CHARACTERISATION OF 3D PITTING CORROSION KINETICS OF STAINLESS STEEL IN CHLORIDE CONTAINING ENVIRONMENTS

Transcription

1 CHARACTERISATION OF 3D PITTING CORROSION KINETICS OF STAINLESS STEEL IN CHLORIDE CONTAINING ENVIRONMENTS A thesis submitted to The University of Manchester for the degree of Doctor of Philosophy in the Faculty of Science and Engineering 2017 Fahd A. Almuaili School of Materials

2 TABLE OF CONTENTS TABLE OF CONTENTS... 2 LIST OF TABLES... 7 LIST OF FIGURES... 8 ABBREVIATIONS ABSTRACT DECLARATION COPY RIGHT STATEMENT ACKNOWLEDGEMENTS INTRODUCTION Motivation Aim Objectives LITERATURE REVIEW Stainless steels Austenitic stainless steels Types of austenitic stainless steel Microstructure Effect of alloying elements Effect of temperature Heat treatments Carbide precipitation Effect of deformation Corrosion of stainless steel Electrochemical corrosion Corrosion thermodynamics Kinetics of electrochemical corrosion Corrosion kinetics control Pitting corrosion of stainless steel Characteristics of pitting corrosion Passive film on stainless steel Pit initiation Metastable pit growth Stable pit growth

3 2.3.6 Solution chemistry inside the pit Solution concentration inside the pit Stable pit dissolution Stable pit transition Growth rate of stable pits Salt film formation in stable pits Shape and morphology of pit growth Metal covers The effect of inclusions on pitting corrosion The effect of stress on pitting corrosion The effect of temperature on pitting corrosion Application of X-ray tomography Summary References EXPERIMENTAL PROCEDURES Materials Microstructure Sample preparation Optical microscopy (OM) ImageJ Scanning electron microscopy Electrochemical techniques Sample preparation Open circuit potential Electrochemical polarisation Electrochemical cells X-ray tomography experiments Principles of X-ray tomography X-ray sources Sample preparation Electrochemical experiments X-ray CT Set-up References RESULTS - OVERVIEW References EFFECT OF MICROSTRUCTURE AND INCLUSIONS ON PITTING CORROSION SUSCEPTIBILITY

4 5.1 Key findings and results On the relationship between pitting corrosion and microstructural inclusion density in austenitic stainless steels (Manuscript A) Abstract Introduction Experimental Results and Discussion Metallography and microstructure SEM and EDX analysis of inclusions Electrochemical measurement Effect of chloride concentration Effect of applied potential on pit density Conclusions Acknowledgements References OBSERVATION AND ESTIMATION OF 3D PITTING CORROSION KINETICS Key findings and results Application of a Quasi In-situ Experimental Approach to Estimate 3-D Pitting Corrosion Kinetics in Stainless Steel (Manuscript B) Abstract Introduction Experimental Results and Discussion Electrochemical polarisation Pit geometry and estimation of growth kinetics Current density and stability product Pit diffusion product estimations Conclusions Acknowledgements References Estimation of 3D Growth Kinetics of Multiple Pits in Austenitic Stainless Steel (Manuscript C) Abstract Introduction Experimental Set-up Results and Discussion Pit morphology Electrochemical polarisation Pit geometry

5 Pit growth kinetics Current density Stability product Estimation of pit diffusion product Conclusions Acknowledgements References EFFECT OF STRAIN ON THREE- DIMENSIONAL PITTING CORROSION KINETICS Key findings and results Strain-induced Re-activation of Corrosion Pits in Austenitic Stainless Steel (Manuscript D) Abstract Introduction Experimental Set-up Results and Discussion Electrochemical observation Pit geometry and dimensions Lacy cover morphology Reactivation of pits Pit growth kinetics Pit stability product Diffusion coefficient and pit diffusion product Conclusion Acknowledgements References Time-dependent Re-activation of Corrosion Pits in Austenitic Stainless Steel (Manuscript E) Abstract Introduction Experimental Set-up Results and discussion Experiment (I) Pit morphology Electrochemical polarisation Pit geometry and estimated growth Reactivation of pit Pit current density and stability product Pit diffusion product Experiment (II) Electrochemical polarisation Morphology of pit growth Pit geometry

6 7.3.5 Conclusions Acknowledgements References SUMMARY AND CONCLUSION References FURTHER WORK APPENDIX Final word count: 73,324 words 6

7 LIST OF TABLES Table 2-1: Types of stainless steel compared [34] Table 2-2: Types of austenitic stainless steel [39] Table 2-3: General effects of alloying elements on austenitic stainless steel [46] Table 3-1: Chemical composition (wt%) of austenitic stainless steel alloys used Table 5-1: Chemical composition (wt%) of austenitic stainless steel alloys used Table 6-1: Summary of in-situ electrochemical polarisation experiment Table 6-2: Measured pit geometries Table 6-3: Summary of in-situ electrochemical polarisation experiment Table 6-4: Geometry and size of pits generated by potentiodynamic (pits 1-4) and potentiostatic polarisation (pits 5-8) Table 7-1: Summary of in-situ electrochemical polarisation experiment Table 7-2: Measured pit dimensions Table 7-3: Summary of in-situ electrochemical polarisation experiments I and II Table 7-4: Measured pit geometries of the four growth approaches Table 7-5: Measured pit geometries for experiment II

8 LIST OF FIGURES Figure 2-1: Austenitic stainless steel family [40] Figure 2-2: Effect of alloy composition on microstructure (Schaeffler diagram).[44] Figure 2-3: Microstructural features inducing pitting and SCC development [45] Figure 2-4: Austenitic stainless steel time-temperature-transformation diagram [47] Figure 2-5: Influence of carbon concentration (C wt%) on sensitization time and temperature [40] Figure 2-6: Sketch and Cr profile showing chromium depleted zone adjusted to grain boundary [48] Figure 2-7: Types of grain boundary resistance to sensitisation as a function of temperature and time [40] Figure 2-8: Overpotential regions of electrochemical control reactions under polarisation [52] Figure 2-9: Pitting corrosion process [55] Figure 2-10: Schematic diagram of polarisation curve (forward and backward) of active-passive metal [57] Figure 2-11: Typical transient current in a pit nucleation event (unstable pit) on a type 304L stainless steel microelectrode [68] Figure 2-12: Passive film breakdown mechanisms: (a) penetration, (b) film breakdown, (c) adsorption [70] Figure 2-13: Transient current of metastable pit nucleation and growth on type 304L stainless steel [68] Figure 2-14: Galvele s one-dimensional model of pit dissolution [80] Figure 2-15: The relation of x.i to H +, cation concentration and hydroxide of Fe and Fe(OH + ) [80] Figure 2-16: Changes to solution chemistry inside an artificial pit with time [92] Figure 2-17: (a) Time of pit passivation after potential step-down and (b) current density over surface concentration [52]

9 Figure 2-18: Effect of various amount of FeCl 2 on anodic polarisation curves of stainless steel in 1 N HCl [94] Figure 2-19: Effect of solution concentration of pre-dissolved 304 and 316 stainless steel on the anodic polarisation curve of new samples [100] Figure 2-20: Schematic of anodic dissolution showing two periods of current-time behaviour[102] Figure 2-21: Steady state points between diffusion and rate of dissolution [52] Figure 2-22: (a) Two states of pit shape (polishing and etching surface) observed through anodic polarisation. (b) Transition of two conditions (active and passive) of pit from polishing state by potential decay, depending on pit size [138] Figure 2-23: Sketch of lacy metal cover and pit shape development as a function of concentration inside the pit. Passive film on the metal surface is shown by a thick line [28] Figure 2-24: Radiograph images of pit growth on foil of type 304 stainless steel in chloride solution at 650 mv vs Ag/AgCl [27] Figure 2-25: Sketch shows the position of salt film in (a) horizontal orientation and (b) vertical orientation samples Figure 3-1: (a) Standard electrochemical cell; (b) mini-electrochemical cell.93 Figure 3-2: Two X-ray sources: (a) micro-focus source with 2D detector, showing cone-beam geometry and image magnification, (b) synchrotron source with parallel beam geometry [5] Figure 3-3: (a) In-situ miniature electrochemical cell with straining rig, (b) schematic diagram of electrochemical cell parts Figure 4-1: Schematic overview of all manuscripts included in this results section Figure 5-1: Micrographs of the microstructure of (a) type 303 bar, (b) type 304 plate and (c) type 304L wire after electroetching with 10% oxalic acid. 108 Figure 5-2: Analysis of type 303 microstructure, with (a) summary of inclusion count versus size of inclusions, and (b) showing a backscattered micrograph(unetched) used for image analysis Figure 5-3: A typical inclusion in type 303 with corresponding EDX line profile

10 Figure 5-4: Type 304 plate microstructure (a) showing a backscattered micrograph (unetched) used for image analysis and (b) a typical inclusion in type 304 with corresponding EDX line profile Figure 5-5: Type 304L wire microstructure (a) showing a backscattered micrograph (unetched) used for image analysis and (b) a typical inclusion in type 304L with corresponding EDX line profile Figure 5-6: Electrochemical properties of type 303 bar, type 304 plate and type 304L wire in 0.1 M NaCl solution summary of OCP values after 15 min and pitting potential (E pit ) after potentiodynamic polarisation at a scan rate of 1 mv.s Figure 5-7: OCP over time of type 303 bar, type 304 plate and type 304L wire in 0.1 M NaCl solution Figure 5-8: Pit density of type 303 bar, type 304 plate and type 304L wire at various end potentials of the potentiodynamic polarisation scan at 1 mv.s - 1 scan rate in (a) 0.1 M NaCl solution and (b) 1 M NaCl solution Figure 5-9: Relationship between pit density and inclusion density in type 303 bar, type 304 plate and type 304L wire in (a) 0.1 M at +400 mv vs. SCE and (b) 1 M NaCl solution at+300 mv vs. SCE Figure 6-1: (a) Photo of the miniature electrochemical cell with the capability to apply strain for in-situ x-ray tomography experiments with a type 304L wire sample, (b) in-situ cell during an X-ray CT experiment Figure 6-2: (a) Reconstructed X-ray CT data volume of the wire after the 1 st electrochemical polarisation scan, (b) X-ray CT data volume after the 2 nd potentio-dynamic polarisation scan, and (c) SEM image of the wire sample with the three pits Figure 6-3: (a) Current evolution vs. time of the 1 st and 2 nd potentiodynamic polarisation scan with (b) measured depth (r), width (w) and height (h) of all three pits from X-ray CT data Figure 6-4: The values in Table 6-2 of the three approaches A, B, and C are applied to show (a) pit 1 current density vs. time, (b) pit 2 and 3 current density vs. time, (c) pit stability product vs. time of pit 1, and (d) pit stability products vs. time of pit 2 and

11 Figure 6-5: Effect of time on (a) (pit depth) 2 to obtain the diffusion parameters of these curves from the slopes, (b) effect ive diffusion coefficient (D) estimated with a constant metal ion concentrations of 4.2 M. 144 Figure 6-6: Image of the miniature electrochemical cell and a type 304L wire sample inside the X-ray CT equipment Figure 6-7: Reconstructed X-ray CT images of the two sides of the wire: (a, b) after the 1 st electrochemical polarisation scan; (c, d) after the 2 nd potentiodynamic polarisation scan Figure 6-8: Cross-sectional isosurface view of wire sample: (a) all pits, (b) close-up view of pit 5 morphology Figure 6-9: Tomogram section of 2D pit from middle slice of pit 5 at (a) height and (b) width view Figure 6-10: 2D tomograms of pits along the middle axis of each pit, showing the shape as a function of pit orientation. All pits grew during exposure of the wire in the in-situ cell, with the vertical axis shown by the arrow Figure 6-11: SEM images of the two sides of the wire sample, showing all eight pits Figure 6-12: (a) Current vs. time during 1 st and 2 nd potentiodynamic and potentiostatic polarisation cycles (see appendix C for potential vs current); (b) depth, width and height of pits calculated from X-ray CT data Figure 6-13: The estimated current evolution of each pit based on the total current. The initiation point for pits with (a, b) potentiodynamic and (c, d) potentiostatic polarisation are shown. More information of the current separation for each pit is given in Appendix C Figure 6-14: Current density over time for pits 1-4 under potentiodynamic polarisation by (a) approach A and (b) approach B, and for pits 5-8 under potentiostatic polarisation by (c) approach A and (d) approach B Figure 6-15: Stability product over time of multiple pits on 304L stainless steel wire in 0.1 M NaCl solution grown under potentiodynamic polarisation by (a) approach A and (b) approach B, and under potentiostatic polarisation by (c) approach A and (d) approach B

12 Figure 6-16: Graphs of depth 2 over time and values of diffusion product obtained from their gradient for multiple pits grown under (a) potentiodynamic and (b) potentiostatic polarisation Figure 6-17: Diffusion coefficient over time, obtained by approach A, for multiple pits generated on 304L stainless steel wire in 0.1M NaCl solution, under (a) potentiodynamic and (b) potentiostatic polarisation Figure 7-1: (a) Image of the electrochemical cell with straining rig for in - situ X-ray CT experiments, (b) cross sectional optical micrograph of the annealed type 304L stainless steel wire after etching in 10% (wt.) oxalic acid at 6V Figure 7-2: Reconstructed X-ray CT data of the wire sample: (a) after the 1 st electrochemical polarisation scan (X-ray CT scan A), (b) after ~5% plastic strain (X-ray CT scan B), (c) after the 3 rd electrochemical polarisation cycle (X-ray CT scan C). The resolution of the scans is not sufficient to show the lacy metal covers Figure 7-3: (a) Current evolution with time during the first and third polarisation cycles and (b) summary of measured pit dimensions taken from X-ray CT data Figure 7-4: SEM images (a) of the type 304L wire sample showing the two pits, (b) pit 1 lacy cover morphology, and (c) pit 2. Note: cracks in the lacy covers are also shown Figure 7-5: Segmented pits from X-ray CT data (a) Pit 1 generated by 1 st polarisation scan, (b) pit 1 after ~5% strain, (c) the difference between (a) and (b), (d) reactivated pit 1 and pit 2 generated during potentiostatic polarisation at the end of the 3 rd cycle, with (e) showing the transparent volume of pit 1 reactivated and pit 2 relative to pit 1 before reactivation. 196 Figure 7-6: (a) Pit 1 current density over time estimated using surface area values calculated via approaches A and B (Table 7-2), (b) current density of pits 1 and 2 over time after strain and repolarisation, calculated using surface area values obtained via approaches A and B Figure 7-7: Pit stability product vs. time estimated from data obtained via approaches A and B (Table 7-2) for (a) pit 1, (b) reactivated pit 1 and pit

13 Figure 7-8: (a) (depth) 2 vs. time and the diffusion product (slope) for pit 1 before and after reactivation and for pit 2, (b) diffusion coefficient of all pits assuming a constant salt concentration of 4.2 M Figure 7-9: Image of the electrochemical cell with miniature straining rig and load cell for in-situ X-ray tomography experiments with a type 304L wire sample Figure 7-10: Reconstructed X-ray CT data of the wire sample: (a) after first electrochemical polarisation scan (step 2), (b) after ~5% plastic strain (step 4), (c and d) after third electrochemical polarisation cycle (step 6) Figure 7-11: SEM images of the wire sample showing the four pits: (a) pit 1; (b) pit 2, showing lacy cover morphology wit h cracks; (c) pit 3; (d) pit Figure 7-12: Morphology of pits grown in experiment (I) showing (a) tomogram section of 2D view of pits 1, 3 and 4; and (b) 3D surface view of pits 1, 2 and Figure 7-13: (a) Current evolution over time of the potentiodynamic polarisation cycles before and after applying strain; (b) final pit geometry.216 Figure 7-14: Estimated current evolution of each pit from total current evolution of the second polarisation Figure 7-15: Tomogram section of 2D pit 1 (a) before strain, (b) after strain and (c) after reactivation Figure 7-16: Pit 1 (a) current density over time, (b) stability product vs. time; estimated by methods A, B, C and D ( Table 7-4) Figure 7-17: Current density vs. time for reactivated pit 1 and pits 2, 3 and 4 after strain and repolarisation, calculated by methods (a) A, (b) B, (c) C and (d) D Figure 7-18: Stability product vs. time for reactivated pit 1 and pits 2, 3 and 4 after strain and repolarisation, calculated by methods (a) A, (b) B, (c) C and (d) D Figure 7-19: (a) Relationship between (depth) 2 vs. time and diffusion product (slope) for all pits before and after the application of strain; (b) diffusion coefficient of all pits, assuming a constant salt concentration of 4.2 M; (c) metal ion concentration over time of all pits Figure 7-20: Reconstructed X-ray CT data of wire sample (a) after 1 st electrochemical polarisation (step 2), (b) with ~10% plastic strain (step 4) 13

14 and (c1-c3) after 2 nd electrochemical polarisation cycle (step 6). The data for figures (c1-c3) were acquired after removing the wire from the solution, producing a reconstructed voxel size of 0.71 µm Figure 7-21: (a) Current evolution over time for 1 st potentiodynamic polarisation scan, with evolution of dissolved volume; (b) current evolution of 1 st polarisation compared to 2 nd polarisation scan after 10% strain was applied Figure 7-22: SEM images (a) of the wire sample showing the growing pits; (b) close-up view of pits with lacy cover containing cracks and fractures; and (c) close-up view of large open pit growing close to beeswax resulting in a crevice and connection to other pits Figure 7-23: X-ray CT data visualising the 3D morphology of the inside of the wire, showing (a) 3D cross-sectional view of all pits, (b) 3D vertical view of the outside

15 ABBREVIATIONS Ag/AgCl Silver/silver chloride electrode AISI American iron and steel institute ASTM American society for testing and materials BCC Body-centric cubic BSE Backscattered electron BSI British Standard institute CCD Charge-coupled device CPT Critical pitting temperature EDX Energy-dispersive X-ray spectroscopy FCC Face-centred cubic HCP Hexagonal close-packed IGC Intergranular corrosion IGSCC Intergranular stress corrosion cracking IR Potential drop OCP Open circuit potential OM Optical microscopy OPS Oxide polishing suspension PDM Point defect model PDP Potentiodynamic polarisation PREN Pitting resistance equivalent number SCC Stress corrosion cracking SCE Saturated calomel electrode SE Secondary electron SEM Scanning electron microscope SHE Standard hydrogen electrode UNS Unified numbering system X-ray CT X-ray computed tomography XRF X-ray fluorescence analysis XPS X-ray photoelectron spectroscopy 15

16 ABSTRACT The research reported in this PhD thesis provides a novel approach to estimate 3D pitting corrosion kinetics of austenitic stainless steel with exposure to chloridecontaining aqueous environments. A quasi-in-situ X-ray computed tomography (X-ray CT) approach was developed, with the aim of providing an experimental methodology to estimate 3D pitting corrosion kinetics under different exposure conditions. The first part summarises a set of preliminary investigations to identify the pitting corrosion behaviour of three austenitic stainless steels (type 303 bar, type 304 plate and type 304L wire) with different inclusion contents. All observed pit densities were related to the inclusion contents, providing confidence in moving to the next stage of the project, for conducting in-situ corrosion studies using X-ray CT. The second section describes the construction of an in-situ electrochemical cell for X- ray CT studies, the aim being to provide an experimental methodology to estimate 3D pitting corrosion kinetics. Pit growth kinetics of individual pits were estimated from segmented 3D X-ray CT data. The evolution of pit current densities, associated pit stability products, and diffusivity parameters over time were obtained. The study also showed that the kinetics of multiple pits could be estimated using this novel approach, based on separating the current response of each pit over time. This was obtained by electrochemical polarisation control and measuring the total current evolution. The third section discusses the effect of plastic strain on 3D pitting corrosion kinetics. Several in-situ X-ray CT experiments were conducted, with a focus on obtaining 3D pit growth, passivation, and re-activation kinetics, to elucidate the effect of applied strain on pit stability and growth. This section explains a possible mechanism for the reactivation of pre-existing corrosion pits, showing that pits grew more rapidly during reactivation than those grown before plastic strain was applied. A marked difference in pit morphology with fractured lacy metal covers was observed with the application of strain. The implications of this observation are discussed in light of stress corrosion crack nucleation mechanisms. 16

17 DECLARATION No portion of the work referred to in this thesis has been submitted in support of an application for another degree or qualification of this, or any other university, or other institute of learning 17

18 COPY RIGHT STATEMENT (i) The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the Copyright ) and he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes. (ii) Copies of this thesis, either in full or in extracts, and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. (iii) The ownership of certain copyrights, patents, designs, trademarks and other intellectual property (the Intellectual Property ) and, any reproductions of copyright works in the thesis, for example graphs and tables ( Reproductions ), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. (iv) Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy (see in any relevant Thesis restriction declarations deposited in the University Library, The University Library s regulations (see and in The University s policy on Presentation of Theses. 18

19 ACKNOWLEDGEMENTS First of all, my grateful thanks to God for giving me the health and strength to complete my thesis. Secondly, I would like to express my great thanks to Dr. Dirk Engelberg, my academic supervisor, for supporting and encouraging me to complete this work. My great thanks are also extended to my academic co-supervisor, Prof. Philip Withers, for his help and support. I am also grateful to those from whom I received initial support for my work: Paul Jordon, Steve blatch and Harry Pickford. During the work, I am grateful to Dr. Tony Cook for his valuable discussion and support in electrochemical study, and to Michael Faulkner and Shirley for their training and help in using the scanning electron microscope. I would like to thank all the staff in the workshop of the School of Materials for their help during design and fabrication of the mini electrochemical cell. I deeply appreciate Dr. Sam McDonald for his support and training on the X-ray tomography equipment. I also appreciate the valued comments of Prof. Bob Akid. I extend my gratitude to all my lab mates and colleagues: Dr. Fred, Dr. James, Dr. Cem, Ahmed, Sanskar, Choen May, Mickel, Dr. Philip Platt, Pier, Yanli and Ruth for the friendly work environment. It was a pleasure to work with them. I must acknowledge the Saline Water Conversion Corporation of Saudi Arabia for sponsoring my PhD study. I am eternally grateful to my parents, my wife Maha, my daughter Lujain, my sons Omar and Faris, my sisters and my brothers for their patient support and encouragement throughout my studies. 19

20 1. INTRODUCTION Austenitic stainless steels are widely used in many industrial applications due to their good corrosion resistance, strength, excellent formability, weldability and toughness in a wide range of temperatures [1]. However, in the presence of corrosive species, localised corrosion often occurs and can lead to accelerated attack, such as pitting corrosion, crevice corrosion, stress corrosion cracking (SCC), or corrosion fatigue. Localised damage then threatens the integrity of components and may cause unexpected failures, especially if combined with applied or residual stress or strain [2]. The process of local attack is often described in two stages, known as initiation and propagation, with a number of mechanisms proposed for each stage [3]. During the first stage, a passive film protects the surface. The film is very thin (typically a few nm) and in corrosive environments a continuous process of breakdown and repassivation may occur. The mechanism of pit initiation under the process of film breakdown is still not fully understood and is difficult to predict, often being referred to as stochastic in nature [4]. The alloy microstructure, including the distribution of elements or second phases/inclusions, is an important factor which can significantly affect the corrosion resistance of stainless steel. MnS-type inclusions are well known to act as preferential sites for pits to initiate and a number of mechanisms for initiation have been proposed, but the effects of individual parameters on the overall process remain subject to further investigation [4]. For example, the relationship between pitting corrosion resistance and type of inclusion (type, size, composition) has been investigated on a macro and micro scale. However, aspects of pit initiation at inclusions, such as whether they dissolve or function as crevice nucleation sites, are still under investigation. Other factors of interest include the potential at which the dissolution of inclusions starts, the presence of Crdepleted microstructure regions around inclusions and stresses in the oxide covering inclusions [5-8]. The process of pit propagation is somewhat better understood, but there is no universal mechanism describing the effect of different factors on subsequent pitting corrosion attack which can be used to predict the severity of pitting. For example, the effects of microstructural parameters such as secondary phases and MnS inclusions, and of 20

21 superimposed stress or strain are often not considered at all in mechanistic pit propagation studies. This is because the environment inside a pit is complicated and there can be a number of simultaneous processes (e.g. dissolution and passivation), which are also dependent, for example, on the local environment and the chemical composition of the material, such as the nickel and molybdenum content [9, 10]. The dissolution and passivation processes inside the pit also depend on pit geometry, which affects the mass transport of species between the bulk environment and the pit interior [11, 12]. The effects of electrochemical potential and strain on susceptibility to pitting corrosion have been investigated [13-16]; however, the influence of electro-chemically polarising samples containing pits is not understood and it is not known whether the presence of applied, residual, or dynamic strain may affect propagation kinetic. Some attention has been given to conditions that act as precursors of pit-to-crack transition, to identify critical steps in the nucleation of SCC, but there is a need for these to be thoroughly investigated [17, 18]. Furthermore, the dynamic relationship between pitting corrosion kinetics in 3D, observed using in-situ techniques such as X-ray computed tomography (X-ray CT), has only recently become accessible to laboratory-based research. Synchrotron X-ray CT has been used for more than a decade to investigate corrosion processes [19-27]. Measurements of pit growth kinetics are generally made using electrochemical methods combined with supporting investigations, based either on pit depth measurements, assuming a hemispherical pit shape, or on observations of boundary movement using 2D geometry of pits grown in foils or under a capillary electrode [24, 27-29]. Beside the importance of these measurements, it is not clear whether such observations simulate real 3D pit growth, because pit geometry influences pit chemistry and associated transport processes, which in turn control the rate of pit growth. Therefore, it is important for a better understanding of pit growth kinetics to conduct measurements based on real 3D components, ideally with exposure to the bulk environment. Such measurements, taken in situ in realistic conditions close to those of actual exposure, have not been reported so far. In this work, austenitic stainless steels of types 304, 304L and 303 were chosen for a study of pitting corrosion. The types 304 and 304L are commonly used in desalination 21

22 plants, in the chemical industry and in other high-temperature applications [1, 30]. First, a systematic laboratory study was made of the influence of microstructural inclusions and the number of pit initiation sites on the severity of pitting corrosion, to better understand the behaviour of austenitic stainless steels in aqueous NaCl electrolyte for in-situ studies. All in-situ measurements undertaken in this study then focused on 3D pit growth in type 304L wire samples, with exposure to bulk electrolyte, with and without applied strain, and under electrochemical polarisation control. 1.1 Motivation The susceptibility of austenitic stainless steel to pitting corrosion depends on the influence of microstructure, which varies according to inclusions and environmental conditions. For instance different inclusion size and composition shows different resistance to pitting corrosion depending on the concentration of chloride concentration and temperature. In the first part of the study, the relationship between inclusion density and susceptibility to pitting corrosion was investigated to better understand this corrosion system for in-situ 3D pitting corrosion control. The relationship between pit density and inclusion density was investigated using three austenitic stainless steels with different sulphur and carbon contents. A study of the relationship between pit density and pitting susceptibility was conducted in the laboratory at different applied potentials and various chloride concentrations. The aim was to develop a system that could be controlled for growing a small number of pits for in-situ X-ray CT. An X-ray CT study of pitting corrosion allows the measurement of 3D morphology and kinetics. It is a non-destructive technique that allows observation and measurement of real pit shape, which can be used to determine pit growth kinetics. More precise models of pit growth kinetics can then be developed and used to predict the severity of pitting attacks. For this purpose, an experimental methodology to investigate 3D pitting corrosion in bulk electrolyte was developed, and the kinetics of 3D pitting corrosion compared to 2D studies. A miniature X-ray CT setup was developed to perform these measurements, based on a miniature three-electrode electrochemical cell. Samples were electrochemically (re-)polarised without removing them from the electrolyte, to study the influence of polarisation potential on generating and/or reactivating pits. Test were performed to assess possible mechanisms for reactivating corrosion pits, to understand how applied strain might affect the integrity of pre-existing pits, whether they then act 22

23 as precursors for pit-to-crack transition, or whether pit growth is facilitated. For example, higher applied potential and strain were reported to accelerate the dissolution of corrosion pits [16, 31, 32]. However, accelerated pit growth may then also prevent SCC from nucleating, in the best case even leading to a reduction in strain/stress. The role of stress, in this case, is to facilitate pit initiation by increase the number of active sites. Therefore, in-situ experiments were performed for the 3D visualisation of pitting corrosion under electrochemical polarisation and applied strain. 1.2 Aim The overarching aim of the PhD project is to develop a better understanding of pitting corrosion of austenitic stainless steel in a chloride-containing environment. The first goal was therefore to develop an experimental methodology to obtain information about ex-situ pitting corrosion attacks on three conventional stainless steel materials: types 304, 304L and 303. Work was conducted to investigate the relation between susceptibility to pitting corrosion and density of microstructural inclusions. The first part of the in-situ study then aimed to investigate 3D pitting corrosion kinetics on exposure to bulk electrolyte. For this purpose, a miniature three-electrode electrochemical cell was developed for carrying out a quasi in-situ X-ray CT studies. A study was conducted to grow a small number of pits and different geometrical assumptions reported in the literature were explored. The second goal was to obtain 3D pit growth kinetics and compare those to literature reports of 2D and 3D kinetics. This included in-situ assessment of a single pit, the growth of two pits, as well as multiple pits, in a wire sample using X-ray CT studies. The final goal was to carry out a systematic study of the effects of repolarisation and applied strain on pitting corrosion kinetics and pit reactivation characteristics. No SCC was observed in this study, but brittle cracks in lacy metal covers were observed, and their potential effect on SCC is briefly discussed. 1.3 Objectives To investigate a localised corrosion system for types 303, 304 and 304L steel, in a chloride-bearing aqueous environment, for subsequent in-situ studies of 3D pitting corrosion. This includes a study of the influence of inclusion contents on 23

24 the pitting corrosion attack under electrochemical potential control and in different chloride concentrations. To develop a method to carry out in-situ 3D pitting corrosion studies using X- ray CT. This involved the construction of an electrochemical cell which would fit inside the X-ray CT instrument to study pitting corrosion in bulk solution under electrochemical control. The cell was also designed to apply tensile strain on wire electrodes. To carry out in-situ 3D pitting corrosion studies in bulk electrolyte and to determine the kinetics of pitting corrosion under electrochemical control. For this purpose, electrochemical polarisation was conducted to grow single and multiple pits for in-situ studies. To carry out an advanced study of the effects of repolarisation potential on the kinetics of 3D pitting corrosion and pit reactivation morphology, to learn about the synergetic effects of strain on pitting corrosion kinetics, pit reactivation morphology and crack nucleation. 24

25 2. LITERATURE REVIEW 2.1 Stainless steels Stainless steels are alloys of iron and chromium with the addition of other alloying elements used to modify their structure and properties. Generally based on Fe-Cr, Fe-Cr-C or Fe-Cr-Ni, they contain at least 11% chromium as the main alloying element. This chemical composition is necessary to develop a chromium oxide layer, which acts as a barrier, preventing the metal surface from staining or corroding in an atmospheric or aqueous environment [33, 34]. This chromium oxide layer forms spontaneously in the presence of oxygen, but its stability can be limited by inclusions and it may show some mechanical or chemical deterioration, depending on environmental conditions. For example, a chloride-rich environment induces pitting corrosion and SCC of stainless steel, which may develop under certain conditions by breaking down the passive layer [2]. Stainless steels comprise a large group of alloys with a wide range of chemical compositions involving many different elements. This variety produces alloys with widely differing microstructures and distinct mechanical properties. In the literature, their microstructure is normally classified into five main types: ferritic, austenitic, martensitic, duplex and precipitation hardening stainless steel. Table 2-1 summarises the advantages and disadvantages of each type [34]. Stainless steels are widely used in the chemical industry, in desalination, in power plants and in gas and oil fields, because of their attractive lifecycle cost, good mechanical properties and corrosion resistance [1, 30, 35]. 25

26 Table 2-1: Types of stainless steel compared [34] Type Examples Advantages Disadvantages Ferritic 410S, 430 Austenitic 304, 316 Martensitic 420, 431 Duplex 2205 Precipitation Hardening 17/4PH Low cost, moderate corrosion resistance and good formability. Widely available, good toughness, excellent formability and weldability Low cost, hardenable by heat treatment with high hardness. Good SCC resistance, good mechanical strength in the annealed condition Hardenable by heat treatment, but with better toughness and corrosion resistance than martensitic. Limited corrosion resistance, formability and elevated temperature strength compared to austenitic. Work hardening can limit formability and machinability. Limited resistance to SCC. Less corrosion resistance than austenitic and less formability than ferritic. Limited weldability. Application temperature range more restricted than austenitic. More expensive and less widely available than austenitic. Limited availability, corrosion resistance, formability and weldability restricted compared to austenitic Austenitic stainless steels Austenitic stainless steels form the largest and most important group, based on the number of alloys and the volume of usage since the 1920s [36]. They occupy this position because of their high corrosion resistance and strength, excellent formability, weldability and toughness within a wide range of temperatures, from cryogenic to high temperatures [37]. Cr and Ni are the principle chemical elements in austenitic stainless steel and are generally added in the range of 15-27% Cr and 8-35% Ni, but other elements such as Mo, N, Mn and W are also added for specific properties. This type has a non-magnetic structure and about 30% less thermal conductivity and 50% higher thermal expansion coefficient than carbon steel [37]. As to its general mechanical properties, it has yield strength of MPa, ultimate tensile strength of MPa and elongation of 30 to 60% with a good toughness at low temperature [38]. There are two main groups of austenitic stainless steels: the 300 and 200 series. The 200 series has less Ni but more Mn and N than the 300 series, so the latter develops higher strength through cold working. Austenitic stainless steels are widely used in a variety of engineering applications such as in nuclear power plants, automobiles, the oil and gas industry and desalination plants [30, 36]. 26

27 2.1.2 Types of austenitic stainless steel Austenitic stainless steel is generally subdivided into five groups, based on composition, as shown in Table 2-2 [39]: Grade 304 is the most popular alloy, from which most austenitic stainless steels can be derived by modifying the elements included and their proportions, as presented in Figure 2-1 [33]. Table 2-2: Types of austenitic stainless steel [39] S No Type of stainless steel Examples 1 Conventional austenitic Types 302, 303, 304 and Stabilised austenitic Types 321, 347 and Low carbon grade austenitic Types 304L, 316L and 317L 4 High nitrogen austenitic Types 304N, 316N, 201 and High alloy austenitic Types 317LM, 317LX, 904L, AL6XLN and 254SMO Figure 2-1: Austenitic stainless steel family [40]. 27

28 2.1.3 Microstructure Austenitic stainless steel has a gamma (γ) phase whose crystal structure is face-centred cubic (FCC). Austenitic structures develop at room temperature from the primary solidifying phase of ferrite or austenite, depending on the composition balance, which controls the microstructure through two groups of elements known as ferrite and austenite stabiliser elements [41]. The group of alloy elements which maintain the austenitic structure is characterised by the nickel weight percentage as an equivalent number, while elements in the other group, which maintain the ferrite structure, are represented by the chromium weight percentage. These are shown as equivalent numbers in equations 2-1 and 2-2 [42]: Cr% equivalent = Cr + 2Si + 1.5Mo + 5V + 5.5Al Nb + 1.5Ti W (2-1) Ni% equivalent = Ni + Co + 0.5Mn Cu + 25N + 30C (2-2) The effects of these element groups are indicated in the Schaeffler diagram (Figure 2-2), which was originally developed for welding and used to predict microstructure at room temperature. Therefore, the solidification and transformation from liquid (L) to solid state in austenitic stainless steel can be divided into three types [43]. (1) Austenitic (γ): L L+γ γ (2) Austenitic-ferritic (γ+δ): L L+γ L+δ+γ γ+δ (3) Ferritic-austenitic (δ+γ): L L+δ L+δ+γ δ +γ 28

29 Figure 2-2: Effect of alloy composition on microstructure (Schaeffler diagram).[44] Furthermore, the primary solidification phases can be predicted by using the ratio between the Cr and Ni equivalent numbers. For example, at ratios between 1.9 and 2, only the primary ferrite solidification phase develops, but at room temperature some ferrite may remain in austenitic stainless steel [38]. Full austenitic stainless steel, such as type 310, is susceptible to hot cracking during welding, but the addition of a small amount of ferrite can reduce this effect [1]. This ferritic microstructure, which is known as delta (δ) ferrite or high temperature ferrite, is formed at high temperature and remains in solid solution on cooling. During the welding process, solidification of primary ferrites is recommended, because it reduces the hot cracking tendency compared to primary austenitic solidification [38]. This effect is a consequence of the narrow range of critical solidification temperatures and high solubility of impurities such as S and P in δ-ferrite compared to the austenitic phase [1]. The microstructure of austenitic stainless steel is an important factor which determines susceptibility to pitting and SCC. A number of microstructural features may be introduced during processing or welding. Figure 2-3 shows the features which are reported to play the most significant roles in pitting corrosion and SCC resistance [45]. 29

30 Figure 2-3: Microstructural features inducing pitting and SCC development [45] Effect of alloying elements The addition of alloying elements contributes differently to the microstructural stability, mechanical properties and corrosion resistance of austenitic stainless steel. For example, carbon, as an austenite stabiliser, enhances strength, but is considered an impurity which reduces ductility and intergranular corrosion resistance due to carbide formation at elevated temperatures. Sulphur has a negative impact on corrosion resistance, but improves alloy machining. Austenitic stability is increased by increasing the proportions of austenite-forming elements (N, C, Ni, Co, Cu and Mn) and reduced by the addition of elements which promote δ-ferrite formation (Al, V, Cr, Mo, Si and W). Table 2-3 summarises the main contributions of those elements to the properties of austenitic stainless steel. Developing alloying materials for specific applications takes into consideration the general optimum balance between them, which gives a desired property without significant drawbacks in regard to other properties. This can normally be achieved by choosing suitable processing and heat treatment conditions [46]. 30

31 Table 2-3: General effects of alloying elements on austenitic stainless steel [46] Element Chromium (Cr) Nickel (Ni) Carbon (C) Dominant characteristic Ferrite former Carbide former Austenite former Strong austenite former Influence in austenitic stainless steel Improves corrosion and scaling resistance, destabilises austenite and in high concentration forms brittle σ phase with iron Stabilises austenite and stress corrosion cracking peaks at content of 17% Causes weld decay unless stabilised and stabilises austenite Manganese (Mn) Austenite former Increases strength and stabilises austenite Nitrogen (N) Strong austenite former Increases strength and stabilises austenite Silicon (Si) Deoxidiser Improved scaling resistance Molybdenum (Mo) Niobium (Nb) Titanium (Ti) Ferrite former Carbide former Strong carbide former Strong carbide former (with aluminium) Improves corrosion resistance and strongly increases strength at high temperatures Stabilises against weld decay and increases strength at high temperature Stabilises against weld decay and increases strength at high temperature, especially by age hardening Impurity, but added to Reduces cleanliness and reduces ductility. At Sulphur (S) improve machinability about 0.3% improves machinability Phosphorus (P) Impurity Reduces ductility and cleanliness Improves corrosion resistance and can Copper (Cu) Normally impurity increase strength at high temperature Effect of temperature The microstructure of austenitic stainless steel is thermodynamically unstable, which leads to its transformation. The microstructural transformation and phase precipitation processes are related to the thermomechanical process and reaction kinetics, or to environmental conditions such as temperature and stress [41] Heat treatments Wrought austenitic stainless steel can be produced thermomechanically under two conditions: cold-worked and annealed. The cold-worked process is used to produce hard alloys, because austenitic stainless steel, unlike carbon steel, is not hardenable by heat treatment, but only by hard working or by the addition of alloying elements such as Ti and Nb. Annealing is a process normally performed above 1050 C after cold-working or welding in order to dissolve chromium carbide or undesirable phases. This heat treatment improves corrosion resistance by eliminating precipitated secondary phases, such as chromium-carbide at grain boundaries. The latter typically reduce the intergranular corrosion resistance and optimises mechanical properties. The microstructural stability of austenitic stainless steel during thermomechanical treatment depends on the 31

32 composition of alloying elements, temperature and deformation [41]. Beside the effect of alloying elements and composition discussed earlier, temperature and ageing time also affect stability range, phase transformation and the metallic and non-metallic phase precipitation of austenitic stainless steel microstructure. At a temperature range of C, the microstructure of austenitic stainless steel shows various precipitate phases, depending on composition and time. For example, carbide is normally formed at a temperature range of C at the grain boundary and intermetallic compounds such as laves phase Fe 2 Mo and the sigma phase. It is a brittle, hard phase which reduces the ductility and toughness of stainless steel. It normally forms at a temperature range of C and increases with ferrite stabilising elements Cr, Mo and Si, but reduces with austenite forming elements Ni, N and C [38]. The time-temperature-transformation diagram of austenitic stainless steel is shown in Figure 2-4, which presents the common precipitation phases in austenitic stainless steel as a function of temperature and time. The composition of precipitated phases is also related to alloying element composition, which promotes different types and degrees of phase composition [37]. For this reason, wrought austenitic stainless steel is heat treated after cold work at a typical temperature range of C, followed by rapid cooling to dissolve those phases and relieve stress [41]. 32

33 Figure 2-4: Austenitic stainless steel time-temperature-transformation diagram [47] Carbide precipitation Carbide precipitation affects both the mechanical properties and corrosion resistance of austenitic stainless steel, reducing both ductility and resistance to intergranular corrosion (IGC). Carbide formation generally occurs when austenitic stainless steel is exposed for a long time to a moderate service temperature or for a short time to welding temperature. This occurs when segregation of the alloying elements is induced by a chromium-depleted zone. The rate of carbide precipitation depends mainly on carbon content, temperature and time [37]. Sensitisation is the term generally used to indicate the sensitivity of a material s microstructure to IGC, which occurs in stainless steels when Cr-depleted zones are formed adjacent to the grain boundary [48]. Precipitation of carbide phases such as M 23 C 6 is strongly related to carbon concentration. In austenitic stainless steel, carbon is considered an impurity and has a solubility above 0.4% at solidification temperature, but solubility decreases with cooling. The general relation between temperature and solubility of carbon is shown by equation 2-3 [40]: Log (C ppm) = T (K) (2-3) 33

34 Austenitic stainless steel with low carbon content, such as 0.03% C at room temperature, is considered to be a super-saturated solid solution, but carbide is not formed because of low diffusion of both carbon and chromium. Figure 2-5 illustrates the effects on carbide precipitation of temperature and carbon concentration with time [40]. Figure 2-5: Influence of carbon concentration (C wt%) on sensitization time and temperature [40]. At high temperatures, the diffusion rates of Cr and C are higher at the grain boundary than within the grain interior. The crystallographic misalignment at grain boundaries introduces higher concentrations of defects, which typically result in higher diffusion rate. This causes the carbides to precipitate at grain boundaries, eventually leading to the depletion of chromium at the grain boundary, which in turn reduces the concentration of chromium at the grain boundary below the minimum protective concentration value (11% Cr). This phenomenon, known as sensitisation, can leads to IGC or intergranular corrosion stress corrosion cracking (IGSCC). The depleted zone at the grain boundary is shown in Figure 2-6. In addition, the austenite is unstable in this zone, making it prone to martensite transformation. Chromium depletion can also be caused by oxide, nitride and sulphide formation and by alpha and chi phase precipitation (Figure 2-3). The rate of carbide precipitation also depends on the type of grain boundary. Figure 2-7 shows that coherent grain boundaries are less sensitisation resistant than non-coherent ones. This difference depends on temperature and exposure time. Increasing the temperature is associated with the reduction of sensitisation time, which is dependent on the type of grain boundary. For instance, non-coherent grain boundaries far less resistance to sensitisation compared to coherent grain boundary, 34

35 such as coherent twins. This is attributed to the lower grain boundary surface area/energy due to the geometric fit [40, 48]. Figure 2-6: Sketch and Cr profile showing chromium depleted zone adjusted to grain boundary [48] Figure 2-7: Types of grain boundary resistance to sensitisation as a function of temperature and time [40] Effect of deformation Plastic deformation in stainless steel is related to dislocation movement through a preferential slip system (close packed planes and direction). There are 12 slip systems in 35

36 an austenitic FCC structure related to [111] planes and <001> directions [38]. Deformation induces martensitic transformation of metastable stainless steel, which is also a non-diffusional phase transformation and leads to deformation-induced martensite [49]. Epsilon martensite (ε) is the dominant type, with low stacking fault energy. This deformation occurs when the atom arrays within the FCC structure (ABCABC) change to denser [111] planes by forming an ABCA/CBA arrangement. The slip process introduces ε-martensite (paramagnetic), which is known as a stacking fault hexagonal close-packed (HCP) structure [1]. Unstable austenite alloys can be transformed into hard α'-martensite, which has a BCC (ferromagnetic) structure at room temperature under tensile transformation. This deformation occurs by the non-diffusion shear mechanism, increasing strain hardening and ductility by shifting the necking point [41]. 2.2 Corrosion of stainless steel Corrosion is the main threat to the integrity of stainless steel materials during service in environments that are contaminated, especially with chloride ions. It is defined as deterioration under corrosive environmental attack. The process begins with a chemical or electrochemical reaction on the surface of the alloy. This reaction is generally dependent on the potential, the ph or both. It is generally classified in terms of environmental conditions as either dry or wet (aqueous) and either high or low temperature corrosion [2]. Corrosion is also classified as general or localised, depending on how it attacks the material. General corrosion is characterised by a uniform attack of the exposed surface, associated with uniform metal dissolution, such as that of stainless steel in acid or a hot caustic environment. In localised attack, by contrast, the dissolution of stainless steel in a chloride environment is limited to small sites [50]. Such localised attack is more detrimental to the integrity of stainless steel and usually involves one or more of pitting corrosion, crevice corrosion, IGC and SCC Electrochemical corrosion In aqueous environments, electrochemical corrosion develops through a cell reaction which requires four components to proceed: an anode, a cathode, an electrolyte, through which ions move, and a connector path for electron transfer. Each of these is essential to 36

37 the corrosion process, although the anodic and cathodic reactions may take place at the same electrode. In an anodic reaction, the dissolution of metal (M) is based on the general reaction shown in equation 2-4 [2]: M M n+ + n e (2-4) In a cathodic reaction, depending on the nature of the environment, reduction can take any of the following three forms [2]: 1) Oxygen reduction, in neutral and basic solutions: O 2 + 2H 2 O + 4e 4OH (2-5) 2) Hydrogen and oxygen reduction, in an acid solution: 2H + + 2e H 2 (2-6) O 2 + 4H + + 4e 2H 2 O (2-7) 3) Metal ion reduction: M n+ + e M (n 1)+ (2-8) Corrosion thermodynamics Thermodynamically, the corrosion process (reaction) is described as the transition from a relatively unstable high energy state to a more stable low energy state. This reaction is associated with an energy change (chemical potential) in the system known as Gibbs free energy. Free energy changes indicate the nature of the electrode reaction, i.e. a spontaneous reaction is expected when the energy change is negative, while a nonspontaneous reaction requires external energy to proceed. The Gibbs free energy of reaction 2-9, with reactants A and B and products C and D, can be estimated using equation 2-10 [51]: aa + bb cc + dd (2-9) G = G o + RTln ( a Product a Reactant ) (2-10) where G o is the standard-state free energy, R is the ideal gas constant (8.134 J.mol -1.K -1 ), T is temperature (Kelvin) and a is activity of the reactant/product, or concentration if the solution is ideal or dilute. 37

38 At equilibrium in reaction 2-9, the change of free energy reaches zero ( G = 0). G o is related to the reaction constant (K eq ) in equation 2-11 at standard state, where the activity of each component is equal to one: = a c d C.aD a Reactant K eq = a Product a A a.a B b (2-11) The relation between G o o and standard cell potential E cell or corrosion potential E cor are given by the equations: G o o = RTlnK eq = nfe cell (2-12) o E cell o = E cathode o E anode (2-13) o o where E cathode is the standard half-cell reduction potential, E anode is the standard half-cell oxidation potential, n is the electron valance number and F is the Faraday constant, representing the charge on one mole of electrons (96485 C.mol -1 ). The deviation of corrosion cell potential (E cell ) from standard cell potential (E o cell ) due to polarisation (activation reaction) is governed by the Nernst equation [51]: o E cell = E cell Kinetics of electrochemical corrosion ( RT nf ) log a oxid a red (2-14) The rate of a corrosion reaction is first controlled by the electrical double layer which builds up at the interface between electrode and electrolyte through fast oxidation when the material is immersed in a solution. At open circuit potential (OCP), the corrosion potential (E cor ) and corrosion current density (i corr ) represent the mixed potential of anodic and cathodic electrode reactions where the exchange current density values of anodic and cathodic reactions are equal. Polarisation can be performed using a Potentiostat, which induces current flow. It is known as anodic polarisation or anodic overpotential (η anod ) if the potential is shifted above E cor and cathodic polarisation or cathodic overpotential (η cathod ) if it is below that value. The overpotential equation is shown as equation 2-15 [51]: η anod/cathod = E anod/cathod E cor (2-15) 38

39 As a result of polarisation, the corrosion reaction is activated and current flows from the high potential electrode (cathode) to the lower potential anode. This technique allows the measurement of the kinetics of corrosion rate based on the current density (charge transfer) at the interface, which equals the contributions of both anodic and cathodic reactions. The total current density flow caused by applying electrochemical polarisation can be estimated from the Tafel equations for the cathodic and anodic reactions, using the Butler-Volmer equation (equations 2-16 and 2-17). This combines the anodic and cathodic electrochemical corrosion reactions when the exchange current density i o of each reaction is the same i corr. i app = i a + i c = i corr [exp ( 2.3 (E E cor) ) exp ( 2.3 (E E cor) )] (2-16) β a β c β a = 2.303RT αnf & β c = 2.303RT (1 α)nf (2-17) where i app is the applied current density, i a is the anodic current, i c is the cathodic current, β a and β c are the Tafel coefficients for anodic and cathodic reactions respectively and α is the anodic transfer coefficient ( or symmetry coefficient), usually taken to be 0.5 for a single step reaction Corrosion kinetics control Under electrochemical polarisation, the kinetics of the electrochemical reaction are controlled by three processes of overpotential: electrode activation (charge-transfer process) control, solution resistance and concentration (mass-transport process) control. Therefore, the total overpotential (η Total ) in equation 2-18, which controls the corrosion reaction, is a combination of these factors. η Total = η activation + η resistance + η concentration (2-18) For activation polarisation (η activation ), the anodic (oxidation) and cathodic (reduction) reactions control the charge transfer through the electrode/solution interface. Therefore, the overpotential of both half-cell reactions (anode and cathode) depends on the electrode material and the reduction environment. The activation overpotential under anodic current or cathodic current is presented in equation 2-19 [51]: η activation = η a/c = β a/c log i a/c i o (2-19) 39

40 However, at high anodic polarisation, the effect of cathodic activation is generally neglected, because anodic current i a is larger than cathodic current i c. The resistance of the solution (R) to current flow can introduce overpotential resistance, especially with a non-conductive solution. This resistance is manifested by a potential drop (IR) when current (I) passes through the solution, which obeys Ohm s law : η resistance = IR (2-20) In localised attack such as pitting corrosion (section 2.3), the solution resistance is also a function of pit depth and increases with increasing metal dissolution, due to the potential drop inside the pit relative to the bulk solution resistance (R b ), which is almost constant at the passive metal surface. Hence, with increasing pit depth, the pit solution resistance (R p ) increases linearly [52]. It is therefore suggested that the electrolyte resistance (R) in a 1D pit is a function of the specific resistance (resistivity) of the solution inside the pit ρ 1 and outside the pit ρ 2 and of system geometry, as in equation 2-21: R = R p + R b = ρ 1 ( d πr 2) + ρ 2 ( 1 4r ) (2-21) where d is total pit depth and r in pencil-like pits is the wire radius; the value of ρ 2 was estimated at ohm.cm for 1 mol.l -1 of NaCl solution at 25 C [52, 53]. In case of concentration control, the reaction depends on the concentration of species, so three mechanisms of mass transport control the reaction; these are diffusion, convection and migration. Among these, the main effects come from diffusion and convection, which control the transfer of reactant species from the bulk solution to the electrode surface or pit bottom, and of product species in reverse. At the interface between electrode and solution, the region near the electrode is controlled by diffusion and the region of the bulk electrolyte is governed by convection. In a reaction with a high rate of reduction, the depletion of ions such as hydrogen or oxygen at the electrode surface may reach a maximum value, limiting the reaction rate. In this case, the limiting reaction rate is known as the limiting diffusion current density (i L ). At this current density, the rate of reaction comes under mass transport (diffusion) control, while at lower current densities, the reaction rate is governed by both mass 40

41 transport and charge transfer [51]. The concentration overpotential under a limiting diffusion current density is presented in equation 2-22: η concentration = η c = 2.3RT log (1 i ) (2-22) n i L where R is the gas constant (8.314 J.K -1.mol -1 ), T is absolute temperature, n is the number of electrons involved in the reaction, i is the reaction current density and i L is the limiting diffusion current density. The relationship between the limiting current density of the cathodic reaction under concentration control (diffusion) and species concentration is given in equation 2-23 [2]: i L = DnFC b δ d (2-23) where D is the diffusion coefficient of the reducing agent, C b is the concentration of reacting ions in the bulk, δ d is the thickness of the diffusion layer, n is the number of electrons involved in the reaction and F is the Faraday constant. For the anodic reaction, no limiting current density applies because of a shortage of metal atoms or metal dissolution, but at high dissolution rates the saturation concentration of dissolved product leads to the precipitation of a salt film on the electrode surface, especially at the pit bottom (section ). At this stage, current density is limited and ohmic drop occurs as a consequence of this salt film formation. The transport of corrosion products out of the pit can also be influenced by pit geometry, while the stability of the salt film depends on the potential which leads to the dissolution rate of metal ions being higher than the diffusion rate [51]. Therefore, the limiting dissolution current during pit growth under conditions of salt film precipitation is given by the modified equation i L = DnF C s C b d+δ d (2-24) where d is pit depth, δ d is the thickness of the diffusion layer, C s and C b are the concentration of metal ions at the metal surface and in bulk. In contrast, the above three regions of overpotential induce electrochemical control of corrosion reactions, as demonstrated for pitting corrosion kinetics under anodic polarisation in Figure 2-8. First, there is an activation-controlled region (I), where the 41

42 current is very low, then a region (II) of potential drop (IR) or solution resistance, which controls the region where current increases exponentially with increasing potential (potential dependent). Finally, the diffusion-controlled region (III) is where a salt layer is precipitated and the current is both limited and independent of potential [52]. Figure 2-8: Overpotential regions of electrochemical control reactions under polarisation [52]. 2.3 Pitting corrosion of stainless steel Pitting corrosion in stainless steel occurs by localised attack on a small area, starting with a breakdown of the passive oxide film, leaving a small area exposed to the environment. This pit then acts as an anode compared to the intact surface, leading to perforation of the metal surface, as shown in Figure 2-9. The larger the potential difference between anode and cathode, the more rapid is the pitting process. Growing pits may then act as stress concentrating features where SCC can develop under certain conditions. Pitting attacks cause undetectably small weight loss and induce unexpected failure of structures such as tubes. While general corrosion is generally a slow process, the penetration rate of pitting can reach 100 times that of uniform corrosion [3]. Therefore, in order to measure and characterise resistance to pitting corrosion in a controlled environment, the corrosion process has to be accelerated, by widely used electrochemical techniques such as potentiodynamic, potentiostatic and galvanostatic methods [54]. 42

43 Figure 2-9: Pitting corrosion process [55]. The pitting resistance equivalent number (PREN) is often used to rank the pitting resistance of stainless steel, based on its composition of elements. For example, Mo and N alloys with PREN values above 32 have good pitting resistance in seawater environments. The general equation which is commonly used to determine PREN is 2-25 [1]: PREN (%) = %Cr %Mo + 16 %N (2-25) Characteristics of pitting corrosion The general behaviour of active-passive materials such as stainless steel under potentiodynamic polarisation is shown in Figure It consists of active, passive and transpassive regions characterised by the commonly measured parameters of pitting kinetics. Pitting corrosion or potential (E p or E pit ) and repassivation potential (E rp ) are often used to evaluate pitting corrosion. It is generally accepted that materials with higher values of E p and E rp will have superior corrosion resistance in similar test environments. In neutral chloride solutions, the anodic polarisation curve of stainless steel generally increases without the active loop characteristic of acidic chloride solutions [12, 56]. 43

44 T P A Figure 2-10: Schematic diagram of polarisation curve (forward and backward) of active-passive metal [57]. Pitting potential (E p ) is defined by a sudden and continuous increase in current density caused by the breakdown of the passive film under anodic polarisation. This potential is used to characterise the resistance of an alloy to pitting corrosion or to define the influence of various environmental conditions on this resistance. Pitting potential has also been found to depend on scan rate and a positive semi-logarithmic relation was reported between the pitting potential and the concentration of aggressive ions such as chloride. This relation also depends on the ratio of corrosive species to non-corrosive ones which can act as inhibitors. Chloride ions make the pitting potential of type 304 steel more negative, whereas SO 4 --, ClO 4 -, OH - and NO 3 - have the opposite effect. This inhibitive effect on pitting corrosion is quantified by equation 2-26 [3, 12, 58]: E p = A B log Cl C inh (2-26) where A and B are constants, and C inh is inhibitor concentration. The repassivation potential (E rp ) can be obtained after the pitting potential from the reverse scan of the anodic polarisation curve; it is defined as the intersection with the forward polarisation curve in the passive current region (Figure 2-10). This potential is characterised by no metastable events or stable pit growth occurring below it; therefore, it is known as the protective potential. Furthermore, E rp depends on the scan rate and pit depth or the value of current density before conducting the reverse scan [12, 56]. 44

45 Induction time (t in ), which is often used to characterise pitting corrosion, measures the time taken for a stable pit to form in the exposed alloy. At constant polarised potential, t in is reduced by increasing the concentration of aggressive species such as chloride ions. It is also dependent on potential, temperature and oxide film thickness. A linear relation between (t in ) -1 and concentration of chloride ions was reported, as shown in equation 2-27: where m and K are constants [12, 59]. 1 t in = K[Cl ] m (2-27) Passive film on stainless steel On the surface of stainless steel exposed to air, a thin layer of Cr 2 O 3, known as a passive film, forms spontaneously. This adherent oxide layer, several nm thick, creates a barrier between the metal surface and the environment, thus reducing corrosion significantly [11]. At steady state, the point defect model (PDM) is used to simulate passive film development in aqueous environments. It assumes that the passive film comprises two layers: a primary rigid oxide layer from which cations are transmitted into the upper layer, formed by the precipitation and hydrolysis of cations [60]. The nature of the passive film and its corrosion resistance in an aqueous environment were found to be related to alloy composition, environmental conditions, exposure time and potential [4, 61]. When stainless steel is anodically polarised, a protective passive film, which is a poor current conductor, is formed and grows linearly [62]. This process is independent of potential and characterised by passive current density. The protectiveness of the passive film may be indicated by a low passive film current with a wide range of stability potential or time [61]. Conversely, the stability of the protective passive film is reduced as chloride ion concentration increases [63]. Isaacs [54] reports that in chloride solution the dissolution kinetics of passive films on stainless steel and nickel-base alloys depend on the active iron and nickel respectively. Devasenapathi and Asawa [64] studied the effects of applied potential on the passive film of Mn austenitic stainless steel (9 wt. %Mn) and type 304 in 1 M HCl, using X-ray photoelectron spectroscopy (XPS). They found that the nature of the film changed with potential: a thick black film formed at OCP, but under anodic polarisation general dissolution occurred, forming a rough 45

46 surface. High Fe content relative to Cr was also reported in the film by increasing potential relative to OCP. High-strain, cold deformation (40-50%) was found to enhance localised corrosion by inducing martensite formation and residual stress, which increased the number of anodic activity points [65]. The effects of tensile strain (at 0-24%) on passive film properties were studied for type 304 in a borate solution with and without 5000 ppm chloride ions. It was found that percentage of α'-martensite increased with strain. Increasing the strain reduced the resistance of the passive film in chloride solution, which was attributed to the reduction in the thickness of the passive film [66]. To conclude, the electrochemical stability of a passive oxide film can be related to its chemical and electronic properties. Thus, it has been proposed that anodic passivation behaves like a semiconductor. Breakdown of the passive film may be caused by aggressive anions, in the case of a thin film, or an electron avalanche in a thick film [67]. Therefore, a small area of the passive film is normally attacked and this accelerates pitting corrosion. Moreover, various studies of stainless steel show that passive film stability is reduced with increasing potential above defects, around or above inclusions and in the second phase relative to the matrix. The stability of the passive film can be affected by the difference in Cr/Fe ratio that occurs between the alloy matrix on one hand and metallic inclusions, alloy oxide and the second phase on the other. The breakdown of the passive film is considered the first stage of pitting initiation, which generally leads to metastable pit growth or propagation to a stable pit [3]. These steps are discussed in the following subsections Pit initiation When stainless steel is exposed to a corrosive environment, the breakdown of passivity causes pit initiation, but this does not often lead to pit propagation, as many events are quickly passivated. Detection of such low-value events requires sensitive current measurement to differentiate transient current from background noise, because the pit current events are very small and brief. It has been reported that pit nucleation is an unstable process which occurs as a result of microscopically violent events. A transient current typical of such nucleation is shown in Figure 2-11, where the nucleation of an unstable pit is indicated by a short, sharp current peak followed by sudden drop current without pit growth. This minor peak can lead to the dissolution of a volume of metal 46

47 below 0.01 μm 3. However, pit nucleation may also be followed by pit growth, to form either a metastable or a stable pit [68, 69]. Figure 2-11: Typical transient current in a pit nucleation event (unstable pit) on a type 304L stainless steel microelectrode [68]. The main mechanisms of film breakdown and pit initiation discussed in the literature are penetration, mechanical breakdown and adsorption of the passive film [3, 12, 70]. Figure 2-12 shows sketch diagrams of these mechanisms. The penetration mechanism (Figure 2-12a) is based on the movement of aggressive anions through the passive film to the interface between the metal and the oxide film. This movement, which leads to metal dissolution, is governed by the film potential and induction time, but the breakdown of the passive film and pit initiation depend on critical chloride concentration at the interface between metal and oxide [3, 70]. The breakdown of the film (Figure 2-12b) suggests that mechanical stress at weak places leads to local breakdown of the oxide film [3, 12]. The adsorption theory (Figure 2-12c) suggests competitive adsorption between chloride ions and oxygen atoms in the passive film, but chloride ions may reduce film thickness by enhancing the movement of cations from the oxide to the electrolyte. The thinning of the oxide film then leads to its breakdown under the effect of a local electric field [12]. 47

48 (a) (b) (c) Figure 2-12: Passive film breakdown mechanisms: (a) penetration, (b) film breakdown, (c) adsorption [70]. Szklarska-Smialowska [71] proposes that the passive film undergoes electrical breakdown with increasing potential and suggests that chloride ions support this process by injecting electrons into the semiconducting passive film and reducing the electrical breakdown potential. In contrast, the mechanism of film breakdown may involve a continuous breakdown and repassivation process, but at certain sites such as regions above inclusions, flaws or under mechanical stress, the surface film becomes weak and may break easily in the presence of corrosive ions without repassivation [3, 54]. 48

49 2.3.4 Metastable pit growth Metastable pit growth has limited life and induces insignificant deterioration of stainless steel integrity, but it is widely investigated because each current event is limited to one metastable pit. Moreover, it can be considered a precursor state of stable pit growth with a similar control mechanism [54, 69, 72-74], but the initiation and growth of metastable pits occurs below the pitting potential before it shows sudden repassivation [72]. A linear relationship was found between metastable and stable pitting potential with different materials, suggesting that metastable pitting potential could be used as an early indication to evaluate the pitting susceptibility of materials prior to damage or leaks which may be caused by stable pitting [75]. Metastable pits can be detected in chloride solution by monitoring the OCP for any sudden reduction in potential, marking transition to the cathodic direction [73]. Under the anodic polarisation of stainless steel, sudden current fluctuation above the passive current with increasing potential indicates metal dissolution and metastable pit growth. The typical transient current of metastable pit nucleation and growth is shown in Figure The current can be seen to rise slowly with pit growth for a few seconds, then a small sharp increase is followed by a very sudden reduction to passive current, indicating complete repassivation (pit death) [68]. Figure 2-13: Transient current of metastable pit nucleation and growth on type 304L stainless steel [68]. Most metastable pits appearing at OCP have a lifetime of less than five seconds. This is believed to be due to the sudden dilution of the corrosive environment after loss of the 49

50 perforated cover above the pit. Below the pitting potential, increasing the polarisation potential generally leads to larger transient currents and longer time events [73, 74]. Longer lifetimes up to 30 s [76] and larger metastable pit diameters estimated for hemispherical pit at 30 µm were reported on 904L stainless steel in 1M NaCl below the critical pitting temperature (CPT) [77]. Metastable pit growth associated with metal dissolution was detected by imaging metastable pits at OCP using scanning electrochemical microscopy and it was found that an anodic transient was associated with the oxidation of dissolved Fe 2+ [73]. At an early stage of pit growth, it is generally agreed that metastable pit growth is supported by the perforated cover, which serves to retain the aggressive environment inside the pit [72, 74, 78]. Frankel et al. [74] believe that metastable pit growth is mainly under ohmic control, whereby the porous cover provides a resistive layer, since the drop in potential in small pits is insignificant [79]. Alternatively, Pistorius and Burstein [72] proposed a diffusion controlled model of metastable pit growth, where the pit cover acts as a barrier to the diffusion of dissolved metal cations. Nevertheless, all agree that losing the cover of a growing pit is the main reason for the sudden repassivation of a metastable pit. This indicates that the cover sustains metastable pit growth by providing a barrier that maintains corrosive pit chemistry before sudden passivation. Therefore, it is suggested that diffusion control of pit growth is related to both the pit cover and surface geometry [69, 72]. It is believed that loss of the pit cover occurs prior to a critical concentration of dissolved metal ions which maintains the low ph necessary for continuous dissolution [80]. Laycock [29] suggests that many metastable pits grow with a salt film before the metal cover is lost, whereas Frankel [74] believes that precipitation of a salt film before the metal cover ruptures will lead to stable pit growth. In another study, Laycock et al. [78] established a difference between salt films precipitated before and after CPT. In the case of metastable pits, the salt film was found to be an intermediary in oxide passivation similar to the film of iron oxide in H 2 SO 4 solution. When the current density of metastable pit growth on stainless steel and iron was calculated, assuming the pit shape to be hemispherical, current density over time was approximately constant [74] or fluctuated around a constant mean value until the pit was passivated [72]. Therefore, metastable pit growth is independent of potential but limited 50

51 by the existence of the metal cover [69]. Laycock and Newman [29] measured the current density of metastable pits in stainless steel between 0.5 and 2.5 A.cm -2. However, Tian et al. [81] found that metastable pits grown on type 304 stainless steel in 3.5% NaCl changed from cone shaped to dish shaped and that current density changed as pit potential increased, suggesting an ohmic potential drop, while pit growth was influenced by diffusion control. In addition, Frankel [74] suggests that the cover of a metastable pit may rupture under high current density due to osmotic pressure above the cover. Moayed and Newman [76] investigated the effect of temperature on the transient current of metastable pits using 904L below CPT. They found that current followed a power relation (I~t n ), where n increased with temperature from 0.5 to 1.5. Pistorius and Burstein [72] proposed that before losing their metal covers, metastable pits did not reach stability because their stability product (i.r) was below 0.3 A.m -1. This value was suggested for 304SS based on the 75% concentration of dissolved metal cations which was required to maintain ph at active dissolution. The pit was assumed to be open and hemispherical, while the saturated metal salt inside the pit was considered to be 4 M FeCl 2. Park et al. [82] studied pitting corrosion in high chloride solutions and found that above 5 M chloride ion concentration and 90 C, metastable pits did not typically form; instead, stable pits formed directly. The effects of surface roughness on metastable pits were investigated using 304 stainless steel in chloride solution and it was found that a rougher surface promoted metastable pit formation, but narrow and deep sites activated metastable pits faster at lower potential in a similar chloride environment [72]. Similarly Zuo et al. [83] studied the influence of surface roughness using 316LSS in 0.01 M NaCl solution, reporting that the rate of metastable pit events and current peak values were reduced by reducing surface roughness, but average growth rate indicated a slight increase. Burstein et al. [68, 84] define exhausted sites as those where metastable pit growth has re-nucleated many times and growth kinetics are of the first order. These sites depend on surface roughness and are gradually annihilated. In other work, metastable pitting was found to initiate around sulphide inclusions at lower potential [85]. The lifetime of metastable pits was shown to be related to the size of inclusions, but after surface treatment using laser melting, the frequency and lifetime 51

52 of metastable pits declined [86]. Park and Böhni [87] measured ph values associated with the dissolution of shallow MnS inclusions in 304 stainless steel in 0.1 M NaCl and report that when metastable events occurred, ph fluctuated between 2 and 4.2, with a potential range of mv vs SCE. The reduction in ph started after one second when the current increased, but it returned to the bulk ph value after the pitting events have ceased Stable pit growth It is generally agreed that stable pit growth occurs above the pitting potential or CPT, where a high rate of active anodic dissolution follows the breakdown of the passive film, while the entire metal surface is passive. Unlike metastable pits, stable pit growth shows no termination behaviour and when a stable pit is formed, it can be propagated below the pitting potential but not below the repassivation potential (E rp ). However, the early growth behaviour of a stable pit is similar to metastable pit growth, but without passivation [56]. Generally, a pit is considered to act as an anode relative to the passive metal surface; however, part of the pit may turn passive despite the presence of corrosive solution inside it [88] and about 5% of anodic current was suggested to be consumed inside the pit [72]. In earlier work, Hoar [89] proposed low ph inside the pit as a main reason for pit growth and active metal dissolution. Pickering and Frankenthal [90] then investigated the influence of the potential drop on active metal dissolution at the bottom of the pit relative to applied potential, while the hydrolysis reaction of metal dissolution induces a drop in ph. A large potential difference inside the pit of 1.6 V SHE was reported in an Fe- 20Cr sample between the applied potential of 1.4 V SHE and the measurement at the pit bottom of -0.2 V SHE at ph 1. This unexpectedly large drop in potential was explained by gas bubbles which restricted current flow at the pit bottom. At around the same time, Isaacs [91] studied the influence of a salt film (resistive layer) formed inside the pit on the kinetics of metal dissolution and concluded that the thickness of the salt layer was a function of the applied potential and the rate of diffusion of dissolved metal from the salt layer/solution interface. This layer, containing chloride ions and metal cations, introduces ohmic resistance and a thin layer of 100 Å was found to account for 95% of the voltage drop, with a current resistivity of 52

53 10 8 ohm.cm. The potential drop through the layer was assumed to be independent of the current flow, due to self-regulation of the salt layer thickness. Galvele [80] developed a one-dimensional model (Figure 2-14) to study the mechanism of active pit growth, based on diffusion control of mass transport inside the pit, while migration and convection mass transport were neglected. Anodic dissolution reaction was assumed to occur in the pit bottom while the pit wall was passive. For dissolved metals, the simple hydrolysis reaction in (2-28) was considered to apply, while complex hydrolysis reactions were ignored. 2M n+ + H 2 O + OH 2M(OH) (n 1)+ + H + (2-28) Pit depth Figure 2-14: Galvele s one-dimensional model of pit dissolution [80]. Galvele then used Fick s first law to calculate the flux of species at steady state between the diffusion boundary of the pit mouth and the pit bottom. This gave a critical value (x.i), pit depth (x) multiplied by current density (i) to determine the composition of cations and hydroxide, which developed through active metal dissolution based on the critical ph. Figure 2-15 shows plot of iron hydroxide concentration against (x.i) In the plot, hydrogen concentration marked with (+) indicates the critical ph concentration of equilibrium; above this concentration, which also defines the critical value of (x.i), dissolved cations start to be active. From the plot in Figure 2-15, it is possible to 53

54 determine the composition of metal cations and hydroxide if the pit depth or current density is known. Figure 2-15: The relation of x.i to H +, cation concentration and hydroxide of Fe and Fe(OH + ) [80] Solution chemistry inside the pit Wilde and Williams [92] measured the ph of dissolved product inside an artificial pit grown on type 304 stainless steel wire in 1 M NaCl at 25 C after dissolving 0.32 cm of the wire under applied potential between +500 and -200 mv vs SCE. Liquid nitrogen was then used to freeze the corrosion products and a glass microelectrode was used to measure the ph, which was found to decrease from 3.6 to 0. The concentrations of Fe(II) and Fe(III) ions were measured, suggesting that this ph change was associated with increasing the Fe(II)/Fe(III) ratio from 1.59 to 9.8 over time. The low ph was attributed to hydrolysis of dissolved ions, which is considered the main cause of active metal dissolution, while increasing chloride migration inside the pit was suggested to have accelerated pit growth. Figure 2-16 shows the ph value and its relation to the ratio of ferrous to ferric ions over time for metal dissolution occurring under different applied potentials. 54

55 Figure 2-16: Changes to solution chemistry inside an artificial pit with time [92]. Suzuki et al. [93] measured the composition of solution in artificial pits in austenitic stainless steel types 304L, 316L and 18Cr-16Ni-5Mo, exposed to 0.5 N NaCl electrolyte at 70 C. About 200 μl of analyte was taken from the pits by pipette and analysed by an atomic absorption spectrochemical technique. They report ph values of 0.6 to 0.8 for 304L, for 316L and to 0.8 for 18Cr-16Ni-5Mo, with chloride concentrations of 6.47 to 3.78 N for active pits. Low ph was attributed mainly to the hydrolysis of chromium ions and molybdenum ions, while the formation of hydroxylchloro complexes of dissolved metal ions and the presence of high chloride concentration were suggested as explaining the lower ph value which was unexpected from the thermodynamic calculation. Mankowski and Szklarska-Smialowska [94] also suggest that low ph inside the pit is a consequence of high chloride ion concentration. Peterson and Linnox [95] report a ph of 1.2 to 2 inside active crevices on 304L stainless steel in seawater. This value of ph was simulated using acidified 0.6 M NaCl solution, indicating that a shift in ph to an alkaline solution through cathodic protection potential is affected by crevice geometry, protection potential and time. Park and Böhni [87] studied the dissolution of MnS inclusions on 304 stainless steel in 0.1 M NaCl using a microchemical cell and report that the ph value of metastable pits at the interface of MnS inclusions fluctuated between 2 and 4, but for stable pits the ph value was close to 2. 55

56 2.3.7 Solution concentration inside the pit Sato [96] proposed that there is a critical concentration of metal salt ( C ) such that a pit can grow if the concentration of metal ions inside it is higher than this ( C > C ), while below C the pit would turn passive. This concentration can be estimated from the build-up concentration using an equation developed earlier for the mass transfer of a hemispherical open pit at steady state [97]. At constant potential C 2 Kmol.m -3 (2 M) was reported for stainless steel. Another result supported the suggestion that pit stability is related to ion concentration, but there was disagreement about which ions make the main contribution to pit stability [96]. Later, FeCl 2 salt was identified as an important factor in stable pit propagation [11]. Kuo and Landolt [98] found that the saturation concentration of FeCl 2 (4.25 M) was related to limited current at the iron electrode surface and was controlled mainly by diffusion of Fe 2+ away from the anode surface, where the effective diffusion coefficient was cm 2.s -1. Mankowski and Szklarska-Smialowska [94] report that increasing the concentration of Cl - ions inside the pit with time leads to slow growth and indicate a concentration of 12 M Cl - as a maximum value, but as pit volume increases, the concentration of chloride ions is reduced. At room temperature, the metal ion concentration below 70% was suggested to limit pit propagation in type 304 stainless steel [99]. In 1986, Gaudet et al. [52] investigated the dependence of the concentration of dissolved products on the anodic dissolution rate (current density) using 1D artificial pits on 304 stainless steel, conducted at different applied potentials. Various pit depths were obtained at +400 mv under limiting current dissolution, then the potential was stepped down to -100 mv vs SCE while the current reduced over time was recorded (Figure 2-17a). A numerical equation derived from a simple pit diffusion model was then used to determine surface concentration. Figure 2-17(b) shows mean curves of current density vs surface concentration for various applied potentials after IR correction for each pit depth. Error bars represent the measurement error for current density difference after IR correction of each pit depth. The results show that the salt concentration of active dissolution (film free) is around 30-80% of the saturation concentration and that below this concentration passivation occurs, while above it, a salt film is formed and the diffusion rate controls pit growth. 56

57 Figure 2-17: (a) Time of pit passivation after potential step-down and (b) current density over surface concentration [52]. The chemical composition inside the pit was related to the kinetics of pit dissolution. Inside the pit, a build-up of dissolved metals during pit growth affects the rate of metal dissolution. Mankowski and Szklarska-Smialowska [94] studied the effects on pit kinetics of various concentrations of FeCl 2 added to 1 N HCl solution. Figure 2-18 shows the resultant behaviour of polarisation curves and indicates that increasing the concentration of FeCl 2 above 3.5 N limits the current density, due to salt film precipitation, while at lower concentration, dissolution is reactivated. Figure 2-18: Effect of various amount of FeCl 2 on anodic polarisation curves of stainless steel in 1 N HCl [94]. 57

58 Hakkarainen [100] investigated the relation of solution concentration to dissolution kinetics by simulating the conditions of a growing pit, using chemically dissolved solutions of 304 and 316 stainless steel in 10 M HCl as electrolytes to polarise new samples. The composition of each solution was analysed by atomic adsorption spectrometry before new samples were polarised anodically up to saturation of metal chloride. The results in Figure 2-19 indicate that increasing solution concentration reduces saturation conditions of polarisation potential and current density. About 80% of metal ion concentration inside the pit was considered important for stable pit growth under diffusion control. Moreover, it was found that the Mo content of 316 stainless steel had no influence on dissolution rates. Isaacs et al. [101] measured the composition of pit solutions using X-ray fluorescence analysis (XRF) and found that they contained concentrations of about 11 M chloride ions and 5 M metal ions, consisting of 3.5 M Fe, 1.1 M Cr and 0.5 M Ni. The highest effective diffusion coefficient is reported for nickel ions and the lowest for chromium. Figure 2-19: Effect of solution concentration of pre-dissolved 304 and 316 stainless steel on the anodic polarisation curve of new samples [100] Stable pit dissolution Isaacs [91] studied resistive layer in localised corrosion of stainless steels indicate that current density increased to the maximum value before declining with increasing time 58

59 of pit growth. At the region in which current density decreased with time, it varied with t -0.5 and a linear relation was obtained by plotting (1/i 2 ) over time. It was shown that the product of diffusivity (diffusion coefficient and concentration gradient) is related to (1/Slope). In this region, the solution inside the pit cavity became dark. The current behaviour suggests that it depends on the time, potential and number of pits. The results show that the time to maximum current (initiation period) reduced with increasing potential, while the peak value of current increased. In another study, Tester and Isaacs [102] simulated localised corrosion solutions of Ni and stainless steel using various compositions of metal salts at concentrations from 0.5 to 10 M. These were used as bulk solutions to study current-time behaviour above pitting potential (0.5-1 V vs SCE). The resulting schematic of pit dissolution over time (Figure 2-20) shows a clear division into a transient region and a quasi steady-state region. In the transient period, current density increased to a limiting value while active polarisation controlled the dissolution of metal ions and the hydration reaction. During the quasi steady-state period, precipitation of a resistive chloride layer following saturation or supersaturation caused current density to fall gradually from the maximum value. Therefore, diffusion control took place between the metal/solution interface and the bulk solution. The ohmic resistance of the metal salt layer depends on its thickness, which increases with increasing current dissolution. Figure 2-20: Schematic of anodic dissolution showing two periods of current-time behaviour[102]. 59

60 A model of the quasi-steady state region was developed by assuming a steady state in the solution between current dissolution and mass transport through the salt layer. In solution, diffusion flux was taken into account, while migration and convection mass transport were neglected. The Nernst-Einstein equation of mass transport (2-29) was used, by considering diffusion flux under concentration and potential gradient as driving force in the solution. i nf = D [ C x nf Φ + C D ] (2-29) RT x where D is the diffusion coefficient of species, C is the concentration of species ions, x is the diffusion layer thickness, Ф is the potential drop in the solution relative to the reference electrode, i is the anodic current density, n is the number of electrons involved in the reaction, R is the gas constant, T is the temperature and F is the Faraday constant. The above equation was simplified to equation 2-30 assuming that the potential gradient inside the solution is not significant relative to the concentration gradient: i nf = D C x (2-30) In this region, the limiting current under diffusion control for the pit cavity is estimated as in equation Tester and Isaacs [102] conclude that their diffusion model makes it possible to estimate the dissolution rate based on diffusion control. They point out that at concentrations below 3 M chloride ions, the dissolution of chloride salts is independent of concentration, but below 5 M, dissolution of the chloride salt layer is related to chloride activity. However, above this concentration, the viscosity of chloride ions affects the diffusion rate and reduces the dissolution rate. Gaudet et al. [52] identified a steady state between the dissolution reaction and diffusion rate by plotting current density vs surface concentration (Figure 2-21). Two isopotential lines were plotted after IR correction, based on pit depth: low potential (line B'G) and high potential (line BIHG). There are three steady state points (I, H and G) in contact with the straight line of diffusion current (GA), which indicates the diffusion rate and whose slope decreases as pit depth increases. The stability at points G and I can be obtained by either potentiostatic or galvanostatic control, but only by galvanostatic control at point H. 60

61 Figure 2-21: Steady state points between diffusion and rate of dissolution [52] Stable pit transition Pistorius and Burstein [72] propose criteria of stable pit transition based on the pit stability product (i.r), where i is pit current density and r is pit depth, assuming 3 M as the minimum concentration of dissolved cations required to maintain ph at active dissolution and 6 M as the maximum concentration. These concentrations present 75% and 150% of saturation concentration of 4 M FeCl 2 respectively. A steady state was assumed for an open hemispherical pit of radius r with diffusion mass transport, while migration and convection were ignored. The concentration of dissolved metal ions was then calculated at the pit bottom using equation 2-31: C = ( 2π ) i. r (2-31) 3nFD where C is build-up ion concentration, i is metal dissolution current density, r is pit radius, D is the diffusion coefficient of metal ions, taken as cm 2.s -1, F is the Faraday constant and n is the valency of dissolving metal ions. The resulting range of values for stable pit growth was 0.3 A.m -1 < i.r < 0.6 A.m -1. Since stable pits grow in a similar way to metastable pits, it was believed that below 0.3 A.m -1 pits would grow with the additional support of a metal cover but loss of the cover would cause a metastable pit to die. Above 0.3 A.m -1, the pit would become stable and the metal cover would have no important function as a diffusion barrier. Stable pit 61

62 transitions with respective stability products of 0.4 A.m -1 reported by Frankel et al. [74] and Williams et al. [103]. and 0.6 A.m -1 were also Conditions of transition from a metastable to a stable pit (stability criteria) were also suggested as being similar to a stable crevice transition, based on the critical aspect ratio of the diameter to the depth of the micropit, satisfying the process of IR dissolution when IR > ΔФ * [104]. This was suggested to occur as a sequence of breakdown and repassivation happening at the same place until the aspect ratio of the micropit exceeded a critical value such as the critical depth of active micropit dissolution. In a crevice study [105], increasing the applied potential showed a linear relation between drop potential (IR) and potential range of passive to active transition (ΔФ * ). A recent study of pitting corrosion on 316 stainless steel by Al Ameri [106] using potentiodynamic polarisation (PDP) and potentiostatic polarisation (PSP) demonstrated the importance of applied potential for stable pit transition, providing sufficient current to achieve the critical value of pit stability product Growth rate of stable pits Stainless steel develops a passive film that keeps the rate of general surface attack very low (for instance, μm/year), but localised pitting corrosion can reach more than 12 mm/year [107]. However, pit growth over time is a dynamic process subject to various control methods (section 2.2.4) and depending on a wide range of environments condition and materials. These control methods lead to different relationships of pit growth over time with potential, current and current density. The rate of pit growth is often expressed in terms of pit depth, or pit radius assuming a hemispherical shape, and can be used to estimate the lifetime of alloys and the effects of different aggressive environments. Pit growth kinetics can also be determined using electrochemical methods based on measurements of pit current (I) or current density (i) as a consequence of oxidation and metal dissolution to estimate the size of single stable pits, whereas for more than one pit, the mean value of pit depth can be estimated [59, 108]. An alternative method based on the penetration of a material (foil) of known thickness was used to estimate the rate of pit growth in an Al alloy. This was done to avoid a significant amount of anodic current being consumed by hydrogen evolution inside the pit [3]. 62

63 The above methods have been used to study pit growth and to derive general relationships, but resultant growth rates vary and a number of growth laws have been proposed [11, 12, 59]. Below, we review a number of studies reporting estimates of pit growth rate over time. Rozenfeld and Danilov [109] studied pit growth on 18Cr-10Ni-Ti in 0.1 N NaCl solution, using a galvanostatic method. They found that mean pit radius over time was related to r = kt 0.37, where k is constant. The current density decay was also estimated to be proportional to ~t 2/3. The authors claim that the reduction in current density with time indicates that the current was not constant. However, without current control the reduction in current density over time of pit growth was related to i~t 0.5. Pit growth under potentiostatic conditions was investigated by Engell and Stolica [110], who estimated the increase in pit current over time as I = kt b, where k and b are constants dependent on the environment and materials. They suggested that b=2 for a constant number of pits or a single pit over time and that b=3 for an increasing number of pits over time. Assuming hemispherical pits, they suggested that current density was constant and that pits grew faster in width than in depth. Forchhammer and Engell [111] used a movie camera to observe pit growth on 18Cr- 10Ni-0.2Mo alloy in 1 N NaCl, reporting that pit growth in radius was given by r = kt 1/3 and that for a constant number of pits, increase in current was related to time by I~t They found linear relationships between potential and number of pits and between chloride concentration and pit initiation time. In addition, the largest pit radius was proportional to time, while the number of pits showed a power relation. Earlier researchers had proposed that current of pit growth was given by I = ar 2 iz, where a is constant, r is pit radius, i is current density and z is the number of pits. At low chloride concentrations, there was a linear relationship with rate of pit growth, but this was reduced to the square root at high chloride concentrations [110, 112]. Herbsleb and Engell [113] used a microscope to observe pit growth in iron at constant potential, reporting a linear relation between pit radius and time. Assuming that pit shape is hemispherical, pit growth was suggested to occur at constant current density [114]. In similar work, Strehblow and Wenners [115] imaged single pit growth for a short period above pitting potential and reported that at early stages of pit growth, the radius was in linear relation with time at constant current density, assuming hemispherical pit shape. 63

64 Current density was estimated at 17 A.cm -2 for Fe, with pit radius ranging from 0.3 μm to 0.6 μm. Hunkeler and Bohni [116] studied pit growth by measuring the time required for complete penetration of stainless steel foil of different thicknesses. Assuming pit growth was hemispherical, they found that pit depth changed with t 0.5 and that current density was related to t They suggest that mixed ohmic/charge transfer control applied at low potential, with diffusion control at high potential. By limiting the exposed area of 304 stainless steel to 3 mm 2 using lacquer, Newman and Franz [117] grew a single pit under constant potential in 1 M NaCl with 0.04 M sodium thiosulfate. They found that pit current was related to t 0.5 below 100 s and above 300 s, whereas between these times the relationship was linear. Assuming hemispherical pit shape, growth in pit radius was proportional to t 0.5 below 100 s and above 300 s, and to t 2/3 between these times. A linear relation between potential and current of single pit growth suggested ohmic control. Since the current density was not calculated, Frankel [114] suggests that current density of pit growth is related to t -0.5 below 100 s and to t -1/3 after that time. Alkire and Wong [118] exposed a small area of stainless steel 100 μm in diameter to 1 N H 2 SO 4 containing 0.1 M NaCl by masking a sample with photoresist, then mounting it face upwards in the solution at an applied potential of 600 mv vs SCE. This method produced a single pit in 80% of all samples. The pit current density was obtained by using two measurement methods to estimate pit surface area. The first was based on final depth, assuming hemispherical shape, while the second used the formula A = 4r 2, because measurements revealed a ratio of 0.5 between pit depth and mouth radius, indicating shallow pits. The researchers found that current density decreased with time to t -0.5 and suggested that the corrosion rate was controlled by mass transport of dissolved corrosion products. The current density obtained from single pit growth was estimated in the range of A.cm -2 using the first method, and A.cm -2 using the second. The difference between the two methods was thus approximately 10%. Anodic current density in the range of A.cm -2 was also obtained, based on the estimated pit surface area on Fe-16Cr alloy polarised in 0.7 NaCl solution with 0.7 N Na 2 SO 4 in deaerated solution at ph 4 [119, 120]. Frankel [74] also reported an average current density between 2 and 4 A.cm -2 using SS 302 in a chloridecontaining solution. Current density was later estimated in the range of 2-8 A.cm -2 using 304 stainless steel polarised in 0.8 M NaCl/0.2 M HCl solution [72]. Using rotating 64

65 stainless steel electrodes polarised at constant potential in 0.2 M NaCl/0.1 M Na 2 SO 4 solution, Sato [121] found that the current density of pit growth was 8 A.cm -2 and a function of the area of the pit mouth but independent of potential. In earlier work, Isaacs [91] assumed that the gradient through the salt film layer was constant, then used equation 2-30 to estimate the limiting current density in stainless steel based on diffusion rate where D= cm 2.s -1. The concentration at the X diffusion boundary was assumed to be given by C s = C sat. At x = 0 the initial surface δ t concentration C s =0 and at the pit bottom (x=δ t ) the surface concentration C sat = 4.2 M. Changes in current density were found to be related to t -0.5 and associated with increasing pit size. Laycock and Newman [29] state that pit growth under an active or salt film depends on current density and that the transition potential to a stable pit is a function of chloride concentration. The transition current density of 1D pit growth on stainless steel was reported in the range of 1-5 A.cm -2. Furthermore, a 20% reduction in the limiting current density of 1D pit growth in stainless steels was achieved by increasing the concentration from 0.1 M to 1 M NaCl with no effect on Mo content [122]. The addition of 0.5 sulfate to 1 M NaCl solution reduced the transit current density of 1D pit growth in 302 stainless steel from 8.8 to 2.5 A.cm -2. The pits were grown for more than 700 seconds and underwent current decay, ending with values below 1 A.cm -2. The value of (D.C sat ) was also reduced from to mol.cm - 1.s -1 [123]. The current density in 2D pit growth was measured along the pit perimeter using in-situ X-ray synchrotron radiography based on the speed of movement of the pit boundary. Three regions were identified from the results: a salt film where the current density was 1-3 A.cm -2, a passive region and an active region with maximum current between 3-5 A.cm -2, related to lateral active lobe growth [27]. Pit growth was observed in a vertical sample of stainless steel in 3.5% NaCl, where maximum current density was measured at 30 A.cm -2 after 18 s, followed by decay to a stable range of 2-8 A.cm -2 [124]. In another study of pit growth at constant potential of 350, 150 and 50 mv, current density reached approximate maximum values of 41, 26.5 and 6 A.cm -2 respectively, then decayed to constant values close to 1.8, 4.8 and 3 A.cm - 2 [125]. Hisamastsu et al. [126] indicated that when estimating the current density based on pit surface area, the assumption of a hemispherical shape may lead to an overestimate of pit surface area and thus to a lower current density. 65

66 Salt film formation in stable pits A salt film precipitates inside a pit once the corrosion products of metal dissolution exceed their saturation concentrations or the solubility limit of the metal salt, which depends on temperature. In pitting corrosion, Alkire et al. [127, 128] suggest that a salt film forms at an early stage of pit initiation, within a small region of the oxide-free metal surface and it is believed to be necessary for stable pit transition. Isaacs [91] studied the behaviour of the resistive layer (salt film) which developed inside pits on an artificial electrode of 304 stainless steel after polarisation using constant potentials of 0.5, 0.8 and 1 V vs SCE. This layer was reported to have a complex chemistry rich in iron and it was found that its thickness depended on the potential, solubility and diffusion of metal ions away from the layer/solution interface. The resistivity of a layer of 0.1 μm thickness was estimated by measuring the impedance at 10 8 ohm.cm and it was suggested that it would restrict the rate of metal dissolution. Inside the pit, the potential drop (IR) across the salt layer was found to have reached 95%. A linear relation was observed between the applied potential and the potential across the salt layer, equal to the current, and its resistance indicated the ionic conductivity of the salt layer. During the quasi-steady state period (Fig 2-20), pit growth became independent of potential, which is explained by the self-regulating thickness of the salt layer, whereby its ohmic resistance was readjusted by the diffusion flux of ions from the interface. However, at a higher concentration of 5 M, the activity and viscosity of the chloride solution increased and affected the solubility of the salt layer and diffusion flux, thus limiting the dissolution rate. It has been proposed that an oxide film is formed under the salt film [129]; Newman and Ajjawi [130] assert that this occurs in the presence of nitrate in chloride solution and is temporary without nitrate. Laycock [78] suggests that below CPT, a salt film is formed but is mediated by oxide passivation. Dissolution of the oxide film formed beneath the salt film was expected when the salt layer thickness was reduced, depending on the ph of the solution. Rayment et al. [131] studied the salt film inside artificial pits on 316L stainless steel in 1 M NaCl using in-situ synchrotron X-ray diffraction and conclude that the salt film formed is FeCl 2.4H 2 O as the main compound. 66

67 Shape and morphology of pit growth Earlier, Schwenk [112] studied the shape of pits grown on 304 stainless steel in 1 M NaCl with and without 1 N H 2 SO 4 and 0.5 M NaNO 3, and found that they initially grew hemispherically, with porous metal covers, but that this changed with time. Below pitting potential (E p ), the pits were regularly etched and often grew with either circular, hexagonal or square edges. It is believed that crystallographic planes such as (111), which has the lowest dissolution rate in austenitic steel, affect the pit shape because the current is low. Above the pitting potential, a higher dissolution current leads to both polished and dull isotropic features with smooth surfaces. However, Ernst and Newman [28] argue that changes in the pit shape are caused by the formation of holes in the metal cover during pit growth, rather than by non-uniform current density inside the pit. The sample position and the gravity effect were found to influence the accumulation of chloride ions inside the pits and leads to various pit morphology. Mankowski and Szklarska-Smialowska [94, 132] studied the shape of pits grown in stainless steel for 5 hrs using three different positions of samples: vertical, and horizontal facing up and down. For all three positions, pits were hemispherical at first, then became cup-like with a pit diameter/depth ratio around 4. Subsequent changes pit morphology depended on sample position. In the vertical sample, the cup shape became ellipsoidal and deformed on the lower side, which grew less than the upper side. In the horizontal sample facing up, the pit morphology showed prolonged growth with an almost flat bottom surface, while in the downward-facing sample a distinct cavity developed at the bottom of an elongated pit shape. The researchers also observed a convex shape to the salt film at the bottom of the pits and new holes appearing around the pit mouth. They suggest that a change in film tightness may lead to different growth rates and pit shapes. This observation was explained by the varying distribution of chloride ions inside the pit, which changes the thickness of the salt film on the pit surface. The concentration of chloride ions is reduced at the edge of the pit, relative to its interior, by mixing with solution from the new holes. Therefore, there is a rise in anodic current density at the edge of the pits, increasing the rate of dissolution and broadening the pit shape. Hisamatsu et al. [133] state that dissolution inside a pit on a horizontal sample facing upward is related to potential and anion species inside the pit and that when pit growth is faster radially relative to depth, this leads to shallow pit morphology. Similarly, shallow pit growth on iron in 1 N H 2 SO 4 solution was observed with increasing chloride 67

68 concentration, suggesting that pit growth is related to chloride concentration and independent of potential [113, 134]. Tousek [135] found that pits grown in Cr-Ni stainless steel in 0.5 M NaCl at ph 8.4 had a rotary ellipsoid shape and that the ratio of depth to radius was 0.5. Tomashov et al. [136] used rotating electrodes to study pitting on stainless steel in a chloride-containing acid solution and report that the ratio of pit diameter to depth was , depending on the Mo content of the steel. Szklarska- Smialowska and Mankowski [137] studied the effects of temperature on pitting shape in Cr18-Ni12-Mo2-Ti stainless steel using samples oriented vertically in 0.5 N NaCl and 0.1 N H 2 SO 4 solution. They report that after 90 min of pit growth at 20 C, the diameter-to-depth ratio increased gradually to 5, while at 35 C and 50 C the ratio reached only 3 and 2 respectively. At a similar polarisation potential, the number of pits was found to increase with increasing temperature. Sato et al. [121] studied the shape of pits grown under constant potential in H 2 SO 4 containing chloride ions using SUS-27; 18.5Cr-9Ni-1Mn commercial stainless steel. They found that pits were approximately hemispherical and that the ratio of diameter to depth depended on potential and increased linearly with time. Reiser and Alkire [134] measured pit shape by scanning electron microscope (SEM) photogrammetry. They took micrograph images from two angles to determine the 3D characteristics of the pits, which were grown on pure iron at 0.65 V (Ag/AgCl) in 1 N H 2 SO 4 containing N NaCl. After 76 measurements, they concluded that the pits ranged in size between 7 and 200 μm and that those below 50 μm were hemispherical, while those above this size were shallow. In an earlier study, Sato [138] measured pit shape in stainless steel at various potentials and identified the two states of pit growth (active and passive) shown in Figure 2-22(a). At high potential (above pitting potential), pits grew with a polished surface while chloride concentration remained above the critical value of metal salt. This produced near-hemispherical pits. At low potential, active pits were etched in irregular shapes related to the microcrystal planes in the crystallographic structure of the pit surface. Because ph inside the pits was maintained below the critical value, their behaviour was similar to that of crevice corrosion. The author proposed that the passivationdepassivation potential (E pit ) inside the pit acts as a boundary between the polishing state and the active state, and is related to the critical chloride concentration. The transition from polishing state to active state or passive mode is believed to occur by 68

69 potential decay, which depends on pit size. Thus, if the size of the pit is below a critical value the pit will turn passive, while above the critical size the pit will propagate become active (Figure 2-22b). Above the passivation potential, the potential range of the polishing pit state increases with increasing chloride concentration. Moreover, increasing the pit size increases the range of active state potential in the direction of the noble potential, while the ph value is reduced. (a) Figure 2-22: (a) Two states of pit shape (polishing and etching surface) observed through anodic polarisation. (b) Transition of two conditions (active and passive) of pit from polishing state by potential decay, depending on pit size [138] Metal covers Metal covers on pits consist of a number of holes forming rings around the pit mouth in a sequence of stable pit development that has been observed and reported by many researchers [102, 112, 130, 139]. Ernst et al. [28, 140, 141] propose a mechanism whereby lacy metal covers are developed in 2D, as shown in Figure The development of the lacy metal cover was studied in 50 µm thick 304 stainless steel foil in 1 M NaCl solution. The foil was sandwiched between two optical glass sheets, fixed with epoxy resin and oriented upwards. Pit growth under a polarised potential of 0.6 V vs SCE was monitored using a video camera attached to a microscope. The development of the lacy metal cover started under the passive surface where a hemispherical cavity initiated, while the surface around the pit rim remained passive (Figure 2-23a). This is believed to occur when the concentration of dissolved metal falls (b) 69

70 below a critical concentration (b and c). Dissolution of the metal then undercuts the passive surface, forming lobe features which grow laterally until further holes emerge at the metal surface (d and e). These holes allow water to enter, diluting the pit solution around them. Cations also diffuse out and the solution drops below the critical concentration. This step induces quick passivation of the metal around the hole. The process of undercutting is repeated to form new holes at the metal surface, leading to further dilution and passivation (f). While this model applies to samples facing up, it is proposed that in those facing downward, the solution inside the pit will emerge through the holes under gravity. Figure 2-23: Sketch of lacy metal cover and pit shape development as a function of concentration inside the pit. Passive film on the metal surface is shown by a thick line [28]. Ernst et al. [141] also suggest that new pits with lacy covers may form at the bottom of old pits, by repolarisation after passivation caused by a potential drop at room temperature. Repeating this process may then lead to the formation of pits within pits, in a Russian doll structure. Ernst and Newman [142] indicate that the morphology of pit covers may change under the influence of chloride concentration on the CPT of stainless steel. At high chloride concentration, open pits with no lacy covers were formed, while classic lacy covers developed at low concentration. They suggest that no pit covers form at low critical fraction of saturation (R c ), because high chloride concentration affects passivation of the pit rim. The value of R c is given by R c = C C, where C and C sat are the critical and saturation concentration respectively; it sat was suggested that C* falls between 70% and 80% [52]. Ernst and Newman also found 70

71 that the pit bottom had a rough surface while a viscous bulk electrolyte suggested no salt film. In another study, Ernst et al. [140] reported a flask-shaped pit without lacy cover when R c =1 and suggests that the spacing between the holes in lacy metal covers may be related to pit depth and R c The effect of inclusions on pitting corrosion It is generally accepted that the chemical structure of inhomogeneities such as surface defects, inclusions (intermetallic and non-metallic phases), precipitates and second phases plays an important role in the breakdown of passivity of stainless steels in chloride-containing solutions [143]. They also tend to have different electrochemical potential and low area ratio relative to the alloy matrix; therefore, localised attack may be accelerated electrochemically [144]. In addition, pits were not preferentially nucleated in chloride solution by mechanical damage to the passive film using a scratch technique at fixed potential. Therefore, the number of pits initiated and growing is believed to be related to inclusions [145]. Non-metallic MnS inclusions have often been found to act as preferential sites for pit initiation on stainless steel in chloride environments [86, 146, 147]. Three mechanisms of attack are proposed, based on inclusion dissolution, chromium depletion and oxide stress [144]. Alloy microstructure and system geometry, for example surface roughness and specimen orientation, are also important in explaining pit initiation sites; however, the initiation stage is still not fully understood. In earlier work, Eklund [146] suggested a mechanism of inclusion dissolution based on thermodynamic calculation and claimed that at MnS inclusions, pits were unstable above -100 mv vs SCE. Dissolution of inclusions and matrix metal were reported in chloride solution, but chloride ions were preferentially adsorbed onto the MnS inclusions during immersion in ferric chloride solution, indicating chloride catalysis. Alternatively, dissolution of inclusions may only occur at a certain potential but without pit dissolution of the adjacent matrix, such as in 1 M Na 2 SO 4 solution [87, 148]. Williams et al. [85] state that the electrochemistry of inclusions appears to be chloride-catalyzed. Lott and Alkire [149] propose that MnS dissolution produces thiosulfate as a by product which catalyses pit initiation, while Brossia and Kelly [150] argue that the release of sulphide species during the chemical dissolution of inclusions leads to depassivation and microcrevice initiation, whereas chloride ions only support the process. The influence of thiosulfate was later studied by Webb and Alhire [8], who found that thiosulfate increased the dissolution of MnS and 71

72 that stable pit formation occurred at a critical concentration of thiosulfate when chloride solution was added. To explain the sequence of MnS inclusion dissolution, Baker and Castle [151] propose salt film precipitation with MnCl 2. Suter et al. [148] studied pitting resistance relative to sulphur content in four alloys and reported that increasing the sulphur content from to 0.3% reduced the pitting potential from 790 to 170 mv vs SCE. However, Stewart and Williams [86] related pit initiation to MnS size, rather than sulphur content. Rossi [152] reports that in high sulphur stainless steel (0.29% S), both MnS and mixed CrOx/MnS inclusions were present, but the MnS was not protected by the passive film. Therefore, when the alloy was immersed in FeCl 3 solution, inclusions underwent either complete or partial dissolution. This was also observed in stainless steel of low sulphur content (0.003% S). MnS inclusions were reported to become more active when combined with metal oxide or silicate with or without chloride solution [153, 154]. Muto et al. [155] used a microcell to study the electrochemical dissolution of MnS inclusions and pitting of type 303 stainless steel in 1 M Na 2 SO 4 and 0.1 M NaCl. Dissolution of MnS was found to occur at V (Ag/AgCl, 3.33 M KCl) in 1 M Na 2 SO 4 and at V in 0.1 M NaCl. MnS dissolution was also observed at V in 0.1 M NaCl, which is below the potential of metastable events. It was concluded that the dissolution of inclusions does not lead directly to pit initiation independently of the critical concentration of dissolved species. It is believed that the synergetic effect of released sulphur and chloride ions leads to active dissolution and pit initiation, while at certain concentrations it leads to stable pit transition. Polygonal metastable pits of around 1 μm diameter were reported to initiate at the boundary of MnS inclusions, while stable pits formed at 0.5 V. In another study, Muto et al. [5] compared the polarisation behaviour of CrS inclusions with manganese-rich (Mn,Cr)S inclusions using stainless steel type 304 with three Mn contents (0.01, 0.76 and 1.67%) in three concentrations (0.1, 0.5 and 3 M) of NaCl solution. Macroscopic polarisation showed that pitting potential was reduced and that metastable pitting events increased with increasing Mn content. It was also shown that the dissolution of CrS inclusions occurred in the transpassive region at 1.25 V, while (Mn,Cr)S inclusions started to dissolve in the passive region and stable pits were observed on MnS inclusions in the potential range of V in 0.1 M NaCl. It was 72

73 concluded that the passive film above CrS inclusions became four times thicker and more protective with increasing potential relative to the oxide film above (Mn,Cr)S inclusions which consisted of about 10 wt.% Cr. Earlier, Krawiec et al. [156, 157] studied three different structures of type 303 stainless steel (forged, hot rolled and cold worked) and reported that chemical dissolution of (Mn,Cr)S inclusions enriched in chromium (30-40 wt.%) in 1 M NaCl and 1 M NaClO 4 solution at ph 3 was slightly higher relative to MnS inclusions. Chromium-enriched inclusions underwent no dissolution in 1 M NaCl at ph 3, but dissolution was observed in 1 M NaClO 4 at ph 3 above 500 and 600 mv vs SCE in forged steel. However, dissolution of chromium-enriched inclusions in some areas was also observed at 100 mv. This observation was interpreted as related to the depleted zone at the inclusion/matrix boundary. In another study, Chiba [158] reported that stable pits grew deeply in the matrix and beneath inclusions at the MnS/steel boundary. Stable pit formation was found to be related to the orientation of inclusions. Webb et al. [159] reported that stable pits are more strongly related to deep inclusions than to large shallow inclusions, even when both are associated with SiO particles. In addition, Ke and Alkire [160] investigated 200 inclusions in 304 stainless steel in 0.1 NaCl solution and report that inclusions above 0.7 μm were more likely to be associated with stable pit growth. Pitting attacks were found to occur in MnS and oxide inclusions. In previous studies, it was reported that in NaCl and Na 2 S 2 O 3 solution, MnS inclusions larger than 1 μm tended to form stable pits [161]. Pit initiation at the interface between MnS inclusions and the metal matrix was observed by Park et al. [82] and Muto et al. [162], while Baker and Castle [151] observed attack both at the edges of MnS inclusions and in small active regions in the middle of the inclusions. Inclusion attack were found to form micro-crevices between the metal matrix and MnS inclusions in 304 stainless steel after polarisation in chloride solution [159]. Schmuki et al. [163] studied 200 MnS inclusions and observed different morphologies of attack. They found that 25% of attacks occurred inside an inclusion and 20% at the interface, while 10% were mixed attacks and 40% of MnS inclusions showed no sign of being attacked. Ryan et al. [164] propose that attack at the MnS inclusion/matrix interface occur because chromium-depleted zones develop during cooling when the materials are 73

74 manufactured. This finding is discussed by Meng et al. [165], who investigated the same materials but did not observe chromium-depleted zones around inclusions. Similarly, Schmuki et al. [163] investigated 27 MnS inclusions and reported finding no chromium-depleted zones around them. On the other hand, Williams et al. [6] observed a narrow halo of half-width typically 100 nm around MnS inclusions enriched with FeS. They suggest that this formed by cooling during manufacture and may play a role in pit initiation by accumulating corrosive species and chloride ions after dissolution. A study of pitting in 316L SS found that pit initiation started at the interface between oxide inclusions and the matrix in a sulphur environment containing chloride ions, but corrosion resistance was found to vary according to the type of oxide inclusion (Mg, Al, Ca). The result of Kelvin probe force microscopy indicate that a lower surface potential was associated with reduced corrosion resistance [166] The effect of stress on pitting corrosion Suter et al. [167] studied the influence of stress (at 60% and 80% of yield strength) and inclusion morphology on pitting behaviour using a microelectrochemical cell. They used type 304 stainless steel with three different sulphur contents (0.001, and wt.% S) in 1 M Na 2 SO 4 and 1 M NaCl solutions. The samples with the highest and lowest sulphur contents were in plate form and the other was a rod. The authors report that applying stress shifted the pitting potential negatively by 150 mv and induced stable pits on shallow inclusion in 1 M NaCl, while sites with no inclusions showed no pitting, either with or without stress, in low sulphur content steel. However, the growth rate of metastable pits was reported to increase under stress [168]. Recently, Shimahashi [169] used microelectrochemical polarisation measurement to investigate the effects of applied stress (60% of the 0.2% proof stress) on MnS dissolution in type 304 stainless steel in 1.5 M MgCl 2 solution. The results indicate that micro-cracks perpendicular to the tensile stress occur on the surface of MnS inclusions. This cracking was attributed to the breakdown of the oxide film, leading to stable pit formation. Horner [170] suggested a minimum pit size for pit-to-crack transition and a crack was also observed to emerge from the pit in 3 NiCrMoV disc steel in a sodium chloride environment, with features of finger-like cracks [23]. Turnbull et al. [18] studied strain around single pits on cylindrical specimens using finite element analysis and found that strain was concentrated at the pit mouth. They suggest that plastic strain 74

75 occurs in the pit wall due to the application of high stress and is concentrated just below the pit mouth, while plastic strain rates affect pit growth. The dynamic plastic strain developed at the pit shoulder under stress was identified as a new factor in pit-to-crack transition and SCC, in addition to pit chemistry and potential [18, 171]. In another experiment, Turnbull and colleagues found that 50% of cracks initiated from the pit shoulder, 7% from the pit bottom and the rest between shoulder and bottom [172]. Gutman et al. [31] studied the influence of strain with plastic deformation on anodic polarisation (the mechanochemical effect) under constant load using type 316L stainless steel in 0.1 N Na 2 SO 4 + 5% H 2 SO 4. The result show that maximum current density occurs with increasing plastic strain in the passive and transpassive regions, depending on potential. Hoar and Scully [32] showed that under yielding stress, the rate of anodic dissolution increased by a factor of 10 4 for 18Cr-8Ni stainless steel in 42 wt% MgCl 2 solution at 145 C under the control potential of -140 mv vs SHE. Zhu et al. [14] studied the effect of very low stress of 20 MPa on pit-to-crack transition in a sample of single crystal type 316L in 45 wt% magnesium chloride solution at boiling point under OCP. The test indicated that cracks nucleated preferentially at the shoulders of the pit The effect of temperature on pitting corrosion At room temperature or below a specific value of temperature, some materials do not show stable pitting at any potential or exposure time in some environments, but manifest only a potential breakdown relating to the dissolution of a transpassive region [142]. Therefore, pitting occurs in these alloys only above their critical pitting temperature, which depends on the environmental conditions. Increasing both chloride concentration and temperature significantly reduces the pitting potential. CPT ranges in stainless steel from about 10 to 100 C and can be used to rank stainless steel resistance [173]. Increased temperature was observed to accelerate the dissolution of MnS inclusions at lower potential in a macro- and micro-electrochemical study of AISI 304 [82]. The morphology of pit also changed with temperature and chloride concentration, as discussed in section A study of types 302, 304 and 316 austenitic stainless steel shows that pitting potential is reduced with increasing temperature and this is attributed to a reduction in activation energy over potential [174]. Szklarska-Smialowska [59] reports a reduction of about 75

76 200 mv in pitting potential associated with increasing temperature in the range of 30 to 100 C. Wang et al. [175] observed the same effect of temperatures up to 200 C and conclude that increasing temperature reduces the protective effect of the oxide film by increasing its porosity. 2.4 Application of X-ray tomography X-ray tomography has been used in-situ and ex-situ to study and visualise microstructural features (pores, phases, inclusions, cracks, intergranular corrosion and pit corrosion) under controlled environmental conditions (temperature, stress/strain, corrosive environment and potential). The in-situ application of X-ray CT is limited by any changes occurring in the corrosion process during the scan itself, because each 3D tomogram can take up to several hours. Therefore, X-ray CT can be used only to capture images of evolutionary features which undergo insignificant changes during the time of the scan. For instance, imaging of microstructural evolution during experiments was performed in situ under mechanical testing (compression and tension) and thermal treatment (sintering, forming of metal foams) [176]. Synchrotron X-ray tomography was used differentiate gas pores from the pores caused by micro-shrinkage inside cast aluminium alloys [177] and to conduct in-situ tensile tests on composite materials [178]. Maire et al. [176] investigated defects in aluminium alloys using the absorption mode and Ludwig et al. [179] studied the interaction between short crack fatigue and grain boundaries in cast aluminium alloys. Marrow et al. [19] observed the development of short fatigue cracks in low alloy austempered ductile cast iron and both intergranular corrosion and stress corrosion cracking on 302 sensitised austenitic stainless steel [20]. Localised corrosion and environment-assisted cracking were also investigated in aluminium alloys [21, ]. Ghahari et al. [24, 25] used in-situ X-ray synchrotron microtomography to study corrosion pits grown at the tip of a pin electrode of type 304 stainless steel using a micro-electrochemical cell under galvanostatic and potentiostatic control. In another study [27], in-situ synchrotron X-ray radiography was used to study pit propagation in stainless steel foil under galvanostatic and potentiostatic control in different chloride solutions. A 2D image (for example Figure 2-24) was recorded over time to estimate pit propagation kinetics based on displacement of the pit boundary over time due to metal dissolution. Recently, Burnett et al. conducted a 3D microtomography study of pitting 76

77 corrosion and intergranular corrosion in sensitised type 316H stainless steel in 0.1 M NaCl solution. The results show the occurrence of both intergranular corrosion and pitting attack. Pits nucleated at the surface of the electrochemically polarised sample, with the aggressive pit solution resulting in intergranular attack emanating along grain boundaries from within the pit [183]. Figure 2-24: Radiograph images of pit growth on foil of type 304 stainless steel in chloride solution at 650 mv vs Ag/AgCl [27]. 77

78 2.5 Summary The process of pitting corrosion in stainless steel involves pit initiation and propagation. The initiation of corrosion pits occurs by localised attack on a small area, starting with breakdown of the passive oxide film, leaving a small area exposed to the environment. Then, initially the pit (as a whole) acts as an anode compared to the intact cathodic surface. Oxide film breakdown is commonly related to the position of inclusions, in particular MnS. Anodic dissolution reaction and hydrolysis reaction of dissolved metal ions lead to a decrease in ph inside the pit. The low ph is considered as the main reason for pit growth under active metal dissolution, which is in parallel related to a critical concentration of metal ions (typically, above 60% of the saturation concentration). Such an environment leads to active metal dissolution and stable pit growth, which is primarily depended on the stability product (i.r). The latter is defined by the pit depth (r) and current density (i), with stable pit growth associated with 0.3 A.m -1. Below the critical pitting potential, breakdown of the passive film leads to nucleation of metastable pits, which are characterised by short growth periods, follow by sudden repassivation. The rate of metal dissolution at this potential is not sufficient to provide critical metal ion concentration inside the pit for stable pits to grow without metal lacy cover. A lacy metal cover often develops, which acts as a diffusion barrier, resulting in an increase ion metal ions inside the pit. When these covers are lost, the pits can either suddenly passivate, or if they are stable/deep enough, pits can grow. However, above the critical pitting potential, stable pit growth can occur with high rates of local metal dissolution. This process leads to formation of salt film at the pit bottom, and the pit growth then changes from activation control (potential dependent) to diffusion control (potential independent). Then, the pit starts propagating by the under-cutting mechanism, often referred to as lobe formation. Maintaining active corrosive environment inside the pit is a function of the balance between the rate of dissolution inside the pit, and the rate of diffusion of metal cations out of the pit. The active dissolution process under applied potential is influence by the 78

79 potential drop in the solution, and through the salt film. However, the current resistance inside the pit is reduced by increasing the metal dissolution, which affects the solution conductivity inside the pit. Inside the pit, three regions (Figure 2-25a) on the pit perimeter can develop, and these include: (i) a salt film at the pit bottom, (ii) passive regions close to the top edge of the pit, and (iii) in between a free salt film at the middle-height region. These regions show diffusion and active growth control relative to presence of salt films, and lead to shape change of pit growth from hemispherical to elongated dish shape. Similarly on samples exposed vertically (Figure 2-25b), the lower side of the pit will be covered with a salt film up to the open pit mouth, with the free salt film under active dissolution control, and the upper side of the pit is passive. This also depends on the size of the pit where holes may cover or uncover by the oxide film b a M n+ Chloride solution Passive film Passive film Free salt film Sample Salt film Free salt film M n+ Salt film Chloride solution Sample Figure 2-25: Sketch shows the position of salt film in (a) horizontal orientation and (b) vertical orientation samples. 79

80 2.6 References 1. Lo, K.H., C.H. Shek, and J.K.L. Lai, Recent developments in stainless steels. Materials Science and Engineering: R: Reports, (4 6): p Fontana, M.G., Corrosion Engineering, 3/E. 2005: Tata McGraw-Hill Education. 3. Frankel, G., Pitting corrosion of metals a review of the critical factors. Journal of the Electrochemical Society, (6): p Frankel, G.S. and N. Sridhar, Understanding localized corrosion. Materials Today, (10): p Muto, I., S. Kurokawa, and N. Hara, Microelectrochemical investigation of anodic polarization behavior of CrS inclusions in stainless steels. Journal of the Electrochemical Society, (11): p. C395-C Williams, D.E., et al., Composition changes around sulphide inclusions in stainless steels, and implications for the initiation of pitting corrosion. Corrosion Science, (11): p Meng, Q., et al., Metallurgy (communication arising): Stainless-steel corrosion and MnS inclusions. Nature, (6947): p Webb, E.G. and R.C. Alkire, Pit initiation at single sulfide inclusions in stainless steel - I. Electrochemical microcell measurements. Journal of the Electrochemical Society, (6): p. B272-B Oldfield, J., Crevice Corrosion of Stainless Steels in Seawater. acom, 1988(1): p Malik, A.U., et al., The effect of dominant alloy additions on the corrosion behavior of some conventional and high alloy stainless steels in seawater. Corrosion Science, (10): p Newman, R.C., 2001 W.R. Whitney Award Lecture: Understanding the Corrosion of Stainless Steel. Corrosion, (12): p Soltis, J., Passivity breakdown, pit initiation and propagation of pits in metallic materials - Review. Corrosion Science, : p Guan, L., et al., Effects of cyclic stress on the metastable pitting characteristic for 304 stainless steel under potentiostatic polarization. Corrosion Science, (0): p Zhu, L.K., et al., Stainless steel pitting and early-stage stress corrosion cracking under ultra-low elastic load. Corrosion Science, (0): p Begum, Z., et al., Studies of the tensile and corrosion fatigue behaviour of austenitic stainless steels. Corrosion Science, (4): p LÜ, G., et al., Effect of Strain and Chloride Concentration on Pitting Susceptibility for Type 304 Austenitic Stainless Steel. Chinese Journal of Chemical Engineering, (2): p Huang, Y., et al., Quantitative correlation between geometric parameters and stress concentration of corrosion pits. Engineering Failure Analysis, (0): p Turnbull, A., L. Wright, and L. Crocker, New insight into the pit-to-crack transition from finite element analysis of the stress and strain distribution around a corrosion pit. Corrosion Science, (4): p Marrow, T.J., et al., High resolution X-ray tomography of short fatigue crack nucleation in austempered ductile cast iron. International Journal of Fatigue, (7): p Babout, L., et al., X-ray microtomographic observation of intergranular stress corrosion cracking in sensitised austenitic stainless steel. Materials Science and Technology, (9): p Connolly, B.J., et al., X-ray microtomography studies of localised corrosion and transitions to stress corrosion cracking. Materials Science and Technology, (9): p

81 22. Steuwer, A., et al., In situ analysis of cracks in structural materials using synchrotron X- ray tomography and diffraction. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, (1): p Turnbull, A., et al., Technical note: Visualization of stress corrosion cracks emerging from pits. Corrosion, (7): p Ghahari, S.M., et al., In situ synchrotron X-ray micro-tomography study of pitting corrosion in stainless steel. Corrosion Science, (9): p Ghahari, S.M., In Situ Synchrotron X-Ray Characterisation and Modelling of Pitting Corrosion of Stainless Steel, in A thesis Singh, S.S., et al., In situ experimental techniques to study the mechanical behavior of materials using X-ray synchrotron tomography. Integrating Materials and Manufacturing Innovation, (1): p Ghahari, M., et al., Synchrotron X-ray radiography studies of pitting corrosion of stainless steel: Extraction of pit propagation parameters. Corrosion Science, : p Ernst, P. and R.C. Newman, Pit growth studies in stainless steel foils. I. Introduction and pit growth kinetics. Corrosion Science, (5): p Laycock, N.J. and R.C. Newman, Localised dissolution kinetics, salt films and pitting potentials. Corrosion Science, (10-11): p Oldfield, J.W. and B. Todd, Environmental aspects of corrosion in MSF and RO desalination plants. Desalination, (1 3): p Gutman, E.M., G. Solovioff, and D. Eliezer, The mechanochemical behavior of type 316L stainless steel. Corrosion Science, (7): p Hoar, T. and J. Scully, Mechanochemical anodic dissolution of austenitic stainless steel in hot chloride solution at controlled electrode potential. Journal of the Electrochemical Society, (3): p Davis, J.R., Alloy digest sourcebook: stainless steels. 2000: ASM International. 34. Lula, R.A. and J.G. Parr, Stainless steel. 1986: American Society for Metals. 35. Llewellyn, D. and R. Hudd, Steels: Metallurgy and Applications: Metallurgy and Applications. 1998: Elsevier Science. 36. McGuire, M.F., et al., Austenitic Stainless Steels, in Encyclopedia of Materials: Science and Technology (Second Edition). 2001, Elsevier: Oxford. p Totten, G.E., Steel Heat Treatment Handbook. 2007: Taylor & Francis. 38. Handbook, A.S.M., Welding, brazing and soldering. ASM Handbook, Handbook, A.S.M., Heat treating, vol. 4. ASM International, Materials Park, OH, 1991: p McGuire, M.F., Stainless steels for design engineers. 2008: ASM International. 41. Marshall, P., Austenitic stainless steels: microstructure and mechanical properties. 1984: Springer. 42. Bhadeshia, H. and R. Honeycombe, Steels: Microstructure and Properties: Microstructure and Properties. Butterworth-Heinemann. 43. Padilha, A.F., C.F. Tavares, and M.A. Martorano. Delta ferrite formation in austenitic stainless steel castings. in Materials Science Forum Trans Tech Publ. 44. Twi. [cited th June]. 45. Cramer, S.D. and B.S. Covino, ASM Handbook Vol. 13 A Corrosion: Fundamentals, Testing, and Protection. Materials Park, OH: ASM International, , Smith, E.H., Mechanical Engineer's Reference Book. 1998: Elsevier Butterworth- Heinemann. 81

82 47. Plaut, R.L., et al., A short review on wrought austenitic stainless steels at high temperatures: processing, microstructure, properties and performance. Materials Research, (4): p Engelberg, D.L., Intergranular Corrosion, in Shreir's Corrosion, B. Cottis, et al., Editors. 2010, Elsevier: Oxford. p Roth, I., et al., Crack initiation and short crack growth in metastable austenitic stainless steel in the high cycle fatigue regime. Procedia Engineering, (1): p Fox, P.D. and R. آ آ: Editor-in-Chief Hubert, CORROSION, in Encyclopedia of Dairy Sciences. 2002, Elsevier: Oxford. p Marcus, P., Corrosion mechanisms in theory and practice. 2002: CRC Press. 52. Gaudet, G.T., et al., Mass-Transfer and Electrochemical Kinetic Interactions in Localized Pitting Corrosion. Aiche Journal, (6): p Steinsmo, U. and H.S. Isaacs, Dissolution and Repassivation Kinetics of Fe-Cr Alloys in Pit Solutions.1. Effect of the Surface Salt Layer. Journal of the Electrochemical Society, (3): p Isaacs, H.S., The localized breakdown and repair of passive surfaces during pitting. Corrosion Science, (2 3): p Revie, R.W., Corrosion and corrosion control. 2008: John Wiley & Sons. 56. Szklarska-Smialowska, Z. and M. Janik-Czachor, The analysis of electrochemical methods for the determination of characteristic potentials of pitting corrosion. corrosion science, (12): p Szklarska-Smialowska, Z. and ZS-Smialowska, Pitting and crevice corrosion. 2005: NACE International Houston, TX. 58. Leckie, H.P. and H.H. Uhlig, Environmental factors affecting the critical potential for pitting in 18-8 stainless steel. Journal of The Electrochemical Society, (12): p Szklarska-Smialowska, Z., Review of Literature on Pitting Corrosion Published Since Corrosion, (6): p Macdonald, D.D. and M. Urquidimacdonald, Theory of Steady-State Passive Films. Journal of the Electrochemical Society, (8): p Schmuki, P., From Bacon to barriers: a review on the passivity of metals and alloys. Journal of Solid State Electrochemistry, (3): p Olsson, C.O.A. and D. Landolt, Passive films on stainless steels chemistry, structure and growth. Electrochimica Acta, (9): p Saadawy, M., Kinetics of Pitting Dissolution of Austenitic Stainless Steel 304 in Sodium Chloride Solution. ISRN Corrosion, Devasenapathi, A. and M. Asawa, Effect of applied potential on the nature of surface film and SCC of a high Mn stainless steel in 1 M HCl. Journal of Materials Science, (23): p Hamdy, A.S., E. El-Shenawy, and T. El-Bitar, The corrosion behavior of niobium bearing cold deformed austenitic stainless steels in 3.5% NaCl solution. Materials Letters, (13): p Jinlong, L. and L. Hongyun, Electrochemical investigation of passive film in predeformation AISI 304 stainless steels. Applied Surface Science, (0): p Sato, N., An overview on the passivity of metals. Corrosion Science, (0): p Burstein, G.T., Understanding Localized Corrosion through Electrochemical Measurements. ECS Transactions, (31): p Burstein, G.T., P.C. Pistorius, and S.P. Mattin, The Nucleation and Growth of Corrosion Pits on Stainless-Steel. Corrosion Science, (1-4): p

83 70. Böhni, H., Breakdown of passivity and localized corrosion processes. Langmuir, (6): p Szklarska-Smialowska, Z., Mechanism of pit nucleation by electrical breakdown of the passive film. Corrosion Science, (5): p Pistorius, P.C. and G.T. Burstein, Metastable Pitting Corrosion of Stainless Steel and the Transition to Stability. Philosophical Transactions of the Royal Society of London Series a-mathematical Physical and Engineering Sciences, (1662): p González-Garcıá, Y., et al., Imaging metastable pits on austenitic stainless steel in situ at the open-circuit corrosion potential. Electrochemistry Communications, (7): p Frankel, G.S., et al., Metastable Pitting of Stainless Steel. Corrosion, (7): p Tang, Y., et al., The metastable pitting potential and its relation to the pitting potential for four materials in chloride solutions. Corrosion Science, (0): p Moayed, M.H. and R.C. Newman, Evolution of current transients and morphology of metastable and stable pitting on stainless steel near the critical pitting temperature. Corrosion Science, (4): p Moayed, M.H., N.J. Laycock, and R.C. Newman, Dependence of the Critical Pitting Temperature on surface roughness. corrosion science, (6): p Laycock, N.J., M.H. Moayed, and R.C. Newman, Metastable pitting and the critical pitting temperature. Journal of the Electrochemical Society, (8): p Vetter, K.J., Initiation of Corrosion Pits and Its Theoretical Problems. Corrosion, (2): p Galvele, J.R., Transport Processes and Mechanism of Pitting of Metals. Journal of the Electrochemical Society, (4): p Tian, W., et al., Metastable pitting corrosion of 304 stainless steel in 3.5% NaCl solution. Corrosion Science, : p Park, J.O., S. Matsch, and H. Bohni, Effects of temperature and chloride concentration on pit initiation and early pit growth of stainless steel. Journal of the Electrochemical Society, (2): p. B34-B Zuo, Y., H.T. Wang, and J.P. Xiong, The aspect ratio of surface grooves and metastable pitting of stainless steel. Corrosion Science, (1): p Burstein, G.T. and S.P. Vines, Repetitive nucleation of corrosion pits on stainless steel and the effects of surface roughness. Journal of the Electrochemical Society, (12): p. B504-B Williams, D.E., T.F. Mohiuddin, and Y.Y. Zhu, Elucidation of a trigger mechanism for pitting corrosion of stainless steels using submicron resolution scanning electrochemical and photoelectrochemical microscopy. Journal of the Electrochemical Society, (8): p Stewart, J. and D.E. Williams, The Initiation of Pitting Corrosion on Austenitic Stainless- Steel - on the Role and Importance of Sulfide Inclusions. Corrosion Science, (3): p. 457-&. 87. Park, J.O. and H. Bohni, Local ph measurements during pitting corrosion at MnS inclusions on stainless steel. Electrochemical and Solid State Letters, (9): p Rosenfeld, I.L. and I.S. Danilov, Electrochemical aspects of pitting corrosion. Corrosion Science, (3): p Hoar, T.P. and W.R. Jacob, Breakdown of Passivity of Stainless Steel by Halide Ions. Nature, (5122): p &. 83

84 90. Pickering.Hw and Frankenthal.Rp, Mechanism of Localized Corrosion of Iron and Stainless-Steel I. Electrochemical Studies. Journal of the Electrochemical Society, (10): p &. 91. Isaacs, H.S., Behavior of Resistive Layers in Localized Corrosion of Stainless-Steel. Journal of the Electrochemical Society, (11): p Wilde, B.E. and E. Williams, Use of Current/Voltage Curves for Study of Localized Corrosion and Passivity Breakdown on Stainless Steels in Chloride Media. Electrochimica Acta, (11): p &. 93. Suzuki, T., M. Yamabe, and Y. Kitamura, Composition of Anolyte within Pit Anode of Austenitic Stainless-Steels in Chloride Solution. Corrosion, (1): p Mankowski, J. and Z. Szklarska-Smialowska, Studies on Accumulation of Chloride-Ions in Pits Growing during Anodic Polarization. Corrosion Science, (8): p Peterson, M.H. and T.J.L. JR., A Study of Cathodic Polarization and ph Changes in Metal Crevices. Corrosion, (10): p Sato, N., The Stability of Pitting Dissolution of Metals in Aqueous-Solution. Journal of the Electrochemical Society, (2): p Hisamatsu, Y., Passivity and Its Breakdown on Iron and Iron Base Alloys. NACE, Houston, 1976: p Kuo, H.C. and D. Landolt, Rotating-Disk Electrode Study of Anodic Dissolution or Iron in Concentrated Chloride Media. Electrochimica Acta, (5): p Hultquist, G., et al., The dissolution behaviour of iron, chromium, molybdenum and copper from pure metals and from ferritic stainless steels. Corrosion Science, (9): p Hakkarainen, T.J. Factors Determining the Dissolution Rate in Pitting Corrosion of Stainless Steels. in Materials Science Forum Trans Tech Publ Isaacs, H., et al., In Situ X Ray Microprobe Study of Salt Layers during Anodic Dissolution of Stainless Steel in Chloride Solution. Journal of the Electrochemical Society, (4): p Tester, J.W. and H.S. Isaacs, Diffusional Effects in Simulated Localized Corrosion. Journal of the Electrochemical Society, (11): p Williams, D., et al., Critical Factors in Localized Corrosion. ECS (4th edn), : p Pickering, H.W., The Role of Electrode Potential Distribution in Corrosion Processes. Materials Science and Engineering a-structural Materials Properties Microstructure and Processing, (1-2): p Abdulsalam, M.I. and H.W. Pickering, Effect of the applied potential on the potential and current distributions within crevices in pure nickel. Corrosion Science, (2): p Al Ameri, M., et al., Critical conditions for pit initiation and growth of austenitic stainless steels. Corrosion Science, (0): p González-Sánchez, J., et al., Corrosion pit growth on austenitic stainless steels in chloride containing solution: a quantitative approach. Anti-Corrosion Methods and Materials, (5): p Rao, K.S. and K.P. Rao, Pitting corrosion of heat-treatable aluminium alloys and welds: a review. Transactions of the Indian Institute of Metals, (6): p Rozenfeld, I. and I. Danilov, Behaviour of the stainless steel KH 18 N 10 T on galvanostatic anodic polarisation. Zashchita Metallov, (1): p Engell, H. and N.Z. Stolica, phys. Chem. NF, : p Forchhammer, P. and H.J. Engell, Untersuchungen über den Lochfraß an passiven austenitischen Chrom Nickel Stählen in neutralen Chloridlösungen. Materials and Corrosion, (1): p Schwenk, W., Theory of Stainless Steel Pitting. Corrosion, (4): p. T129-&. 84

85 113. Herbsleb, G. and H. Engell, Formation and growth of pits by corrosion on passivated iron electrodes. Z. physik. Chem, : p Frankel, G.S., The Growth of 2-D Pits in Thin-Film Aluminum. Corrosion Science, (12): p Strehblow, H.H. and J. Wenners, Investigation of the processes on iron and nickel electrodes at high corrosion current densities in solutions of high chloride content. Electrochimica Acta, (4): p Hunkeler, F. and H. Bohni, Mechanism of Pit Growth on Aluminum under Open Circuit Conditions. Corrosion, (10): p Newman, R.C. and E.M. Franz, Growth and Repassivation of Single Corrosion Pits in Stainless-Steel. Corrosion, (7): p Alkire, R.C. and K.P. Wong, The Corrosion of Single Pits on Stainless-Steel in Acidic Chloride Solution. Corrosion Science, (4): p. 411-& Szummer, A., Z. Szklarska-Smialowska, and M. Janik-Czachor, Electron microprobe study of the corrosion pits nucleation on Fe-16cr single crystals. Corrosion Science, (11): p. 827-IN Janik-Czachor, M. and Z. Szklarska-Smialowska, Pitting corrosion of single crystals of the Fe-16 Cr alloy in solutions containing Cl ions. Corrosion Science, (4): p Sato, N., et al., Generation and Propagation of Chloride-Pits on Rotating Stainless Steel Electrode in Acid Solution. Transactions of the Japan Institute of Metals, (2): p Carcea, A.G., et al., Anodic Kinetics of NiCr [Mo] Alloys During Localized Corrosion: I. Diffusion-Controlled Dissolution. Journal of the Electrochemical Society, (6): p. C215-C Moayed, M.H. and R.C. Newman, Deterioration in critical pitting temperature of 904L stainless steel by addition of sulfate ions. Corrosion Science, (11): p Tian, W.-M., et al., Pitting Kinetics of 304 Stainless Steel Using ESPI Detection Technique. Acta Metallurgica Sinica (English Letters), (4): p Tian, W., et al., Effects of applied potential on stable pitting of 304 stainless steel. Corrosion Science, (0): p Hisamatsu, Y., et al., Localized Corrosion. NACE, Houston, 1974: p Alkire, R., D. Ernsberger, and T.R. Beck, Occurrence of salt films during repassivation of newly generated metal surfaces. Journal of the Electrochemical Society, (9): p Beck, T.R. and R.C. Alkire, Occurrence of salt films during initiation and growth of corrosion pits. Journal of the Electrochemical Society, (10): p Beck, T.R., Formation of salt films during passivation of iron. Journal of the Electrochemical Society, (11): p Newman, R. and M. Ajjawi, A micro-electrode study of the nitrate effect on pitting of stainless steels. Corrosion Science, (12): p Rayment, T., et al., Characterisation of salt films on dissolving metal surfaces in artificial corrosion pits via in situ synchrotron X-ray diffraction. Electrochemistry Communications, (6): p Mankowski, J. and Z. Szklarskasmialowska, Effect of Specimen Position on Shape of Corrosion Pits in an Austenitic Stainless-Steel. Corrosion Science, (9): p Hisamatsu, Y. and K. Ichikawa, Etch-Pitting as a Faceting Dissolution Process. Anticorrosion technology, (3): p

86 134. Reiser, D.B. and R.C. Alkire, The measurement of shape change during early stages of corrosion pit growth. Corrosion Science, (7): p Toušek, J., The localized corrosion of Cr-Ni stainless steel. Corrosion Science, (1): p Tomashov, N., G. Chernova, and O. Markova, Mechanism of pit formation in steels 18 Cr-14Ni and 18Cr-14 Ni-2.5Mo. Prot Met, (2): p Szklarska-Smialowska, Z. and J. Mankowski, Effect of temperature on the kinetics of development of pits in stainless steel in 0 5n NaCl + 0 in H2SO4 solution. Corrosion Science, (12): p Sato, N., The Stability of Localized Corrosion. Corrosion Science, (12): p Williams, D.E., J. Stewart, and P.H. Balkwill, The Nucleation, Growth and Stability of Micropits in Stainless-Steel. Corrosion Science, (7): p Ernst, P., et al., The mechanism of lacy cover formation in pitting. Corrosion Science, (6): p Ernst, P., et al., Shapes of corrosion pits in metals. Proceedings of the International Symposium on Pits and Pores: Formation, Properties, and Significance for Advanced Luminescent Materials, (7): p Ernst, P. and R.C. Newman, Explanation of the effect of high chloride concentration on the critical pitting temperature of stainless steel. Corrosion Science, (9): p Sedriks, A., Effects of alloy composition and microstructure on the localized corrosion of stainless steels. Advances in Localized Corrosion, 1987: p Wijesinghe, T.L.S.L. and D.J. Blackwood, Real time pit initiation studies on stainless steels: The effect of sulphide inclusions. Corrosion Science, (4): p Newman, R.C., H.S. Isaacs, and B. Alman, Effects of Sulfur-Compounds on the Pitting Behavior of Type-304 Stainless-Steel in near-neutral Chloride Solutions. Corrosion, (5): p Eklund, G.S., Initiation of Pitting at Sulfide Inclusions in Stainless-Steel. Journal of the Electrochemical Society, (4): p Wranglen, G., Pitting and sulphide inclusions in steel. Corrosion Science, (5): p Bohni, H., T. Suter, and A. Schreyer, Microtechniques and Nanotechniques to Study Localized Corrosion. Electrochimica Acta, (10): p Lott, S.E. and R.C. Alkire, The Role of Sulfide Inclusions on Initiation of Crevice Corrosion of 304-Ss. Journal of the Electrochemical Society, (8): p. C355-C Brossia, C.S. and R.G. Kelly, Occluded solution chemistry control and the role of alloy sulfur on the initiation of crevice corrosion in type 304ss. Corrosion Science, (11): p Baker, M.A. and J.E. Castle, The Initiation of Pitting Corrosion at Mns Inclusions. Corrosion Science, (4): p Rossi, A., et al., XPS, AES and ToF-SIMS investigation of surface films and the role of inclusions on pitting corrosion in austenitic stainless steels. Surface and Interface Analysis, (7): p Suter, T. and H. Böhni, A new microelectrochemical method to study pit initiation on stainless steels. Electrochimica Acta, (20 22): p Noh, J.S., et al., Effects of nitric acid passivation on the pitting resistance of 316 stainless steel. Corrosion Science, (12): p Muto, I., Y. Izumiyama, and N. Hara, Microelectrochemical measurements of dissolution of MnS inclusions and morphological observation of metastable and stable 86

87 pitting on stainless steel. Journal of the Electrochemical Society, (8): p. C439- C Krawiec, H., et al., Dissolution of chromium-enriched inclusions and pitting corrosion of resulfurized stainless steels. Metallurgical and Materials Transactions a-physical Metallurgy and Materials Science, A(5): p Krawiec, H., et al., Influence of the dissolution of MnS inclusions under free corrosion and potentiostatic conditions on the composition of passive films and the electrochemical behaviour of stainless steels. Electrochimica Acta, (16): p Chiba, A., et al., A Microelectrochemical System for In Situ High-Resolution Optical Microscopy: Morphological Characteristics of Pitting at MnS Inclusion in Stainless Steel. Journal of the Electrochemical Society, (8): p. C341-C Webb, E.G., T. Suter, and R.C. Alkire, Microelectrochemical measurements of the dissolution of single MnS inclusions, and the prediction of the critical conditions for pit initiation on stainless steel. Journal of the Electrochemical Society, (5): p. B186-B Ke, R. and R. Alkire, Initiation of corrosion pits at inclusions on 304 stainless steel. Journal of the Electrochemical Society, (12): p Ke, R. and R. Alkire, Surface-Analysis of Corrosion Pits Initiated at Mns Inclusions in 304 Stainless-Steel. Journal of the Electrochemical Society, (6): p Muto, I., D. Ito, and N. Hara, Microelectrochemical investigation on pit initiation at sulfide and oxide inclusions in type 304 stainless steel. Journal of the Electrochemical Society, (2): p. C55-C Schmuki, P., et al., The composition of the boundary region of MnS inclusions in stainless steel and its relevance in triggering pitting corrosion. Corrosion Science, (5): p Ryan, M.P., et al., Why stainless steel corrodes. Nature, (6873): p Meng, Q., et al., High-resolution characterization of the region around manganese sulfide inclusions in stainless steel alloys. Corrosion, (4): p Zheng, S., et al., Mechanism of (Mg, Al, Ca)-oxide inclusion-induced pitting corrosion in 316L stainless steel exposed to sulphur environments containing chloride ion. Corrosion Science, (0): p Suter, T., et al., Pit initiation on stainless steels in 1 M NaCl with and without mechanical stress. Journal of the Electrochemical Society, (5): p. B174-B Wang, H. and E.-H. Han, Simulation of metastable corrosion pit development under mechanical stress. Electrochimica Acta, (0): p Shimahashi, N., et al., Effect of Applied Stress on Dissolution Morphology and Pit Initiation Behavior of MnS Inclusion in Stainless Steel. ECS Transactions, (31): p Horner, D.A., et al., Novel images of the evolution of stress corrosion cracks from corrosion pits. Corrosion Science, (11): p Turnbull, A., D.A. Horner, and B.J. Connolly, Challenges in modelling the evolution of stress corrosion cracks from pits. Engineering Fracture Mechanics, (5): p Turnbull, A., L.N. McCartney, and S. Zhou, Modelling of the evolution of stress corrosion cracks from corrosion pits. Scripta Materialia, (4): p Frankel, G.S., Pitting corrosion of metals; A summary of the critical factors. Proceedings of the International Symposium on Pits and Pores: Formation, Properties, and Significance for Advanced Luminescent Materials, (7): p

88 174. Laycock, N.J. and R.C. Newman, Temperature dependence of pitting potentials for austenitic stainless steels above their critical pitting temperature. corrosion science, (6): p Wang, J.H., C.C. Su, and Z. Szklarska-Smialowska, Effects of Cl-concentration and temperature on pitting of AISI 304 stainless steel. Corrosion, (10): p Maire, E., et al., On the application of X-ray microtomography in the field of materials science. Advanced Engineering Materials, (8): p Coléou, C., et al., Three-dimensional snow images by X-ray microtomography. Annals of glaciology, (1): p Buffiere, J.-Y., et al., Characterization of internal damage in a MMC p using X-ray synchrotron phase contrast microtomography. Acta Materialia, (5): p Ludwig, W., et al., Study of the interaction of a short fatigue crack with grain boundaries in a cast Al alloy using X-ray microtomography. Acta Materialia, (3): p Eckermann, F., et al., In situ monitoring of corrosion processes within the bulk of AlMgSi alloys using X-ray microtomography. Corrosion Science, (12): p Knight, S.P., et al., In situ X-ray tomography of intergranular corrosion of 2024 and 7050 aluminium alloys. Corrosion Science, (12): p Eckermann, F., et al., In Situ Microtomographically Monitored and Electrochemically Controlled Corrosion Initiation and Propagation in AlMgSi Alloy AA6016. Journal of the Electrochemical Society, (1): p. C1-C Burnett, T., et al., Correlative Tomography. Scientific reports,

89 3. EXPERIMENTAL PROCEDURES 3.1 Materials The materials used in this study were three conventional austenitic stainless steels: type 303 bar, 304 plate and 304L wire. The bar sample had a dimension of 15 mm 14 mm and the plate sample had a dimension of 13 mm 13 mm. The type 304L wire, with an average diameter of 500 µm, was supplied in a reel. The chemical compositions of the alloys are shown in Table 3-1. Table 3-1: Chemical composition (wt%) of austenitic stainless steel alloys used Alloys Cr Ni C Mn P S Si Fe Type 303 bar max 2 max 0.2 max 0.3 min 1 max Bal. Type 304 plate* Bal. Type 304L wire* Bal. * Manufacturer s chemical analysis protocol; Composition based on standard specification (BSI 303S31; ASTM UNS S30300). 3.2 Microstructure The microstructure of each as-received alloy was revealed by etch testing with 10% oxalic acid at 1 A.cm -2 for s as per Practice A of the ASTM A standard. Inclusions in alloy microstructure were also investigated using Scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDX). The samples were prepared as follows Sample preparation Samples of types 303 and 304, with a thickness of 4 mm thick, were cut from the bar and lengths of about 5 mm were cut from the type 304L wire. The samples were cold mounted inside acrylic moulds of 3 cm diameter using a slow setting epoxy (Araldite) with a 4:1 ratio of resin to hardener. The wire samples were mounted so that both crosssectional and longitudinal faces could be polished. After mounting, the samples were ground using a grinding machine with a speed-controlled wheel and water as lubricant. Progressively finer silicon carbide grinding papers were used: first 320, then 400, 800, 1200 and 4000 grit. The samples were next polished with 3 µm and 1 µm diamond 89

90 paste. After polishing, all samples were rinsed with ethanol and deionised water, then dried gently in a stream of hot air before etching with oxalic acid. In preparation for the SEM inclusion study, a finer surface finish was achieved by polishing the samples with ¼ µm diamond paste followed by an oxide polishing suspension containing colloidal silica. 3.3 Optical microscopy (OM) Optical microscopy was used to observe and to take micrograph images of the surface morphology of the samples. OM uses light reflected from the sample surface and its resolution is limited to the wavelength of the light. Therefore, it has a limited magnification. It is often used to observe the microstructure of alloys. In this study, a Zeiss optical microscope attached to a 5 megapixel digital camera and a computer running Axio Vision V4.81 image analysis software was used to take micrographs of etched microstructure and to observe pitting corrosion. All images were taken using bright field imaging mode, with 10x, 20x and 50x objective lenses. 3.4 ImageJ ImageJ software [1] was used to analyse micrograph images of inclusions in the alloys to determine their microstructure. All micrograph images were converted to 8-bit and scale bars converted to pixels using ImageJ. Next, based on contrast differences between inclusions and matrix, inclusions were selected by adjusting the threshold values. The images were then processed and inclusions counted, excluding those of area below 0.5 µm Scanning electron microscopy In SEM, a narrow electron beam is generated and accelerated under a voltage up to 30 kv. Therefore, the image resolution is higher than that obtained by OM. The interaction of the incident electron beam with the sample causes electrons to scatter and induces the emission of elastic and inelastic electrons from the sample. Two types of image can be collected from the sample, depending on the type of electrons emitted: secondary electron (SE) and backscattered electron (BSE) images. SE images are generated by low energy electrons emitted from the surface of the sample and reveal details of morphology and topography. BSE images are obtained from higher energy electrons, emitted from deeper levels. Therefore, the image contrast is obtained from differences 90

91 in the phase composition of the sample, relative to the difference in atomic number between elements. X-rays are emitted by the relaxation of excited electrons in the inner shell orbits of the atoms. The energy difference and intensity of alloy elements are used to define and estimate the elemental composition of the sample [2]. In this work an FEI Quanta 650 SEM, equipped with an energy dispersive X-ray detector (EDX) running Aztec software and supplied by Oxford Instruments, was used to study the microstructure and surface morphology of pitting corrosion under high magnification. Elemental analysis was performed using EDX analysis at an accelerating voltage of 20 kv and 10 mm working distance. 3.6 Electrochemical techniques Electrochemical measurements are quick techniques commonly used to study pitting corrosion under controlled environmental conditions because pitting corrosion is a slow process. In this study, an Ivium CompactStat Potentiostat instrument was used to measure the OCP and to carry out potentiodynamic and potentiostatic polarisation Sample preparation For electrochemical measurements, samples were cut into 6 mm 6 mm (L W), then copper wire was spot-welded onto the back of each sample and insulated with PVC tubing prior to moulding. A small amount of fast-cure Araldite, prepared by mixing equal amounts of epoxy and hardener, was used to cover the outer edge of the sample, then left for 3 hours to harden. This provided a good quick bonding to the samples in order to eliminate crevice corrosion during the electrochemical polarisation study. Next, the samples were placed in moulds and slow setting Araldite was added and left for 24 hours to cure. The sample surfaces were prepared by grinding with SiC paper up to 1200 grit, then cleaned with ethanol, rinsed using deionised water and dried in air. For type 304L wire, short sections of 70 mm were cut from the reel and the surface manually ground using 1200 grit SiC paper, followed by rinsing in deionised water. The wire surface was coated in a 3:1 mixture of beeswax and colophony, with a length of 1-2 cm left uncovered, and one end was connected to the potentiostat Open circuit potential The electrode potential was measured with respect to two reference electrodes: a saturated calomel electrode (SCE) and a mini-reference silver/silver chloride (Ag/AgCl, 91

92 3 M NaCl) electrode. To allow comparison of measurements in some results, conversions between the values of electrode potential were made. In this work, OCP measurements were taken before electrochemical polarisation for 5 or 15 minutes at room temperature (typically C) Electrochemical polarisation In this study, anodic polarisation from OCP was performed at a 1 mv.s -1 scan rate to different values of end potential above the pitting potential. E pit was defined by current response showing a continuous increase above 100 µa.cm -2. The current response was recorded during polarisation at a rate of 1 Hz. The aim of this test was to investigate the relationship between pit density and applied potential using the three stainless steels varying in inclusion contents and under different chloride concentrations. A set of end polarisation potentials between +200 and +600 mv vs. SCE were used and the obtained pits were counted through OM using a 100 magnification lens Electrochemical cells Electrochemical cells of two sizes were used as shown in Figure 3-1(a, b): a standard 500 ml round-bottomed flask cell and an in-house fabricated mini-electrochemical cell made from Perspex. The mini-cell had a length of 60 mm, inner diameter of 24 mm and approximate volume of 18 ml. The mini-electrochemical cell was used to polarise the wire samples to simulates the planed experiments. Electrochemical measurements were performed using a three-electrode set-up. In the standard cell, an SCE was used as reference electrode and platinum as a counter electrode. The mini-electrochemical cell had a miniature reference electrode (Ag/AgCl, 3 M NaCl) of 6 mm diameter and 2 cm length, and a miniature platinum counter electrode of 3 mm outer diameter, 1.6 mm inner diameter and 6 cm length. Both electrodes were manufactured by BAS Inc., Japan. The electrolyte concentrations used in the study were 0.1 and 1 M NaCl, prepared using deionised water. 92

93 Pt counter electrode Ag/AgCl reference electrode Working electrode Working electrode a Pt counter electrode Saturated calomel reference electrode b Figure 3-1: (a) Standard electrochemical cell; (b) mini-electrochemical cell. 3.7 X-ray tomography experiments X-ray CT is a non-destructive imaging technique, which can be used to characterise 3D features of the internal structure of bulk materials at micron and submicron resolution [3]. X-ray tomography is based on large numbers of radiography images collected from the object while it rotates through 360. These 2D images are then reconstructed to reveal the 3D shape of the object [3-5]. The reconstructed data can then be visualised and further analysis can be performed using software such as Avizo or ImageJ Principles of X-ray tomography X-ray radiography images are created by passing an X-ray beam through a sample and detecting areas where the beam is transmitted relative to the absorbed beam. The image contrast depends on beam energy and the attenuation coefficient along the beam relative to the material composition. A charge-coupled device (CCD) camera is used to record the image information from the object volume in 2D projection. This image provides information in one direction, then when the sample is rotated, more information can be obtained from a different direction. Multiple 2D images are collected, then 93

94 reconstructed using, for example, a filtered back-projection algorithm to obtain a 3D tomogram [3, 5] X-ray sources X-rays are generated either by a micro-focus X-ray tube or by synchrotron radiation. In an X-ray tube, the beam is emitted from a metallic target through a focused point. The beam is divergent or conical, as shown in Figure 3-2(a), and thus produces a magnified image on the CCD camera. A large divergent angle may create geometric artefacts and the field of view of the detector is a compromise between sample field of view and resolution. The scan time is also related to the size of the CCD camera, sample size and resolution. (a) (b) Figure 3-2: Two X-ray sources: (a) micro-focus source with 2D detector, showing cone-beam geometry and image magnification, (b) synchrotron source with parallel beam geometry [5]. In synchrotron radiation (Figure 3-2b), the distance between the source and the sample is very large and X-ray flux is high. A monochromatic beam can be created by placing a monochromatic filter between the source and the sample. Such a parallel and monochromatic beam gives exact magnification, with no beam hardening or geometric artefacts, thus allowing quantitative reconstruction. In X-ray CT, a scintillator is used to convert X-radiation into visible light projected onto the CCD camera [6]. The image contrast may be obtained by different modes such as the absorption contrast mode and the phase contrast mode. The absorption contrast mode is attained by varying the linear attenuation coefficients through the sample corresponding to the atomic number of the elements and the density. In this case, high 94

95 X-ray absorption leads to bad photon statistics, while high transmission reduces the contrast between features of the sample. Therefore, about 10% beam transmission is used to optimise these features of the reconstructed image. The phase contrast mode is attained by increasing the distance between the sample and the detector; when the X-ray beam is partially coherent, image contrast arises from interference. The difference in phase retardation after the X-ray beam allows small defects in the sample to be detected, but the image is sometimes difficult to segment [5] Sample preparation The type 304L stainless steel wire with a diameter of 500 μm was used in X-ray tomography experiments with a chemical composition shown in Table 3-1. Short wire sections of 70 mm were cut and manually ground using 1200 grit SiC paper, followed by rinsing in deionised water. The wire was mounted vertically in a minielectrochemical tensile cell in a bulk solution of 0.1 M NaCl to conduct an in-situ study of pitting corrosion. The surface of the wire was coated with beeswax by immersion, leaving an uncoated length in the range of mm, equivalent to an exposed surface area of mm 2. The exposed lengths was measured using micrometer ruler and magnifying glass. This small area of pitting corrosion was then observed under X-ray CT Electrochemical experiments In this work, a miniature electrochemical cell with straining rig shown in Figure 3-3(a) was designed for in situ tomography experiments of pitting corrosion in bulk solution. The inner diameter of the cell is 24 mm and the outer diameter and length are 30 mm. The electrochemical cell made from Perspex material consisted two parts, the lower part containing the electrochemical cell with a volume of 9 ml and the upper part the tensile cell, where unidirectional tensile strain could be applied. The working electrode wire was vertically mounted inside an electrochemical cell. The wire was clamped between two bolts and strain was applied to it by tightening a nut at the top of the cell. The length of working electrode wire was fixed at 50 mm before strain was applied. The elongation of the wire relative to its initial length was used to approximate the applied strain Using a digital calliper. A miniature reference electrode (Ag/AgCl, 3 M NaCl) and a miniature platinum counter electrode were used for polarisation measurement. The upper part containing a 95

96 straining rig used to apply a tensile load on the working electrode wire. The cell parts were shown in schematically in Figure 3-3(b). Straining part was design to investigate the effect of plastic strain on kinetics of pit growth, pit morphology, and for assessing the nucleation of stress corrosion cracking. Two steel nuts Pt counter electrode Load cell Ag/AgCl reference electrode Perspex cell Counter electrode X-ray Source Working electrode In-situ cell electrolyte Beeswax coating Working electrode Reference electrode Aluminium holder Bolt (a) Figure 3-3: (a) In-situ miniature electrochemical cell with straining rig, (b) schematic diagram of electrochemical cell parts. Electrochemical polarisation tests were performed at a scan rate of 1 mv.s -1 at room temperature with an Ivium CompactStat Potentiostat and after each polarisation scan or applied strain an X-ray CT scan was performed to observe pit formation. OCP was measured for 15 min before the first polarisation and for 5 min before each subsequent cycle of polarisation. During polarisation experiments, the current response was recorded at a rate of 1 Hz. Electrochemical polarisation was performed to grow corrosion pits and X-ray tomography was used to obtain 3D data on pitting corrosion. Cycles of repolarisation were conducted without removing the sample from the solution, either by repeating the scan with a similar polarisation potentials, with and without increasing the time of polarisation, or by increasing the polarisation potential. These steps were also repeated with and without applied strain. These procedures are also described in more detail in each of the fallowing manuscripts. The tests were conducted to assess the effects of repolarisation potential on the kinetics of 3D pitting (b) 96

97 corrosion and pit reactivation morphology. The synergetic effect of repolarisation after the application of plastic strain was also investigated to examine the effect of strain on pitting corrosion kinetics, pit reactivation morphology and crack nucleation X-ray CT Set-up X-ray CT measurements were carried out using a Zeiss Xradia 400 microtomography instrument at an accelerating voltage of 120 kv, and 721 projections were recorded in 2 2 binning mode with an exposure time of 10 seconds. This resulted in a reconstructed voxel size of 1.83 µm 3, with a 2300 µm 2300 μm field of view. Other parts of the measurement were performed using a Zeiss Xradia 520 Versa microtomography instrument with an accelerating voltage between 110 and 130 kv, and 721 projections were recorded in 2 2 binning mode at 10 optical magnification, with an exposure time of 12 s for each projection. This resulted in a reconstructed voxel size of 1.1 µm 3 with a field of view of 1100 µm 1100 µm. Each scan took about 2-3 hours using above setup. The data were reconstructed using the Feldkamp-Davis-Kress (FDK) approach [7] and the Xradia system was used to reconstruct the virtual slices by adjustment of centre shift and beam hardening to obtain the best signal-to-noise ratio. Images were segmented and visualised using the Avizo 9 software and a median filter was applied for data smoothing. The image processing may have reduced the theoretical resolution by ~2 μm. After segmentation of each pit, measurements were made of pit volume and pit surface area. The depth, width and height of pits were obtained by selecting the middle 2D slices of the pit with respect to the pit edges. These slices represent the pit centre with the maximum pit depth. This approach provided a snapshot of pit dimensions after each polarisation cycle, with the current response recorded during the polarisation experiment providing in-situ pit growth information. 97

98 3.8 References 1. ImageJ. [cited th October]. 2. Goldstein, J., et al., Scanning electron microscopy and X-ray microanalysis: a text for biologists, materials scientists, and geologists2012: Springer Science & Business Media. 3. Maire, E., et al., On the application of X-ray microtomography in the field of materials science. Advanced Engineering Materials, (8): p Stock, S.R., Recent advances in X-ray microtomography applied to materials. International Materials Reviews, (3): p Salvo, L., et al., X-ray micro-tomography an attractive characterisation technique in materials science. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, (0): p Withers, P.J., X-ray nanotomography. Materials Today, (12): p Feldkamp, L., L. Davis, and J. Kress, Practical cone-beam algorithm. JOSA A, (6): p

99 4. RESULTS - OVERVIEW The results reported in this PhD thesis provide new insight into pitting corrosion in austenitic stainless steels exposed to chloride-containing environments. These results are set out in five manuscripts (an alternative thesis format), which are aligned to three key research directions, presented below in chapters 5, 6 and 7. Chapter 5 addresses the effect of electrochemical potential and microstructural inclusion density on pitting corrosion susceptibility. The aim of the work reported in this chapter was to obtain information about the relationships among microstructural inclusions density, pitting susceptibility, and exposure parameters. The corrosion behaviour of three austenitic stainless steels were explored, to obtain confidence in moving to the next stage of the project, to carry out in-situ pitting corrosion studies using X-ray CT. Chapter 6 describes the construction of an in-situ electro-chemical cell for X-ray CT studies, to provide an experimental methodology to estimate 3D pitting corrosion kinetics. It reports the nucleation and growth of a single pit, two pits in parallel, and multiple pits. The results obtained are compared to the literature on 1D and 2D pitting corrosion studies, as well as 3D investigations involving 3D video microscope [1-4]. Chapter 7 discusses experiments to elucidate the effect of strain on 3D pitting corrosion kinetics. Characterisation was carried out using the same set-up as in the previous chapter, with a focus on 3D pit shape and re-passivation kinetics. Figure 4-1 gives an overview of all manuscripts, with Chapter 5 based on one manuscript, and Chapter 6 and 7 on two manuscripts each. Part of this work was also presented as a poster and a conference presentation: i. Poster at 2015 Faraday Discussions, London, with the title On the Application of X-ray Computed Tomography to Investigate Re-activation Behaviour of Pitting Corrosion in Austenitic Stainless Steel. ii. Presentation at Electrochem 2015, Durham, with the title Determination of 3D Pitting Corrosion Kinetics in Stainless Steel. 99

100 Chapter 5: Effect of Microstructure and Inclusion on Pitting Corrosion Susceptibility Manuscript (A): On the relationship between pitting corrosion and microstructural inclusion density in austenitic stainless steels. (not submitted yet) TITLE Characterisation of 3D Pitting Corrosion Kinetics of Stainless Steel in Chloride-Containing Environments Chapter 6: Observation and Estimation of 3D Pitting Corrosion Kinetics Manuscript (B): Application of a Quasi In-situ Experimental Approach to Estimate 3-D Pitting Corrosion Kinetics in Stainless Steel. Published paper in Journal of the Electrochemical Society, 2016, DOI: / jes Manuscript (C): Estimation of 3D Growth Kinetics of Multiple Pits in Austenitic Stainless Steel. (not submitted yet) Chapter 7: Effect of Strain on 3D Pitting Corrosion Kinetics Manuscript (D): Strain-induced Re-activation of Corrosion Pits in Austenitic Stainless Steel. Submitted to the Corrosion Science. Manuscript (E): Time-dependent Re-activation of Corrosion Pits in Austenitic Stainless Steel. (not submitted yet) Figure 4-1: Schematic overview of all manuscripts included in this results section. 100

101 4.1 References 1. Laycock, N.J. and R.C. Newman, Localised dissolution kinetics, salt films and pitting potentials. Corrosion Science, (10-11): p Ghahari, M., et al., Synchrotron X-ray radiography studies of pitting corrosion of stainless steel: Extraction of pit propagation parameters. Corrosion Science, : p Tian, W., et al., Effects of applied potential on stable pitting of 304 stainless steel. Corrosion Science, (0): p Alkire, R.C. and K.P. Wong, The Corrosion of Single Pits on Stainless-Steel in Acidic Chloride Solution. Corrosion Science, (4): p. 411-&. 101

102 5. EFFECT OF MICROSTRUCTURE AND INCLUSIONS ON PITTING CORROSION SUSCEPTIBILITY The susceptibility to pitting corrosion of three different stainless steel microstructures has been investigated using electrochemical polarisation tests (Manuscript A, section 5.2). The work presented in this paper focuses on the relationship between applied electrochemical potential and pitting corrosion susceptibility, as a function of microstructural different inclusion densities. These inclusions are chemically inhomogeneous and known to act as preferential sites for pitting corrosion to initiate. Type 303, type 304 and type 304L were investigated and the relationship between pit density and inclusion density obtained and compared to the chemical compositions of these steels. The influence of chloride concentration was also investigated. 5.1 Key findings and results - A relationship between microstructural inclusion density and susceptibility to pitting corrosion was observed. - An experimental methodology for carrying out in-situ pit nucleation and growth studies was explored. 102

103 5.2 On the relationship between pitting corrosion and microstructural inclusion density in austenitic stainless steels (Manuscript A) F. A. Almuaili 1 & D. L. Engelberg 1 1 Corrosion and Protection Centre, School of Materials, The University of Manchester, Sackville Street, Manchester, M13 9PL, United Kingdom Abstract The pitting corrosion susceptibility of types 304, 304L and 303 austenitic stainless steels has been investigated. The alloy microstructures revealed the presence of different inclusion densities with variations in their chemical composition in all microstructures investigated. The pitting response was a function of the inclusion content, with the density of pits and the pitting potential related to inclusion density. Higher applied electrochemical potentials showed an exponential relationship with pit density, and increasing the chloride concentration resulted in a shift of the critical pitting potential for all microstructures investigated. Keywords: Pitting Corrosion, Inclusion, Potentio-dynamic Polarisation, Stainless Steels Introduction Austenitic stainless steels are widely used for their good corrosion resistance in corrosive environments. They owe this property to the protective effect of the surface oxide layer, a few nanometres thick, which forms spontaneously in air. However, the integrity of this passive film is limited in chloride-containing environments by the occurrence of localised forms of attack, such as pitting and crevice corrosion. Localised attack is well known to be related to inclusion and initiated by the breakdown of the passive film, leading to pit nucleation. Pits tend to propagate once they achieve a certain internal stability [1, 2], but the random characteristics of pitting corrosion make it difficult to determine when or where they will initiate [3]. Pit nucleation starts with the breakdown of the passive film, which is often related to the inhomogeneous chemistry of the microstructure below the oxide film, and may lead to the growth of metastable or stable pits [1, 4]. A number of factors have been 103

104 investigated and found to reduce the resistance of the passive film, related to the environment, microstructure or alloy composition. Environments containing halide ions play an important role in accelerating pitting attack. For instance, microstructure sensitisation by inappropriate heat treatment or welding induces chromium depletion at the grain boundary and leads to corrosion initiation. In highly alloyed grades of austenitic stainless steel, pitting corrosion is induced only if the critical pitting temperature (CPT) is reached, even where chloride concentration is high [5-8]. In austenitic stainless steel, minor elements such as sulphur are added to improve machinability; however, other types of microstructural inclusions such as oxides, sulphides and silicates may form during solidification [9]. The thermodynamic instability of such inclusions [10] may influence the corrosion resistance of stainless steel. For instance, non-metallic inclusions are well known as preferential sites for pitting corrosion due to associated reduced passive film stability [11]. It is suggested that micro-crevices initiate at the matrix/inclusion interface due to chemical differences as well as inclusion dissolution [12-14]. By-products released from dissolved inclusions are also suggested to support pit propagation [15-17]. The pitting potential of stainless steel is strongly related to the types of inclusions present. In addition, inclusion size and composition were found to play important roles in pit initiation. Inclusions with dimensions in excess of 0.7 μm were reported to induce stable pit formation in chloride solutions [15]. Manganese sulphide inclusions in type 303 were reported to dissolve at a potential range of V Ag/AgCl in Na 2 SO 4 and V Ag/AgCl in 0.1 M NaCl. The synergistic effects of corrosion by-products of dissolved inclusions and chloride ions were suggested to influence the initiation of pitting corrosion [18]. The quantity of sulphur released was found to be related to sulphide concentration inside inclusions on types 303 and 304 stainless steel [16]. The frequency of metastable pit events was found to decrease with time in 304L stainless steel at constant potential and chloride concentration, due to the exhaustion of pit initiation sites [19]. In addition, the number of metastable events increased as a function of applied potential in chloride solution within a current density range of A.cm -2 [20]. However, the stable pit current density was found in previous work to be limited to the range of 4-5 A.cm -2 [21]. 104

105 The aim of the present work is to study the relationship between pitting corrosion susceptibility and inclusion density in three conventional stainless steels: types 303, 304 and 304L. Electrochemical polarisation measurements were made using potentiodynamic polarisation in different chloride concentrations, and the relationship between pit density and inclusion density of alloy microstructures correlated with microstructure susceptibility of stainless steel to pitting corrosion. An understanding of the relationship between alloy microstructure and impurity elements concentration, in particular the inclusion content, may help to explain the susceptibility of stainless steel to pitting corrosion attacks Experimental Materials: Table 5-1 lists the chemical composition of the type 303 bar, type 304 plate and type 304L wire used in this study. The sulphur content of type 303 was two orders of magnitude higher than that of types 304 and 304L. Samples of 303 bar and 304 plate were cut from supplied bar and plate, while the wire samples were cut from a reel, and all samples were used in the as-received (solution annealed) condition. The type 304 and type 303 samples measured respectively mm and mm (L W T), where L is parallel to the rolling/drawing direction. The type 304L wire, 500 μm in diameter, was cut into lengths of about 50 mm, of which about 10 mm was exposed to the electrolyte. The two edges were coated with beeswax, except a small part at one edge, which was used to connect the sample to a Potentiostat. The samples of types 304 and 303 were embedded in epoxy resin with a wire connected to the back of each. Prior to exposure to the solution, all specimen surfaces were ground to 1200 grit using SiC papers, rinsed in distilled water and dried in air. The time between the polishing and the starting measurement was typically less than 15 minutes. Table 5-1: Chemical composition (wt%) of austenitic stainless steel alloys used. Alloys Cr Ni C Mn P S Si Fe Type 303 bar max 2 max 0.2 max 0.15 min 1 max Bal. Type 304 plate* Bal. Type 304L wire* Bal. * Manufacturer s analysis; Composition based on standard specification (BSI 303S31; ASTM UNS S30300) 105

106 Microstructure and SEM/EDX: The samples were prepared by grinding and polishing to 1 μm diamond paste finish. The microstructure of each alloy was revealed by electrolytic etching with 10 wt% oxalic acid, using a current density of ca. 1 A.cm -2 for second. Etching was carried out in a small glass beaker with a stainless steel counter electrode, following practice A of the ASTM A standard [22]. The mean grain size of all three samples was determined by the mean linear intercept procedure [23], with the error estimated by calculating the standard deviation between five measurements. Images of the etched microstructures were taken using a Zeiss optical microscope attached to a 5 megapixel digital camera and a computer running Axio Vision V4.81 image analysis software. Inclusions on the surface of samples were observed by scanning electron microscopy (SEM). Samples were finely ground to 4000 grit, followed by diamond paste polishing to 0.25 μm, and a oxide polishing suspension (OPS) of colloidal silica solution (without etching). Backscattered images were taken using an FEI Quanta 650 SEM equipped with energy dispersive X-ray (EDX) spectroscopy and Aztec analysis software. EDX analysis was performed using line scan profiles across inclusions identified in SEM micrographs. Inclusion densities were determined by grey-value and area size thresholding of SEM images, analysed using ImageJ software. The SEM images with pixel resolution were taken between 1000 and 4000 magnification. The image scale bar was used to set image size using ImageJ (Analysis> set scale bar), then images were converted to 8 bit using (Image.> Type> 8-bit). Inclusions were adjusted according to the threshold values with image resolution of 0.15 μm/pixel, then selected and processed using (Process> Binary> Fill holes). Images were cropped to all areas of particles, then the particles were counted using (Analyze>Analyze particles). The inclusion density was obtained by dividing the number of inclusions by the image area. The inclusion density was determined for inclusions larger than 0.5 μm 2, with smaller areas not considered in the count. Electrochemical measurement: Electrochemical measurements were made using a three-electrode cell. Samples of 303 bar and 304 plate were measured in a standard 0.5 L cell using an SCE reference electrode and a platinum counter electrode. Measurements of type 304L wire were made in a small electrochemical cell of 18 ml 106

107 volume using a miniature reference electrode (Ag/AgCl, 3 M NaCl) and a miniature platinum counter electrode. The miniature reference electrode (Ag/AgCl, 3 M NaCl) has an offset potential of 42 mv vs. SCE. An Ivium CompactStat potentiostat and IviumSoft acquisition software were used to take measurements in 0.1 M NaCl and 1 M NaCl solution, prepared using reagent grade NaCl and deionised water at ambient temperature. The current response was recorded during polarisation at a rate of 1 Hz. The open circuit potential (OCP) of all specimens was recorded for 15 min prior to a potentiodynamic each polarisation sweep. The sweep was carried out from OCP towards the anodic direction at a scan rate of 1 mv.s -1 to end potentials vs Ag/AgCl (+350,+400,+450,+500,+550 and +600 mv) for 304L and for 303 and 304 vs SCE (+100,+200,+300,+400,+500 and +600 mv). After the end potential was reached, the samples were removed from the electrolyte, rinsed in deionised water, and dried before the pits were counted using OM at100x magnifications Results and Discussion Metallography and microstructure Figure 5-1(a, b, c) shows the etched microstructure of cross section samples of types 303 bar, 304 plate and longitudinal section of 304L wire, respectively. The samples had mean grain sizes of about 15±2 μm, 35±4 μm and 25±3 μm for types 303, 304 and 304L, respectively. The oxalic acid etch is a very severe etching method, and attacks most microstructural features [24]. The images of type 303 bar indicate an austenitic microstructure with numerous annealing twin grain boundaries, and a large number of precipitates or inclusions, distributed inside grains and at grain boundaries. The 304 plate sample has fewer etch features, sparsely distributed at grain. Some intergranular etch features are also apparent. The 304L wire has a typical fine austenitic structure (Figure 5-1c). 107

108 (a) (b) (c) Figure 5-1: Micrographs of the microstructure of (a) type 303 bar, (b) type 304 plate and (c) type 304L wire after electroetching with 10% oxalic acid. 108

109 SEM and EDX analysis of inclusions Microstructural examination of type 303 bar is shown Figure 5-2(a, b), also summarising the inclusion density. A large numbers of inclusions of different sizes and irregular shapes were observed in type 303, with the size distribution calculated using ImageJ, revealing inclusions ranging in size from ~1-4 μm 2, with the majority below 1 μm 2, however one inclusion with size reached 14 μm 2 was found. The chemical composition of different inclusions (about 10 inclusions) was determined by EDX profile analysis, with a typical profile shown in Figure 5-3. The line profiles show that the inclusion in type 303 bar contained mainly S and Mn, indicating that MnS was the dominant compound. The type 304 plate is shown Figure 5-4(a, b), indicating fewer inclusions compared to type 303 bar. The inclusion density was 14 times below that of 303 bar, with inclusions ranging in size from ~1-5 μm 2 ; however, a few inclusions reached 8 μm 2. The type 304 profile typically shows no S peak and this may indicate a mixed-phase inclusion of elemental oxides, mainly of Cr and Mn. However, a small S peak was observed in a few inclusions (not shown). Dissolved Sulphur promotes the formation of aggressive species, such as thiosulphates and/or hydrogen sulphide. These are then believed to facilitate dissolution reactions. The type 304L wire is shown in Figure 5-5(a, b), indicating fewer inclusions compared to 304 plate, with inclusion density half of that of 304 plate, but with inclusions of approximately the same size. However, one elongated inclusion was found to reach 23 μm 2. The type 304L profile shows a mainly oxide phase inclusion, but with less Cr than matrix relative to the 304 sample, which had high Cr content in its inclusions. The type 303 material has far more inclusions relative to types 304 and 304L. The size of inclusions also depends on the sample process orientation. The samples process orientation was the rolling directions for plate/bar material, and the drawing direction for the wire samples. Differences in chemistry, size and density of inclusions can affect the susceptibility of stainless steel microstructures to pitting corrosion. Some inclusions have been found to be more electrochemically active than others at certain potentials [25-30]. The relationship of applied electrochemical potential to the frequency of metastable pit events on stainless steel has already been investigated, and individual current events have been related to the nucleation of metastable pits [31]. The approach 109

110 chosen in our work looks at the overall microstructure susceptibility, by comparing the total number of inclusions to corresponding number of attacked sites after electrochemical testing. (a) (b) Figure 5-2: Analysis of type 303 microstructure, with (a) summary of inclusion count versus size of inclusions, and (b) showing a backscattered micrograph(unetched) used for image analysis. 110

111 Figure 5-3: A typical inclusion in type 303 with corresponding EDX line profile. (a) 111

112 (b) Figure 5-4: Type 304 plate microstructure (a) showing a backscattered micrograph (unetched) used for image analysis and (b) a typical inclusion in type 304 with corresponding EDX line profile. (a) 112

113 (b) Figure 5-5: Type 304L wire microstructure (a) showing a backscattered micrograph (unetched) used for image analysis and (b) a typical inclusion in type 304L with corresponding EDX line profile Electrochemical measurement Figure 5-6 shows the OCP of the three materials immersed in 0.1 and 1 M NaCl solution, with the values measured after the specimens had been exposed to the solution for 15 minutes. This time was set for all in situ X-ray computed tomography experiments in the next chapters. The results show that type 304L had a slightly higher potential than types 303 and 304. The differences in OCP may be a consequence of the nature of the oxide film developed at the sample surface relative to the time to reach OCP stabilisation, which depends on the composition of the sample and the environmental conditions to which it was exposed. The variation in composition of inclusions also plays a role in the stability differences and fluctuations in OCP between the alloys. It may influence in particular the local passivity of the oxide film. 113

114 Figure 5-6: Electrochemical properties of type 303 bar, type 304 plate and type 304L wire in 0.1 M NaCl solution summary of OCP values after 15 min and pitting potential (E pit ) after potentiodynamic polarisation at a scan rate of 1 mv.s -1. With longer immersion, the OCP (Figure 5-7) was found to increase gradually over time in the anodic direction, indicating passive film growth up to a stable potential range, characterised by a slight increase in potential over time. This observation is consistent with published results for type 304 [32]. However, OCP in the stable range showed a number of potential fluctuations in type 303 bar and type 304 plate, while the 304L wire showing only a few sudden small drops in potential. The fluctuations in potential may be ascribed to the breakdown and repair of the passive layer on the sample surface over time or metastable pits. This observation may also indicate that the protection afforded by the passive layer is reduced over time, probably by chloride ions diffusing through the film, particularly at the site of defects or weak points, such as above inclusions or at the interface between inclusions and matrix, which is also related to the inclusion density. The diffusing of chloride ions through oxide film is governed by the film potential and induction time. This difference may indicate a more active dissolution of inclusions in the passive potential range of type 303, relative to inclusions in other alloys. This is similar to previously reported results which show active dissolution and 114

115 pitting potential in the range of +100 mv to +200 mv vs. Ag/AgCl in 0.1 M NaCl [18]. De Wit et al. [33] investigated pitting corrosion on stainless steel in 0.1 M NaCl solution and reported that small inclusion dissolution was not necessarily associated with current spikes. Figure 5-7: OCP over time of type 303 bar, type 304 plate and type 304L wire in 0.1 M NaCl solution. Figure 5-6 also shows the pitting potential obtained by applying a potentiodynamic polarisation test at a scan rate of 1 mv.s -1 from OCP. The pitting potential was taken when a sudden and continuous increase in the passive current was recorded, reaching current densities of up to 100 µa.cm -2. Typical polarisation curves are shown in appendix A. The results indicate that pitting potential reduced in the order 304L>304>303. This observation is consistent with the increasing inclusion density values reported in the previous section. It indicates that increasing inclusion content, size or different composition in type 303 relative to type 304L may contribute to the dissolution of inclusions and pit activation as polarisation potential increases. This is consistent with previous observations [14, 15, 26, 34]. Inclusion dissolution with increasing potential is also related to chemical composition and size of inclusions, which was found to play an important role in reducing pitting potential and enhancing the activation of inclusion dissolution, leading to attack at the interface between 115

116 inclusion and matrix [4, 11, 15, 35, 36]. In a previous study [37], both accumulated charge and scan rate were found to influence the pitting potential under potentiodynamic polarisation. The difference between pitting and corrosion potential (E pit -E cor ) has often been used to indicate the stability range in different corrosive environments and to compare materials. In 0.1 M NaCl, the value of E pit -E cor was lowest for type 303, at 370 mv, compared with 520 mv and 548 mv for types 304 and 304L respectively. The results of this study show that passivity range increased with a reduction of inclusion content in the alloys and that the sample with the highest inclusion density (type 303 bar) had the lowest passive range. This shows that reduction of inclusion contents increased the pitting resistance and the stability of the oxide passive film of conventional stainless steel. This difference can be attributed to the influence of inclusion contents, which act as sites of weakness under the passive film [38]. Therefore, the lower pitting potential of type 303 compared to type 304 can be attributed to a higher inclusion content, as it contains more sulphur [25]. The result may indicate that especially large MnS inclusions were responsible for lower pitting resistance [30]. Another study attributed the pitting of types 303 and 304 to the dissolution of sulphide inclusions in the range of 1 to 20 μm [39], whereas an earlier study associated inclusions above 0.7 µm with pitting corrosion in type 304 [36] Effect of chloride concentration Increasing the chloride concentration from a moderate value of 0.1 M to 1 M reduced the pitting potential at room temperature. The results (Figure 5-6) show total reductions of 20, 66 and 95 mv for types 303, 304 an 304L, respectively, suggesting no direct relation to inclusion densities. This may indicate that the dissolution of a large number of inclusions in type 303 was activated at a lower chloride concentrations, while increasing chloride concentration had more effect on the activation of different types of inclusions in the other microstructures. This difference can be attributed to the types of inclusion and their composition observed in the analysis of EDX profiles, which appears to play an important role for the pitting potential. Non-metallic inclusions were reported to be less stable than oxide and mixed-phase inclusions, the dissolution of MnS inclusions was reported to occur at OCP and the dissolution rate appears to vary with chloride concentration [25, 28, 32]. 116

117 Furthermore, the protective effect of the passive film is a function of temperature and chloride concentration; it has been proposed that increasing the chloride concentration changes the oxide film properties by increasing the charge transfer, which reduces the resistance of the oxide film [40]. Ernst and Newman [41] relate the effects of both chloride concentration and CPT on pitting potential to critical pit chemistry, specifically the critical concentration of dissolved cations which is necessary to maintain active dissolution. MnS dissolves in both chloride and non-chloride solutions and pits are believed to initiate in chloride solution at the interface between inclusion and matrix. Higher chloride concentrations were found to initiate pits at this interface by increasing metal dissolution and hindering metastable pit passivation, which is also related to inclusion size [15, 42, 43]. Recently, Chiba et al.[44] demonstrated the importance of interstitial carbon atoms in improving pitting potential, which could be another explanation for the slightly lower pitting potential of type 303 relative to types 304 and 304L with increasing chloride concentration. The role of interstitial carbon atoms hinders dissolution of Fe atoms; so C is considered to increase pitting resistance Effect of applied potential on pit density Figure 5-8(a, b) show the effect of chloride concentration on pit density as a function of end potential. Potentio-dynamic scans were carried out from OCP to different end potentials in 0.1 M NaCl and 1 M NaCl, and the number of attacked sites obtained by image analysis. The results indicate that pitting density increased with increasing polarisation potential for type 303 bar and 304 plate, while only a slight increase was observed for type 304L wire in 0.1 M NaCl solution. Therefore, inclusions can lead to increasing pit density in all microstructures, but at different ratios depending on inclusion type, size and density. The pitting potential reported in the previous section showed that type 304L wire had a more noble pitting potential than type 303 bar. This difference could be related to the MnS inclusions which were dominant in 303 relative to 304L, as indicated by the EDX line profile analysis in Figure 5-3, Figure 5-4 and Figure 5-5. Micro-electrode studies suggested a relationship between the pitting potential, MnS, and the applied stress [45]. It was also reported that beside a reduction in pitting corrosion, the linear relation between chloride concentration and corrosion rate was associated with an increase in the number of pits [46]. 117

118 (a) (b) Figure 5-8: Pit density of type 303 bar, type 304 plate and type 304L wire at various end potentials of the potentiodynamic polarisation scan at 1 mv.s -1 scan rate in (a) 0.1 M NaCl solution and (b) 1 M NaCl solution. 118

119 Increasing the chloride content from 0.1 M NaCl to 1 M NaCl shifted all pitting potential more negative, leading to an increase in inclusion dissolution, possibly the activation of more inclusion sites, and a subsequent increase in the pit density of all materials. The number of pits is related to applied potential, chloride concentrations, and the inclusions type, composition and size. This observation indicates that increasing both chloride concentration and applied potential play important roles in pit formation, a result which is consistent with previously reported data [8, 46]. In addition, lower potential dissolution of inclusions was suggested to reduce the activation energy of dissolution and inhibit reformation of alloy passive film [47]. In contrast, inclusion density showed a strong relationship with pit density. Dissolution sites may turn to stable pits, depending on applied potential and environmental conditions. Figure 5-9 shows the relationship between inclusion densities and pit density at the end polarisation potentials of +400 mv in 0.1 M NaCl and +300 mv in 1 M NaCl. The results indicate that pitting density increased with increasing inclusion density. For type 304L wire, the applied potential of +300 mv resulted in few nucleation of pits. At the higher chloride concentration, pit density in types 303 and 304 was higher than at the lower chloride concentration, probably due to more inclusion sites being activated by anodic polarisation, leading to stable pit growth. However, pit density increased by 47% in type 303 and only by 22% in type 304 with exposure to higher chloride concentrations, implying that the influence of inclusion content on the severity of stainless steel attack depends on both, the type of inclusion and the corresponding environmental exposure conditions. 119

120 (a) (b) Figure 5-9: Relationship between pit density and inclusion density in type 303 bar, type 304 plate and type 304L wire in (a) 0.1 M at +400 mv vs. SCE and (b) 1 M NaCl solution at+300 mv vs. SCE. 120

121 5.2.5 Conclusions 1- Different inclusion densities and variation of their chemical compositions were observed in all investigated microstructures. 2- Higher chloride concentrations reduced the pitting corrosion resistance, resulting in more pit nucleation sites. 3- The pitting response was a function of the inclusion content, but the relationship between pit density and inclusion density varied with type of stainless steel. 4- Higher applied electro-chemical potentials showed an exponential relationship with pit density Acknowledgements The authors would like to thank the Saline Water Conversion Corporation, Saudi Arabia for financial support and acknowledge the use of Image J software References 1. Frankel, G., Pitting corrosion of metals a review of the critical factors. Journal of the Electrochemical Society, (6): p Soltis, J., Passivity breakdown, pit initiation and propagation of pits in metallic materials - Review. Corrosion Science, : p Newman, R.C., 2001 W.R. Whitney Award Lecture: Understanding the Corrosion of Stainless Steel. Corrosion, (12): p Williams, D.E., et al., Composition changes around sulphide inclusions in stainless steels, and implications for the initiation of pitting corrosion. Corrosion Science, (11): p González-Sánchez, J., et al., Corrosion pit growth on austenitic stainless steels in chloride containing solution: a quantitative approach. Anti-Corrosion Methods and Materials, (5): p Gnanamoorthy, J.B., Electrochemical investigations for understanding and controlling corrosion in nuclear reactor materials. Transactions of the Indian Institute of Metals, (5): p Ernst, P. and R.C. Newman, Pit growth studies in stainless steel foils. II. Effect of temperature, chloride concentration and sulphate addition. Corrosion Science, (5): p Tian, W., et al., Effects of applied potential on stable pitting of 304 stainless steel. Corrosion Science, (0): p Lo, K.H., C.H. Shek, and J.K.L. Lai, Recent developments in stainless steels. Materials Science and Engineering: R: Reports, (4 6): p Eklund, G.S., Initiation of Pitting at Sulfide Inclusions in Stainless-Steel. Journal of the Electrochemical Society, (4): p

122 11. Szklarska śmialowska, Z. and E. Lunarska, The effect of sulfide inclusions on the susceptibility of steels to pitting, stress corrosion cracking and hydrogen embrittlement. Materials and Corrosion, (11): p Lott, S.E. and R.C. Alkire, The Role of Sulfide Inclusions on Initiation of Crevice Corrosion of 304-Ss. Journal of the Electrochemical Society, (8): p. C355-C Webb, E.G., et al., Experimental and theoretical studies of pit initiation at single MnS inclusions in stainless steels. Passivity and Localized Corrosion, (27): p Webb, E.G., T. Suter, and R.C. Alkire, Microelectrochemical measurements of the dissolution of single MnS inclusions, and the prediction of the critical conditions for pit initiation on stainless steel. Journal of the Electrochemical Society, (5): p. B186-B Ke, R. and R. Alkire, Surface-Analysis of Corrosion Pits Initiated at Mns Inclusions in 304 Stainless-Steel. Journal of the Electrochemical Society, (6): p Paik, C.H., H.S. White, and R.C. Alkire, Scanning electrochemical microscopy detection of dissolved sulfur species from inclusions in stainless steel. Journal of the Electrochemical Society, (11): p Webb, E.G. and R.C. Alkire, Pit initiation at single sulfide inclusions in stainless steel - II. Detection of local ph, sulfide, and thiosulfate. Journal of the Electrochemical Society, (6): p. B280-B Muto, I., Y. Izumiyama, and N. Hara, Microelectrochemical measurements of dissolution of MnS inclusions and morphological observation of metastable and stable pitting on stainless steel. Journal of the Electrochemical Society, (8): p. C439- C Burstein, G., et al., Origins of pitting corrosion. Corrosion Engineering, Science and Technology, (1): p Laycock, N.J. and R.C. Newman, Localised dissolution kinetics, salt films and pitting potentials. Corrosion Science, (10-11): p Almuaili, F.A., et al., Application of a Quasi In Situ Experimental Approach to Estimate 3-D Pitting Corrosion Kinetics in Stainless Steel. Journal of the Electrochemical Society, (13): p. C745-C A262-10, A., Standard Practices for Detecting Susceptibility to Intergranular Attack in Austenitic Stainless Steels Standard, A., E112: Standard Test Methods for Determining Average Grain Size. West Conshocken, Engelberg, D.L., Intergranular Corrosion, in Shreir's Corrosion, B. Cottis, et al., Editors. 2010, Elsevier: Oxford. p Krawiec, H., et al., Dissolution of chromium-enriched inclusions and pitting corrosion of resulfurized stainless steels. Metallurgical and Materials Transactions a-physical Metallurgy and Materials Science, A(5): p Stewart, J. and D.E. Williams, The Initiation of Pitting Corrosion on Austenitic Stainless- Steel - on the Role and Importance of Sulfide Inclusions. Corrosion Science, (3): p. 457-&. 27. Park, J.O., T. Suter, and H. Bohni, Role of manganese sulfide inclusions on pit initiation of super austenitic stainless steel's. Corrosion, (1): p Muto, I., S. Kurokawa, and N. Hara, Microelectrochemical investigation of anodic polarization behavior of CrS inclusions in stainless steels. Journal of the Electrochemical Society, (11): p. C395-C Zheng, S., et al., Mechanism of (Mg, Al, Ca)-oxide inclusion-induced pitting corrosion in 316L stainless steel exposed to sulphur environments containing chloride ion. Corrosion Science, (0): p

123 30. Pisarek, M., et al., Characterization of the effects of hydrostatic extrusion on grain size, surface composition and the corrosion resistance of austenitic stainless steels. Materials Characterization, (9): p Pistorius, P.C. and G.T. Burstein, Metastable Pitting Corrosion of Stainless Steel and the Transition to Stability. Philosophical Transactions of the Royal Society of London Series a-mathematical Physical and Engineering Sciences, (1662): p Krawiec, H., et al., Influence of the dissolution of MnS inclusions under free corrosion and potentiostatic conditions on the composition of passive films and the electrochemical behaviour of stainless steels. Electrochimica Acta, (16): p De Wit, J., E. Jansen, and L. Jacobs. A comprehensive electrochemical approach to the relation between pitting, passivity and inclusions in stainless steel. in Materials Science Forum Trans Tech Publ. 34. Wijesinghe, T.L.S.L. and D.J. Blackwood, Real time pit initiation studies on stainless steels: The effect of sulphide inclusions. Corrosion Science, (4): p Muto, I., D. Ito, and N. Hara, Microelectrochemical investigation on pit initiation at sulfide and oxide inclusions in type 304 stainless steel. Journal of the Electrochemical Society, (2): p. C55-C Ke, R. and R. Alkire, Initiation of corrosion pits at inclusions on 304 stainless steel. Journal of the Electrochemical Society, (12): p Yi, Y., et al., Potentiodynamic polarization behaviour of AISI type 316 stainless steel in NaCl solution. Corrosion Science, : p Virtanen, S. and H. Bohni, On the stability of passive films on stainless steels. Passivation of Metals and Semiconductors, : p Park, J.O. and H. Bohni, Local ph measurements during pitting corrosion at MnS inclusions on stainless steel. Electrochemical and Solid State Letters, (9): p Wang, J.H., C.C. Su, and Z. Szklarska-Smialowska, Effects of Cl-concentration and temperature on pitting of AISI 304 stainless steel. Corrosion, (10): p Ernst, P. and R.C. Newman, Explanation of the effect of high chloride concentration on the critical pitting temperature of stainless steel. Corrosion Science, (9): p Pardo, A., et al., Influence of ph and Chloride Concentration on the Pitting and Crevice Corrosion Behavior of High-Alloy Stainless Steels. Corrosion, (4): p Chiba, A., et al., A Microelectrochemical System for In Situ High-Resolution Optical Microscopy: Morphological Characteristics of Pitting at MnS Inclusion in Stainless Steel. Journal of the Electrochemical Society, (8): p. C341-C Chiba, A., et al., Microelectrochemical Aspects of Interstitial Carbon in Type 304 Stainless Steel: Improving Pitting Resistance at MnS Inclusion. Journal of the Electrochemical Society, (6): p. C270-C Suter, T., et al., Pit initiation on stainless steels in 1 M NaCl with and without mechanical stress. Journal of the Electrochemical Society, (5): p. B174-B Malik, A.U., et al., The influence of ph and chloride concentration on the corrosion behaviour of AISI 316L steel in aqueous solutions. Corrosion Science, (11): p Marcus, P., A. Teissier, and J. Oudar, The influence of sulphur on the dissolution and the passivation of a nickel-iron alloy I. electrochemical and radiotracer measurements. Corrosion Science, (4): p

124 6. OBSERVATION AND ESTIMATION OF 3D PITTING CORROSION KINETICS The work presented in this chapter demonstrates the application of a quasi in-situ X-ray computed tomography approach using laboratory X-ray equipment to estimate 3D pitting corrosion kinetics under electrochemical polarisation control. The study of 3D pit growth was conducted in bulk electrolyte using a miniature electrochemical cell. This set-up was developed to conduct in-situ electrochemical measurements of pitting corrosion under electrochemical polarisation control, with the progress of corrosion observed, quasi in-situ, using X-ray CT. The first study (Manuscript B, section 6.2) revealed the nucleation and development of three discrete pits with exposure to chloride-containing electrolyte and under polarisation control. One single pit was observed, and compared to the growth of two pits during the second polarisation cycle. The results of these 3D corrosion experiments were explored, using different pit shapes to approximate pit growth kinetics. The kinetics were obtained, based on measured pit volumes and surface areas from X-ray CT, rather than by estimating pit shapes. Large fluctuations of the mean current density (averaged current over the total internal surface area of a pit) were observed during the pit nucleation stage, with pit stability products in the order of 0.3 to 0.6 A.m -1, and estimated diffusion product (DΔC) of 1.68 to mol.cm -1.s -1 for pit growth. In the second paper (Manuscript C, section 6.3), the kinetics of multiple pits were estimated. A novel approach was taken to allow the current obtained from each pit to be separated from the total evolution of current, based on the predicted volume evolution. Two distinct regions of mean current density were observed during 3D pit development, with maximum current densities of up to 14 A.cm -2. This was followed by pit growth with typical mean current densities of ~1-2 A.cm -2. Associated pit stability products of stable pit growth in 3D were also above 0.3 A.m -1 and DΔC was estimated at 2.73 to mol.cm -1.s

125 6.1 Key findings and results 3D pit growth kinetics can be estimated via quasi in-situ measurements. The pit volumes obtained by X-ray CT showed a good fit with the volume of metal dissolution calculated using Faraday s law. 1- The surface areas of pits measured from X-ray CT data were larger than those calculated by assuming hemispherical pit growth, with pits approaching elongated dish shapes rather than a hemispherical shape. 2- Large fluctuations of current density were observed during the pit nucleation stage, with a drop in current before stable pit growth was achieved. 3- Typical mean current densities of 1 to 3 A.cm -2 with pit stability products of 0.3 to 0.6 A.m -1 have been estimated for stable pit growth. 4- DΔC of mol.cm -1.s -1 and up to mol.cm -1.s -1 were estimated. 125

126 6.2 Application of a Quasi In-situ Experimental Approach to Estimate 3-D Pitting Corrosion Kinetics in Stainless Steel (Manuscript B) F. A. Almuaili 1, S. A. McDonald 2, P. J. Withers 2, D. L. Engelberg 1 1 Corrosion and Protection Centre, School of Materials, The University of Manchester, Sackville Street, Manchester, M13 9PL, United Kingdom 2 Manchester X-ray Imaging Facility, School of Materials, The University of Manchester, Sackville Street, Manchester, M13 9PL, United Kingdom Abstract Pitting corrosion kinetics of type 304L stainless steel have been obtained using a quasi in-situ X-ray computed tomography (X-ray CT) approach. A miniature electro-chemical cell was constructed to allow imaging during potentio-dynamic polarization of a wire specimen in chloride solution. The formation of three discrete pits was observed, allowing comparison of pit volume, surface area and depth of the real (measured) pit geometry with different geometrical assumptions to estimate pit growth kinetics. The pit volumes obtained by X-ray CT showed a good fit with the volume of metal dissolution calculated using Faraday s law. Large fluctuations of the mean current density were observed during the pit nucleation stage, followed by pit growth with mean current densities of 1-3 A.cm -2. Stability products associated with these pits were of the order of A.m -1, with a diffusivity parameter (DΔC) of mol.cm -1.s -1. Diffusion coefficients for stable pit growth of cm 2.s -1 were estimated for metal ion concentrations of 4.2 M. Keywords: Pitting Corrosion, Potentio-dynamic Polarisation, Stainless Steels, X-ray Computed Tomography Introduction Pitting or crevice corrosion can occur on stainless steels exposed to halide containing environments. The breakdown of the passive surface film results in the nucleation of local attack, which is followed by pit growth, with a number of mechanisms proposed for each stage [1, 2]. Electrochemical polarisation above the critical pitting potential leads to formation of pits, in the form of localised metal dissolution, with corrosive 126

127 electrolyte developing inside the nucleated pit due to hydrolysis of dissolved metal ions, and chloride ions being attracted from bulk solution to maintain charge neutrality [3]. At high applied electrochemical potentials, pits grow with a polished inner surface morphology, associated with salt film formation, whereas pits grown at lower applied potentials display more irregular etch morphologies of the internal pit surface [4]. An undercutting mechanism was suggested for pit growth at high electrochemical potentials, leading to the development of dish shaped pits covered by metal lacy covers [5-10]. Pit propagation is controlled by diffusion of species through the salt film at the inner surface of pits. A high anodic dissolution rate, which facilitates critical ion concentrations inside a pit, is the criterion for pits to propagate [11]. The anolyte concentration inside a pit influences pit growth kinetics and the pit shape. For example, a 3 M concentration of dissolved metal ions was proposed as the minimum concentration to turn an open hemispherical meta-stable pit into a stable pit. However, pits developed with lacy metal covers grow at far lower metal ion concentrations, since these covers are believed to provide an effective diffusion barrier [12] and an increased ohmic resistance between pit interior and the bulk electrolyte [1]. Based on these observations for the transition between meta-stable and stable pits, the pit stability product (0.3 A. m -1 i.r 0.6 A. m -1 ) was proposed, with the latter based on the evolution of pit current density (i) and pit depth (r) [12]. To sustain stable pit growth under anodic polarisation, a minimum current density is required, with typical values for stainless steel in the region of 1-5 A.cm -2 [13, 14]. For pit initiation, however, far higher current densities were observed [15], followed by a reduction with increasing pit volume which corresponds to an increase in active pit surface area over time. The electrochemical potential at the pit bottom remains above the repassivation potential resulting in active dissolution conditions. Therefore, initial pit growth studies suggested that associated kinetics were potential dependent under charge transfer control, resulting in a hemispherical pit shape. After the formation of a salt film inside the pit due to limited solubility of cations, pit propagation is considered to be under diffusion-control and potential independent, where the pit shape changes into more elongated dish shape over time. The sides of the pit, typically referred to as lobes, grow faster than the bottom part. Furthermore, due to this mechanism the pit aspect ratio (depth vs. width) can change, suggesting that pits grow under diffusion 127

128 control, with ohmic resistance controlling lateral pit growth [7]. Typical current densities between 1-2 A.cm -2 are related to the precipitated salt film at the bottom of the pit, whereas, salt film free regions (side walls) show a maximum current in the range of 3-5 A.cm -2 [9]. The difference in potential between pit bottom and sidewall is related to potential drop through the salt film. The growth kinetics of a single pit in type 304 stainless steel under potentiostatic polarisation control in 1 M NaCl with 0.4 M Na-thiosulfate solution showed a parabolic behaviour [16]. Pit volumes measured by means of mechanical grinding and sectioning followed by optical microscopy showed a good fit with the dissolved volume determined by current measurements using Faraday s law and assuming a hemispherical pit shape. The study also showed that pit growth under potentiostatic control in 3.5% wt. NaCl and seawater was similar to pits grown under open circuit conditions in 1 M FeCl 3 solution, both indicating an exponential relation of pit growth with time [17]. One dimensional (1D) pit growth investigations using pencil electrodes [18] and twodimensional (2D) pit geometry estimated using thin foils and in-situ recording or radiography images [9] have been used to allow estimation of pit stability product. These methods allow pits underneath the metal surface to be observed, but the developed pit shape does not necessarily represent pit growth in 3D. The effect of constraining pits to grow in 1D or 2D may not accurately reflect 3D growth kinetics. It may affect transport processes of dissolved metal ions between pit anolyte and the bulk solution, which can subsequently affect local pit chemistry and associated growth dynamics. For pencil electrode (1D) measurements, pit growth occurs in depth by maintaining a constant surface area related to the diameter of the pencil electrode [14, 19, 20]. These 1D studies revealed a transition of pit growth kinetics from ohmic/dissolution control to diffusion control at high potentials. A transition current density between both growth regimes in the region of 1-5 A.cm -2, with the latter a function of the chloride concentration inside the pit, was also observed [14]. A potential step experiment with 1D electrodes further revealed the effect of Cr content and solution chemistry on the dissolution and passivation kinetics, suggesting that salt film precipitation controls the metal dissolution rate[20]. The effect of electrolyte species also has a profound effect on pit growth kinetics with, for example, the addition of sulphate to NaCl environment 128

129 significantly reducing the current density for pit growth [21]. In this study, pit chemistry was determined based on the minimum concentration at the pit surface that maintains pit growth without passivation. The dissolved metal salt concentration was then estimated by assuming a constant diffusion coefficient (D), yielding diffusivity parameters (DΔC) of mol.cm -1.s -1 for NaCl, and mol.cm -1.s -1 for NaCl with sulphate [21]. Pistorius and Burstein also reported a reduction of current density in presence of sulphate for stable pitting on type 304 stainless steel [22]. The development of pit shape and current density over time was recently obtained in stainless steel foils using in-situ X-ray synchrotron radiography experiments [23]. This study showed that an elongated dish-shape developed under potentiostatic control, with the pit shape remaining near-hemispherical under galvanostatic control. Current densities measured, based on pit boundary movement over time using a sequence of 2-D radiography images, showed that the distribution of current density along the pit perimeter was not uniform and was influenced by the concentration of metal cations and the pit shape. Therefore higher values of current density at active lobes drive the pit growth through an undercutting process. Furthermore, the smooth surface at the bottom of the pits show lower current density, while the pit surfaces near the bulk solution tend to passivate. Pit growth showed a fluctuation of the mean pit stability product with values between 0.3 and 0.4 A.m -1 or above [23]. Pit diffusivity parameters (DΔC) for 2D pit growth in 0.1 M NaCl was estimated from the gradient between square depth and time, resulting in values of mol.cm -1 s -1 under potentiostatic control, and mol.cm -1 s -1 under galvanostatic control. For 1D pit growth in 1 M NaCl a diffusivity parameters of mol.cm -1 s -1 was determined, with the difference attributed to the perforation factor between an open pit (1D) and a covered pit (2D) [9]. Corrosion pits initiated at the tip of a type 304 stainless steel pin using a capillary micro-cell have also been observed in 3D using X-ray synchrotron micro-tomography. Pits grown under current and potential control in 1 M NaCl electrolyte showed similar pit morphologies as reported in earlier 2D pit growth studies on foils [8]. Another 3D X-ray micro-tomography study was conducted recently, to investigate the relationship between pitting corrosion and intergranular corrosion in sensitised type 316H stainless steel exposed to 0.1 M NaCl electrolyte [24]. The results showed the occurrence of both forms of corrosion attack. Pits nucleated at the surface of the electo-chemically polarised wire, with the aggressive pit anolyte leading to intergranular attack along 129

130 grain boundaries within the pit. Earlier 3D studies also investigated the propagation of intergranular corrosion and stress corrosion cracking in sensitised austenitic stainless steel and aluminium alloy 5083, highlighting the application of 3D imaging on corrosion and crack growth kinetics [25-27]. The aim of our study was to develop an experimental methodology to investigate 3D pitting corrosion kinetics during exposure to bulk electrolyte, by using a miniature 3- electrode electrochemical cell combined with quasi in-situ X-ray computed tomography (X-ray CT). The second goal was to obtain 3D pit growth kinetics, and compare those to literature reports of pit growth in 2D and 3D, obtained from experiments using miniature capillary probes Experimental A solution annealed type 304L stainless steel wire with a diameter of 500 μm was used in this study, with a chemical composition of (wt. %) 18.4 Cr, 8.7 Ni, 0.02 C, 1.4 Mn, 0.34 Si, 0.04 N, 0.03 P and S. Short wire sections of 70 mm were cut and manually ground using 1200 grit SiC paper, followed by a rinse in deionised water. The surface of the wire was coated with beeswax exposing a cylindrical surface area of 2.83 mm 2. The wire section was mounted vertically in a miniature electrochemical cell shown in Figure 6-1(a), and the set-up was then placed in a ZEISS Xradia 400 Versa Micro tomography instrument (Figure 6-1b). The diameter of the cell was 24 mm, consisting of a lower part containing the electrolyte with a volume of approximately 9 ml for electrochemical polarisation measurements, and an upper part with a straining rig for applying a tensile load along the length of the wire. The latter was used in another study to investigate the effect of strain on pit growth kinetics and for assessing the nucleation of stress corrosion cracking. The bottom part of the cell housed a miniature reference electrode (Ag/AgCl, 3 M NaCl) and a miniature platinum counter electrode. Electrochemical polarisation tests were performed with a scan rate of 1 mv/s in aerated 0.1 M NaCl solution using an Ivium CompactStat Potentiostat. Prior to polarisation, the open circuit potential (OCP) was monitored for 15 min. The current response was recorded during the polarisation experiment at a rate of 1 Hz. Table 6-1 gives a summary of the in-situ experiment with associated polarisation scans. After each scan, one X-ray CT scan was performed to visualise the progress of pitting 130

131 corrosion over time. X-ray CT scans were recorded at OCP, with each scan taking approximately 2-3 hrs. The sample remained in the solution throughout the whole experiment. For the X-ray CT measurements an accelerating voltage of 120 kv was used, and 721 projections recorded with 2 2 binning. This resulted in a reconstructed voxel size of 1.83 µm 3, with a field of view of 2300 μm 2300 μm. Load Cell Counter electrode In situ cell electrolyte Working electrode Reference electrode (a) X-Ray Source Sample inside the cell Detector (b) Figure 6-1: (a) Photo of the miniature electrochemical cell with the capability to apply strain for in-situ x-ray tomography experiments with a type 304L wire sample, (b) insitu cell during an X-ray CT experiment. 131

132 The data was reconstructed using a Feldkamp-Davis-Kress (FDK) approach [28], and images were segmented and visualised using Avizo software. After segmentation of each pit, the total volume, total pit surface area, pit depth, width, height, aspect ratio and shape were obtained. The pit depth, width and height were obtained by using 2D slices of the 3D data-set taken from the geometrical pit centre. This approach provided a snapshot of pit dimensions after each polarisation scan, with the current response recorded during the polarisation experiment providing in-situ pit growth information. The wire was then removed from the in-situ cell after step 9 (Table 6-1), rinsed in water and images of the pits obtained using a FEI Quanta 650 scanning electron microscope (SEM). Table 6-1: Summary of in-situ electrochemical polarisation experiment. Step Polarisation Cycles 0 X-ray CT scan (reference / without electrolyte) 1 OCP measurement (15 min.) 2 Potentio-dynamic polarisation from OCP to +644 mv vs. Ag/AgCl 3 X-ray CT scan (1) at OCP 4 OCP measurement (5 min) 5 2nd potentio-dynamic polarisation from OCP to +644 mv vs. Ag/AgCl 6 X-ray CT scan (2) at OCP 7 OCP measurement (5 min) 8 3rd potentio-dynamic polarisation from OCP to +700 mv vs. Ag/AgCl 9 X-ray CT scan (3) at OCP Results and Discussion The first X-ray CT scan (step 0) was recorded before the sample was exposed to the electrolyte, to obtain reference data of the investigated wire volume. This scan confirmed that no corrosion or mechanical damage was present before the sample was polarised. After completing this scan, the NaCl electrolyte was introduced into the insitu cell. The 1 st potentio-dynamic polarisation (step 2 in Table 6-1) resulted in the formation of one corrosion pit (pit 1), with the 2 nd potentio-dynamic scan (step 5 in Table 6-1) leading to the nucleation and growth of two new pits (pit 2 and 3). Figure 6-132

133 2(a) shows a 3D view of the reconstructed tomography data of the wire with pit 1, and Figure 6-2b gives the same volume with all 3 pits, recorded after the 2 nd polarisation cycle. The final polarisation cycle (step 8) did not result in the formation or further growth of corrosion pits. A SEM image of the wire after the test is shown in Figure 6-2(c). Pit 2 Pit 1 Height (h) Pit 1 Pit 3 Width (w) (a) Pit 2 (b) Pit 1 Pit 3 (c) Figure 6-2: (a) Reconstructed X-ray CT data volume of the wire after the 1 st electrochemical polarisation scan, (b) X-ray CT data volume after the 2 nd potentiodynamic polarisation scan, and (c) SEM image of the wire sample with the three pits. 133

134 Electrochemical polarisation The OCP prior to the first polarisation was +129mV vs Ag/AgCl (Step 1), with an OCP prior to the second scan of +151mV (step 4). The current evolution over time of the first and second potentio-dynamic scan are shown in Figure 6-3(a). The current evolution over time of second polarisation resulted in two pits (pit 2 and pit 3). During the first polarisation cycle, the current started to rise at +616 mv up to the applied max. potential of +644 mv vs. Ag/AgCl. The second polarisation resulted in a current increase starting at +596 mv but with a far steeper rise of current over time, resulting in approximately double the gradient compared to the current evolution observed during the first polarisation cycle. Both curves show a drop of current after a few seconds followed by a continuous rise again. The shift in OCP after the first polarisation is either due to the growth of the passive surface film, often associated with anodic polarisation of passive material, or alternatively an effect of the high energy X-ray beam, causing chemical changes of the film/electrolyte interface. This shift is also observed after the second X-ray CT scan (step 6) with an OCP recorded of +178 mv vs. Ag/AgCl. A third polarisation cycle (step 8) was performed from OCP up to +700 mv, but no measurable current increase over time was observed. No further pit was nucleated during this cycle, which may be related to the limited number of active sites of inclusions at this potential. 134

135 (a) (b) Figure 6-3: (a) Current evolution vs. time of the 1 st and 2 nd potentio-dynamic polarisation scan with (b) measured depth (r), width (w) and height (h) of all three pits from X-ray CT data Pit geometry and estimation of growth kinetics The pit dimensions at the end of the polarisation with depth, width and height obtained from measurements of the X-ray CT data are shown in Figure 6-3(b). All pit volume and surface parameters measured via segmentation are summarised in Table 6-2. Beside the direct measurements from X-ray CT data, two additional geometrical approaches (B and C) were also explored to estimate pit growth kinetics. Table 6-2 lists the parameters of all 3 approaches, which were applied to simulate pit growth kinetics. 135

136 Table 6-2: Measured pit geometries. A measured data from B - depth measured by C- Faraday approach Methods/pit parameters X-ray CT scans X-ray CT + assumption of a hemi-spherical pit shape + assumption of a hemispherical pit shape Pit 1 Pit 2 Pit 3 Pit 1 Pit 2 Pit 3 Pit 1 Pit 2 Pit 3 Depth, r (μm) * 35.8* 35.8* Width, w (μm) * 60* 62* 45.4* 71.6* 71.6* Height, h (μm) * 60* 62* 45.4* 71.6* 71.6* Volume, V (μm 3 ) * 56520* 62362* Surface area, A (μm 2 ) * 5652* 6035* 3225* 8060* 8060* * assumed values in bold In approach A, all values were measured from segmented X-ray CT data. The area of the pit surface was obtained by measuring the surface area of the segmented pit volume, excluding the area of the pit mouth. This represented the real area of the pit internal surface assuming the pit cover is passive. In the case of pits 2 and 3, the similar size of both segmented volumes from X-ray CT data, in combination with the current evolution over time implied that both had nucleated at the same time during the 2 nd polarisation scan. This assumption of simultaneous pit growth then allowed the current in Figure 6-3(a) to be equally divided between pit 2 and 3, which is supported by the observed current over time gradient, i.e. approximately double that observed for the growth of pit 1. In approach B, the pit volume and surface area were calculated by assuming the pit shape is hemi-spherical with only the pit depth (r) obtained from the X-ray CT measurements. The computed volume (V) of the hemisphere followed the equation V = 2/3.π.r 3, with the development of the surface area determined by assuming isotropic growth from a local initiation point at the centre of the hemi-sphere diameter. To determine the evolution of pit surface area over time, it was assumed that pits were active at the end of the polarisation cycle. This is supported by the current response in Figure 6-3(a). The assumption of symmetric growth of the pit volume then allowed back extrapolation of the pit surface area from the end of the polarisation scan to the point where the current started to rise. This assumption is based on a mean current 136

137 density over the entire pit surface area to satisfy homogeneous growth of both, the pit surface area and pit volume. However, this also allowed comparison of the measured current response in Figure 6-3(a), by computing the measured current over the estimated surface area for each point in time, shown as current density plots in Figure 6-4(a, b). Comparison of the behaviour of pits observed in this study then allowed trends of current density and associated stability product values over time to be identified (Figure 6-4(c, d)). In approach C, the pit volume was calculated from the charge passed during the period of pit growth in Figure 6-3(a) and converted into mass using Faraday s law. The metal dissolution assumed an average metal cation charge of n = 2.19, atomic weight M = g.mol -1 and density of ρ = 7.97 g.cm -3, and Faraday s constant F = coulomb/mol [12]. Also a hemispherical pit shape was assumed, with a radius based on the dissolved volume (Table 6-2). The development of the pit surface area for estimating the corresponding current densities was also based on a homogeneous symmetric growth, as described for approach B. From the current-time response in Figure 6-3(a), the pit current density and pit stability product over time were estimated (Figure 6-4). Table 6-2 indicates that pit volumes calculated via segmentation (approach A) are close to those calculated via Faraday s law (approach C), with a difference of less than 6%. This difference may be explained with uncertainty in segmented X-ray CT data. Likewise, the small contribution to the overall current density from the passive surface of the wire was also not considered in these calculations. By only considering the measured X-ray CT data of approach A, the relationship of growth in pit depth over time can be determined, corresponding to r = a. t x, where (r) represents pit depth, (a) and (x) empirical pit growth constants, and (t) time. Computing the pit depths for all 3 pits gives similar values for (x), confirming that pit growth kinetics for all 3 pits were similar. For achieving t 0.5 a pit growth constant (a) of between m.s -1 would need to be assumed, which is in the range of typical growth constants reported. [29] This indicates that the pit depth growth rate in 3-D was possibly under diffusion control, similar to the reported t 0.5 of diffusion control for 1D [30] and for 2-D [7] pit depths. The 3-D pit lateral growth by assuming (r) is the width (w) and height (h), show a relation with time of > t 0.5. This implies that the width grew 137

138 in a less than saturated solution faster than the depth, which is also supported by the overall dimensions obtained in Figure 6-3(b). Figure 6-3(b) shows a slight variation between pit width and height for pit 1 and 2 compared to pit 3. This indicates that all the pits shape are not hemispherical, with 2D X-ray CT slices and SEM images showing an semi-ellipsoid dish shape (see appendix B). Figure 6-3 shows the ratio of pit depth over width and height is in the range of 0.41 and 0.47, fitting well with reported values of 0.4 to 0.5 [13]. This suggests that pits grow under diffusion control, with ohmic resistance controlling lateral pit growth Current density and stability product In order to obtain pit growth rates, the current densities in Figure 6-4(a) have been calculated by using the obtained current response of Figure 6-3(a), divided by internal pit surface area. Three values of pit current density over time are presented based on the three approaches A, B and C outlined in Table 6-2. Figure 6-4(a) shows mean current density as a function of time for pit 1. Two regions can be distinguished: the first where the current density increases rapidly and the second where it starts to drop over time. Fluctuations in current density also become smaller over time. The two regions consist of a transient and quasi-steady state period which have been reported for pit growth [30]. The maximum current density of pit 1 reached 5 A.cm -2 at the transient region for approach A, reducing to a mean value of 1.8 A.cm -2 in the steady state region. Figure 6-4(b) shows the results for pit 2 and 3. The behaviour of pit current density over time is similar to pit 1, but the time for pit growth was almost doubled. The results also show a limiting current density of 5 A.cm -2 for approach A, similar to the maximum value observed for pit 1. At the end of the polarisation scan, a mean current density of 1.3 A.cm -2 was obtained. 138

139 (a) (b) (c) 139

140 (d) Figure 6-4: The values in Table 6-2 of the three approaches A, B, and C are applied to show (a) pit 1 current density vs. time, (b) pit 2 and 3 current density vs. time, (c) pit stability product vs. time of pit 1, and (d) pit stability products vs. time of pit 2 and 3. In Figure 6-4(a, b), approach A has the lowest current densities of all three approaches, indicating that the real pit surface area must be larger relative to the assumptions in approach B, with approach C overestimating the pit depth compared to the real pit dimensions (Table 6-2). The difference in current density between approach A and B is nearly 50%, whereas for A and C it is only 6%. An increase of current density with increasing polarisation time up to the maximum current value under activation polarisation in the transient stage, and after that the thickness of salt film precipitation affects pit growth [31]. A sharp drop in current density was observed at the transition between both stages, which may indicate that dissolved metals ions reach saturation and salt film precipitation occurs, reducing the measured current at this stage to very low values. Longer pit growth periods clearly show a quasi-steady state of current density. The reduction of current density over time at this location is a result of the increase in pit size. The difference in calculated current density between approach A and approach B and C show that longer pit growth periods in Figure 6-4(b) reduce the difference between them to less than 3%. Figure 6-4(c, d) show the value of stability products as a function of time for all three pits obtained by the three approaches. Stable pits should have a stability product range of 0.3< i.r <0.6 A.m -1 [12]. Pit stability values obtained by the three methods for all pits 140

141 increased over time, and fit in the range of stable pit criteria. However, the result of approach A relative to approaches B and C show lower stabilities of 25% and 50% respectively. The results also show that pit stability product was initially below the stable pit growth criteria, and then increased with time. At an early stage of pit growth, the pit stability product was below 0.3, and pit growth is then believed to be supported by a diffusion barrier in the form of lacy metal covers [12]. It should be noticed that the above criteria were developed by Pistorius and Burstein for stainless steel in chloride electrolytes, assuming hemispherical open pit growth and 3 M concentration as a minimum for metal ion dissolution for metastable to stable pit transition. The measurement also assumed a diffusion coefficient of D = cm 2.s -1 [12, 22]. Comparing the above three approaches in our study, the results indicate that the current density and pit stability product show slight differences. This variation can be related to pit shape differences and the effect of non-uniform dissolution on the local chloride concentration. It appears that using estimated methods B and C leads to an overestimate in the pit depth growth rate as compared to that obtained from a 3D analysis of current density and associated stability products Pit diffusion product estimations Figure 6-5(a) shows the relationship between the square of the pit depth (r 2 ) over time (t) using equation 1 (derived from Fick s first law and Faraday s second law), which suggests that pit growth with salt layer is under diffusion control. The equation shown below was used to estimate pit diffusion product.[9, 12] r 2 = ((3. M. D. C)/π. ρ). t (6-1) Where, D is the effective diffusion coefficient and C is concentration difference between pit bottom and mouth with atomic weight M =55.79 g.mol 1 and density of ρ =7.97 g.cm 3. The gradients in Figure 6-5(a) provide the diffusion product (DΔC) with , and mol.cm -1 s -1 for pits 1, 2 and 3, respectively, by using the pit depths of the segmented volumes in approach A. These show linear relationship between square depth of pits and time, suggesting diffusion controlled pit growth in all cases. The obtained slopes are lower than those reported for diffusion controlled pit 141

142 growth in 1D,[7, 32] but similar to 2D pit growth studies facing upward diffusion [9]. Open pit growth in 1D with pits facing upwards, typical diffusion product values close to mol.cm -1.s -1 were obtained in 1 M NaCl solution [21], whereas, 2D pit growth in 0.1 M NaCl with pits facing upwards and lacy metal covers showed values similar to our results for pits 2 and 3 [9]. It may be possible that 3D pit growth and the vertical sample orientation allows the electrolyte to be more easily infused by gravity, relative to 1D and 2D experimental set-ups, where the electrolyte within the pit is constrained by pit geometry and sample position [33]. In our study, pit 1 shows a lower diffusion product (DΔC) relative to pits 2 and 3, possibly due to the influence of the lacy cover observed for pits 2 and 3 (Figure 6-2c), affecting the anolyte concentration inside the pit. A number of factors were suggested to influence the gradient of pit growth, such as the area of perforation of the metal lacy covers, chloride concentration, applied potential, type of electrolyte and temperature, as well as pit growth direction, with respect to the sample position [7, 9, 14, 21, 34-37]. For example, a difference of 15% was reported between pit growth with the sample facing upwards versus downwards in 2D [7]. In our study, all pits grew vertical at the wire, and minor differences in diffusivity parameters estimated for pits 2 and 3 may even be a result of the difference in perforated area of both pit covers [9]. The vertical section and horizontal section of tomogram and SEM images (see appendix B) show similar morphology between horizontal and vertical tomogram. It indicates elongated dish shape while SEM images clearly show lacy cover on pit 2 and 3. At high potential, polarization for short period may leads to fast polished surface and make the influence of gravity on pit shape less significant. If the diffusion coefficient (D) is assumed to be constant at cm 2.s -1 [9, 19], the mean metal salt concentration inside the pits can be estimated from the diffusivity parameters given above. The mean concentration would therefore be around 1.97 M, 3.34 M and 3.57 M for pits 1, 2 and 3, respectively. These values indicate that the mean concentration at the pit bottom for pit 1 is below the 3 M (75% of saturation concentration of FeCl 2 ) suggested for stable growth facing upwards in type 304 stainless steel [12]. It was also suggested that stable pits can propagate without lacy covers, as the pit depth also provides a diffusion barrier to maintain the corrosive solution for active pit dissolution. This would mean that in our case pit 1 propagates without lacy metal cover in 1.97 M metal salt concentration with a pit depth of 19 μm 142

143 acting as effective diffusion barrier to support stable pit propagation after the pit lost its lacy cover. This concentration is close to the reported values of M, defined as critical ion concentration for stable pit transition in stainless steel at constant potentials [38]. Furthermore, it is shown that losing the lacy metal cover certainly affects the ion concentrations inside pits, since pit 2 and pit 3 show far higher concentrations of 3.34 M and 3.57 M, respectively, with both growing to similar depths of roughly 30 μm. However, comparison of the growth kinetics of pits inferred by assuming a constant diffusion coefficient is not correct, since the effect of chloride concentration and its change over time need to be taken into consideration. The effect of chloride concentration can be shown, for example, by assuming a constant metal ion concentration inside the pit of 4.2 M throughout the pit growth period [34]. This would result in our case in mean diffusion coefficients of for pit 1, and cm 2.s -1 for pits 2 and 3, based on the above diffusivity products of (DΔC) , and mol.cm -1 s -1. (a) 143

144 (b) Figure 6-5: Effect of time on (a) (pit depth) 2 to obtain the diffusion parameters of these curves from the slopes, (b) effective diffusion coefficient (D) estimated with a constant metal ion concentrations of 4.2 M. Alternatively, the variation of effective diffusivity (D) with pit growth over time can be estimated using equation 6-2,[22] with the results summarised in Figure 6-5(b). C = (2π/(3n. F. D )) i. r (6-2) This gives a realistic approach for the effect of time, by considering (D) as a key variable, and using the evolution of current density (i) and pit depth (r) over time from approach (A) as input parameters. The saturation concentration ( C) was kept constant at 4.2 M. This assumption does not reflect changes in metal ion concentration and pit growth regimes that are not under diffusion control during the initial pit nucleation stage and is therefore not fully correct. However, the results reflect the latter by showing large variations of D at the early stages of pit growth. The diffusion coefficients then converge with longer exposures, resulting in pit growth under diffusion control. Figure 6-5(b) clearly shows a large variation of diffusion coefficients during the initial transient stage, which is most likely caused by overestimating the metal ion concentration inside the pit. In reality, lower metal ion concentrations than the one used in this estimation will be present, ultimately resulting in the formation of lacy metal covers. Interestingly, effective diffusion coefficients of for pit 1 and cm 2.s -1 for pit 2 and 3 are obtained for the steady state pit growth period in Figure 6-5(b), which 144

145 is close to the estimated diffusion coefficient for 1D and 2D pit growth studies [22, 30]. The drop in diffusion coefficient between 7-9 seconds for all pits may indicate salt film precipitation and for pit 1 this may be associated with the time where pit 1 lost its lacy metal cover as seen in Figure 6-2(c). This study shows that quasi in-situ X-ray CT experiments provide an effective tool to study pit growth kinetics and to probe assumptions for optimising pit kinetics for predicting material behaviour. The advantage of the 3D approach over previous 1D and 2D in-situ approaches lies in the reduction of geometrical constraints, by studying real 3D systems. Further experiments are currently conducted with this approach, by inducing nucleation and growth of multiple pits, to understand whether these affect each others growth kinetics, and the influence of strain on pit growth Conclusions 1-3D pit growth kinetics can be estimated via quasi in-situ measurements from electrochemical polarisation tests with information of pit dimensions using X- ray CT data. The pit volumes obtained by X-ray CT showed good fit with the volume of metal dissolution calculated using Faraday s law. 2- The measured surface area of pits from X-ray CT is larger than those calculated by assuming hemispherical pit growth, with pits approaching elongated dish shapes rather than perfect hemispherical shapes. 3- Typical mean current densities of 1-3 A.cm -2 with pit stability product of A.m -1 have been estimated for stable pit growth. Diffusivity parameter (DΔC) between mol.cm -1.s -1 were obtained. 4- Pit growth rates in 3D indicate that the pit depth is under diffusion control, whilst lateral growth occurred faster. 5- Effective diffusion coefficients (D) from the pit base to the pit mouth were estimated at and cm 2.s -1 for pit growth with assumed metal ion concentrations of 4.2 M Acknowledgements The authors acknowledge the provision of beam time at Henry Moseley X-ray imaging Facility (HMXIF) at the University of Manchester, UK, established with funding from EPSRC through grants EP/F007906, EP/I02249X and EP/F The authors would also like to thank Saline Water Conversion Corporation (SWCC), Saudi Arabia for 145

146 financial support. Valuable discussions with Dr Anthony Cook (University of Manchester) are also acknowledged References 1. Frankel, G., Pitting corrosion of metals a review of the critical factors. Journal of the Electrochemical Society, (6): p Szklarska-Smialowska, Z., Review of Literature on Pitting Corrosion Published Since Corrosion, (6): p Hoar, T.P. and W.R. Jacob, Breakdown of Passivity of Stainless Steel by Halide Ions. Nature, (5122): p &. 4. Sato, N., The Stability of Localized Corrosion. Corrosion Science, (12): p Isaacs, H. and G. Kissel, Surface preparation and pit propagation in stainless steels. Journal of the Electrochemical Society, (12): p Mankowski, J. and Z. Szklarska-Smialowska, Studies on accumulation of chloride ions in pits growing during anodic polarization. Corrosion Science, (6 12): p Ernst, P. and R.C. Newman, Pit growth studies in stainless steel foils. I. Introduction and pit growth kinetics. Corrosion Science, (5): p Ghahari, S.M., et al., In situ synchrotron X-ray micro-tomography study of pitting corrosion in stainless steel. Corrosion Science, (9): p Ghahari, M., et al., Synchrotron X-ray radiography studies of pitting corrosion of stainless steel: Extraction of pit propagation parameters. Corrosion Science, : p Laycock, N.J. and S.P. White, Computer simulation of single pit propagation in stainless steel under potentiostatic control. Journal of the Electrochemical Society, (7): p. B264-B Laycock, N.J., et al., Perforated covers for propagating pits. Journal of the Electrochemical Society, (4): p Pistorius, P.C. and G.T. Burstein, Metastable Pitting Corrosion of Stainless Steel and the Transition to Stability. Philosophical Transactions of the Royal Society of London Series a-mathematical Physical and Engineering Sciences, (1662): p Alkire, R.C. and K.P. Wong, The Corrosion of Single Pits on Stainless-Steel in Acidic Chloride Solution. Corrosion Science, (4): p. 411-&. 14. Laycock, N.J. and R.C. Newman, Localised dissolution kinetics, salt films and pitting potentials. Corrosion Science, (10-11): p Tian, W.-M., et al., Pitting Kinetics of 304 Stainless Steel Using ESPI Detection Technique. Acta Metallurgica Sinica (English Letters), (4): p Newman, R.C. and E.M. Franz, Growth and Repassivation of Single Corrosion Pits in Stainless-Steel. Corrosion, (7): p González-Sánchez, J., et al., Corrosion pit growth on austenitic stainless steels in chloride containing solution: a quantitative approach. Anti-Corrosion Methods and Materials, (5): p Cook, A.B., et al., Pit Propagation in Pure Aluminum Investigated via the 1D Artificial Pit Technique: Growth Regimes, Surface Morphology and Implications for Stability Criteria. ECS Transactions, (25): p Gaudet, G.T., et al., Mass-Transfer and Electrochemical Kinetic Interactions in Localized Pitting Corrosion. Aiche Journal, (6): p Steinsmo, U. and H.S. Isaacs, The Dissolution and Repassivation Kinetics of Fe-Cr Alloys in Pit Solutions. Corrosion Science, (1-4): p

147 21. Moayed, M.H. and R.C. Newman, Deterioration in critical pitting temperature of 904L stainless steel by addition of sulfate ions. Corrosion Science, (11): p Pistorius, P.C. and G.T. Burstein, Growth of Corrosion Pits on Stainless-Steel in Chloride Solution Containing Dilute Sulfate. Corrosion Science, (12): p Ghahari, S.M., et al., Pitting corrosion of stainless steel: measuring and modelling pit propagation in support of damage prediction for radioactive waste containers. Corrosion Engineering Science and Technology, (2): p Burnett, T., et al., Correlative Tomography. Scientific reports, King, A., et al., Observations of intergranular stress corrosion cracking in a grainmapped polycrystal. Science, (5887): p Marrow, T.J., et al., High-resolution, in-situ, tomographic observations of stress corrosion cracking, in Environment-Induced Cracking of Materials, S.A. Shipilov, et al., Editors. 2008, Elsevier: Amsterdam. p Babout, L., et al., X-ray microtomographic observation of intergranular stress corrosion cracking in sensitised austenitic stainless steel. Materials Science and Technology, (9): p Feldkamp, L., L. Davis, and J. Kress, Practical cone-beam algorithm. JOSA A, (6): p Cavanaugh, M.K., R.G. Buchheit, and N. Birbilis, Modeling the environmental dependence of pit growth using neural network approaches. Corrosion Science, (9): p Tester, J.W. and H.S. Isaacs, Diffusional Effects in Simulated Localized Corrosion. Journal of the Electrochemical Society, (11): p Frankel, G.S., et al., Pit Growth in Nife Thin-Films. Journal of the Electrochemical Society, (8): p Moayed, M.H. and R. Newman, Using pit solution chemistry for evaluation of metastable pitting stability of austenitic stainless steel. Materials and Corrosion, (3): p Mankowski, J. and Z. Szklarskasmialowska, Effect of Specimen Position on Shape of Corrosion Pits in an Austenitic Stainless-Steel. Corrosion Science, (9): p Kuo, H.C. and D. Landolt, Rotating-Disk Electrode Study of Anodic Dissolution or Iron in Concentrated Chloride Media. Electrochimica Acta, (5): p Tian, W., et al., Effects of applied potential on stable pitting of 304 stainless steel. Corrosion Science, (0): p Carcea, A.G., et al., Anodic Kinetics of NiCr [Mo] Alloys During Localized Corrosion: I. Diffusion-Controlled Dissolution. Journal of the Electrochemical Society, (6): p. C215-C Frankel, G.S., The Growth of 2-D Pits in Thin-Film Aluminum. Corrosion Science, (12): p Sato, N., The Stability of Pitting Dissolution of Metals in Aqueous-Solution. Journal of the Electrochemical Society, (2): p

148 6.3 Estimation of 3D Growth Kinetics of Multiple Pits in Austenitic Stainless Steel (Manuscript C) F. A. Almuaili 1, S. A. McDonald 2, P. J. Withers 2, D. L. Engelberg 1 1 Corrosion and Protection Centre, School of Materials, The University of Manchester, Sackville Street, Manchester, M13 9PL, United Kingdom 2 Manchester X-ray Imaging Facility, School of Materials, The University of Manchester, Sackville Street, Manchester, M13 9PL, United Kingdom Abstract The growth of multiple corrosion pits on type 304L stainless steel was estimated by a quasi-in-situ X-ray computed tomography (X-ray CT) approach using electrochemical polarisation. A miniature electrochemical cell was used to polarize a wire sample in 0.1 M NaCl solution. The formation of eight discrete pits varying in size was observed under potentiodynamic and potentiostatic control. For each pit, 3D volume measured using X-ray CT data shows a good fit to the volume obtained using Faraday s law. Current density of 3D pit growth over time shows two regions suggesting stable pit transition under salt film formation and maximum current densities between 4 and 14 A.cm -2. Pit growth with typical mean current densities at steady state were ~1-2 A.cm -2 and associated pit stability products were above 0.3 A.m -1. The diffusivity parameter (DΔC) was estimated at mol.cm -1.s -1. Keywords: Pitting corrosion, potentiodynamic polarisation, stainless steels, X-ray computed tomography Introduction The integrity of stainless steel in halide environments has been found to affect component life and operational safety due to unexpected failures. Localised attack-and in particular pitting corrosion-often occurs due to metal dissolution in small areas of the metal surface, while the rest remains passive. Such attack may accelerate the process of pit penetration and induce leaks by perforation or stress concentration, which is considered a precursor of stress corrosion cracking (SCC). Pitting corrosion is typically related to the presence of non-metallic inclusions, which act as a preferred sites for pits 148

149 to initiate. In chloride solution local metal dissolution occurs after the breakdown of the passive film, which is referred to as pit initiation [1]. The second step is pit propagation, which requires a critical concentration of metal ions to be maintained inside the cavity. Low ph chemistry inside the pit due to hydrolysis is necessary to maintain active pit growth in chloride environments, which is associated with a metal ion concentration of approximately 60% or more [2]. In the 1970s, Galvele [3] proposed a critical quantity of active pit propagation based on current density and pit depth. Pistorious and Burstein [4] later suggested that the criterion for stable pit transition and growth without passivation was the pit stability product i.r, where i is current density and r is the radius of a hemispherical pit. In type 304 stainless steel the stability product for an open pit is given by 0.3 A. m -1 i.r 0.6 A. m -1 ; these estimated values are based on a minimum concentration of 75% and a maximum concentration (supersaturation concentration) of 150% of saturated concentration (4 M) of FeCl 2. Sato [5] differentiated between the shape of pit growth above and below pitting potential. High potential leads to a polished surface of the pit bottom under a salt film, with hemispherical pit shape, while active pit growth with an irregular etched shape develops under low potential. Pit shape is reported to change during anodic polarisation from hemispherical at an early stage of pit growth to an elongated dish shape over time. This is attributed to variation in the thickness of the salt film which is precipitated inside the pit. It is also suggested that the position of the sample influences pit shape because gravity affects the distribution of local environment concentration inside the pit [6]. During an early stage of pit growth, it is suggested that lacy metal cover plays an important role in the stability of pit growth. The cover supports and maintains the aggressive chemistry inside the pit, keeping dissolution active and maintaining the concentration of dissolved metal ions above the critical concentration [4]. The pit depth appears more important after salt film formation, compared to the early stages of pit growth. This then acts as diffusion depth, where a pit can grow without lacy cover. Pit growth has been investigated widely, based on one- and two-dimensional (1D and 2D) pit geometry [7, 8]. Pit growth under the local metal dissolution of electrochemical polarisation is related to ohmic activation (charge-transfer), solution resistance and mass transport of electrochemical reactants. The relationship between pit growth and current 149

150 over time was investigated with anodic polarisation and it was found that local pit concentration affects the kinetics of pit growth [9]. Laycock [10] used a 1D artificial pit electrode to study the transition potential from metastable to stable pit growth and found it to occur at current densities of 1-5 A.cm -2. At higher potential, pit growth was found to be under diffusion control with salt film precipitation, whereas at low potential (salt film free), pit growth occurs under activation and ohmic control. Induction time and current density of pit growth show a relationship with potential and chloride concentration. The total current evolution under polarisation increases rapidly as the number of pits is increased [11]. However, the current density of stable pit growth decreases relative to the early stages of growth, due to increased pit size. The growth kinetics is simply described by a power law (y = at x ) where y is pit depth or current, t is time, a is a constant depending on various parameters and b is a system-independent constant [12-14]. Alkire and Wang [15] measured the current density of single pit growth over time. They report a mean limited current density decay with increasing time following t -1/2 of pit growth and suggest mass transport control under a salt film. Newman and Franz [16] grew a single pit on 304 stainless steel in 1 M NaCl solution containing 0.04 M Na 2 S 2 O 3 at high potential. Their results show a good fit between dissolved pit volume obtained by Faraday s law and volume measured through an optical microscope assuming a hemispherical pit. They found that pit evolution current increased with t 1/2 below 100 s and linearly above 100 s. Assuming a hemispherical pit, Frankel estimated the corresponding current density to be related to t -1/2 below and t -1/3 above 100 s. Moayed and Newman [17] report that the current increase for a stable pit on type 904L stainless steel polarised at 750 mv in 1 M NaCl at 62 C was proportional to t 3/2. They observed a steady increase in current for 4 s, then a drop occurred, followed by a slow increase with time. Metastable pit current obeyed a power relation of I ~ t b, where b varies with increasing temperature from 0.5 to 1.5. Increasing the potential resulted in I ~ t for an open pit and t 0.5 with an occluded cavity [18], however Frankel suggests that increased current with a different potential is related to t 2, where the lifetime of pit growth may be between 5 and 15 s prior to transition [19]. After transition, the current increases with t 1/2 but mean current density is independent of potential [20]. Small single pits investigated by Strehblow and Wenners [21] showed linear growth in the early stages and the current density of iron was estimated for a hemispherical pit at 17 A.cm -2 in KCl 150

151 at ph 5, showing an approximately linear relation with potential. Earlier, the number of pits and their total current were reported to increase linearly with time in acid solution proportional to t 3, while the current from single pits was not considered [22]. Electrochemical measurements with metallographic observation methods were widely used to estimate pit growth in 1D and 2D, and it was suggested that a combination of these methods was useful, especially for analysing multiple pit growth [16]. In addition, real pits grown in 3D, compared to limited growth in 1D and 2D, may act differently in a bulk environment [23, 24]. In a previous study [25], 3D pitting growth was investigated, resulting in the growth of three pits, which were observed by a quasi-insitu 3D method using X-ray CT. This study found i.r above 0.3 A.m -1 and the diffusion product (DΔC) between 1.68 and mol.cm -1.s -1. The pits were of elongated dish shape and grew with mean current densities of 1-3 A.cm -2. González-Sánchez [26] estimated pit growth on 304L and 316 steels in 3.5 wt% NaCl and natural seawater under potentiostatic polarisation in the range of mv vs SCE. The pits were initiated above the pitting potential, then grown below the pitting potential. The pit volume was obtained by electrochemical measurement of current using Faraday s law and pit profile measurement using optical microscopy. The results show a reasonable fit between these methods assuming that the pit is hemispherical and it was suggested that maximum pit depth (r) is given by the equation r = a bc t. In addition, pit growth on 304 and 316 samples in artificial seawater was described by the equations r = 27.9t 0.67 and r = 26t 0.6 respectively. An earlier study of pit growth on 12% Cr martensitic stainless steel in 0.1 M NaCl suggested a power equation of r = 48t 0.53 [27]. Tian et al. [28] studied the kinetics of pit growth using a vertical sample of 304 stainless steel polarised at 350 mv vs SCE in 3.5 wt% NaCl by combining the measurement of pit current evolution and images recorded by the electronic speckle pattern interferometer system. Their results suggest diffusion control of pit growth with dish morphology. Pitting current density for a single pit had a higher value of 30 A.cm -2 at 18 s, followed by s of decay to 2-8 A.cm -2 with salt film precipitation, while the stability product indicated stable pit growth. In another experiment in a similar environment, these researchers studied the effect of varying the applied potential in the range of mv vs SCE, concluding that open pit morphology is related to high 151

152 potential and current density. They suggest that even when the pit was under diffusion control, an ohmic drop occurred between pit bottom and mouth [29]. The aim of the present study was to observe the growth of multiple pits in situ. It combines for the first time in-situ 3D shape information on multiple pits from X-ray CT with electrochemical polarisation measurements in bulk chloride solution. Pit growth characteristics were obtained by estimating current over time. The second objective was to estimate pit growth kinetics, stability product and morphology of multiple pits, then to compare these with data from the literature Experimental Set-up A 70 mm length of annealed type 304L stainless steel wire with a diameter of 500 μm was used to conduct in-situ electrochemical polarisation experiments in aerated 0.1 M NaCl solution. The stainless steel wire had a wt% chemical composition of 18.4 Cr, 8.7 Ni, 0.02 C, 1.4 Mn, 0.34 Si, 0.04 N, 0.03 P and S. The wire sample was surface ground with 1200 grit SiC paper, then rinsed in deionised water. Its surface was coated with beeswax, except for a 1 mm length, leaving a cylindrical area of 1.57 mm 2 which was then exposed to the chloride solution. The wire sample was mounted vertically in a miniature electrochemical cell as shown in Figure 6-6, then the setup was interfaced with a ZEISS Xradia 520 Versa microtomography instrument. The cell, of inner diameter 24 mm, consisted of a lower part containing the electrolyte with a volume of approximately 9 ml, for electrochemical polarisation measurements, and an upper part with straining rig for applying a tensile load along the length of the wire. The upper part was used in another study to investigate the effect of strain on pit growth kinetics. The lower part of the cell housed a miniature reference electrode (Ag/AgCl, 3 M NaCl) and a miniature platinum counter electrode. Electrochemical polarisation tests were performed at a scan rate of 1 mv.s -1 using an Ivium CompactStat Potentiostat. Prior to polarisation, the open circuit potential (OCP) was monitored for 15 min. The current response was recorded during the polarisation experiment at a rate of 1 Hz. Table 6-3 gives a summary of the in-situ experimental methodology. After each polarisation cycle, one X-ray CT scan was performed to visualise the progress of pitting corrosion over time. X-ray CT scans were recorded at OCP, with each scan taking 152

153 approximately 2-3 hrs. The sample remained in the electrolyte throughout the experiment. For the X-ray CT measurements an accelerating voltage of 120 kv was used and 721 projections were recorded with 2 2 binning at 10 optical magnification, allowing 12 seconds for each projection. This resulted in a reconstructed voxel size of 1.1 µm 3 (giving a spatial resolution of 2-3 μm), with a field of view of 1100 μm 1100 μm. The data were reconstructed using the Feldkamp-Davis-Kress approach [30] and images were segmented and visualised using Avizo software. After segmentation of each pit, the total volume, total pit surface area, pit depth, width, height, aspect ratio and shape were obtained. The pit depth, width and height were measured from 2D slices of the 3D dataset taken from the pit centre with respect to their edges. This approach provided a snapshot of pit dimensions after each polarisation cycle, with the current response recorded during the polarisation experiment providing in-situ pit growth information. The wire was removed from the cell after step 9 (Table 6-3) and rinsed in water, then images of the pits were obtained using an FEI Quanta 650 scanning electron microscope (SEM). Table 6-3: Summary of in-situ electrochemical polarisation experiment Step Polarisation experiment 0 OCP measurement for 15 min 1 1 st potentiodynamic polarisation from OCP to +600 mv vs Ag/AgCl 2 X-ray CT scan at OCP (scan 1) 3 OCP measurement for 5 min 2 nd potentiodynamic polarisation from OCP to +600 mv, followed by 4 potentiostatic polarisation at +600 mv for 3 min 5 X-ray CT scan at OCP (scan 2) 6 OCP measurement for 5 min 3 rd potentiodynamic polarisation from OCP to +600 mv, followed by 7 potentiostatic polarisation at 600 mv for 3 min 8 X-ray CT scan at OCP (scan 3) 9 Experiment terminated and sample removed for SEM analysis 153

154 Pt counter electrode X-ray source Straining rig Ag/AgCl reference electrode Working electrode Detector Figure 6-6: Image of the miniature electrochemical cell and a type 304L wire sample inside the X-ray CT equipment Results and Discussion The first potentiodynamic polarisation (step 1, Table 6-3) resulted in the formation of four corrosion pits of different sizes (pits 1-4). Figure 6-7(a, b) shows 3D views of the reconstructed wire from X-ray CT scan 1, indicating pits 1-4. A second polarisation cycle (step 4, Table 6-3) was performed on the same wire without removing it from the solution, generating four new pits (5-8). Figure 6-7(c, d) shows reconstructed tomography images of the wire with all eight pits (1-8). The results indicate that pits 1-4 remained without further growth or noticeable changes after the second polarisation cycle. A third cycle (step 7) was performed but no current evolution was recorded and the subsequent X-ray CT scan showed only the existing eight pits without changes. The spatial resolution of X-ray CT is not sufficient to show the lacy metal covers in Figure

155 Pit 1 Height (h) Width (w) Pit 2 Pit 3 Pit 4 (a) Side A (b) Side B Pit 1 Pit 6 Pit 3 Pit 2 Pit 7 Pit 5 Pit 8 Pit 4 (c) Side A (d) Side B Figure 6-7: Reconstructed X-ray CT images of the two sides of the wire: (a, b) after the 1 st electrochemical polarisation scan; (c, d) after the 2 nd potentiodynamic polarisation scan. 155

156 Pit morphology Figure 6-8(a) is a cross-sectional 3D isosurface view of the interior morphology of all eight pits and Figure 6-8(b) is a magnified view of pit 5, the largest pit, clearly showing the development of 3D lobes. Three lobes appear to have grown below each other and to have failed to emerge completely at the surface. This indicates that one or more lobes can develop underneath the growing pit before emerging completely, which was demonstrated in a 2D study of pit growth on a horizontal sample facing upward [7]. This may be explained by differences in behaviour between vertical and horizontal samples as a result of gravity affecting the chemistry and salt film thickness inside the pit [6]. The morphology of pit 5 was also observed by 2D slice along the pit height and width, depicted respectively in Figure 6-9(a, b), indicating only a single lobe in the pit height view. This apparent difference in morphology indicates the advantage of 3D over 2D observation of pit morphology. The 2D image shows a smooth pit bottom similar to the reconstructed image in Figure 6-7(c), indicating metal growth under a salt film, consistent with polished morphology of pits grown under diffusion control at high anodic potential [5, 8]. Figure 6-10 shows tomogram sections of the wire mounted in the in-situ cell, with the vertical axis shown (arrow). The effect of gravity is assessed by comparing the shape of the pit at lower part with upper part. The result indicate that pits have almost similar edge shape at the lower part and upper part but the propagation of the lobes appears to be faster and wider at upper side relative to the lower side. Pit 6 shows at the lower side far narrower pit morphology relative to upper side. The lobes grow faster at the upper side in both pits 5, 7 and 8. The above results indicate variation between the upper side and the lower side during the pit growth process but without deform in the edge of lower part compare to upper part. It appears that the lobes development affect the shape of the pits. This depends on the thickness of salt film layer inside the pit. The thick layer at the lower part of the pit reduces the growth of the pit compare to the thin layer at upper part. Therefore, the difference of pit morphology can be attributed to the effect of gravity which leads to build up concentration of dissolved metal ions in the lower part of the pit. Build-up of metal ion concentration inside the pit is also depends on the size of the pit and the lacy cover (the size of pit mouth and the holes). In this results, polarization at high potential 156

157 for short time leads to fast polished surface and make the etching in the lower edge of the pit by flow of corrosive solution [6] less significant under effect of gravity. After the test, SEM images of the wire were taken from both sides, to confirm the presence of the eight pits and their surface morphology. The images indicate that pits grew with typical lacy metal covers around the mouth (Figure 6-11(a, b). The images also indicate the position of the beeswax cover at either end of the sample and a mark on one side which may be contamination deposited during handling before imaging. (a) 3D of three lobes Figure 6-8: Cross-sectional isosurface view of wire sample: (a) all pits, (b) close-up view of pit 5 morphology. (b) (a) (b) Figure 6-9: Tomogram section of 2D pit from middle slice of pit 5 at (a) height and (b) width view. 157

158 Pit 1 Pit 2 Pit 3 Pit 4 Pit 5 Pit 6 Pit 7 Pit 8 Figure 6-10: 2D tomograms of pits along the middle axis of each pit, showing the shape as a function of pit orientation. All pits grew during exposure of the wire in the in-situ cell, with the vertical axis shown by the arrow. 158

159 Pit 1 Pit 6 Pit 2 Pit 3 Pit 7 Pit 5 Pit 8 Pit 4 (a) Side A (b) Side B Figure 6-11: SEM images of the two sides of the wire sample, showing all eight pits Electrochemical polarisation After OCP had stabilised at -29 mv vs Ag/AgCl (step 0, Table 6-3), the first potentiodynamic polarisation (step 1) was performed from OCP to +600 mv at a 1 mv.s -1 scan rate. Figure 6-12(a) shows the time-dependent current response obtained during this scan; the current started to increase abruptly at +485 mv and took 117 seconds. The OCP measurement (step 3) was +135 mv prior to the second polarisation (step 4) using potentio-dynamic polarisation to +600 mv vs Ag/AgCl, followed by potentiostatic control at +600 mv for 3 min. Figure 6-12(a) shows current response during the second polarisation; the current started to rise after 38 s (zero time in Figure 6-12a) during the potentiostatic control period, which lasted for 141 seconds. A third polarisation cycle (step 7) was performed from OCP +160 mv but no current response was observed, which means that no pits were formed during this cycle. This may be attributed to a limited number of active sites such as inclusions or the development of more protective oxide film after polarisation. The OCP shifted by +149 between the first and second polarisations and by +25 mv between the second and third. This potential shift may be a consequence of the passive film growth which is often associated with anodic polarisation, or alternatively of the influence of the high energy X-ray beam, which may have induced chemical changes in the passive film. However, the fact that the OCP shift between the second and third 159

160 polarisations was smaller than that between the first and second may indicate that the passive film which formed was more compact and dense in the former case. Such shifts in OCP were also observed after polarisation without removing the specimen from the solution in a previous study [25]. The difference between the two curves in Fig 6-12(a) shows that higher polarisation potential leads to higher current value. This leads to faster metal dissolution and the growth of larger pits size. A significant change in the gradient of current over time was used to indicate the initiation of new pits, as reported in the next section. (a) 160

161 (b) Figure 6-12: (a) Current vs. time during 1 st and 2 nd potentiodynamic and potentiostatic polarisation cycles (see appendix C for potential vs current); (b) depth, width and height of pits calculated from X-ray CT data Pit geometry Figure 6-12(b) shows the maximum pit dimensions (depth, width and height) at the end of polarisation, obtained from X-ray CT data. Pits 5-8, generated during potentiostatic polarisation, were larger than pits 1-4, which developed during the first potentiodynamic polarisation. Surface area and volume of all pits were also measured via segmentation and are listed in Table 6-4. Beside the direct measurements from X-ray CT data, another approach B was also explored to obtained pits dimension and used to estimate pit growth kinetics. The detailed of the values obtained in Table 6-4 via approach A and B explained below: In approach A, the measurements were obtained from the segmented X-ray CT data. The area of the pit surface was obtained by measuring the surface area of the segmented pit volume, excluding the area of the pit cover. This represented the real area of the internal pit surface. In approach B, pits volume were estimated from the charge, determined from current vs time data by Faraday s law. Before conducted this measurement, the current over time for each pit was obtained from the current response curve (Figure 6-12a) using assumption below: 161

162 From Figure 6-12(a) the current estimated for each pit based on the significant gradient change, which considered as initiated time for a new pit. Therefore, the following assumptions were considered to separate the contribution of new pit current over time to the total evolution current: 1. The largest pit was assumed to nucleate first, to stay active until the end of the polarisation cycle and to grow at a constant rate. 2. The second pit was assumed to initiate at the time when a significant gradient change occurred. This pit then stayed active until the end of polarisation and was smaller size than the first. This assumption also applies to the third pit. 3. Pits of similar volume were assumed to initiate at the same time and to grow at similar rates; therefore the current was divided between them equally. 4. After identification of the starting time of each pit, the current over time was divided after the first gradient change between the first and second pits using a fixed ratio. The above assumptions are justified by the finding of a good fit between segmented pit volume from X-ray CT data (approach A) and the dissolved metal volume of each pit obtained by Faraday s law (approach B) after the total current was allocated among the pits over time. The fixed ratios which satisfied these assumptions were a quarter (1/4) of total current taken by each new pit initiated over time during potentiodynamic polarisation and a half (1/2) in case of potentiostatic polarisation. Figure 6-13(a, b, c, d) shows the result of the current separation over time for the first potentiodynamic polarisation and during the potentiostatic region of the second cycle. Figure 6-13(a, c) shows the position of gradient changes with gradient value and total volume of dissolved metal, while Figure 6-13(b, d) depicts the separated curves of each pit current over time. The drop in current associated with an early stage of new pit initiation appears as an artefact of applying separated current using fixed ratio. 162

163 (a) (b) 163

164 (c) (d) Figure 6-13: The estimated current evolution of each pit based on the total current. The initiation point for pits with (a, b) potentiodynamic and (c, d) potentiostatic polarisation are shown. More information of the current separation for each pit is given in Appendix C. Pit volume was determined by integration of the transferred current charge to obtain the mass of dissolved metal using Faraday s law. The calculation assumed a stoichiometric electron-exchange value of n = 2.19, atomic weight M = g.mol -1, density of ρ = 164

165 7.97 g.cm -3 and Faraday s constant F = coulomb/mol [4]. Current efficiency was assumed to be 100%. The volume of the growing pit over time was derived from the evolution of the pit current over time for each pit. Then the growth of the pit in radius (depth) over time and the surface area evolution curve were estimated assuming pit growth to be hemispherical. Table 6-4 summarises the results obtained by approaches A and B. Table 6-4: Geometry and size of pits generated by potentiodynamic (pits 1-4) and potentiostatic polarisation (pits 5-8) Methods A: measurements taken from X-ray CT scans Depth, r Width, w Height, h Volume, V Pits (μm) (μm) (μm) (μm 3 ) Surface area without lacy cover (μm 2 ) Pit Pit Pit Pit Pit Pit Pit Pit Pit 1 56* 112* 112* * B: Faraday approach + assumption of hemispherical shape Pit 2 30* 60* 60* * Pit 3 22* 44* 44* * Pit 4 22* 44* 44* * Pit 5 66* 132* 132* * Pit 6 48* 96* 96* * Pit 7 48* 96* 96* * Pit 8 48* 96* 96* * * Estimated values in bold The above values of surface area and depth evolution over time for each pit obtained via approach B were used to calculate the pit surface area and depth over time for approach A. The calculation relied on back extrapolation of pit surface area and depth values from the end polarisation measurement of approach A in Table 6-4, assuming symmetrical pit growth. Pit surface area and depth over time were then used to 165

166 determine the current density and stability product for each pit via approaches A and B (see sections and ). The results in Table 6-4 show a good fit between the segmented pit volume of approach A and the volumes obtained via Faraday s law in approach B, with a typical difference of less than 10%. This difference may be explained by cathodic reactions occurring within the pit, which are not accounted for in these calculations and which would compensate part of the anodic response by reducing the measured anodic current density. For example, 5% of cathodic current was assumed to occur inside the pit by hydrogen evolution [4]. Another source of error may be inaccurate pit segmentation from the X-ray data Pit growth kinetics The relationship of pit growth geometry to time can be estimated from the equation r = a. t b, where r represents pit depth, a and b are empirical pit growth constants and t is time. The results show values of pit growth in depth for potentiodynamic and potentiostatic polarisation of r = t 0.43 and r = t 1.29 respectively. Similarly, lateral pit growth in width (w) and height (h) was found to be w = 11.7t 0.47, h = 11.61t 0.49 for potentiodynamic polarisation and w = 0.57 t 1.14, h = 0.22t 1.34 for potentiostatic polarisation. These results show that pits grow faster laterally than in depth, leading to elongated dish shapes with aspect ratios of pit depth to width and height of and respectively. This suggests that pits grow under diffusion control, with ohmic resistance controlling. Aspect ratio show slight variation related to the pit size as shown in Figure 6-12b. At high anodic potential, pit growth occurs under the formation of a salt film, but vertical sample position can cause the accumulation of dissolved metal ions at the lower side of the pit due to gravity. In the presence of a passive oxide film, a metal cover can act as a physical barrier that maintains the metal solution at saturation concentration. This may prevent the solution from escaping from the pit holes until the lacy metal cover cracks or collapses at its lower side when the pit increases in size, as can be seen in pit 1 (Figure 6-11). Similar fractures were observed in metal lacy covers over pit mouths in a vertical sample in acid solution [31]. Alternatively, a new hole may form in the metal cover as a consequence of pit growth by an undercutting lobe, allowing metal ions to diffuse out rapidly from the lower side of the pit in a vertical sample due to 166

167 gravity [28]. Pit growth in depth is more likely to be under diffusion control compared to lateral pit growth [7]. Slight differences in growth between width and height appear to result from the influence of gravity on a vertical sample, while drawing direction may have a similar effect [28]. The higher potential of pit growth under potentiostatic polarisation is associated with faster growth and larger pits compared to potentiodynamic pit growth. This may also indicate that high metal dissolution potential leads to larger 3D pit growth compared to 2D [7]or 1D [9] growth due to constrained geometry which leads to fast saturation. However, the results indicate that slow pit growth is associated with diffusion control and that rapid growth does not occur under diffusion control. Lateral growth was found to be twice as great as depth growth under potentiodynamic polarisation, with slightly less difference under potentiostatic polarisation. The polished pit bottom surfaces and elongated dish shapes in Figure 6-9(a, b) imply the presence of a salt film, which is often associated with pit growth under diffusion control. Lateral pit growth appears to be under ohmic control; however, a vertical sample position can cause supersaturation of metal salts at the lower side of the pit, affecting pit growth. Figure 6-12(b) shows a slight variation between pit width and height for all smaller pits compared to larger pits under potentiodynamic polarisation, while all pits under potentiostatic control tended to be larger than pits under potentiodynamic control. A difference in lateral growth was also observed. This clarifies that the pit shape is not necessarily hemispherical, with 2D X-ray CT slices showing an elongated dish shape (Figure 6-9(a, b)). The ratio of pit depth to width was in the range of 0.4 to 0.48 for potentiodynamic polarisation and for potentiostatic polarisation, fitting closely to reported values of 0.4 and 0.5 [15, 32] Current density The current densities associated with pit growth over time were obtained by dividing the current response of each pit, as given in Figure 6-13(b, d), by the internal pit surface area as listed in Table 6-4. The calculation of pit surface area and depth over time via approaches A and B is explained in section Figure 6-14(a, b, c, d) shows the values of mean current density over time for pits 1-4 under potentiodynamic control and pits 5-8 under potentiostatic control, using approaches A and B. Two regions of pit growth over time can be distinguished from the 167

168 current density curve. First, there is a sharp increase in current to a maximum value at an early stage of pit growth, followed by a second region where the current decays over time and fluctuations in current density are reduced over time. These regions represent successive transient and quasi-steady-state periods, which have been reported to occur because current density is limited by salt film formation [9]. Figure 6-14(a, b) plots the current density of pits under potentio-dynamic control, obtained by approaches A and B respectively. For pit 1, the maximum current density reached at the end of the transient region, according to approach A, was 4.4 A.cm -2, reducing to a mean value of 1 A.cm -2 in the a quasi-steady state region at the end of polarisation. Pits 2, 3 and 4 initiated at higher potential than pit 1 and the graphs show maximum current densities between 14 and 13 A.cm -2, but these all underwent rapid decay with increasing pit size, and mean current density at the end of polarisation reached values between 1.3 and 1.6 A.cm -2. The green trace in each graph represents average current density for all four pits over time, obtained by dividing the total current response by the total pit surface area. These results indicate a slight increase in total current density after each new pit initiation. (a) 168

169 (b) (c) 169

170 (d) Figure 6-14: Current density over time for pits 1-4 under potentiodynamic polarisation by (a) approach A and (b) approach B, and for pits 5-8 under potentiostatic polarisation by (c) approach A and (d) approach B. Figure 6-14(c, d) plots the current density of pits 5, 6, 7 and 8, grown under potentiostatic control, using approaches A and B. For pit 5, current density reached a maximum of 6.5 A.cm -2, higher than for pit 1, which was generated at a lower potential. As pit size increased, this value declined to 0.74 A.cm -2 at the end of the experiment. The three remaining pits (6, 7 and 8) are assumed to have grown simultaneously, because they had almost similar volumes and maximum current densities of A.cm -2 which decayed rapidly to a steady state region of A.cm -2. Such rapid decay may indicate quick salt film precipitation, causing pit growth to fall under diffusion control. It may also indicate that pit growth initially proceeded by lobes undercutting the metal surface. The maximum difference in measured current density between approaches A and B was found to be 6% at current transition and at the end of polarisation in the case of potentio-dynamic polarisation, with equivalent values of 16% for potentiostatic polarisation. The difference between approaches A and B was also found to increase with increasing pit size. This phenomenon, insofar as it applies to values at the end of polarisation, may be attributed to the fact that approach A took account of the actual elongated pit shape, whereas approach B assumed that pits were hemispherical. 170

171 Stability product The stability product for stable pit growth in 2D on 304 stainless steel is reported to be in the range 0.3 < i.r < 0.6 A.m -1 [4]. Figure 6-15(a,b, c, d) shows the stability products obtained from 3D pit growth over time using approaches A and B, for pits generated under potentio-dynamic and potentiostatic control. The results show that pit stability values at an early stage of pit growth were below stability criteria. This suggests that pit growth was supported by lacy metal cover acting as a diffusion barrier and maintaining a critical concentration of corrosive environment [4, 33]. This is confirmed by the SEM images in Figure 6-11, which show that all pits developed with lacy metal covers. After that, pit stability increased over time above 0.3 A.m -1 and remained within the range of stable pit criteria. Figure 6-15(a, b) shows that the stability product values of pits 1-4, generated under potentiodynamic polarisation, were lower than for pits 5-8, generated under potentiostatic polarisation (Figure 6-15(c, d)). This appears to be a consequence of higher polarisation potential under potentiostatic than potentiodynamic polarisation. This effect can also be observed in pit 5, which reached stability criteria faster than pit 1, indicating higher and faster metal dissolution over time. The stability product of pit 1 fell briefly below the stable pit criteria due to new pit initiation, which appears to be an artefact of current density from the current separation method described above at the early stage of pit growth. However, it then soon increased to within the stable range, while the value for the new pit fell steeply from its maximum. The pit stability values obtained by approach A are lower by around 0.1 A.m -1 than those calculated by approach B, a difference consistent with our earlier observations [25]. This difference appears to be due to the overestimation of pit depth by method B relative to the real pit depth employed in method A. Therefore, this variation can be related to the difference pit shape between real pit dimensions (approach A) and the estimated hemispherical pit shape (approach B). It appears that 2D methods of measuring pit kinetics reported in the literature overestimate the current density and underestimate the pit stability product by assuming pits to be hemispherical. 3D pit geometries using actual surface areas and volumes give far better estimations of current density and associated stability products. 171

172 (a) (b) 172

173 (c) (d) Figure 6-15: Stability product over time of multiple pits on 304L stainless steel wire in 0.1 M NaCl solution grown under potentiodynamic polarisation by (a) approach A and (b) approach B, and under potentiostatic polarisation by (c) approach A and (d) approach B Estimation of pit diffusion product Figure 6-16(a, b) shows the square of pit depth (r 2 ) over time (t) for pits grown under potentio-dynamic and potentiostatic polarisation respectively. The results were obtained using Equation 6-3 (derived from Fick s first law and Faraday s second law), suggesting a linear relation under diffusion control [4, 8]. r 2 = 3M D C πρ t (6-3) Where atomic weight M = g.mol -1, density of ρ = 7.97 g.cm -3, D is the effective diffusion coefficient and C is the difference in concentration between pit bottom and pit mouth. Table 6-5 lists diffusion products (DΔC) for all of the pits, obtained from the gradients in Figure 6-16(a, b). It also lists the values of D between pit bottom and mouth, estimated by assuming a saturated salt concentration at the pit bottom of 4.2 M and neglecting the concentration at the mouth. 173

174 (a) (b) Figure 6-16: Graphs of depth 2 over time and values of diffusion product obtained from their gradient for multiple pits grown under (a) potentiodynamic and (b) potentiostatic polarisation. 174

175 Table 6-5: Diffusion product (DΔC) and effective diffusion coefficient (D) Pits D ΔC D (cm 2.s -1 ), assuming ΔC=4.2 M (mol.cm -1.s -1 ) Gradient in equation 6-3 From the gradient From equation 6-4 (at the end of polarisation) Pit Pit Pit Pit Pit Pit Pit Pit The diffusion product results are consistent with those of earlier experiments conducted under the same conditions [25] and with reported results of diffusion-controlled pit growth in 2D [8], but lower than values for 1D pit growth [7] except for pit 5, which had a high diffusion product of mol.cm -1.s -1, close to reported values for 1D pit growth [34]. This may be considered a consequence of the effect of gravity on the largest pit volume generated in 3D in the vertical position under a high potential, allowing the dissolved metal ions to escape more rapidly than when constrained by the 2D or 1D morphology of pits grown facing upward [6]. This effect can be observed by comparing pit 5 with pit 1, which was half its size but was the largest of the pits grown under a lower potential and had the lowest diffusion product ( mol.cm -1.s -1 ). In addition, the greater size of pit 5 relative to pits 6-8 is associated with an increased size of pit mouth, which often reduces the diffusion barrier effect of the lacy metal cover, allowing more dissolved metal ions to escape from the pit. Thus, three variables appear to play important roles in increasing the diffusion product: the effect of gravity on vertical samples, the higher dissolution rate under a higher potential and the increased size of the pit mouth and holes at lacy metal cover. If D is assumed to be cm 2.s -1 [35], the mean saturation concentration inside the pit can be estimated from the diffusion product values in Table 6-5. The mean concentration would therefore be between 3 and 3.9 M, except for pit 5, which had a metal salt concentration of 8 M. However, these values are above the 150% saturation 175

176 concentration of FeCL 2, which is often considered a criterion for the stable transition of an open pit in stainless steel [4] The diffusion coefficient was also obtained from the diffusion product values listed in Table 6-5, assuming a saturated metal concentration of 4.2 M at the pit bottom while neglecting the effect of concentration at the pit mouth. The results, listed in Table 6-5, show D between and cm 2.s -1 for all pits except pit 5, for which D= cm 2.s -1. This result is slightly higher than the value reported for a 1D pit grown vertically [9]. The variation of diffusivity with pit growth over time can also be estimated using equation 6-4 [4], assuming a 4.2 M saturation concentration. The values of diffusion coefficient over time are plotted in Figure C = 2π 3nFD i. r (6-4) These results are correct for pit growth after transition, which was under diffusion control, but do not take accurate account of the pit initiation period, because this was not under diffusion control. The diffusion coefficient results in Figure 6-17(a, b) at the end of the steady state period of pit growth are listed in Table 6-5. The D values are close to the estimated diffusion coefficients for 1D and 2D pit growth, except for pits 1 and 5, where values were higher but consistent with 1D pit growth in the vertical position [9]. The diffusion coefficient of 3D pit growth on vertical samples was found to increase with increasing pit size. This suggests that pit growth may not be under full diffusion control, due to differences in the thickness of salt film developed inside the pit. The pit depth effect is reduced in vertical samples relative to horizontal ones, as diffusion length reduces the retraction by diffusion of dissolved metal ions. In addition, a small variation in diffusion coefficient between pits of similar size may be attributed to the perforated cover, which restricts metal ion diffusion [8]. 176

177 (a) (b) Figure 6-17: Diffusion coefficient over time, obtained by approach A, for multiple pits generated on 304L stainless steel wire in 0.1M NaCl solution, under (a) potentiodynamic and (b) potentiostatic polarisation Conclusions 1-3D pit growth kinetics can be estimated for multiple pits via quasi-in-situ X-ray CT measurements and electrochemical polarisation tests. Pit volumes obtained by X-ray CT showed a good fit with estimates obtained via metal dissolution by Faraday s law. 2-3D lobes undercutting the metal surface were observed to be a feature of pit growth and 3D pit growth rates revealed faster growth laterally than in depth, suggesting that pit depth is under diffusion control. 177

178 3- Current density of pit growth over time can be characterised by two regions corresponding to transition regions, suggesting the formation of a salt film. 4- Typical mean current densities of ~1-2 A.cm -2 were obtained, with pit stability product above 0.3 A.m -1, consistent with stable pit growth. 5- Effective diffusion coefficients between and cm 2.s -1 were obtained for the diffusion-controlled region of pit growth Acknowledgements The authors acknowledge the provision of beam time at Henry Moseley X-ray imaging Facility at the University of Manchester, UK, established with funding from EPSRC through grants EP/F007906, EP/I02249X and EP/F We would also like to thank the Saline Water Conversion Corporation of Saudi Arabia for its financial support References 1. Soltis, J., Passivity breakdown, pit initiation and propagation of pits in metallic materials - Review. Corrosion Science, : p Frankel, G.S., Pitting corrosion of metals; A summary of the critical factors. Proceedings of the International Symposium on Pits and Pores: Formation, Properties, and Significance for Advanced Luminescent Materials, (7): p Galvele, J.R., Transport Processes and Mechanism of Pitting of Metals. Journal of the Electrochemical Society, (4): p Pistorius, P.C. and G.T. Burstein, Metastable Pitting Corrosion of Stainless Steel and the Transition to Stability. Philosophical Transactions of the Royal Society of London Series a-mathematical Physical and Engineering Sciences, (1662): p Sato, N., The Stability of Localized Corrosion. Corrosion Science, (12): p Mankowski, J. and Z. Szklarska-Smialowska, Effect of Specimen Position on Shape of Corrosion Pits in an Austenitic Stainless-Steel. Corrosion Science, (9): p Ernst, P. and R.C. Newman, Pit growth studies in stainless steel foils. I. Introduction and pit growth kinetics. Corrosion Science, (5): p Ghahari, M., et al., Synchrotron X-ray radiography studies of pitting corrosion of stainless steel: Extraction of pit propagation parameters. Corrosion Science, : p Tester, J.W. and H.S. Isaacs, Diffusional Effects in Simulated Localized Corrosion. Journal of the Electrochemical Society, (11): p Laycock, N.J. and R.C. Newman, Localised dissolution kinetics, salt films and pitting potentials. Corrosion Science, (10-11): p Hunkeler, F. and H. Bohni, The Significance of Pit Induction Times. Corrosion, (10): p Rosenfeld, I.L. and I.S. Danilov, Electrochemical aspects of pitting corrosion. Corrosion Science, (3): p

179 13. Böhni, H., Breakdown of passivity and localized corrosion processes. Langmuir, (6): p Bohni, H., T. Suter, and A. Schreyer, Microtechniques and Nanotechniques to Study Localized Corrosion. Electrochimica Acta, (10): p Alkire, R.C. and K.P. Wong, The Corrosion of Single Pits on Stainless-Steel in Acidic Chloride Solution. Corrosion Science, (4): p. 411-&. 16. Newman, R.C. and E.M. Franz, Growth and Repassivation of Single Corrosion Pits in Stainless-Steel. Corrosion, (7): p Moayed, M.H. and R.C. Newman, Evolution of current transients and morphology of metastable and stable pitting on stainless steel near the critical pitting temperature. Corrosion Science, (4): p Ezuber, H. and R. Newman, Growth-rate distribution of metastable pits. Critical Factors in Localized Corrosion, 1991: p Frankel, G.S., et al., Metastable Pitting of Stainless Steel. Corrosion, (7): p Moayed, M.H. and R.C. Newman, The relationship between pit chemistry and pit geometry near the critical pitting temperature. Journal of the Electrochemical Society, (8): p. B330-B Strehblow, H.H. and J. Wenners, Investigation of the processes on iron and nickel electrodes at high corrosion current densities in solutions of high chloride content. Electrochimica Acta, (4): p Szklarska-Smialowska, Z., Review of Literature on Pitting Corrosion Published Since Corrosion, (6): p Frankel, G.S., The Growth of 2-D Pits in Thin-Film Aluminum. Corrosion Science, (12): p Frankel, G., Pitting corrosion of metals a review of the critical factors. Journal of the Electrochemical Society, (6): p Almuaili, F.A., et al., Application of a Quasi In Situ Experimental Approach to Estimate 3-D Pitting Corrosion Kinetics in Stainless Steel. Journal of the Electrochemical Society, (13): p. C745-C González-Sánchez, J., et al., Corrosion pit growth on austenitic stainless steels in chloride containing solution: a quantitative approach. Anti-Corrosion Methods and Materials, (5): p Zhou, S. and A. Turnbull, Influence of pitting on the fatigue life of a turbine blade steel. Fatigue and Fracture of Engineering Materials and Structures, (12): p Tian, W.-M., et al., Pitting Kinetics of 304 Stainless Steel Using ESPI Detection Technique. Acta Metallurgica Sinica (English Letters), (4): p Tian, W., et al., Effects of applied potential on stable pitting of 304 stainless steel. Corrosion Science, (0): p Feldkamp, L., L. Davis, and J. Kress, Practical cone-beam algorithm. JOSA A, (6): p Wu, K., W.-S. Jung, and J.-W. Byeon, In-situ monitoring of pitting corrosion on vertically positioned 304 stainless steel by analyzing acoustic-emission energy parameter. Corrosion Science, : p Ulrich, R.K. and R.C. Alkire, The initiation of single corrosion sites on iron by focused laser illumination. Corrosion Science, (11): p Pistorius, P.C. and G.T. Burstein, Growth of Corrosion Pits on Stainless-Steel in Chloride Solution Containing Dilute Sulfate. Corrosion Science, (12): p

180 34. Moayed, M.H. and R. Newman, Using pit solution chemistry for evaluation of metastable pitting stability of austenitic stainless steel. Materials and Corrosion, (3): p Gaudet, G.T., et al., Mass-Transfer and Electrochemical Kinetic Interactions in Localized Pitting Corrosion. Aiche Journal, (6): p

181 7. EFFECT OF STRAIN ON THREE- DIMENSIONAL PITTING CORROSION KINETICS The work presented in this chapter investigates the effect of strain on 3D pitting corrosion behaviour. For the first study (Manuscript D, section 7.2), the effects of 5% plastic strain followed by electrochemical repolarisation of an existing pit were characterised, and then the electrochemical current response was recorded to estimate pitting corrosion kinetics. The study indicated the reactivation of an existing pit, with the pit size obtained during the first polarisation cycle increasing significantly after reactivation. The growth rate during re-activation was greater than that of the pit grown before plastic strain was applied. Fracturing of the metal lacy covers above corrosion pits and during the formation of new pits was also observed, with the implications of this observation discussed in light of SCC nucleation and growth. The second manuscript in this chapter (Manuscript E, section 7.3) reports work to investigate the influence on pit reactivation of polarisation time and plastic strain of ~5% and ~10%. The results confirm observations obtained in previous studies, showing reactivation of pits after the synergetic effect of strain and re-polarisation. An increase in applied strain from 5% to 10% led to a larger number of activated pits, often yielding open pit morphologies. The effect of strain on pit morphology led to faster pit growth kinetics, resulting in fracture and collapse of metal lacy covers. The pit morphology showed elongated dish shapes with lobes indicating stable pit growth by the undercutting process. Pit growth and morphology also appear to be influenced by sample position and the integrity of the metal cover. 7.1 Key findings and results The effects of plastic strain and electrochemical polarisation on the reactivation of corrosion pits in type 304L stainless steel wire were investigated using X-ray computed tomography. 181

182 1. The re-activation of corrosion pits was observed, after the application of strain followed by electro-chemical re-polarisation. The formation of a pit at the circumference of an existing pit was observed. 2. The pit stability product indicates stable pit growth above 0.3 A.m -1, with the pit shape evolving into an ellipsoidal morphology. A 3D lobe morphology associated with the propagation of stable pits by undercutting of the metal surface was observed. 3. Pit growth kinetics estimated before and after pit reactivation indicated an increase in dissolution rate. The diffusion product (DΔC) of 3D pit growth increased from mol.cm -1.s -1 to mol.cm -1.s -1 with the application of strain. 182

183 7.2 Strain-induced Re-activation of Corrosion Pits in Austenitic Stainless Steel (Manuscript D) F.A. Almuaili 1, S.A. McDonald 2, P.J. Withers 2, A.B. Cook 1, D.L. Engelberg 1 1 Corrosion and Protection Centre, School of Materials, The University of Manchester, Sackville Street, Manchester, M13 9PL, United Kingdom 2 Manchester X-ray Imaging Facility, School of Materials, The University of Manchester, Sackville Street, Manchester, M13 9PL, United Kingdom Abstract The reactivation of a corrosion pit under the synergetic effect of strain and electrochemical polarisation has been observed in a type 304L stainless steel using X-ray computed tomography. The pit reactivation process was associated with the formation of a new pit, directly adjacent to a pre-existing pit. Pit growth kinetics were estimated, revealing an increase of the diffusivity parameter (DΔC) from mol.cm -1.s -1 to mol.cm -1.s -1 with the application of strain, indicating higher metal dissolution rates. Applied strain resulted in fractured lacy metal covers, and its effect on pit growth kinetics is discussed. Keywords: Pitting corrosion, reactivation, electrochemical polarisation, stainless steel, X-ray computed tomography Introduction Austenitic stainless steels are widely used in petrochemical and nuclear applications, due to their good corrosion resistance and excellent mechanical properties. However, localised attack, such as pitting or crevice corrosion, is often found on exposure of such steels to halide-containing environments. Corrosion pits grow beneath the metal surface, and can act as stress concentration sites from which cracks can nucleate and subsequently propagate [1-3]. Stainless steels derive their corrosion resistance from a passive layer comprising chromium oxide, and the protective nature of this oxide layer is typically improved with increasing chromium content [4, 5]. The microstructure of austenitic stainless steel can 183

184 also influence the stability of the passive oxide film and its corrosion resistance. For example, the microstructure may contain second phases and inclusions, among which non-metallic inclusions are common sites where localised attack initiates. Local strain concentrations, introduced either through thermomechanical processing or during service exposure, can also affect microstructure and oxide film stability [6, 7]. Localised breakdown of the passive film with rapid metal dissolution leads to stable pit growth, with the latter occurring at electrochemical potentials in excess of the critical pitting potential (E pit ). This parameter depends on the chloride concentration of the bulk electrolyte and electrolyte temperature, and is frequently used as a measure of localized corrosion resistance of alloys. Below E pit, metastable pitting can occur; this is characterised by short periods of pit nucleation and growth, followed by repassivation [8, 9]. Furthermore, a stable passive film is characterised by a protection potential, also called repassivation potential (E rep ), where pits cease to grow and passivity of the alloy surface is re-established. E rep varies with the depth of existing active pits and the potentiodynamic sweep rate. Below this repassivation potential, no metastable pits can initiate, and previously grown stable pits will turn passive and cease to grow [7]. Galvele [10] addressed critical factors for a one-dimensional model of active pit dissolution, based on current density (i) and pit depth (r). The model is based on local acidification, with a 1D geometry used to represent an imperfection in a passive film. Galvele assumed that dissolution was followed by hydrolysis generating an acidic environment adjacent to the dissolving interface. By considering the transport of ionic species in and out of the pit, he calculated a critical value of i.r above which the ph at the pit bottom could maintain active dissolution and the metal would remain in the active state. Gaudet et al. [11] used artificial pits to study the relationship between current density and metal ion concentration (C), suggesting that a metal ion concentration of 60% of its saturation limit is required for active dissolution inside a pit in a low ph environment. Pistorius and Burstein [12] propose a pit stability product based on a critical minimum and maximum metal ion concentration threshold inside an open hemispherical pit, assuming a diffusion coefficient (D) of metal ions in water of cm 2.s -1. A pit stability product of 0.3 A.m -1 i.r 0.6 A.m -1 is often assumed [13], yielding a minimum concentration of metal ions of 3 M for stable pit growth. 184

185 The susceptibility of pitting corrosion under stress was investigated by Suter [14] in 1 M NaCl, using type 304 stainless steel with varied sulphur contents. Electrochemical polarisation at 60-80% of yield stress showed that cracks and stable pits were formed. However, without applied stress only meta-stable pit growth was observed under the same conditions. This observation was attributed to the effect of local stress, which is believed to increase the metal dissolution rate. Shimahashi et al. [15, 16] studied the influence of strain using a micro-electrochemical cell in a 1.5 M MgCl 2 and 1 M Na 2 SO 4 electrolyte. The results showed that micro-cracks could be induced from (Mn, Cr)S inclusions at 60% of yield stress. Micro-cracks initiating perpendicular to the loading axis lead to stable pit growth by exposing the metal surface beneath the inclusion to corrosive solution. Moreover, a higher anodic polarisation was found to increase the dissolution of Mn-containing stainless steel with application of 75% of the yield stress in 1 M NaCl solution [17]. With the application of plastic strain, pitting corrosion resistance was significantly reduced in various chloride concentration of 90 mg L -1, whereas below this concentration the effect of strain on corrosion resistance was found to be insignificant [18]. The application of 3D imaging methods using synchrotron-based X-ray CT to observe the behaviour of bulk materials in extreme environments and to study corrosion processes is now well established [19-22]. Snapshots of dynamic processes, as well as important details from 3D images, can now easily be obtained to better understand the kinetics of pit growth [23]. In-situ X-ray CT of pitting corrosion, intergranular corrosion, and stress corrosion cracking (SCC) of stainless steels and aluminium alloys have been reported in the literature [24-27]. The technique has also been employed to assess SCC under atmospheric exposure conditions in aged aluminium alloy [28]. The 3D topography of pits obtained using X-ray CT images can be used to investigate nucleation sites and to provide information about conditions leading to pit growth, pitto-crack transition and about SCC propagation mechanisms [29-31]. The work reported in this paper was undertaken with a view to determining whether the application of strain leads to a greater incidence of pits and can assist in their reactivation. Experiments were conducted under electrochemical potential control with and without applied plastic strain. In-situ X-ray CT was employed to obtain 3D information about the effect of electrochemical polarisation on pit morphology and dissolution kinetics. 185

186 7.2.3 Experimental Set-up A solution annealed type 304L stainless steel wire was used to perform the in-situ electrochemical polarisation experiment in 0.1 M NaCl solution at room temperature. The wire had a diameter of 500 μm with a chemical composition of (wt.%) 18.4 Cr, 8.7 Ni, 1.4 Mn, 0.34 Si, 0.04 N, 0.02 C, 0.03 P and S. A specimen of 70 mm length was cut and the surface was ground using 1200 grit SiC paper, washed with deionised water, and dried in a stream of hot air. The surface of the wire was then coated with beeswax, leaving a cylindrical area of 1.57 mm 2 exposed. The wire was mounted in an electrochemical cell and exposed to NaCl electrolyte. The cell was interfaced to a ZEISS Xradia 520 Versa micro-tomography instrument as shown in Figure 7-1(a). A cross-sectional optical micrograph of the type 304L stainless steel wire used in this work is shown in Figure 7-1(b). The cell designed for in-situ X-ray tomography had an outer diameter of 30 mm and inner diameter of 24 mm, and was designed to allow the application of strain without removing the wire from the electrolyte. The cell consists of two parts, the lower part containing the electrochemical cell with an electrolyte volume of 9 ml and the upper part containing the straining set-up that allowed a unidirectional tensile strain to be applied. For polarisation measurement, a miniature reference electrode (Ag/AgCl, 3 M NaCl) and miniature platinum counter electrode were used. Table 7-1 gives a summary of the in-situ experimental methodology. Potentiodynamic polarisation tests were performed using a scan rate of 1 mv.s -1 with an Ivium CompactStat Potentiostat. The open circuit potential (OCP) was recorded for 15 minutes after the solution was added into the in-situ cell (step 0). The OCP was also recorded for 5 min before the start of all other subsequent polarisation cycles (steps 5 and 7). The current was recorded at a rate of 1 Hz. Strain was applied to the wire, which was clamped between two bolts, by tightening a nut at the top of the in-situ cell. The elongation of the wire relative to its initial length was used to approximate the applied strain. The first potentio-dynamic polarisation was performed from OCP to +650 mv vs. Ag/AgCl. When this potential was reached the polarisation was stopped and the first X- ray CT scan (scan A) was performed at OCP to visualise the pits formed. Approximately 5% tensile strain was then applied, and another X-ray CT scan (scan B) 186

187 carried out to obtain information about the changes in pit shape effected by the applied strain. In order to investigate whether pits can be re-activated after straining, a second polarisation scan was then performed from OCP to +650 mv; this potential was maintained for 180 seconds beyond the end of the scan. A third polarisation scan was performed from OCP to +650 mv, followed by a potentio-static hold at +650 mv for 130 seconds. A final X-ray CT scan (scan C) was performed at OCP to visualise the pits after the third polarisation cycle was completed. For the X-ray CT measurements, an accelerating voltage of 120 KV was used, and 721 projections recorded with 2 2 binning at 10 optical magnification, with an exposure time of 12 seconds for each projection. This resulted in a reconstructed voxel size of 1.1 m 3 (giving a spatial resolution of 2-3 m) with a field of view of 1100 µm 1100 µm. Each scan took about 2-3 hrs. The data was reconstructed using a Feldkamp-Davis- Kress approach [32], and images were segmented and visualised using Avizo software. A min./max. threshold segmentation was carried out, and the total pit volume, total internal pit surface area, pit depth and width, and associated aspect ratios and shapes were obtained. To determine pit depth, width and height, individual 2D slices of the reconstructed X-ray CT data were taken from the geometrical centre of the pit volume. These measurements provide a snap-shot of the overall pit dimensions. The current response recorded over time provided in-situ pit growth information. At the end of the experiment, the wire was removed from the in-situ cell and the sample imaged using a FEI Quanta 650 scanning electron microscope (SEM). 187

188 Table 7-1: Summary of in-situ electrochemical polarisation experiment Step Polarisation potential cycles 0 OCP measurement for 15 min 1 1 st Potentiodynamic polarisation from OCP up to +650 mv vs. Ag/AgCl 2 X-ray CT at OCP Scan (A) 3 Application of 5% tensile strain at OCP 4 X-ray CT at OCP Scan (B) 5 OCP measurement for 5 min 2 nd potentio-dynamic polarisation from OCP to +650 mv, followed by potentiostatic polarisation at +650 mv for 180 seconds 6 7 OCP measurement for 5 min 3 rd potentio-dynamic polarisation from OCP to +650 mv, followed by potentiostatic polarisation at +650 mv for 130 seconds. 8 9 X-ray CT at OCP Scan (C) 10 Experiment terminated & sample removed for SEM analysis X-Ray Source Straining rig Detector Wire sample exposed to electrolyte Figure 7-1: (a) Image of the electrochemical cell with straining rig for in-situ X-ray CT experiments, (b) cross sectional optical micrograph of the annealed type 304L stainless steel wire after etching in 10% (wt.) oxalic acid at 6V. 188

189 7.2.4 Results and Discussion Successive X-ray CT scans were acquired in a time lapse manner to image the development of corrosion pits in the stainless steel wire sample. Figure 7-2(a) shows a 3D image of the wire reconstructed from X-ray CT scan (A), indicating nucleation and growth of a single pit (pit 1). The pit grew for 55 second during the first polarisation (step 1), with the current response shown in Figure 7-3(a). Figure 7-2(b) shows the effect of a 5% tensile strain on the volume and shape of pit 1. The max strain is marked at the edges of pit normal, to the straining direction. The image was obtained during scan (B), and a change in pit shape is clearly visible. Figure 7-2(c) shows the re-constructed image from X-ray CT scan (C), performed after the third polarisation scan and hold (step 8) revealing the presence of two large pits. One new pit (pit 2) had nucleated next to the existing pit 1. It is evident from the reconstructed X-ray CT data of pit 1 in Figure 7-2(c) that the pit volume has increased compared with Figure 7-2(b). The current response during the third polarisation scan is shown in Figure 7-3(a). The spatial resolution of X-ray CT is not sufficient to show the lacy metal covers in Figure 7-2. Strain Pit 2 Width (w) Maximum Strain Height (h) Pit 1 Pit 1 strained Pit 1 reactivated (a) (b) (c) Figure 7-2: Reconstructed X-ray CT data of the wire sample: (a) after the 1 st electrochemical polarisation scan (X-ray CT scan A), (b) after ~5% plastic strain (X-ray CT scan B), (c) after the 3 rd electrochemical polarisation cycle (X-ray CT scan C). The resolution of the scans is not sufficient to show the lacy metal covers. 189

190 Electrochemical observation The OCP measurement (step 0) stabilised at +27 mv vs. Ag/AgCl, before the anodic polarisation to +650 mv was carried out (step 1). Figure 7-3(a) shows the time dependent current response obtained during this scan; the current started to increase abruptly at a potential just short of +600 mv. After strain was applied (step 3), the second polarisation was conducted from OCP (+162 mv) to +650 mv vs Ag/AgCl, and the potential kept constant at +650 mv for 3 min (step 6). This polarisation scan showed no current response, which means that no pit formation occurred during this scan. The polarisation was repeated in polarisation scan 3 from OCP (+284 mv) to +650 mv followed by holding the final potential for 130 second (step 8). The current started to rise during the potentio-static period, and the experiment was terminated after a further 70 seconds of pit growth. The current response is shown in Figure 7-3(a). The OCP shifted by +135 mv between the first and second polarisation scans. This potential shift may be a result of passive surface film thickening during polarisation and/or a result of the influence of the X-ray beam on the nature of the passive film. OCP measurements before the second and after the third polarisation scan showed a further anodic shift of +122 mv. This indicates that the OCP shift was dominated by the film thickening rather than interference with the X-ray CT beam, since no X-ray CT analysis was carried out between these two polarisation scans. (a) 190

191 Figure 7-3: (a) Current evolution with time during the first and third polarisation cycles and (b) summary of measured pit dimensions taken from X-ray CT data. (b) Pit geometry and dimensions Figure 7-3(b) shows pit dimensions (depth, width and height) obtained from X-ray CT data. Pit 1 had a near hemi-spherical shape. The results also show an elongation of pit height from 75 to 85 μm with the application of strain, but no change in depth or width of pit 1. After repolarisation (3 rd polarisation scan, step 8), the change in geometry of pit 1 indicates that further metal dissolution had taken place, suggesting that the pit had reactivated. The measured volume of pit 2 was slightly larger than the volume of the reactivated pit 1. Table 7-2 summarises the measurement of all pit dimensions obtained from X-ray CT time lapse data, including total segmented pit volumes and measured surface areas of pits 1 and 2. Dimensions determined from X-ray CT scan (C) show that the volume of pit 1 increased significantly, to more than twice its volume at the end of polarisation scan 1 after the third potentio-dynamic scan and potentiostatic hold (step 8). This means that pit 1 reactivated during the third polarisation scan. Table 7-2 also lists pit dimensions estimated by Faraday s law from the charge determined from current vs. time (data to be used in determining pit surface area via Approach B- details of the approach to be described later). To do this, the total charge 191

192 associated with the growth of each pit was obtained and converted via Faraday s law into a volume of dissolved metal. The valence of metal dissolution was assumed to be 2.19, the atomic weight M = g.mol -1, density of ρ = 7.97 g.cm -3 and Faraday s constant F = C.mol -1 [12]. Current efficiency was assumed to be 100%. A assuming a hemispherical pit shape, the pit radius may be determined from volume of metal dissolved and consequently the pit depth, width, height and corresponding surface area determined. In order to obtain information about the share of charge for both pits during the third polarisation cycle in Figure 7-3(a), the following assumptions were made: (i) (ii) Both pits are active at the end of their polarisation cycles, and their volumes increase at a constant rate; In the current vs time profile associated with the third polarisation procedure (Figure 7-3a), a change in gradient is observed at ~11 second. This is the instant when the reactivation of pit 1 is assumed to start. The total current from this point in time is then divided equally between the reactivated pit 1 and pit 2. To support the latter assumption, the charge consumed by each pit during the third potentio-dynamic scan was converted into a dissolved volume using Faraday s law, and then compared to the dissolved volume measured by X-ray CT of scan (C). This comparison shows a reasonably good match, with a difference of only 17% in volume for the re-activated pit 1, and 6% for the volume of pit 2. However, the Faraday-based assumption slightly over-estimates pit depth relative to the measurement of real pit depth in approach A. 192

193 Table 7-2: Measured pit dimensions Calculated from the charge by Parameter Measured from reconstructed X-ray CT data (to be used in determining pit surface area via approach A) Faraday s law, assuming hemispherical pit growth (to determine pit surface area via approach B) Pit 1 Pit 1 Pit 1 Pit 2 Pit 1 Pit 1 Pit 2 (strained) (reactivated) (reactivated) Depth, r (μm) * 45* 46* Width, w (μm) * 90* 92* Height, h (μm) * 90* 92* Volume, V (μm 3 ) Surface area, 7955 (μm 2 ), without (11958) lacy cover (with cover) *Estimated values in bold * 12742* 13439* (12302) Lacy cover morphology Figure 7-4(a, b, c) shows SEM images of the wire after the X-ray CT experiment was terminated (step 10). The lacy cover morphology of pit 1 shows two fractures in the middle of the cover, with cracks emanating from holes at the periphery. The lacy cover has a slightly different morphology relative to the classic lacy covers reported in the literature [33-35]. The pit mouth was not in the centre of the perforated ring of the lacy cover. Examination of Figure 7-4(a, b) indicates that pit 1 actually made-up of two adjacent pits, with one large hole and circular lacy cover, augmented by a second small hole and semi-circular cover. The pits seem to have coalesced. 193

194 Pit 1 Crack Width (w) Height (h) (b) Pit 2 (a) (c) Figure 7-4: SEM images (a) of the type 304L wire sample showing the two pits, (b) pit 1 lacy cover morphology, and (c) pit 2. Note: cracks in the lacy covers are also shown. Pit 2 shows a lacy cover and its mouth is elongated in height, compared to its width. The collapse of the cover around the pit mouth is clearly visible. The lacy cover also shows small cracks around the pit mouth, normal to the straining direction. The lacy cover is a thin metallic cover which aids pit growth by restricting diffusion of metal ions out of the pit, hence maintaining the corrosive environment inside the pit [35]. The cracks observed in these covers appear to start from the pit mouth, which has a thinner metal cover, and propagate through perforation to the pit edge. At the circumference of the covers, sharp cracks are also apparent, and fracture of the cover appears brittle. In SCC it has been reported that crack initiated from pit bottom or the pit walls at the edge of pits [36]. The cracks observed to have developed in the lacy metal cover close to the edge of the pits may also provide sites for SCC to propagate further into the matrix metal. The brittle cracks observed in the lacy cover may also act as precursor sites for 194

195 SCC, with crack propagation rather than crack initiation being the predominant mechanism that needs to be considered in such a case Reactivation of pits Figure 7-5(a, b, c, d, e) summarises the segmented volumes of both pits from X-ray CT data. The figure shows the volume of pit 1 after the first polarisation scan (Figure 7-5a). The segmented data of X-ray CT scan B is shown in Figure 7-5(b) after the application of 5% plastic strain. The difference between the strained (scan B) and unstrained pit (scan A) is shown in Figure 7-5(c). The result of scan C is shown in Figure 7-5(d), after the reactivation of pit 1 and the nucleation and growth of pit 2. The composite image in Figure 7-5(e) clearly shows the change in volume before and after pit reactivation, which occurred in the potentiostatic region of the third polarisation scan. The latter image was obtained by making the pits in Figure 7-5(d) transparent, with an overlay of the pit volumes from Figure 7-5(c). Pit re-activation seemed to occur as a result of a synergetic effect between strain and electrochemical polarisation. Re-activation is related in this case to the nucleation of a pit directly at the circumference of an existing pit (pit 1, Figure 7-4b). The acidity generated by metal dissolution within it and subsequent hydrolysis most likely generates the local environmental conditions to reactivate the original cavity of pit 1. This would be consistent with changes in the dimensions (width, depth, height) of pit 1. The original cavity of pit 1 is seen to have expanded, it is not just a case of growth via the newly nucleated adjacent pit (Figure 7-5e). The observed reactivation may have been facilitated by the application of plastic strain which may have caused fractures in the surface films or around inclusions at which the new pit could initiate. Though, since the wire remained exposed to the environment, the local chemistry inside the pit may not have changed significantly over time, with the nucleation of an adjacent pit further enhancing local conditions to re-active part of the pre-existing pit surface. Breakdown of the film inside the pit may have exposed the bar surface to the corrosive environment, with aggressive environmental conditions additionally promoted by growth of the adjacent pit, resulting in pit reactivation. 195

196 Pit 2 Pit 1 Pit 1 strain Pit1 reactivated (a) (b) (c) (d) (e) Figure 7-5: Segmented pits from X-ray CT data (a) Pit 1 generated by 1 st polarisation scan, (b) pit 1 after ~5% strain, (c) the difference between (a) and (b), (d) reactivated pit 1 and pit 2 generated during potentiostatic polarisation at the end of the 3 rd cycle, with (e) showing the transparent volume of pit 1 reactivated and pit 2 relative to pit 1 before reactivation Pit growth kinetics The current density associated with pit growth as a function of time was estimated using two methods; these are explained in more detail in [23]. Figure 7-6(a) shows the estimated current density over time for pit 1, which was obtained by dividing the current at each time step of the first polarisation scan by the estimated pit surface area at the corresponding current. The pit surface area was obtained via approach A, in which the final volume of the pit, obtained from segmented X-ray CT data, was considered to be that of a hemisphere and the area at each point determined by assuming a constant, isotropic, hemi-spherical growth rate varying linearly with time. This allowed the current at each point in the current vs. time profile of Figure 7-3(a) to be divided by an estimate of pit surface at each point in time. In approach B, the current density was based on the total dissolved volume calculated from the charge consumed at each point and Faraday s law, assuming hemispherical pit growth. Comparison of approaches (A and B) provides an estimate of the uncertainty in the segmented X-ray CT data compared to the measured current response. 196

197 Figure 7-6(a, b) shows current density based on pit surface area. The current density calculated via approach A for pit 1 reached a maximum of 5 A.cm -2, whereas, the current density of the reactivated pit 1 and pit 2 reached 10 A.cm -2 and 6.5 A.cm -2, respectively. The difference in maximum current density relative to pit 1 before reactivation could be related to an effect of the plastic strain, which has been reported to accelerate metal dissolution rate [17]. After reaching the maximum current density, all pits show a drop in current density, possibly due to the precipitation of a film on the surface inside pits. The current density then further decays to constant values close to 1 A.cm -2. The current densities reported here are consistent in magnitude with those reported in other studies [23, 37]. Figure 7-6(a, b) also shows the current density inferred using Faraday s law via approach B in Table 7-2. Only minor differences are apparent between the two approaches for the growth of pit 1, cross-validating the methods. Figure 7-6(b) indicates far higher current densities via approach B, which suggests a far larger degree of divergence between measured X-ray CT data and the dissolved volume. The current response in Figure 7-3(a) indicates current initially rising over time, followed by a significant change in gradient after approximately 13 seconds for polarisation scan 1 and after 8 seconds for polarisation scan 3. Such behaviour has been related to metastable-to-stable pit transition behaviour [38, 39]. (a) 197

198 (b) Figure 7-6: (a) Pit 1 current density over time estimated using surface area values calculated via approaches A and B (Table 7-2), (b) current density of pits 1 and 2 over time after strain and repolarisation, calculated using surface area values obtained via approaches A and B Pit stability product Pit stability in type 304L requires a pit stability product between 0.3 and 0.6 A. m -1 [12]. The variation in pit stability product with time was calculated from estimations of current density and pit depth via approaches A and B [23]. The results calculated using approaches A and B are shown in Figure 7-7(a, b). Values calculated from values obtained via both approaches show that the pit stability product at the early stage of pit growth lies below the stability criteria, indicating the pit growing metastable initially and emphasizing the role of the lacy metal cover which maintains the critical chemistry within the pit cavity. The time dependences of the stability product observed in this work are consistent with those reported in the literature [12]. The results in Figure 7-7 show a difference in stability product of 0.1 A.m -1 between the two approaches for pit 1. The difference increased to ~0.2 A.m -1 after the application of plastic strain and repolarisation. This is most likely due to an overestimation of pit depth via approach B, relative to the value obtained by approach A. It also indicates that the pit was more dish-shaped than hemispherical. In addition, the SEM image in Figure 7-4 shows that the pits had metal lacy cover supports, indicating that early pit growth 198

199 occurred with a pit stability product below the lower bound calculated by Pistorius and Burnstein [11]. (a) (b) Figure 7-7: Pit stability product vs. time estimated from data obtained via approaches A and B (Table 7-2) for (a) pit 1, (b) reactivated pit 1 and pit Diffusion coefficient and pit diffusion product The product of the effective diffusion coefficient and the critical concentration of metal ions at the corroding interface (C crit ) of pit growth can be estimated from the slope of the relationship between the square of pit depth and time of pit growth using equation 7-199

200 1 [12, 40]. The difference between C crit and the concentration at bulk or pit opening is defined as ΔC. r 2 = 3M D C πρ t (7-1) Figure 7-8(a) illustrates graphically this relationship and the value of the diffusion product for pits 1 and 2. The difference in slope indicates the effect of changing the dissolution rate under strain. The diffusion product (DΔC) of pit 1 is calculated from data related to the first polarisation scan as mol.cm -1.s -1. After applying strain, the reactivated pit 1 and the pit 2 show larger diffusion products of and mol.cm -1.s -1, respectively. The latter is in the range of reported values of the diffusion product for 1D pit growth [35, 41]. The diffusion product of pit 1 (without strain) is in the range of diffusion products reported for 2D studies of pit growth [40]. This result is also consistent with a previous result [23] of mol.cm -1.s -1 which was obtained for pit growth in the presence of a lacy metal cover using similar experimental conditions (without strain). The difference in the diffusion product between unstrained and strained pits may indicate increased metal dissolution rates due to the influence of applied strain as reported elsewhere in the literature [15, 16, 18]. However, this would also mean either that the diffusion of species through a precipitated salt film inside pits may be accelerated in the presence of strain, or that faster dissolution can occur along pathways of damage or cracks in this salt film. Pit growth is generally believed to follow an ohmic growth law below a certain potential with a transition to diffusion control above it. Under diffusion control, the metal salt concentration at the pit bottom assumes saturation at 4.2 M, while the effect of concentration at the pit mouth is typically neglected. With these assumptions of ΔC, the mean diffusion coefficients (D) can be estimated from diffusion product values as and cm 2.s -1 for pit 1 without and with strain, and cm 2.s -1 for pit 2. The variation of diffusivity in pit growth over time can also be estimated using a modified form of equation 7-1, shown as equation 7-2 [12]. C = 2π 3nFD i. r (7-2) The value of diffusion coefficient (D) at each time step, assuming a 4.2 M saturation concentration (ΔC) is shown in Figure 7-8(b). This assumption does not reflect changes 200

201 in metal ion concentration and pit growth regimes that are not under diffusion control during the initial pit nucleation stage and is therefore not fully correct. However, the results reflect the latter by showing large variations of D at the early stages of pit growth. The diffusion coefficients then converge with longer exposures, resulting in pit growth under diffusion control. Effective diffusion coefficients of cm 2.s -1 for pit 1 (without applied strain), with cm 2.s -1 for reactivated pit 1 and cm 2.s -1 for pit 2 were obtained for the period of diffusion controlled pit growth. (a) (b) Figure 7-8: (a) (depth) 2 vs. time and the diffusion product (slope) for pit 1 before and after reactivation and for pit 2, (b) diffusion coefficient of all pits assuming a constant salt concentration of 4.2 M. 201

202 7.2.5 Conclusion The following conclusions can be derived from this time-lapse study. 1- Reactivation of a corrosion pit under the synergetic effect of strain and electrochemical polarisation was observed using X-ray CT. 2- The pit morphology and associated lacy metal covers showed marked differences before and after plastic strain was applied. Cracks were observed in the lacy metal covers. 3- Pit stability products between 0.3 A.m -1 and 0.6 A.m -1 were calculated via two methods and are consistent with values reported for stable pit growth. 4- The product of effective diffusivity and metal ion concentration (DΔC) of 3D pit growth was mol.cm -1.s -1, with the introduction of strain resulting in an increase to mol.cm -1.s -1 and mol.cm -1.s Effective diffusion coefficients (D) of for pit 1, cm 2.s -1 for reactivated pit 1 and cm 2.s -1 for pit 2 were obtained for the diffusion controlled pit growth regime Acknowledgements The authors acknowledge the provision of beam time at the Henry Moseley X-ray imaging Facility (HMXIF) of the University of Manchester, UK, established with funding from the EPSRC through grants EP/F007906, EP/I02249X and EP/F FAA and DLE would like to thank the Saline Water Conversion Corporation (SWCC), Saudi Arabia for financial support References 1. Turnbull, A., et al., Technical note: Visualization of stress corrosion cracks emerging from pits. Corrosion, (7): p Horner, D.A., et al., Novel images of the evolution of stress corrosion cracks from corrosion pits. Corrosion Science, (11): p Zhu, L.K., et al., Stainless steel pitting and early-stage stress corrosion cracking under ultra-low elastic load. Corrosion Science, (0): p Olsson, C.O.A. and D. Landolt, Passive films on stainless steels chemistry, structure and growth. Electrochimica Acta, (9): p Schmuki, P., From Bacon to barriers: a review on the passivity of metals and alloys. Journal of Solid State Electrochemistry, (3): p

203 6. Jun, J., K. Holguin, and G.S. Frankel, Pitting Corrosion of Very Clean Type 304 Stainless Steel. Corrosion, (2): p Soltis, J., Passivity breakdown, pit initiation and propagation of pits in metallic materials - Review. Corrosion Science, : p Frankel, G., Pitting corrosion of metals a review of the critical factors. Journal of the Electrochemical Society, (6): p Laycock, N.J. and R.C. Newman, Localised dissolution kinetics, salt films and pitting potentials. Corrosion Science, (10-11): p Galvele, J.R., Transport Processes and Mechanism of Pitting of Metals. Journal of the Electrochemical Society, (4): p Gaudet, G.T., et al., Mass-Transfer and Electrochemical Kinetic Interactions in Localized Pitting Corrosion. Aiche Journal, (6): p Pistorius, P.C. and G.T. Burstein, Metastable Pitting Corrosion of Stainless Steel and the Transition to Stability. Philosophical Transactions of the Royal Society of London Series a-mathematical Physical and Engineering Sciences, (1662): p Kuo, H.C. and D. Landolt, Rotating-Disk Electrode Study of Anodic Dissolution or Iron in Concentrated Chloride Media. Electrochimica Acta, (5): p Suter, T., et al., Pit initiation on stainless steels in 1 M NaCl with and without mechanical stress. Journal of the Electrochemical Society, (5): p. B174-B Shimahashi, N., et al., Effects of Corrosion and Cracking of Sulfide Inclusions on Pit Initiation in Stainless Steel. Journal of the Electrochemical Society, (10): p. C494-C Shimahashi, N., et al., Effect of Applied Stress on Dissolution Morphology and Pit Initiation Behavior of MnS Inclusion in Stainless Steel. ECS Transactions, (31): p Devasenapathi, A. and M. Asawa, Effect of applied potential on the nature of surface film and SCC of a high Mn stainless steel in 1 M HCl. Journal of Materials Science, (23): p LÜ, G., et al., Effect of Strain and Chloride Concentration on Pitting Susceptibility for Type 304 Austenitic Stainless Steel. Chinese Journal of Chemical Engineering, (2): p Eckermann, F., et al., In situ monitoring of corrosion processes within the bulk of AlMgSi alloys using X-ray microtomography. Corrosion Science, (12): p Rayment, T., et al., Characterisation of salt films on dissolving metal surfaces in artificial corrosion pits via in situ synchrotron X-ray diffraction. Electrochemistry Communications, (6): p Ghahari, S.M., et al., In situ synchrotron X-ray micro-tomography study of pitting corrosion in stainless steel. Corrosion Science, (9): p Burnett, T., et al., Correlative Tomography. Scientific reports, Almuaili, F.A., et al., Application of a Quasi In Situ Experimental Approach to Estimate 3-D Pitting Corrosion Kinetics in Stainless Steel. Journal of the Electrochemical Society, (13): p. C745-C Babout, L., et al., X-ray microtomographic observation of intergranular stress corrosion cracking in sensitised austenitic stainless steel. Materials Science and Technology, (9): p Marrow, T.J., et al., Three dimensional observations and modelling of intergranular stress corrosion cracking in austenitic stainless steel. Journal of Nuclear Materials, (1 3): p

204 26. Marrow, T.J., et al., High-resolution, in-situ, tomographic observations of stress corrosion cracking, in Environment-Induced Cracking of Materials, S.A. Shipilov, et al., Editors. 2008, Elsevier: Amsterdam. p Knight, S.P., et al., In situ X-ray tomography of intergranular corrosion of 2024 and 7050 aluminium alloys. Corrosion Science, (12): p Singh, S.S., et al., In situ experimental techniques to study the mechanical behavior of materials using X-ray synchrotron tomography. Integrating Materials and Manufacturing Innovation, (1): p Acuña, N., et al., Analysis of the stress intensity factor around corrosion pits developed on structures subjected to mixed loading. Scripta Materialia, (4): p Zhou, S. and A. Turnbull, Influence of pitting on the fatigue life of a turbine blade steel. Fatigue and Fracture of Engineering Materials and Structures, (12): p Kondo, Y., Prediction of Fatigue Crack Initiation Life Based on Pit Growth. Corrosion, (1): p Feldkamp, L., L. Davis, and J. Kress, Practical cone-beam algorithm. JOSA A, (6): p Ernst, P., et al., The mechanism of lacy cover formation in pitting. Corrosion Science, (6): p Laycock, N.J., et al., Perforated covers for propagating pits. Journal of the Electrochemical Society, (4): p Ernst, P. and R.C. Newman, Pit growth studies in stainless steel foils. I. Introduction and pit growth kinetics. Corrosion Science, (5): p Turnbull, A., L.N. McCartney, and S. Zhou, Modelling of the evolution of stress corrosion cracks from corrosion pits. Scripta Materialia, (4): p Alkire, R.C. and K.P. Wong, The Corrosion of Single Pits on Stainless-Steel in Acidic Chloride Solution. Corrosion Science, (4): p. 411-&. 38. Frankel, G.S., et al., Metastable Pitting of Stainless Steel. Corrosion, (7): p Moayed, M.H. and R.C. Newman, Evolution of current transients and morphology of metastable and stable pitting on stainless steel near the critical pitting temperature. Corrosion Science, (4): p Ghahari, M., et al., Synchrotron X-ray radiography studies of pitting corrosion of stainless steel: Extraction of pit propagation parameters. Corrosion Science, : p Moayed, M.H. and R. Newman, Using pit solution chemistry for evaluation of metastable pitting stability of austenitic stainless steel. Materials and Corrosion, (3): p

205 7.3 Time-dependent Re-activation of Corrosion Pits in Austenitic Stainless Steel (Manuscript E) F. A. Almuaili 1, S. A. McDonald 2, P. J. Withers 2, D. L. Engelberg 1 1 Corrosion and Protection Centre, School of Materials, The University of Manchester, Sackville Street, Manchester, M13 9PL, United Kingdom 2 Manchester X-ray Imaging Facility, School of Materials, The University of Manchester, Sackville Street, Manchester, M13 9PL, United Kingdom Abstract The effects of plastic strain and electrochemical polarisation on the reactivation of corrosion pits in type 304L stainless steel has been investigated using X-ray computed tomography (X-ray CT). Strain-induced reactivation of corrosion pit was observed, which was linked to a time-dependent re-activation of corrosion pits under potentiodynamic/static control. Cracks in the lacy metal covers, resulting in sharp cracks at the cover edges, and brittle fractures normal to the direction of applied strain were found and discussed. Pit growth kinetics were estimated before and after reactivation, showing an increase in pit dissolution rate and changes to pit morphology due to the application of plastic strain. The development of 3D lobe morphologies, associated with the propagation of stable pits by undercutting of the metal surface was observed. The diffusivity parameter of 3D pit growth before the application of strain was mol.cm -1.s -1 and with the application of strain increased to mol.cm -1.s -1. The pit stability product indicates stable pit growth above 0.3 A.m -1, where pit growth appears to be under diffusion control. Keywords: Pitting corrosion, Pit Reactivation, Electrochemical Polarisation, Stainless steel, X-ray computed tomography, Plastic strain Introduction Stainless steels show good corrosion resistance, which is attributed to the passive oxide film that forms on the metal surface, which makes these materials attractive for industrial applications. However, stainless steels are also subject to localised attack, such as pitting or crevice corrosion, often observed in halide-containing environments. 205

206 The passive oxide layer develops spontaneously in air and provides a protective barrier between the corrosive environment and the metal surface, but when it is broken, metal dissolution spreads from small active areas (anodes), while the remaining passive metal surface becomes cathodic. Thus, the corrosion resistance of stainless steel is mainly a function of the stability of the oxide layer, which depends on the corrosivity of the environment, as well as the chemical composition and microstructure condition of the material [1]. Susceptibility to pitting corrosion in corrosive environments is enhanced by increasing the applied potential, the presence of applied or residual strain, and exposure time. The pitting potential (E pit ) is commonly used to characterise pitting corrosion resistance, with local attack often related to the electro-chemistry of the microstructure, such as the difference between non-metallic inclusions and the alloy matrix. Most inclusions show reduced electrochemical stability relative to the alloy matrix as the applied potential is increased [2]. The influence of stress on the pitting corrosion susceptibility of 304 stainless steel with different sulphur contents was reported to increase and induce the cracking of inclusions. This was attributed to the presence of strain, which increased the rate of metal dissolution. Therefore, the effect of stress was suggested to enhance stable pit growth, by exposing new surface of the metal to the corrosive solution [3, 4]. The kinetics of pitting corrosion have been investigated at high applied potential, using one-dimensional (1D) and two-dimensional (2D) electrodes [5-9]. Pit propagation mechanisms in 2D were characterised by an undercutting mechanism in the form of lobes, which developed from the pit bottom,. The repetition of this lobe formation process leads to the formation of a perforated lacy metal cover around the pit mouth. The study of the morphology of pit growth over time shows a change from hemispherical shape, to an elongated dish shape, suggesting that lateral pit growth is faster than growth in depth [6]. This is in contrast to early pit growth, and it has been suggested that pit growth in depth is under diffusion control after the precipitation of a salt film, while lateral pit growth is under activation/ohmic control [6, 8, 9]. The chemistry inside the pit that is necessary to maintain active dissolution and pit growth is ascribed to low ph, with in parallel a critical concentration of dissolved metal ions. A critical concentration of metal ions of 60% of the saturation concentration was reported for active pit dissolution in stainless steel [10]. Therefore, pit stability is 206

207 controlled by the mass transport process of the corrosive solution diffusing out from pits and the associated pit morphology. Pistorius and Burstein [11] investigated the transition criteria of stable pit growth based on the current density and pit depth, reporting a minimum stability value of 0.3 A.m -1 for open pits in type 304 stainless steel. Stable pit growth was associated with a saturation concentration of 75% of FeCl 2. However, Frankel et al. report a value of 0.4 A.m -1 [12]. When stable pits grow under diffusion control, the diffusion product or diffusivity parameter (DΔC) suggests a linear relationship between the square of pit depth (square diffusion length) and the time of pit propagation in 1D and 2D stainless steel electrodes. A difference in diffusion product between 1D and 2D studies was attributed to inherent pit morphologies differences. Therefore, a perforation factor was suggested as the ratio between the open pit (1D) and the perforated pit cover (2D) [8]. However, 3D pit growth geometry and sample position may also contribute to the pit diffusion product [8, 13-15]. Moreover, a recent study which monitored pit growth in NaCl at ph 2 using acoustic emission attributed the effect of pit morphology to the fact that gravity caused the corrosive solution to accumulate at the lower end of the growing pit in vertical samples [16]. In our previous study, pit reactivation under potential control and plastic strain was observed [17]. The present study investigates the influence of polarisation time and plastic strain on the re-activation of existing pits. As in the previous study, in-situ X-ray tomography was employed to obtain 3D information on pitting corrosion at different stages of electrochemical polarisation time, strain and repolarisation. The geometry of 3D pit observation and electrochemical data were combined to determine the 3D pit dissolution kinetics Experimental Set-up A type 304L stainless steel wire of 500 μm diameter was used for the electrochemical polarisation study in 0.1 M NaCl at room temperature. The chemical composition of the steel in wt% was 18.4 Cr, 8.7 Ni, 0.02 C, 1.4 Mn, 0.34 Si, 0.04 N, 0.03 P and S. Wire specimens of 70 mm length were cut and ground using 1200 grit SiC paper, washed with deionised water, then dried in a stream of hot air. The surface of the wire was coated with beeswax, except for an area of 1.1 to 1.57 mm 2, which was exposed to the electrolyte. The wire was vertically mounted inside an electrochemical cell and left 207

208 for 16 hr before the experiment was started. The in-situ cell was then interfaced with a Zeiss Xradia 520 Versa micro-tomography instrument, using the setup shown in Figure 7-9. This setup was designed for in-situ X-ray microtomography experiments. The cell had an outer diameter of 30 mm and inner diameter of 24 mm, designed to be able to apply strain without removing the sample from the electrolyte. The cell consisted of two parts, the lower part containing the electrochemical cell with a volume of 9 ml and the upper part the tensile cell, where unidirectional tensile strain could be applied. For polarisation measurements, a miniature reference electrode (Ag/AgCl, 3 M NaCl) and miniature platinum counter electrode were used, interfaced with the lower part of the insitu cell setup. Table 7-3 gives a summary of the two in-situ experiments (I and II) with polarisation cycles. The wire was clamped between two bolts and strain was applied to it by tightening a nut at the top of the cell. The length of working electrode wire was fixed at 50 mm before strain was applied. The elongation of the wire was used to measure the strain relative to the initial length. Electrochemical polarisation were performed using a scan rate of 1 mv.s -1 with an Ivium CompactStat Potentiostat. After the solution had been added to the in-situ cell, OCP was recorded for 15 min before the first polarisation scan was conducted. The OCP was recorded for 5 min ahead of all subsequent polarisation scans. The current was acquired at a rate of 1 Hz. The first potentiodynamic polarisations was performed from OCP to +607 mv vs Ag/AgCl for exponent (I), and to +604 mv vs Ag/AgCl for experiment (II). X-ray CT scans were then performed at OCP to visualise the developed pits. Uni-directional tensile strain of approximately ~5% and ~10% was applied to sample in experiment I and II, respectively, without removing the sample from the in-situ cell, followed by X- ray CT scans to observe changes in pit shape brought about by the strain. In order to investigate whether the pits could be reactivated, a second polarisation scan was then performed from OCP up to +600 mv (I and II), with this potential maintained for 330 seconds for experiment I, and 198 seconds for experiment II. X-ray CT scans were performed to visualise the pits after this repolarisation scan. For all X-ray CT measurements, an accelerating voltage of 120 KeV was used, and 721 projections were recorded with 2 2 binning at 10 optical magnification, with an exposure time of 12 seconds for each projection. This resulted in a reconstructed voxel 208

209 size of 1.1 µm 3 (giving a spatial resolution of 2-3 µm) with a field of view of 1100 µm 1100 µm. Each scan took about 2-3 hrs. The data were reconstructed using the Feldkamp-Davis-Kress (FDK) approach [18], and images were segmented and visualised using the Avizo 9 software. A min/max data segmentation was undertaken and the total volume, total internal surface area, depth and width, aspect ratio, and shape of each pit were obtained. Single slices were taken from the middle of each pit, with respect to its edges, to determine pit depth, width and height, thus providing a snapshot of pit dimensions. The current response recorded over time provided in-situ pit growth information. At the end of the experiment, the wire was removed from the in-situ cell and the sample morphology imaged using an FEI Quanta 650 scanning electron microscope (SEM). Table 7-3: Summary of in-situ electrochemical polarisation experiments I and II Step Polarisation potential cycles 0 OCP measurement for 15 min 1 First potentiodynamic polarisation from OCP up to (I) +607 and (II) +604 mv vs Ag/AgCl 2 X-ray CT scan 1 at OCP 3 Unidirectional tensile strain of 5% (I) and 10% (II) at OCP 4 X-ray CT scan 2 at OCP 5 OCP measurement for 5 min 6 Second potentiodynamic polarisation from OCP up to +600 mv, followed by potentiostatic polarisation at +600 mv for experiment (I) 330 seconds, and experiment (II) 198 seconds 7 X-ray CT scan 3 at OCP 8 Sample removed for SEM analysis 209

210 Straining rig X-Ray Source Wire sample exposed to electrolyte Detector Figure 7-9: Image of the electrochemical cell with miniature straining rig and load cell for in-situ X-ray tomography experiments with a type 304L wire sample Results and discussion Experiment (I) 3D X-ray CT scans were used to image the stainless steel wire after electrochemical polarisation and after the application of strain. Figure 7-10(a) shows a 3D image of the reconstructed wire from scan 1, indicating nucleation and growth of a single pit (labelled pit 1). The pit was grown for 40 seconds by the first cycle of potentiodynamic polarisation (Table 7-3, step 1). The result of X-ray CT scan 2 in Figure 7-10(b) shows the effect of applying 5% tensile strain on the shape of pit 1. Figure 7-10(c, d) shows the result of the X-ray CT scan performed after the second polarisation cycle (Table 7-3 step 6), indicating the presence of four large pits (three pits on front side and one pit on back side). Three new pits (pits 2, 3 and 4) were formed and the existing pit 1 was reactivated. Pit 1 was considerably enlarged, which is evidence of pit reactivation These changes occurred as consequences of the synergetic effect of strain and electrochemical repolarisation scan, consistent with the results observed in our previous study [17]. 210

211 Width (w) Pit 1 Height (h) Pit 1 strained (a) (b) Beeswax Beeswax Pit 1 reactivated Pit 2 Pit 3 Pit 4 (c) Front side Beeswax (d) Back side Beeswax Figure 7-10: Reconstructed X-ray CT data of the wire sample: (a) after first electrochemical polarisation scan (step 2), (b) after ~5% plastic strain (step 4), (c and d) after third electrochemical polarisation cycle (step 6) Pit morphology Figure 7-11(a, b, c, d) shows SEM images of pits morphology. Figure 7-11(a) shows the reactivated pit 1 growing along the beeswax crevice at the upper edge of the exposed wire, but it shows no evidence of the pit growing underneath the beeswax when compared with Figure 7-10(c). The upper part of the lacy metal cover is shown and it appears twisted due to fracture and pull-out from the pit mouth. The fracturing of the 211

212 lacy cover seems to be a consequence of applying strain before the pit was reactivated. The morphology of the reactivated pit shows that the lacy cover was partially developed, while the other part, closer to the beeswax, shows evidence of open pit morphology. Beeswax cover Cracks (a) Pit 1 (reactivated) (b) Pit 2 (c) Pit 3 Beeswax cover (d) Pit 4 Beeswax cover Figure 7-11: SEM images of the wire sample showing the four pits: (a) pit 1; (b) pit 2, showing lacy cover morphology with cracks; (c) pit 3; (d) pit 4. Pit 2 has the largest size of all pits, with a typical diameter in excess of 100 μm. Figure 7-11(b) shows its morphology with a metal lacy cover, consistent with pit growth at higher anodic polarisation [6, 7, 19]. Interestingly, the pit has a wide mouth, elongated in the direction of the applied strain. The lacy cover also has two regions with cracks between the holes, normal to the strain direction, at the periphery of the lacy cover. One 212

213 appears to start at the pit mouth, indicating the pulling up of the metal cover, and the second is below the pit mouth. The crack starting at the pit mouth appears to be due to mechanical fracture under stress, leading to propagation through perforations in the direction of the pit edge normal to the applied strain. The vertical sample position appears to have influenced the perforations below the pit mouth, as gravity influenced the accumulation of corrosive chemistry at the lower end of the pit. This is consistent with previous studies which suggest differences in pit morphology as a consequence of pit growth relative to sample position [15, 16]. For SCC it has been reported that cracks initiate from the pit bottom or the walls at the edge of pits [20]. In this study, small cracks were observed to start developing from the thin lacy metal cover, reaching to the edge of the pit, so they may also provide sites for further SCC to propagate at the pit shoulders [20]. The brittle cracks in the lacy covers may act as precursor sites for SCC, with SCC crack propagation being the predominant mechanism to consider. The morphology of the wire surface (Figure 7-11) shows horizontal marks which appeared when the sample was prepared by surface grinding. Figure 7-11(c, d) shows pits 3 and 4, which both initiated near the beeswax crevice, similar to the pit 1. They both formed at the lower end of the exposed wire. Morphologically, these were wide-open cavities with a brightly polished inner surface consistent with the bottom surface of a stable pit developed at high potential, which Sato [21] suggests will grow with a salt film. The upper parts of these pits have perforated covers (away from the crevice), indicating the direction of pit growth. This is the opposite morphology to that of reactivated pit 1. There is also no evidence that pit growth occurred underneath the beeswax, similar to pit 1. The metal covers appear to be kept intact by remnants of the oxide film, preventing them from falling into the pit. This is consistent with results reported for a horizontal sample, where pit growth was in an upward direction [22]. The internal morphology of pit growth was observed as shown in Figure 7-12(a, b). The result shows the formation of 3D lobes, which represent the undercutting mechanism of pit propagation in pit 2. Figure 7-12(a) shows the cross-sectional view of tomogram taken of three pits (reactivated pit 1, pits 3 and 4) which grew near the beeswax, using 2D slicing. It can be seen that the pit at the top grew downward, away from the upper edge of the exposed wire, while the two pits at the bottom grew upward, away from the 213

214 lower exposed edge. This result is consistent with previous reports of 2D pit growth in foil facing upward and growing close and below to the edges of the lacquer [8]. The contrasting fact that no pit growth was observed under the beeswax in the present work may be attributed to the use of beeswax instead of lacquer, which appears to have the advantage of eliminating crevice growth underneath it. Stable pit growth is a consequence of the lobe undercutting mechanism, which leads to perforation of the lacy metal cover, shown in the SEM images (Figure 7-11). Figure 7-12(b) shows a longitudinal 3D cross-sectional view of the wire revealing the internal morphology of three pits: reactivated pit 1, pit 2 and pit 3. The images show clearly the 3D lobe features, indicating three different growth behaviours. The reactivated pit 1 appears to grow downwards, away from beeswax, while the pit 3 lobes grow upwards, also away from beeswax. This pit growth in opposite directions appears to be due to the presence of beeswax at the edge of exposure surface, since the direction of pit growth in horizontal samples showed no difference compared to results for horizontal samples reported in [8]. The 3D view of pit 2 indicates that the lobes grew in all directions. 214

215 Beeswax Beeswax Pit 1 (reactivated) Pit 2 Pit 4 (a) Beeswax Pit 3 Figure 7-12: Morphology of pits grown in experiment (I) showing (a) tomogram section of 2D view of pits 1, 3 and 4; and (b) 3D surface view of pits 1, 2 and 3. (b) Beeswax Electrochemical polarisation Figure 7-13(a) shows current response vs time for the first polarisation and the second polarisation scan. The OCP measurement (step 0) stabilised at +93 mv vs Ag/AgCl before potentiodynamic polarisation to +607 mv vs Ag/AgCl (step 1). During first polarisation, the current started to increase steadily at +567 mv vs Ag/AgCl for 40 second before polarisation potential was terminated. This period of current evolution led to the growth of pit 1 (Figure 7-10a). After plastic strain was applied (step 3), the second polarisation was conducted. The measured OCP was +179 mv, followed by polarisation to +600 mv vs Ag/AgCl, with this potential maintained for 330 s. The current started to rise under potentio-static control at +600 mv after 96 s of exposure, and continued for 236 s, before the experiment was terminated. The results show that the gradient of the current vs time curve increased after strain was applied, relative to the current response without strain. This difference may indicate an increased rate of metal dissolution under the influence 215

216 of strain, either due to simultaneous formation and growth of several pits or simply accelerated dissolution. The OCP measurement indicates a potential shift by 86 mv in an anodic direction between the first and second polarisation cycles. This difference in potential may be a result of passive surface film growth, in line with observations reported in previous work [17]. (a) (b) Figure 7-13: (a) Current evolution over time of the potentiodynamic polarisation cycles before and after applying strain; (b) final pit geometry. 216

217 Pit geometry and estimated growth Figure 7-13(b) shows the pit geometries (depth, width and height) obtained at the end of polarisation from X-ray CT data, assuming that the middle slices of the pits represent the maximum dimensions. The pit parameters measured via segmentation, including volume and surface area, were also obtained and are listed in Table 7-4. Beside those direct measurements from 3D X-ray CT data, three additional geometrical methods (B, C and D in Table 7-4) were also explored to compare the common methods often used in the literature to estimate pit growth kinetics. In method A, all pit parameter values were taken at the end of polarisation from segmented X-ray CT data (pit geometry, volume and surface area). The surface area of each pit was obtained from segmented pit volume excluding the area of the metal cover, as it is considered passive during stable pit growth. The evolution of pit volume, surface area, radius, width and height were obtained assuming symmetrical pit shape over time, similar to method B (described below), using back extrapolation. In method B, pit volume was calculated from the charge passed during pit growth using the current-time curve in Figure 7-13(a). Faraday s second law equation was used to obtain the dissolved mass, finding total charge by integrating the area under the currenttime curve, and pit volume using V = QM/nFρ where Q is the charge in coulombs [5, 23]: The calculation assumed the average metal cation charge of metal dissolution n = 2.19, atomic weight M = g.mol -1, density of ρ = 7.97 g.cm -3 and Faraday s constant F = coulomb.mol -1 [11]. For surface area measurement, hemispherical pit shape was assumed, with radius of each pit based on the dissolved volume. The evolution of pit volume over time was used to obtain the evolution of pit radius (r) over time using the equation r = (3V 2π)^(1/3) and surface area (A) was obtained by assuming hemispherical pit shape and using the equation A = 2πr 2. The evolution of pit surface area over time was used to estimate current densities vs time assuming a homogeneous symmetric growth rate. From the current-time response curves in Figure 7-13(a), the first polarisation current over time was found to correspond to the growth of one pit; therefore, the above 217

218 methods can be used directly to estimate pit dimensions. In the case of the second polarisation scan, the current-time curve corresponded to the growth of three new pits as well as reactivation of the pre-existing pit 1. Therefore, numbers of assumptions were made to derive individual dissolution currents for each pit over time from the total current response. Based on the fit between each segmented pit volume obtained from X-ray CT data (approach A) and the dissolved metal volume of each pit obtained by Faraday s law (approach B) after the total current separated between the pits over time. Thus, the following steps were taken to estimate the current-time for each pit from the total current response over time: 1. The largest pit size was assumed to nucleate first, to stay active until the end of the polarisation cycle and grows with a constant rate. 2. The second pit was assumed to initiate at the time when a gradient change occurred. This pit then stayed active until the end of polarisation and was smaller size than the first. This assumption also applies to the third pit. 3. Pits of similar volume were assumed to initiate at the same time and to grow at similar rates; therefore the current was divided between them equally. 4. After identification of the starting time of each pit, the current over time was divided after the first gradient change between the first and second pits using a fixed ratio. The above assumptions are justified by the finding of a good fit between segmented pit volume from X-ray CT data (approach A) and the dissolved metal volume of each pit obtained by Faraday s law (approach B) after the total current was allocated among the pits over time. The fixed ratios which satisfied these assumptions were a half (1/2) of total current taken by each new pit initiated over time. In this method, half of the current was divided between pit 2 and pit 3 after 63 s of growth of the larger pit (pit 2), when the gradient changed, until 126 s, where another gradient change occurred. This change was sharp and corresponded to two pits (reactivated pit 1 and pit 4), because they had similar volumes. Using the above ratio, the current was divided equally between the old and the new pits at 126 s. Thus, each pit was assigned a quarter of the total value of current response over time after 126 s. 218

219 The result of current separation over time in Figure 7-14 was used to determine each pit volume using method A. Comparison of the volumes estimated with this approach, and the volume obtained by X-ray CT data revealed a good fit, with maximum differences between them less then 18%. In method C, pit volume and surface area were calculated from pit depth (r) obtained from the X-ray CT measurement at the end of polarisation, with the assumption that pits were hemispherical, with volume (V) given by V = 2 3. π. (r)3. The evolution of pit volume and surface area were determined by assuming symmetric pit growth from a local initiation point at the centre of the hemisphere and using back calculation methods over time, corresponding to method A. In method D, pit volume and surface area were calculated on the alternative assumption of semi-ellipsoid shape with pit depth (r), width (w) and height (h) obtained from X-ray CT measurements at the end of polarisation. The volume (V) of the semi-ellipsoid was estimated using the formula V = π/6 r. w. h and surface area was estimated from Equation 7-3 [19]: A = 2π [ 1 {(w 3 2 ) r ( w 2 ) ( h 2 ) r ( h 2 ) }] (7-3) The pit volume and surface area were obtained at the end of polarisation. Then, the aspect ratios between pit depth/width and depth/height were maintained during the evolution of pit geometry by back extrapolation. Table 7-4 lists the results of the four methods of measuring pit geometry, volume and surface area, which indicate that at the end of polarisation the pit volumes calculated via segmentation (method A) are close to those calculated via Faraday s law (method B), with a difference less than 18%. The difference may be explained by the uncertainty in obtaining segmented pit volume by X-ray CT data or in methods of current separation over time. In this analysis, the difference in dissolution rates between pits growing close to the beeswax crevice and those unaffected by the crevice was not considered, as it induced a different pit morphology. Likewise, the contribution of current consumption inside the pit due to cathodic reaction, which is typically suggested to be 5% of the total current, was also not considered [11]. 219

220 Table 7-4: Measured pit geometries of the four growth approaches Methods Pits Depth, r (μm) Width, w (μm) Height, h (μm) Volume, V (μm 3 ) Surface area without lacy cover (μm 2 ) Pit Pit 1 (strained) x x A: Data from X-ray CT Pit 1 (reactivated) Pit Pit Pit B: Faraday approach + assumption of hemispherical pit shape C: Depth measured by X-ray CT + assumption of hemispherical pit shape Pit 1 34* 68* 68* * Pit 1 (reactivated) 70* 140* 140* * Pit 2 88* 176* 176* * Pit 3 81* 162* 162* * Pit 4 70* 140* 140* * Pit * 58* 51054* 5281* Pit 1 (strained) 26 52* 52* 36792* 4245* Pit 1 (reactivated) 35 70* 70* 89752* 7693* Pit * 166* * 43263* Pit * 124* * 24140* Pit * 74* * 8597* Pit * 6762* D: Data from Pit 1 (strained) * 6950* X-ray CT + Pit 1 (reactivated) * 17981* assumption of Pit * 52151* semi-ellipsoid pit shape Pit * 35402* Pit * 23677* *Estimated values in bold 220

221 Figure 7-14: Estimated current evolution of each pit from total current evolution of the second polarisation Reactivation of pit 1 There is clear evidence of a synergetic effect of strain and repolarisation potential (steps 3 and 6, Table 7-3) on the existing pit 1 (Figure 7-15a), as its volume increased more than five times, from 93,691 μm 3 to 618,199 μm 3 (Table 7-4) and Figure 7-15c). After the application of strain, the pit was elongating in the direction of applied strain, with an increase in the pit height from 70 to 78 μm (Table 7-4) and Figure 7-15b. The result of X-ray CT scan 3 shows the reactivated pit 1 and the formation of three new pits (Figure 7-10c). The reactivation of pit 1 occurred under the synergetic effects of strain and electrochemical polarisation, and pit growth occurred adjacent to the beeswax covered area, similar to pits 3 and 4. In a previous study, repolarisation without applied strain was found to activate new pits but not to reactivate pre-existing pits [24]. The results show that the reactivated pit 1 (Figure 715a) grew in the direction away from the beeswax covered area with lobe features. Fractures of the metal lacy cover appear due to the introduction of strain before repolarisation. Strain appears to influence pit reactivation by weakening the passive film, especially above inclusions, and when the film breaks down, active pit dissolution begins to accelerate the local attack. This is also consistent with pitting potential being reduced under strain and affecting the rate of 221

222 pit dissolution [4, 25, 26]. Our assumption is that pit 1 was reactivated after the activation of the larger pits 2 and 3. The delay in reactivation could be attributed to a drop in potential inside the existing pit relative to the metal surface, reducing the driving force of applied potential at the pit bottom, resulting in a later activation compared to the larger pits. c Beeswax a b Figure 7-15: Tomogram section of 2D pit 1 (a) before strain, (b) after strain and (c) after reactivation Pit current density and stability product Current density over time was estimated by dividing current by pit surface area, calculated by methods A, B, C and D (Table 7-4). Figure 7-16(a) presents mean current density as a function of time for pit 1 generated by first polarisation (step 1, Table 7-3), for each of the four methods. The values based on actual pit surface area (method A) are seen to be lower than for any of the other methods (B, C and D) which are commonly used to estimate mean current density [5, 13, 14, 23, 27]. For all four methods, the behaviour of current density over time can be divided into two distinct regions of the graph: it first increases rapidly, reaching a maximum value after a few seconds, then decays for the rest of the time. The maximum current density can be attributed to the 222

223 precipitation of a salt film induced by diffusion control of stable pit growth, which causes the current density to decay due to the increasing size of the pit over time. These two regions are consistent with a transient and quasi-steady-state period, which has been reported for pit growth [28]. Pit 1 reached a maximum current density of 4.8 A.cm -2 at the transient region as calculated by method A, decaying to a mean value of 1.1 A.cm -2 after 40 s of pit growth. Current density is seen to fluctuate as it decays, suggesting the process of activation and passivation associated with the development of lobes at the pit bottom or the formation of new holes in the perforated cover. The estimation methods (B, C and D) gave values that were higher than the values obtained for method A, by 23%, 74%, and 36% respectively. The current decay associated with 3D pit growth was less distinct than that reported for a 1D pit. This appears to result from a higher dissolution rate relative to pit geometry, which accelerates the process of salt film precipitation [29]. The current densities obtained by method B were closer to the method A values. This may be explained by the overestimation of pit depth by method B, giving higher values for pit surface area. The results obtained by methods C and D indicate that the pit shape was closer to a semiellipsoid (elongated dish) than a perfect hemisphere, which is supported by the observations in Figure

224 (a) (b) Figure 7-16: Pit 1 (a) current density over time, (b) stability product vs. time; estimated by methods A, B, C and D ( Table 7-4). Figure 7-16(b) shows the stability product of pit 1 over the time of pit evolution obtained via methods A, B, C and D. The pit stability product is derived by multiplying the current density by pit depth. Transition to a stable pit has been suggested to occur above 0.3 A.m -1 [11]. Below this value, it is suggested that open pit turn passive as a consequence of metal ion concentration lower than the critical value needed to maintain active pit dissolution. This may occur when the metal dissolution rate falls below the diffusion rate. At an early stage of pit growth, all methods show the stability product below the critical value of 0.3 A.m -1, suggesting that pit growth was supported by the metal lacy cover acting as a diffusion barrier [11]. After about 4 s, the stability product began to exceed 0.3 A.m -1 and it remained thereafter in the range of stable pit criteria [11]. The stability product value was lower for method A than for any of the three estimated methods; values were about 50% higher for methods B and D, while the difference increased to 100% with method C. Similarly, Figure 7-17 (a, b, c, d) shows mean current densities as a function of time for the four pits (reactivated pit 1, pits 2, 3 and 4) obtained by the second polarisation scan, after strain was applied (step 6, Table 7-3). The current density of reactivated pit 1 was 224

225 calculated with the assumption to be a new pit, showing two regions of pit growth current density, similar to pit 1 in the first polarisation (Figure 7-16a). For pit 2, the results for using method A show a maximum current density of 6 A.cm -2, an increase of 25% relative to pit 1, with the latter for pit 1 obtained without applied strain. This difference can be attributed to the effect of strain. Pit 3 initiated after 63 s and reached a maximum current density of 16 A.cm -2, while pit 4 and the reactivated pit 1 grew at the same time, reaching respective maximum current densities of 19 and 18 A.cm -2. The current density then decayed, reaching stable values below 1 A.cm -2. These results show that pit 2 grew first, with a lower maximum current density than that of the pits which were nucleated or activated after time. The absolute magnitude of maximum currents observed are affected by the way the currents are separated at the time of pit initiation and before salt formation. Therefore, when a new pit nucleates it is exposed to a large current (assumed to be half of the total current) relative to pit area, which would induce a high current density during new pit growth and accelerate pit dissolution. (a) 225

226 (b) (c) 226

227 (d) Figure 7-17: Current density vs. time for reactivated pit 1 and pits 2, 3 and 4 after strain and repolarisation, calculated by methods (a) A, (b) B, (c) C and (d) D. The patterns of current density estimated by methods B, C and D were similar to those obtained by method A, but with variations in maximum current density. For pit 2, the differences of maximum current density at the end of pit growth were 19%, 34% and 11% for methods B, C and D respectively. This indicates that method D, assuming a semi-ellipsoid pit shape, gave results closest to method A, with the longer growth time of 236 s, whereas for the pit obtained by the first polarisation scan, with only 40 s of pit growth, method B gave values closest to method A. This suggests that increasing the duration of pit growth leads to a more elongated dish shape. For pits 1, 3 and 4, which grew near the beeswax perimeter, there were large differences in the values obtained by methods C and D compared to method A, due to the effect of the shape of pits. The difference in maximum current between methods A and B was below 22% for pit 1 obtained during the 1 st polarisation scan. The difference in maximum current density for pit 1 before and after reactivation could be explained both by the introduction of plastic strain, which has been reported to accelerate metal dissolution [30], and by the influence on pit growth morphology along the beeswax crevice. 227

228 After reaching a maximum value, the current density dropped for all pits, associated with pit growth until current density stabilised below 1 A.cm -2 according to method A. However, the overall range of current densities obtained by the different estimation methods is not consistent with reported pit current density [24, 27] obtained by method A, indicating that real current density is lower due to the variation in pit surface area. The values of stability product over time for pits grown by the second polarisation, scan, estimated by method A, B, C and D, are plotted in Figure 7-17(a, b, c, d). At an early stage of pit growth, the values were below the stability product criterion of 0.3 A.m -1, emphasizing the role of the lacy metal cover in retaining corrosive solution inside the pit. Stability products then increased with time to reach the 0.3 A.m -1 threshold, which is consistent with reported values of stable pit growth [11]. The influence of strain seems also to increase the maximum stability product compare with a pit grown without the effect of strain. Strain can increase stability product by exposing a new surface to corrosive solution and increased metal dissolution rates. However, in the early stages of pit growth, strain may introduce cracks in lacy cover and affect the pit stability negatively, if the rate of metal dissolution is low before the crack. The fluctuation in pit stability observed when new pits grow, as well when pit 1 is reactivated, is an artefact of the mean current density separation between pits, described in the previous section. Therefore, the stability product increased sharply to a high value due to the high current density when a new pit was initiated, then reduced as pit size increased. The stability product obtained by all four methods was mostly above 0.3 A.m -1. Values for pit 2, for example, estimated by methods B, C and D, differed from the stability product obtained using method A by 26%, 19% and 11% respectively. This indicates that among the three estimations, methods D, assuming a semi-ellipsoid pit shape, gave a result closest to method A. However, the difference was greater for pits which developed near the beeswax crevice, most likely due to a larger difference between real pit shape and that assumed by the estimation methods. 228

229 (a) (b) 229

230 (c) (d) Figure 7-18: Stability product vs. time for reactivated pit 1 and pits 2, 3 and 4 after strain and repolarisation, calculated by methods (a) A, (b) B, (c) C and (d) D. 230

231 Pit diffusion product A linear relationship between square of pit depth (r 2 ) and time of pit growth has been used to estimate the diffusion product of stable pit growth under diffusion control. The diffusion product (DΔC) can be estimated from the slope of pit growth using Equation 7-4 [8, 11] r 2 = 3M D C πρ t (7-4) Figure 7-19(a) shows the result of this relationship and the value of DΔC for all pits. The diffusion product for pit 1 was mol.cm -1 s -1 after first polarisation, then after the application of strain and repolarisation, pit 2 had a larger value of mol.cm -1 s -1, while pits grown near the beeswax crevice (reactivated pit 1, pits 3 and 4) had lower values of , and mol.cm -1 s -1 respectively. The difference between pit growth before and after strain can be observed between pit 1 and pit 2, where the increase in diffusion product was similar to our previous result [17], which can be attributed to an increased rate of metal dissolution. This value of DΔC is in the range reported for 1D pit growth facing upward [6, 31]. The diffusion products of pit 1 (without strain) and pits grown near the periphery/beeswax were in the range of values reported in 2D studies [8]. The difference between pits growing in parallel (reactivated pit 1 and pit 4) appears to occur due to the assumption that pit 1 grew as a new pit, similar to pit 4. The result for pit 1 before reactivation is also consistent with our previous result [24] of mol.cm -1 s -1 which was obtained for pit growth with a metal lacy cover under similar conditions. The difference in the diffusion product between unstrained and strained wire may indicate increased metal dissolution rates due to strain [4, 25, 32]. The diffusion product of pit 2 obtained after strain and repolarisation is similar to 1D pit growth. However, in a 1D pit, diffusion is constrained by the geometry of a pit growing in one dimension but without a cover. The position of the sample in this study may also have increased the diffusion product by allowing dissolved ions to escape from the pit by gravity, compared with pit growth upwards in 1D or 2D studies [8, 31]. Therefore, in 3D pits grown under strain, accelerated dissolution resulting in increased concentration gradient (ΔC) will increase the diffusion product relative to pits without strain. 231

232 Assuming that pits grow under diffusion control with a metal salt concentration of 4.2 M at the bottom and neglecting the concentration at the pit mouth [11], from the previous value of DΔC, the diffusion coefficient (D) at the pit bottom can be estimated for pit 1 (without strain) and pit 2 (with strain) at and cm 2.s -1 respectively. An increase in diffusion coefficient suggests a more open pit cover as a result of strain, which reduces the diffusion barrier. However, in Figure 7-18(a) the linear relation and changes in slope indicate the effect of changing the dissolution rate after strain. Alternatively, the variation of effective diffusivity in pit growth over time can be estimated using equation 7-5: [11] C = 2π 3nFD i. r (7-5) Figure 7-19(b) shows diffusion coefficient over polarisation time assuming 4.2 M saturation concentration of ions. It shows that the diffusion coefficient increased to a maximum value, then remained fairly constant over time, probably due to the formation of a salt film. The higher value of the diffusion coefficient over time for pit 2 appears to be due to the effect of strain on dissolution during pit growth. This may indicate the influence of strain on the metal lacy cover. The results show that the effective diffusion coefficient at maximum growth for pit 1 before applied strain was cm 2.s -1, while the value for pit 2 after applied strain was cm 2.s -1. The diffusion coefficient of pit 2 dropped to lower values after new pits (pit 3, reactivated pit 1 and pit 4) began to grow, but all values remained at a constant range. The very high values at early stages of pit growth is an artefact of the current separation methods, as discussed in the previous section On the other hand, if we assume a constant diffusion coefficient of cm 2.s -1 [33], the metal dissolution concentration of pit growth over time can be estimated as shown in Figure 7-19(c). This shows an increase to a constant saturation concentration of 4 M for pit 1, whereas for pit 2 (after strain), it reached 6 M, implying an increase in metal dissolution rate under the influence of strain. The graph also shows a reduction to 4 M in metal ion concentration in pit 2 over time when new pits grew, while the new pits grow with metal ion concentrations between 3 and 5 M. 232

233 (a) (b) 233

234 (c) Figure 7-19: (a) Relationship between (depth) 2 vs. time and diffusion product (slope) for all pits before and after the application of strain; (b) diffusion coefficient of all pits, assuming a constant salt concentration of 4.2 M; (c) metal ion concentration over time of all pits Experiment (II) In the second experiment, the exposure time of polarisation and applied strain were increased relative to experiment (I), to examine the influence of these parameters on the reactivation of existing pits. Figure 7-20(a) shows a 3D image of the reconstructed wire from X-ray scan 1, indicating the growth of a single pit. Figure 7-20(b) shows the deformation of pit geometry after the application of ~10% plastic strain (step 3, Table 7-3). It was not possible to obtain a good reconstructed image of this step due to blurring of the X-ray image, which may have been caused by a slight relaxation of strain during scanning. Figure 7-20(c1-c3) shows the result obtained by X-ray CT scan performed after the wire was removed from the electrolyte cell to examine the exposed area after elongation, which had moved out of the field of view. Therefore, a repeated X-ray scan 4 was conducted with 0.71 μm 3 voxel size resolution in air by readjusting the distances of both the X-ray source and the detector from the wire sample with a field of view of µm 2 and projections recorded with 1 1 binning with an exposure time of 10 seconds for each projection. It shows again reactivation of pre-existing pit 1, with however only a slight increase in pit size; however, in this experiment a significantly increased numbers of new pits were activated. It can also be seen that three pits 234

235 activated (pit 8,9 and 10) during the second polarisation scan grew until they merged. with reactivated pit 1. Lacy metal covers were also developed throughout pit growth. Pit 1 Pit 1 strained (a) (b) Pit 2 Pit 4 Pit 6 Pit 5 Pit 1 Pit 7 Pit 7 Pit 8 Pit 9 Pit 10 Pit 11 (c1) Front view (c2) Side view (c3) Back view Figure 7-20: Reconstructed X-ray CT data of wire sample (a) after 1 st electrochemical polarisation (step 2), (b) with ~10% plastic strain (step 4) and (c1-c3) after 2 nd electrochemical polarisation cycle (step 6). The data for figures (c1-c3) were acquired after removing the wire from the solution, producing a reconstructed voxel size of 0.71 µm

236 Electrochemical polarisation The result of electrochemical measurements show that the OCP prior to the first polarisation was -51 mv vs Ag/AgCl (Table 7-3, step 0), changing to +177 mv vs Ag/AgCl prior to the second polarisation (step 6). A similar shift was observed in previous experiments [17, 24]. The OCP prior to the second polarisation was almost the same as in experiment I, which suggests an insignificant influence of increasing plastic strain on OCP relative to anodic polarisation. The current response of the first polarisation cycle, performed from OCP to a maximum potential of +607 mv, shows an increase in current over 76 s. The current-time curve in Figure 7-21(a) shows a gradual increase in current with small fluctuations over time. These fluctuations are associated with pit growth, and may be attributed to both active and passive processes, i.e. the formation of new holes in the perforated cover on the metal surface and the undercutting mechanism of lobes at the pit bottom[8]. (a) 236

237 (b) Figure 7-21: (a) Current evolution over time for 1 st potentiodynamic polarisation scan, with evolution of dissolved volume; (b) current evolution of 1 st polarisation compared to 2 nd polarisation scan after 10% strain was applied. Figure 7-21(b) shows the evolution of current during the second polarisation scan (step 6) performed from OCP up to +600 mv and held at this potential for 198 s before the experiment was terminated. It shows that at +456 mv, the current began to increase gradually, in the potentiodynamic polarisation region, but with a higher gradient than the first polarisation until 25 s. This difference appears to be due to the increasing dissolution rate after the application of strain prior to the second polarisation. This observation is consistent with result obtained in experiment I. The results show that current response during the second polarisation started to increase at a potential below the first polarisation potential by 72 mv, possibly due to the effect of strain. This is consistent with a previous study reporting a marked effect of strain on pitting susceptibility [32]. The current response during the second polarisation scan increased continuously for 144 s under potentiodynamic polarisation and for 198 s under potentiostatic polarisation, with a total pit growth time of 342 seconds. The drop spike in the current-time curve (Figure 7-21b) is an artefact of switching the potentiostat from potentiodynamic to potentiostatic control during this experiment. 237

238 Morphology of pit growth The morphology of the wire surface and pits observed by SEM is shown in Figure 7-22 (a, b, c) after the X-ray CT experiment was terminated (step 8, Table 7-3). Figure 7-22(a) shows pits grown on one side of the exposed wire. Clearly visible are both edges of the exposed wire, the beeswax cover and a number of pits distributed across the surface, having lacy metal covers with fractures and cracks in them. Figure 7-22(b) is a magnified image of the upper section of the wire shows three pits with classic lacy covers, but with brittle fractures, with pit 5 showing cracks through the pit mouth and the perforated cover normal to the applied strain. Two pits (2 and 3), not shown in this figure, grew also at the interface to the beeswax. Comparing these results with those of experiment I reveals the effect of increasing strain from 5% to 10% on fracture morphology and crack formation in metal lacy covers, as well as the number of pits grown. Figure 7-22(c) is a magnified image of the lower section of the wire, showing a large open pit (pit 9) growing close to the beeswax, resulting in crevice. Pits 8, 9 and 10 merged and grew into one large pit/crevice. A remaining section of a lacy metal cover on the edge of pit 9 indicates that the direction of pit growth was away from the beeswax crevice. Pit 9 has a smooth, bright surface, with indications of undercutting and lobes. However, pit 9 may also be a result of several pits merging into one pit. The reactivated pit 1 also shows a wide, open morphology, with a lacy cover section at the upper edge. It can be seen that the wall between reactivated pit 1 and pit 8 has begun to be perforate, which indicates that they begun to merge below the surface. Pit 8 also shows signs of pit nucleation within the existing pit cavity. Pit 10 has a lacy cover with fractures, possibly as a consequence of strain applied to the wire. 238

239 Pit 4 Pit 6 Pit 5 (b) Reactivated Pit 1 (a) Pit 10 Pit 8 Pit 9 (c) Figure 7-22: SEM images (a) of the wire sample showing the growing pits; (b) close-up view of pits with lacy cover containing cracks and fractures; and (c) close-up view of large open pit growing close to beeswax resulting in a crevice and connection to other pits. The 3D morphology of all pits was extracted using horizontal and vertical crosssectional views as shown in Figure 7-23(a, b). The red dashed line in Figure 7-23(a) marks the position of the cross-sectional view with result is shown in Figure 7-23(b). It can be seen that 3D lobes grew on the upper side of pit 9, similar to those observed in experiment I. Also clearly visible is a pit growing inside pit 8 (Figure 7-23a). This pit is also shown in Figure 7-23(b), which indicates that growth occurred below the 3D lobes developed at the bottom of pit 8. This may indicate active pit growth, which is similar to the first step of early pit growth after the breakdown of the passive film prior to undercutting by lobes at the start of lateral growth. Figure 7-23 also indicate that pits 10, 9, 8 and reactivated pit 1 are connected below the surface. There is a 3D lobe growing at the bottom of pit 6 and 8, which appears to have reached the metal surface, except on one side. Such 3D lobes were also observed in 239

240 experiment I and are consistent with reported undercutting mechanism of pit propagation observed by 2D radiography of pit growth in 2D foil [8]. 3D lobes grow below pit 6 bottom Pit 2 Pit inside pit 8 shows initiation of 3D pit prior to lobe formation (a) Figure 7-23: X-ray CT data visualising the 3D morphology of the inside of the wire, showing (a) 3D cross-sectional view of all pits, (b) 3D vertical view of the outside Pit geometry Table 7-5 lists all pit dimensions after polarisation and applied strain with maximum depth, width and height obtained from the X-ray CT data. It also lists the segmented volume and surface area of the pits. Pits 8, 9 and 10 and reactivated pit 1 were connected beneath the surface and are therefore represented by single values of volume and surface area. Beside the direct measurements from X-ray CT data, electrochemical data were used to calculate the total pit volume from the charge passed during the period of pit growth, as shown in Table 7-5. One-sided growth of 3D lobes close to crevice of pit 9 The table shows that the volume of pit 1 calculated by segmentation was only half of that obtained by Faraday s law after the first polarisation. This implies that, most likely, one other pit was not in the field of view during this X-ray CT scan. This assumption is confirmed by calculating the total dissolved volume after the second polarisations scan after increasing the field of view, indicating a good fit between the methods volumes measured by data segmentation and calculated via Faraday s law. The difference (b) Reactivated pit 1 Pit

241 between the two methods is in the range of 3%, which shows a good fit similar to the result obtained in previous experiments I. Table 7-5: Measured pit geometries for experiment II Methods A: Data from X-ray CT B: Faraday approach + assumption of hemispherical pit shape Depth, r Width, w Pits (μm) (μm) Height, h Volume, V Surface (μm) (μm 3 ) area (μm 2 ) Pit Pit 1 (strained) Blurred image Pit Pit Pit Pit Pit Pit Pit 1 (reactivated) Pit Pit Pit Pit All pits Pit * 82* * All pits The geometry of pit 1 after strain was introduced shows a slight elongation along the straining direction (height), with reduced pit depth and width. These measurements were taken from a blurred image, as described above; however, after application of a second polarisation scan the measurements shows an increase in pit geometry relative to the original values, indicating re-activation, consistent with the results discussed in method A, but with somewhat a smaller activation pit size. This difference may be attributed to the influence of a large number of activated pits in this scan, which may also be a result of the magnitude of applied strain and the open morphology of pit 1, causing rapid dilution of the corrosive environment under the effect of gravity. Pitting 241

242 kinetics were not estimated for these data sets, since the presence of pits not captured in the tomography data sets introduces further uncertainties Conclusions 1- The synergetic effect of strain and electrochemical polarisation induced reactivation of corrosion pit 2- Differences in the morphology of pits and their associated lacy metal covers were found, with cracks in the lacy metal covers, resulting in sharp cracks at the cover edges, and brittle fractures normal to the direction of applied strain. 3- The application of higher strain resulted in pits nucleating at lower applied potentials, with growth of multiple smaller sized pits 4- The estimated maximum current density increased after the application of strain, and pits grew with a stability product above or close to 0.3 A.m The diffusivity parameter (DΔC) of 3D pit growth before strain was mol.cm -1.s -1, but after strain was applied, this increased to mol.cm -1.s The 3D pit growth rate and diffusion product indicate diffusion control of the growth of pit depth, adding to the effect of strain on SCC Acknowledgements The authors acknowledge the provision of beam time at the Henry Moseley X-ray imaging Facility (HMXIF) of the University of Manchester, UK, established with funding from the EPSRC through grants EP/F007906, EP/I02249X and EP/F FAA and DLE would like to thank the Saline Water Conversion Corporation (SWCC), Saudi Arabia for financial support References 1. Olsson, C.O.A. and D. Landolt, Passive films on stainless steels chemistry, structure and growth. Electrochimica Acta, (9): p Soltis, J., Passivity breakdown, pit initiation and propagation of pits in metallic materials - Review. Corrosion Science, : p Suter, T., et al., Pit initiation on stainless steels in 1 M NaCl with and without mechanical stress. Journal of the Electrochemical Society, (5): p. B174-B

243 4. Shimahashi, N., et al., Effects of Corrosion and Cracking of Sulfide Inclusions on Pit Initiation in Stainless Steel. Journal of the Electrochemical Society, (10): p. C494-C Newman, R.C. and E.M. Franz, Growth and Repassivation of Single Corrosion Pits in Stainless-Steel. Corrosion, (7): p Ernst, P. and R.C. Newman, Pit growth studies in stainless steel foils. I. Introduction and pit growth kinetics. Corrosion Science, (5): p Laycock, N.J., et al., Perforated covers for propagating pits. Journal of the Electrochemical Society, (4): p Ghahari, M., et al., Synchrotron X-ray radiography studies of pitting corrosion of stainless steel: Extraction of pit propagation parameters. Corrosion Science, : p Laycock, N.J. and R.C. Newman, Localised dissolution kinetics, salt films and pitting potentials. Corrosion Science, (10-11): p Gaudet, G.T., et al., Mass-Transfer and Electrochemical Kinetic Interactions in Localized Pitting Corrosion. Aiche Journal, (6): p Pistorius, P.C. and G.T. Burstein, Metastable Pitting Corrosion of Stainless Steel and the Transition to Stability. Philosophical Transactions of the Royal Society of London Series a-mathematical Physical and Engineering Sciences, (1662): p Frankel, G.S., et al., Metastable Pitting of Stainless Steel. Corrosion, (7): p Tian, W., et al., Effects of applied potential on stable pitting of 304 stainless steel. Corrosion Science, (0): p Tian, W.-M., et al., Pitting Kinetics of 304 Stainless Steel Using ESPI Detection Technique. Acta Metallurgica Sinica (English Letters), (4): p Mankowski, J. and Z. Szklarska-Smialowska, Effect of Specimen Position on Shape of Corrosion Pits in an Austenitic Stainless-Steel. Corrosion Science, (9): p Wu, K., W.-S. Jung, and J.-W. Byeon, In-situ monitoring of pitting corrosion on vertically positioned 304 stainless steel by analyzing acoustic-emission energy parameter. Corrosion Science, : p Almuaili, F.A., et al., Strain-induced Re-activation of Corrosion Pits in Austenitic Stainless Steel Submitted to the Corrosion Science, Feldkamp, L., L. Davis, and J. Kress, Practical cone-beam algorithm. JOSA A, (6): p Ernst, P., et al., The mechanism of lacy cover formation in pitting. Corrosion Science, (6): p Turnbull, A., L.N. McCartney, and S. Zhou, Modelling of the evolution of stress corrosion cracks from corrosion pits. Scripta Materialia, (4): p Sato, N., The Stability of Localized Corrosion. Corrosion Science, (12): p Eklund, G.S., Initiation of Pitting at Sulfide Inclusions in Stainless-Steel. Journal of the Electrochemical Society, (4): p González-Sánchez, J., et al., Corrosion pit growth on austenitic stainless steels in chloride containing solution: a quantitative approach. Anti-Corrosion Methods and Materials, (5): p Almuaili, F.A., et al., Application of a Quasi In Situ Experimental Approach to Estimate 3-D Pitting Corrosion Kinetics in Stainless Steel. Journal of the Electrochemical Society, (13): p. C745-C

244 25. Shimahashi, N., et al., Effect of Applied Stress on Dissolution Morphology and Pit Initiation Behavior of MnS Inclusion in Stainless Steel. ECS Transactions, (31): p Guan, L., et al., Effects of cyclic stress on the metastable pitting characteristic for 304 stainless steel under potentiostatic polarization. Corrosion Science, (0): p Alkire, R.C. and K.P. Wong, The Corrosion of Single Pits on Stainless-Steel in Acidic Chloride Solution. Corrosion Science, (4): p. 411-&. 28. Tester, J.W. and H.S. Isaacs, Diffusional Effects in Simulated Localized Corrosion. Journal of the Electrochemical Society, (11): p Moayed, M.H. and R.C. Newman, Deterioration in critical pitting temperature of 904L stainless steel by addition of sulfate ions. Corrosion Science, (11): p Devasenapathi, A. and M. Asawa, Effect of applied potential on the nature of surface film and SCC of a high Mn stainless steel in 1 M HCl. Journal of Materials Science, (23): p Moayed, M.H. and R. Newman, Using pit solution chemistry for evaluation of metastable pitting stability of austenitic stainless steel. Materials and Corrosion, (3): p LÜ, G., et al., Effect of Strain and Chloride Concentration on Pitting Susceptibility for Type 304 Austenitic Stainless Steel. Chinese Journal of Chemical Engineering, (2): p Kuo, H.C. and D. Landolt, Rotating-Disk Electrode Study of Anodic Dissolution or Iron in Concentrated Chloride Media. Electrochimica Acta, (5): p

245 8. SUMMARY AND CONCLUSION (i) Pitting corrosion susceptibility of austenitic stainless steel and its relationship to inclusion content The relationship between microstructural inclusion content, susceptibility to pitting corrosion, and the number of corrosion pits was examined in three conventional austenitic stainless steels (types 303, 304 and 304L) exposed to chloride-containing solution. Measurements of OCP over time indicate an effect of contents and chemical composition of inclusions on the stability and behaviour of passive surface films. Microstructures with higher inclusion contents have less noble OCPs, with far larger potential fluctuations over time. This indicates inclusion dissolution and breakdown of the passive film, which is the first step in pit initiation. Higher electrochemical polarisation potential (above the pitting potential) activates more sites for pit nucleation. The results also show that the density of pits is related to the chloride concentration and suggest a direct relationship between pitting density and inclusion density in the microstructures investigated. (ii) Application of a quasi in-situ (3D) method to characterise pitting corrosion A quasi in-situ experimental approach has been developed to study the 3D pitting corrosion of stainless steel in bulk solution containing chloride, using a miniature electrochemical cell. Electrochemical polarisation was conducted by potentio-dynamic and potentio-static methods, while X-ray CT was used to visualise 3D pitting corrosion. The 3D pit geometries with associated current response were used to estimate the kinetics of 3D pitting corrosion, and results compared with 1D and 2D pitting corrosion kinetics from the literature. Pit morphology and growth: Observation of polished internal pit surfaces in conjunction with elongated dish-shaped pits indicated stable pit growth under diffusion control. However, faster lateral growth in width and height than in 245

246 depth suggests activation/ohmic control. 3D lobes with the typical undercutting pit growth were observed. The vertical sample position appears to affect the location of lobes initiation, allowing pits to grow in depth with increasing polarisation time and applied potential. Pit current density and stability product: Current density of 3D pit growth characterise with two distinct regions: a transient and then quasi-steady state period distinguished by maximum current density due to salt film formation. Current decay followed limited current associated with increasing pit size over the time of pit growth. There were also two regions of stability product corresponding to current behaviour. Pit stability was below 0.3 A.m -1 during early pit growth but increased over time to a stable region above this value indicating stable pit transition. The measurement of current density and stability product based on real pit geometry compared with estimated pit geometry, indicated overestimation or underestimation by as much as 50%, however all values were in the range of those reported for 1D and 2D pit growth. Two methods were used to estimate diffusivity parameter (D C) of 3D pit growth. The first obtained from the gradient of the linear relation between square of pit depth and time of pit growth; the second used, the stability product at the end of pit growth. The former indicated an average value while the later give a spontaneous value. Using the second method, the variation in diffusion coefficient over time was also estimated assuming diffusion control under a salt film with a saturation concentration of 4.2 M at the pit bottom, while concentration at the pit mouth was neglected. The diffusivity parameters of pit growth using first method show different values relative to pit morphology and dissolution rate over time (pit size). Small and open pits grown in a vertical sample showed lower values than larger pits with metal lacy covers, suggesting an influence of the lacy cover on maintaining the anolyte concentration necessary for an active metal dissolution rate in vertical samples. However, in horizontal sample where pits faced upwards, it was suggested that a lacy cover induced a diffusion barrier thus reducing the diffusion product relative to 1D pit growth with open morphology [1, 2]. In vertical samples, the diffusivity parameter is expected to be increased by gravity 246

247 [3, 4] allowing dissolved metal to escape from the pit through the pit mouth or through holes with no covering of oxide film [5]. These effects were observed with more in larger pits than small ones. Thus, in vertical samples, pits below 30 µm diameter grown in moderate chloride concentration appear not to be fully under diffusion control, whereas larger pits may become partially under diffusion control (at the lower end). Larger pits had larger diffusion products on vertical samples, implying the influence of lacy metal cover development and the increased size of the pit mouth in vertical relative to horizontal samples. The second method gave diffusion products of mol.cm -1.s -1 for an open pit and mol.cm -1.s -1 for pits with metal covers. These are higher than values obtained from the gradient but within the range reported for 2D pits and lower than those for 1D pits grown in horizontal samples. Assuming saturation concentration at the end of pit growth resulted in similar diffusion coefficients. This indicates the effect of pit morphology on chemistry inside the pit and the dissolution rate, which is influenced by sample position. This leads to different values compared with 1D or 2D diffusion parameters under diffusion control. (iii) Synergetic effects of applied strain and repolarisation potential on reactivated pit growth using 3D X-ray CT The synergetic effects of applied strain and repolarisation potential clearly show reactivation growth of existing pits beside activation of new pits. The size of reactivated pits was show to be related to strain and the time of repolarisation potential in vertical samples. The morphology of reactivated pits was characterised by fractures and cracking of the metal lacy cover at the pit mouth, which increased as applied strain was increased from 5% to 10%, while the reverse was true of the size of reactivated pits. However, increasing the strain activated large sites of new pits. The effect of increasing applied strain on metal covers may be one reason for the small size of activated pits and for the large numbers of pits activated over time. Newly activated pits had wide open mouths, elongated normal to the direction of strain, with cracks developing increasingly between holes when strain was increased by 10%, apparently as a consequence of residual stress. The effect of strain on SCC, especially pit-to-crack transition, may begin in the thin lacy metal pit cover. 247

248 The maximum current density of pit growth increased with strain by 30%, from 5 A.cm -2 before strain to 6.5 A.cm -2 after strain, while there was a 100% increase for the reactivated pit, apparently an artefact of the separation current assumption at early stages of pit growth, before transition to salt film control. However, at steady pit growth the current density dropped as pit size increased, approaching 1 A.cm -2 at the end of pit growth. This region of pit growth suggests salt film diffusion control. The pit stability product reached a stable value above 0.3 A.m -1 ; however, at the early stages of pit growth the value was below the stability criteria, suggesting that pit growth is supported by the metal lacy cover acting as a diffusion barrier. The stability product increased after strain by 30% due to an increased metal dissolution rate. In addition, the stability product was higher in reactivated pits, due to the higher current of separation assumption at the early stages of pit growth, but this was reduced to within the range of stable pit criteria. The difference between estimated methods and measurement methods in 3D pits indicates that overestimation of pit depth with hemispherical pit shape leads to close values of current density but a difference of about 16% in stability product, increasing with strain for reactivated pits. The diffusivity parameters of new and reactivated pits increased after applied strain due to the increased metal dissolution rate and diffusion rate of vertical samples close to 1D pits. 248

249 8.1 References 1. Ghahari, M., et al., Synchrotron X-ray radiography studies of pitting corrosion of stainless steel: Extraction of pit propagation parameters. Corrosion Science, : p Ernst, P. and R.C. Newman, Pit growth studies in stainless steel foils. I. Introduction and pit growth kinetics. Corrosion Science, (5): p Wu, K., W.-S. Jung, and J.-W. Byeon, In-situ monitoring of pitting corrosion on vertically positioned 304 stainless steel by analyzing acoustic-emission energy parameter. Corrosion Science, : p Mankowski, J. and Z. Szklarska-Smialowska, Effect of Specimen Position on Shape of Corrosion Pits in an Austenitic Stainless-Steel. Corrosion Science, (9): p Moayed, M.H. and R.C. Newman, Deterioration in critical pitting temperature of 904L stainless steel by addition of sulfate ions. Corrosion Science, (11): p

250 9. FURTHER WORK The following work is recommended for further studies in bulk environments, using the miniature electrochemical cell described in this thesis: 1- The effect of sample position on 3D pitting corrosion of stainless steel appears to influence the growth rate and morphology of pits. Testing samples facing either upwards or downwards during pit growth would give information on how gravity and pit chemistry influence the parameters and shape of growing pits relative to the results obtained in this study. 2- Pitting in more highly-alloyed materials could be compared with conventional stainless steel to explore the influence of alloying elements on the kinetics of pit propagation and pit morphology. 3- The influence of strain on 3D pitting corrosion kinetics of stainless steel and their transition to environment-assisted cracking could be assessed, to investigate the interaction between pits and SCC. 4- The effects of strain on the nucleation of cracks in lacy metal covers could be investigated to determine whether such cracks can propagate into the bulk material due to the synergetic effects of plastic strain and repolarisation. This can be achieved by conducted slow strain tensile test under SEM using single pit with lacy cover generated on mini tensile sample. 250

251 APPENDIX Appendix A A-1: Polarisation curves of 304L wire. A-2: Polarisation curves of 304 plate showing metastable pit events. 251

252 A-3: Polarisation curves of 303 bar. 252

253 Appendix B Tomogram section of 2D pit from middle slice of pits 1, 2 and 3 at at height and width view. SEM images of the pits Pit 1, height Pit 1 Pit 1, width Pit 2, height Pit 2 Pit 2, width 253

254 Pit 3, height Pit 3 Pit 3, width 254