochem. Soc., 140 (1993), 1571. 19. J. Ourdar and P. Marcus, Appl. Surf. Sci., 3 (1979), 48. 20. P. Marcus and J. Ourdar, in Fundamental Aspects of Corrosion Protection by Surface Modification, PV 84-3, 173, ed. E. McCafferty, C. R. Clayton and J. Ourdar. The Electrochemical Society Softbound Proceedings Series, Pennington, NJ, 1984. 21. M. Faraday, Experimental Researches in Electricity, Vol. 2, 234, Dover, Mineola, NY (reprinted 1965). 22. T. S. Gronboy and L. L. Shrier, Electrochim. Acta, 11 (1966), 895. 23. D. D. Macdonald and G. A. Cragnolino, ‘Corrosion and erosion – Corrosion of materials in steam cycle systems’, Chapter 9 in Water Technology for Thermal Power Systems, ed. P. Cohen. ASME, New York, NY, 1989. 24. D. D. Macdonald, Pure Appl. Chem., 71 (1999), 951. 25. M. Pourbaix, Atlas of Electrochemical Equilibria, NACS International, Houston, TX, 1976. 26. G. B. Naumov, B. N. Ryzhenko and I. L. Khodakosky, Hand book of Thermodynamic Data, USGS Transl. USGS-WRD-74-001, US Geological Survey, 1974. 27. Bulletin of the United States Bureau of Mines, 584 (1960). 28. National Bureau of Standards, Technical Note, No. 270. US GPO, Washington, DC. 29. D. D. Macdonald, J. Electrochem. Soc., 153(7) (2006), B213. 30. D. D. Macdonald and O. Azizi, unpublished results from research carried out at Pennsylvania State University for ONDRAF-NIRAS, 2006–2008.

3 Corrosion mechanisms and material performance in environments containing hydrogen sulfide and elemental sulfur Liane Smith Intetech Ltd, Salmon Court, Rowton Lane, Rowton, Chester, CH3 6AT, UK [email protected]

Bruce Craig MetCorr, 100 Fillmore Pl. Suite 500, Denver, Colorado, 80206, USA

3.1 3.1.1

Performance of carbon steels in environments containing H2S and S Protectiveness of the sulfide scale

It is important first to discuss the protectiveness of iron sulfide scales before the role of S is addressed. Iron sulfide scales (also referred to as films) are of great interest from a practical standpoint since protective scales reduce the corrosion rate and in many cases eliminate the need for corrosion inhibitors. However, the protectiveness of corrosion product layers is not easily studied or quantified. In areas such as Canada and the Middle East where wells produce oil and gas with H2S but no CO2, field experience has shown that the corrosion product layer can indeed be quite protective even at very high concentrations and partial pressures of H2S. On the other hand, many laboratory investigators have suggested that iron sulfides cannot be protective since many are non-stoichiometric and, therefore, unstable. Not all laboratory studies have evaluated the protectiveness of iron sulfide scales but there have been some efforts in this area. Smith and Pacheco [1] described the nature of H2S-controlled corrosion as being typified by discrete pitting attack. Most of the research they reviewed was concerned with high levels of H2S and there was also elemental S present – a notorious initiator of pitting corrosion. Furthermore, high concentrations of H2S have been produced in oil and gas fields without pitting so it is not inevitable that localized attack will occur. Other factors such as chlorides, O2 and CO2 can increase the likelihood of localized attack. The role of chlorides in sour systems has not been specifically studied to any great extent but is usually included to some degree in studies aimed at H2S–CO2 systems. Sridhar et al. [2] examined the effect of chlorides up to 160 000 ppm on the corrosion rate of steel in H2S + CO2 + O2 and reported very little impact on the average corrosion rate. However, at concentrations greater than 5% (50 000 ppm) chlorides, they observed localized corrosion. No further characterization was made even though this was a crucial observation. 46

Corrosion mechanisms and material performance

47

3.1 Corrosion product layers inside the pits including a thin FeCl2 layer

It has been found in field failures of wells, when very high concentrations of chlorides and high concentrations of H2S are present, that severe pitting can occur. Hamby [3] noted the morphology of the pitting in this type of environment where the H2S content ranged from 28% to 46% and CO2 from 3% to 8%. Figure 3.1 shows a reproduction of Hamby’s pit morphology and sequence of corrosion products that formed on pitted tubing in these wells. The presence of a thin layer of iron chloride at the bottom of the pit has been found to increase the rate of pit penetration significantly. Ho-Chung-Qui and Williamson [4] found the same sequence of corrosion products and severe pitting in flowlines that carried H2S with water that contained chlorides at 65 000 ppm. There does not appear to be any laboratory studies of H2S corrosion in very high chloride environments (>100 000 ppm) which would be of great importance since the acidizing of wells with HCl or the use of brine-water, kill fluids will likely cause the sort of severe pitting noted in Fig. 3.1. The presence of a thin iron chloride layer at the bottom of pits in H2S environments has been noted by Kasnick and Engen [5] and the stability of this compound was established by Kesavan and Wilde [6]. FeCl2 is stable at pH values of 3 or less, confirming the very low pH present at the bottom of these pits. Oxygen contamination in sour systems is a significant problem from a corrosion standpoint. Mechanistically, there have been numerous studies, all of which generally agree on the reactions that occur. For example, Craig [7] demonstrated that, in H2S saturated solutions at ambient temperature, the sulfide that forms is mackinawite (Fe9S8) but with the introduction of oxygen into the test, or by simply exposing the Fe9S8 corrosion product to air, it was oxidized to γ-FeO(OH) and then to Fe3O4. The S was liberated as either elemental S or a polysulfide. The resulting corrosion rates were not particularly high, 0.19 mm/year, but the time was not sufficient to allow pitting which may be the more likely outcome under these conditions. While some in the industry understand the oxidation path of iron sulfides, it is remarkable how many researchers mistakenly assume that the presence of iron oxides in their corrosion analysis is a normal product when in fact it signifies that their

48

Sulphur-assisted corrosion in nuclear disposal systems

laboratory procedures were unsatisfactory allowing oxygen entry into the system during testing. This of course nullifies the test results. It is, therefore, extremely important to remove all oxygen in a test system when testing in H2S and that this complete elimination of oxygen be maintained for the entire test, otherwise the results are meaningless. Not only does oxygen change the corrosion products in sour systems but it has a significant effect on the corrosion rate. Almost all of the literature on the subject shows a corrosion rate for steel in aerated wet H2S of less than 1 mm/year and most often about 0.5 mm/year. However, these rates can be misleading, since they are uniform or average corrosion rates, and do not highlight the important fact that oxygen in the system can induce severe localized attack in the form of pitting and hydrogen induced cracking (HIC). Several important field failures have been caused by this problem [8,9]. Recently Hausler [10] showed that when 800 ppm O2 (equal to 16 ppb in the water phase) was bled into an H2S purge gas during corrosion testing of steels, the corrosion rate jumped from 0.68 to 2.8 mm/year. If the oxygen concentration was increased to 2000 ppm in the H2S stream, the onset of rapid localized corrosion occurred, particularly in the presence of a high concentration of chloride ions. The effect of oxygen contamination on the corrosion products and their stability is one aspect but oxygen will also lead to changes in the solution chemistry. For example, Crolet et al. [11] indicate that trace amounts of oxygen in a sour system will produce thiosulfate. This will not be important to the overall corrosivity of the environment but, in the absence of H2S, the role of thiosulfate can be significant to the overall corrosion process. Furthermore, in the gas phase the following reaction can occur: H2S + O2 → S + H2O The generation of S and water can have catastrophic effects on corrosion and hydrogen cracking of steels. Until quite recently, laboratory test work was hampered by the lack of facilities for replenishing gases in the test cell, so reducing corrosion rates are noted as the H2S and CO2 charged in the autoclave was consumed in the reactions. This limits the value of much laboratory test work. Knowing how difficult it is to carry out really well-controlled experiments in H2S systems, with full exclusion of air and sufficient replenishment of gases which are consumed in the corrosion reaction, there is a lot of doubt about the validity of many of the test results quoted. It is expected that tests with once-through, flowing systems, more typical of real service conditions, would show different results. Some examples of these corrosion rates are noted below. In general, the corrosion rates in systems that only contain H2S are relatively low, in the order of 0.5 mm/year or less. From the work of Shoesmith et al. [12] at 1 bar H2S, the corrosion rate at pH 4 was 0.31 mm/year and at pH 5 was 0.11 mm/year. These results are consistent with more recent work of Cheng et al. [13] that found the same rates of corrosion at similar pH. Thomason [14] found corrosion rates of steel in H2S that never exceeded 0.8 mm/ year. Likewise, Lino et al. [15] observed rates of 0.46 mm/year at pH 5.2 and 0.65 mm/ year at pH 4. Shannon and Boggs [16] performed corrosion tests with H2S/N2 mixtures. They noted different scales formed with different concentrations of H2S and a gradual

Corrosion mechanisms and material performance

49

reduction in corrosion rates with time. Corrosion rates increased in going from distilled to chloride-containing water, up to about 1% NaCl. Above this, the corrosion rate dropped off and above 6% up to 20% NaCl, the corrosion rate remained steady at about 80% of the distilled water value. The first noted increase in corrosion rates when adding chloride ions may reflect the impact of chloride ions in destabilizing the sulfide film. Increasing amounts of chloride ions in solution reduce the solubility of acid gas in solution, effectively reducing the activity of the gas. Thus, from these examples and other reports, it appears that the literature is quite consistent with regard to the corrosion rate of steel in H2S alone (no CO2 or O2): that the rates are generally about 0.5 mm/year or less with some few exceptions up to 1 mm/year. The addition of CO2, however, can significantly affect the corrosion rate. Consideration of the thermodynamics of carbonate and sulfide scaling leads to the relationship: KFeS/FeCO3 = C (aCO2/aH2S) This is the source of the general rule of thumb in the industry that there is a ratio of CO2/H2S below which the equilibrium scale formed on the surface of steel shifts from being carbonate to sulfide. For example, Rhodes [17] suggests that a ratio of CO2/ H2S<500:1 should result in sulfide scales forming and he also suggested that the corrosion rate was independent of the flow rate. More recently, this 500:1 ratio has been revised downwards. For example, Simon-Thomas and Loyless [18] proposed: CO2/H2S >200 Æ CO2 dominated (flow conditions are important in influencing corrosion rates) CO2/H2S <200 Æ H2S dominated (flow is much less critical to scale stability and therefore to corrosion rate) They suggest that the 200 figure may be further revised to 50–100. They state that the corrosion rate is dominated by CO2, but iron sulfides are very stable and cathodic to steel and can therefore tend to initiate localized corrosion, i.e. if there is a little H2S, there is a problem of destabilization of carbonate scales and a risk of pitting with high local corrosion rates (potentially faster than CO2 corrosion rates – driven by the cathodic scale). So, slightly sour systems require careful monitoring. Higher H2S levels result in stable scales and a reduction in corrosion rate as long as the sulfide film can be maintained. This follows normal observations made in the field internationally. 3.1.2

Corrosion rate of carbon steel in presence of sulfur

It is recognized that dry solid S does not produce corrosion of carbon steel. It is only when moisture is present that S induces corrosion. Moreover, the pH of the solution and the presence of chlorides are significant contributors to the corrosion of steels in S-containing environments. Potential–pH diagrams, while useful from a thermodynamic standpoint, are generally for fresh water environments so are not complete without considering the impact of chlorides. Hyne et al. [19] demonstrated that S can be a very potent corrodant depending on pH when chlorides are present. Figure 3.2 shows the effect on corrosion as a function of pH without chlorides and with 1000 ppm chlorides in solution. Figure 3.3 also indicates the effect of pH and chloride content on the corrosion rate in a S-containing environment.

50

Sulphur-assisted corrosion in nuclear disposal systems

3.2 Corrosion rate of steel (mpy) in S as a function of pH and chloride content [19]. Corrosion rate in mm/year is mpy figure divided by 39.37

3.3 Corrosion rate of steel in a S-containing environment as a function of pH and chloride content of the fluid [19]

The corrosion reactions that occur for steel in the presence of wet S are not completely understood and are disputed. The most accepted reaction, according to several investigators, has been published by Macdonald et al. [20]: (x–1)Fe + Sy–1 · S2− + 2H+ → (x–1)FeS + H2S + Sy–x However, this reaction has been criticized by Schmitt [21] since it implies that the cathodic reduction of polysulfide (Sy2−) should occur without direct contact of S with the metal surface which is contradictory to observations.

Corrosion mechanisms and material performance

51

Boden and Maldonado-Zagal [22] proposed the following: 4S + 4H2O → 3H2S + H2SO4 These authors found that the pH dropped to 1.8 when flowers of S were stirred into distilled water. The generation of H2S and sulfuric acid would explain the lower pH and formation of FeS when S is in contact with steel as well as the high corrosion rates. Similar acidification, although stabilizing at higher values, was reported by Fang et al. [23]. While the net result of FeS formation is the same, the difference in corrosion rates between polysulfides and sulfuric acid is substantial. Also, polysulfides lead to localized corrosion, whereas sulfuric acid produces more general corrosion. More work needs to be performed to elucidate this mechanism. Figure 3.4 shows the results of work carried out in the absence of oxygen [24]. There is a significant increase in corrosion rate from S as the chloride content increases above 0.01 mol/l to a maximum of 15 mm/year. As will be shown later in this study, some field failures have demonstrated pitting corrosion rates in excess of 25 mm/year. 3.1.3 Background to autocatalytic mechanism of corrosion of steel by elemental sulfur Corrosion test work Macdonald et al. [20] investigated the corrosion behavior of carbon steel with wet elemental sulfur under both aerobic and anaerobic conditions. They measured corrosion rates by the change in resistance of wires made of mild steel. With time, they note that at pH 5.79, the change in resistance curves with time tend to be concave upwards, thereby indicating that the rate of corrosion increases with time. This behavior strongly suggests autocatalysis by a corrosion product. It appears that this autocatalysis mechanism is more severe under anaerobic conditions than under aerobic conditions.

3.4 Effect of chloride concentration on corrosion of steel in solution of S/NaCl at room temperature 23

52

Sulphur-assisted corrosion in nuclear disposal systems

Visual examination of the wires removed at pH 5.79 showed that they were severely etched and uniformly thinned over the entire length in contact with sulfur. Furthermore, the wires were frequently found to have broken at a point of high stress, where the wires were bent through a small radius. This suggested the operation of a stress-related corrosion phenomenon, in agreement with previous observations for the corrosion of iron in systems containing iron sulfides, H2S and sulfur. However, in lower pH (1.82) and higher pH (9.10) systems, the rate of corrosion is fairly constant or even decreases over the later period beyond 60 days. This sort of behavior would arise in situations where a non-protective but non-catalytic product is formed, or when the product confers corrosion resistance to the underlying metal. The general corrosion rate observed in these systems provides good evidence that the corrosion films are protective to some extent, i.e. the corrosion reactions result in formation of passive films at the extremes of pH. These results were confirmed by the lack of corrosion products (iron oxides or sulfides) in the sulfur water slurries at the end of the experiment at extreme pH values, while there were large amounts of black iron sulfide in the solution of experiments carried out at pH 5.79 under anaerobic conditions. Looking more closely at the pH range 3.88 to 6.14, it can be seen that the rates of corrosion in this range generally increase with time confirming this autocatalytic mechanism. However, there clearly exists an induction period for the onset of catastrophic corrosion, which is dependent upon pH (increasing for pH values above 5) in an anaerobic system. The results carried out with oxygen present show that this reduces the time elapsed before the onset of catastrophic corrosion, particularly if the initial pH of the system is low. These relationships suggest that the induction period involves a non-catalytic corrosion process, which is sensitive to oxygen, but one which will, nevertheless, eventually occur under anaerobic conditions. It is also noted that it was only after the induction time that there was formation of black iron sulfide (mackinawite) and the evolution of H2S in the solution. Tests were made with different sizes of sulfur particles in the suspension. In the anaerobic system, the induction period was found to be strongly dependent upon particle size, such that smaller particles exhibit the shorter induction times. Also the rate of corrosion is greater with decreasing particle size. It was also noted that if direct contact between the steel and the sulfur was prevented, no increase in the corrosion rate was observed, even at pH 5.79. This is in general agreement with the observations of many other workers who indicate that intimate contact of the sulfur with the steel is necessary for it to result in corrosion. Mechanism Any mechanism which is proposed to explain the corrosion of mild steel in wet elemental sulfur must account for the following observations. • • • •

An induction time exists before the onset of catastrophic corrosion. The induction time decreases with increasing initial pH over the range 3.88 to 5.79. The induction time also decreases as the sulfur particle size distribution shifts to smaller values. At the onset of catastrophic corrosion, both H2S and mackinawite are produced.

Corrosion mechanisms and material performance • • • •

53

The catastrophic corrosion process is autocatalytic. The onset of catastrophic corrosion produces a shift in the corrosion potential to a more positive value. Direct contact between steel and sulfur is necessary for catastrophic corrosion to occur (the likelihood of which may be presumed to increase as the sulfur particle size decreases). The pH increases with time.

The following reactions are proposed to explain the observed catastrophic corrosion phenomenon: Sy–1.S2 + 2xH+ + 2(x–1)e− Æ xH2S + Sy–x (cathodic reaction in the presence of FeS) (x–1)Fe Æ (x–1)Fe2+ + 2(x–1) e− (anodic) (x–1)Fe2+ + (x–1)H2S Æ (x–1)FeS + 2(x–1)H+ (anodic) which give the overall corrosion reaction as: (x–1)Fe + Sy–1.S2− +2H+ Æ (x–1)FeS + H2S + Sy–x An additional cathodic reaction, viz H+ + e− Æ Hads Æ ½H2 presumably also occurs, which accounts for the hydrogen embrittlement of steel in aqueous systems containing hydrogen sulfide. In the above scheme, the species Sy–1.S2− is considered to be formed by chemisorption on the surface of a particle of elemental sulfur. It is also possible that the corrosive species involves adsorbed polythionate ions (i.e. O3S.Sx.SO2−), which may be formed by the reaction of elemental sulfur with water. Irrespective of the exact identity of the corrosive species, contact between the steel or steel/FeS and sulfur is therefore necessary for electron exchange to occur, and hence for the reaction to proceed. The above scheme also accounts for the increase in pH and the formation of both H2S and FeS (mackinawite). The mechanism proposed above also explains the observed autocatalysis, if it is assumed that the overall corrosion process is controlled by the cathodic reaction, and that this reaction is catalyzed by mackinawite. Cathodic catalysis is also consistent with the observed positive shift in the corrosion potential at the onset of catastrophic corrosion. Since the corrosion potential is determined by equality of the partial anodic and cathodic currents, then catalysis of the cathodic process is expected to shift the corrosion potential in the positive direction, from E2 to Ev as observed Fig. 3.5. Catalysis of the anodic reaction would be expected to have the reverse effect. The mechanism by which iron sulfides (including mackinawite) catalyze cathodic processes in sulfur-containing systems has not been established although it has been attributed to their good electronic conductivity, low over potential for hydrogen evolution, noble electrode potentials and defect structures [25]. Furthermore, a number of studies have indicated that the composition and protectiveness of iron sulfide scales depends upon the pH of this system, with the least protective (mackinawite) being formed when the pH of the medium lies between 6.5 and 8.8. Outside this range, protective scales of pyrrhotite and/or pyrite are apparently formed and the overall corrosion reaction is subject to anodic control (e.g. by

54

Sulphur-assisted corrosion in nuclear disposal systems

3.5 Schematic current/potential curves for (a) cathodic catalysis and (b) anodic catalysis

ionic diffusion through the sulfide film). This suggests that the induction period may be due to the formation of a protective film on the metal surface which then breaks down at some critical pH value to form non-protective mackinawite. At this point, the reaction becomes cathodically controlled, and consequently is subject to cathodic catalysis. 3.2 Performance of corrosion-resistant alloys in environments containing H2S and S In considering sour gas production conditions, the conservative assumption should be made that there is some chloride ion content in the stream. Even where conditions are gas producing and very little liquid phase is expected, occasional droplets of formation water will be carried in the produced gas and these will generally contain chloride ions. Thus, for natural gas producing conditions, the selection of corrosion-resistant alloys (CRAs) should be made from established limits for stainless steels and nickel alloys. These limits have been made available in publications by Craig [26], reviews

Corrosion mechanisms and material performance

55

Table 3.1 Pitting corrosion of Alloys 825 and 2550 in S and Cl solutions Alloy 825, 2550 825, 2550 825, 2550 825, 2550 825 825, 2550 825, 2550 825, 2550 825, 2550

Temp, °C

NaCl, ppm

S, 1 g/l

Pitting attack

66 66 66 66 93 93 93 93 93

10 000 10 000 50 000 50 000 0 10 000 10 000 50 000 50 000

No Yes No Yes Yes No Yes No Yes

No Yes No Yes No No Yes No Yes

by Gooch and Gunn [27] and as incorporated into Intetech Ltd material selection software, ‘The Electronic Corrosion Engineer’. These limits have been used for materials selection for more than 10 years and are continuously checked against published failure data. To date, there have been no incidences of failures where materials have been used within these limits and so they can be presented with confidence as a reliable guideline for sour service conditions. As is the case for steels, dry S does not attack stainless steels or nickel-based alloys (CRAs) in the temperature range encountered in oil and gas production. 3.2.1

Corrosion of corrosion-resistant alloys in the presence of sulfur

The fact that Cr offers protection from corrosion by S is true only in the absence of chlorides. Table 3.1 shows the effects of chlorides, S and temperature on the pitting of Alloy 825 and Alloy 2550 in an aerated solution [28]. The test duration was 1 month. Few studies have been performed considering the corrosion of CRAs in S alone without chlorides. This is because the major concern in the petroleum industry has been the great potential for stress corrosion cracking (SCC) of CRAs in H2S–S–Cl environments and the need to define those alloys that would be suitable for these environments. A large amount of laboratory evaluation was carried out from the mid-1980s to the mid-1990s evaluating various CRAs in S-containing environments primarily because the industry fully expected high-pressure high-temperature (HPHT) wells in Mobile Bay, Alabama, USA to produce S at some point in their lives. To date, some 15 years after the first well was completed and began producing, none of the numerous wells in that area have ever produced S. Still the output of data was, and is, useful, especially since there are wells around the world that do produce S. Wilken [29] showed the increase in corrosion rate for Alloy 28 and Incoloy 925 compared to steel in a sour gas environment containing various concentrations of S, both as a liquid and a solid (Fig. 3.6). There was no significant difference between the different states of S and the resulting corrosion rate. The corrosion rates of the CRAs were less than 0.1 mm/year when S was present as a solid and less than 0.2 mm/year when it was present as a liquid even at 30 g/l S. The test environment was 10 bar H2S, 10 bar CO2 and 80 bar N2. The test duration was 120 h. It is unclear what the chloride content was, but, probably 5% NaCl.

56

Sulphur-assisted corrosion in nuclear disposal systems

3.6 Corrosion rate of three alloys in a sour environment containing S at two different temperatures. Note that the vertical line denotes the transition from gaseous to solid/liquid sulfur

Ikeda et al. [30] observed a similar relationship for the corrosion rate of SM 2550 to the S content in a 25% NaCl solution at 177°C with 1 MPa H2S and 1 MPa CO2 (Fig. 3.7). There has been a tendency for the industry to use 1 g/l S additions when testing CRAs based on the limiting solubility of S in solution according to the work of Eills and Giggenbach [31] as shown in Fig. 3.8. However, as can be seen in Fig. 3.5, there was a very significant increase in the corrosion rate for all alloys above the 1 g/l S industry standard (note logarithmic corrosion rate scale). Thus, testing with 1 g/l S may give unrealistically low corrosion rate values from what may be experienced in practice. Since the melting point of S8, the most common allotrope, is 112.8–114.6°C, corrosion testing for down-hole applications covers the range of phases from solid to liquid

Corrosion mechanisms and material performance

57

3.7 Effect of S on corrosion rate of SM 2550

3.8 Effect of temperature on the solubility of S in water

to gas, which requires careful attention to the method of adding S to the test environment to ensure S participates in the corrosion tests. As indicated by the preceding discussion, S has the greatest impact on corrosion of alloys when it is in intimate contact with the alloy surface. Thus, considerable effort has been expended in configuring test cells to guarantee contact of the S with the surface. In some cases, this has been achieved by placing a cup that holds the liquid S around the slow strain rate specimens [32].

58

Sulphur-assisted corrosion in nuclear disposal systems

3.2.2 Stress corrosion cracking of corrosion resistant alloys in the presence of sulfur It has been found that not only does S cause serious pitting attack of many CRAs but that it also leads to stress corrosion cracking (SCC). Nickel alloys Figure 3.9 shows the increasing corrosion rate, localized attack and, ultimately, SCC of Alloy 825 in the presence of S as a function of chloride content at 250°C. The more corrosion-resistant Alloy 625 did not crack but did pit in this environment [33]. Similar results were found by Chaung et al. [34] who showed cracking of Alloy 825 with 1000 ppm S, at 149°C, with 25% NaCl. Tests by Craig et al. [35] in 1991 showed that Alloy 825 could pass a slow strain rate test (SSRT) but under less severe conditions (100 ppm S, 5% NaCl in an environment of 62 bar H2S and 107 bar CO2 at a temperature of 218°C). Craig specifically noted that samples were fine-grained and had no twinning and were also resistant to pitting, crevice corrosion and C-ring tests under the same conditions. Extensive testing by Martin et al. [32] using SSRT techniques confirmed the deleterious effect of temperature and chloride content on the SCC of nickel-based alloys exposed to S (Fig. 3.10). The alloys tested were Hastelloy G50 (UNS N06950), Hastelloy G3 (UNS N06985) and SM2550 (UNS N06255). The method of adding S, by filling a cup attached to the SSRT specimen, makes quantifying the S content of the environment impossible. However, the data in Fig. 3.10 are expected to be

3.9 Effect of NaCl concentration on the corrosion and cracking of Alloys 825 and 625 in an environment containing S

Corrosion mechanisms and material performance

59

3.10 SCC resistance of Alloy G50 (UNS N06950), Alloy 2550 (UNS N06255) and Alloy G3 (UNS N06985) as a function of chloride content in an environment containing S

quite conservative and as such, present a good means to select CRAs for S-bearing environments. Craig [36] also presented data on the effect of chlorides and S content on the corrosion rate of Alloy 825 and cracking of Alloy G3 (see Tables 3.2 and 3.3). Cracking was observed in Alloy G3 only at the highest temperature and concentration of chloride ions and elemental S. Because of the good resistance to SCC at high temperatures in S-containing environments with moderate chlorides, G3 type alloys have been widely used in Mobile Bay and other areas of the world with great success where S production has been considered a risk. Using SSRT, Wilhelm and Oldfield [37] determined that Alloy 825 and Alloy G were acceptable in a sour solution (27.6 bar H2S) with 1 g/l S up to 204°C when the chloride content was 90 500 ppm. Hibner and Tassen [38] demonstrated that Alloy 825 failed at 177°C when 0.5% acetic acid was also present with elemental S and the chloride content was raised to 165 000 ppm. Figure 3.11 shows one scheme for ranking CRAs in a sour environment containing 1 g/l S with 0.5% acetic acid and 25% NaCl [39]. Table 3.2 Corrosion rate of unstressed samples of Alloy 825 in sour environments after 30 days H2S, psi 195 195 360 360 360

CO2, psi

Temp, °C

S, g/l

Cl, ppm

Corr. rate, mm/year

335 335 600 600 600

121 121 149 149 177

10 10 10 10 10

1500 50 000 1500 50 000 1500

0 0 0 0 0

60

Sulphur-assisted corrosion in nuclear disposal systems

Table 3.3 SSRT results for Alloy G-3 in 1100 psi H2S, 1900 psi CO2, and deaerated NaCl at the concentrations and temperatures shown Temp (°F)/Cl, ppm

S, g/l

% Tf ratio

% RA ratio

375/10 000 425/10 000 425/100 000 425/150 000

1.2 1.2 10.0 10.0

1.03 0.92 0.81 0.13

1.04 0.95 0.90 0.09

Secondary cracking None/None None/None None/None Yes/Yes

3.11 Suitable alloys for service in environments containing S as a function of alloy composition and temperature

Thus, it can be seen that, in the presence of chlorides and S, there is a limiting temperature for each alloy that is largely related to the Mo content. Even Alloy C276 has a limit of about 275°C. This is a much higher temperature limit than found by Vaughn and Greer [40] under the same conditions. These latter authors set the limit for C276 as 177°C. However, Craig et al. [35] did not observe SCC of C276 when tested to 210°C. Regardless of the exact limit, there does appear to be a limit for the nickel-based alloys. However, the titanium alloys offer the potential for much higher resistance to SCC in the presence of chlorides and S at temperatures above 210°C. Titanium alloys Considerable work has been carried out by RMI Titanium, and some limited work by Sumitomo, with regard to the suitability of Ti alloys for environments containing

Corrosion mechanisms and material performance

61

H2S and S. For high-strength, down-hole tubulars and components, the beta alloys, such as Beta C and Beta III, were initially considered but, more recently, the alpha beta alloys such as Ti–6Al–4V–Ru, Ti–3Al–2.5V–Ru and Ti–6Al–2Sn–4Zr–6Mo have been tested since they are less expensive, are heat-treatable and have better SCC resistance than do the beta alloys. In the same environment as for the C276 above (1 g/l S with 25% NaCl and acetic acid), the Ti alloys were completely resistant to SCC and localized corrosion up to 260°C. In the absence of S and with lower chlorides, they are resistant up to 330°C. Therefore, the Ti alloys represent the next higher level of resistance to corrosion and cracking in S with chlorides above the nickel-based alloys. 3.3 3.3.1

Practical application of steels and CRAs Fields that produce H2S with no sulfur

This section addresses selection of materials for oil and gas wells that contain H2S with no S and H2S with CO2 at levels low enough that CO2/H2S<200 resulting in sulfide scale formation dominating – but this does not necessarily mean that stable sulfide scales will be present if other conditions (pH, chloride, etc.) are unfavorable. Both laboratory and field experience reviewed in this paper indicate that sulfide film breakdown may result in pitting attack with potential for extremely rapid failure. Thus the forms of corrosion control in sour systems are primarily targeted at maintaining stable filming conditions and preventing pit initiation. Across a wide pH range and a considerable variation in H2S concentration (partial pressure), the predominant corrosion product that will form on steels is mackinawite (Fe(1+x)S), that may or may not be protective but will generally not produce a high corrosion rate. At temperatures above about 100°C, pyrrhotite forms (sometimes with a thin cover of pyrite) and is very protective. However, high chlorides can interfere with the protectiveness of mackinawite and pyrrhotite and induce localized pitting. The combined action of H2S and CO2 can influence the composition, stability and protective nature of the corrosion product scale. The formation of iron sulfide corrosion product is known to form a sulfide scale on the surface which can be very protective and persistent and can, therefore, significantly reduce the corrosion rate relative to systems without H2S present. When corrosion does take place, it tends to be localized, with the surrounding sulfide scale acting as a large cathode area. This can give very high corrosion rates. Many of the discrepancies in field experience and the reported influence of H2S in the literature are probably due to the cathodic nature of iron sulfide scales with respect to the steel substrate. Hence, if the scale cracks or spalls when it reaches a critical thickness, high rates of pitting can ensue due to local galvanic effects. Another factor that can also have a similar effect in causing local scale breakdown is the physical removal of iron sulfide corrosion product by mechanical means. The localized removal of sulfide scale would cause a direct galvanic effect with the exposed steel substrate acting as an anode to the surrounding cathodic scale. The resulting high rates of localized attack could continue for considerable periods thereafter due to the difficulty in redeveloping protective scale at these locations. A further potential source of corrosion may be the ingress of oxygen. Traces of oxygen in systems containing H2S may be sufficient to oxidize the H2S and result in

62

Sulphur-assisted corrosion in nuclear disposal systems

free S, which is highly aggressive, and will lead to serious pitting attack as well as hydrogen induced cracking (HIC) or stress-oriented hydrogen induced cracking (SOHIC) in flowlines. Carbon steel will be the main material considered for equipment as long as there is confidence that initiation of pitting corrosion can be prevented or is unlikely to occur (i.e. low chlorides and moderate pH). It is assumed that once pitting corrosion commences, the rate of attack is too rapid for it to be detected and controlled before perforation of the steel has taken place. Thus the whole approach to considering carbon steel shifts from one of trying to estimate a corrosion rate, to one of evaluating the risk of pit initiation. Factors which influence pit initiation in a sulfide-filmed surface: • • • • • • • •

high ratio of CO2/H2S, i.e. CO2/H2S >200 chloride ion concentrations in excess of 50 000 ppm pre-corrosion of surface of steel, particularly prior exposure to chloride containing fluids such as completion brines or hydrochloric acid from acidizing treatment low pH conditions, e.g. from acidizing returns lack of inhibition or low frequency batch inhibition erosive flow oxygen ingress through air entrained in injected chemicals or from mechanical operations damage directly to sulfide layer by inspection equipment.

Consideration of the factors listed above may make it immediately clear that the risks involved in using carbon steel may be rather high, in which case there may be an immediate preference for choosing a CRA material. Other situations may be less clear-cut and there may be scope for using carbon steel with operational procedures carefully controlled to prevent sulfide film breakdown. The risk of pit initiation is important to address, but the consequences of such a pit and likely leak also have to be addressed. When the decision is taken to select a CRA for sour oilfield environments, the choice can generally be made between Alloy 825 and Alloy 625. Both have been used as cladding or lining in flowlines and vessels or as solid piping in facilities. 3.3.2

Fields that produce H2S with sulfur

One of the most difficult problems when S is expected in the produced fluids is accurately predicting its occurrence. If S does not appear but was predicted, the cost of completions can be extraordinarily expensive when lesser alloys would have sufficed. On the other hand, if S occurs unexpectedly, failure can be rapid. Sulfur prediction is still very difficult and not accurate. Alberta Sulfur Research (ASR) has been active in this area for many years and is considered one of the leaders in S precipitation modeling. One of the important factors concerning whether S precipitates is if hydrocarbon condensate is present in the well or not. Field experience has demonstrated that S often precipitates when the production was dry gas and no liquid hydrocarbons were present. Once it is determined, or suspected, that S will be produced and the phase in which the S will be present is known, the following are recommended.

Corrosion mechanisms and material performance

63

All of the considerations of the previous section apply, but in addition, there is the question to address whether the S will actually be in contact with the steel surface. If it is, it can be assumed that perforation of the wall will be rather rapid (Bich and Goerz of Shell Canada quote one case with corrosion rates as high as 30 mm/year [42]). It should also be considered that there is a greater sensitivity towards the presence of chloride ions, so when there are expected to be chloride ions in excess of about 5000 ppm, combined with elemental S production, there should be careful consideration of the preference for CRA material selection. Similarly, the presence of liquid hydrocarbon may be considered beneficial in helping to take and keep S in solution. Systems which do not produce liquid hydrocarbon, but do have elemental S, should be considered particularly aggressive. 3.4

Conclusions

Although there are considerable data in the literature concerning corrosion from H2S and S (and the same is true for field experience), there is essentially no accepted set of guidelines for the selection of materials in these environments. To date most oil companies have based materials selection in these environments solely on their own field experience around the world or that of other operators. As seen in the details of the above report, there are many factors that determine the corrosion rate of steels and CRAs in H2S- and S-bearing environments. In fact, so many factors that simply basing materials selection decisions on experience can easily lead to the wrong choices and in some cases, premature failure. Generally speaking, the issue of material choice and corrosion control approach is strongly influenced by the consideration of the factors that will enhance the stability of the sulfide film, or will tend to destabilize it and encourage pit initiation. The evaluation of all of the influencing factors, based on the review of the literature and best practice from many operators, gives a good basis for evaluation of new producing conditions with aggressive environments. It is considered that it is possible to handle aggressive, high H2S- and S-containing environments safely and with expectation of long service life if care is given to the materials selection at the beginning of a project. References 1. S. N. Smith and J. L. Pacheco, ‘Prediction of corrosion in slightly sour environments’, in NACE CORROSION 2002, Paper 224. 2. N. Sridhar, D. S. Dunn, A. M. Anderko, M. M. Lencka and H. U. Schutt, Corrosion, 57 (2001), 221. 3. T. W. Hamby, J. Petrol. Technol., 33(5) (1981), 792–798. 4. D. F. Ho-Chung-Qui and A. I. Williamson, ‘Corrosion experiences and inhibition practices in wet sour gas gathering systems’, in Corrosion 87, Paper No. 46, NACE 1987. 5. M. A. Kasnick and R. J. Engen, ‘Iron sulfide scaling and associated corrosion in Saudi Arabian Khuff Gas Wells’, in SPE Middle East Oil Tech. Conf., Paper 17933, SPE, Bahrain, 1989. 6. S. Kesavan and B. E. Wilde, Corrosion, 46 (1990), 19. 7. B. D. Craig, Corrosion, 35 (1979), 136. 8. M. G. Hay and M. D. Stead, ‘The hydrogen induced cracking failure of a seamless sour gas pipeline’, in NACE Western Canada Conference, Calgary, Canada, 1994.

64

Sulphur-assisted corrosion in nuclear disposal systems

9. B. D. Craig, T. V. Bruno and W. M. Buehler, ‘Complexities of failure analyses in sour systems’, in CWS/ASM International Conference on Sour Service and Low Temperature Material and Welding Properties, Calgary, Canada, November 1997. 10. R. H. Hausler, ‘Contribution to the understanding of H2S corrosion’, in Corrosion 2004, Paper No. 04732, 2004. 11. J.-L. Crolet, M. Pourbaix and A. Pourbaix, ‘The role of trace amounts of oxygen on the corrosivity of H2S media’, in Corrosion’91, Paper No. 22, NACE, 1991. 12. D. W. Shoesmith, P. Taylor. M. G. Bailey and D. G. Owen, J. Electrochem. Soc. (1980), 1007. 13. X. L. Cheng, H. Y. Ma, J. P. Zhang, X. Chen, S. Chen and H. Q. Yang, Corrosion, 54 (1998), 369. 14. W. H. Thomason, ‘Formation rates of protective iron sulfide films on mild steel in H2S saturated brine as a function of temperature’, in Corrosion ’78, Paper No. 41, NACE, 1978. 15. M. Lino, N. Nomura, H. Takezawa and T. Takeda, ‘Engineering solutions to the H2S problem in linepipes’, in Current Solutions to Hydrogen Problems in Steels, 159. Proc. Int. Conf. on Current Solutions to Hydrogen Problems in Steels, ASM Int., 1982. 16. D. W. Shannon and J. E. Boggs, Corrosion, 15 (1959), 209–302. 17. P. R. Rhodes, ‘Corrosion mechanism of carbon steel in aqueous H2S solutions’, The Electrochemical Society, Inc, held in Las Vegas, Nevada, Fall 1976. 18. M. J. J. Simon-Thomas and J. C. Loyless, ‘CO2 corrosion in gas lifted oil production correlations of predications and field experience’, Corrosion/93, Paper No.79, 1993. 19. J. B. Hyne, C. L. Labine and N. I. Dowling, ‘Corrosion of steel by wet sulfur and its mitigation’, in Symposium on Effects of Hydrogen Sulfide on Steel, 22nd Annual Conf. of Metallurgists, CIMM, Canada, 1983. 20. D. D. Macdonald, B. Roberts and J. B. Hyne, Corros. Sci., 18 (1978), 411. 21. G. Schmitt, Corrosion, 47 (1991), 285. 22. P. J. Boden and S. B. Maldonado-Zagal, Br. Corros. J., (1982), 17(3), 116–120. 23. H. Fang, D. Young and S. NešiÇ, ‘Corrosion of mild steel in the presence of elemental sulfur’, NACE CORROSION 2008, Paper 08637. 24. W. Kuster, H. Schlerkmann, G. Schmitt, W. Schwenk and D. Steinmetz, Werkst. Korros., 35 (1984), 556. 25. J. S. Smith and J. D. A. Miller, Br. Corros. J., 10 (1975), 136. 26. B. D. Craig, Corrosion Resistant Alloys in the Oil and Gas Industry, NiDI Technical Series Publication 10073, Updated 2000. 27. T. G. Gooch and R. N. Gunn, ‘Alloy materials for sour service environments – a critical review’, TWI joint industry sponsored project, June 1992. 28. S. E. Mahmoud, H. E. Chaung and C. W. Petersen, ‘Localized corrosion of corrosionresistant alloys in sulfur-chloride-containing environments’, in Corrosion ’90, Paper No. 70, NACE, 1990. 29. G. Wilken, ‘Effect of environmental factors on downhole sour gas corrosion’, Corrosion ’96, Paper No. 76, NACE, 1996. 30. A. Ikeda, et al. ‘On the evaluation methods of Ni-base corrosion resistant alloy for sour gas exploration and production’, in Corrosion ’88, Paper No. 65, NACE, 1988. 31. A. J. Eills and W. Giggenbach, Geochim. Acta, 35 (1971), 247. 32. C. J. Martin, H. S. Ahluwalia, F. Blanchard and J. Skogsberg, ‘The use of slow strain rate testing to determine the effect of elemental sulfur on stress corrosion cracking of nickel based alloys’, in Corrosion ’94, Paper No. 70, NACE, 1994. 33. A. Miyasaka, K. Denpo and H. Ogawa, Corrosion, 45 (1989), 771. 34. H. E. Chaung, M. Watkins and G. A. Vaughn, ‘Stress corrosion cracking resistance of stainless alloys in sour environments’, in Corrosion ’85, Paper 227, 1985. 35. B. D. Craig, J. C. Collins, R. L. Patrick and T. Gilbert, ‘Testing and evaluation of corrosion-resistant alloys’, in Materials Selection and Design, 51–55. December 1991.

Corrosion mechanisms and material performance

65

36. B. D. Craig, ‘The significant effect of chlorides on the selection and application of CRAs’, in 9th Middle East Corrosion Conference, NACE, Bahrain, February 2001. 37. S. M. Wilhelm and J. W. Oldfield, Effects of Elemental Sulfur on the Performance of Alloy 825 in Deep Sour Gas Well Production, Cortest Labs Report, UK, 1987. 38. E. L. Hibner and C. S. Tassen, ‘Corrosion resistant oil country tubular goods and completion alloys for moderately sour gas service’, Eurocorr 2000. 39. Special Metals Bulletin, 2002. 40. G. A. Vaughn and J. B. Greer, ‘High strength nickel alloy tubulars for deep sour gas well applications’, in SPE 55th Annual Technical Meeting, Paper No. 9240, 1980. 41. N. Sridhar and S. M. Corey, ‘The effect of elemental sulfur on stress cracking of nickel base alloys’, in Corrosion ’89, Paper No. 12, NACE, 1989. 42. N. N. Bich and K. Goerz, ‘Caroline pipeline failure: findings on corrosion mechanisms in wet sour gas systems containing significant CO2’, in NACE Corrosion ’96, Paper 26, 1996.

4 Lifetime prediction of metallic barriers in nuclear waste disposal systems: overview and open issues related to sulphur-assisted corrosion Damien Féron Commissariat à l’Energie Atomique, Direction de l’Energie Nucléaire, DPC/SCCME, Bâiment. 458, PC50, 91191 Gif-Sur-Yvette, France

4.1

Introduction

The generally accepted strategy for dealing with High Level Nuclear Waste (HLNW) is deep underground burial in stable geological formations. The multi-barrier concept, which involves the use of several natural and engineered barriers to retard and/or to prevent the transport of radio-nuclides into the biosphere, is applied in all geological repositories over the world. The extraordinarily long time frame (several tens or hundreds of thousands years) involved in disposing of high-level nuclear waste leads to many questions about the ability of present knowledge in materials science and corrosion to anticipate waste package behaviour in the distant future. These issues have already been discussed, compared, and explored with the corrosion community which has to face new challenges for corrosion prediction over millenniums on a scientific and technical basis. During specific workshops, the scientific and experimental approaches have been compared between various organisations worldwide for predicting long-term corrosion phenomena, including corrosion strategies for interim storages and geological disposals [1–3]. From these exchanges, it appears that strategies to predict life times over millenniums include four main steps: Experimentation and analyses are performed in laboratories to obtain preliminary and first data for design and to identify possible corrosion mechanisms. Modelling and simulations are performed based on semi-empirical models or on a more theoretical mechanistic approach which includes a phenomenological description of corrosion phenomena. The objectives are to validate mechanisms when compared to experimental data and to obtain long-term extrapolation. Validations include integrated experiments where coupling of events and interactions are taken into account. In situ experiments, in repository and representative conditions, are an example of these types of validation. The study of historical and archaeological artefacts gives access to far longer periods and may serve to validate average corrosion rates and the understanding of the corrosion mechanisms. This approach is iterative and allows progressively orienting the choices toward solutions offering the greatest robustness with respect to the evolution of knowledge. During these iterations, some processes which have not been considered may be included. This could be the case for sulphur species and sulphur assisted corrosion which up to now, have not really been taken into account for lifetime prediction of 66

Lifetime prediction of metallic barriers in nuclear waste disposal systems

67

metal and alloys in underground repository conditions. As corrosion is an interaction between materials and media, sulphur assisted corrosion includes sulphur species which may be: in the material itself: we will focus on steels and nickel alloys in which sulphur inclusions are known to be important for localised corrosion phenomena, and dissolved sulphur in the alloy may also play a role with regard to the corrosion behaviour of passive alloys in the media which include, in the case of geological disposal systems, the soil itself (sulphur solid species such as pyrite for instance) and the water contained in the more or less porous soil, with numerous soluble sulphur species, the oxidation degrees of which are from −2 (sulphides) to +7 (sulphates). As very often when two main parameters are involved (material and media here), when a third parameter occurs, it may cause trouble: in sulphur assisted corrosion, this third element could be stress, bacteria or irradiation, as far as nuclear wastes are concerned. These interactions are summarised in Fig. 4.1. This paper, based on literature data, will give an overview of the influences of sulphur (in the metal or in the environment) on the corrosion behaviour of steels (mainly) and nickel alloys in conditions encountered in geological storage. It will include firstly the importance of sulphur inclusions and dissolved aqueous sulphur in the alloys with regard to the long corrosion behaviour. Emphasis will then be given to the role played by dissolved sulphur species on the corrosion behaviour with and without bacteria. Some data coming from integrated experiments (short term) and from iron archaeological artefacts (long term behaviour) will also be analysed. 4.2

Material: sulphur in the alloys

Today, it is possible to elaborate alloys with very low sulphur content (below 0.005%). Nevertheless, the sulphur content in most commercial alloys is generally much higher

4.1 Sulphur-assisted corrosion: schematic diagram of parameters and their interactions

68

Sulphur-assisted corrosion in nuclear disposal systems

4.2 AFM images of a MnS inclusion: (a) and (c) before activation, (b) and (d) after activation, (e) same as ‘d’ but after ultrasonic rinsing (to remove the precipitates inside the cavity formed by the inclusion dissolution) – from Ref. 5

due to the manufacturing properties: for instance, the machining of stainless steels generally needs between 0.02% and 0.025% sulphur. Below 0.01%, the stainless steel is claimed to be difficult to machine while sulphur is added up to 0.035% when specific machining properties are needed. On the other hand, manganese sulphide (MnS) inclusions are known to be precursor sites for localised corrosion phenomena and are clearly involved in the initiation of pitting. The pitting of stainless steels which occurs in environments containing chloride usually initiates at manganese sulphide inclusions. Using atomic force micrographs on stainless steels, it has been shown that pits are initiated at the edge of the inclusions, in the metallic matrix side (Fig. 4.2). The scanning vibrating electrode technique, used to map the current distribution over an activated MnS inclusion, has revealed anodic zones around the inclusion, whereas the inclusion itself was cathodically polarised. The anodic current is clearly ascribed to the breakdown of passivity induced by adsorbed sulphur coming from MnS dissolution, whereas two hypotheses are postulated for the origin of cathodic current: the reduction of the Mn2+ or the reduction of Fe3+ into, respectively Mn0 and Fe2+ [5]. Non-metallic inclusions and particularly manganese sulphides are initiation sites for hydrogen induced cracking (HIC) and are particularly harmful to high-strength alloys. For instance, high-strength steels used in environments where HIC may occur

Lifetime prediction of metallic barriers in nuclear waste disposal systems

69

4.3 Tritium accumulation around MnS inclusion – from Ref. 6

have very low sulphur content, often lower than 0.001%. Atomic hydrogen which diffuses inside the alloy is recombined into molecular hydrogen on manganese sulphide inclusions and very high hydrogen pressure may be reached which leads to the rupture of the material. Illustration of the accumulation of hydrogen on MnS inclusions is given in Fig. 4.3. It could also be noted that the form of the inclusions is important: long and large MnS inclusions are more harmful than round and small ones [6]. In aqueous media where the hydrogen activity is high (for example, in acid solutions with hydrogen sulphide, H2S, or under aqueous conditions with a strong cathodic protection), some steels may crack without applied stresses: the cracks start in areas where inclusions occur. MnS inclusions are considered as the worst type of inclusion and lead to blistering. With applied stresses, the cracks may link blisters and lead to what is called stress oriented hydrogen induced cracking (SOHIC), as shown in Fig. 4.4.

4.4 Internal cracks in low alloyed steel exposed to hydrogen sulphide; (a) blister at a manganese sulphide inclusion; (b) SOHIC crack – from Ref. 7

70

Sulphur-assisted corrosion in nuclear disposal systems

Sulphur inclusions are also involved in the propagation of stress corrosion cracks: in water at high temperature, the increase in the crack growth rate of low alloyed steels is explained generally by the dissolution of the sulphur inclusions which are encountered by the corrosion-fatigue cracks. The dissolved sulphur pollutes the crack media and increases the anodic dissolution rate [7]. For the prediction of the long-term behaviour of alloys, the dissolved sulphur in the alloys (and not only the sulphur inclusions) may also play a role. It has been shown that dissolved sulphur in the bulk of nickel or nickel alloys accumulates on the surface during anodic dissolution of these materials and leads to the formation of a sulphur film which is non-protective and even prevents the passivation [8]. It is not clear if the same behaviour is obtained with active dissolution of steels or of copper or copper alloys. In the passive state, at the alloy–passive film interface, will anodic segregation of sulphur also occur? Slow dissolution and formation of the passive films occur continuously. For passive films mainly composed of chromium oxides and hydroxides, the behaviour of bulk impurities is still under investigation. For sulphur in nickel and nickel–iron alloy monocrystals, investigations have been made with radioactive sulphur: the results gave direct evidence of sulphur accumulation at the interface of the metal and the passive film. The sulphur enrichment was shown to be proportional to the sulphur content in the metal, among other parameters. Above a threshold concentration which is close to one monolayer of sulphur, breakdown of the passive film was observed. After passive film rupture, the presence of sulphur on the surface hinders the repassivation [8]. A schematic illustration of the breakdown of the passive film by sulphur accumulation is shown in Fig. 4.5. With nickel base alloys containing chromium and doped with sulphur, the alloy (Ni–21Cr–8Fe (at.%) with 0.009 at.% S) is passivated even with sulphur at the metal–passive film interface, which shows that chromium strongly counteracts the detrimental effect of sulphur. Evidence has been produced that there is competition between the growth on the metal surface of nickel sulphide and chromium oxide. The coverage of the surface by nickel sulphide remains low enough (16%) to allow the passive film to grow and cover the whole surface [10], as illustrated in Fig. 4.6. If we assume that all the dissolved sulphur contained in an industrial alloy (stainless steel or nickel alloy) accumulates at the interface between the alloy and the passive film, the time required to accumulate a mono-layer of sulphur is about 50 years (sulphur concentration 0.001%, passive current density 0.01 μA/cm2). The question remains whether the most probable scenario is given by Fig. 4.5 or by Fig. 4.6.

4.5 Schematic diagram of the mechanism of anodic segregation of sulphur at the metal–passive film interface and breakdown of the passive film (nickel or nickel–iron alloys) – from Ref. 9

Lifetime prediction of metallic barriers in nuclear waste disposal systems

71

4.6 Schematic diagram of the mechanism of the antagonistic roles of chromium and sulphur on the behaviour of the passive film of a Fe–Cr nickel base alloy – from Ref. 10

To summarise, sulphur is dissolved and/or may precipitate and form inclusions (mainly MnS) in the alloys. In both cases, the corrosion behaviour is influenced by the following: Sulphur inclusions may be involved in pitting (initiation) phenomena, in hydrogen induced cracking (initiation) and in stress corrosion cracking (propagation). These issues are quite well known today and care has to be taken with regard to the material specification (sulphur content, heat treatments, etc.). Dissolved sulphur may influence the long-term corrosion behaviour of nonpassive and passive materials. This is the case on nickel and nickel–iron alloys. For steels and alloys containing chromium, more investigations are needed to understand the influence of dissolved sulphur either on the general corrosion rate and/or on the passivity of these materials over a long period. More generally, the future of dissolved impurities in alloys is a major issue for the long-term behaviour of passive materials and passive films in the context of high level nuclear waste repositories. 4.3

Media: sulphur species in aqueous solutions

The speciation of sulphur in aqueous solution is very complex as the degree of oxidation may vary between −2 (S2−, sulphide) to +7 (SO42−, sulphate). Following the redox potential determined by oxidising species (oxygen, H2O2, etc.) or the reducing species (H2, etc.) in the solution, the degree of oxidation of soluble sulphur species will change. If some of the equilibrium constants are well known, some others are still missing. Thus, further work is needed on the speciation of sulphur species in aqueous solutions. In underground nuclear waste repository conditions, these evolutions of the degree of oxidation of sulphur species may also be affected by radiolysis. Very few data are available with regard to the effect of radiation on the evolution of dissolved sulphur species. These effects have to be investigated and progress is needed in conjunction with a better knowledge of the aqueous chemistry of sulphur.

72

Sulphur-assisted corrosion in nuclear disposal systems

Pyrite (FeS2) is one of the most common minerals and has been found in clays where underground repositories are planned for nuclear wastes: for instance, pyrite has been found in the Boom Clay formation (Belgium) and in the Callovo-Oxfordian formation (Bure, France). During excavation, oxidation of pyrite will occur according to FeS2 + 7/2O2 + H2O → Fe2+ + 2SO42− + 2H+ and leads to the release of two moles of H+ per mole of oxidising pyrite and so to the acidification of the environment. The acidification can be further enhanced by the oxidation of iron according to FeS2 + 15/4O2 + 7/2H2O → Fe(OH)3(s) + 2SO42− + 4H+ Ferric iron, produced by the above reaction, is also known to be a strong oxidant of pyrite under acidic conditions. Pyrite oxidative dissolution is a function of pH and no consensus has yet emerged on a well-established oxidation mechanism. Under acidic conditions (pH<3), the formation of S0 has been reported [11], while in carbonated aqueous solutions, with carbonate concentrations not far from those observed in the interstitial solution of Boom clay, thiosulphates (SO32− and S2O32−) are the major sulphur soluble compounds. Under these carbonated conditions, pyrite oxidation seems to be limited to the following reaction FeS2 + 3/2O2 = Fe2+ + S2O32− while the ferrous cation leads to the formation of siderite (FeCO3) which is further oxidised to goethite and lepidocrocite [12]. The rate of pyrite oxidative dissolution may also be increased by bacteria (Ferrobacillius ferro-oxidans, Thiobacillus thiooxidans, Acidithiobacillus ferrooxidans, etc.). The pyrite bio-oxidation processes have been widely studied to determine the kinetics of the reactions and the composition of dissolved product sulphur and iron species [13,14]. As far as corrosion of metallic materials is concerned, under aerobic conditions, the (bio-)oxidative dissolution of pyrite leads to acidic environments which may be as low as pH 1. Metals and alloys will then be exposed locally to acid corrosion and are generally not designed to sustain very low pH. However, this aerobic period is generally considered in performance evaluation of the metallic material behaviour. Nevertheless, more detailed considerations are needed as there will be a period of one to several decades while the repository is under construction or operation and during which pyrite oxidation will occur and very low pH may be reached locally. After the initial aerobic period which occurs during the construction and operational phase of the disposal, and after the closure of the system, the disposal conditions will evolve towards anaerobic conditions. Under anaerobic conditions, sulphide (S2−) and probably other sulphur compounds are dissolved in the pore water coming from the dissolution of pyrite and/or the reduction of oxidised sulphur species (for example, SO42−) by bacteria (for example, sulphate reducing bacteria). Aqueous dissolved sulphides are known to have a negative effect on the behaviour of passive layers. One of the most common explanations is linked to the competitive adsorption of OH− and HS− and the subsequent charge transfer and proton transfer processes involving the adsorbed OH and HS species. The former leads to the passivation of the metal by an oxide layer while the second species (HS) leads to a poorly protective layer [15]. Generally speaking, the presence of sulphide decreases the pitting potential

Lifetime prediction of metallic barriers in nuclear waste disposal systems

73

4.7 Polarisation curves for mild steel in deaerated pH 13.4 NaOH solution with added chloride (1 M), malic acid (MA) and different levels of sulphide anion – from Ref. 16

of passive alloys (steels, stainless steels, nickel alloys) and increases the passive current. This is illustrated by the polarisation curves obtained on passive mild steel in a chloride solution at pH 13.4 and shown in Fig. 4.7. The same observations have been made on stainless steels, with a chromium rich oxide layer and, on these materials, the repassivation potential generally also decreased when sulphide concentration increased in the solution. Two main points may summarise the effect of soluble sulphur species on the corrosion behaviour of metals and alloys under geological disposal conditions: •

•

4.4

Under aerobic conditions (construction and operational phases of deep geological disposal), the oxidative dissolution of pyrite, which is catalysed by bacteria, may lead to strongly acidic environments. It has to be checked if this may occur in deep geological disposals, as metals and alloys which are planned to be used will not sustain strongly acidic solutions. Under anaerobic conditions, sulphide (S2−) and probably other sulphur compounds will be in aqueous pore fluids (water inside the clay pores) and will lead to a decrease in the passive properties of the steels used. Possible sulphide concentrations (and concentrations of other sulphur species) have to be known to predict the behaviour of alloys. Influence of a third parameter: bacteria

Stainless steels are widely used under de-aerated conditions. The development of sulphate reducing bacteria (SRB) is of concern and sulphide concentrations will increase locally. Results obtained to tackle the issues related to SRBs and stainless steels in anaerobic seawater conditions [17,18] may be useful for geological disposals. Monitoring the corrosion potential and the intensity-potential curves measured in the presence of SRB has shown that the electrochemical behaviour of stainless steels is essentially linked to the sulphide concentration and the pH, more than to the presence of SRB: for example, there is no difference in the free corrosion potential of 316L grade steel in the presence of Desulfovibrio vulgaris or Desulfovibrio gigas for

74

Sulphur-assisted corrosion in nuclear disposal systems

equivalent pH values and sulphide contents in the culture medium. Given the relatively low potentials obtained (−500 to −300 mV/SCE), it is necessary to determine the depassivation and repassivation potentials to find out if the stainless steels are likely to show localised corrosion. However, it is difficult to compare depassivation and repassivation potentials, given the uncertainties with regard to their determination. A statistical experimental method has been developed, which consists of using a minimum of 10 samples for each determination. Since crevice corrosion is the phenomenon of most concern on stainless steels in general, the critical corrosion potential through the crevice effect (or initiation potential of crevice corrosion) constituted the criterion for comparing the different experimental conditions tested with and without bacteria. Additional tests were performed to obtain the cumulative distribution of the breakdown potentials for crevice corrosion of UNS S31603 samples exposed to sterile aerated seawater. The aim of these tests was to appreciate, by comparison, the magnitude of the effect of the local presence of biologically generated sulphides on the probability of crevice corrosion onset. Figure 4.8 shows the cumulative distribution of the breakdown potentials in sterilised aerated seawater and the cumulative distribution already observed in a SRB culture. It can be seen that a very important decrease, close to 400 mV, of the anodic resistance is induced by the activity of SRB bacteria. Other tests were also performed to verify if the cumulative distribution of the breakdown potentials for crevice corrosion changes if a SRB culture is substituted with de-aerated seawater in which suitable doses of Na2S and HCl are added in such a way that the same total sulphide concentration and the same pH of the SRB culture is obtained. Figure 4.9 shows the comparison of the cumulative distribution of the breakdown potentials, respectively obtained in a SRB culture and in an Na2S solution; total sulphide concentration was close to 400 ppm and pH was close to 7 in both solutions. It can be seen that no appreciable differences are observed between the two cumulative distributions. From cumulative distributions, the most frequently observed breakdown potential and the standard deviation can be calculated. Table 4.1 summarises these values measured on UNS S31603, UNS S31254 and UNS S31803 exposed to sulphides

4.8 Cumulative distributions of breakdown potentials for UNS S31603 in SRB culture and in aerated sterile seawater – from Ref. 17

Lifetime prediction of metallic barriers in nuclear waste disposal systems

75

4.9 Cumulative distributions of breakdown potentials for UNS S31603 in SRB culture and in deaerated seawater with added Na2S – from Ref. 17

Table 4.1 Sulphide source SRB Na2S

Breakdown potentials of stainless steels under anaerobic conditions Sulphide conc., ppm

pH

UNS S31603, mV (SCE)

UNS S31254, mV (SCE)

UNS S31803, mV (SCE)

400–450 400

7.1 7.0

–240±15 –210±15

+40±50 +20±50

+60±100 +20±130

UNS S31603: 17Cr–11Ni–2Mo, UNS S31254: 20Cr–18Ni–6Mo, UNS S31803: 22Cr–6Ni– 3Mo – wt%.

produced by SRB or added in solution as inorganic sulphides. It can be seen that the same conclusions reached for UNS S31603 are also reached for the other SSs tested in this work: sulphides added as Na2S to sterile de-aerated seawater can well simulate the effect of biologically generated sulphides provided that pH and total sulphide concentrations are equal in the two solutions. These conditions are not so far from those encountered in geological disposal environments. However, usually, the sulphide contents are not as high, and the chloride concentrations are lower. Nevertheless, the effect of SRB on the behaviour of stainless steels can be simulated by a pH value and sulphide contents equivalent to those measured in the presence of SRB. The effect of SRB on the stainless steels is therefore essentially linked to the chemical modifications (pH and presence of sulphides) that SRBs impose on the surrounding environment and that lead to a significant decrease in the initiation potential of the localised corrosion. In geological repository conditions, just after the closure, the conditions will evolve from aerobic to anaerobic. In other words, at the beginning of the post-closure period, zones under aerated conditions and others under anaerobic conditions may be simultaneously present at the surface of metallic materials. In other applications, this is the case for instance when the biofilm has been established for a long time with locations where SRB type anaerobic bacteria have developed, whereas other locations are still aerobic. These mixed conditions are probably the worst conditions for passive materials. In order to explain this, let us schematically summarise the previous results and take the case of a UNS S31603 stainless steel:

76

Sulphur-assisted corrosion in nuclear disposal systems

4.10 Behaviour of UNS S31603 in aerated sterile seawater – from Ref. 18

•

•

•

Under aerated conditions, the corrosion potential of stainless steel is around 0 mV/ECS at the beginning of exposure or if there are no bacteria, whereas its crevice corrosion initiation potential is above +200 mV/ECS: there is little likelihood of crevice corrosion developing, as illustrated in Fig. 4.10. Under aerated conditions, the corrosion potential of the passive stainless steels is raised to +300 mV/SCE due to the aerobic biofilm. The material is therefore at potentials greater than or equal to the crevice corrosion initiation potential and therefore, this type of corrosion develops in natural seawater for instance (Fig. 4.11). Under mixed conditions, the potential of the stainless steel is raised to relatively high values by the aerobic biofilm (the enzymes present in the aerobic zones of

4.11 Behaviour of UNS S31603 in the presence of aerobic biofilm alone (possible situation in the presence of a young biofilm) – from Ref. 18

Lifetime prediction of metallic barriers in nuclear waste disposal systems

77

4.12 Behaviour of UNS S31603 in the presence both aerobic biofilm on cathodic areas and anaerobic biofilm under shielded areas (possible situation in post-closure period of a geological disposal) – from Ref. 18

the biofilm according to the enzymatic model), whereas in the anaerobic zones, sulphide (reduction of sulphates by SRBs) causes the crevice corrosion initiation potential to drop: therefore, the higher the sulphide content, the more easily this type of corrosion will develop (Fig. 4.12). So, mixed conditions (aerated and anaerobic areas on the surface of the same material, at the same time) are even more dangerous for the passivable materials: their corrosion potential is increased by the action of aerobic bacteria, whereas the presence of sulphate reducing bacteria in the anaerobic niches leads to a sharp decrease in the localised corrosion of the passivable alloy. In other words, the cathodic reaction is furthered by the aerobic biofilm, whereas the anodic reaction is affected by the presence of sulphate reducing bacteria. The increase in the rate of the cathodic reaction leads to an increase in the corrosion potential whereas the presence of sulphides leads to a drop in the pitting potential in the zones where sulphate reducing bacteria are present. As far as bacteria are concerned, the mixed conditions (aerated and anaerobic conditions) are probably the more dangerous exposure conditions for the passive alloys and these mixed conditions have to be identified (duration, concentrations of reducing and oxidising species, etc.). At least two other parameters have also to be evaluated in the presence of sulphur: stress corrosion cracking and radiolysis. Even though some studies have tackled these subjects, the quantitative evaluation of the impact of soluble aqueous sulphur species is still lacking when stress is applied and under irradiation. 4.5

Integrated experiments and archaeological artefacts

As claimed in the introduction, strategies to predict lifetimes over millenniums include not only laboratory experiments and modelling, but also integrated experiments and archaeological artefacts. This is of course also valid for sulphur-assisted

78

Sulphur-assisted corrosion in nuclear disposal systems

corrosion. Two main results have been published with regard to sulphur-assisted corrosion issues. The first publication [19] is related to an in situ test carried out in Grimsel (Switzerland). Two heaters, simulating the canister and the heat generated, were installed inside the guide tubes or liners and surrounded by highly compacted bentonite blocks. Coupons of several candidate metals for manufacturing HLW containers were introduced into these bentonite blocks, as well as sensors to monitor different physicochemical parameters during the test. After 6 years, several corrosion coupons and sensors were extracted. Corrosion coupons were made of carbon steel and stainless steel, titanium, copper and cupro-nickel alloys. Results obtained in the study indicate a slight generalised corrosion of corrosion coupons. The sensors, made of stainless steel (17Cr–11Ni–2Mo) show, however, important corrosion damage (Fig. 4.13). The sulphur-rich corrosion products and the presence of sulphate reducing bacteria (SRB) in the bentonite covering the sensors indicate a corrosion phenomenon induced by bacteria. An interesting fact is that no bacteria are found in the bentonite near the coupons while aerobic bacteria and sulphate reducing bacteria are found in the bentonite near the damaged sensors. The most interesting fact is that, in the presence of pure culture of sulphate reducing bacteria, nearly no damage is observed on stainless steels during laboratory experiments performed to reproduce the observed degradation on sensors [20]. This is probably an illustration that the worst conditions are those when aerobic and anaerobic zones are encountered together, as both aerobic and anaerobic bacteria have been observed near the damaged sensors. The long-term corrosion of iron artefacts in waterlogged soil at the archaeological site of Nydam in Denmark has been widely studied [21]. In brief, the analyses of the environment show that the conditions are anoxic, slightly acidic (pH 6–7), with no oxygen (below the detection limit, i.e. below 0.1 mg/L) and some sulphide (maximum 2.2 mg/L, a mean value of 0.18 mg of S2−/L). Sulphate reducing bacteria are also present. Corrosion rate estimations have been made. For instance, on 151 lances, corrosion depths as low as 50 μm have been found (average of 20 measurements on a lance). At 1700 years of age this corresponds to an average corrosion rate of 0.03 μm/ year. The average corrosion depth for the 151 lances was 300 μm or 0.2 μm/year. These corrosion rates are among the lowest found on artefacts even compared with aerobic conditions. It seems that the anaerobic conditions even with the presence of sulphide and sulphate reducing bacteria are not corrosive and may even preserve the iron artefacts.

4.13 Damage observed on a stainless steel sensor after 6 years of in situ experiment in bentonite – from Ref. 19

Lifetime prediction of metallic barriers in nuclear waste disposal systems 4.6

79

Conclusions

From this short overview of literature data, it is clear that sulphur (in the metal or in the environment) will influence the corrosion behaviour of metals and alloys under conditions encountered in geological storage. The presence of sulphur inclusions in alloys is well known to initiate localised corrosion phenomena (pitting, SCC, HIC) on passive alloys and mild steels: these phenomena occur rather at the beginning of exposure. For long-term exposure, even low levels of sulphur in the metallic materials (nickel or iron–nickel alloys) may accumulate in the passive layer and lead to the destruction of the passivity. For alloys containing chromium, the sulphur coming from the metal does not lead to the passivity breakdown, due to passivity of chromium oxides. This result has probably to be confirmed for various chromium contents in the alloy and over a long time. The effect of sulphur species in the environment on the corrosion behaviour of metals and alloys under geological disposal conditions has been divided into two main categories: •

•

In aerobic conditions (construction and operational phases of deep geological disposal), the oxidative dissolution of pyrite, which is catalysed by bacteria, may lead to strongly acidic environments. It has to be checked if this may occur in deep geological disposals, as metals and alloys which are planned to be used will be significantly degraded in such environments. Under anaerobic conditions, sulphide (S2−) and probably other sulphur compounds will be in aqueous pore fluids (aqueous solutions inside clay pores) and will lead to a decrease in the passive properties of the steels used. Possible soluble sulphide concentrations (and concentrations of other sulphide species) have to be known to be able to predict the behaviour of alloys.

The role played by sulphate reducing bacteria under anaerobic conditions has also been addressed. On stainless steels and in chlorinated environments, mineral sulphides or sulphides produced by sulphate reducing bacteria have the same effects on the passive layer which is much less resistant with than without sulphides, the worse conditions for metallic passive alloys being the coupling of areas exposed to sulphide (anaerobic) and to oxidising conditions in other parts. The few data coming from one integrated experiment and from archaeological iron artefacts (one site) exposed to sulphide conditions are in agreement with the above analysis of laboratory experiments. Sulphur-assisted corrosion has to be included for lifetime prediction of metallic barriers in nuclear waste disposal systems. The effects of stresses and of radiolysis have not been discussed here and should be integrated in a complete overview. Nevertheless, it seems that sulphur-assisted corrosion will be under control if care is taken in the elaboration of the alloys and if mixed conditions (aerobic and anaerobic) are minimised. References 1. International Workshop ‘Prediction of long term corrosion behaviour in nuclear waste systems’ (Cadarache, France, 26–29 November 2001), Proceedings published in the EFC Series 36, ed. D. Féron and D. D. Macdonald. Maney, London, UK, 2002. 2. Second International Workshop ‘Prediction of long term corrosion behaviour in nuclear waste systems’, during Eurocorr’2004 (Nice, France, September 2004), Proceedings

80

3.

4.

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

18. 19.

20.

21.

Sulphur-assisted corrosion in nuclear disposal systems published in ANDRA Scientific and Technical Series, published by ANDRA, ChatenayMalabry, France, 2005. Third International Workshop on ‘Long-term prediction of corrosion damage in nuclear waste systems’ (University Park, PA, USA, 14–18 May 2007), ed. D. Féron and D. D. Macdonald. J. Nucl. Mater., 379(1–3) (2008). A. A. Sagüés and C. A. W. Di Bella (eds), Proceedings from an International Workshop on Long-Term Passive Behavior (Arlington, Virginia, 19–20 July 2001). U.S. Nuclear Waste Technical Review Board, Arlington, VA, December 2001. B. Vuillemin, X. Philippe, R. Oltra, V. Vignal, L. Coudreuse, L. C. Dufour and E. Finot, Corros. Sci., 45 (2003), 1143–1159. A.-M. Brass and J. Chêne, Mater. Sci. Eng., A242 (1998), 210–221. Corrosion des métaux et alliages, mécanismes et phénomènes; sous la direction de G. Béranger et H. Mazille, Hermes Science publication, Lavoisier, Paris, 2002. P. Marcus and V. Maurice, ‘Passivity of metals and alloys’, in Corrosion and Environmental Degradation, ed. M. Schütze, Wiley-VCH, 2000. P. Marcus and H. Talah, Corros. Sci., 29(4) (1989), 445–463. P. Marcus and J. M. Grimal, Corros. Sci., 31 (1990), 377. M. Descostes, P. Vitorge and C. Beaucaire, Geochim. Cosmochim. Acta, 68(22) (2004), 4559–4569. M. Descostes, P. Vitorge and C. Beaucaire, Geochim. Cosmochim. Acta, 70 (2006), A138. C. Pisapia, M. Chaussidon, C. Mustin and B. Humbert, Geochim. Cosmochim. Acta, 71 (2007), 2474–2490. J. Telegdi, Zs. Keresztes, G. Palink, E. Kalman and W. Sand, Appl. Phys. A, 66 (1998), S639–S642. R. C. Salvarezza, H. A. Videla and A. J. Arvia, Corros. Sci., 22(9) (1982), 815–829. M. Holloway and J. M. Sykes, Corros. Sci., 47 (2005), 3097–3110. D. Féron and A. Mollica, ‘From the mechanisms of the biocorrosion of stainless steels in seawater to corrosion test standards and methods’, in International Conference on Biocorrosion of Materials, BIOCORYS 2007 (Paris, France, 11–14 June 2007). D. Féron, Contribution à l’étude des phénomènes de biocorrosion des matériaux métalliques. Commissariat à l’Energie Atomique, rapport CEA-R-6064, 2004. V. Madina, I. Azkarate and M. Insausti, ‘Corrosion of several components of the in situ test performed in a deep geological granite disposal site’, in Proceedings of the Second International Workshop ‘Prediction of long term corrosion behaviour in nuclear waste systems’, in ANDRA Scientific and Technical Series, published by ANDRA, ChatenayMalabry, France, 2005, 61–67. V. Madina, I. Azkarate, L. Sanchez and M. A. Cunado, ‘Experimental investigation on microbial corrosion for repository conditions’, in SACNUC workshop, Brussels, 21–23 October 2008. H. Matthiesena, L. R. Hilbertb, D. Gregorya and B. Sørensena, ‘Long term corrosion of iron at the waterlogged site Nydam in Denmark: studies of environment, archaeological artefacts, and modern analogues’, in Proceedings of the Second International Workshop ‘Prediction of long term corrosion behaviour in nuclear waste systems’, in ANDRA Scientific and Technical Series, published by ANDRA, Chatenay-Malabry, France 2005, 114–127.

5 The anaerobic corrosion of carbon steel and the potential influence of sulphur species Nicholas R. Smart Serco Technical Consulting Services, Culham Science Centre, Abingdon, Oxfordshire, OX14 3DB, UK

5.1

Introduction

In many countries that operate nuclear power plant, the preferred option for disposal of radioactive waste is to adopt the multi-barrier approach to minimise the risk of release of radionuclides to the biosphere. In this system, the waste is initially placed in a metallic container, which after a period of interim surface storage is placed in a deep underground repository and surrounded by a backfill material that is able to retard the transport of radionuclides to the surrounding geosphere and biosphere. Carbon steel is a candidate container material in a number of international concepts [1–23]. One of the concerns in assessing the safety aspects of a geological repository is the possible release of gas from the waste packages, since gas release would provide a means of transporting radioactive gases out of the repository and may damage the engineered barriers, such as the backfill material, if pressurisation were to occur. The major source of gas in a repository is expected to be the anaerobic corrosion of ferrous materials. The issue of gas generation by the anaerobic corrosion of steel has been the subject of extensive research in a number of laboratories and the results of these research activities are briefly reviewed in this paper. There is significant interest in the role that sulphur species in a repository environment may have on the corrosion behaviour of waste package materials and the main purpose of this paper is to consider how sulphur species may interact with the anaerobic corrosion of carbon steel, particularly in relation to the Belgian Supercontainer concept (Fig. 5.1). In this concept [4], the spent fuel or vitrified waste will be packaged in a stainless steel container, which will be surrounded by a carbon steel overpack that will in turn be surrounded by a cementitious buffer material. The whole package will be surrounded by a stainless steel liner and the complete Supercontainer assembly will be placed within tunnels in a Belgian Boom clay geological formation. It is expected that sulphur-based species will be present within the tunnels in the Belgian repository and there is a requirement to determine how these species may affect the anaerobic corrosion behaviour of the carbon steel in the Supercontainer. The carbon steel will be exposed to alkaline porewater in the cementitious buffer, or in the alkaline material that may be used as a filler between the cementitious buffer and the carbon steel overpack. This is a similar situation to the most extensively considered concept for an ILW (Intermediate Level Waste) repository in the UK, where cementitious grout material would be used as the encapsulant material for carbon steel waste and for the repository backfill. After a relatively short period, the interface between the iron and alkaline porewater will become anoxic as oxygen is consumed by aerobic corrosion processes and microbial activity within the tunnel [5]. 81

82

Sulphur-assisted corrosion in nuclear disposal systems

5.1 Top: Longitudinal section of the Supercontainer showing the two canisters contained in the overpack, surrounded by the buffer and enclosed within the liner. Bottom: Cross section of the Supercontainer. The Supercontainer is located in excavated Boom clay, which is supported by the tunnel lining, composed of concrete blocks [4]

There is an extensive database of information about the anaerobic corrosion of iron in near-neutral and alkaline conditions, under sulphur-free abiotic conditions. The possible effects of sulphur species on the anaerobic corrosion of steel have been less thoroughly investigated. This paper (i) provides an overview of the abiotic anaerobic corrosion processes affecting carbon steel in repository environments, (ii) summarises the properties of the sulphur-based species that may be present in the aqueous phase in contact with iron-based waste containers, (iii) reviews the literature data available on the effect of sulphur on the anaerobic corrosion process, and (iv) identifies areas requiring further research.

The anaerobic corrosion of carbon steel and the potential inﬂuence 5.2

83

Mechanisms of anaerobic corrosion of steels

The following two sections summarise the thermodynamic and kinetic aspects of the anaerobic corrosion of iron. 5.2.1

Thermodynamic aspects of anaerobic corrosion

In the absence of oxygen, the corrosion potential for steel in aqueous solutions is determined by the anodic (oxidation) and cathodic (reduction) reactions involved in the corrosion process. The simplified form of the anodic reaction for dissolution of iron is: Fe0 → Fe2+ + 2e–

[5.1]

for which the standard electrode potential is –440±40 mV (versus normal hydrogen electrode, NHE) [6]. However, instead of direct transfer of two electrons as implied, in reality, there is a single electron transfer sequence. One possible reaction sequence, involving two single-electron transfer reactions [7] is: Fe + H2O → FeOHads + H+ + e−

[5.2]

FeOHads → FeOH + e

[5.3]

FeOH + H → Fe + H2O

[5.4]

+

+

+

−

2+

where the subscript ‘ads’ denotes an adsorbed species. Reaction 5.4 is the rate determining step. This sequence was identified as the most probable mechanism as the reaction rate is known to increase with hydroxyl concentration [7]. (Note – the corrosion rate depends on the reaction rate and several other factors.) An alternative to reactions 5.3 and 5.4, involving deprotonation rather than protonation, is: FeOHads → FeO + H+ + e−

[5.5]

The hydrated form of FeO is then Fe(OH)2. Whether or not reaction 5.5 proceeds in preference to reactions 5.3 and 5.4 depends on such factors as pH, temperature and potential. An alternative reaction that could be envisaged for alkaline conditions, following reaction 5.3, would be: FeOH+ + OH− → Fe(OH)2

[5.6]

Regardless of which of the mechanisms is followed, the overall anodic reaction in alkaline conditions is therefore [8]: Fe + 2H2O → Fe(OH)2 + 2H+ + 2e−

[5.7]

Fe(OH)2 is highly insoluble in moderately alkaline conditions and therefore forms a film on the surface of the iron. Depending on the temperature and pH of the surrounding environment, the ferrous hydroxide may transform, via the following reaction, known as the Schikorr reaction, into magnetite, leading to the evolution of additional hydrogen: 3Fe(OH)2 → Fe3O4 + 2H2O + H2↑

[5.8]

This reaction is favoured by increasing temperature because magnetite is thermodynamically more stable than ferrous hydroxide at higher temperatures. This is shown by the fact that iron corroding in deaerated water at 25°C produces an Fe(OH)2 film, but that magnetite is formed at 60°C [9]. The presence of certain impurities, such as nickel, copper and cobalt, either in solution or from the metal, can catalyse the

84

Sulphur-assisted corrosion in nuclear disposal systems

Schikorr reaction [10]. However, even with high amounts of impurities, only 10–20% of the initial oxide is transformed to magnetite. The balancing cathodic reaction is the reduction of water: H2O + e− → OH− + ½H2↑

[5.9]

This latter reaction is referred to as either the water reduction or the hydrogen evolution reaction. The overall reaction for the anaerobic corrosion of iron under moderately alkaline conditions is therefore: Fe + 2H2O → Fe(OH)2 + H2↑

[5.10]

Or, if the Schikorr reaction occurs, the overall reaction is: 3Fe + 4H2O → Fe3O4 + 4H2↑

[5.11]

The predominant iron oxide species and the regions of soluble ion formation are shown in the Pourbaix diagrams for iron, as a function of potential (vs. normal hydrogen electrode, NHE, which is −242 mV vs. saturated calomel electrode, SCE) and pH, in Fig. 5.2 [11]. The two diagrams show the regions of stability for iron in the form of oxides (Fig. 5.2a) or hydroxides (Fig. 5.2b). At very high pH values, a soluble HFeO2− species is predicted on thermodynamic grounds, although there is little evidence of it arising in a cementitious environment, where magnetite is normally the oxide formed on the surface of steel and the region of stability of the HFeO2− ion is small. The diagrams also show the regions of stability of water: at potentials below line (a), water is reduced to hydrogen and hydroxyl ion (equation 5.9), and at potentials above line (b), water is oxidised to oxygen and protons. Since hydrogen is a product of the anaerobic corrosion reaction, it is possible that the hydrogen overpressure could influence the rate of the anaerobic corrosion reaction. The equilibrium pressure depends on the composition of the corrosion product; at 25°C, the equilibrium pressures for Fe(OH)2 and Fe3O4 are 39 atm and 760 atm, respectively [12]. The possibility of the corrosion rate being inhibited by the presence of hydrogen was investigated experimentally during previous work [12], but no inhibition was found. In the initial stages of corrosion, random oxidation of the surface occurs until the first monolayer of corrosion product is complete, after which a process of film thickening occurs by ion transport through the film under the influence of the electric field gradient established in the film. As the film grows, the field strength decreases and the rate of film growth declines. A steady state is reached when the rate of film thickening equals the rate of film dissolution, giving a net constant rate of metal loss [13]. The rest potential of corroding steel is fixed at the point of charge neutrality where the rates of anodic reactions (oxidation, electron producing) and cathodic reactions (reduction, electron consuming) are equal. The electrode potential of carbon steel will therefore be at a value between the equilibrium potentials (Eø) for reactions 5.5, 5.8 and 5.9, which are given, with respect to the normal hydrogen electrode (NHE), by: Eø Fe/Fe(OH)2 = −0.47 − 0.0591 pH

[5.12]

E Fe(OH)2/Fe3O4 = −0.197 − 0.0591 pH

[5.13]

Eø H2O/H2 = 0 − 0.0591 pH − 0.0295 log [PH2]

[5.14]

ø

where PH2 is the partial pressure of hydrogen. The production of hydrogen is thermodynamically not promoted at potentials positive to the equilibrium potential of the water reduction reaction (as determined from equation 5.14).

The anaerobic corrosion of carbon steel and the potential inﬂuence

85

5.2 Potential (E, V vs. NHE)–pH equilibrium diagrams (Pourbaix diagrams) for the system iron–water at 25°C (top) considering as solid substances only Fe, Fe3O4 and Fe2O3 and (bottom) considering as solid substances only Fe, Fe(OH)2 and Fe(OH)3 [11]

86

Sulphur-assisted corrosion in nuclear disposal systems

For a pH within a cement backfill porewater of 12.5, the equilibrium potentials (V vs. NHE) given by equations 5.12, 5.13 and 5.14 are −0.786, −0.936, and −0.737 or −0.619 (for PH2 1 atm and 0.0001 atm), respectively. The maximum overpotential available to drive the corrosion of iron to form Fe(OH)2 in the complete absence of oxygen in alkaline conditions at a pressure of 1 atm hydrogen is therefore less than 50 mV (i.e. the difference between the potentials for the anodic reaction (equation 5.12) and the cathodic reaction 5.14). 5.2.2

Kinetics of anaerobic corrosion

In practice, the kinetics of anaerobic corrosion will be controlled by the rate of film formation, but if this factor is ignored, the maximum possible corrosion rate can be estimated from a consideration of the activation controlled kinetics of reactions 5.1 and 5.9. At pH 12.5 and a temperature of 25°C, the rate of corrosion, in the absence of any film retardation effects, is estimated to be 9.8 μm year−1 at a potential of −824 mV vs. NHE [14]. The most relevant data in assessing the anaerobic corrosion rate of carbon steel in repository environments are those generated over very long periods by hydrogen evolution experiments. Such experiments performed by Grauer and colleagues [15–17] and Kreis and Simpson [18–20] show that the corrosion rate in deaerated groundwater and cementitious porewaters (i.e. predominantly KOH, Ca(OH)2 and NaOH) may fall as low as 0.07 μm year−1 at 20ºC, after an exposure period of several thousand hours. This is a reflection of the very low solubility of iron oxides at high pH and the low concentration of H+ at high pH. The present author has been involved in obtaining anaerobic corrosion rate data for both the UK programme on ILW and the Swedish programme on HLW (High Level Waste). In the UK programme, the main issue is related to the rate of gas generation due to the anaerobic corrosion of carbon steel waste materials, rather than container materials, since most ILW waste containers will be fabricated from 316L stainless steel. The main focus of the work has been on measuring the anaerobic corrosion rate of carbon steel in simulated alkaline porewaters, through monitoring the production of hydrogen using glass gas cells. The methodology for making such measurements and the results obtained are summarised in Ref. 21. If the material is pickled initially to remove the air-formed film there is an initial peak in corrosion rate, which then decreases with time to a low value as a layer of magnetite or ferrous hydroxide builds up. The general form of the curve relating hydrogen evolution rate (or corrosion rate) to time has been observed both in the weakly alkaline conditions that are relevant to the Swedish programme (i.e. in bentonite or bentonite porewater simulant, pH ~8.4 (Fig. 5.3 [22–24]) and in the strongly alkaline conditions that are relevant to the UK cementitious repository concept (Fig. 5.4 [21]). Initially, the corrosion rates in alkaline conditions are higher at higher temperature, but the long-term rates are not very sensitive to temperature. The long-term rates are higher in the weakly alkaline conditions examined in the Swedish programme than in the highly alkaline, cementitious porewater simulants used for the UK programme. A number of Japanese workers [25–28] have also measured the corrosion rates of carbon steel in artificial cementitious porewaters containing up to 5000 ppm chloride and recorded long-term anaerobic corrosion rates of the order of 0.004–0.2 μm year−1, at temperatures up to 45°C, over exposure periods of several thousand hours. The

The anaerobic corrosion of carbon steel and the potential inﬂuence

87

5.3 Example of data from experiments on the anaerobic corrosion of carbon steel in Swedish granitic groundwater at 30ºC and 50ºC [24]

corrosion rate was found to increase at pH 14, possibly due to the formation of soluble HFeO2− species. Hydrogen was also formed under oxic conditions when localised corrosion and acidification occurred. On non-pickled specimens, which would have had an initial air-formed oxide present on the surface, there was an incubation period observed before gas generation started [21]; the incubation period was shorter at higher temperatures. Surface analytical measurements (X-ray diffraction and Raman spectroscopy) have shown that the corrosion product formed on carbon steel in both strongly alkaline and weakly alkaline conditions, with no carbon dioxide present in the cover gas at the start of the experiment, is magnetite [21,29]. Siderite would be expected in the presence of carbon dioxide. Repassivation experiments that involved abrading the surface of carbon steel electrodes in anoxic alkaline solutions prepared using isotopically labelled water (H2O18) have shown that carbon steel repassivates rapidly by reaction with the water [30]. The detailed compositions and structures of the passive films formed on carbon steel in alkaline solutions are still a matter of debate, despite much research. The films may have a multi-layer structure, the exact composition of which depends on the electrochemical potential [31]. Several workers have investigated the reduction of the oxide films using electrochemical techniques [32]. Jelinek and Neufeld [33] reported that a two-layer corrosion product formed on iron in deaerated water. The outer layer was friable and the inner layer was adherent and compact. X-ray diffraction (XRD) analysis identified Fe3O4, but there were several other unidentified peaks. Mass spectroscopy of the gas released while heating the corrosion product revealed that hydrogen was trapped in it, indicating that hydrogen was formed during the anaerobic corrosion process. A two-layer oxide has also been observed by Smart et al. [21]. In recent years, the Swedes and Finns have considered the possibility of placing copper HLW canisters horizontally rather than vertically in their repository designs. However, such a layout would involve using an outer carbon steel container to hold

88

Sulphur-assisted corrosion in nuclear disposal systems

5.4 Anaerobic corrosion rate of carbon steel as a function of time in 0.1 M NaOH + NaCl at 30ºC, 50ºC and 80ºC. Note: Bottom figure has expanded scale for corrosion rate [21]

the bentonite around the copper canister and consequently the effect of the iron corrosion products on the physico-chemical properties of the bentonite is an area of recent research. The corrosion rate of carbon steel embedded in compacted bentonite has been measured using the same gas cell technique as that used to measure the corrosion rate of steel in aqueous solutions [34,35] and the effect of the Fe2+ ions released by corroding iron on the physico-chemical properties of the bentonite has been examined using a range of analytical techniques [36–38]. Experiments designed to investigate the effect of radiation on the anaerobic corrosion rate of steel in repository environments have been carried out as part of the Swedish programme [39] and similar experiments in support of the Belgian Supercontainer design are in progress. At dose rates of 300 Gy h−1 it was found that the gas

The anaerobic corrosion of carbon steel and the potential inﬂuence

89

generation rate due to anaerobic corrosion was higher than without radiation, but the effect was less marked at 11 Gy h−1. 5.3 Sulphur-based species in contact with carbon steel containers in a Belgian repository In general, the predominant sulphur-containing species present in a repository groundwater will be sulphate. In the proposed Belgian repository [40,41], the sulphur near the waste containers will be present as sulphide, in the form of the mineral pyrite, FeS2, and sulphate as a dissolved species in the Boom clay groundwater, at a maximum concentration of up to 6 mg/l. However, disturbing the Boom clay leads to the formation of higher concentrations of sulphate, through the oxidation of pyrite, which is present at concentrations of ~5 wt% in the Boom clay. This may lead to sulphate concentrations as high as 20 g/l sulphate, together with 720 mg/l thiosulphate in the disturbed zone. Examples of the calculated concentrations of the sulphur species at the surface of the overpack are given in Fig. 5.5; these data refer to an intermediate case between Boom clay porewater and the porewater in the disturbed zone. There may also be a small amount of sulphur in the organic matter present in the groundwater. Groundwater taken from possible geological formations in other repository concepts may contain much higher concentrations of sulphate, for example, the groundwater at Äspö underground laboratory in Sweden contains of the order of 500 mg/l sulphate [42] and a borehole at Sellafield in the UK contained 1130 mg/l sulphate [43]. Pyrite particles are also present in the bentonite backfill used in the Swedish programme. Sulphide concentrations are generally orders of magnitude lower than the concentration of sulphate and may be below the detection limit in specific groundwaters; for example, the sulphide concentration in Äspö groundwater has recently been measured as being of the order of 0.02 mg/l. Thiosulphate concentrations are not generally present in measurable concentrations in groundwaters. Another potential source of sulphur is the sulphur that is present in carbon steel itself, which is normally at a concentration of <0.05 wt%. With respect to the possible effect of sulphur species on corrosion, the most significant species are sulphate, thiosulphate, sulphide and elemental sulphur. The Pourbaix diagram for sulphur [11] is shown in Fig. 5.6, which shows that the predominant species in the neutral to alkaline pH range that is of interest in the radioactive waste disposal situation are sulphide at low oxidation potentials and sulphate at higher oxidation potentials. Although the sulphides are the stable species at low potentials, it is found in practice that sulphates are stable, even in the absence of oxidising species, unless the reduction reaction is catalysed, for example by microbial activity (e.g. by sulphate reducing bacteria). Recent thermodynamic calculations by Macdonald [4] indicate that a number of polysulphides and polythionates are thermodynamically stable under the proposed repository conditions; an example of the calculated potential–pH diagram for the sulphur–water system is shown in Fig. 5.7. If one considers the neutral to alkaline pH range, in addition to SO42− at oxidising potentials and HS− or S2− at reducing potentials, there are also thermodynamically predicted domains of stability for S4O3, HS2O8 and S3O32−. However, the actual presence of any of these species under repository conditions has not been demonstrated.

90

Sulphur-assisted corrosion in nuclear disposal systems

5.5 Calculated concentration evolution of chloride, sulphate, sulphide, thiosulphate and carbonate at the overpack surface [41]. This is the ‘mixed case’, which is intermediate between the Boom clay porewater and the worst case. All concentrations are in mmol/kg

5.4 Literature data on the effect of sulphur species on the anaerobic corrosion of carbon steel 5.4.1

Thermodynamic stability of iron-sulphur species

When iron is introduced into the S–H2O system, thermodynamic modelling [4,44] predicts the thermodynamically stable phases shown in Figs. 5.8 and 5.9. In this situation, it is interesting to note that iron sulphides, FeS or FeS2, become an option for the corrosion product as well as magnetite, Fe3O4, and this could have an influence on the kinetics of the corrosion reactions. However, the inclusion of carbon, chlorine, phosphorus and sulphur-based species, indicates that the stable species are as shown in Fig. 5.10.

The anaerobic corrosion of carbon steel and the potential inﬂuence

91

5.6 Potential (E, V vs. NHE)–pH equilibrium diagram (Pourbaix diagram) for the stable equilibria of the system sulphur–water, at 25°C [11]

5.7 Calculated potential–pH diagrams for the sulphur–water system at 50°C for a dissolved sulphur activity of 0.031. Macdonald in Ref. 4

92

Sulphur-assisted corrosion in nuclear disposal systems

5.8 Calculated potential–pH diagrams for the iron–sulphur–water system at 50°C for dissolved activities of 10−6 and 0.031, respectively. Macdonald in Ref. 4

5.9 Potential–pH diagram for iron in high salinity brine at 25°C in the presence of 10 ppm total dissolved sulphide (H2S + HS− + S2−). Activities of HSO4− and SO42− = 10−6 mol/kg. Activities of dissolved iron species = 10−4 mol/kg [44]

The anaerobic corrosion of carbon steel and the potential inﬂuence

93

5.10 Calculated potential–pH diagrams for the iron–carbon–chloride– phosphorous–sulphur–water system at 50°C for dissolved activities of 10−6, 1.0, 10−4, 10−4 and 10−4 mol/kg, respectively [4]

Under some conditions, it is possible to form ‘green rusts’ as the corrosion products; these are Fe(II)–Fe(III) mixed hydroxides with a layered structure, consisting of Fe(OH)2-like hydroxide sheets which alternate regularly with interlayers composed of anions and H2O molecules. Their composition depends on the anions present in the solution, for example Cl−, CO32− or SO42−. The general formula for Green Rust 1 (GR1(Cl−)) [45,46] can be written as [Fe(II)3–2.2 Fe(III) (OH)8–6.4.]+.[Cl.2H2O]−, whereas Green Rust 2 (GR2(SO42−)) has the formula [Fe(II)4 Fe(III)2 (OH)12]2+.[SO4.2H2O]2− [47,48]. Detailed analytical work, using particularly Mössbauer analysis, has enabled E–pH diagrams to be constructed showing the regions of stability of GR1 and GR2, as shown in Fig. 5.11. These diagrams indicate that, in neutral or alkaline solutions, such as are of interest in waste disposal environments, neither GR1 or GR2 will be formed under anoxic conditions, when the potential of the iron would be expected to fall below the hydrogen evolution line (line (a) in the E–pH diagrams). They are generally believed to be reactive metastable phases that are transients in the formation of stable end products such as magnetite [49]. A number of authors have reported the observation of Green Rust in corrosion products formed in seawater [50–54] in conjunction with the activity of sulphate reducing bacteria in biofilms. Sagoe-Crentsil and Glasser [55] has constructed a stability diagram for iron in cement at pH 12 and a temperature of 25ºC, as a function of chloride concentration, showing the presence of green rusts at intermediate chloride concentrations and ferrous chloride at high chloride concentrations. In situ measurements of the compositions of oxides formed on carbon steel in simulated anoxic groundwaters [56,57] identified siderite in sodium carbonate/ bicarbonate solutions, and showed that in mixed carbonate–sulphate–chloride solutions, compact deposits of carbonate-containing, and to a lesser degree, sulphatecontaining green rusts were formed, along with small amounts of magnetite.

94

Sulphur-assisted corrosion in nuclear disposal systems

5.11 Potential–pH diagram for iron in sulphate-containing aqueous media and chloride-containing aqueous media, showing domain of stability of (a) Green Rust 1, GR1(Cl−) [45] and (b) Green Rust 2, GR2(SO42−) [47]

The anaerobic corrosion of carbon steel and the potential inﬂuence

95

5.4.2 The possible effect of sulphur-based species on the anaerobic corrosion of carbon steel under abiotic conditions The experiments described above (Section 5.2) to measure the anaerobic corrosion rate of carbon steel by hydrogen evolution measurements investigated the anaerobic corrosion behaviour of carbon steel in purely inorganic abiotic (i.e. microbe-free) conditions, although sulphate was present in some of the test groundwaters [24]. A small number of experiments have been conducted within the Swedish programme to investigate the effect of different concentrations of dissolved sulphate on the anaerobic corrosion rate of carbon steel [24]. Increasing anaerobic corrosion rates were observed in solutions containing a higher concentration of sulphate, but this may have been due to an overall increase in the ionic strength, rather than an effect of the SO4 per se, or due to a change in pH. The compositions of the synthetic groundwaters used are shown in Table 5.1; in some experiments, FeSO4 was also added. The results from these experiments are shown in Fig. 5.12 and Fig. 5.13. Table 5.1 Composition (mM) of artificial groundwaters used in anaerobic corrosion experiments [24] Ion

Allard

Äspö

Na+ K+ Ca2+ Mg2+ Cl– Total carbonate SO42– SiO2 pH

2.84 0.10 0.45 0.18 1.96 2.00 0.10 0.21 8.1

131.3 0.19 109.5 2.06 339.6 0.180 7.40 7.0–8.0

Bentonite-equilibrated 560 0 0 0 540 10 0 10.5

5.12 Rate of evolution of hydrogen produced by anaerobic corrosion of pickled carbon steel in anoxic artificial Allard groundwater at 50°C, showing the effect of adding 0.1 M FeSO4 or 0.1 M K2SO4, compared to Allard water with 0.1 mM sulphate

96

Sulphur-assisted corrosion in nuclear disposal systems

5.13 Rate of evolution of hydrogen produced by the anaerobic corrosion of pickled carbon steel in anoxic artificial groundwaters [24]

Similarly, Gould and Evans [58] found that aggressive anions (Cl− and SO42−), at concentrations of up to 0.01 M, had no effect on the corrosion of iron in deaerated neutral conditions. The following part of the paper summarises the results of electrochemical investigations described in the literature on the effect of sulphur-based species on the corrosion behaviour of carbon steel in abiotic alkaline conditions. Shoesmith et al. studied the electrochemical behaviour of carbon steel in alkaline sulphide solutions. Their first paper [59] contains a summary of the literature up to 1978 with regard to the substances formed on the surface of iron over a range of pH values in the presence of dissolved sulphur species. The survey showed that various peaks in cyclic voltammograms have been attributed to the formation of iron sulphide (pyrite) and the deposition of sulphur at more anodic potentials on top of a layer of magnetite and/or haematite films. The experimental work reported in the paper investigated the species formed in NaOH/H2S solutions over a range of pH values and sulphide concentrations. The paper contains a detailed experimental study of the electrochemical behaviour of iron in deaerated 0.1 M NaOH solution, dosed with dissolved H2S to give pH values in the range 6.5 to >13. The peak at ~−484 mV vs SCE is attributed to the deposition of sulphur. Other more oxidised sulphur species may be formed at higher potentials (e.g. sulphite, sulphate, thiosulphate or polythionates). Polysulphides were observed at potentials above the sulphur deposition potential (−100 to +200 mV vs SCE). NaFeS2 is also formed at higher anodic potentials. The presence of dissolved sulphide tends to suppress the formation of an oxide film in current transient measurements carried out at −400 mV vs SCE (i.e. the presence of sulphide inhibits the formation of an iron oxide film). It is suggested that competitive adsorption occurs between SH− and OH−, which affects the formation of the oxide film. It is likely that the presence of sulphide will affect the corrosion rate, but the corrosion rate was not specifically measured in the paper. The presence of deposited sulphur blocks the pores in the iron oxide film. It should be noted that the peaks observed in the cyclic voltammograms occur at potentials above the corrosion potential expected for carbon steel in anoxic alkaline conditions, such as would be

The anaerobic corrosion of carbon steel and the potential inﬂuence

97

expected in the Supercontainer concept. Furthermore, the pH values examined were below those that will occur in a cementitious porewater. In a subsequent paper [60], the electrochemical behaviour of iron was examined in the pH range 9–12. Films grown at potentials more anodic than −400 mV vs SCE were composed of mackinawite (FeS1−x). Transient measurements suggest that initially a thin film of magnetite is formed on the metal surface. This is consistent with other anaerobic corrosion studies (e.g. [21]). Salvarezza et al. [61] showed that, at pH 12, the anodic current is higher in the presence of sulphide across the whole potential range. Potentiostatic experiments were carried out at anodic potentials and the formation of black deposits and pits was investigated. Pits were also observed at potentials in the active region. It remains to be seen whether similar behaviour occurs at the higher pH values expected in cementitious porewater. The SH/OH ratio is important in determining the corrosion behaviour. No experiments were reported for pH values higher than 12. Yeske [62] studied the corrosion of carbon steel in alkaline sulphide liquors in relation to the Kraft process used in the paper pulp industry. The paper is concerned with the corrosion of carbon steels in ‘white liquor’, which is an alkaline solution of sodium hydroxide (NaOH, 70–150 g/l), sodium sulphide (Na2S, 20–50 g/l), sodium carbonate (Na2CO3, 10–60 g/l), and a few g/l of sodium sulphite (Na2SO3), sodium sulphate (Na2SO4), sodium chloride (NaCl), sodium thiosulphate (Na2S2O3) and sodium polysulphide (Na2Sx). This is mainly used at a temperature of 85–95°C. The corrosivity of the liquor is believed to be due to the minor components, particularly thiosulphates and polysulphides. A thick conductive sulphide-rich film is reported to form. It was found that the corrosion potential became more negative with increasing concentrations of sulphide. A change from active to passive behaviour was observed in moderate concentrations of caustic (60–80 g/l) and sulphide (10–30 g/l), but in more concentrated solutions, no passivation occurred and the corrosion rate was approximately 125 μm year−1 at a corrosion potential of ~−900 mV SHE. This value was below the calculated hydrogen evolution potential for the test solutions. The corrosion rate appears to have been related to the OH−/S2− concentration ratio. In the passive condition, the corrosion potential was ~−560 mV SHE (i.e. above the hydrogen evolution potential). It was suggested that the carbon steel can act as an inert host for reactions involving sulphur redox reactions. It was found that thiosulphate additions to the NaOH/Na2S liquors caused a marked increase in corrosion rate, in agreement with earlier work by Wensley and Charlton [63], accompanied by the formation of a greenish-black NaFeS2 film, whereas the corrosion product in the absence of thiosulphate was assumed to be mackinawite, FeS1−x. Additions of small amounts of elemental sulphur to form polysulphides in the liquor increased the corrosion rates, but at higher concentrations (>2 g/l S0) passivation ensued. It was found that the corrosion rates were very similar in both aerated and deaerated solutions; this may be due to the consumption of oxygen by the liquor to form sulphoxy compounds. The amorphous structure of the surface layers observed in tests with polysulphides was suggested to be due to the formation of amorphous sulphur deposits. Wensley et al. found that polysulphides have the ability to control the potential of the carbon steel, but do not significantly affect the corrosion rate. Crowe and Tromans [64] also carried out studies in support of the pulp and paper industry and determined the high-temperature polarisation behaviour of carbon steel in 2.5 M NaOH/0.5 M Na2S solutions and 3 M NaOH solutions at 150°C. Their calculated

98

Sulphur-assisted corrosion in nuclear disposal systems

Pourbaix diagrams for the Fe–S–H2O system are shown in Fig. 5.14 for 25°C and 100°C, respectively. Cyclic voltammetry was used to investigate the electrochemical behaviour of iron in concentrated NaOH solutions, with and without sulphide ions present. In the presence of sulphide, two new peaks were observed, corresponding to the formation of FeS and its oxidation to FeS2. It was suggested that sulphide ion is incorporated into the surface film, which only becomes protective at higher potentials when the iron sulphide is oxidised to FeOOH or Fe2O3. The iron–sulphur interactions are complex and strongly dependent on kinetics. The hydrogen evolution currents were observed to increase in the presence of sulphide and this was attributed to the fact that iron sulphides are known to catalyse the hydrogen evolution reaction [65]. Some disagreements with the interpretation given in the paper by Shoesmith et al. [59] were recorded. It was concluded that sulphide interferes with passivation of the iron surface, due to the formation of iron sulphides which reduce the protectiveness of the surface layer. Vasquez Moll et al. [66] investigated the pitting behaviour of carbon steel in alkaline solutions containing a range of sulphur species. Cyclic voltammetry was carried out on 1020 SAE carbon steel in 0.002 M solutions of NaOH containing Na2SO4, Na2S2O3, Na2S, Na2SO3 or potassium thiocyanate, KSCN, in the concentration range 10−5 to 2 × 10−3 M. Thiosulphate is metastable and in alkaline conditions, it decomposes to form sulphide and sulphate. Pitting was observed in the presence of thiosulphate, with the pitting potential decreasing with increasing thiosulphate concentration and the production of black iron sulphide in the pits. Both Crowe and Tromans, and Vasquez Moll et al., mention the work of Horowitz [67], who found evidence for the formation of a soluble iron (II)–thiosulphate complex. Kannan and Kelly [68] also found that carbon steel pits at much lower potentials in alkaline conditions than is normal for chloride-contaminated alkaline solutions and that the SH/OH ratio is critical in determining the occurrence of pitting. Jayalakshmi and Muralidharan [69,70] investigated the effect of dissolved elemental sulphur (0.1–1 M) on the passivation of iron in deoxygenated KOH solutions (0.1–10 M), using cyclic voltammetry. The results of the cyclic voltammetry investigation are explained on the basis of competitive adsorption between SH− and OH−. The activation energy for the oxidation of iron is lower in the presence of SH− ion compared to OH−. The authors suggest that the stable polysulphides at high pH are S42− and S32−. They proposed the following reaction sequence for the effect of sulphide ion on the corrosion of iron: Fe + SH− → FeSHads + e−

[5.15]

(FeSH)ads → FeSH+ads + e−

[5.16]

(FeSH)ads + OH− + xFe → Fe(1+x)S + H2O

[5.17]

FeS1−x, mackinawite (≡Fe(1+x)S), is unstable and converts to FeS, thus: Fe(1+x)S → FeS + xFe2+ + 2x e−

[5.18]

At higher potentials the FeS is converted to FeOOH: FeS + 2H2O → FeOOH + S2− + 3H+ + e−

[5.19]

Zamaletdinov et al. [71] showed that in a simulated alkali pulp liquor, the corrosion potential falls by ~0.55 V in the presence of >3 g/l Na2S, indicating that surface

The anaerobic corrosion of carbon steel and the potential inﬂuence

99

5.14 Potential–pH diagram for Fe–S–H2O system at (a) 25°C, and (b) 100°C in the presence of 0.5 M HS−, 10−2 M S2O3 and 10−6 M dissolved iron [64]

100

Sulphur-assisted corrosion in nuclear disposal systems

activation occurs. Abd El Haleem and Abd El Aal [72,73] investigated the effect of sulphide ions on the electrochemical behaviour of iron in alkaline conditions. They suggested that many different stages occur during the corrosion of iron in alkaline sour conditions and that it is difficult to evaluate the rates and mechanisms accurately. They measured cyclic voltammograms for iron in 0.1, 0.5 and 1 M NaOH in the presence of 0.1 M S2− ions and proposed that several iron sulphide and iron oxide species are formed as the potential is increased in the anodic direction. Some of the sulphide compounds are non-stoichiometric. There are a number of possible iron sulphide compounds that have been reported in real corrosion scales, including FeS (troilite), Fe2S3, Fe1−xS (pyrrhotite), FeS1−x (mackinawite) and FeS2 (pyrite). It is important to consider the longer-term processes that might occur in a corrosion product layer, rather than just the features that are observed over a short timescale in electrochemical polarisation experiments. It was found that at higher sulphide concentrations, the corrosion potential became more negative, corresponding to the formation of iron sulphide films, rather than iron oxide films. There was a Nernstian relationship between the sulphide concentration and the corrosion potential. From this brief survey of the literature on the electrochemical aspects of the corrosion of iron in alkaline conditions in the presence of sulphur species, it can be seen that a range of iron oxides, hydroxides and sulphides can be formed, and it is not possible to predict with certainty how protective such films would be in a repository situation. In principle, iron sulphide films could be less protective than iron oxyhydroxides and act as superior cathodes and hence increase the hydrogen production rate resulting from the anaerobic corrosion of steel in a repository. In the context of the Belgian Supercontainer concept, the main question arising from this review, therefore, is whether there are any sulphur-containing films formed on iron at the free corrosion potential under the abiotic, anoxic, alkaline conditions that may arise on the surface of the carbon steel overpack and if so, how do they affect the corrosion rate and hydrogen generation rates? 5.4.3

Microbially induced corrosion under anoxic conditions

The database of existing anaerobic corrosion gas generation experiments does not take any account of the possible effect of sulphide that is produced through microbial activity. In the Belgian Supercontainer concept, sulphide will probably not be generated by microbial action near the surface of the carbon steel container, because it will be protected by the alkaline porewater of the cementitious buffer material, and there would be a number of factors which would tend to inhibit microbial activity, including a low water supply because of the low hydraulic conductivity of the surrounding Boom clay, an elevated temperature, at least initially, and the presence of a radiation field, although the latter would probably be insufficient to prevent microbial activity. It may be possible for sulphide that is generated externally to the Supercontainer package by the enhanced activity of sulphate reducing bacteria (SRBs) in the Engineered Disturbed Zone (EDZ) [40] to diffuse through the cementitious buffer material and hence to come into contact with the carbon steel container, in which case the question becomes does this sulphide have any effect on the corrosion behaviour of the carbon steel under alkaline conditions? Most literature dealing with the topic of anaerobic corrosion of iron in the presence of sulphur species revolves around the effects of sulphate reducing bacteria, rather than corrosion in purely inorganic systems. In repository systems where carbon steel

The anaerobic corrosion of carbon steel and the potential inﬂuence

101

is exposed directly to groundwaters containing sulphate and no method exists for reducing the activity of SRBs, such as an alkaline cementitious porewater, there seems little doubt that carbon steel will suffer from microbially influenced corrosion (MIC), as discussed below. Evidence from the Swedish programme indicates that in compacted bentonite, there is insufficient water available to support an active population of SRBs [42]. Similarly, the microbial activity within the Boom clay used in the Belgian concept is believed to be very low, partly because the pore size in the clays is too small to accommodate the bacteria. However, in a disturbed region, microbial activity may increase. It should be noted that the oil and gas production industry handles anoxic conditions with various sulphide species, both naturally present and microbially generated. That literature is the subject of review in another paper in this volume. Whilst not comparable in exact conditions to a repository scenario, it does, like the pulp and paper experience, provide another source of information on corrosion rates and hydrogen generation. Sulphate-reducing bacteria (SRB) SRBs require anoxic conditions (i.e. they are obligate anaerobes) and use sulphate, sulphite, thiosulphate, sulphur or nitrate as an electron acceptor in anaerobic respiration [74]. They are a well known cause of corrosion of cast iron pipelines in soils [75]. The risk of MIC of radioactive waste containers has been considered by Pritchard [76] and King [77]. The mechanism of SRB corrosion is shown schematically in Fig. 5.15. Sulphate-reducing bacteria produce mainly hydrogen sulphide. The iron sulphide film formed on steels by reaction with hydrogen sulphide is a good catalyst for hydrogen reduction and this facilitates the introduction of hydrogen into steels, making them more susceptible to hydrogen cracking and increasing the corrosion rate. The hydrogen generation rate would also be expected to increase, due to the reduction

5.15 Schematic diagram of the mechanism in an FeS corrosion cell created by the action of SRB. Iron sulphide sets up a galvanic couple with steel, sustained and extended by the further action of SRB. The bacteria use electrons from the corrosion process, possibly in the form of cathodic hydrogen, to reduce sulphate. Acid producing bacteria (APB) may also have a role in providing nutrients to SRB [78]

102

Sulphur-assisted corrosion in nuclear disposal systems

of water or hydrogen sulphide. The exact mechanisms of SRB corrosion and the metabolic pathways by which SRBs cause corrosion under anoxic conditions are still the subject of considerable conjecture [79,80] and a number of intermediates have been postulated, including the production of a corrosive phosphorus-based compound [81,82]. Lee and Characklis [80] have postulated that ferrous ion is a necessary precursor for corrosion by SRB activity. The FeS film formed has a low hydrogen overpotential and so galvanic coupling with unoxidised steel can result in an increased corrosion rate. In a recent review, Videla et al. [83] summarised the process by which SRB attack of carbon steel occurs. The activity of the SRBs leads to the formation of various sulphur-containing species (e.g. sulphides, bisulphides, hydrogen sulphide) or intermediate products (e.g. thiosulphates, polythionates, thiocyanate) which can lead to corrosion of the iron. The effects of sulphides can range from being protective, when adherent and continuous films are formed in the presence of low concentrations of soluble iron (probably mackinawite), to corrosive, when the film is loosely associated with high concentrations of soluble iron (e.g. pyrrhotite (Fe1−xS), or greigite, Fe3S4 or Fe(II)Fe(III)2S4). Videla reports that in biotic environments, FeS predominates, whereas in abiotic environments, FeS2 is the major type of iron sulphide present and it is more loosely attached. De Romero et al. [84] have shown that sulphide produced by SRB activity poisons the hydrogen evolution reaction and causes more hydrogen to enter the steel, possibly resulting in embrittlement. Pankhania [85] and De Silva et al. [82] have discussed the importance of hydrogen utilisation at metal surfaces during MIC processes; some experiments have shown that hydrogen consumption by SRB or hydrogenase increases the corrosion rate of mild steel. Corrosion rates of >7 mm year−1 have been reported for carbon steel in soils [75]. Li et al. [75] report that, in abiotic systems, anaerobic corrosion of iron in the presence of H2S initially results in the formation of mackinawite (FeS1−x), but this readily cracks and spalls and results in the formation of a loose precipitate as the sulphur content of the corrosion product increases to form greigite (Fe3S4) or pyrrhotite (Fe1−xS). The kinetics of such processes depend on the local environment (i.e. solution chemistry, redox potential, etc.). SRBs can use a range of sources of organic carbon, which can include carbon provided by the activity of other bacterial species, such as autotrophic acetogens, which are able to metabolise hydrogen that could in turn be produced by the abiotic anaerobic corrosion of carbon steel. There is therefore the possibility of an autocatalytic feedback loop, viz. Autotrophic acetogens: 4H2 + 2CO2 ⇒ CH3COO− + H+ + 2H2O Sulphate reducing bacteria: CH3COOH + SO42− ≈ ⇒ H2S and Fe 2+ + S2− ⇒ FeS There could be a beneficial effect of microbial activity because it will tend to consume the hydrogen produced by inorganic corrosion processes and therefore reduce the amount of gaseous hydrogen released by a waste container, but on the other hand, it may increase the corrosion rate of the waste container materials. The presence of SRB can enhance the hydrogen evolution reaction. It has been suggested that this could be by a process described as ‘cathodic depolarisation’, involving the action of an enzyme, hydrogenase, but this is just one of a number of mechanisms that could lead to enhanced corrosion rates in the presence of SRBs [74]. The relationship between the various iron sulphides that can form as a result of bacterial and chemical means is summarised in Fig. 5.16.

The anaerobic corrosion of carbon steel and the potential inﬂuence

103

5.16 Chemical and biological inter-relationships between iron sulphides (from Ref. 74, originally from Ref. 86)

Sulphur oxidising bacteria Sulphur oxidising bacteria (e.g. Thiobacillus) oxidise sulphides, elemental sulphur and other sulphur compounds to sulphuric acid, and can generate and tolerate pH values of 2 or lower [74,87]. Metals and concrete are both attacked. Sulphuroxidising bacteria can produce general corrosion by producing sulphuric acid. Under some circumstances, they may form a consortium with SRB. Many thiobacilli can fix carbon dioxide. In order to avoid microbially influenced corrosion of carbon steel waste containers, it is necessary to provide an environment around the waste container which will inhibit the activity of microbes, since microbial activity in deep geological repository groundwaters is inevitable. This could be by limiting the amount of available water, for example, by using a highly compacted bentonite, or increasing the pH to a high value, for example, by using a cement-based buffer. The presence of a radiation flux is unlikely to be sufficient to eliminate possible microbial activity. In the author’s opinion, it is important to carry out in situ tests, for example, in underground laboratories, to establish the likelihood of SRB activity affecting the corrosion behaviour of candidate waste container materials, since the microbial activity is very dependent on local chemistry and this cannot be adequately reproduced in a laboratory situation. Furthermore, the effects of mixed bacterial populations are important and they cannot be adequately simulated in a surface laboratory [74]. 5.5

Areas requiring further research

Thermodynamically, anaerobic abiotic corrosion to produce hydrogen is possible in the presence of sulphide species. During this review, a number of possible areas requiring further investigation were identified and these can be summarised as follows:

104 •

•

• • • •

5.6

Sulphur-assisted corrosion in nuclear disposal systems

There is a need to establish whether sulphur-containing species which may be present in a repository (e.g. S2−, S2O3 and SO42−, complex sulphides) are present in sufficiently high concentrations to affect the long-term kinetics of anaerobic corrosion. There is a need to explore the electrochemical behaviour of iron in the presence of potential sulphur species, in anoxic alkaline and neutral conditions. This should include repassivation measurements to determine the film-forming capabilities. There is a need to characterise the films formed after long-term exposure in alkaline, anoxic conditions, in the presence of sulphur species. There is a need to determine which sulphide films are protective against anaerobic corrosion and whether the corrosion rates are higher than in the absence of sulphur species, over a range of representative temperatures. There is a need to determine whether radiolysis affects the sulphur speciation and how it affects corrosion rate. There is a need to investigate the effect of sulphur species on the anaerobic corrosion behaviour by applying ‘in situ’ techniques, to take count of processes that cannot be adequately simulated in the laboratory (e.g. microbial processes [74]). Conclusions

The main conclusions from this review are as follows: 1. There is an extensive database with regard to the anaerobic corrosion of iron in neutral and alkaline conditions, such as would be expected in radioactive waste repositories. 2. There are a number of investigations reported in the literature with regard to the thermodynamic stability of iron–sulphur species and the electrochemical characteristics of iron in alkaline sulphur-containing environments. However, there are very few data on the effect of sulphur on the anaerobic corrosion rate of iron or the resulting hydrogen evolution rate. A range of iron oxides, hydroxides and sulphides can be formed, and it is not possible to predict with certainty how protective such films would be in a repository situation. It is likely that the existence of iron sulphide films would accelerate the rate of generation of hydrogen under reducing conditions. 3. There is a need for additional experimental studies to determine the protectiveness of iron sulphide films in simulated repository environments. This should also take account of radiation effects and the generation of sulphur species through microbial activity. Acknowledgements The author gratefully acknowledges NIRAS/ONDRAF for financial support for preparing and presenting this paper. References 1. N. R. Smart, Corrosion, 65(3) (2009), 195–212. 2. B. Kursten, E. Smailos, I. Azkarate, L. Werme, N. R. Smart and G. Santarini, COBECOMA State-of-the-art document on the COrrosion BEhaviour of COntainer Materials, Final Report for EU contract report N° FIKW-CT-20014-20138, 2004.

The anaerobic corrosion of carbon steel and the potential inﬂuence

105

3. B. Kursten, E. Smailos, I. Azkarate, L. Werme, N. R. Smart, G. Marx, M. A. Cuñado and G. Santarini, Corrosion Evaluation of Metallic HLW/Spent Fuel Disposal Containers – Review, presented at Eurocorr 2004, Nice, September, 2004. 4. Corrosion Expert Panel, A Review of Corrosion and Materials Selection Issues Pertinent to Underground Disposal of Highly Active Nuclear Waste in Belgium, Niras-Ondraf Report, NIROND 2004-02, 2004. 5. I. Puigdomenech, J.-P. Ambrosi, L. Eisenlohr, J.-E. Lartigue, S. A. Banwart, K. Bateman, A. E. Milodowski, J. M. West, L. Griffault, E. Gustafsson, K. Hama, H. Yoshida, S. Kotelnikova, K. Pedersen, V. Michaud, L. Trotignon, J. Rivas Perez and E.-L. Tullborg, O2 Depletion In Granitic Media. The REX Project, SKB Report SKB TR-01-05, 2001. 6. A. J. Bard, R. Parsons and J. Jordan (eds), Standard Potentials in Aqueous Solution, IUPAC, 1985. 7. L. L. Shreir, ‘Corrosion in aqueous solutions’, Chapter 1.4 in ‘Corrosion’, 1:105, 3rd edition, ed. L. L. Shreir, R. A. Jarman and G. T. Burstein. Butterworth-Heinemann, 1994. 8. U. R. Evans and J. N. Wanklyn, Nature, 162 (1948), 27. 9. V. J. Linnenbom, J. Electrochem. Soc., 105 (1958), 321. 10. R. Grauer, B. Knecht, P. Kreis and J. P. Simpson, ‘Hydrogen evolution from corrosion of iron and steel in intermediate level waste repositories’, in Scientific Basis for Nuclear Waste Management XIV, ed. T. A. Abrajano and L. H. Johnson, Mater. Res. Soc. Symp. Proc. 212 (1991), 295–302. 11. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, 2nd edition, National Association of Corrosion Engineers, Houston, USA, 1974. 12. C. C. Naish, P. H. Balkwill, T. M. O’Brien, K. J. Taylor and G. P. Marsh, The Anaerobic Corrosion of Carbon Steel in Concrete, Nirex Report NSS/R273 (available from UK Nuclear Decommissioning Authority, NDA), 1990. 13. L. Young, Anodic Oxide Films, Academic Press, London, 1961. 14. A. Atkinson and G. P. Marsh, Engineered Barriers: Current Status, Nirex Report NSS/ G102 (available from UK Nuclear Decommissioning Authority, NDA), 1988. 15. R. Grauer, B. Knecht, P. Kreis and J. P. Simpson, Werkst. Korros., 42 (1991), 637. 16. R. Grauer, B. Knecht, P. Kreis and J. P. Simpson, ‘Hydrogen evolution from corrosion of iron and steel in intermediate level waste repositories’, in ‘Scientific Basis for Nuclear Waste Management XIV’, 295, ed. T. A. Abrajano and L. H. Johnson, Mater. Res. Soc. Symp. Proc., 1991. 17. R. Grauer, The Corrosion Behaviour of Carbon Steel in Portland Cement, NAGRA Technical Report 88-02E, 1988. 18. P. Kreis, Wasserstoffentwicklung durch Korrosion von Eisen und Stahl in anaeroben, alkalischen Medien im Hinblick auf ein SMA-Endlager, NAGRA-NTB-93-27, 1993. 19. P. Kreis, Hydrogen Evolution from Corrosion of Iron and Steel in Low/Intermediate Level Waste Repositories, NAGRA Technical Report 91-21, 1991. 20. P. Kreis and J. P. Simpson, ‘Hydrogen gas generation from the corrosion of iron in cementitious environments’, in Corrosion Problems Related to Nuclear Waste Disposal, European Federation of Corrosion Publication number 7, Institute of Materials, 1992. 21. N. R. Smart, D. J. Blackwood, G. P. Marsh, C. C. Naish, T. M. O’Brien, A. P. Rance and M. I. Thomas, The Anaerobic Corrosion of Carbon and Stainless Steels in Simulated Cementitious Repository Environments: A Summary Review of Nirex Research, AEA Technology Report (available from UK Nuclear Decommissioning Authority, NDA), AEAT/ERRA-0313, 2004. 22. N. R. Smart, D. J. Blackwood and L. Werme, The Anaerobic Corrosion of Carbon Steel and Cast Iron in Artificial Groundwaters, SKB Report TR-01-22, 2001. 23. N. R. Smart, D. J. Blackwood and L. Werme, Corrosion, 58(7) (2002), 547. 24. N. R. Smart, D. J. Blackwood and L. Werme, Corrosion, 58(8) (2002), 627. 25. R. Fujiwara, I. Yasutomi, K. Fukudome, T. Tateishi and K. Fujiwara, Mater. Res. Soc. Symp. Proc. 663 (2001), 497.

106

Sulphur-assisted corrosion in nuclear disposal systems

26. R. Fujisawa, T. Cho, K. Sugahara, Y. Takizawa, Y. Horikawa, T. Shiomi and M. Hironaga, Mater. Res. Soc. Symp. Proc. 465 (1997), 675. 27. M. Mihara, T. Nishimura, R. Wada and A. Honda, Saikuru Kiko Giho 15 (2002), 91–101. 28. T. Nishimura, R. Wada and K. Fujiwara, R D Kobe Seiko Giho, 53(3) (2003), 78–83. 29. N. R. Smart, P. A. H. Fennell, R. Peat, K. Spahiu and L. Werme, ‘Electrochemical measurements during the anaerobic corrosion of steel, in Scientific Basis for Nuclear Waste Management XXIV, ed. K. P. Hart and G. R. Lumpkin. Mater. Res. Soc. Symp. Proc., 663 (2001), 487–495. 30. N. R. Smart and N. J. Montgomery, The Repassivation of Carbon Steel and Stainless Steel in Deaerated Alkaline Conditions: An Electrochemical and Surface Analytical Investigation, AEA Technology Report AEAT/R/ENV/0232 (available from UK Nuclear Decommissioning Authority, NDA), 2001. 31. S. Haupt and H. H. Strehblow, Langmuir, 3 (1987), 873. 32. A. M. Riley and J. M. Sykes, Corros. Sci., 28(8) (1988), 799. 33. J. Jelinek and P. Neufeld, Corrosion, 38(2) (1982), 99. 34. N. R. Smart, A. P. Rance and L. O. Werme, ‘Anaerobic corrosion of steel in bentonite’, presented at MRS 2003 (Kalmar, Sweden, 15–18 June 2003), in Scientific Basis for Nuclear Waste Management XXVII, ed. V. M. Oversby and L. O. Werme. Mater. Res. Soc. Symp. Proc., 807 (2004), 441–446. 35. N. R. Smart, A. P. Rance, L. Carlson and L. O. Werme, ‘Further studies of the anaerobic corrosion of steel in bentonite’, presented at MRS 2005 (Ghent, September 2005), Mater. Res Soc. Symp. Proc., 932 (2006), 813. 36. L. Carlson, O. Karnland, V. M. Oversby, A. P. Rance, N. R. Smart, M. Snellman, M. Vähänen and L. O. Werme, ‘Experimental studies of the interactions between anaerobically corroding iron and bentonite’, presented at conference on Clays in Natural and Engineered Barriers for Waste Confinement (Tours, France, 14–18 March 2005). Phys. Chem. Earth, 32(1–7) (2007), 334–345. 37. L. Carlson, O. Karnland, S. Olsson, A. P. Rance and N. R. Smart, Experimental Studies on the Interactions Between Anaerobically Corroding Iron and Bentonite, Posiva Working Report 2006-60, 2006. 38. N. R. Smart, F. Bate, L. Carlson, T. G. Heath, A. R. Hoch, F. M. I. Hunter, O. Karnland, S. J. Kemp, A. E. Milodowski, A. M. Pritchard, A. P. Rance, B. Reddy and L. O. Werme, Interactions Between Iron Corrosion Products and Bentonite – Final Report, Serco/TAS/ MCRL/19801/C001 Issue 2, Final Report for EU contract N° FI6W-CT-2003-02389, NF-PRO WP2.3/D2.3.9, 2008. 39. N. R. Smart, A. P. Rance and L. O. Werme, ‘The effect of radiation on the anaerobic corrosion of iron’, presented at workshop on Long-term Prediction of Corrosion Damage in Nuclear Waste Systems (Pennsylvania State University, May 2007). J. Nucl. Mater., 379 (2008), 97–104. 40. M. Van Geet, L. Wang, P. De Boever and M. De Craen, Geochemical Boundary Conditions for In-Situ Corrosion Experiments, SCK-CEN Report CEN-R-4308, 2006. 41. L. Wang, Near-field Chemistry of a HLW/SF Repository in Boom Clay – Scoping Calculations Relevant to the Supercontainer Design, SCK-CEN Report CEN-ER-17, 2006. 42. F. King, L. Ahonen, C. Taxen, U. Vuorinen and L. Werme, Copper Corrosion Under Expected Conditions in a Deep Geological Repository, SKB Report TR-01-23, 2001. 43. K. A. Bond and C. J. Tweed, Groundwater Compositions for the Borrowdale Volcanic Group, Boreholes 2,4 and RCF3, Sellafield, Evaluated Using Thermodynamic Modelling, Nirex Report NSS/R397 (available from UK Nuclear Decommissioning Authority, NDA), 1995. 44. D. D. MacDonald and B. C. Syrett, Corrosion, 35(10) (1979), 471–474. 45. P. Refait and J.-M. R. Genin, Corros. Sci., 34(5) (1993), 797.

The anaerobic corrosion of carbon steel and the potential inﬂuence

107

46. P. H. Refait, M. Abdelmoula and J.-M. R. Genin, Corros. Sci., 40(9) (1998), 1547–1560. 47. J.-M. R. Genin, A. A. Olowe, Ph. Refait and L. Simon, Corros. Sci., 38(10) (1996), 1751– 1762. 48. C. Ruby, A. Gehin, R. Aissa and J.-M. R. Genin, Corros. Sci., 48(11) (2006), 3824–3837. 49. P. Refait, A. Gehin, M. Abdelmoula and J.-M. R. Genin, Corros. Sci., 45(4) (2003), 659–676. 50. A. A. Olowe, Ph. Bauer, J.-M. R. Genin and J. Guezennec, Corrosion, 45(3) (1989), 229. 51. J. Chen, Z. Cai, Z. Wang and H. Zhang, ‘A study of the barrier layer rust formed on low alloy steels in sea water by conversion electron Mössbauer spectroscopy’, in Proc. ICAME, 264. Jaipur, India, 1981. 52. P. Refait, J.-B. Memet, C. Bon, R. Sabot and J.-M. R. Genin, Corros. Sci., 45(4) (2003), 833–845. 53. S. Pineau, R. Sabot, L. Quillet, M. Jeannin, Ch. Caplat, I. Dupont-Morral and Ph. Refait, Corros. Sci., 50(4) (2008), 1099–1111. 54. A. Zegeye, L. Huguet and F. Jorand, ‘Microbial formation of Fe(II)–Fe(III) hydroxysulphate green rust, a product from corrosion in sea water’, in EUROCORR 2004: Long Term Prediction & Modelling of Corrosion (Nice, France, 12–16 September 2004). 55. K. K. Sagoe-Crentsil and F. P. Glasser, Corrosion, 49(6) (1993), 457. 56. C. T. Lee, M. S. Odziemkowski and D. W. Shoesmith, J. Electrochem. Soc., 153(2) (2006), B33–B41. 57. C. T. Lee, Z. Qin, M. Odziemkowski and D. W. Shoesmith, Electrochim. Acta, 51(8–9) (2006) 1558–1568. 58. A. J. Gould and U. R. Evans, J. Iron Steel Inst., 155 (1947), 195. 59. D. W. Shoesmith, P. Taylor, M. G. Bailey and B. Ikeda, Electrochim. Acta, 23 (1978), 903. 60. D. W. Shoesmith, M. G. Bailey and B. Ikeda, Electrochim. Acta, 23 (1978) 1329. 61. R. C. Salvarezza, H. A. Videla and A. J. Arvia, Corros. Sci., 22(9) (1982), 815–829. 62. R. A. Yeske, ‘Measurements of corrosion rates of carbon steels exposed to alkaline sulfide environments’, in NACE Corrosion 84, paper 245, NACE, Houston, TX, 1984. 63. D. A. Wensley and R. S. Charlton, ‘Corrosion studies in kraft white liquor: potentiostatic polarization of mild steel in caustic solutions containing sulfur species’, in NACE Corrosion ’79, paper 187 and Corrosion, 36(8) (1980), 385. 64. D. C. Crowe and D. Tromans, Corrosion, 44(3) (1988), 142–148. 65. R. L. Martin and R. R. Arnand, Corrosion, 36(5) (1981), 297. 66. D. V. Vasquez Moll, R. C. Salvarezza, H. A. Videla and A. J. Arvia, Corros. Sci., 24(9) (1984), 751–767. 67. H. H. Horowitz, Corros. Sci., 23(4) (1983), 353–362. 68. S. Kannan and R. G. Kelly, ‘The role of the interaction between oxygen and catechol in the pitting corrosion of steel in alkaline sulfide solutions’, paper no. 578, presented at NACE Corrosion ’95, NACE, Houston, TX, 1995. 69. M. Jayalakshmi and V. S. Muralidharan, Corrosion (USA), 48(11) (1992), 918–923. 70. M. Jayalakshmi and V. S. Muralidharan, Corros. Rev., 12(3–4) (1994), 359–375. 71. I.I. Zamaletdinov, G. V. Khaldeev and V. M. Ermasheva, Prot. Met. (Russia) (USA), 34(2) (1998), 147–153. 72. S. M. Abd El Haleem and E. E. Abd El Aal, Corros. Eng. Sci. Technol., 43(2) (2008), 173–177. 73. S. M. Abd El Haleem and E. E. Abd El Aal, Corros. Eng. Sci. Technol., 43(3) (2008), 225–230. 74. D. Thierry and W. Sand, ‘Microbially influenced corrosion’, Chapter 16 in Corrosion Mechanisms in Theory and Practice, ed. P. Marcus. Marcel Dekker, New York, 2002. 75. S. Y. Li, Y. G. Kim; K. S. Jeon, Y. T. Kho and T. Kang, Corrosion, 57(9) (2001), 815– 828.

108

Sulphur-assisted corrosion in nuclear disposal systems

76. A. M. Pritchard, ‘An approach to assessing the effects of microbially-influenced corrosion in nuclear waste systems’, in International Workshop on Prediction of Long Term Corrosion Behaviour in Nuclear Waste Systems (Cadarache, 2001), EFC Publication No. 36, 2003. 77. F. King, Corrosion, 65(4) (2009), 233–251. 78. T. R. Jack, ‘Biological corrosion failures’, in ASM Handbook Vol. 11: Failure Analysis and Prevention, ASM International 2002. 79. W. P. Iverson, Int. Biodeter. Biodegrad., 47(2) (2001), 63–70. 80. W. Lee and W. G. Characklis, Corrosion, 49(3) (1993), 186–199. 81. W. P. Iverson, Mater. Perform., 37(5) (1998), 46–49. 82. L. De Silva; A. Bergel and R. Basseguy, Corros. Sci., 49(10) (2007), 3988–4004. 83. H. A. Videla, L. K. Herrera and R. G. Edyvean, ‘An updated overview of SRB induced corrosion and protection of carbon steel’, in NACE Corrosion 2005, Paper 488, 2005. 84. M. De Romero, Z. Duque, L. Rodriguez, O. De Rincon, O. Perez, I. Araujo and M. A. Mendez, ‘A study of microbiologically induced corrosion by SRB on carbon steel using hydrogen permeation’, in Corrosion 2003, San Diego, CA, 2003. 85. I. P. Pankhania, Biofouling, 1(1) (1988), 27–47. 86. D.T. Rickard, ‘The microbiological formation of iron sulphides’, in Proc. Contribution to Geology, Stockholm, Sweden, 67–72, 1969. 87. D. E. Hughes, ‘The microbiology of corrosion’, Chapter 2, in Corrosion, ed. L. L. Shreir, R. A. Jarman and G. T. Burstein, Volume 1, 3rd edition, Butterworth-Heinemann, Oxford, 1994.

6 The influence of chloride on the corrosion of copper in aqueous sulfide solutions J. M. Smith Kinectrics, 800 Kipling Avenue, Toronto, ON, M8Z 6C4, Canada

Z. Qin and F. King Integrity Corrosion Consulting, Nanaimo, BC, V9T 1K2, Canada

D. W. Shoesmith Department of Chemistry, University of Western Ontario, London, ON, N6A 5B7, Canada

6.1

Introduction

A proposed method of disposal of Swedish/Finnish/Canadian high-level nuclear waste is to place it in corrosion-resistant containers and bury it approximately 500 m to 1000 m deep in a granitic environment [1–3]. One option is that the containers be emplaced in bore holes and surrounded by compacted bentonite. The residual excavated space would then be backfilled with a mixture of bentonite and crushed granite. Copper is selected primarily because of its thermodynamic stability in the aqueous anoxic environments anticipated in such repositories [4], and the design of the container has been discussed elsewhere [5,6]. A model based on mixed potential principles has been developed to predict container lifetimes [7]. This model shows that corrosion during the early repository lifetime, when a significant O2 concentration exists, should be minimal, since >80% of the available O2 in the repository will be consumed by reaction with Fe(II) minerals and organic materials [7]. This model predicts a conservative maximum depth of general corrosion and pitting of 7.6 mm after 106 years [7]. However, possible components of the immediate repository environment, as well as the bentonite clay itself, contain pyrite (FeS2) and sulfate (SO42−) both of which are potential sources of sulfide, the latter following reduction by sulfate-reducing bacteria, which can convert sulfates to sulfides [8]. Various factors will ensure that there is negligible microbial activity in the vicinity of the container [8]; however, remotely produced sulfide could be slowly transported through the compacted buffer to the copper container surface. In the presence of sulfides, Cu becomes a base metal since its corrosion to produce extremely stable and insoluble copper sulfides can be sustained by the reduction of water [9,10]. Whether or not this causes significant corrosion will depend on the supply of sulfide and the protectiveness of the sulfide films formed on the copper surface. The corrosion and electrochemistry of copper and copper alloys (especially Cu/Ni alloys) in sulfide-containing solutions have been studied primarily with an emphasis 109

110

Sulphur-assisted corrosion in nuclear disposal systems

on their behavior in polluted seawater. These studies have been reviewed [11]. The possibility of sulfide-induced copper container corrosion has been considered [12–14]. Early models [13,14] were based primarily on thermodynamic principles and the assumption that the corrosion rate would be controlled by the transport of sulfide to the copper surface. More recently, the mixed potential model of King and Kolar [7] was extended to include an indirect effect of sulfide [15]. According to thermodynamic considerations, a combination of high chloride concentrations, low pH (<1), high temperature (80 to 100°), and O2 free conditions could support the general corrosion of Cu [4,16] by reaction with water Cu + 2Cl− + H+ ' CuCl2− + 1/2 H2

[6.1]

To sustain corrosion, the flux of the CuI chloride complex and the dissolved H2 away from the copper surface would displace this reversible reaction to the right. The model considers the consequences of increasing this flux, and hence the corrosion rate, by precipitating Cu2S at locations remote from the container surface by reaction with sulfide minerals (MeS) 2CuCl2− + MeS → Cu2S + Me(II) + 4Cl−

[6.2]

However, none of these models consider the direct influence of sulfide on the corrosion process, which has recently been shown to be important [16]. Our previous studies show that the properties of the sulfide surface film exert a key influence on the corrosion rate [16], and that there is a possibility that Cl− could influence the properties of this film. Electrochemical studies show that, in the absence of diffusion effects, the formation of Cu2S films on Cu proceeds via a rapid equilibrium adsorption step Cu + SH− = Cu(SH)ads + e−

[6.3]

followed by a slower, rate-determining formation of Cu2S Cu + Cu(SH)ads + SH− → Cu2S + H2S + e− −

I

[6.4] −

However, Cu is also soluble as CuCl2 and, under conditions of low [SH ] and high [Cl−], the complexation and dissolution of CuI as CuCl2− could compete with the film formation, step 6.4 Cu(SH)ads + 2Cl− → CuCl2− + SH−

[6.5]

In this study, we have investigated the influence of chloride concentration on the properties of Cu2S films on Cu, under both electrochemical and natural corrosion conditions. 6.2 6.2.1

Experimental results Electrochemical cell and anaerobic chamber

All experiments were performed in a Pyrex cell using a conventional three-electrode configuration. The counter electrode was a Pt sheet and the reference electrode a commercial saturated calomel electrode (SCE, 241 mV vs. SHE). All potentials are quoted on the SCE scale. Electrochemical experiments were performed in the open laboratory in Ar-purged solutions. Longer term natural corrosion experiments were performed within an Ar-purged anaerobic chamber (Canadian Vacuum Systems

The inﬂuence of chloride on the corrosion of copper

111

Ltd.), which was maintained at a positive pressure (2–4 mbar) by a glove box control system (MBraun). An MBraun oxygen probe, installed within the chambers gas circulation system, verified the operability of the online O2 removal catalyst, thereby ensuring experiments were performed under relatively anoxic conditions (<1 ppm in the gas phase). Electrochemical measurements were made with either a Solartron 1287 or 1284 potentiostat running Corrware software (Scriber Inc.). Electrochemical impedance spectroscopy (EIS) measurements were performed with a Solartron 1255B frequency response analyzer, using a ±10 mV potential perturbation over a frequency range from 106 Hz to 10−3 Hz. Steady-state conditions were verified by recording a small number of data points on a reverse frequency scan. Rotating disc experiments were performed using an analytical rotator (Pine Instruments) or a low noise portable rotator (Radiometer). 6.2.2

Electrode and solution preparation

Disk electrodes were fabricated from Cu machined from a bulk sample of oxygenfree, phosphorus-doped Cu, supplied by the Swedish Nuclear Fuel and Waste Management Company, Stockholm. Electrodes were then either painted with an insulting lacquer or encased in cylindrical Teflon holders with epoxy resin, to ensure that only a flat circular face was exposed to the solution. Before experiments, the electrodes were polished with SiC paper (down to 1200 grit) and then to a mirror finish with a series of alumina-silicate suspensions (to 500 Å). Samples were then rinsed and sonicated in deionized water to remove polishing residue, and dried in a stream of Ar. Before all experiments, electrodes were cathodically cleaned to remove air-formed oxides by polarizing at –1.15 V for 60 s. Electrolyte solutions were prepared with ultra-pure deionized water (18.2 Mohm. cm), obtained from a Milli-Q® Millipore System, and reagent grade chemicals. To achieve anoxic conditions, solutions used in open laboratory experiments were Ar-purged before and throughout the experiment. For experiments within the anaerobic chamber, solutions were prepared within the chamber using Ar-purged water to avoid evaporative concentration of solutions at the low pressure experienced in the chamber introduction port. 6.2.3

Surface analytical instrumentation

X-ray diffraction was performed with a Bruker D8 DISCOVER 2D GADDS microdiffractometer. Spectra were recorded using a Cu κα1 + κα2 X-ray source at a power of 40 keV and 40 mA with a beam diameter of 500 μm. Scanning electron microscopy was performed with a Hitachi S-4500 scanning electron microscope. Images were recorded with an accelerating voltage of 10 keV and a beam current of 20 μA at a 15 mm working distance. 6.3 6.3.1

Results and discussion Natural corrosion experiment

Figure 6.1 shows a series of SEM micrographs recorded on specimens allowed to naturally corrode for various time periods in solutions containing 10−3 mol/L Na2S

112

Sulphur-assisted corrosion in nuclear disposal systems

6.1 SEM micrographs recorded on Cu electrodes after natural corrosion in solutions containing 10−3 mol/L Na2S and various amounts of NaCl for various exposure times: (A) 0.1 mol/L NaCl; (a) 1 h, (b) 5 h, (c) 30 h; (B) 1.0 mol/L; (a) 1 h, (b) 5 h, (c) 30 h; (C) 5.0 mol/L; (a) 1 h, (b) 5 h, (c) 30 h

and either 0.1 mol/L, 1.0 mol/L, or 5.0 mol/L NaCl. The progression of the corrosion process in all cases is clear, and the morphology of the films formed is clearly dependent on the chloride concentration. In 0.1 mol/L solution, EDX/SEM results indicate a thin layer of sulfide exists on the surface after only 1 h exposure. The micrographs in Fig. 6.1A show the presence of a base layer with some porosity (a) and, with time, an outer deposited layer (b, c) is formed which uniformly covers the electrode surface. At higher chloride concentrations (Fig. 6.1B and C), a similar progression from an initially formed base layer to the deposition of an outer layer is also observed. However, the morphologies of the layers vary with Cl− concentration. As the Cl− concentration is increased, the rapidly formed base layer present after 1 h of exposure increases considerably in roughness and porosity and the outer deposited layer appears less well-formed and less uniform. XRD analyses of layers grown under these conditions indicate the coexistence of chalcocite (Cu2S) and digenite (Cu1.8S) in all cases. Since Cu2S is always the dominant phase, it seems likely that this is the thicker, outer deposited layer, while the thin inner

The inﬂuence of chloride on the corrosion of copper

113

base layer is Cu1.8S. A similar claim has been made by previous authors [18]. Thermodynamic calculations [18] and potential–pH diagrams [10] show that stability fields exist for a range of copper sulfides; namely, chalcocite (Cu2S), djurleite (Cu1.934S), digenite (Cu1.8S), and anilite (Cu1.75S) and, for slightly more oxidizing conditions, covellite (CuS). Of these phases, the stability field for chalcocite extends beyond (i.e. to lower potentials than) the stability field for water in the presence of sulfide at concentrations as low as 10−5 mol/L, making Cu2S the most stable phase under the conditions of the experiments described here. Given that the base layer formation is expected to involve solid state growth via a defect transport process, the presence of the highly non-stoichiometric Cu1.8S is not surprising. The slower formation of the outer layer would involve transport of CuI species over a considerable distance, i.e. from the reacting metal surface to the deposition site. Since this process involves the deposition of solution or surface diffusing species, and occurs considerably more slowly than base layer formation, the formation of the more thermodynamically stable Cu2S is expected. 6.3.2

EIS measurements

A series of natural corrosion experiments was performed in the same three solutions as described in Section 6.3.1 and an EIS spectrum was periodically recorded. The spectra recorded in 0.1 mol/L NaCl are shown in Fig. 6.2. Two clear time constants are observed and the increase in total impedance (|Z|) at the low frequency limit (10−3 Hz) indicates a small but steady increase in the impedance of the Cu/Cu2S/SH− solution interface with time. The higher frequency time constant (at ~10 Hz) is attributed to charge transfer at the Cu metal surface and the lower frequency time constant to the properties of the Cu2S surface film. As shown for the shortest and longest exposure times in Fig. 6.3, the spectra can be accurately fitted to the two time constant equivalent circuit shown in Fig. 6.4 provided that constant phase elements are used to account for the non-ideality in the capacitances. As described above, the surface film comprises a flawed base layer and a deposited outer layer. Thus, in this equivalent circuit, Rct, is the charge transfer resistance at the base of fault (pore) sites in the base layer and Cdl is the double layer capacitance. Since Cu2S films are electronically conductive, a large fraction of the potential drop across the interface will polarize the Cu2S/SH− solution interface [18]. For this reason, Cdl is placed at this interface in the equivalent circuit rather than in parallel with Rct at the base of the flaws. The nature of the flaw sites in the base layer varies with Cl− concentration (Fig. 6.1) and is not well defined. It is likely that such sites contain polarizable charge-carrying species including both solution-soluble and surface-adsorbed species. In the equivalent circuit, the properties of these flaws are represented by a parallel combination of a capacitance, Cpore, and a resistance, Rpore, the latter representing the dimensional constraints of the pore. It is acknowledged that this combination may not fully represent the process determining the interfacial properties since diffusional effects could be incorrectly incorporated into the Rpore Cpore combination. However, the inclusion of a porous structure in the equivalent circuit provided a better fit to the experimental data than a Warburg impedance used previously to fit EIS data in Cl− free solutions [16]. The values associated with the constant phase elements show that this circuit is a reasonable representation at the lowest Cl− concentration; however,

114

Sulphur-assisted corrosion in nuclear disposal systems

6.2 EIS spectra recorded after natural corrosion for various times in a solution containing 10−3 mol/L Na2S + 0.1 mol/L NaCl: (a) phase angle (h) vs. log frequency; (b) log impedance (Z) vs. log frequency

there is a significant deviation from ideality at the higher chloride concentrations, as discussed below. The parameter values obtained by fitting the spectra are shown in Figs. 6.5 and 6.6. The observed increases in Rct and especially Rpore, indicate a closing of the flaws in the deposit with time, as observed in Fig. 6.1. With the exception of the value for the shortest experiment, Cdl varies little with time as is expected if this capacitance is associated with the Cu2S/solution interface. The decrease in Cpore with time is also consistent with the closing of pores since this would decrease the number of polarizable species within these locations. The trend in all parameters towards time-independent values suggests that the deposited layer is compact and eventually enforces a steady-state protective condition. The impedance spectra obtained in the 1.0 ml/L NaCl solution are shown in Fig. 6.7. With the exception of the spectra recorded for the shortest exposure time, there is very little change in the spectra over the 38.5 h of the experiment. As observed in 0.1 mol/L NaCl, the spectra contain two distinct time constants, both at slightly higher values, ~40 Hz and ~10−1 Hz, than at the lower concentration. While these

The inﬂuence of chloride on the corrosion of copper

115

6.3 Examples of the fits of the EIS spectra in Fig. 6.6 to the electrical equivalent circuit in Fig. 6.4

6.4 Electrical equivalent circuit used to fit EIS spectra: Rct, charge transfer resistance; Cdl, double-layer capacitance; Rpore, capacitance associated with polarizable species in pores; RS, resistance of the bulk solution

6.5 Charge transfer resistance (Rct) and double-layer capacitance (Cdl) as a function of the duration of natural corrosion of Cu exposed to 10−3 mol/L Na2S + 0.1 mol/L NaCl. Values obtained by fitting the EIS spectra plotted in Fig. 6.2 to the electrical equivalent circuit shown in Fig. 6.4

116

Sulphur-assisted corrosion in nuclear disposal systems

6.6 Pore resistance (Rpore) and pore capacitance (Cpore) as a function of the duration of natural corrosion of Cu exposed to 10−3 mol/L Na2S + 0.1 mol/L NaCl. Values obtained by fitting the EIS spectra plotted in Fig. 6.2 to the electrical equivalent circuit in Fig. 6.4

spectra could be fitted to the equivalent circuit in Fig. 6.4, the fits were not as good as at the lower concentration, especially for the spectra recorded after the longest exposure time (Fig. 6.8). As indicated in the figure, the fit to this spectrum was limited to frequencies >10−1 Hz in an attempt to minimize errors. The parameter values obtained from these fits are plotted in Figs. 6.9 and 6.10. The most obvious difference between the impedance behavior in the two Cl− concentrations is found in the pore characteristics. Whereas, in 0.1 mol/L NaCl, Rpore increased from ~30 kohm.cm2 to ~100 kohm.cm2, the value recorded in 1.0 mol/L decreases from a similar early value (~35 kohm.cm2) to a final value of ~12 kohm. cm2. This indicates the occurrence of a pore opening process at the higher concentrations as opposed to the pore sealing process observed at the lower concentration, consistent with the SEM micrographs in Fig. 6.1. The increase in Cpore, while small, supports this suggestion since an increase in polarizable species would be expected within the opening pore. The spectra recorded in 5.0 mol/L NaCl are shown in Fig. 6.11. These spectra cannot be fitted to the equivalent circuit in Fig. 6.4. The inclusion in the equivalent circuit of an additional parallel RC combination does lead to visually well-fitted spectra but unreasonably large capacitance values (3 to 5 mF.cm−2) are obtained, a clear indication that diffusive processes are important at this concentration. Despite this inability to fit the spectra, a number of pertinent qualitative observations can be made. 1. The impedance at the low frequency limit (10−3 Hz) is an order of magnitude lower than at the two lower Cl- concentrations, suggesting a much lower degree of surface protection of the surface by the deposited sulfide layer. 2. If the intermediate time constant (at ~1 Hz) can be attributed to the Rpore Cpore combination, then the small decrease in impedance in this region indicates that there is a slight tendency for the pores in the base layer to open with time.

The inﬂuence of chloride on the corrosion of copper

117

6.7 EIS spectra recorded after natural corrosion for various times in a solution containing 10−3 mol/L Na2S + 1.0 mol/L NaCl: (a) phase angle (h) vs. log frequency; (b) log impedance (Z) vs. log frequency

3. The time constant at high frequencies (~103 Hz) indicates an unimpeded charge transfer at the base of the faults in the base layer. The value of Rct (estimated by visual inspection of Fig. 6.11) is in the range 1 to 3 ohm.cm2. Such a low value indicates that the metal surface at the base of flaws in the base layer is unprotected at all times by any deposition process in the flaws. This is consistent with the SEM micrographs in Fig. 6.1, which show the presence of an exposed porous base layer and a non-protective deposited layer at all times. 6.3.3

Cyclic voltammetry

The above experiments clearly show that as the Cl− concentration is increased from 0.1 mol/L to 5.0 mol/L, the Cu1.8S base layer becomes more porous and the outer deposited Cu2S layer less protective. This suggests that the coupling of reactions 6.3 and 6.5 competes with the coupling of reactions 6.3 and 6.4 as the Cl− concentration is increased. To investigate this possibility more thoroughly, a series of voltammetric scans to various anodic limits was performed in 0.1 mol/L and 5.0 mol/L NaCl (Figs. 6.12 and 6.13).

118

Sulphur-assisted corrosion in nuclear disposal systems

6.8 Examples of the fits of the EIS spectra in Fig. 6.7 to the electrical equivalent circuit in Fig. 6.4

6.9 Charge transfer resistance (Rct) and double-layer capacitance (Cdl) as a function of the duration of natural corrosion of Cu exposed to 10−3 mol/L Na2S + 1.0 mol/L NaCl. Values obtained by fitting the EIS spectra plotted in Fig. 6.2 to the electrical equivalent circuit shown in Fig. 6.4

The anodic behavior observed on the forward scan is similar at both concentrations although the currents were higher at the higher Cl− concentration. An oxidation peak with the shape typical of a diffusion-controlled process is observed. Although not shown here, scans performed on rotating disc electrodes yield considerably larger currents and a current independent of potential at positive potentials, confirming that the anodic process is controlled by the diffusive flux of SH− to the electrode

The inﬂuence of chloride on the corrosion of copper

119

6.10 Pore resistance (Rpore) and pore capacitance (Cpore) as a function of the duration of natural corrosion of Cu exposed to 10−3 mol/L Na2S + 1.0 mol/L NaCl. Values obtained by fitting the EIS spectra plotted in Fig. 6.2 to the electrical equivalent circuit shown in Fig. 6.4

surface. However, the reverse voltammetric scans show that the films formed anodically in the two solutions are significantly different. At the lower Cl− concentration, scans to various anodic potential limits produce a single reduction peak at a potential between –1.0 V and –1.1 V. The exact position of this peak depends on the extent of sulfide film formation, which is proportional to the area under the cathodic reduction peak. The presence of a single peak indicates the formation of a compact, dense layer of sulfide in good electrical contact with the electrode surface, which is consistent with the SEM observations in Fig. 6.1. Although unimportant in terms of the present study, it is worth noting that the anodic formation and subsequent reduction of this layer leads to a very large enhancement of the current for water reduction for potentials <–1.3 V. This indicates that the reduction of the sulfide film leads to a fine particulate Cu layer with a very large surface area. For the higher Cl− concentration, a similar single reduction peak is observed providing the anodic scan limit is not too positive (Fig. 6.13 (a, b)). This is consistent with the formation of a Cu1.8S base layer in good electrical contact with the Cu surface and hence reduced at a similar potential to the layer formed at the lower Cl− concentration (–1.0 V to –1.1 V). However, when the anodic potential limit is extended to more-positive values, not only does this peak become larger but a second reduction peak is observed at more-negative reduction potentials. The observation of two distinct reduction peaks suggests the reduction of two separate layers as opposed to the single compact layer formed at the lower Cl− concentration. Thus, the reduction occurring at more negative potentials can be attributed to a more loosely attached deposit formed on top of the base layer but not necessarily within the pores of the base layer. In this regard, the observed behavior is consistent with the SEM observations in Fig. 6.1, for natural corrosion experiments. For potential scans to more-positive anodic limits, the anodic film formation process is under diffusion control. This means that the SH− concentration at the

120

Sulphur-assisted corrosion in nuclear disposal systems

6.11 EIS spectra recorded after natural corrosion for various times in a solution containing 10−3 mol/L Na2S + 5.0 mol/L NaCl: (a) phase angle (h) vs. log frequency; (b) log impedance (Z) vs. log frequency

metal surface will be very low compared with the Cl− concentration, especially in the 5 mol/L NaCl solution. Thus, the maintenance of open porosity in the base layer leading to the formation of a less compact and less-readily reducible outer deposit in 5.0 mol/L is consistent with the coupling of reactions 6.3 and 6.5. This would lead to the transport of CuI species away from the metal surface and their deposition as Cu2S on the outer surface where the SH− concentration would approach bulk solution values. This enhanced CuI transport via soluble CuCln(n−1)− species does not appear to compete with sulfide film formation at 0.1 mol/L NaCl leading to the formation of a more compact protective sulfide deposit. In an attempt to determine whether dissolution of CuI plays an important role in the overall anodic process at higher [Cl−], the anodic (QA) and cathodic (QC) charges were calculated by numerical integration of the anodic and cathodic areas in Figs. 6.12 and 6.13. The ratio QC/QA, which is a measure of the fraction of anodic charge recovered by reduction of the sulfide films formed during the anodic part of the scan, is plotted as a function of anodic potential limit in Fig. 6.14. For short anodic excursions (i.e. to potentials ≤ –0.90 V), this ratio is small, indicating that dissolution, as opposed to sulfide film formation, is dominant at the low anodic

The inﬂuence of chloride on the corrosion of copper

121

6.12 Cyclic voltammograms recorded on Cu immersed in a solution containing 10−3 mol/L Na2S + 0.1 mol/L NaCl. The plots were recorded at a potential scan rate of 2 mV/s and reversed at (a) –0.92 V; (b) –0.90 V; (c) –0.85 V; (d) –0.80 V; (e) –0.70 V. The curves are offset by –0.4 mA/cm2, as indicated by the vertical arrow

6.13 Cyclic voltammograms recorded on Cu immersed in a solution containing 10−3 mol/L Na2S + 5.0 mol/L NaCl. The plots were recorded at a potential scan rate of 2 mV/s and reversed at (a) –0.95 V; (b) –0.93 V; (c) –0.90 V; (d) –0.85 V; (e) –0.80 V; (f) –0.75 V; (g) –0.70 V. The curves are offset by –0.25 mA/cm2, as indicated by the vertical arrow

122

Sulphur-assisted corrosion in nuclear disposal systems

6.14 The cathodic to anodic charge ratio (QC/QA) as a function of anodic scan limit. The charges were calculated by integrating the anodic reactions (QA) and cathodic film reduction reactions (QC) of the voltammetric scans plotted in Figs. 6.12 and 6.13

currents achieved up to these potential limits. For more-positive potential limits, sulfide film formation dominates and the ratio approaches 1, i.e. the great majority of oxidized Cu becomes trapped in the Cu1.8S/Cu2S surface layers and is hence available for cathodic reduction on the reverse scan. While this last plot clearly suggests that dissolution contributes significantly at potentials close to those prevailing in natural corrosion experiments (≤ –0.95 V), it does not confirm that higher Cl− concentrations lead to more extensive dissolution, since no significant difference in the QC/QA ratio is observed between the two solutions. 6.4 • • • • •

Summary and conclusions The possibility that the corrosion of copper in aqueous sulfide solutions is influenced by chloride present in groundwaters in a nuclear fuel waste repository has been investigated under natural corrosion and electrochemical conditions. Sulfide films were observed to grow initially as a thin base layer of Cu1.8S which developed porosity allowing the further growth of a much thicker outer deposited layer of Cu2S. In 0.1 mol/L chloride solution, this outer layer was dense and compact and eventually sealed the pores in the base layer, a situation which would lead to low corrosion rates. When the chloride concentration was increased to 1.0 mol/L, the base layer became more porous and the outer layer less dense and compact, and the pores in the base layer were not sealed. At 5.0 mol/L chloride, EIS data indicate that open pores are present in the base layer allowing rapid reaction to produce CuI species (as surface adsorbed CuSH species), a situation that could lead to a significant increase in corrosion rate.

The inﬂuence of chloride on the corrosion of copper •

123

A possible explanation for this influence is that, when the chloride to sulfide concentration ratio is high, CuI species, originally formed as surface adsorbed CuSH species can be complexed by chloride to produce soluble species (CuCln(n−1)−). Transport of these species away from the metal surface would maintain porosity in the base layer and limit the deposition of the outer protective Cu2S deposit.

Acknowledgements This research was funded by the Swedish Nuclear Fuel and Waste Management Company (SKB), Stockholm. We are grateful to Surface Science Western (University of Western Ontario) for the use of their scanning electron microscope. References 1. F. King, Corrosion of Copper in Alkaline Chloride Environments. TR-02-25. Swedish Nuclear Fuel and Waste Management Company, Stockholm, 2002. 2. J. McMurry, D. A. Dixon, J. D. Garroni, B. M. Ikeda, S. Stroes-Gascoyne, P. Baumgartner and T. W. Melnyk, Evolution of a Canadian Deep Geologic Repository: Base Scenario, Ontario Power Generation Report No. 06819-REP-01200-10092-R00, 2003. 3. J. McMurry, B. M. Ikeda, S. Stroes-Gascoyne and D. A. Dixon, Evolution of a Canadian Deep Geologic Repository: Defective Container Scenario, Ontario Power Generation Report No. 06819-REP-01200-10127-R00, 2004. 4. I. Puigdomenech and C. Taxen, Thermodynamic Data for Copper. Implications for the Corrosion of Copper under Repository Conditions. TR-0-13. Swedish Nuclear Fuel and Waste Management Company, Stockholm, 2000. 5. C. Anderson, Development of Fabrication Technology for Copper Canisters with Cast Inserts. Status Report in August 2001, TR-02-07. Swedish Nuclear Fuel and Waste Management Company, Stockholm, 2002. 6. P. Maak, Used Fuel Container Requirements, Ontario Power Generation Report No. 06819-PDR-01110-R01, 2001. 7. F. King and M. Kolar, The Copper Container Corrosion Model Used in AECL’s Second Case Study, Ontario Power Generation Report No. 06819-REP-01200-10041-R00, 2000. 8. K. Pedersen, Microbial Processes in Radioactive Waste Disposal, TR-00-04. Swedish Nuclear Fuel and Waste Management Company, Stockholm, 2004. 9. F. King and S. Stroes-Gascoyne, ‘Microbially influenced corrosion of nuclear fuel waste disposal containers, in Proc. 1995 International Conference on MIC, 35/1–35/14. NACE International, Houston, Texas, USA, 1995. 10. M. Pourbaix and A. Pourbaix, Geochim. Cosmochim. Acta 56 (1992), 3157. 11. J. M. Smith, ‘The corrosion and electrochemistry of copper in aqueous, anoxic sulphide solutions’, Ph.D. thesis, The University of Western Ontario, London, Canada, 2007. 12. SKB, Final Storage of Spent Nuclear Fuel. KBS3, Volumes I–IV. Swedish Nuclear Fuel and Waste Management Company, 1983. 13. L. Werme, P. Sollin and N. Kjellbert, Copper Canisters for Nuclear High Level Waste Management Company, Report No. TR-92-26, 1992. 14. F. King, L. Ahonen, C. Taxen, U. Vuorinen and L. Werme, Copper Corrosion Under Expected Conditions in a Deep Geologic Repository, Report No. TR-01-23. Swedish Nuclear Fuel and Waste Management Company, Stockholm, 2001. 15. B. Beverskog and I. Puigdomenech, Pourbaix Diagrams for the System Copper–Chlorine at 5–100°C, SKI Report No. 98:19. Swedish Nuclear Power Inspectorate, 1998. 16. J. M. Smith, Z. Qin, F. King, L. Werme and D. W. Shoesmith, Corrosion, 63 (2007), 135. 17. M. R. G. de Chialvo and A. J. Arvia, J. Appl. Electrochem., 15 (1985), 685–696. 18. K. Rahmouni, M. Keddam, A. Srhiri and H. Takenouti, Corros. Sci., 47 (2005), 3249.

7 Interactions between sulphide species and components of rust Ph. Refait, J. A. Bourdoiseau, M. Jeannin, R. Sabot and C. Rémazeilles Laboratoire d’étude des matériaux en milieux agressifs (LEMMA), EA-3167, Université de La Rochelle, Fédération de recherche en environnement et développement durable, FR CNRS 3097, Bât. Marie Curie, Av. M. Crépeau, F-17042 La Rochelle cedex 01, France

J. A. Bourdoiseau ANDRA, Parc de la Croix Blanche, 1/7 rue Jean Monnet, F-92298 Châtenay-Malabry, France

7.1

Introduction

This study is concerned with nuclear waste disposal. In France, it is envisaged that high-level radioactive wastes will be confined in a glass matrix, stored in a stainless steel canister, itself placed in a carbon steel overpack. The wastes will then be stored at a depth of ∼500 m in a deep geological repository, drilled in a very stiff (indurated) clay (argillite) formation. The kinetics of corrosion expected for the overpack in this disposal concept are low and will remain low if the protective rust layer that will develop initially on the steel surface remains undamaged. Local changes of the physico-chemical conditions may however degrade this layer, inducing localised corrosion processes that could, provided that they are autocatalytic, lead to catastrophic kinetics of corrosion, similar to those encountered on steel structures in marine environments [1], often associated with sulphate reducing bacteria (SRB) [2,3]. In this case, it is generally assumed that the localised nature of the corrosion process is a consequence of the heterogeneous nature of the biofilm formed on the metal surface. The local development of colonies of SRB induces large local sulphide concentrations and acidic conditions at the metal/electrolyte interface. However, if the bacterial growth occurs once the metal surface is already covered by a thick rust layer, the modifications of the environment induced by the metabolic activity of microorganisms may not reach the metal surface. This could be the case if the rust layer is compact and adherent, not porous and/or if the interactions between the sulphide species produced by SRB and the components of rust prevent the transport of the sulphide species towards the metal. More generally, the corrosion rate of steel depends on the properties of the rust layer. For instance, if cracks are present, the aggressive species and the oxidising agent can reach the metal surface more easily; if some components of the rust layer are conductors, the cathodic reaction can take place in the outer part of the rust layer so that the oxidising agent would not have to reach the metal surface, etc. Therefore, the reactivity of the rust layer with respect to sulphide species should similarly play a role on the corrosion rate of the underlying metal. 124

Interactions between sulphide species and components of rust

125

So, to estimate if localised corrosion phenomena could be induced by sulphide species on the steel surface of the overpack, the interactions between sulphide and the main expected components of rust have to be studied. Two kinds of effects could be involved, depending on the oxidation number of Fe: (1) the reductive effect of S(–II) could lead to a dissolution, at least partial, of the Fe(III) compounds (e.g. α-FeOOH or γ-Fe2O3) and (2) the low solubility of iron sulphides could lead to a dissolution of Fe(II) compounds (e.g. FeCO3) and reprecipitation of FeS. In the present study, goethite α-FeOOH and lepidocrocite γ-FeOOH were prepared. The powders obtained were placed in Na2S solutions under anoxic conditions. Various [Fe(III)]/[S(–II)] ratios were considered. The products resulting from the interactions of S(–II) with Fe(III) oxyhydroxides were analysed by micro-Raman spectroscopy. Similarly, suspensions of FeCO3 were synthesised from FeSO4 and NaHCO3 solutions. The products obtained after addition of Na2S were analysed by micro-Raman spectroscopy and X-ray diffraction. These experiments were performed at room temperature. Micro-Raman spectroscopy was chosen as the main analytic tool for this study since it is the most suitable method for the characterisation of the corrosion products generated by localised corrosion processes. As the Raman data for iron sulphides are scarce, it proved necessary to synthesise and characterise iron sulphides to interpret our results. It was demonstrated that the Raman spectrum of FeS varied with dehydration, crystallisation and oxidation [4]. New results and, in particular, results related to synthesis and ageing of mackinawite at 80 and 96°C are described in this paper. 7.2 7.2.1

Experimental methods Synthesis of components of rust

Fe(II) sulphides were precipitated from FeCl2 · 4H2O and Na2S · 8–9H2O solutions. All chemical products were provided by Aldrich® with a 99% minimum purity. The first set of precipitates was obtained at room temperature by mixing 100 mL of the Fe(II) solution with 100 mL of the Na2S solution. The Fe(II) concentration in the resulting 200 mL was set at 0.1 mol L−1. Only the Na2S concentration was varied and the results are then given as functions of the concentration ratio Fe/S = [FeII]/[S−II] = [FeCl2]/[Na2S]. Ratios of 1 and 3/2 were considered here. The precipitates were analysed once obtained or after various ageing times at room temperature under anoxic conditions. In this last case, the whole suspension was poured into a flask. The flask was completely filled with the suspension and sealed carefully to avoid infiltration of air. During the ageing period, the suspension was kept static. FeS samples were also prepared the same way at 80°C and aged for 50 days at 96°C. The ratio Fe/S was set at 1. The pH of the resulting suspensions was about 6.5–7. Siderite can also be prepared by precipitation from a Fe(II) solution. Therefore, the competition between the formation of FeCO3 and that of FeS could be studied by mixing FeSO4 · 7H2O, Na2S · 8–9H2O and NaHCO3 solutions. The concentrations of the reactants, given with respect to the 250 mL of solution after mixing, were [Fe(II)] = 0.08 mol L−1, [S(–II)] = 0.026 mol L−1 and [HCO3−] = 0.08 mol L−1. The precipitate obtained was aged for 6 weeks at room temperature before analysis. An alternative experiment was envisioned. FeCO3 was precipitated first from FeSO4 · 7H2O and NaHCO3 solutions. Na2S · 8–9H2O was added to the suspension

126

Sulphur-assisted corrosion in nuclear disposal systems

1 week later. The concentrations chosen were the same as above. The FeCO3 + Na2S system was then aged for 8 days before the analysis of the precipitate. Fe(III) oxyhydroxides can be prepared by oxidation of Fe(II) precipitates. Goethite could be obtained by mixing FeSO4 · 7H2O and NaOH solutions, with [FeSO4 · 7H2O] = 0.50 mol L−1 and [NaOH] = 0.20 mol L−1. Its formation resulted from the oxidation in air at 45°C of the initial Fe(II) precipitate. Details can be found in Ref. 5. Lepidocrocite was obtained using a similar procedure, after the oxidation at 25°C of the precipitate obtained by mixing FeCl2 · 4H2O and NaOH solutions with [FeCl2 · 4H2O] = 0.24 mol L−1 and [NaOH] = 0.40 mol L−1 [6]. During the oxidation, the suspensions were stirred vigorously using a magnetic rod turning at 760 rpm. Once obtained, the Fe(III) oxyhydroxides were filtered and dried. The resulting powder was washed thoroughly so as to remove the salts (FeSO4, FeCl2, Na2SO4 or NaCl) remaining with the FeOOH compound. The precipitate was then dried once more. Then 1 g of each FeOOH phase was set in a Na2S solution and the FeOOH + Na2S system was aged at room temperature. 7.2.2

Characterisation of the solids

Raman analysis of synthetic precipitates was carried out with a Jobin Yvon High Resolution Raman spectrometer (LabRAM HR) equipped with a microscope (Olympus BX 41) and a Peltier-based cooled charge coupled device (CCD) detector. The spectra were recorded with LabSpec acquisition software at room temperature with a resolution of about 2 cm−1. Excitation was provided by a He–Ne laser (632.8 nm). Its power was varied between 1.94 and 0.07 mW to prevent excessive heating that could induce transformation of the analysed sample. The spot under the ×50 objective had a diameter of ∼3 μm. A procedure was designed to avoid any oxidation of the Fe(II) precipitates: a small amount of the suspension was sampled at the bottom of the flask, where the solid had accumulated by decantation. It was immediately poured in a small quartz cell (40×10×1 mm). The cell was filled with the suspension, sealed carefully and set under the microscope of the Raman apparatus for the analysis. The action of S(–II) on FeOOH phases can lead to a partial dissolution of the solid phase. This phenomenon, called reductive dissolution, can be written as follows, if we consider that the oxidation of S(–II) leads to elemental sulphur: 2 FeIIIOOH + HS− + H2O → 2Fe2+aq + S(0) + 5OH−

[7.1]

In this case, it was necessary to analyse the suspension as it was, without filtration, by Raman spectroscopy, so as to obtain information on both dissolved and solid species. The precipitates were also analysed by powder X-ray diffraction (XRD) with a Bruker-AXS D8 Advance, using the Cu Kα wavelength (λ = 0.15406 nm) and a vertical axis counter diffractometer in Bragg-Brentano geometry. They were filtered just before the analysis. Since Fe(II) compounds are sensitive to the oxidising action of air, they were sheltered by a plastic membrane during filtration. The wet pastes obtained were immediately placed on the sample holder and coated with glycerol before the analysis. This procedure limits oxidation of the sample during the acquisition of the pattern.

Interactions between sulphide species and components of rust 7.2.3

127

Analyses of iron archaeological artefacts immersed in seawater

In the recently published work devoted to mackinawite [4], Raman spectra of synthetic samples were compared to some of those obtained during the analysis of the rust layers of a Roman iron ingot. Another spectrum is presented here. It will be discussed in the light of the new information described in the present paper. The ingot was raised from a shipwreck, dated from the Gallo-Roman period, immersed at a depth of 11 m and 1.5 miles from the coast near Les Saintes Maries de la Mer (France) in the Mediterranean Sea. It was discovered by A. Chabaud in 1996 [7]. The ingot examined for the specific study described here was 30 cm long, 6 cm wide and 4 cm thick. It was partially covered by calcareous concretions and sediments. It was never exposed to the atmosphere so as to avoid modification of the corrosion products before the analyses. More details about these ingots can be found in Ref. 8. For these analyses, the ingot was quickly dried, mounted in epoxy resin and cut so that cross sections could be observed. It was polished in hexane with SiC papers with decreasing particle size (down to 5 μm). The morphology of the rust layers was observed using optical and scanning electron microscopes (SEM). Compositions were determined by energy dispersive spectroscopy (EDS) coupled to SEM (acceleration voltage: 20 kV). Raman spectroscopy was used to characterise the iron sulphides present inside the rust layers. 7.3 7.3.1

Results and discussion Precipitation and ageing of FeS

The initial precipitate obtained was, whatever the value of the Fe/S concentration ratio, nanocrystalline mackinawite [4,9–11]. The corresponding Raman spectrum shows two sharp peaks at ∼208 and ∼282 cm−1 [4,12,13]. When Fe/S ≥ 1, crystallisation takes place and the Raman spectrum changes. Figure 7.1 illustrates these effects

7.1 Raman spectra of the iron(II) sulphide obtained by mixing FeCl2 · 4H2O and Na2S solutions, with [Fe(II)]/[S(–II)] = 3/2 at 25°C after various ageing times at room temperature

128

Sulphur-assisted corrosion in nuclear disposal systems

for the precipitate obtained at Fe/S = 3/2. For the Raman spectrum obtained after an ageing time of 95 min, the bands at 208 and 282 cm−1 are still predominant. But two other bands can already be seen. A small one develops at 256 cm−1, and a shoulder at 290 cm−1 appears on the right side of the main peak. The Raman spectrum obtained after 6 days of ageing is that of crystallised mackinawite. The additional band at 256 cm−1 is now clearly visible, and the main peak has shifted from 282 to 297 cm−1. The peak at 208 cm−1 is not affected by crystallisation and is still present. The two small peaks at 125 and 355 cm−1 are due to a slight oxidation of iron, and the presence of Fe(III)–S(–II) bonds [4]. Since the temperature of the steel overpack is expected to reach 100°C, similar experiments were performed between 80°C and 96°C. The initial precipitate, obtained by mixing solutions of FeCl2 · 4H2O and Na2S with Fe/S = 1 at 80°C, was analysed by X-ray diffraction. The pattern (Fig. 7.2a) is still characteristic of nanocrystalline

7.2 XRD patterns (Cu Kα wavelength) of the iron(II) sulphide obtained by mixing FeCl2 · 4H2O and Na2S solutions, with [Fe(II)]/[S(–II)] = 1 at 80°C. (a) Initial precipitate and (b) precipitate obtained after 50 days of ageing at 96°C

Interactions between sulphide species and components of rust

129

mackinawite: only the three main diffraction lines can be clearly identified. So a high temperature does not change the initial state of the precipitated iron(II) sulphide. Another sample, prepared the same way at 80°C, was aged for 50 days at 96°C under anoxic conditions, at pH ~ 6.5–7. Its XRD pattern, presented in Fig. 7.2b, is that of well crystallised mackinawite. The diffraction lines are much sharper than those of samples aged at 25°C ([4] and see also Fig. 7.3a). So, the main effect of temperature is to accelerate the crystallisation of mackinawite from its nanocrystalline state, and to improve its crystallinity. Since larger and ordered particles are more thermodynamically stable, the increase in the temperature from 25°C to 96°C would then be favourable to the formation of mackinawite.

7.3 XRD pattern (Cu Kα wavelength) (a) and Raman spectrum (b) of the precipitate obtained by mixing FeSO4 · 7H2O, Na2S and NaHCO3 solutions, aged 6 weeks at 25°C. Shkl and Mhkl are the diffraction lines of siderite and mackinawite, respectively, with the corresponding Miller index

130 7.3.2

Sulphur-assisted corrosion in nuclear disposal systems Formation of FeS in carbonated media

Among the anions present in groundwater that would leach from the argillite to reach the surface of the steel overpack, the only one that promotes the formation of a rather insoluble Fe(II) compound is HCO3− Fe2+ + HCO3− → FeCO3 + H+

[7.2]

The equilibrium conditions at 25°C are given in Ref. 14: [Fe ][HCO3 ]/[H ] = 10−0.47. Note that the metabolic activity of sulphate reducing bacteria (SRB) can also lead to the oxidation of organic matter into carbonate species. The presence of S(–II) dissolved species in the groundwater, whether they are produced by the dissolution of pyrite in argillite or by SRB, could induce the formation of FeS instead of FeCO3. The reaction is: 2+

Fe2+ + HS− → FeS + H+

−

+

[7.3]

The equilibrium conditions at 25°C between the dissolved species and ‘amorphous’ FeS, which is in fact nanocrystalline mackinawite, are given in Ref. 15: [Fe2+][HS−]/[H+] = 10−3.00. This shows that even in its nanocrystalline state, mackinawite is less soluble than siderite. The precipitate obtained by mixing FeSO4, Na2S and NaHCO3 solutions such that Fe/S = 3 and Fe/HCO3 = 1 illustrates this difference in solubility. Figure 7.3 presents the XRD pattern and the Raman spectrum of this precipitate after 4 weeks of ageing at 25°C. The XRD pattern (Fig. 7.3a) is composed of very sharp diffraction lines, due to siderite, and of broad lines, due to mackinawite. In perfect agreement, the Raman spectrum (Fig. 7.3b) displays the main peak of siderite, at 1085 cm−1 [16], and the characteristic peaks of crystallised mackinawite, found here at 211, 257 and 299 cm−1. The peak at 355 cm−1 is due to a slight oxidation of mackinawite [4] and that at 984 cm−1 corresponds to SO42− ions in solution [17]. In this case, the carbonate species concentration was sufficient to precipitate all of the Fe(II) cations in the form of FeCO3. But a large amount of FeS was obtained. Since the Fe/S ratio was equal to 3, it can be proposed that siderite was obtained only because all S(–II) dissolved species were consumed by the precipitation of FeS. The composition of the solid obtained would then be FeS + 2FeCO3. Another way to illustrate the lower solubility of FeS is to precipitate FeCO3 first, from FeSO4 and NaHCO3 solutions, then to add Na2S to the system. In the experiment described here, the suspension of FeCO3 was aged 1 week at room temperature before the addition of Na2S. Its Raman spectrum at that time is spectrum (a) of Fig. 7.4. Only two sharp peaks are clearly seen, the main one of siderite at 1085 cm−1, and that of SO42− ions in solution at 983 cm−1. After addition of Na2S, the Fe/S and Fe/HCO3 ratios were 3 and 1, respectively. The suspension was aged for one more week. The corresponding Raman spectrum is spectrum (b) of Fig. 7.4. Except for the peak of SO42− ions, it only shows those of mackinawite. The transformation of siderite seems to be complete while there was not enough sulphide species to precipitate integrally the Fe2+ cations in the form of FeS. It can be proposed that the mackinawite crystallites, forming at the siderite/solution interface, have formed a thick FeS layer around the siderite crystals, preventing the laser beam from interacting with FeCO3. 7.3.3

Reductive dissolution of Fe(III) oxyhydroxides

Another effect of sulphide species could be a deterioration of the protective layer of Fe(III) compounds formed on the steel surface of the overpack during the initial oxic

Interactions between sulphide species and components of rust

131

7.4 Raman spectra (a) of siderite precipitated from FeSO4 · 7H2O and NaHCO3 solutions, aged 7 days and (b) of the solid present in the same suspension after addition of a Na2S solution and 7 days of ageing at room temperature

conditions. This effect is due to the reduction of Fe(III) by S(–II). So as to illustrate the consequence of this phenomenon, called reductive dissolution, 1 g of lepidocrocite was placed in a Na2S solution so that the Fe/S concentration ratio was equal to 1.3. The surface of the particles immediately turned to green and, after 24 h, the suspension became a green solution, with no solid phase visible. Its pH was 11.9; it increased since the dissolution of FeOOH produced OH− ions. This solution was analysed by Raman spectroscopy. In agreement with the visual observation, the spectrum (Fig. 7.5a) does not display any Raman band of lepidocrocite, and in particular, its main sharp peak at 250 cm−1 [18,19]. It is composed of a sharp peak at 360 cm−1, and additional smaller peaks at 120, 287 and 321 cm−1. A broad band is also visible around 640 cm−1. Such a spectrum has not yet been reported. Note that after evaporation, a green powder could be obtained from this solution. Its Raman spectrum was similar to that of the whole solution. A similar experiment was achieved using 1 g of goethite instead of lepidocrocite. The Fe/S concentration ratio was also changed and increased up to 3. As observed with lepidocrocite, the particles turned to green while immersed in the Na2S solution. This indicates the reduction of Fe(III) to Fe(II) at the surface of the goethite particles. But after 1 month, the particles were still present. The pH of the solution was measured at 9.5. The Raman spectrum of the suspension is displayed in Fig. 7.5b. As expected, the Raman bands of goethite, at 243, 299, 385, 479 and 550 cm−1 [18,19], are present, confirming that the reductive dissolution was only partial. The Raman bands of the species produced by this reductive dissolution are the same as for lepidocrocite dissolution: a main band at 362 cm−1, and others at 120, 290, 320 and 640 cm−1. These experiments confirm that interactions between Fe(III) compounds and S(–II) species can induce important transformations that may modify the protective

132

Sulphur-assisted corrosion in nuclear disposal systems

7.5 Raman spectra of the species obtained by reductive dissolution of FeOOH phases in Na2S solutions. (a) Lepidocrocite left for 24 h in a solution so that Fe/S = 1.3 and (b) goethite left for 4 weeks in a solution so that Fe/S = 3. G are the Raman bands of goethite

properties of rust layers. At the present time, it is not possible to describe the mechanisms of these interactions, since the Fe and S containing species they produce are not identified. It is finally interesting to note that similar spectra were obtained during the analysis of the rust layers of Roman iron ingots that remained for 2000 thousand years at the bottom of the Mediterranean Sea. Such a spectrum is presented in Fig. 7.6. It is essentially composed of an intense peak at 350 cm−1 and a smaller one at 322 cm−1. A broad band is also visible at 640 cm−1. These are typical features of the compound obtained via the reductive dissolution of FeOOH phases (see Fig. 7.5). The other small peaks may correspond to other phases. Note that the spectrum in Fig. 7.6 is also very similar to that of greigite, Fe3S4 [20].

Interactions between sulphide species and components of rust

133

Figure 7.7 shows the rust layer surrounding the zone where this Raman spectrum was obtained. The EDS analysis (Fig. 7.7d) revealed a locally important S concentration. The matrix surrounding this Fe and S rich zone was essentially made of siderite and magnetite, but S was found scattered all over this region. Closer to the iron surface, the rust layer is mainly composed of β-Fe2(OH)3Cl (with spots of akaganéite β-FeOOH) which indicates that this region of the ingot was maintained in reductive and anoxic conditions. Note that these S-rich regions appeared only here and there in the outer part of the rust layer. This confirms the localised nature of the process, in this case due to the sulphide species biogenerated by SRB. 7.4

Conclusions

Micro-Raman spectroscopy is one of the main analytical tools for the identification of the products resulting from localised corrosion processes. But the characterisation of iron sulphides is complex, as their Raman spectral features proved to be very sensitive to their physico-chemical state. As an example, three Raman spectra characterised mackinawite, the iron sulphide that precipitates from Fe(II) and S(–II) dissolved species. One spectrum is for stoichiometric (FeS) well crystallised mackinawite, one is for nanocrystalline mackinawite, formerly denoted as ‘amorphous FeS’, and the last one is for Fe(III)-containing mackinawite [4]. At the present time, various Raman spectra of Fe and S containing compounds present in rust layers of archaeological artefacts cannot be interpreted. The preliminary study of the interactions between siderite, lepidocrocite, or goethite and S(–II) species confirmed that important modifications of the initially protective rust layers could take place. Dissolution of FeCO3 in sulphide solutions led to the formation of mackinawite crystallites around the large crystals of FeCO3.

7.6 Raman spectrum obtained in a S-rich local zone of the rust layer of a Roman iron ingot

134 Sulphur-assisted corrosion in nuclear disposal systems

7.7 SEM analysis of a Roman iron ingot. (a) SEM photograph of an expanded zone, showing remaining iron (left), rust layer and resin surrounding the sample (upper right corner), (b) S mapping of the rectangular zone shown in (a): the more abundant S is, the lighter the image, (c) detail of the square zone shown in (a), and (d) EDS spectrum of the circular zone shown in (c). See Ref. 8 for details

Interactions between sulphide species and components of rust

135

The reductive dissolution of FeOOH phases also led to Fe and S containing species that have not yet been identified. However, their Raman spectrum is similar to that obtained from the Raman analysis of the rust layers of a Roman iron ingot left for 2000 years at the bottom of the Mediterranean Sea, and also similar to that of greigite, Fe3S4 [20]. The next step in this study should be realised with steel electrodes covered by FeCO3, α-FeOOH and FeCO3/ α-FeOOH layers, at temperatures around 80°C. The effects of a confined and localised source of sulphide, simulating the localised influence of SRB observed on archaeological objects, could thus be examined. Moreover, the understanding of Raman signatures of FeS and related compounds is not yet complete and research in this direction will be continued. Acknowledgements This work was supported by the French national radioactive waste management agency (ANDRA) as part of the PhD thesis of J.A. Bourdoiseau. The archaeological ingot was studied in the frame of the French Culture Ministry research programme ODéFA. References 1. J. M. Moulin, E. Marsh, W. T. Chao, R. Karius, I. Beech, R. Gubner and A. Raharinaivo, Prevention of Accelerated Low-Water Corrosion on Steel Piling Structures due to Microbially Influenced Corrosion Mechanisms, EUR 20043EN, European Commission, Final Report, 2001. 2. H. A. Videla and L. K. Herrera, Int. Microbiol., 8 (2005), 169–180. 3. I. B. Beech, S. A. Campbell and F. C. Walsh, ‘Microbial aspects of the low water on the corrosion of carbon steel’, in Proceedings of the 12th International Corrosion Congress, Vol. 5B, NACE, Houston, TX, 1993. 4. J. A. Bourdoiseau, M. Jeannin, R. Sabot, C. Rémazeilles and Ph. Refait, Corros. Sci., 50 (2008), 3247–3255. 5. F. Gilbert, Ph. Refait, F. Lévêque, C. Remazeilles and E. Conforto, J. Phys. Chem. Solids, 69 (2008), 2124–2130. 6. Ph. Refait and J.-M. R. Génin, Corros. Sci., 34 (1993), 797–819. 7. L. Long, ‘Au Large des Saintes Maries de la Mer’, in Bilan scientifique du Département des Recherches Archéologiques Subaquatiques et Sous Marines (DRASSM) de 2002, 51–57, 2003. 8. E. Guilminot, D. Neff, C. Rémazeilles, S. Reguer, F. Nicot, Ph. Dillmann, F. Mirambet, Ph. Refait, L. Bertrand, N. Huet, F. Mielcarek and J. Rebière, ‘Dechlorination of archaeological iron artefacts: dechlorination efficiency assessment assisted by physicochemical analytical high-tech methods’, in Proceedings of the 15th triennial ICOM-CC conference, New Delhi, India, 22–26 September 2008, Vol. I, 435–443. 9. M. Wolthers, S. J. Van der Gaast and D. Rickard, Am. Min., 88 (2003), 2007–2015. 10. H. Y. Jeong, J. H. Lee and K. F. Hayes, Geochim. Cosmochim. Acta, 72 (2008), 493–505. 11. H. Ohfuji and D. Rickard, Earth Planet. Sci. Lett., 241 (2006), 227–233. 12. A. Boughriet, R. S. Figueiredo, J. Laureyns and P. Recourt, J. Chem. Soc., Faraday Trans., 93 (1997), 3209–3215. 13. E. B. Hansson, M. S. Odziemkowski and R. W. Gillham, Corros. Sci., 48 (2006), 3767–3783. 14. J. Bruno, P. Wersin and W. Stumm, Geochim. Cosmochim. Acta, 56 (1992), 1149–1155. 15. W. Davison, N. Phillips and B. J. Tabner, Aquat. Sci., 61 (1999), 23–43.

136

Sulphur-assisted corrosion in nuclear disposal systems

16. M. C. Bernard, S. Duval, S. Joiret, M. Keddam, F. Ropital and H. Takenouti, Prog. Org. Coat., 45 (2002), 339–404. 17. J. Dünnwald and A. Otto, Corros. Sci., 29 (1989), 1167–1176. 18. D. L. A. De Faria, S. V. Silva and M. T. D. Oliveira, J. Raman Spectrosc., 28 (1997), 873–878. 19. D. Neff, L. Bellot-Gurlet, Ph. Dillmann, S. Reguer and L. Legrand, J. Raman Spectrosc., 37 (2006), 1228–1237. 20. C. Rémazeilles, M. Saheb, D. Neff, E. Guilminot, K. Tran, J.-A. Bourdoiseau, R. Sabot, M. Jeannin, H. Matthiesen, Ph. Dillmann and Ph. Refait, J. Raman Spectrosc., in press.

8 Experimental investigation of the impact of microbial activity on the corrosion resistance of candidate container materials Virginia Madina, Iñaki Azkarate and Laura Sánchez INASMET-Tecnalia, Mikeletegi Pasealekua, 2 E-20009 San Sebastián, Spain

Miguel Ángel Cuñado ENRESA, Emilio Vargas, 7 E-28043 Madrid, Spain

8.1

Introduction

Certain environmental factors such as temperature, pH, radiation, nutrient supply and, in particular, lack of water could severely limit the microbial activity in repositories. Most microorganisms require water activities (aw) above 0.9 to support active metabolism which corresponds to 60–80% soil saturation [1–3]. Thus, drying the environment by heating will probably eliminate the microbes initially present in the vicinity of the container. The concrete and bentonite barrier will also limit the survival and motion of microorganisms towards the bentonite due to the high pH and the very small pore size of the highly compacted bentonite blocks [3–5]. Even though microbial activity in the repository could be expected to be low, due to the above mentioned limiting factors, microbial activity inside the repository cannot be discarded altogether. Some microbial cells, especially those far away from the container, could resist restrictive conditions as dormant bacteria, and move closer to the container when the environmental conditions are less severe [2,6]. Bacteria naturally occurring in groundwater and rocks could also migrate into the repository through fractures in the concrete and bentonite barriers, thus increasing the risk of microbiologically influenced corrosion (MIC) of the metal canister. It is also important to point out that a significant MIC of the metallic container could occur indirectly from sulphides produced by sulphate reducing bacteria (SRB), at locations remote from the container surface, where microbial activity is not inhibited [7,8]. Several studies suggest however that this corrosion process would be severely limited by the slow transport rate of sulphide inside compacted bentonite [7,9]. In this respect, the results obtained in the Full-scale Engineered Barriers Experiment project (FEBEX) and described briefly in section 8.2, were significant [10]. The hydration of some bentonite blocks close to the rock, probably due to the penetration of water through cracks in the concrete/rock, produced significant corrosion damage in certain components. This corrosion damage was principally induced by microbial activity. Thus, even though microbial activity in the repository could be expected to be low, MIC cannot be discounted as a potential corrosion mode for the metallic container. 137

138

Sulphur-assisted corrosion in nuclear disposal systems

The presence of microorganisms on a metal surface or in its vicinity, often leads to highly localized changes in the concentration of the electrolyte constituents, pH and oxygen levels. These microorganisms and their metabolic activity can therefore influence the corrosion process by destroying the protective films of corrosionresistant alloys, stimulating localized forms of corrosion such as pitting, crevice formation or stress corrosion cracking (SCC) [11,12]. Patchy microbial deposits/ tubercles/biocorrosion products can also form a discontinuous biofilm that creates conditions for the development of differential aeration cells [13,14]. It is important to point out that microorganisms are able to drastically change the electrochemical conditions at the metal/solution interface by biofilm formation. These changes can range from induction or acceleration of corrosion, to corrosion inhibition [12,15,16]. The present study was designed to gain a better understanding of the influence of aerobic and anaerobic bacteria on the corrosion behaviour of three candidate container materials, by electrochemical test methods. Corrosion features were also evaluated using optical and scanning electron microscopy (SEM) techniques. Energy dispersive spectroscopy (EDS) analyses were carried out to chemically characterise the corrosion products and biodeposits generated during the tests. 8.2

Background: The FEBEX in situ test*

The FEBEX project, carried out in Grimsel (Switzerland), was initiated by ENRESA in 1995. It was designed to demonstrate the technical feasibility of manufacturing, handling and installing the engineered barriers of a deep geological disposal facility for high-level radioactive waste. The corrosion damage experienced by several components of the in-situ test was assessed. A heater, simulating a waste container was buried in a granitic formation surrounded by highly compacted bentonite blocks. The average water content and dry density values of the bentonite blocks were 14.4% and 1.69 g/cm3, respectively [17]. The FEBEX bentonite selected by ENRESA was extracted from a deposit in the Cabo de Gata region (Almería). Coupons of several candidate metals for manufacturing high-level waste (HLW) containers were introduced into these bentonite blocks, as well as sensors to monitor different physicochemical parameters during the test. The in-situ test began in July 1996 and, in June 2002, one of the heaters, a section of the liner, several corrosion coupons and four sensors were extracted. The heater was a carbon steel cylinder 4.5 m long, 100 mm thick and had welded lids. Corrosion coupons were made of carbon steel, stainless steel, titanium, copper and cupro-nickel alloys. The extensometer type sensors extracted had a 316L stainless steel sheath. All of these components were visually inspected, and analysed by optical and scanning electron microscopy including EDS and X-ray diffraction (XRD) analyses of corrosion products, to assess the degree of corrosion suffered. This work was complemented by the microbiological characterization of bentonite samples. Results obtained in the study indicated a slight general corrosion damage in the heater, guide tube and corrosion coupons. However, analysis of the sensors revealed significant pitting corrosion damage, advancing from the outer to the inner surface and occasionally perforating the tube wall (Fig. 8.1) as well as under tubercles or * The authors wish to acknowledge ANDRA´s for publishing the papers of the 2nd International Workshop Prediction of Long Term Corrosion Behaviour in Nuclear Waste Systems, Nice, September 2004, Eurocorr 2004.

Experimental investigation of the impact of microbial activity

139

8.1 Damage observed in one extensometer type sensor from the FEBEX in situ study. The extensometer outer tube was manufactured from AISI 316 stainless steel. (a) Photograph of a sensor inside a block of bentonite. Diffusion of corrosion products coming from the sensor can be seen in the bentonite surrounding it. (b) Detail of biodeposits/tubercles on outer tube. (c) Optical micrograph of the cross-section of outer tube. (d) Optical micrograph of the cross-section of outer tube showing transgranular SCC

biodeposits. Tube cracking was also detected, with cracks starting at the outer surface, sometimes from previous pitting damage. These cracks had a ramified morphology with transgranular progression, typical of SCC in stainless steel. Energy dispersive spectrometry analysis carried out on corrosion products and biodeposits appearing on two sensors, revealed the presence of sulphur in almost all of the corrosion products analysed in the form of pyrrhotite (Fe1−xS), as confirmed by XRD analysis. Sulphur compounds such as sulphates and sulphites, commonly detected in the presence of SRB, were not observed. Most of these compounds are amorphous and do not give rise to a diffraction signal. Samples of the bentonite housing the corrosion coupons and sensors were characterized microbiologically. The determination of the presence of aerobic and anaerobic bacteria was carried out by microbiological count. Assessment of SRB was carried out by the most probable number (MPN) method. No microbial groups were

140

Sulphur-assisted corrosion in nuclear disposal systems

detected in the bentonite surrounding the coupons, whereas 4.2 × 103 cfu/g aerobic bacteria and 3.8 × 102 cfu/g SRB were detected in the bentonite housing the sensors. The water content of the bentonite samples was 13.1% for the bentonite close to the coupons, and approximately 21.5% for the bentonite blocks housing the sensors. The water content of the bentonite was determined on sample arrival by heating the samples in an oven at 115ºC until constant weight was achieved. The sulphur-rich corrosion products, the presence of SRB in the bentonite covering the sensors, as well as the morphology of the damage, indicated a corrosion phenomenon induced by bacteria. This factor, together with the high humidity content of the bentonite blocks housing the sensors, was responsible for the significant corrosion damage observed in these components. This hydration was probably due to a flow of water through cracks in the concrete/rock and in the bentonite. In the case of the heater and corrosion coupons, the almost complete absence of humidity as a result of heating was responsible for the slight general corrosion observed. Microbiologically influenced corrosion usually acts as precursor to localized corrosion phenomena such as SCC and pitting. Concerning the SCC damage observed in certain components, the combined and synergistic interaction of mechanical stresses and bacterial corrosion, usually known as microbiologically assisted stress corrosion cracking, is not an uncommon phenomenon and it has been reported by several authors [12–15]. The presence of depassivating corrosion agents resulting from the metabolic activity of the SRB, such as S2− and HS−, can induce SCC. It has also been suggested that SRB can cause hydrogen embrittlement [18–22]. The role of bacteria in embrittlement of metals is not fully understood. The presence of H2S, which can be produced by SRB, is known to retard formation of molecular hydrogen on the metal surface, and to enhance adsorption of atomic hydrogen by the metal [12,20]. 8.3 8.3.1

Materials and methods Specimen preparation

The experiments were conducted with disc-shaped coupons made of S355 carbon steel (1.0570), 316L stainless steel (1.4404) and Cu–OF alloy (UNS C10200). Specimens were embedded in epoxy resin for electrical isolation and firm attachment. The metal specimens had a total exposed area of 1.8 cm2. To create working electrodes, an electrical contact was provided by a length of steel wire connected to the back of each specimen. Experimental coupons were abraded using 600-grit (approximately 14 μm according DIN ISO 6344) silicon carbide metallurgical paper, degreased in acetone, washed with sterilized distilled water and dried and stored in a desiccator. 8.3.2

Bacterial cultivation

Several SRB and sulphur oxidizing bacteria (SOB) species were chosen for the electrochemical tests. These types of bacteria have been detected in the Spanish reference bentonite known as FEBEX-bentonite [23]. In addition, these bacteria are known to promote biological corrosion mechanisms. Some of them are also able to withstand harsh repository conditions such as high pH due to the presence of concrete in the emplacement. The main properties of these two types of bacteria are given below. Bacterial strains were ordered from the German Collection of Microorganisms [Deutsche Sammlung von Mikroorganismen (DSM)]:

Experimental investigation of the impact of microbial activity

141

1. Sulphate reducing bacteria Strict anaerobic, SRB play a major role in metal corrosion by consuming hydrogen with the production of metal sulphides. Two strains of SRB were selected for laboratory tests: • Desulfovibrio desulfuricans DSM 4369. This bacterium has an optimum temperature for growth of 30ºC, and an optimum growing pH of 7.8. They can be considered as neutrophiles, with a pH range of 5.5 to 8.5 [1]. • Desulfotomaculum alkaliphilum DSM 12257. Optimum temperature for growth of 50ºC and optimum pH for microbial growth ranging between 8.7 and 9. This bacterium can tolerate relatively high pH values (alkalophiles). 2. Sulphur oxidizing bacteria These aerobic bacteria use reduced sulphur compounds such as elemental sulphur, thiosulphates, metal hydrogen sulphide or metal sulphides, and oxidize them to sulphate thus producing sulphuric acid and/or ferric ions. Amongst SOB, a slightly alkalophile strain, was selected. • Thiobacillus versutus DSM 582. This bacterium has an optimum temperature for microbial growth of 30ºC and optimum pH for microbial growth of 8.5. The culture media used for bacterial growth were formulated following the procedure indicated by the DSM supplier. Strain DSM 4369 was grown anaerobically in the Desulfovibrio Medium (medium 63, DSM Catalogue of strains). The pH was adjusted to 7.8 with NaOH (10 N). The medium was under a stream of oxygen-free N2 gas. Strain DSM 12257 was incubated anaerobically in Desulfotomaculum Alkaliphilum Medium (medium 866, DSM Catalogue of strains). This medium was prepared anaerobically under oxygen-free N2 gas. The pH of the autoclaved medium was kept between 8.7 and 9.0. SOB DSM 582 culture was grown aerobically on Thiobacillus Novellus Medium (medium 69, DSM Catalogue of strains). The pH was adjusted to 8.5 with NaOH (0.5 N). Growth of bacteria was monitored by periodic counts of bacteria by a MPN method. The tests were performed in nine dilution levels containing SRB or SOB media with five repetitions at each level. The detection limit for MPN was less than 1.6 cells/mL. 8.3.3

Medium and inoculation

Synthetic bentonite water with and without inoculated bacteria was used as the testing environment. This water simulates the water in equilibrium with the bentonite barrier following saturation of the latter, which, in turn, is the water that will reach the canister. It is saline water with a chloride concentration of 6500 mg/kg and a sulphate concentration of 1500 mg/kg. A certain volume of the cultured bacteria was used to inoculate the simulated water in test reactors. Inoculated water was maintained under aerobic/anaerobic conditions at the optimum temperature for growth of bacteria, for approximately 14 days. After this period, bacterial counts were carried out to check bacterial survival in bentonite water. The total number of cells at the beginning of the experiment ranged from 105 to 7 10 cells per mL of bentonite water. Bacteria cell densities were also estimated at the end of the experiment so as to ensure bacteria survival during the test (Table 8.1).

142

Sulphur-assisted corrosion in nuclear disposal systems

Table 8.1 Bacterial concentration determined by MPN in inoculated simulated water at the beginning and at the end of the corrosion tests Bacteria strain

T (ºC) pH

SRB DSM 4369

30

7.8

SRB DSM 12257 SOB DSM 582

50 30

8.7 8.5

Tested material S355 316L Cu–OF S355 S355

Total cells (cells/mL) Total cells (cells/mL) t=0 t = 1450 h 5.1×105 5.1×105 N.D. 9.7×107 2.2×106

8.8×108 5.7×108 N.D. 6.4×104 1.3×106

N.D.: Not determined. Detection limit for MPN: <1.6 cells/mL.

8.3.4

Electrochemical measurements

Electrochemical measurements were carried out in 1.5 L glass reactors with rubber stoppers containing ports for gas, electrodes and sensors. The corrosion cell contained the testing media, the working electrode (test specimen), the platinum counter electrode and a salt bridge connected to a saturated calomel reference electrode (SCE). Thermostatic baths were used to maintain the desired temperature. During the test, argon or air were continuously bubbled into the solution to create anaerobic (0.25–0.40 mg/L of O2) or aerobic solutions, respectively. Before the addition of bacteria, the entire test reactors including the solutions and electrodes were autoclaved at 121ºC for 30 min. An identical set of sterile corrosion cells were tested for comparison. Potentiodynamic polarization and linear polarization resistance (LPR) electrochemical techniques were used to obtain information on the corrosion rate, pitting susceptibility and passivity of the three studied alloys in sterile and inoculated simulated water. Potentiodynamic polarization tests were performed in test reactors after approximately 500 and 1450 h of exposure in the inoculated and sterile simulated water following specifications provided in standard ASTM G-59. A potentiostat was used to control the potential and to measure the current. The potential scan started at 200 mV below the corrosion potential (Ecorr) and was increased at a rate of 0.6 V/h. Once the current density reached 10 mA/cm2 (or 2 V), the direction of the potential scan was reversed at the same scan rate. Linear polarization resistance measurements were periodically carried out in test reactors. A potentiostat performed potential scans from 20 mV below Ecorr to 20 mV above Ecorr. ‘Instant’ general corrosion rates were calculated following specifications provided in ASTM G-102. This technique is very suitable for detection of changes in corrosion rates due to the presence of bacteria [18]. These electrochemical tests were conducted in sterile and inoculated media, under aerobic and anaerobic conditions at different temperatures and pH values corresponding to the optimum conditions for growth of bacteria (Table 8.1). 8.3.5

Post-exposure analysis

Post-exposure analyses are essential to validate the data obtained from the electrochemical tests. Tested specimens were examined by optical microscopy and SEM

Experimental investigation of the impact of microbial activity

143

to evaluate the morphology and extent of corrosion damage. Coupons were also examined for biofilm generation using SEM. In order to characterize chemically the corrosion products and/or deposits generated on tested specimens, EDS analyses were carried out using a microanalyser coupled to the SEM microscope. 8.4 8.4.1

Results S355 Carbon steel: SRB DSM 12257 and SRB DSM 4369

Comparisons were made between polarization curves recorded in sterile media with those obtained in the presence of bacteria. No sensibility to pitting corrosion was observed for carbon steel S355 under sterile or SRB inoculated conditions after the potentiodynamic tests had been carried out. Corrosion rates measured by the LPR technique were higher in SRB media compared with values obtained under sterile conditions. As can be seen in Fig. 8.3, corrosion rate values for the carbon steel in the presence of SRB 4369 strain were higher than those measured in SRB 12257 cultures. It can also be concluded that the corrosion rates on S355 coupons in water inoculated with SRB DSM 12257 reached a ‘steady state’. This could indicate exhaustion of the media or build-up of toxic end-products. This would explain the significant loss of bacteria population detected for this SRB strain at the end of the experiment (Table 8.1). In any case, biodeposits were present in all carbon steel coupons tested in the presence of this bacteria strain, indicating a significant bacterial activity during a portion of the experiment (Figs. 8.2, 8.4 and 8.5). Visual examination of tested specimens revealed certain differences in the corrosion morphology of samples exposed to SRB. Corrosion in the presence of DSM

8.2 Test reactor containing SRB DSM 12257 inoculated medium and carbon steel coupons after exposure for 360 h. Bacteria are active, as they are capable of reducing sulphate and producing black-coloured iron sulphide precipitates

144

Sulphur-assisted corrosion in nuclear disposal systems

8.3 Corrosion rates of carbon steel coupons in synthetic bentonite water inoculated with SRB or left sterile, measured by the LPR technique

8.4 Appearance of S355 carbon steel coupons after polarization scans performed after 1450 h exposure in SRB inoculated and sterile water

12257 bacteria was characterized by tubercle or biodeposit generation, with little general corrosion on the exposed surface, whereas a more general corrosion was observed in the presence of DSM 4369 bacteria. Metallographic studies of tested specimens showed a slight corrosion penetration under the biodeposits, thus increasing the risk factors for the initiation of localized corrosion. Scanning electron microscopy examination of the biodeposits revealed that they were mainly made of corrosion products, bacteria and their metabolites. Bacteria cells were sometimes difficult to discern because they were totally embedded or partially covered by corrosion precipitates. Energy dispersive analysis of these deposits indicated significant amounts of sulphur, not detected under sterile conditions. The development of non-continuous microbial deposits on the surface of carbon steel coupons could be directly correlated with localized corrosion areas under them. These deposits can create differential concentration cells that increase the risk of localized corrosion phenomena such as crevice or pitting corrosion. Once started, these corrosion phenomena can continue without the presence of bacteria.

Experimental investigation of the impact of microbial activity

145

8.5 (a) SEM micrograph showing biodeposit formed on S355 carbon steel surface tested in the presence of SRB. (b) Optical micrograph of S355 cross section showing corrosion beneath a biodeposit

8.4.2

S355 Carbon steel: SOB DSM 582

Anodic polarization curves obtained for carbon steel under sterile and SOB inoculated conditions show, in all cases, more negative Ecorr values in the sterile medium, with little risk of localized corrosion for both environments. The extent of the corrosion damage was significantly higher in samples tested in the aerated sterile media (Table 8.2 and Fig. 8.6). Further tests with SOB were performed with the addition of 0.017 M thiosulphate (Na2S2O3) to the test media. The corrosion rates of steel in the sterile and inoculated media increased with addition of S2O32−, but even with this compound in the solution, the corrosion rate was higher under sterile conditions. SEM examinations revealed a bacteria colonized surface in carbon steel specimens, but rather than increasing the corrosion damage, this biofilm appeared to have inhibitory properties (Fig. 8.6). This is not uncommon and has been reported previously by several authors [12,15,16,24–26]. Microorganisms can induce corrosion inhibition according to two general mechanisms or their combination: (i) neutralizing

146

Sulphur-assisted corrosion in nuclear disposal systems

Table 8.2 Instant corrosion rates for S355 carbon steel coupons in SOB inoculated and sterile simulated water Alloy Medium S355 Sterile simulated water (30ºC) SOB DSM 582 inoculated water Sterile + Na2S2O3 (0.017 M) SOB DSM 582 + Na2S2O3 (0.017 M)

Test duration Instant corrosion rates* (LPR) (h) (μm/year) 500 500 1450 1450

373 6.3 1047 798

*LPR corrosion rate measurements obtained at test conclusion.

8.6 (a) Appearance of S355 carbon steel coupons after polarization scans performed following 1450 h exposure in SOB inoculated and sterile water, without thiosulphate. (b) SEM micrograph showing biofilm on carbon steel surface tested in the presence of SOB

the effects of corrosive substances and (ii) forming or stabilizing protective films on a metal surface [15]. 8.4.3

316L Stainless steel: SRB DSM 4369

No significant differences were observed in the anodic polarization curves obtained for stainless steel coupons tested in sterile and inoculated conditions with SRB

Experimental investigation of the impact of microbial activity

147

8.7 Anodic polarization curves of stainless steel 316L tested under sterile and SRB inoculated media

4369 strain (Fig. 8.7). Similar pitting potential values were obtained under both conditions. Corrosion rates were less than 3 μm/year for all test conditions. In all cases, tested specimens showed pitting corrosion after potentiodynamic scans, both in sterile and bacterial environments. The number and morphology of these pits were similar in both media. They were mainly generated due to anodic polarization as a consequence of the high chloride content present in synthetic bentonite water. Figure 8.8 shows the typical morphology of chloride induced corrosion pits in 316L stainless steels, generated by anodic polarization [12,25,26].

8.8 SEM micrograph showing pitting in 316L stainless steel coupon after anodic polarization in SRB inoculated bentonite water

148

Sulphur-assisted corrosion in nuclear disposal systems

EDS analysis performed inside the pits indicated higher sulphur contents, not detected in pits developed under sterile conditions. Sulphur, probably in the form of sulphides due to the metabolic activity of SRB, appears to make little or no contribution to pit development. SEM examinations revealed no biofilm generation on the surface of 316L tested coupons, its absence being probably the main reason for the lack of bacterial corrosion damage in these specimens. 8.4.4

Cu–OF alloy: SRB DSM 4369

Results obtained in the electrochemical tests with copper alloy specimens showed a significant increase in the corrosion rate when tested under SRB 4369 inoculated conditions, with respect to tests performed under sterile conditions (Fig. 8.9). The anodic polarization curves obtained for the copper alloy in sterile and inoculated media indicate no susceptibility to pitting corrosion. Metallographic studies of these specimens show deeper uniform general corrosion for copper alloys tested in the presence of SRB. SEM examinations revealed a significant biofilm growth on the surface of copper alloys tested in the presence of

8.9 (a) Appearance of Cu–OF coupons after polarization scans performed following 1450 h exposure in SRB inoculated and sterile water. (b) SEM micrograph showing biofilm on Cu–OF surface tested in the presence of SRB

Experimental investigation of the impact of microbial activity

149

8.10 Optical micrographs of Cu–OF cross sections of specimens tested in (a) sterile and (b) SRB inoculated water

bacteria. This biofilm was quite uniform across the metal surface (Fig. 8.10). Analysis of the biofilm by EDS indicated a significant enrichment in sulphur, not detected in the EDS analyses performed on the exposed surface of specimens tested under parallel sterile conditions. 8.5

Discussion

The lack of impact of SRB inoculated media in 316L coupons is not well understood. SRB are known to be associated with stainless steel corrosion in numerous industrial systems. It is believed that a consortium of bacteria, such as that existing in the FEBEX in situ test, could render more realistic information. A common feature of biofilms associated with MIC produced under real conditions is the existence of a microbial community. It is the interactive growth activity of this consortium of bacteria that in many cases stimulates the corrosion process [27,28]. In this study, the experiments have been done using pure cultures of SOB or SRB, so this approach could have some limitations. A more realistic test condition would

150

Sulphur-assisted corrosion in nuclear disposal systems

require the evaluation of the effect of SRB and SOB bacteria in consortia with other bacteria. In certain experiments carried out in this study, a significant decrease in the bacterial population has been measured. It cannot be discounted that this behaviour could be due to the exhaustion of the media, especially for long-term experiments. This could be avoided by using flow-through systems, rather than batch experiments, which is also a more realistic condition. In any case, the methodology used provides a means of rapidly assessing the influence of certain bacterial species in the corrosion behaviour of carbon steels and Cu–OF alloys. Further work should include the above mentioned improvements over the initial approach. 8.6

Conclusions

The following conclusions can be drawn from the experimental results of this investigation: S355 carbon steel • Corrosion rates of carbon steel S355 increased with exposure to SRB inoculated media, with respect to corrosion rates measured in parallel sterile conditions. Bacteria increased not only the general corrosion rates, but also the risk for initiation of localized corrosion. The increase in corrosion rates, the generation of biodeposits and the enrichment of sulphur in the corrosion products were indicative of corrosion activity due to SRB. Consequently, S355 carbon steel was susceptible to experiencing MIC due to SRB under the experimental conditions used in this investigation. • Measured corrosion rates for carbon steel coupons were lower in media inoculated with SOB than in the sterile control. This corrosion inhibition is mainly attributed to the formation of a protective biofilm on the metal surface. 316L • No significant differences were observed in the corrosion resistance of 316L stainless steel in the sterile and SRB inoculated media. This was mainly attributed to the absence of a biofilm on the metal surface. Accordingly, 316L steel was relatively stable in the presence of SRB, at least under the test conditions used in these experiments. Cu–OF • Cu–OF alloy experienced sensitivity to MIC due to SRB activity. The development of a biofilm in the copper exposed surface was responsible for the important increase in the uniform corrosion rate, with respect to values observed in the sterile control media. • The damage resulting from the corrosion of Cu–OF by SRB was in the form of uniform corrosion. Initiation of localized corrosion damage such as pitting was not observed. References 1. K. Pedersen and F. Karlsson, SKB 95-10, 1995. 2. J. M. Horn and A. Meike, UCRL-ID-122256, Lawrence Livermore National Laboratory, 1995.

Experimental investigation of the impact of microbial activity

151

3. BSC (Bechtel SAIC Company), DOC.20041118.0005, 2004. 4. K. Pedersen, M. Motamedi, O. Karnland and T. Sandén, J. Appl. Microbiol., 89 (2000), 1038. 5. S. Fukunaga, T. Jintoku, Y. Iwate, M. Nakayama, T. Tsuji, N. Sakaya, K. Mogi and M. Ito, Geomicrobiol. J., 22 (2005), 361. 6. S. Stroes-Gascoyne and J. M. West, FEMS Microbiol. Rev., 20 (1997), 573. 7. D. W. Shoesmith, Corrosion, 62(8) (2006), 703. 8. S. Stroes-Gascoyne and F. King, ‘Microbially influenced corrosion issues in high level nuclear waste repositories’, in Proc. Topical Research Symp., CORROSION/2002, NACE, Houston, TX, 2002. 9. J. Smith, Z. Qin, F. King, L. Werme and D. W. Shoesmith, Corrosion, 63(2) (2006), 135. 10. ANDRA, Science and Technology Series, 2004, 61. 11. S. C. Dexter, D. J. Duquette, O. W. Siebert and J. A. Videla, Corrosion, 47(2) (1991), 308. 12. B. J. Little and J. S. Lee, Microbiologically Influenced Corrosion, Wiley, New York, 2007. 13. H. A. Videla, Int. Biodeterior. Biodegrad., 29 (1992), 195. 14. S. W. Borenstein, Mater. Perform., 27 (1991), 62. 15. H. A. Videla and L. K. Herrera, Int. Biodeterior. Biodegrad., 63 (2009), 896. 16. A. Pedersen and M. Hermansson, Biofouling, 3 (1991), 1. 17. ENRESA, Technical Report 05-0/2006, 2006. 18. M. Walch and R. Mitchell, ‘Biologically induced corrosion’, in Proc. International Conference on Biologically Induced Corrosion, ed. S. C. Dexter. NACE, Houston, TX, 1986. 19. R. K. Javaherdashti, C. Singh Raman, C. Panter and E. V. Pereloma, Int. Biodeterior. Biodegrad., 58 (2006), 27. 20. M. V. Biezma, Int. J. Hydrogen Energy, 26 (2001), 515. 21. H. A. Videla, ‘Electrochemical aspects of biocorrosion’, in Bioextraction and Biodeterioration of Metals, 85. ed. C. Gaylarde and H. A. Videla. Cambridge University Press, Cambridge, UK, 1995. 22. C. J. Thomas, R. G. J. Edyvean and R. Brook, Biofouling, 1 (1988), 65. 23. INASMET 36.0056.0 – Final Report: Corrosión de materiales para cápsulas de RAA en formaciones arcillosas. Métodos y viabilidad de fabricación de cápsulas, 2009. 24. A. M. El-Shamy, T. Y. Soror, H. A. El-Dahan, E. A. Ghazy and A. F. Eweas, Mater. Chem. Phys., 114 (2009), 156. 25. A. Jayaraman, J. C. Earthman and T. K. Wood, Appl. Microbiol. Biotechnol., 47 (1997), 62. 26. Kh. M. Ismail, A. Jayaraman, T. K. Wood and J. C. Earthman, Electrochim. Acta, 1999, 4685. 27. S. E. Werner, C. A. Johnson, N. J. Laycock, P. T. Wilson and B. J. Webster, Corros. Sci., 49 (1998), 465. 28. R. T. Huang, B. L. McFarland and R. Z. Hodgman, ‘Microbial influenced corrosion in cargo oil tanks of crude oil tankers’, in CORROSION/97, paper no. 535, NACE, Houston, TX, 1997.

9 Reactive-transport modelling of the sulphide-assisted corrosion of copper nuclear waste canisters Fraser King Integrity Corrosion Consulting Ltd, Nanaimo, BC, Canada V9T 1K2

Miroslav Kolar LS Computing, Nanaimo, BC, Canada V9T 2N6

Marjut Vähänen Posiva Oy, Olkiluoto, FI-27160 Eurajoki, Finland

9.1

Introduction

One of the significant advantages of the use of copper as a canister material for the disposal of spent fuel is that it is thermodynamically stable in water in the absence of oxygen. As a consequence, once all of the initially trapped atmospheric O2 (and the Cu(II) produced by the oxidation of Cu(I) by O2) has been consumed and in the absence of oxidising radiolysis products, corrosion will theoretically cease. However, the anodic dissolution of Cu in the presence of sulphide occurs at a potential more negative than that of the H2/H2O equilibrium line. Therefore, corrosion of copper canisters will continue during the long-term anaerobic phase if sulphide ions are present at the canister surface. There are a number of sources of sulphide in potential deep geological repositories in Finland and Sweden. First, Finnish and Swedish groundwaters contain dissolved sulphide ions at concentrations ranging from <1 mg/L at repository depth at the Laxemar and Forsmark sites in Sweden [1] to 12 mg/L at Olkiluoto in Finland [2]. Second, the MX-80 and Deponit CA-N bentonite proposed for use in the repository both contain pyrite impurities [2]. Third, sulphate present in the groundwater and in the bentonite porewater (as a result of the dissolution of gypsum or anhydrite mineral impurities) can be reduced to sulphide by sulphate-reducing bacteria (SRB) [3]. Given the possible sources of sulphide in the repository, lifetime predictions invariably account for corrosion in the presence of sulphide during the long-term anaerobic phase [4]. Mass-transport arguments are used to predict the rate of corrosion on the assumption, subsequently confirmed experimentally [5], that the rate of corrosion is controlled by the rate of supply of HS− to the canister surface. Under these conditions, the various sources of sulphide listed above result in <4 mm of wall loss in a period of 100 000 years [1,4]. This relatively small amount of corrosion (compared with the proposed canister wall thickness of 50 mm) is a consequence of the low solubility of HS− in the repository environment and the slow rate of mass transport through the compacted bentonite buffer material. 152

Reactive-transport modelling of the sulphide-assisted corrosion

153

Apart from the effect on the corrosion of the canister, sulphide species also play a role in determining the evolution of the redox conditions in the repository. Some of the initially trapped O2 will be consumed by the oxidation of pyrite. Once the repository has become anoxic, the redox potential (EH) is likely to be controlled by a redox process involving dissolved HS− ions. This evolution of EH will be reflected in the evolution of the corrosion potential (ECORR). Figure 9.1 shows the evolution of ECORR determined experimentally using both bare and clay-covered Cu electrodes immersed in 1 mol/L NaCl solution at 25°C [6]. Initially, the solution was aerated and the ECORR allowed to reach a steady-state value. At points A, B, and C, the purge gas was sequentially changed to 2 vol.% O2/N2, 0.2% O2/N2, and pure N2, respectively. With each decrease in O2 partial pressure, the value of ECORR shifted to more negative values. At point D, Na2S was added to the solution to produce a dissolved HS− concentration of 10 mg/L. The ECORR of the bare Cu electrode (curve (a) in Fig. 9.1) decreased immediately by ~500 mV to a potential close to the H2O/H2 equilibrium value at the pH of the amended solution (~pH 10.5). However, after an initial decrease in ECORR of the clay-covered electrode (curve (c), Fig. 9.1), thought to be due to the consumption of residual O2 in the clay layer, the potential apparently stabilised until a second addition of Na2S at point E to produce a dissolved HS− concentration of 100 mg/L. After a further decrease in potential and levelling off, the value of ECORR of the clay-covered electrode also dropped precipitously to a value similar to that of the bare Cu electrode when also exposed to a solution of 100 mg/L HS−.

9.1 Time dependence of the corrosion potential (ECORR) of copper electrodes in 1 mol/L NaCl solution at 25°C in the presence of dissolved oxygen or sulphide [6]. Curve (a) is for a bare copper electrode and curves (b) and (c) are for a copper electrode covered by a 1-mm-thick layer of compacted Na-bentonite. At points A, B, and C, the original air purge was changed for gas mixtures of 2 vol.% O2/N2, 0.2% O2/N2, and pure N2, respectively. At points D and E, Na2S was added to the solution to produce a dissolved HS− concentration of 10 mg/L and 100 mg/L, respectively

154

Sulphur-assisted corrosion in nuclear disposal systems

On a much contracted timescale, the evolution of ECORR exhibited in Fig. 9.1 is thought to reflect that expected for the corrosion potential of a copper canister in the repository. Initially, both ECORR and EH are determined by the presence of O2. As that O2 is consumed by corrosion of the copper canister (electrode), aerobic respiration, and/or the oxidation of pyrite, however, both ECORR and EH shift to more negative values. This paper describes the development of, and preliminary simulations using, a reactive-transport model for the corrosion of copper in sulphide-containing chloride environments. The model is an extension of previous models developed to describe the corrosion behaviour of copper canisters in a deep geological repository sealed with bentonite-based materials and saturated with saline groundwater [7]. Various aspects of the development of the model are described, including: the reaction scheme, details of the reactions involving sulphide species, a brief description of other processes considered, and the nature of the mass-balance equations, spatial grid, and initial and boundary conditions. Finally, the results of preliminary simulations of the evolution of the corrosion behaviour of a copper canister in a KBS-3V type repository at the Olkiluoto site are presented and discussed. 9.2

Model development

Reactive-transport modelling of the corrosion of spent fuel canisters involves the coupling of interfacial electrochemical reactions to various mass-transport, redox, precipitation, and sorption processes occurring in the near- and far-field environments [7]. Such models can be used to predict the evolution of the repository environment and the consequent effects on the corrosion behaviour of the canister. The model described here has been termed the Copper Sulphide Corrosion Model Version 1.0 (CSM V1.0). 9.2.1

Reaction scheme

Figure 9.2 shows the reaction scheme on which the model is based. The CSM builds on previous reactive-transport models developed to predict the long-term corrosion behaviour of copper canisters [7–9]. The basic reaction scheme described by King et al. [7] has been extended to include appropriate interfacial and homogeneous reactions involving dissolved sulphide ions and various sulphur-containing species. 9.2.2

Reactions involving sulphur species

Five S-containing species are explicitly included in the model, including pyrite (FeS2), dissolved sulphide (HS−), a precipitated iron sulphide phase (FeS), precipitated cuprous sulphide (Cu2S), and dissolved sulphate ion (SO42−). In addition, an adsorbed sulphide intermediate species (Cu(HS)ADS) is implicitly included in the interfacial sulphidation reaction. These species are involved in the following processes: Pyrite oxidation In the presence of O2, pyrite is oxidised (ultimately) to ferric species and sulphate ions

Reactive-transport modelling of the sulphide-assisted corrosion

155

9.2 Reaction scheme for the Copper Sulphide Corrosion Model Version 1.0 (CSM V1.0)

FeS2 + 3.75O2 + 3.5H2O → Fe(OH)3 + 2SO42− + 4H+

[9.1]

The rate of pyrite oxidation is treated as ½-order with respect to [O2] and proportional to the surface area of pyrite [10]. Pyrite dissolution Pyrite dissolves accompanied by the disproportionation of the polysulphide ion to produce sulphide and thiosulphate (S2O32−) FeS2 + 0.75H2O → Fe(II) + 1.5HS− + 0.25S2O32−

[9.2]

Thiosulphate ions do not impact the corrosion behaviour of copper [4] and are not tracked further in the model. Microbial reduction of sulphate to sulphide Sulphate-reducing bacteria (SRB) will reduce SO42− to HS− using either organic carbon or H2 as an electron donor. The reduction reaction can be written as SO42− + 5H2O + 8e− → HS− + 9OH−

[9.3]

Masurat et al. [3] have demonstrated recently that the microbially-mediated reduction of SO42− can occur in highly compacted bentonite under conditions which previously were thought to preclude microbial activity. The rate of sulphide production is observed to decrease with increasing bentonite density, with the rate at a

156

Sulphur-assisted corrosion in nuclear disposal systems

bentonite density of 2.0 Mg m−3 insufficient to lead to penetration of the canister wall within a period of 100 000 years [3]. It is assumed in the CSM V1.0 that the rate of sulphate reduction is not limited by the availability of electron donors, but may be limited by the availability of water (and, of course, the availability of SO42−). The possible limitation due to water availability (activity) is implemented by defining a threshold degree of saturation below which microbial activity is not possible [8]. Precipitation of amorphous iron sulphide Sulphide ions may precipitate with Fe(II) to form an amorphous iron sulphide (‘FeS’). The stoichiometry for this reaction is given by Fe(II) + HS− → FeS + H+

[9.4] −

The rate of precipitation is assumed to be proportional to [HS ] and the degree of supersaturation of Fe(II). Anodic dissolution of copper as Cu2S Based on the results of electrochemical studies of the anodic dissolution of copper in sulphide solutions, the overall interfacial reaction can be written as [5,11] Cu + HS− = Cu(HS)ADS + e−

fast

[9.5]

slow

[9.6]

followed by a slow second one-electron process Cu + Cu(HS)ADS + HS− → Cu2S + H2S + e− Cathodic reduction of HS− The nature of the cathodic reaction accompanying the sulphidation process (reactions 9.5 and 9.6) is currently uncertain. In the CSM V1.0, it is assumed that the reaction involves the reduction of HS− 2HS− + 2e− → H2 + 2S2−

[9.7]

Diffusive mass transport of HS− ions In the model, sulphide ions are assumed to be free to diffuse through the bentonite and rock layers towards and away from the canister surface. 9.2.3

Other processes included in the model

Apart from the various reactions involving S species described above, the reaction scheme on which the model is based includes a number of other processes, including: • • • •

the interfacial dissolution of copper as CuCl2− ions the precipitation and dissolution of Cu2O the homogeneous oxidation of CuCl2− to Cu2+ by O2 the precipitation and dissolution of Cu(II) as CuCl2⋅3Cu(OH)2

Reactive-transport modelling of the sulphide-assisted corrosion • • • • • • • •

157

the adsorption and desorption of Cu(II) on mineral surfaces, particularly bentonite the homogeneous reduction of Cu(II) to CuCl2− by Fe(II) the diffusive mass transport of CuCl2− and Cu(II) the interfacial reduction of Cu2+ to Cu(I) the interfacial reduction of dissolved O2 on the canister surface the partitioning of O2 between aqueous and gaseous phases the diffusive mass transport of O2 the consumption of O2 by aerobic bacteria.

Details of the treatment of each of these processes are given elsewhere [7–9]. 9.2.4

Structure of the model

Mass-balance equations Mathematically, the reaction scheme in Fig. 9.2 is described by a series of onedimensional reaction-diffusion (mass-balance) equations, of the general form ea

n ∂(Sci ) ∂ Ê ∂c ˆ = tf ee SDi i ˜ + ea S Â R j Á ∂t ∂x Ë ∂x ¯ j =0

[9.8]

where ea and ee are the accessible and effective porosity for mass transport of each of the various layers in the model (see next sub-section), respectively, tf is the tortuosity factor, S the degree of saturation, ci and Di are the concentration and bulk-solution diffusion coefficient of species i, and Rj is the rate of the jth reaction in the bentonite, backfill, and rock. A mass-balance equation is written for each of the 11 species considered in the model, namely: gaseous O2, dissolved O2, dissolved CuCl2−, precipitated Cu2O, dissolved Cu2+, precipitated CuCl2⋅3Cu(OH)2, adsorbed Cu(II), dissolved Cl−, dissolved Fe(II), precipitated FeS, dissolved HS−, pyrite FeS2, and dissolved SO42−. The amount of precipitated Cu2S corrosion product is also tracked. In addition, a heat-transport equation is used to predict the spatial and temporal variation of the temperature in the repository. Spatial grid A one-dimensional spatial grid is used to represent the various mass-transport barriers in the repository. It has proven more flexible to use a linear 1-D geometry, as opposed, for example, to a radial geometry, to represent the complex layout of the various buffer and backfill materials in different repository designs (e.g. in-room and borehole disposal), despite the limitations inherent in a 1-D formulation. These barriers can include: a precipitated porous layer of Cu2S, highly compacted bentonite, backfill material, excavation disturbed and damaged zones, room and tunnel seals, host rock. These layers can be further sub-divided, for instance to represent different time-dependent saturation behaviour of buffer or backfill materials or different degrees of fracturing of the host rock. Each layer is defined by the following properties: •

thickness (in the case of the precipitated Cu2S layer, the thickness is timedependent)

158 • • • • • • • • • •

Sulphur-assisted corrosion in nuclear disposal systems

porosity (total, accessible, and effective porosity for mass transport) tortuosity factor density initial concentrations of gaseous O2, dissolved O2, Cl−, Fe(II), HS−, pyrite, and SO42− maximum adsorption capacity for Cu(II) exposed surface area of pyrite threshold degree of saturation for microbial activity time-dependent saturation saturation-dependent thermal conductivity heat capacity.

A total of six layers were used for the simulations described here, including (from the canister surface outwards): 1. a porous layer of precipitated Cu2S with variable (increasing) thickness 2. highly compacted bentonite (MX-80 bentonite, dry density 1.65 Mg m−3) 3. tunnel backfill (70% crushed rock/30% CA-N Deponit bentonite, dry density 2.1 Mg m−3) 4. excavation damaged zone EDZ (thickness 0.5 m) [2] 5. excavation disturbed zone EdZ (thickness 2.0 m) [2] 6. fractured host rock (thickness 50 m). The thicknesses of the layers representing the buffer and backfill materials were calculated to conserve the same buffer/backfill volume/canister surface area ratio as the KBS-3V design [2]. For the current simulation, all layers were assumed to be saturated at all times. Boundary and initial conditions Mathematical boundary and initial conditions are required to solve the series of 12 mass-balance equations (one for each of the 11 chemical species and the heatconduction equation). Boundary conditions are only required for species that diffuse (including temperature). By convention, the left-hand boundary is defined as the canister surface and the right-hand boundary by a fracture in the far-field rock. For those species participating in either anodic or cathodic corrosion processes, the corresponding Butler–Volmer (B–V) expressions serve as one set of boundary conditions. For example, for the interfacial dissolution of Cu to form Cu2S [5,11] Cu + HS− = Cu(HS)ADS + e−

[9.9a]

Cu + Cu(HS)ADS + HS− → Cu2S + H2S + e−

[9.9b]

the B–V expression is given by Ï (1 + aS )F ¸ Ï F ¸ 0 + aS3 ES30 ) ˝ ( ES12 iS (t ) = nS efilm FkS [HS - ]02 exp Ì E ˝ exp ÌÓ RT ˛ ˛ Ó RT

[9.10]

where iS(t) is the time-dependent current density, E is the potential, E0S12 and E0S3 and aS and aS3 are the standard potentials and transfer coefficients for Reactions 9.9a and 9.9b, respectively, nS is the number of electrons transferred in the rate-determining

Reactive-transport modelling of the sulphide-assisted corrosion

159

step (nS = 1), efilm is the Cu2S film porosity, kS is the electrochemical rate constant for Reaction 9.9b, [HS−]20 is the interfacial sulphide ion concentration, and F, R, and T are the Faraday constant, gas constant, and absolute temperature, respectively. Similar left-hand boundary conditions are used for other species involved in interfacial electrochemical reactions, such as CuCl2−, Cu2+, and O2 (Fig. 9.2). The rates of the interfacial reactions are linked to the fluxes of reactants and products to and from the canister surface. Thus, the anodic and cathodic current densities for the formation of Cu2S (iS) and reduction of HS− (iE) are related to the interfacial flux of HS− by -

iS iE ∂[HS - ]0 + = - FS tf ee DHSnS ne ∂x

[9.11]

In addition to the individual B–V expressions (e.g. equation 9.10) and interfacial fluxes (e.g. equation 9.11), the other important left-hand boundary condition is that the time-dependent current densities of the various electrochemical reactions should sum to zero iA(t) + iC(t) + iD(t) + iE(t) + iS(t) = 0

[9.12]

where iA, iC, and iD are the current densities for the dissolution of copper as CuCl2−, the reduction of O2, and the reduction of Cu2+, respectively, and iE and iS have been defined above. Other mobile species that are not involved in interfacial electrochemical reactions, such as gaseous O2, dissolved Fe(II), and SO42− are assigned a zero-flux left-hand boundary condition. The left-hand condition for the temperature is defined by the time-dependent canister temperature. The right-hand boundary is defined as a groundwater-bearing fracture. Species that are not naturally present in the groundwater, such as dissolved O2, CuCl2−, and Cu2+, are assigned a zero-concentration boundary condition. Species that are naturally present in groundwater, such as Cl−, dissolved Fe(II), HS−, and SO42−, are assigned a constant concentration right-hand boundary condition determined by the groundwater composition. The initial conditions are determined by the properties of the sealing materials and the composition of groundwater in the EDZ, EdZ, and fractured rock layers. 9.2.5

Computational aspects

The series of 12 mass-balance equations are solved using a finite-difference technique using the TRANSIENT [12] sub-routine. The spatial and temporal concentrations and temperature are predicted at each of the 1895 grid points representing the various layers in the model and for each time step up to a maximum period of 106 years. A geometrical progression is used to determine the grid point spacing, with closer spacing close to the canister surface where concentration gradients are typically the steepest and extending to larger spacings further from the canister. An adaptive time-step algorithm is used to determine the time steps, with the period between calculations increasing unless rapid changes in concentration occur (e.g. at the time at which all of the initially trapped O2 in the repository is consumed). Further details of the discretisation procedure and numerical solution are provided elsewhere [8,9,12].

160 9.3

Sulphur-assisted corrosion in nuclear disposal systems Results of preliminary simulations

A major characteristic of the expected evolution of the corrosion behaviour of a copper canister in a Finnish or Swedish repository is the expected shift to a sulphidedominated system and the accompanying decrease in ECORR. In this respect, the timedependence of ECORR shown in Fig. 9.1 is expected to occur for canisters in the repository, but over longer timescales. A fundamental requirement of the CSM, therefore, is to be able to predict the evolution of ECORR. Figure 9.3 shows the predicted evolution of ECORR of a copper canister in an initially saturated KBS-3V style repository with a saline groundwater containing 10−4 mol/L (3.3 mg/L) HS−. The predicted change in ECORR is similar to that observed using the clay-covered copper electrode, although the drop in ECORR occurs after 90 years instead of a matter of a couple of hundred hours. Because of the slow execution of the code, the simulation was stopped after a period of 3000 years. Initially, corrosion of the canister results from the dissolution as CuCl2− species supported by the reduction of O2 and, more importantly, Cu2+ (Fig. 9.4a). The direct interfacial reduction of O2 occurs for only seconds at which time it is replaced by the reduction of Cu2+ (formed by the homogeneous oxidation of CuCl2− by O2, rate constant k1 in Fig. 9.2). The reduction of cupric ions is the major cathodic reaction for the initial 10–20 years (Fig. 9.4). However, after only 1–2 years, the canister begins to corrode with the formation of Cu2S, with the sulphidation reaction becoming the major cause of corrosion after ~20 years (Fig. 9.4b). The only sources of HS− in the current simulation are the groundwater, the action of SRB in the backfill (it is assumed here that microbial activity is not possible in the highly compacted bentonite), and the anaerobic dissolution of pyrite in the buffer, backfill, and rock.

9.3 Predicted time dependence of the corrosion potential of a copper canister in a KBS-3V style repository in saline groundwater containing 10−4 mol/L sulphide

Reactive-transport modelling of the sulphide-assisted corrosion

161

9.4 Predicted time dependence of the partial anodic and cathodic current densities for (a) the entire simulation period of 3000 years and (b) the period between 1 and 3000 years

An interesting feature of the experimental data (Fig. 9.1), and one that appears also in the simulated data, is the slight decrease in ECORR followed by an arrest immediately before the rapid decrease in potential. In the description of the experimental data above, this intermediate decrease in ECORR was attributed to the consumption of

162

Sulphur-assisted corrosion in nuclear disposal systems

residual O2 in the clay. However, an examination of the predicted time dependence of the partial current densities for the various anodic and cathodic processes suggests that it is related to changes in the predominant anodic and cathodic reactions. Figure 9.4b shows the time dependence of iA, iD, iE, and iS between 1 year and 3000 years in more detail, the magnitude of iC being insignificant during this time period. As noted above, the sulphidation of the copper canister to form Cu2S (iS) begins after 1–2 years, at which time it is supported by the cathodic reduction of Cu2+ (iD). The value of ECORR starts to decrease at this time (see the inset to Fig. 9.3) as a result of the decrease in [Cu2+] and the more negative equilibrium potential for reaction 9.9 compared with that for dissolution as CuCl2−. The corrosion potential reaches a temporary minimum of −0.208 VSCE after a period of 20–30 years, which corresponds to the time at which the current density for the dissolution of copper as CuCl2− (iA) switches from positive (anodic) to negative (cathodic) (Fig. 9.4b). For the period from 27 years to 90 years, the formation of Cu2S is supported by the reduction of both Cu(II) and Cu(I), with CuCl2− being the only oxidant for the period 60–90 years. Figure 9.4b also illustrates why ECORR shifts to more negative potentials after 90 years (indicated by the vertical dotted line in the figure). The rapid decrease in ECORR is, specifically, an indication of the onset of H2 evolution due to the cathodic reduction of HS− (iE) and is not, in itself, an indication of the anodic reaction between Cu and HS−. Since the CSM is based on the assumption that the potential is controlled by the relative rates of different anodic and cathodic processes, the ability of the model to replicate the experimental observations in Fig. 9.1 is evidence that the post-transition potential is a true corrosion potential (determined by the relative rates of sulphidation and HS− reduction) as assumed in the model, rather than a redox process involving the Cu2S/HS− couple. Figure 9.5 shows the predicted time dependence of the Cu2S film thickness. Film growth is predicted to start after 26 years, i.e. long before the transition in ECORR to negative potentials. Film growth appears to attain a steady-state rate of ~1.5 nm/year, equivalent to a corrosion rate of 1.3 nm/year for the assumed film porosity of 0.1. This rate compares with rates predicted based on the supply of HS− to the canister surface of <10 nm/year due to HS− in the groundwater in the recent SR-Can assessment [1] and 20 nm/year in a Nagra-design repository [13]. 9.4

Conclusions

A one-dimensional reactive-transport model has been developed to predict the effect of sulphide on the long-term corrosion behaviour of copper nuclear waste canisters. Various sources of sulphide have been considered, including the dissolution of pyrite, the reduction of sulphate by microbial activity in the buffer and backfill materials surrounding the canister, and from the groundwater itself. In addition to participating in corrosion reactions, pyrite will also consume some of the initially trapped O2 in the repository. The model is based on a series of one-dimensional reaction-diffusion equations which describe the various redox precipitation/dissolution, adsorption/desorption, and mass transport processes in the near- and far-fields. These reactions are coupled to interfacial electrochemical reactions on the canister surface which actually determine the corrosion rate. The repository is described by a series of six porous mass-transport barriers representing (from the canister surface outwards): a growing

Reactive-transport modelling of the sulphide-assisted corrosion

163

9.5 Predicted time dependence of the copper sulphide film thickness. The steps in the curve are artifacts due to the use of finite grid spacings for the film thickness and increasing time steps

corrosion product layer of Cu2S, highly compacted bentonite buffer material, crushed rock/bentonite backfill, an excavation damaged zone, an excavation disturbed zone, and the fractured host rock itself. Solution of the reaction-diffusion equations using electrochemical boundary conditions for the canister surface permits the prediction of the corrosion rate and corrosion potential, in addition to the spatial and temporal distribution of the concentrations of each of the 11 chemical species considered in the model. The model has been used to perform preliminary simulations of the evolution of the corrosion behaviour of a copper canister in an KBS-3V style repository at Olkiluoto, Finland. The corrosion potential is predicted to decrease to potentials at which H2 evolution occurs after a period of ~90 years (for the assumptions used for the simulation). The predicted rapid drop in ECORR is similar to that observed, over shorter timescales, in laboratory experiments. Following a period of aerobic corrosion, the long-term rate of anaerobic corrosion due to the presence of sulphide is predicted to be of the order of 1 nm/year. References 1. SKB, Long-term Safety for KBS-3 Repositories at Forsmark and Laxemar – A First Evaluation. Main Report of the SR-Can project. SKB TR 06-09. Swedish Nuclear Fuel and Waste Management Company, 2006. 2. Posiva, Expected Evolution of a Spent Nuclear Fuel Repository at Olkiluoto. Posiva Report 2006-05, revised October 2007. 3. P. Masurat, S. Eriksson and K. Pedersen, Appl. Clay Sci., 47 (2010), 58–64.

164

Sulphur-assisted corrosion in nuclear disposal systems

4. F. King, L. Ahonen, C. Taxén, U. Vuorinen and L. Werme, Copper Corrosion under Expected Conditions in a Deep Geologic Repository. Posiva Oy Report POSIVA 2001-01, 2002. Also published as Swedish Nuclear Fuel and Waste Management Company Report, SKB TR 01-23, 2001. 5. J. Smith, Z. Qin, F. King, L. Werme and D. W. Shoesmith, Corrosion, 63 (2007), 135– 144. 6. F. King and S. Stroes-Gascoyne, ‘Predicting the effects of microbial activity on the corrosion of copper nuclear waste disposal containers’, in Microbial Degradation Processes in Radioactive Waste Repository and in Nuclear Fuel Storage Areas, 149–162, ed. J. H. Wolfram. Kluwer Press, Dordrecht, 1997. 7. F. King, M. Kolar and P. Maak, J. Nucl. Mater., 379 (2008), 133–141. 8. F. King, M. Kolar and S. Stroes-Gascoyne, Theory Manual for the Microbiological Copper Corrosion Model CCM-MIC.0. Ontario Power Generation Nuclear Waste Management Division Report No: 06819-REP-01200-10091, 2002. 9. F. King, Mixed-Potential Modelling of the Corrosion of Copper in the Presence of Sulphide. Posiva Working Report 2007-63, 2007, available from http://www.posiva.fi/en/databank/ working_reports. 10. S. U. Salmon and M. E. Malmström, Appl. Geochem., 19 (2004), 1–17. 11. J. M. Smith, ‘The corrosion and electrochemistry of copper in aqueous, anoxic sulphide solutions’, PhD Thesis, The University of Western Ontario, London, Canada, 2007. 12. TRANSIENT, http://transient.mkolar.org/. 13. L. H. Johnson and F. King, J. Nucl. Mater., 379 (2008), 9–15.

10 Sulphur-related issues in deep underground nuclear waste disposal systems P. De Cannière, B. Kursten, F. Druyts and H. Moors SCK•CEN, The Belgian Nuclear Research Centre (SCK•CEN), R&D Waste Packages Unit, Boeretang 200, B-2400 Mol, Belgium

R. Gens ONDRAF/NIRAS, The Belgian Agency for Radioactive Waste and Enriched Fissile Materials (ONDRAF/NIRAS), Avenue des Arts 14, B-1210 Brussels, Belgium

Panel Discussion The long-term safety of nuclear waste disposal relies on three main cornerstones: (i) the isolation of the waste from man and biosphere to prevent direct exposure to radiation, (ii) the long-term containment of radionuclides to delay and attenuate their release in groundwater, and (iii) the durable preservation of the performance of the disposal system. For high-level nuclear waste and spent fuel, a total engineered containment of the radioactivity is required during centuries or millennia, certainly during the first containment phase when the radiotoxicity present in the inventory has not yet significantly decayed or when the waste still emits non-negligible quantities of heat. The metallic components of the engineered barrier system (EBS) contribute to these three safety functions and are the ultimate barrier in case of human intrusion. As a consequence, they must be robust and durable, and are essential to preserve system integrity and safety. In the different disposal concepts for high-level nuclear waste and spent fuel, corrosion of the metallic barriers, and in particular the overpack/container, is thus a major issue. It is imperative for performance assessment to predict the lifetime of the containers and overpacks. One particular issue hampering the lifetime prediction is the presence of sulphur species that may cause localized corrosion. Therefore, an international workshop on Sulphur-Assisted Corrosion in Nuclear Waste Disposal Systems (SACNUC) was organized to provide an exchange of information on the influence of sulphur species on the corrosion of metallic barriers. During the 3-day workshop, different aspects of sulphur-assisted corrosion (microbial processes, modelling, effect of sulphur species on the corrosion of carbon steel and copper, etc.) were addressed. At the end of the workshop, a panel discussion took place to identify open issues in the investigation of sulphur-assisted corrosion phenomena and how to incorporate these in reliable lifetime prediction of metallic barriers. This paper gives a summary of the main outcomes of these discussions. 165

166

Sulphur-assisted corrosion in nuclear disposal systems

1 Source term: Is it possible to predict the sulphide concentration reliably at the waste overpack surface? Is it correct to assume that sulphur transport is diffusion-controlled? In the absence of buffer materials, the metallic overpack is directly exposed to air during the operational phase and to water in the long term, there is no diffusion barrier and the sulphide concentration expected in the water could be very high because of the possible development of sulphate reducing bacteria (SRB). The role of compact materials (bentonite or cementitious buffer materials) installed around containers is to limit the different processes favourable to corrosion: • •

•

Transport of aggressive sulphur species (HS−, S2O32−) is controlled by diffusion. Microbial activity is limited by several factors, such as: low water activity, space restriction and diffusion-limited transport of nutrients/toxins. At dry densities above 2000 kg/m3, no significant bacterial activity can be observed [1]. Also, at pH higher than 12.5, no microbial activity is expected. In the absence of active microbes, there is no longer production of free sulphides close to the overpack surface. In the presence of Fe2+ in the compact buffer, no high concentrations of HS− are expected in water because of the precipitation of poorly soluble iron sulphides such as FeS or FeS2.

To predict the transport and reactivity of sulphur species at the surface of canisters, it is also necessary to improve our understanding of the water chemistry of reduced sulphur species. A lot is known but not everything, for example, among other things, the nature of adsorbed surface species. 2 How can short-term experimental results be extrapolated to the long time frame of geological disposal of nuclear waste? Sufficient knowledge of elementary mechanisms is necessary to adequately describe the key processes needed to develop models for long-term assessment and to support semi-empirical models mostly used in canister lifetime prediction. Mechanistic understanding for Cu is much more advanced than for carbon steel. Sensitivity analysis with different concentrations of sulphides is needed as for chlorides, i.e. do threshold values exist below which no effects are observed? Then, extrapolation to real conditions should be feasible. 3 For the Belgian concept, which foresees a pH of at least 12.5 for an extended time, can bacterial activity be excluded? Which bacteria could be active at this pH? Under what conditions does microbial activity not play a role? How can we prove this experimentally? No significant microbial activity is expected at high pH (>12.5). It is much more difficult for bacteria to cope with high pH than with low pH. To survive, alkaliphiles must maintain a relatively low alkaline level around pH 8 inside their cells by constantly pumping hydrogen ions (H+) across their cell membranes into their cytoplasm [2]. Energetically and from the viewpoint of mass transfer (very limited number of available H+ ions at high pH), the proton pump in cell membranes functions much

Sulphur-related issues in deep underground nuclear waste disposal

167

less efficiently at high pH than at low pH. Moreover, membrane protein degradation and hydrolysis reactions are more severe under alkaline conditions than in acidic media [3]. Culture experiments conducted with SRB isolated from Boom Clay water revealed no SRB activity above pH 11. Prolonged exposure (4 months) of these SRB to high pH environments (pH 11, 12 and 13) showed that viable SRB remained present [4]. Experiments should be designed to test alkaliphile SRB strains, based on a careful overview of the existing literature. Another argument also deals with poor life evolution at high pH. On Earth, extreme acidic environments are relatively frequent (e.g. volcanic lakes, sulphide ore mine tailings, etc.), while alkaline environments are much rarer. The pH of soda lake (natron-, trona-rich evaporites in desert environments, etc.) does not exceed a value of 12 typical for natural sodium carbonate (Na2CO3) salts [5]. In general, life only evolves to adapt to extreme conditions. As high-alkaline environments are rare in nature, life has not been exposed to such conditions too often and as a consequence, evolution to adapt to high pH has not progressed much. Degradation of cement and concrete is commonly observed under aerobic conditions and could occur in the early phase of repository operations. Sulphur oxidizing bacteria oxidize sulphur, sulphides and thiosulphates under aerobic conditions to produce sulphuric acid. This acid can attack the cement paste matrix by dissolving calcium silicate hydrate gel (CSH) and Ca(OH)2 (portlandite). Direct anaerobic corrosion of concrete is not known [6]. However, microbes can develop at the surface of carbonated cement, i.e. the external surface and the surface of large cracks occurring in concrete blocks. The presence of biofilms developing inside open fissures in old or damaged concrete structures should be investigated. 4

How can the effect of irradiation on sulphide be studied?

There is a lack of knowledge to develop radiolysis models integrating a sulphur chemistry module. A gap exists for the sulphur species between (–I) and (III) valences. Irradiation of blast furnace slag cement pastes in the UK showed the formation of calcium tri-sulphoaluminate hydrate (AFt). The observation that an AFt type phase is only present in the irradiated blends may imply that the oxidation of reduced sulphur species (S2−) into sulphate (SO42−) is accelerated by gamma irradiation [7]. A good knowledge of the fundamental radiochemistry of sulphur is thus essential. 5 What happens when aerobic and anaerobic zones occur concurrently and can this be avoided? In case of disposal system concepts without buffer material installed in close contact with the canisters, an anodic area cannot be ruled out. This is the main disadvantage of leaving an open annular space around the canisters for the sake of retrievability. The only way to overcome the problem is to completely backfill the void of the annulus. Another major advantage of the complete backfilling (with dense swelling bentonite or cementitious materials) is to limit microbial activity and to control the transport of sulphide in the system by diffusion.

168

Sulphur-assisted corrosion in nuclear disposal systems

6 How can we predict what will/can happen during excavation and disposal phases? What is the impact on metal corrosion? It is important to know the spatial extent of the oxidized zone in the disturbed region of the excavation in a clay formation and to assess the quantity of H2SO4 that could be produced by pyrite oxidation around a gallery. The thickness of the oxidized zone is presently estimated not to exceed about 2 m for the galleries excavated at Mol in the Boom Clay [8]. It is also important to evaluate the impact of the salts that have accumulated in open and ventilated galleries on the degradation of cementitious buffers (enhanced risk of external sulphate attack (ESA) by ettringite and thaumasite crystallization) and on the corrosion of steels. The continuous water flow towards an open and ventilated gallery limits the diffusion of oxygen in the clay formation and flushes back sulphate along with chloride towards the gallery. As a consequence, salts accumulate near the gallery as can be observed by efflorescence of sulphate on the walls of ventilated galleries. High salt amounts that have accumulated in the engineered barrier system (EBS) might influence the corrosion of the metallic overpack and should therefore also be taken into account. The possibility of ingress of aggressive species within the EBS and its consequences for key safety functions, particularly the total engineered containment required during the first centuries or millennia, should be assessed to determine the maximum amount of salt that could accumulate in, or around, an open and ventilated gallery. The maximum time duration acceptable to leave a gallery open could be more limited by the consequences of salt accumulation in the near-field of a gallery than by the extent of the oxidized zone in the formation, i.e. near-field effects might be more limiting than far-field implications. Anaerobic corrosion of iron will also produce hydrogen that can easily be used by bacteria as an electron source to reduce CO2 in acetate or to achieve sulphate reduction. Finally, underground human activity can leave various contaminations in, or around, galleries: nitrate as residues of blasting operations, organic matter (cement admixtures, wood beams abandoned behind concrete lining after supporting gallery roof or walls, oil and various technological wastes). Quality assurance should be applied and a strong regulation should be enforced to avoid, or at least minimize, the introduction of organic matter in a repository. The mass balance of inventories of critical materials before and after each operation in which they are used could be helpful to guarantee that they have been totally removed and to minimize possible surprises. 7 How can a better interaction between modellers and experimentalists (boundary conditions, kinetic data, etc.) be achieved? Mathematicians translate their conceptual models in terms of equations, but sometimes they are not sufficiently familiar with lab experiments. Ideally, experimentalists, well aware of key processes and lab test limitations, should develop their own models, or work hand in hand with colleagues with good mathematical skills. In fact, it is not an issue if experimentalists model their own experiments themselves. Often, the problem is more a lack of verification (control of conceptual model applicability to the problem, correctness of equations and quality of resolution algorithms) or of validation (confrontation between model predictions and experimental results). An

Sulphur-related issues in deep underground nuclear waste disposal

169

appropriate design of experiments is also needed to obtain meaningful data and to be able to interpret the model results correctly, and preliminary scoping calculations are priceless to achieve that. 8 Will mass-transport limitations decrease the corrosion rate to very low values? Could the presence of S-species lead to the formation of protective sulphide containing films? Corrosion of metal surfaces in intimate contact with compacted bentonite is a diffusion controlled process as very well illustrated in the paper by King et al. [9]. Coupling point-defect models, or interfacial reaction parameters, at the metal/ bentonite interface to reactive transport calculations of corrosive species migrating in compacted bentonite, is certainly an important step to achieve more reliable long-term predictions on the corrosion of metals. The precipitation of an insoluble metallic sulphide film at the surface of metals, or metal oxides, could contribute to the formation of a protective layer passivating the metal surface. However, the barrier properties and the mechanical adhesion of porous sulphide films are far from being sufficiently understood. Acknowledgements This workshop was co-organized by the Belgian Nuclear Research Centre (SCK•CEN) and the Belgian Agency for Radioactive Waste and Enriched Fissile Materials (ONDRAF/NIRAS) under the auspices of the European Federation of Corrosion (EFC Event N° 311). The organizers of the SACNUC Workshop and the authors wish to thank all participants to the meeting for their fruitful discussions and the exchange of numerous ideas summarized during the panel discussion. References 1. K. Pedersen, M. Motamedi, O. Karnland and T. Sandén, J. Appl. Microbiol., 89 (2000), 1038–1047. 2. K. Horikoshi, Microbiol. Mol. Biol. Rev., 63(4) (1999), 735–750. 3. T. A. Krulwich, ‘Alkaliphilic prokaryotes’, in The Prokaryotes, Vol. 2, 3rd edn, ed. M. Dworkin, S. Falkow, E. Rosenberg, K.-H. Schleifer and E. Stackebrandt. Springer Science, Singapore. 4. S. Aerts, Use of Inhibitors to Prevent Bacterial Artefacts in Experiments, report SCK•CEN-ER-65, SCK•CEN, Belgium, 2008. 5. W. D. Grant, ‘Alkaline environments and biodiversity’, in Extremophilies, ed. C. Gerday and N. Glansdorff. In Encyclopedia of Life Support Systems (EOLSS), developed under the auspices of the UNESCO, Eolss Publishers, Oxford, UK. 6. J. M. West, I. G. McKinley and S. Stroes-Gascoyne, ‘Microbial effects on waste repository materials’, in Interactions of Microorganisms with Radionuclides, ed. M. J. Keith-Roach and F. R. Livens, Elsevier, Oxford, UK, 2002. 7. I. G. Richardson, G. W. Groves and C. R. Wilding, Mater. Res. Soc. Symp. Proc., 176 (1990), 31–37. 8. M. De Craen, M. Honty, M. Van Geet, E. Weetjens, X. Sillen, L. Wang, D. Jacques and E. Martens, Overview of the Oxidation around Galleries in Boom Clay (Mol, Belgium), NF-PRO report D4.3.24, 2008. 9. F. King, M. Kolar and M. Vähänen, ‘Reactive-transport modelling of the sulphideassisted corrosion of copper nuclear waste containers’, presented at the SACNUC2008 Conference, to be published.

INDEX

Index Terms

Links

A Abd El Aal, E.E.

100

Abd El Haleem, S.M.

100

abiotic conditions

82

acidification

72

Alberta Sulfur Research (ASR)

62

alkaline environments

167

Alloy 825

58

Alloy G3

59

anaerobic corrosion

95

62

8

15

51

72

79

81

95

152

163

168

77

83

118

156

127

132

135

152

155

effect of sulphur-based species on

95

kinetics of

86

thermodynamics of

83

anodic polarization curves

38

anodic processes

70

anoxic conditions

101

archaeological artefacts

78

Äspö

89

autocatalysis

51

104

102

B backfill of annular space

167

bentonite

109

Bich, N.N.

63

bio-oxidation

72

137

165

This page has been reformatted by Knovel to provide easier navigation.

Index Terms

Links

Boden, P.J.

51

Boggs, J.E.

48

Boom Clay repositories

2

19

42

81

89

81

88

101 boundary conditions for mass-balance equations Bouniol, P. breakdown potentials Butler–Volmer (B–V) expressions

158

163

5 74 158

C carbon steel

carbonate scaling cathodic depolarisation cathodic reaction

42

49

62

96

143

150

77

84

49 102 53

Chabaud, A.

127

Characklis, W.G.

102

Charlton, R.S.

97

Chaung, H.E.

58

Cheng, X.L.

48

156

Chlorides and corrosion of copper

109

in sour systems

46

chromic oxide barriers

42

44

chromium alloys

70

79

1

4

contained environment concept (CEC) copper alloys

148

copper corrosion

152

Copper Sulphide Model (CSM) of corrosion

154

160

This page has been reformatted by Knovel to provide easier navigation.

Index Terms

Links

corrosion-resistant alloys (CRAs)

54

Cragnolino, G.

24

Craig, B.D.

54

58

crevice corrosion

74

77

Crolet, J.-L.

48

Crowe, D.C.

97

cyclic voltammetry

62

117

D De Romero, M.

102

DeSilva, L.

102

E Eills, A.J.

56

electrochemical impedance spectroscopy (EIS)

40

111

122

143

148

electrochemical measurement techniques energy dispersive spectroscopy (EDS) Engen, R.J. engineered barrier systems (EBSs) Evans, U.R.

8 138 47 165

168

96

F Faraday, Michael Fe(III) oxyhydroxides

21 130

Full-scale Engineered Barriers Experiment (FEBEX)

137

149

This page has been reformatted by Knovel to provide easier navigation.

Index Terms

Links

G gamma radiolysis of water

15

gas generation through corrosion

81

100

1

66

73

138

152

165

geological disposal of nuclear waste

Giggenbach, W.

56

Goerz, K.

63

Gooch, T.G.

54

Gould, A.J.

96

Grauer, R.

86

green rust

94

Greer, J.B.

60

Gronboy, T.S.

24

Gunn, R.N.

54

79

81

84

102

H Hamby, T.W.

47

Hausler, R.H.

48

Hibner, E.L.

59

high-level nuclear waste (HLNW)

66

Ho-Chung-Qui, D.F.

47

Horowitz, H.H.

98

hydrocarbon condensate

62

hydrogen embrittlement

20

hydrogen evolution

71

8

22

27

hydrogen-induced cracking (HIC)

48

61

68

Hyne, J.B.

49

This page has been reformatted by Knovel to provide easier navigation.

Index Terms

Links

I Ikeda, A.

56

intermediate level waste (ILW) repositories

81

Intertech Ltd

55

iron sulphide scaling

46

irradiation effect on sulphide

86

49

53

61

167

J Jayalakshmi, M.

98

Jelinek, J.

87

K Kannan, S.

98

Kasnick, M.A.

47

Kelly, R.G.

98

Kesavan, S.

47

kinetics of corrosion

86

104

124

King, F.

110

154

169

Kolar, M.

110

Kreis, P.

86

Kursten, B.

8

L Lee, W.

102

linear polarisation resistance (LPR) measurements

142

Lino, M.

48

Loyless, J.C.

49

This page has been reformatted by Knovel to provide easier navigation.

Index Terms

Links

M Macdonald, D.D.

24

50

mackinawite

35

44

128

133

Maldonado-Zagal, S.B.

51

manganese sulphide

68

manometric gas cell technique

52

61

140

149

58

67

102

8

martensite

42

Martin, C.J.

58

mass-balance equations

157

Masurat, P.

155

mathematical modellers’ interaction with experimentalists

168

microbial activity in repositories

137

165

microbially-influenced corrosion (MIC)

101

137

micro-Raman spectroscopy

125

133

Mobile Bay oil wells

55

59

multi-barrier approach to waste disposal

81

Muralidharan, V.S.

98

N Neufeld, P.

87

nickel alloys

43

54

1

13

NIRAS/ONDRAF Nydam

78

Nyquist plots

40

70

O Oldfield, J.W.

59

Outo-kumpu HSC–5 Chemistry for

This page has been reformatted by Knovel to provide easier navigation.

Index Terms

Links

Outo-kumpu HSC–5 Chemistry for (Cont.) Windows software oxygen contamination in sour systems

24 47

P Pacheco, J.L. Pankhania, I.P.

46 102

passivation

70

pickling

10

pitting corrosion

62

77

98

68

71

98

138

91

99

147 polarisation curves

38

73

143

potential–pH diagrams

20

35

49

potentiodynamic polarisation

39

142

Pourbaix, Marcel

21

Poyet, S. pyrite

7 72

79

154

R reaction-diffusion equations

162

redox processes

153

Rhodes, P.R. rust layers

49 124

133

see also green rust

S safety of nuclear waste disposal determinants of Salvarezza, R.C. scanning electron microscopy (SEM)

165 97 134

Schikorr reactions

83

Schmitt, G.

50

138

142

This page has been reformatted by Knovel to provide easier navigation.

Index Terms

Links

Shannon, D.W.

48

Shoesmith, D.W.

48

Shrier, L.L.

24

Simon-Thomas, M.J.J.

49

Simpson, J.P.

86

simulation of corrosion behaviour slow strain rate testing (SSRT)

96

160

163

11

16

Smart, N.R.

9

Smith, S.N.

46

sour gas systems

46

98

54

59

100

14

81

68

73

81

41

55

spatial geometry of mass-transport barriers in repositories spent fuel (SF), disposal of

157 1

Sridhar, N.

46

stainless steels

42

54

138

146

steel corrosion by elemental sulphur

51

see also carbon steel; stainless steels stress corrosion cracking (SCC)

of corrosion-resistant alloys

11

16

20

71

138

140

58

stress-strain curves

11

sulphate-reducing bacteria (SRB)

19

73

78

100

124

130

135

137

148

152

155

165

sulphide scaling, protectiveness of

46

sulphide stress corrosion cracking (SSCC)

20

sulphur and anaerobic corrosion

81

in aqueous solutions

71

This page has been reformatted by Knovel to provide easier navigation.

Index Terms

Links

sulphur (Cont.) and corrosion-resistant alloys

55

in metals

70

and stress corrosion cracking

58

sulphur-assisted corrosion SACNUC workshop on sulphur chemistry sulphur-oxidising bacteria supercontainer (SC) concept

66

77

165 19

28

103

141

1

13

34

38

62

43

81

100

T Tassen, C.S.

59

thermodynamic stability

90

Thomason, W.H.

48

titanium alloys

60

Tromans, D.

97

152

U underground workings, contamination of repositories by uniform corrosion rates

168 8

15

V Vasquez Moll, D.V.

98

Vaughn, G.A.

60

Videla, H.A.

102

vitrified high-level waste (VHLW) disposal volt-equivalent diagrams

1

13

19

29

44

This page has been reformatted by Knovel to provide easier navigation.

Index Terms

Links

W Weetjens, E.

6

Wensley, D.A.

97

Wilde, B.E.

47

Wilhelm, S.M.

59

Wilken,G.

55

Williamson, A.I.

47

X X-ray diffraction (XRD) patterns

128

138

Y Yeske, R.A.

97

Z Zamaletdinov, I.I.

98

This page has been reformatted by Knovel to provide easier navigation.