UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Corrosion behavior of API X100 steel in near-neutral pH bicarbonate environments : experimental and modelling… Gadala, Ibrahim M. 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_may_gadala_ibrahim.pdf [ 13.89MB ]
Metadata
JSON: 24-1.0343472.json
JSON-LD: 24-1.0343472-ld.json
RDF/XML (Pretty): 24-1.0343472-rdf.xml
RDF/JSON: 24-1.0343472-rdf.json
Turtle: 24-1.0343472-turtle.txt
N-Triples: 24-1.0343472-rdf-ntriples.txt
Original Record: 24-1.0343472-source.json
Full Text
24-1.0343472-fulltext.txt
Citation
24-1.0343472.ris

Full Text

Corrosion behavior of API X100 steel in near-neutral pH bicarbonate environments: experimental and modelling studies  by  IBRAHIM M. GADALA  B.A.Sc. with distinction, The University of British Columbia, 2010 M.A.Sc., The University of British Columbia, 2012      A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Materials Engineering)       The University of British Columbia (Vancouver)  April 2017 © Ibrahim M. Gadala, 2017  ii Abstract API X100 is a new high-strength low-alloy steel which has garnered great interest from the pipeline industry due to the economic benefits it offers in terms of lower material, transportation, and fabrication costs. However, buried pipeline steels suffer from external corrosion and cracking, the mechanism of which in near-neutral pH environments is not yet fully understood. This dissertation presents and discusses the results of several electrochemical studies and numerical models conducted on X100 steel, contributing to a more complete understanding of the fundamental corrosion processes occurring in these environments. Improved simulation accuracy for corrosion rates and pipeline integrity is achieved. Applications of this research are strongest within the buried oil and gas transmission pipeline field, yet are extendable to other infrastructural and engineering applications such as utility piping systems and steel reinforcements of buried or concrete structures. Results of this work elucidate the criticality of even minor pH variations within the near-neutral pH environment on the corrosion and passivation of X100. Conflicting impacts of [HCO3−] versus %CO2 on corrosion rate are revealed, attributed to the gradual dominance of hydrogen evolution cathodic reactions involving HCO3− species at higher pH. A finer three region subdivision of the near-neutral pH range is proposed based on the dissolution, dissolution-adsorption, adsorption-diffusion, and diffusion-controlled behaviors which appear as pH, [HCO3−], [Cl−], [SO42−], and temperatures are changed. Dynamic electrochemical impedance spectroscopy identifies the role of Cl−/SO42− in increasing corrosion rate, decreasing pre-passive stage diffusion, and decreasing passive layer protection. Increased porosity of FeOOH tubercle structures formed in low dissolved [O2] environments augments diffusion therein, spurring the exclusive formation of Fe2O3 underneath instead of only Fe3O4. Hydrogen diffusivity in X100 steel is found to be 4.4x10-7 cm2/s. During free corrosion, a diffusible hydrogen concentration of 1 atomic ppm is also measured in the alloy. Simulations in the Finite Element models developed  iii indicate ideal anode placements and applied voltages for an underground cathodic protection system. Additionally, the transient evolution of an external corrosion defect which reaches 3 mm deep within 3 years is visualized, wherein the pipeline’s structural integrity is diminished by 8% of the intact pipeline strength.                                       iv Preface The following journal articles, book chapter, and articles in the proceedings of professional conferences have been published from the research work presented in this dissertation. Professor Akram Alfantazi extensively helped with all aspects of the research work.  Peer-reviewed Journals and Book Chapter: 1. I. M. Gadala, M. Abdel Wahab, and A. Alfantazi, “Simulating the burst pressure of externally corroded underground gas transmission pipelines using a finite element corrosion model and stress analysis,” [under review]. 2. I. M. Gadala, H. M. Ha, P. Rostron, and A. Alfantazi, “Formation and evolution of oxide/oxyhydroxide corrosion products on low-alloy steel during exposure to near-neutral pH solutions containing oxygen and nitrate,” Corrosion, vol. 73, no. 3, pp. 221-237, March 2017. 3. H. M. Ha, I. M. Gadala, and A. Alfantazi, “Hydrogen Evolution and Absorption in an API X100 Line Pipe Steel Exposed to Near-Neutral pH Solutions,” Electrochimica Acta, vol. 204, pp. 18-30, June 2016. 4. F. F. Eliyan, I. M. Gadala, H. M. Ha, and A. Alfantazi, “Pipeline Corrosion,” ASTM Fuels and Lubricants Handbook: Technology, Properties, Performance, and Testing, 2nd edition. PA: ASTM International, 2016. 5. I. M. Gadala, M. Abdel Wahab, and A. Alfantazi, “Numerical simulations of soil physicochemistry and aeration influences on the external corrosion and cathodic protection design of buried pipeline steels,” Materials and Design, vol. 97, pp. 287-299, May 2016.  6. I. M. Gadala and A. Alfantazi, “A study of X100 pipeline steel passivation in mildly alkaline bicarbonate solutions using Electrochemical Impedance Spectroscopy under potentiodynamic conditions and Mott-Schottky,” Applied Surface Science, vol. 357A, pp. 356-368, Dec 2015.  7. I. M. Gadala and A. Alfantazi, “Low Alloy X100 Pipeline Steel Corrosion and Passivation Behavior in Bicarbonate-Based Solutions of pH 6.7 to 8.9 with Groundwater Anions: An Electrochemical Study,” Metallurgical and Materials Transactions A, vol. 46, no. 7, pp. 3104–3116, April 2015.  8. I. M. Gadala and A. Alfantazi, “Electrochemical behavior of API-X100 pipeline steel in NS4, near-neutral, and mildly alkaline pH simulated soil solutions,” Corrosion Science, vol. 82, pp. 45–57, May 2014.  v Peer-reviewed Conference Proceedings: 9. I. M. Gadala and A. Alfantazi, “Inhibitive effectiveness of hydrazine oxygen scavenger on low-alloy steel in near-neutral pH underground conditions,” Proceedings of the ACA Corrosion and Prevention Conference, November 13 – 16 (2016), Auckland, New Zealand. 10. I. M. Gadala, M. Abdel Wahab, and A. Alfantazi, “A finite element model of the external corrosion of buried pipeline steel under the combined influence of heat transfer, cathodic protection, and oxygen diffusion in surrounding soil,” Proceedings of NACE CORROSION 2016, March 6 – 11 (2016), Vancouver, Canada.  11. I. M. Gadala, A. Alfantazi, Z. Farhat, and A. M. A. Mohamed, “External corrosion of API-X100 pipeline steels in near-neutral pH soils of variable aeration and nitrate ion content,” Proceedings of the 10th ASME RioPipeline, September 22 – 24 (2015), Rio de Janeiro, Brazil. 12. I. M. Gadala and A. Alfantazi, “An Electrochemical Impedance Spectroscopy study of the corrosion of buried low-alloy steel infrastructure used in energy transportation and storage,” Proceedings of the Energy & Materials Research Conference (EMR), February 25 – 27 (2015), Madrid, Spain. 13. I. M. Gadala and A. Alfantazi, “Bicarbonate, temperature, and pH influences on the passivation of API-X100 pipeline steel in simulated groundwater solutions,” Proceedings of the 10th ASME International Pipeline Conference (IPC), September 29 – October 3 (2014), Calgary, Canada.  The following table identifies the specific publication(s) upon which each corresponding chapter of this dissertation is based: Chapter number in dissertation Publication number(s) from the list above 2 4 5 7, 8, and 13 6 6 and 12 7 2, 9, and 11 8 3 9 1, 5, and 10   vi Table of Contents Abstract ............................................................................................................................... ii Preface................................................................................................................................ iv Table of Contents ............................................................................................................... vi List of Tables ..................................................................................................................... ix List of Figures ................................................................................................................... xii List of Abbreviations ....................................................................................................... xxi List of Chemical Formulae ............................................................................................ xxiii List of Symbols ............................................................................................................... xxv Acknowledgements ......................................................................................................... xxx Dedication ...................................................................................................................... xxxi 1 Introduction ............................................................................................................... 1 1.1 Impact of pipeline corrosion ............................................................................... 2 1.2 Importance for Canada ........................................................................................ 3 1.3 High-Strength Low-Alloy steels ......................................................................... 3 1.4 Environments causing external SCC .................................................................. 5 1.5 Motivation ........................................................................................................... 8 2 Literature review ...................................................................................................... 9 2.1 Ionic and pH influences on corrosion processes in anoxic soil environments . 10 2.2 Time-dependent corrosion product and surface effects in anoxic and oxic soil environments ................................................................................................................. 15 2.3 Hydrogen evolution, absorption, and diffusion processes in steels exposed to simulated soil solutions ................................................................................................. 21 2.4 Numerical modelling of corrosion, CP, and pipeline structural integrity ......... 27  Models and simulations of steel corrosion with CP and gas transport ....... 27  Models of the residual strength and integrity of corroded pipelines .......... 32 3 Objectives................................................................................................................. 36 3.1 Key technical objectives ................................................................................... 37 4 Approach and methodology ................................................................................... 38 4.1 Material and specimen preparation ................................................................... 38  Microstructural evaluation .......................................................................... 39  Preparation of specimens ............................................................................ 41 4.2 Test environments and methods........................................................................ 42  Standard NS4 solution and variants ............................................................ 42  Electrochemical test methods for corrosion and passivation processes...... 43  Electrochemical test method for hydrogen permeation and diffusion ........ 46  Microscopy and chemical characterization methods .................................. 49  vii 4.3 Modelling geometries and meshes .................................................................... 50 5 Electrochemical behavior of X100 pipeline steel in deaerated 𝐇𝐂𝐎𝟑− solutions of near-neutral, mildly acidic, or mildly alkaline pH, ...................................................... 54 5.1 OCP and LPR measurements ............................................................................ 55 5.2 PDP testing and parameter variations based on E-pH ...................................... 58  Passive and passive-like PDP responses ..................................................... 59  Anodic parameter relationships with E-pH................................................. 64 5.3 EIS tests at OCP in deaerated nn-pH conditions .............................................. 74  Subdivision of electrochemical response based on pH range ..................... 83 5.4 Summary ........................................................................................................... 83 6 Quantitative studies of the properties & growth of corrosion products on X100 steel in mildly alkaline deaerated 𝐇𝐂𝐎𝟑− solutions using EIS and Mott-Schottky .... 85 6.1 Details of dynamic EIS, PSP, and Mott-Schottky test methods ....................... 87 6.2 Dynamic step-wise anodizing EIS test ............................................................. 89  Active corrosion at E < Ep1 ......................................................................... 89  Corrosion and passive-like behavior in the Ep1 < E < Ep2 transition ........... 94  Corrosion and passivation processes at E > Ep2 .......................................... 98 6.3 Protective and semiconductive properties of anodized FeCO3 ....................... 102  Current density decay and qualitative analysis of Bode |𝑍| results .......... 102  Semiconductive properties analysis using Mott-Schottky ........................ 105 6.4 Summary ......................................................................................................... 108 7 Extended immersion studies of the formation of oxide/oxyhydroxide corrosion products on X100 steel during exposure to nn-pH 𝐇𝐂𝐎𝟑− solutions with O2, N2H4, and/or 𝐍𝐎𝟑− .................................................................................................................... 110 7.1 Test environments and experimental procedures ............................................ 112 7.2 Short-term electrochemical tests at immersion times ≤ 1 h ............................ 115  OCP < 1 h ................................................................................................. 115  PDP at 1 h ................................................................................................. 119 7.3 Surface analysis following 24 h or 168 h immersions .................................... 126 7.4 Periodic electrochemical tests during 24 h immersions in 0 – 20 ppm [O2] and 0 – 0.015 M [NO3−] conditions ....................................................................................... 134  OCP ........................................................................................................... 134  EIS............................................................................................................. 138  LPR ........................................................................................................... 146 7.5 Summary ......................................................................................................... 149 8 Evaluation of hydrogen evolution, absorption, and diffusion in X100 steel exposed to nn-pH 𝐇𝐂𝐎𝟑− solutions of various ion constituents and temperatures .. 151 8.1 Environments and hydrogen permeation conditions....................................... 152  Hydrogen permeation conditions .............................................................. 154 8.2 Results of OCP and cathodic polarization (E < -1.2 VSCE ) ............................ 155  Environmental effects on cathodic polarization results ............................ 156  Surface deposit effects on cathodic polarization results ........................... 160 8.3 Results of hydrogen permeation experiments ................................................. 161  viii 8.4 Discussion of OCP, cathodic PDP, and permeation results ............................ 165  Hydration of CO2 and [HCO3−-CO32−] in CO2-H2O system ....................... 165  HER on X100 at OCP and cathodic potentials ......................................... 170  Hydrogen permeation in X100 exposed to nn-pH HCO3− solutions .......... 174 8.5 Summary ......................................................................................................... 176 9 FEM of the external corrosion and structural integrity of buried pipelines under the influences of CP, gas diffusion, and environment physicochemistry ..... 178 9.1 Modelling details: governing phenomena and equations................................ 181  Heat transfer, CP, and O2 diffusion phenomena ....................................... 181  Initial conditions and boundary conditions for governing phenomena .... 183  Reaction kinetics at exposed steel surface ................................................ 185 9.2 Polarization experiments and equations for coupled simulations ................... 189  Temperature- and O2-dependent fittings for m-model .............................. 189  Temperature- and %CO2-dependent fittings for mm-model .................... 194 9.3 𝐸𝐹𝑒, 𝐶𝑂2, and 𝑖 simulation results from m-model ........................................... 196  Temperature, φ, and 𝐶𝑂2 contours ............................................................ 196  𝐸𝐹𝑒 ............................................................................................................. 199  𝑖𝐹𝑒 .............................................................................................................. 201  𝐶𝑂2, 𝑖𝑂2, and 𝑖𝐻2 ........................................................................................ 203  Practical CP design applications ............................................................... 206 9.4 𝜑 evolution, defect growth, and stress analysis in mm-model ....................... 209  𝜑 evolution within the disbondment ......................................................... 209  Growth of corrosion defect and implications on structural integrity ........ 210 9.5 Model convergence through mesh sensitivity analysis ................................... 217 9.6 Summary ......................................................................................................... 219 10 Conclusions ............................................................................................................ 222 10.1 Key contributions and broad implications ...................................................... 222 10.2 Summary of important technical findings....................................................... 226 10.3 Suggestions for future work ............................................................................ 231 References ...................................................................................................................... 233 Appendices ..................................................................................................................... 253 Appendix A: supplementary figures for chapter 5 ...................................................... 253 Appendix B: supplementary figures for chapter 6 ...................................................... 260 Appendix C: supplementary figures and tables for chapter 7 ..................................... 262 Appendix D: supplementary tables and figures for chapter 8 ..................................... 269 Appendix E: supplementary tables and figures for chapter 9 ..................................... 272  ix List of Tables Table 1-1: Natural gas transmission pipeline incident summary by cause for 1/1/2002 - 12/31/2003, from US Department of Transportation’s Office of Pipeline Safety [4] ..................... 2 Table 1-2: Characteristics of nn-pH and high pH external SCC in pipelines – adapted from [13]–[15] .................................................................................................................................................. 6 Table 4-1: Chemical composition and carbon equivalent (CE) of API X100 steel used in laboratory tests .............................................................................................................................. 39 Table 4-2: Maximum and minimum element size, maximum element growth rate, and resolution of curvature values for extra coarse to extra fine mesh resolutions, within the soil domain vs. at electrode/ground boundaries in Figure 4-4b .................................................................................. 53 Table 5-1: EIS component values for NS4 solution at 25, 40, or 55 °C purged with 5% CO2/95% N2, or 100% CO2 ........................................................................................................................... 77 Table 5-2: EIS component values for NS4 + 10x HCO3− solution ................................................ 80 Table 5-3: EIS component values for NS4 + 10x HCO3− solutions with added Cl− and/or SO42− at 25 °C, purged with 5% CO2/95% N2 (pH 7.6) .............................................................................. 82 Table 6-1: Anodizing potentials (Ean) in each potential range for step-wise anodizing-EIS routine ....................................................................................................................................................... 87 Table 6-2: EIS component values for 0.1 M [HCO3−] solution with and without Cl−/SO42− at Ean = -0.65 VSCE (early active E region) ................................................................................................. 91 Table 6-3: EIS component values for 0.1 M [HCO3−] solution with and without Cl−/SO42− at -0.5 VSCE (late active potential region), and 0.1 M [HCO3−] solution without Cl−/SO42− at -0.4 VSCE (transition potential region) ........................................................................................................... 93 Table 6-4: EIS component values for 0.1 M [HCO3−] solution with Cl−/SO42− at -0.475 VSCE (transition potential region) ........................................................................................................... 97 Table 6-5: EIS component values for 0.1 M [HCO3−] solution with Cl−/SO42− at -0.1 VSCE (passive formation potential region) and 0.175 VSCE (passive potential region), and without Cl−/SO42− at -0.25 VSCE (passive formation potential region) and 0.5 VSCE (passive potential region) ......................................................................................................................................... 100 Table 6-6: EIS component values for 0.1 M [HCO3−] solution without Cl−/SO42− at 1.1 VSCE (transpassivation potential region) .............................................................................................. 101 Table 7-1: List of [O2] and [NO3−] combinations studied in this chapter, based on the reference NS4 electrolyte of AA*, with corresponding measured CO2_i, pH, and conductivities of solutions ..................................................................................................................................................... 113  x Table 7-2: List of temperature and [N2H4] combinations studied in short term corrosion inhibition tests of this chapter based on reference NS4 electrolyte of environment AA*, with corresponding measured CO2_i ............................................................................................................................. 114 Table 7-3: List of temperature and [N2H4] combinations studied in extended 168 h immersion corrosion inhibition tests of this chapter based on environment AA* in Table 7-2, with corresponding measured CO2_i ..................................................................................................... 114 Table 7-4: Values of 𝐸𝑐𝑜1 and 𝐸𝑐𝑜2 extracted from PDP plots of N2H4 treated environments, with corresponding ranges of the low-current region .......................................................................... 124 Table 7-5: Values of 𝑖𝑎1 and 𝑖𝑎2 extracted from PDP plots of N2H4 treated environments, with corresponding 𝑖𝑚𝑎𝑥 values in low i region and minimum percent decrease in i ......................... 125 Table 7-6: Comparison of Rp values from LPR with Rp* or Rp** from EIS component sums, in solutions with 0, 6, and 20 ppm O2 at 2, 8, 16, and 24 h immersion ........................................... 147 Table 8-1: Approximate concentrations and % fractions of H2CO3, HCO3−, and CO32− in the CO2-H2O system for a temperature of 20 oC and 1 atm CO2 ............................................................... 168 Table 8-2: Summary of steady state permeation current densities and diffusible hydrogen concentrations in the X100 steel exposed to the NS4 solution under different charging conditions ..................................................................................................................................................... 175 Table 9-1a: Governing transport equations and related parameters in each module of the m-model [290] ................................................................................................................................. 181 Table 9-1b: Values or expressions for the thermal, electrical, and diffusion properties of the soil media in the m-model [290] .......................................................................................................1812 Table 9-2: Overall formulations and reaction-specific equations governing reaction kinetics in the uncoupled m-model ............................................................................................................... 187 Table 9-3: Regression fit equations and corrosion parameters governing reaction kinetics in the coupled simulations of the m-model, extracted from polarization plots ..................................... 192 Table 9-4: List of OCP values for environments studied for mm-model, characterized by temperature, %CO2 concentration in deaerating purging gas, and pH ........................................ 194 Table 9-5: C1 and C2 parameter results of regression fits for ic evaluation (ic as a function of 𝐸𝑠(𝑥), 𝐶1, and 𝐶2) ........................................................................................................................ 194 Table 9-6: Comparison of Pb/Pbi ratio results obtained from {E-9.19} versus those from the shell element FEM of Figure 9-11, using corrosion defect values simulated from the corrosion module of the mm-model ......................................................................................................................... 217 Table C-1: Corrosion parameters obtained from Tafel extrapolation of deaerated PDP results of three NO3− concentrations ............................................................................................................ 263  xi Table C-2: Fitting results for linear and nonlinear regressions of final OCP values at 1 h and 24 h (Figure 7-1), valid for [O2] values between 0 and 20 ppm .......................................................... 266 Table C-3: EIS component values for deaerated 0.2 ppm O2 solution (pH 6.7, 0.005 M NO3−) at 2, 8, 16, and 24 h immersion ........................................................................................................... 267 Table C-4: EIS component values for 5.8 ppm O2 solution (pH 7.7, 0.005 M NO3−) at 2, 8, 16, and 24 h immersion ..................................................................................................................... 267 Table C-5: EIS component values for 20.4 ppm O2 solution (pH 8.6, 0.005 M NO3−) at 2, 8, 16, and 24 h immersion ..................................................................................................................... 268 Table D-1: Test conditions and the measured natural pH of the test solutions in chapter 8 ...... 269 Table D-2: Some forms of Henry’s law and values of Henry’s constant for various gases in water at 298 K [277], [278] ................................................................................................................... 271 Table E-1 (for mm-model): Maximum element size, minimum element size, and maximum element growth rate for coarse to fine mesh resolutions, within the electrolyte domain vs. at the exposed steel surface in Figure 4-5 ............................................................................................. 272 Table E-2 (for m-model): Thermal conductivity and volumetric heat capacity parameters for the soil structure and ψ combinations simulated ............................................................................... 272 Table E-3 (for mm-model): List of current, potential, current balance, reaction kinetics (extracted from polarization plots), and regression fit equations governing corrosion simulations, with corresponding numbers, details, and conditions .................................................................. 273 Table E-4 (for mm-model): Maximum radial depth (ddefect_max) and longitudinal length (ldefect) dimensions [mm] of corrosion defects at 1 and 3 year time stages for all temperatures (Twall), applied potentials (Eapp), and coating disbondment opening sizes (w) simulated ....................... 276 Table E-5: Mechanical properties of various grades of HSLA pipeline steels [152] ................. 277             xii List of Figures Figure 1-1: North American network of major liquid transmission pipelines [1] .......................... 1 Figure 1-2: Development flow chart of HSLA pipeline steels from 1965 to 2009 [8] .................. 4 Figure 1-3: Three conditions necessary for pipeline SCC, with corresponding dependencies [13]  ......................................................................................................................................................... 6 Figure 1-4: Areas of near-neutral pH SCC formation on external pipeline surfaces [13] .............. 7 Figure 2-1: The numerous environmental, material, and surface effects on metal corrosion in soil environments, ranging from the macroscale to the smaller surface level scale [34] ....................... 9 Figure 2-2: Schematic showing the effect of CO2 concentration on the environment’s pH, and the resulting influence on aggressiveness (where te/tair is the time to failure ratio in environment over that in air) [24] ............................................................................................................................... 11 Figure 2-3: Potentiodynamic polarization in 0.1, 0.5, and 0.8 M HCO3− solutions at 20 ºC in (a) Cl−-free and (b) Cl−-containing conditions [56] ........................................................................... 12 Figure 2-4: (a) Actual CP levels at different positions from opening mouth (OM) and different times; (b) pH distribution of the electrolyte inside the shielded disbondments at different positions from opening mouth, and different times [25] .............................................................................. 13 Figure 2-5: (a) Surface morphology of CaCO3 layer formed on X70 surface after 40 h immersion in 5% CO2 NS4 solution at 22 °C; (b) icorr map measured on X70 specimen covered with CaCO3 [41] ................................................................................................................................................ 16 Figure 2-6: (a) Ecorr (vs. SCE) of pretreated steel under anoxic conditions in nn-pH saline solutions, where square points show Rp values from LPR; (b) Schematic illustrating the film transition process within an acidified pore, assuming seperation of anode (Fe dissolution) and cathode (HCO3− discharge) [68] ..................................................................................................... 17 Figure 2-7: Variation of embrittlement index (using % reduction in area) with potential for X100 steel [90] ........................................................................................................................................ 21 Figure 2-8: (a) Hydrogen permeation curves of X65 specimens (600-grit finish) at -1200 mV with and without calcium carbonate coatings (b) Hydrogen permeation current vs. applied cathodic potential [93] ................................................................................................................... 23 Figure 2-9: SEM morphologies of the cross-section of X100 specimen and the chemical composition obtained at the individual inclusions (a) an Al-enriched inclusion; (b) a Si-enriched inclusion [108] ............................................................................................................................... 25 Figure 2-10: Steady-state H amount released from an X100 steel specimen as a function of the charging current density, where the 𝐶𝐻∗  needed to initiate HIC is identified [108] ....................... 26  xiii Figure 2-11: (a) Planar cross-section of buried tank and anode system modelled in [125], with meshing of the soil media; (b) Potential distribution at surface of tank for different anode positions (circle: bottom of pit, square: at half depth, triangle: critical potential) ........................ 28 Figure 2-12: (a) Schematic view of a cross-section of the internal CP system investigated in [130], where ra is the wire anode radius, rc is the hollow cylinder (cathode) radius, and d is the offset distance from the central axis (b) Steady-state potential measured at 180° in (a) as a function of d in electrolytes of different conductivities (triangle: 1% NaCl, square: 0.6% NaCl, diamond: 0.3% NaCl, and star: 0.15% NaCl)................................................................................ 30 Figure 2-13: Comparisons between experiments, the linear fit (i.e. {E-2.7}), and predictions from DNV and B31G codes [146] ......................................................................................................... 33 Figure 2-14: (a) Experimental and FEM CIC defect profiles; (b) 3D simulation results for 60% wall thickness CIC defect model, showing spatially-dependent 𝜎𝑉𝑀 from a 5.59 MPa internal pressure [53] .................................................................................................................................. 35 Figure 4-1: Microstructure of API X100 steel sample etched with: (a) 2% nital, with corresponding differentiation of ferrite and bainite phases according to ASTM E562-08; and (b) LePera solution; (c) graphical representation of M-A phase identification using ImageJ analysis software ......................................................................................................................................... 40 Figure 4-2: Schematic of experimental setup for electrochemical tests of corrosion and passivation [162] ........................................................................................................................... 44 Figure 4-3: Electrochemical H permeation and diffusion cell: (a) components and assembly (b) measuring apparatus and settings (with two potentiostat instruments or two channels of one multistat) [154] .............................................................................................................................. 48 Figure 4-4: (a) 3D representation of buried pipeline with CP anodes (right), and axis of model cross-section (left); (b) 2D representation of model geometry and dimensions ............................ 51 Figure 4-5: 2D cross-section of pipeline in longitudinal plane with coating defect (zoomed section), dimensions, and mesh of trapped water region ............................................................... 53 Figure 5-1: Effect of HCO3−, purging environment, and temperature on solution pH .................. 55 Figure 5-2: Open circuit potentials vs. pH in all solutions free of Cl− and/or SO42− additions (varying HCO3−, purging gas, and temperature) ............................................................................. 56 Figure 5-3: Corrosion activity vs. pH based on polarization resistance values of select data ...... 58 Figure 5-4: PDP in 5% CO2/95% N2 purged NS4 + 10x HCO3− at 25, 40, and 55 °C (pH 7.6, 7.9, and 8.1, respectively) ..................................................................................................................... 60 Figure 5-5: PDP profiles in 100% CO2 purged NS4 + 10x HCO3−  at 25, 40, and 55 °C (pH 6.4, 6.6, and 6.7, respectively) .............................................................................................................. 62  xiv Figure 5-6: PDP in 5% CO2/95% N2 purged NS4 + 10x HCO3− at 25 °C with various Cl− and/or SO42− ion content (pH 7.6 for all) .................................................................................................. 63 Figure 5-7: 1 mV/s PDP profiles of specimen in NS4 Cl−- and SO42−-containing solutions with 7.19 < pH < 8.85 ............................................................................................................................ 65 Figure 5-8: Current density peaks Ep1 and Ep2 plotted vs. pH for Cl−- and SO42−-containing solutions at 1 mV/s and 0.5 mV/s, and Cl−- and SO42−-free solutions at 1 mV/s .......................... 66 Figure 5-9: Passive current density ip plotted vs. pH for Cl−- and SO42−-containing solutions at 1 mV/s and 0.5 mV/s, and Cl−- and SO42−-free solutions at 1 mV/s ................................................. 69 Figure 5-10: Potential difference ∆E plotted vs. pH for Cl−- and SO42−-containing solutions at 1 mV/s and 0.5 mV/s, and Cl−- and SO42−-free solutions at 1 mV/s ................................................. 70 Figure 5-11: Breakdown or transpassivation potentials Ebd1 and Ebd2 plotted vs. pH for Cl−- and SO42−-containing solutions at 1 mV/s and 0.5 mV/s, and Cl−- and SO42−-free solutions at 1 mV/s …………………………………………………………………………………………………….71 Figure 5-12: (a) SEM image (x150 magnification) of FeCO3-Fe3O4/γ-Fe2O3 discrete boundary: (zoom) x700 magnification of Fe3O4 growth over FeCO3 crystals; (b) XRD pattern of a specimen removed from a PDP scan in pH 8.36 solution (top profile), and control sample with no corrosion product (bottom profile) ................................................................................................................ 73 Figure 5-13: EIS for NS4 solution at 25, 40, and 55 °C purged with 5% CO2/95% N2 (pH 6.7, 6.8, and 7.0, respectively) or 100% CO2 (pH 5.4, 5.5, and 5.6, respectively): (a) Nyquist impedance representation, (b) proposed EEC for 100% CO2 results ............................................ 76 Figure 5-14: Proposed EEC for 5% CO2/95% N2 purged NS4 solution at 25, 40, and 55 °C (pH 6.7, 6.8, and 7.0, respectively) and 100% CO2 purged NS4 + 10x HCO3− solution (pH 6.4, 6.6, and 6.7, respectively) ........................................................................................................................... 77 Figure 5-15: NS4 + 10x HCO3− solution at 25 °C purged with 100% N2 (pH 8.9), and at 25, 40, and 55 °C purged with 100% CO2 (pH 6.4, 6.6, and 6.7, respectively), and 5% CO2/95% N2 (pH 7.6, 7.9, and 8.1, respectively) (a) Nyquist impedance representation for all cases, (b) enlarged Nyquist impedance representation for 5% CO2/95% N2 and 100% CO2 purged cases ................. 79 Figure 5-16: Proposed EEC for 5% CO2/95% N2 purged NS4 + 10x HCO3− at 25, 40, and 55 °C (pH 7.6, 7.9, and 8.1, respectively), and 100% N2 purged NS4 + 10x HCO3− solution at 25°C (pH 8.9) ................................................................................................................................................. 80 Figure 5-17: Nyquist impedance representation plots for NS4 + 10x HCO3− solutions with added Cl− and/or SO42− at 25 °C, purged with 5% CO2/95% N2 (pH 7.6) ............................................... 81  xv Figure 6-1: Comparative PDP plot for 0.1 M [HCO3−] solution with and without NS4 Cl−/SO42− scanned at 0.5 mV/s or 1 mV/s, labelled for potential regions of step-wise anodizing-EIS routine ....................................................................................................................................................... 88 Figure 6-2: Schematic of the two independent electrochemical test routines conducted in this chapter ........................................................................................................................................... 89 Figure 6-3: Nyquist impedance representation in active corrosion region (OCP < E ≤ Ep1) ....... 90 Figure 6-4: (a) Nyquist impedance representation in transition and passive layer formation region (Ep1 < E ≤ 0 VSCE); (b) Proposed physical occurrences for EIS at -0.4 VSCE in solution without NS4 Cl−/SO42− ............................................................................................................................... 95 Figure 6-5: (a) Nyquist impedance representation in passive and transpassive region (E > 0 VSCE); (b) Proposed corresponding physical occurrences for EIS at -0.25 VSCE, -0.1 VSCE, 0.5 VSCE and 0.175 VSCE ............................................................................................................................... 99 Figure 6-6: (a) Proposed EEC and corresponding physical occurrences for EIS at -0.25 VSCE, -0.1 VSCE, 0.5 VSCE and 0.175 VSCE; (b) Proposed EEC for EIS at 1.1 VSCE in solution without Cl−/SO42− ............................................................................................................................................ 100 Figure 6-7: Current density decay profiles of specimen anodized at 0.5 VSCE in 0.1, 0.25, or 0.5 M NaHCO3 at 25, 50, or 75 ºC (pH 7.8 – 9.3) ............................................................................. 103 Figure 6-8: Bode |Z| profiles of specimen after 1 h PSP in at 0.5 VSCE in 0.1, 0.25, or 0.5 M NaHCO3 at 25, 50, or 75 ºC (pH 7.8 – 9.3) ................................................................................. 104 Figure 6-9: Dependence of Nd1, Nd2, and Efb of passive layer on [NaHCO3] (0.1, 0.25, or 0.5 M) and temperature (25, 50, or 75 ºC) .............................................................................................. 106 Figure 6-10: Dependence of (𝛿𝑠𝑐) with respect to E, [NaHCO3] (0.1, 0.25, or 0.5 M), pH, and temperature (25, 50, or 75 ºC) ..................................................................................................... 108 Figure 7-1: Final OCPs reached after 1 h and 24 h immersion (separate tests) as a function of [O2], with nonlinear fit profiles overlaid ..................................................................................... 115 Figure 7-2: Influence of N2H4 treatment concentration on 1 h OCP values at 25 and 50 °C ..... 118 Figure 7-3: PDP in 0 ppm, 6 ppm, and 20 ppm O2 solutions with 0.015 M added NO3− ........... 120 Figure 7-4: PDP in (a) 6 ppm O2 solutions with 0, 0.005, and 0.015 M added NO3−, and (b) 20 ppm O2 solutions with 0, 0.005, and 0.015 M added NO3− .......................................................... 121 Figure 7-5: PDP profiles of aerated environments at 25 and 50 C, treated with N2H4 ............. 124 Figure 7-6: SEM images of steel specimen surface following immersion in solution containing 6.0 ppm O2 (pH 7.8) and 0.015 M NO3−: (left) low magnification; (top-right) as imaged high magnification with 50 µm scale bar; and (bottom-right) high magnification with 50 µm scale bar analyzed by color threshold ......................................................................................................... 128  xvi Figure 7-7: SEM images of steel specimen surface following immersion in 21.2 ppm O2 (pH 8.9) and 0.015 M NO3−: (left) low magnification; (top-right) as imaged high magnification with 50 µm scale bar; and (bottom-right) high magnification with 50 µm scale bar analyzed by color threshold ...................................................................................................................................... 129 Figure 7-8: Ex-situ XPS spectra of specimen immersed in 6 ppm and 20 ppm O2 solutions free of NO3− ............................................................................................................................................. 130 Figure 7-9: Ex-situ Raman spectra of specimen immersed in 6 ppm and 20 ppm O2 solutions free of NO3− ......................................................................................................................................... 131 Figure 7-10: SEM images and corresponding ex-situ XRD spectra of X100 specimen surface following 168 h immersion in environments periodically treated with N2H4 as described in Table 7-3: (a), (b) 25 °C; (c), (d) 50 °C ....................................................................................... 133 Figure 7-11: Full spectrum OCP transients Table 7-1 conditions during 24 h (same legend as Figure C-1 ) .................................................................................................................................. 135 Figure 7-12: Nyquist impedance representation and fit profiles for deaerated solution containing 0.005 M NO3− (0.2 ppm O2, pH 6.6): (a) at 2, 4, 6, 8, 10, and 12 h immersion times, with proposed EEC; (b) at 14, 16, 18, 20, 22, and 24 h immersion times........................................................... 139 Figure 7-13: Nyquist impedance representation and fit profiles for 5.8 ppm O2 solution (pH 7.7) containing 0.005 M NO3−: (a) at 2, 4, 6, 8, 10, and 12 h immersion times, with proposed EEC; (b) at 14, 16, 18, 20, 22, and 24 h immersion times .......................................................................... 140 Figure 7-14: Nyquist impedance representation and fit profiles for 5.8 ppm O2 solution (pH 7.7) containing 0.005 M NO3−: (a) at 2, 4, 6, 8, 10, and 12 h immersion times, with proposed EEC; (b) at 14, 16, 18, 20, 22, and 24 h immersion times .......................................................................... 144 Figure 7-15: Diagram illustrating the influence of ∆[O2] between: (1) 6 ppm O2 and (2) 20 ppm O2 conditions on the evolution mechanism of the Fe-oxyhydroxide tubercle and sub-tubercle Fe-oxide layer(s), where 𝑑?̃? represents the average pore size in the tubercle .................................. 145 Figure 7-16: Polarization resistance (Rp) vs. immersion time, calculated from periodic LPR scans conducted in: (a) deaerated and 6 ppm O2 solutions containing 0, 0.005, or 0.015 M NO3−; (b) 20 ppm O2 solutions containing 0, 0.005, or 0.015 M NO3− ............................................................. 148 Figure 8-1: Open circuit potential of X100 steel in different test solutions of Table D-1 at 20 °C ..................................................................................................................................................... 155 Figure 8-2: Cathodic behavior of X100 steel in: (a) deaerated (100% N2) NS4 and modified NS4 solutions at 20 °C; (b) open air NS4 and modified NS4 solutions at 20 °C ................................ 157  xvii Figure 8-3: Cathodic behavior of X100 steel in: (a) NS4 solution under different gas atmospheres; (b) NS4 solution under different gas atmospheres during constant potential holds at -0.8, -0.9 and -1.0 VSCE ................................................................................................................ 158 Figure 8-4: Cathodic polarization curves of X100 steel in: (a) 100% CO2 NS4 solution at 20, 40, and 60 °C, natural pHs; (b) NS4 solution with adjusted pH at 20 oC .......................................... 159 Figure 8-5: Characterization of the deposit on X100 steel held at -1.2 VSCE in 100% CO2 NS4 solution: (a) SEM photo after 1 h; (b) EDX/EDS spectrum of a particle formed ....................... 161 Figure 8-6: Permeation results for X100 steel exposed to 100% CO2 NS4 solution (20 C) at a hydrogen charging current density of -500 µA/cm2: (a) permeation current transient; (b) determination of Deff of the X100 steel from permeation current transient using {E-8.1}, where the slope of the log((iperm-ibg)×t0.5) vs. 1/t  plot is obtained by linear fitting ................................ 162 Figure 8-7: The current response in the H entry and exit sides (cathodic icharge and anodic iperm, respectively) during the permeation experiment in NS4 solution. The purging gas in the charging compartment is switched from 100% N2 to 5% CO2 to 100% CO2............................................. 164 Figure 8-8: The current response at the H entry and exit sides during the permeation experiment in 100% CO2 NS4 solution at 20 °C. The surface potential of the steel sample at the entry side is first switched from -0.85 to -1.15 VSCE with 100 mV/step/4 h then is left at unpolarized (i.e. OCP) before the solution in the entry compartment is drained out to leave the sample open to air ..... 165 Figure 8-9: Speciation diagram of the CO2-H2O system calculated at 25 oC using Medusa software ....................................................................................................................................... 167 Figure 8-10: Concentration of dissolved CO2 and carbonate species in solution with respect to pH, with pCO2 = 1 atm at: (a) 20 °C; (b) 40 °C; and (c) 60 °C ................................................... 169 Figure 8-11: OCP of X100 steel in NS4 and modified NS4 solutions under different purging gas atmospheres at 20 oC ................................................................................................................... 171 Figure 8-12: Cathodic polarization curves of scale-covered X100 steel in NS4 solution at 20 oC ..................................................................................................................................................... 174 Figure 9-1: Pipeline wall and initial (ambient) soil temperature fluctuations with respect to soil structure, depth, and time in m-model ......................................................................................... 184 Figure 9-2: (a) PDP profiles of X100 steel sample in deaerated and aerated NS4 solutions at 30 and 50 °C; (b) cathodic regime of PDP profiles (with regression fits) showing E- and temperature-dependent O2 reduction and H evolution kinetics; and (c) LPR fitting results for X100 specimen immersed in NS4 with combinations of deaeration, aeration, and temperature 191  xviii Figure 9-3: Contours of: (a) induced soil temperature [K (°C)] for peat with ψ = 0.60 at hottest conditions; (b) φ [VSCE] for sand with ψ = 0.60 at hottest induced temperatures; and (c) CO2 [mol/m3] for clay with ψ = 0.20 at hottest induced temperatures ................................................ 198 Figure 9-4: EFe evaluated at θ = 0° and θ = 90° in minimum and maximum induced temperature profiles, different soil structures, and ψ using: (a) uncoupled governing equations; and (b) coupled governing equations ....................................................................................................... 200 Figure 9-5: iFe evaluated at θ = 0° and θ = 90° in minimum and maximum induced temperature profiles, different soil structures, and ψ using: (a) uncoupled governing equations; and (b) coupled governing equations ....................................................................................................... 202 Figure 9-6: (a) iO2 and (b) iH2, evaluated at θ = 0° and θ = 90° in minimum and maximum induced temperature profiles, different soil structures, and ψ using coupled governing equations ..................................................................................................................................................... 205 Figure 9-7: Influence of CP design parameters on the model results for simulations at Twall = 323 K, θ = 0°, and ψ = 0.2 or 0.3: (a) EFe and iFe as a function of CP anode distance from exposed pipeline surface; and (b) iO2 and iH2 as a function of Eapp ............................................................ 208 Figure 9-8: Timeline of corrosion defect growth into the exposed steel surface and φ evolution throughout the 150 week exposure, for Eapp = -1 VSCE, 1 mm coating disbondment opening, and Twall = 25 °C ................................................................................................................................. 209 Figure 9-9: Length of corrosion defect (ldefect) [mm] along exposed steel surface and x-position of ddefect_max (maximum defect depth) [mm] reached during 0 – 150 weeks of exposure, for all Eapp, coating disbondment opening, and Twall values simulated: (a) Twall = 25 ˚C, (b) Twall = 50 ˚C .... 211 Figure 9-10: Maximum depth [mm] of corrosion defect (ddefect_max) below original steel surface from 0 – 150 weeks of exposure, for all Eapp, coating disbondment opening, and Twall values simulated: (a) Twall = 25 ˚C, (b) Twall = 50 ˚C ............................................................................... 213 Figure 9-11: 3D representation of: (a) intact pipe FEM with shell elements; (b) corroded pipe section used to simulate structural integrity based on the defect dimensions resulting from the corrosion module ......................................................................................................................... 215 Figure 9-12: Convergence profiles for iO2 and iH2 results (at Twall =323 K, θ = 0°, and ψ = 0.2 or 0.3) using extra coarse to extra fine mesh resolutions in m-model ............................................. 218 Figure A-1: PDP in 5% CO2/95% N2 purged NS4 at 25, 40, and 55 °C (pH 6.7, 6.8, and 7.0, respectively) ................................................................................................................................ 253 Figure A-2: 100% CO2 purged NS4 at 25, 40, and 55 °C (pH 5.4, 5.5, and 5.6, respectively) .. 253 Figure A-3: NS4 at 25 °C with 5% CO2/95% N2, 100% CO2, and 100% N2 purging gas (pH 6.7, 5.4, and 8.9, respectively) ............................................................................................................ 254  xix Figure A-4: 5% CO2/95% N2 purged NS4 at 25 °C with various Cl− and/or SO42− ion concentrations (pH 6.7) ............................................................................................................... 254 Figure A-5: (top-left) E–pH diagram for Fe in water; (top-right) Simplified E–pH diagram for Fe in water indicating the domains of immunity, corrosion, and passivity, including the experimental passivation potential as a central dashed line; (bottom) E–pH diagram of Fe in water with the presence of chlorides, where GR1(Cl-) is chloride green rust ([Fe2+3Fe3+(HO−)8]+ · [Cl−·nH2O]−) [308] ……………………………………………………………………………………………..255 Figure A-6: E-pH diagram for the Fe-HCO3−-CO32−-H2O system at around 50 °C, where GR1 is carbonate green rust ([Fe2+4Fe3+2(HO−)12]2+ · [CO32−·2H2O]2−) [186] ......................................... 256 Figure A-7: Bode 𝜃𝐸𝐼𝑆 for NS4 solution at 25, 40, and 55 °C purged with 5% CO2/95% N2 (pH 6.7, 6.8, and 7.0, respectively) or 100% CO2 (pH 5.4, 5.5, and 5.6, respectively) ...................... 256 Figure A-8: Bode |𝑍| for NS4 solution at 25, 40, and 55 °C purged with 5% CO2/95% N2 (pH 6.7, 6.8, and 7.0, respectively) or 100% CO2 (pH 5.4, 5.5, and 5.6, respectively) ...................... 257 Figure A-9: Bode |𝑍| for NS4 + 10x HCO3− solution at 25 °C purged with 100% N2 (pH 8.9), and at 25, 40, and 55 °C purged with 100% CO2 (pH 6.4, 6.6, and 6.7, respectively), and 5% CO2/95% N2 (pH 7.6, 7.9, and 8.1, respectively) ........................................................................ 257 Figure A-10: Enlarged Bode |𝑍| for NS4 + 10x HCO3− solution at 25, 40, and 55 °C purged with 100% CO2 (pH 6.4, 6.6, and 6.7, respectively), and 5% CO2/95% N2 (pH 7.6, 7.9, and 8.1, respectively) ................................................................................................................................ 258 Figure A-11: Bode 𝜃𝐸𝐼𝑆 for NS4 + 10x HCO3− solution at 25 °C purged with 100% N2 (pH 8.9), and at 25, 40, and 55 °C purged with 100% CO2 (pH 6.4, 6.6, and 6.7, respectively), and 5% CO2/95% N2 (pH 7.6, 7.9, and 8.1, respectively) ........................................................................ 258 Figure A-12: Bode |𝑍| and 𝜃𝐸𝐼𝑆 plots for NS4 + 10x HCO3− solutions with added Cl− and/or SO42− at 25 °C, purged with 5% CO2/95% N2 (pH 7.6) ............................................................... 259 Figure B-1: Solution [NaHCO3], pH, and temperature relationships of test conditions in chapter 6 ..................................................................................................................................................... 260 Figure B-2: |𝑍| vs. frequency in the E > 0 VSCE region (see Figure 6-5a for legend of star data points) .......................................................................................................................................... 260 Figure B-3: Mott-Schottky plots at 1 kHz between -0.5 VSCE and 1 VSCE after 1 h of anodizing in 0.5 M NaHCO3 at 25, 50, and 75 ºC (pH 8.5, 8.9, and 9.3, respectively) ................................... 261 Figure B-4: Mott-Schottky plots at 1 kHz between -0.5 VSCE and 1 VSCE after 1 h of anodizing in 0.1, 0.25, or 0.5 M NaHCO3 at 75 ºC (pH 8.8, 9.2, and 9.3, respectively) .................................. 261 Figure C-1: Full spectrum OCP profiles showing stability approaching 1 h, in all conditions of Table 7-1...................................................................................................................................... 262  xx Figure C-2: OCP transients approaching 1 h of immersion in all test environments of Table 7-2 ..................................................................................................................................................... 263 Figure C-3: PDP in 0 ppm, 6 ppm, and 20 ppm O2 in NO3−-free solutions ................................ 264 Figure C-4: PDP in 0 ppm, 6 ppm, and 20 ppm O2 solutions with 0.005 M added NO3− .......... 264 Figure C-5: SEM images of steel specimen surface following immersion in NS4 solution deaerated with 5% CO2/95% N2 and containing 0.015 M NO3− (0.1 ppm O2, pH 6.7) ............... 265 Figure C-6: Different corrosion product formations and steel substrate at the edge of the tubercle formed in solution containing 6.0 ppm O2 (pH 7.8) and 0.015 M NO3− ...................................... 265 Figure C-7: Ex-situ XRD spectra of specimen immersed in 6 ppm and 20 ppm O2 solutions free of NO3− ......................................................................................................................................... 266 Figure D-1: Cathodic polarization curves of X100 steel in 100% CO2 NS4 and modified NS4 solutions at 20 °C ........................................................................................................................ 269 Figure D-2: Cathodic polarization curves of X100 steel in: (a) 100% CO2 NaHCO3-added NS4 solution at 20, 40, and 60 °C, natural pHs; (b) NaHCO3-added NS4 solution with adjusted pH at 20 oC ............................................................................................................................................ 270 Figure E-1 (for mm-model): Polarization profiles of X100 specimen immersed in different temperature and %CO2 environments, obtained through electrochemical corrosion experiments: (a) LPR fitting results at E < OCP; (b) cathodic regime of PDP results, with regression fits; and (c) anodic regime of PDP results ................................................................................................. 274 Figure E-2 (for m-model): CO2 evaluated at θ = 0° and θ = 90° in minimum and maximum induced temperature profiles, different soil structures, and ψ using: (a) uncoupled governing equations; and (b) coupled governing equations ......................................................................... 275 Figure E-3 (for mm-model): Convergence profiles of interfacial surface-electrolyte potential parameter (Es) averaged over full length of exposed steel surface (Es_average), for coarse, normal, and fine mesh resolutions ............................................................................................................ 277              xxi List of Abbreviations 2D Two-dimension(al) 3D Three-dimension(al) API American Petroleum Institute appm Atomic parts per million ASME American Society of Mechanical Engineers BISO Bilinear Isotropic Hardening CE Carbon Equivalent CIC Crack-in-Corrosion CP Cathodic Protection CPE Constant phase element DNV Det Norske Veristas EDX/EDS Energy-dispersive X-ray Spectroscopy EEC Electrochemical equivalent circuit EIS Electrochemical Impedance Spectroscopy FEM Finite Element Model(ling)/Method HE Hydrogen embrittlement HER Hydrogen evolution reactions HIC Hydrogen induced cracking HSLA High-strength low-alloy ICP-MS Inductively Coupled Plasma Mass Spectrometry LPR Linear Polarization Resistance MA Martensite-retained austenite mm-model Millimeter-scaled model m-model Meter-scaled model mmpy Millimeters per year mpy Milli-inches per year NACE National Association of Corrosion Engineers nn-pH Near-neutral pH OCP Open circuit potential PCIM Pipeline Corrosion Integrity Management PDP Potentiodynamic Polarization PHMSA Pipeline and Hazardous Materials Safety Administration PSP Potentiostatic Polarization ppb Parts per billion ppm Parts per million SCC Stress corrosion cracking SCE Saturated calomel electrode SEM Scanning Electron Microscope/Microscopy  xxii SHE Standard hydrogen electrode SVET Scanning Vibrating Electrode Technique TDS Total dissolved solids US United States USD United States Dollars XPS X-ray Photoelectron Spectroscopy XRD X-ray Diffraction                                          xxiii List of Chemical Formulae  Ag Silver Al Aluminum Ar Argon C Carbon C2H6O Ethanol C3H6O Acetone C6H3N3O7 Picric Acid Ca Calcium CaCl2 Calcium chloride, anhydrous Cl- Chloride CO2 Carbon dioxide CO32- Carbonate Cu/CuSO4 Copper/Copper-Sulfate Fe(OH)2 Iron (II) hydroxide Fe2+ Ferrous Fe3+ Ferric FeCO3 Iron (II) carbonate (siderite) H Elemental hydrogen H+ Hydrogen ion (proton) H2 Diatomic hydrogen, gas H2CO3 Carbonic acid HCO3- Bicarbonate HNO3 Nitric acid KNO3 Potassium nitrate Mg Magnesium MgSO4 Magnesium sulfate MnS Manganese sulfide Na2S2O5 Sodium metabisulfite Na2SO4 Sodium sulfate NaCl Sodium chloride NaHCO3 Sodium bicarbonate NaOH Sodium hydroxide NO3- N2 Nitrate Diatomic nitrogen, gas O Elemental oxygen  xxiv O2 Diatomic oxygen, gas O2- Oxide OH- Hydroxide Pd Palladium S Sulfur S2- Sulfide Si Silicon SiC Silicon carbide SO42- Sulfate                                      xxv List of Symbols   am Monolayer capacity Aw Atomic weight C Capacitance (general) Ca Adsorption capacitance (equivalent from corresponding CPE) Cads Concentration of the adsorbed electroactive species/compound Cdl Double layer capacitance (equivalent from corresponding CPE) CH* Critical H concentration CH,T Hydrogen concentration at a crack tip CO2 Effective O2 diffusivity in soil  Cs Subsurface H concentration at entry side of permeation specimen Csc Space-charge layer capacitance Cv Volumetric heat capacity D Pipeline diameter in FEM ddefect Depth of corrosion defect in FEM ddefect_max Maximum corrosion defect depth in FEM Deff Effective H diffusivity DO2 Soil O2 concentration  E Electrochemical potential of steel electrode e- Electron charge (1.60 × 10-19 C) Eanode Equilibrium potential of the anode-soil half-cell in FEM Eapp Applied voltage between pipeline and anode in FEM Ebd1 First potential of sustained layer breakdown or transpassivation Ebd2 Second potential of sustained layer breakdown or transpassivation Ebulk Trapped water electrolyte potential in bulk solution outside disbondment  Ec Critical potential in Csc vs. E plots Eco Potential of new anodic-cathodic reaction couple   Ecorr Corrosion potential Eeq Electrochemical equilibrium (reversible) potential Efb Flat band potential   xxvi EFe Potential difference at pipeline-soil interface in FEM Eo Overlap potential of two or more polarization curves Ep1 First potential of sustained current density decrease Ep2 Second potential of sustained current density decrease Es(x) Potential difference at electrode-trapped water interface in FEM Etrans Transition potential Ew Withdrawal potential  h Hour i Current density, general I Current, general ia Net anodic current density ibg Background permeation current density ic Net cathodic current density icharge Charging current density for hydrogen permeation Icorr Corrosion current icorr Corrosion current density ilim Limiting current density imax Maximum current density (at/in a specific potential level/range) in Inch iO2 O2 reduction current density ip Passive current density (lowest value if changing with potential) iperm Permeation current density iss H permeation current density at steady state ith,CP CP current density threshold (for HIC) itotal Total net current density in FEM simulations (ia - ic) j Imaginary number (√−1) K Boltzmann constant (1.38 × 10−23 m2 kg s-2 K-1) 𝐾𝑒𝑞𝑖  Adsorbate equilibrium constant kIG Intrinsic Griffin toughness km Kilometers ksi Kilopound per square inch  xxvii Ksp Solubility product Kth Threshold stress intensity factor L Coating disbondment length in FEM, or specimen thickness in H diffusion ldefect Length of corrosion defect in FEM m Meter M Molar concentration mm Millimeter MPa megapascal mV millivolts n Number of electrons in a reaction, or charge of an ion na Charge on dissolved cation from anodic oxidation (typically 2 for Fe) Nd Donor density of n-type semiconductors ni Common valence of a specific element “i” nm Surface phase capacity Pa Pascal Q Heat flux density Qa Constant phase adsorption element Qdl Constant phase double layer element Qf Constant phase film element Ql Charge flux R Gas constant Ra Adsorption resistance Rct Charge transfer resistance Rf Film resistance RO2  O2 depletion flux Rp Polarization resistance Rp* Sums of EIS resistance component values Rp** Sums of EIS resistance and diffusion component values (𝑅𝑃∗ + 𝑊) Rpore Pore resistance Rsteel Penetration rate in the exposed steel in FEM T Temperature, general  xxviii t Time (general), or pipeline wall thickness in FEM tb Breakthrough time tlag Time H permeation current reaches 63% of iss Twall Pipeline wall temperature in FEM VH Molar volume of hydrogen in steel w Coating disbondment width in FEM W Width of modelled soil section in FEM WA Warburg constant wt% Percent mass fraction (aka weight %) 𝑥𝑖𝑙 Molar fraction of adsorbate 𝑖 (∑ 𝑥𝑖𝑙 = 1𝑘𝑖=1 ) Z Impedance, general ZCPE CPE impedance Zim of Z" Impedance, imaginary component Zre of Z' Impedance, real component |Z| Impedance magnitude 𝛼𝑎𝑑𝑠 Adsorption constant in Hill-Langmuir isotherm δsc  Thickness of space charge layer ∆ Change or range in any specific parameter denoted in suffix (e.g. ∆O2) 𝜀  Dielectric constant 𝜀0  Permittivity of free space (8.85 × 10−12 F/m) θ Fractional coverage of adsorbate during adsorption, or pipeline circumferential angle (from 9 o’clock position) in FEM θads Fractional coverage of the adsorbent surface θEIS Phase angle in EIS λ Thermal conductivity 𝜌  Density, general 𝜌𝐹𝑒  Density of Fe 𝜎  Electrical conductivity, general 𝜎𝐻  Crack-tip hydrostatic stress 𝜎𝑈  Ultimate tensile strength  xxix 𝜎𝑦  Yield stress (at an offset strain of 0.5%) 𝜐O2 Stoichiometric coefficient of O2 in reduction reaction φ Soil potential in modelling studies ϕ Air porosity ratio in soil ψ Soil volumetric wetness 𝜔  Frequency of alternating current                           xxx Acknowledgements First and foremost, I wish to sincerely thank my supervisor and mentor, Professor Akram Alfantazi, for his instrumental role in the completion of this work. It has truly been a privilege to work under his supervision and guidance. I appreciate his patience, generous financial support, wise judgement, and continual encouragement throughout my doctoral program.  Funding for the research presented in this thesis has mainly been provided through NPRP Grant 6-027-2-010 from the Qatar National Research Fund (a member of Qatar Foundation). Parts of this work have been conducted during an Erasmus Mundus exchange funded by the European Commission. These financial supports have been very beneficial and are greatly appreciated. Special gratitude goes to all my colleagues in the Corrosion Group at UBC for their valuable friendship, scientific input, and positive influence, in particular: Hung Ha, Ibrahem Abushwashi, Masyam Mohammadi, Lokesh Choudhary, Wei Wang, Hamid Zebardast, Faysal Eliyan, Jing Liu, Michael Mahon, Marie Licausi, Victor Padilla, Shabnam Pournazari, and Davood Nakhaie. I also thank the members of the Materials Engineering staff at UBC for their help and patience during this work, in particular: Jacob Kabel, Michelle Tierney, Fiona Webster, Glenn Smith, Ross McLeod, and Mary Jansepar. I also sincerely appreciate the assistance of Hatim Jefri during the summer of 2014 through his Mitacs Globalink exchange from KSA. Special gratitude and thanks go to Professor Magd AbdelWahab for his supervision and guidance during my 6 month exchange at Ghent University in Belgium.  No words can describe how grateful I feel for the love and backing of all my family members. I am greatly indebted to them all, from my mother Omnia for her unconditional love and concern, my father, Professor Mohamed Gadala, for his wise advice and encouragement, to my beautiful sisters Marwa and Mariam for their affection, even when living abroad. I only wish I am as good a son and brother as they are parents and sisters.  xxxi Dedication       ةَّبِحُملا يَِتِلئاَعل  To my loving family  1 1 1. Introduction Pipelines are the most preferred method of transporting large volumes of crude oil, natural gas, and petroleum products over long distances. Compared to alternative methods of transportation such as by road or rail, pipelines are the safest and most efficient [1]. Yet, pipelines still suffer hundreds of ruptures and spills every year. Government estimates claim that large petroleum pipelines will experience a spill every 16 years for every 1000 km [2]. Based on the vast lengths of pipeline infrastructure currently in operation nationally and internationally, with over 100,000 km in Canada alone [3], this statistic is a serious concern for governments, operating companies, adjacent human communities, and regional ecosystems. Hence, ensuring the safety and integrity of pipelines in service and designing reliable future pipelines is undoubtedly of great importance to governments, energy companies, the general public, and the environment.   Figure 1-1: North American network of major liquid transmission pipelines [1]  2 1.1 Impact of pipeline corrosion Of the many causes and contributors to pipeline failures, corrosion ranks as the most important. Statistics by the US Department of Transportation's Pipeline and Hazardous Materials Safety Administration [4] reveal that pipeline corrosion is, on average, the second leading cause of pipeline failure incidents. As seen in Table 1-1, in terms of the total damage caused by any single contributor to transmission pipeline failure, corrosion often takes the lead over other factors such as natural force damage, equipment failure, or excavation damage. Corrosion of pipelines therefore has a direct and strong effect on their life. Since corrosion is influenced by many different factors, the life of a pipeline is correspondingly a function of these factors, such as quality of construction, coatings, cathodic protection (CP) systems, nature of the transported product, physicochemical properties of the external environment, operating conditions, and maintenance methods.  Table 1-1: Natural gas transmission pipeline incident summary by cause for 1/1/2002 - 12/31/2003, from US Department of Transportation’s Office of Pipeline Safety [4] Reported Cause Number of Incidents % of Total Incidents Property Damages % of Total Damages Fatalities Injuries Excavation Damage 32 17.8 $4,583,379 6.9 2 3 Natural Force Damage 12 6.7 $8,278,011 12.5 0 0 Other Outside Force Damage 16 8.9 $4,688,717 7.1 0 3 Corrosion 46 25.6 $24,273,051 36.6 0 0 Equipment 12 6.7 $5,337,364 8.0 0 5 Materials 36 20.0 $12,130,558 18.3 0 0 Operation 6 3.3 $2,286,455 3.4 0 2 Other 20 11.1 $4,773,647 7.2 0 0 Total 180  $66,351,182  2 13 Pipeline corrosion and subsequent failure have an economic and environmental impact which can hardly be understated. Since the start of the 1990s, concerns regarding the threat of corrosion to pipeline integrity have been high due to the significant economic losses and undesirable environmental damage associated with pipeline failures. Historically, corrosion has been identified as a major cause of reportable pipeline incidents in North America [5] and the  3 main culprit behind a significant pipeline failure in the Gulf of Mexico costing $1 billion (USD) [6]. Internal corrosion along the complete length of pipelines regularly results in costly maintenance programs. The corrosion-related cost to the transmission pipeline industry is approximately $5.4 to $8.6 billion annually, 52% of which is geared towards maintenance programs [4]. This is a large financial burden on pipeline operators and is sought to be minimized as much as possible through research and development.  1.2 Importance for Canada Canadian companies in the pipeline industry such as TransCanada Ltd. and Enbridge Inc. are actively pursuing contracts for future pipeline projects. They strive to offer the best safety and integrity for these projects to reduce costs, protect the environment, and eliminate any human injury or loss of life associated with a failure. They are also keen on developing more efficient and reliable processes and methods for operating and maintaining their existing pipeline infrastructure. Based on the aforementioned historical data and statistics, advancements in the understanding of pipeline corrosion through research and the development of superior corrosion control methods are two sensible means of achieving these goals. Furthermore, scientific research and technological advancements enhancing pipeline safety can contribute greatly in building public confidence regarding the subject of oil and natural gas transportation. The opinion and confidence of the general public regarding the safety of oil and gas transportation is critical for the support and approval of associated projects. For the abovementioned reasons, research on pipeline corrosion is of paramount importance for oil and gas companies in general, and Canadian pipeline companies in specific. 1.3 High-Strength Low-Alloy steels High-strength low-alloy (HSLA) steels are extensively utilized in pipeline construction for their desirable strength-to-weight ratios, a benefit which comes at only a modest price premium vs. conventional low-carbon steels [7]. The development of HSLA pipeline steels was  4 led by advancements in steel rolling processes, where thermo-mechanical (TM) rolling replaced hot rolling and normalizing as shown in Figure 1-2 [8]. Further enhancement of the overall process consisted of an accelerated cooling (AC) step succeeding the TM rolling, making it possible to produce the X80 grade under the American Petroleum Institute (API) classification according to minimum yield strength (80 ksi or 550 MPa).   Figure 1-2: Development flow chart of HSLA pipeline steels from 1965 to 2009 [8]  Today, much of the existing global pipeline infrastructure uses API X70 (483 MPa) and API X80 steel grades. More recently, micro-alloying with molybdenum, copper, and nickel has further increased alloy strength. The resulting API X100 (690 MPa) steel grade has garnered great interest from industry, due to the economic benefits it offers in terms of lower material, transportation, and fabrication costs [9]. Successful installation of the highest grade pipeline steel available in the world today (API X120, 825 MPa) was recently achieved for the first time by TransCanada Ltd., and research on various performance aspects of this steel is currently ongoing [10], [11]. Higher strength steel grades directly improve cost efficiency by withstanding higher operating pressures and greater design throughputs without the need for increasing pipe wall thickness. However, susceptibility of steel to cracking often increases with strength, due to the  5 corresponding reduction in ductility and toughness. The extent of this adverse reciprocal relationship must be further understood and minimized to avoid failures of HSLA pipelines in service [12]. 1.4 Environments causing external SCC Stress corrosion cracking (SCC) is the cracking induced from the combined influence of tensile stress and a corrosive environment on a susceptible material. In SCC, cracks penetrate into the susceptible material while most of the remaining surface stays intact. SCC is therefore classified as a catastrophic form of corrosion, since the detection of these cracks can be very challenging and, even when detected, the extent of their damage on the structure is not easily foreseen. The tensile stress required for SCC may be in the form of the applied stress listed in Figure 1-3, or residual stress introduced from fabrication processes like welding and heat treatment.  The second SCC requirement, a corrosive environment, depends on the location of interest within the pipeline system. On the pipeline’s internal surface, the corrosive environment is the internal media determined by product being transported and its physical/operational conditions such as viscosity and flow rate. This environment is always contact with the susceptible pipeline steel alloy, since internal surfaces are rarely coated. Externally, the corrosive environment for underground pipelines is the surrounding soil, which comes into contact with the susceptible material only at holidays and disbondments in the external coating. Samples of the environment gathered near locations where external SCC occurred on buried pipelines reveal that a carbonate (CO32−) – bicarbonate (HCO3−) electrolyte is responsible for the chemical degradation of the steel. Properties of this potent environment such as temperature, pH, and dissolved CO2 content significantly influence the resulting corrosion and SCC crack morphologies, as shown in Table 1-2. The third SCC requirement, a susceptible material, varies with properties such as composition, amount and size of inclusions, microstructure, and surface state. Even within the  6 narrow spectrum of HSLA pipeline steels (X70 – X120) researched in pipeline corrosion studies, there exists critical differences in material microstructure and inclusion content which change the corrosion behavior (see literature review of chapter 2).  Figure 1-3: Three conditions necessary for pipeline SCC, with corresponding dependencies [13] Table 1-2: Characteristics of nn-pH and high pH external SCC in pipelines – adapted from [13]–[15] Factor External SCC form Near-neutral pH High pH Associated electrolyte pH ~ 5 – 8.5 Dilute HCO3− solution  pH > 9.3 Concentrated CO32−–HCO3− solution  Location • 65% between compressor station and 1st downstream block valve  • 12% between 1st & 2nd valves  • 5% between 2nd & 3rd valves  • 18% downstream of third valve • Typically within 20 km downstream of compressor station • Number of failures falls markedly with increased distance from compressor and lower pipe temperature Temperature • No apparent correlation with temperature of pipe • Appear to occur in the colder climates where [CO2] in groundwater is higher • Growth rate decreases exponentially with temperature decrease Electrochemical potential • At free corrosion potential  • Cathodic protection largely ineffective at SCC sites • -600 to -750 mV (Cu/CuSO4) • Cathodic protection is effective to achieve these potentials Crack path and morphology • Primarily transgranular (across the steel grains) • Wide cracks with evidence of substantial corrosion of crack side wall • Primarily intergranular (between the steel grains) • Narrow, tight cracks with no evidence of secondary corrosion of the crack wall  7 Two types of external SCC have been discovered on pipelines: near-neutral pH (nn-pH) SCC in the 5 – 8 range and high pH SCC in the 9 – 13 range. High pH is the classical form of SCC, occurring with effective CP in the -600 to -750 mV Cu/CuSO4 range and the presence of a concentrated CO32−–HCO3− environment. The resulting cracking morphology is intergranular (IG), with decreasing growth rate at lower temperatures. High pH external SCC has been researched extensively [16]–[18]. There is general agreement regarding the mechanisms and controlling factors of this form of SCC, namely, anodic dissolution at grain boundaries and repeated rupture of passive films that form over the crack-tip. On the other hand, understanding of nn-pH SCC is more limited. In fact, to date, there is no precise mechanism identified to understand nn-pH SCC crack initiation and no predictive model developed to define the rate of crack growth [15], [19]. That being said, there is a considerable amount of evidence that anodic dissolution and hydrogen embrittlement play critical roles in the nn-pH SCC mechanism [20]–[23].   Figure 1-4: Areas of near-neutral pH SCC formation on external pipeline surfaces [13]  8 Near-neutral pH SCC was initially discovered in Canada in the mid-1980s, and occurs in a dilute 10-3 M HCO3− electrolyte containing dissolved CO2 and other groundwater anions like chloride (Cl−) and sulfate (SO42−) [24]. This electrolyte has a low electrical conductivity, on the order of 1000 µS/cm, and is typically found in the regions under disbonded coatings shown in Figure 1-4 [25]. The disbonded coating partially shields the CP from reaching exposed locations deep within these disbondments. Combined with the low conductivity of the electrolyte underneath, CP levels at these locations are insufficient, if not completely absent [26], [27]. The resulting cracking morphology from nn-pH SCC is transgranular, as revealed by the quasi-cleavage morphology, little branching, and significant lateral corrosion resulting in the destruction of the original crack faces [28].  1.5 Motivation This dissertation focuses on the external corrosion of API X100 pipeline steel in aqueous environments of nn-pH between 5 and 9. The broad motivation behind this research is the advancement of a fundamental understanding and simulation capability of environmental, microstructural, and surface/interfacial influences on HSLA steel corrosion. The environmentally-assisted diminishment of structural integrity is directly linked to corrosion and hence is also studied. Applications of this research are strong within the buried transmission pipeline field; however, understanding corrosion behaviors of infrastructural steels in aqueous environments with nn-pH, low conductivity, and CP-deficient or CP-absent characteristics is of benefit in numerous other applications. This includes potable or tap water piping systems [29], closed-loop freshwater piping circuits [30], crevices [31], and buried or concrete-embedded reinforcements [32], [33].      9 2 2. Literature review1 The corrosion behavior of HSLA steels depends on numerous environmental, material, and surface properties which range from the macroscale of the overall environment, such as the soil topology, to the smaller scale of the metal, corrosion product, and electrolyte interfaces as shown in Figure 2-1 [34]. In nn-pH buried environments specifically, aspects which affect HSLA steel corrosion include anion constituents, dissolved CO2 and oxygen (O2) content, pH, CP effectiveness, steel microstructure, and surface effects such as corrosion product growth and morphology [14], [35]–[38]. It is particularly important to understand the electrochemical processes that occur at the metal surface, including the development of anodic and cathodic sites and the rate of reactions at these sites. In addition, the mechanisms with which oxide or passive layers impact electrochemical corrosion activity through the control of specie diffusion to and from reaction sites is a critical area of further study [34], [39]–[41].   Figure 2-1: The numerous environmental, material, and surface effects on metal corrosion in soil environments, ranging from the macroscale to the smaller surface level scale [34]                                                   1 F. F. Eliyan, I. M. Gadala, H. M. Ha, and A. Alfantazi, “Pipeline Corrosion,” ASTM Fuels and  Lubricants Handbook: Technology, Properties, Performance, and Testing, 2nd edition. PA: ASTM  International, 2016.  10 The following is a review of previous studies on the external corrosion of ferrous metals in aqueous nn-pH environments which encompass the interrelated topics illustrated in Figure 2-1. Investigations of secondary processes such as hydrogen (H) ingress, diffusion, and embrittlement are also discussed due to their dependence on corrosion processes and their influence on pipeline integrity. Finally, previous works which present computer models and numerical simulations relevant to external pipeline corrosion and the structural integrity of corroded pipelined are reviewed.  2.1 Ionic and pH influences on corrosion processes in anoxic soil environments The role of environmental influences such as ionic composition, pH, temperature, and CO2/O2 concentrations in the corrosion of buried pipelines is still of interest to researchers. A number of researchers have carried out electrochemical studies in laboratory settings simulating real service conditions [42]–[45]. There are also numerous articles in the literature on the behavior of metals in solutions simulating the various physicochemical conditions which arise in underground environments [46]–[48]. In a comparative investigation of 4 different soil types extracted from areas where nn-pH SCC was found on the TransCanada pipeline system, Chen et al. found that the majority of SCC-tested samples exhibit transgranular, cleavage, or quasi-cleavage morphology perpendicular to the loading axis [24]. It is suggested that the relative aggressiveness of a soil environment to SCC is determinable by the change in pH between CO2–saturated and CO2–free anoxic solutions. More aggressive soils exhibit a narrower range of pH, and higher levels of CO2 reduce pH which increases corrosion leading up to SCC (Figure 2-2). Less aggressive soils hinder significant corrosion dissolution toward crack development, resulting in cracking at 45º to the loading axis of the tensile sample, with elongated ductile dimples. These deformation-induced cracks therefore occur on the plane of highest shear stress, prevailing over the brittle cracks forming perpendicular  11 to the loading axis. Increased corrosion leading to greater SCC susceptibility in environments of greater CO2 concentration was also found by Fang et al. [49].   Figure 2-2: Schematic showing the effect of CO2 concentration on the environment’s pH, and the resulting influence on aggressiveness (where te/tair is the time to failure ratio in environment over that in air) [24]  In the nn-pH conditions of [24], varied levels of Cl− activity appears to have a minor effect on SCC susceptibility. Slow strain rate tensile (SSRT) tests in standard NS4 and NOVA type simulated nn-pH solutions also reveal no significant effect for Cl− at most potentials [50]; the composition of NS4 contains an order of magnitude greater Cl− concentration than that of NOVA. Yet, Cl− attack seems to be a function of pH as suggested by Liang et al. in [51], where it is found to be more severe at higher pH. Anodic dissolution at local sites on the metal surface due to Cl− attack results in pitting, even in extracted soil solutions with a fairly alkaline pH of approximately 9. Likewise, in soil solutions with pH values notably more acidic to the nn-pH range (i.e. < 5), an increased ratio of Cl− to other anions has also been reported to cause pitting [52]. Since SCC cracks initiate readily from the bottom of pits due to the stress concentrating effect of the metal loss at those locations [53], increased pitting results in higher SCC susceptibility. Three essential events occur at the bottom of pits: concentration of applied and/or  12 residual stresses [54], preferential electrochemical dissolution of the metal, and local acidification of the environment [55]. The established chronology between pitting and SCC necessitates that tests without stress are performed in material-environment combinations of interest. In other words, to understand corrosion behavior leading up to pitting and crack initiation, fundamental electrochemical studies under no load are needed.   Figure 2-3: Potentiodynamic polarization in 0.1, 0.5, and 0.8 M HCO3− solutions at 20 ºC in (a) Cl−-free and (b) Cl−-containing conditions [56] Eliyan et al. evaluated the basic electrochemical behavior of X100 in relatively concentrated HCO3− solutions (0.1 – 0.8 M), and described clear enhancements in passive layer development driven by HCO3−, and opposing Cl− attack on passivation combined with augmentation of corrosion rate as seen in Figure 2-3. At open circuit potential (OCP) and using electrochemical impedance spectroscopy (EIS), the diffusion found in Cl−-free solutions is eliminated in Cl−-containing solutions, and adsorptive processes are manifested instead. At anodic sites, Cl− accelerates anodic dissolution and ferrous (Fe2+) release from iron specimen, as discussed by Lorenz and Heusler in [57] and supported experimentally by Zhang et al. [58]. In similar characterizations of passive films formed on steel in HCO3− solutions using EIS, the suppression of diffusion processes with increased Cl− concentrations was also found [59]. Evaluation of the semiconductive properties of these films using Mott-Schottky further reveals increases in passive film thickness in the absence of Cl−, attributed to lower donor densities of the  13 space-charge layer, the thickness of which (𝛿𝑠𝑐) is found through {E-2.1} below. In this potential (E) dependent equation, ε is the dielectric constant of the passive film, ε0 is the permittivity of free space, e is electron charge, 𝑁𝑑 is the donor density (for n-type semiconductors only), 𝐸𝑓𝑏 is the flat band potential, k is the Boltzmann constant, and T is the temperature in Kelvin.     𝛿𝑠𝑐 = √2𝜀𝜀0𝑒𝑁𝑑(𝐸 − 𝐸𝑓𝑏 −𝑘𝑇𝑒)         {E-2.1}  Figure 2-4: (a) Actual CP levels at different positions from opening mouth (OM) and different times; (b) pH distribution of the electrolyte inside the shielded disbondments at different positions from opening mouth, and different times [25] The typical test setup of most nn-pH corrosion and SCC tests neglects the concentration-cell aspect of the corresponding environment under pipeline coating disbondments. Since the chemistry of the solution trapped in the disbondment is altered due to the effect of the gradient in CP [60], a tensile specimen exposed to a single pH and CP environment is not representative of the real conditions present in nn-pH corrosion and SCC. In a novel testing method designed by Eslami et al., large coating disbondments are simulated with a poly methyl methacrylate partial shield between the tensile specimen and the corrosive electrolyte [25]. CP is only applied at the opening (holiday) in the coating, and only enters the disbondment region through this opening due disbondment shielding. The distribution of pH and potential (vs. saturated calomel electrode, SCE), monitored periodically throughout the cell, clearly indicates the variable redox conditions within the disbondment (Figure 2-4). This results in corrosion rates and crack initiation severities a b  14 which follow a nonlinear relationship; between the opening mouth of the disbondment and the furthest point in the shielded disbondment crevice, the largest pit diameters and highest percentage of pit coverage are discovered. This is proportional not only to corrosion, but also to microcracks at the bottom of the pits and SCC severity. Thus, it is neither the region with full CP (at opening mouth) nor the region lacking any CP (at the crevice tip) which is most critical in terms of corrosion. This finding highlights the need for more investigations into the fundamental electrochemical processes occurring on steel surfaces with changes in pH within the nn-pH range, and likewise variations in CP near OCP.  Akin to electrolyte pH differences based on disbondment dimensions and exposure times, albeit still within the nn-pH range, ionic composition of the electrolyte is a variable which varies based on soil physicochemical properties. The ionic constituent differences between the standard simulative electrolytes of NS4 [61], NOVA [50], C1 [62], C2 [25], and others undoubtedly affect corrosion performances of HSLA steels exposed to them. In solutions adjusted to pH 3–7 and containing NaCl, CaCl2, Na2SO4, MgSO4, NaHCO3 and KNO3, weight loss and electrochemical experiments reveal that corresponding parameters, which included the corrosion current and potential (Icorr and Ecorr), were strongly dependent on pH [63]; the greatest mass loss and Icorr occur at pH 3. This is a recurring result [42] which is corroborated by corrosion acceleration in acidic soil environments, such as those acidified from acid rain [47]. In nn-pH underground conditions, the only significant pH-altering electrolyte constituents are HCO3− and dissolved CO2 concentrations, governed by the temperature-dependent carbonic acid (H2CO3) association and dissociation equilibrium of {R-2.1} [64], [65]: CO2 (aq. ) + H2O ↔  H2CO3 (aq. ) ↔ H+ (aq. ) + HCO3− (aq. ) {R-2.1} In electrochemical tests carried out to study the effect of electrolyte composition, the aggressiveness of cations is reported to follow the order K+ > Mg+ > Ca2+, whereas that of the anions follows the order SO42− > HCO3− > NO3− [48]. The differences in corrosion incurred by  15 these various ions were not dramatic, causing a modest increase in Icorr around OCP. In EIS tests conducted in these environments, the analysis of which consisted of a three-element electrochemical equivalent circuit (EEC), all anions decreased the charge transfer resistance (Rct) [47], [48], [63]. Liu et al. argued that while the presence of K+ decreased Rct, Ca2+ and Mg2+ increased it [48]. In the work of Benmoussa et al., it was additionally found that the aggressiveness of solution increases with temperature, and contrariwise Rct increases with time due to the development of a protective film.  Electrochemical tests conducted in the laboratory also reveal a relatively high correlation between polarization resistance (Rp) and mass loss in conditions relevant to buried pipelines [44], [45]. Furthermore, the mass transfer of O2 plays a vital role in the kinetics of corrosion, where the entire process is limited by a combination of activation and diffusion control as postulated in [45]. Yet, it should be importantly noted that great difficulty exists in suitably accelerating corrosion in a laboratory environment, as resistivity values in the field do not match those in the laboratory; hence the electrochemical parameters derived from soil solution tests do not match field corrosivity tests [66]. The stimulating and interplaying effects of solution chemistry, notably pH and ionic composition, are sought to be understood through simulated soil environments. Yet, since solution chemistry is only one of many factors controlling corrosion in buried environments, this must be met with more comprehensive investigation of surface layer growth, diffusion, CP, and auxiliary effects present in reality. This is achievable through dedicated studies of time-dependent surface effects, as reviewed in section 2.2 below, H evolution and diffusion secondary processes, as reviewed in section 2.3, and numerical models of diffusion, CP, and corrosion processes in underground environments, as reviewed in section 2.4.   2.2 Time-dependent corrosion product and surface effects in anoxic and oxic soil environments The early work of Scully and Brandy evaluated soil corrosion of buried steel pipe with linear polarization resistance (LPR) and EIS tests [46]. Corrosion rates calculated from these  16 electrochemical methods decrease with time of exposure. In a comparative study of internal and external corrosion of buried cast iron pipe, Sancy et al. found markedly lower corrosion rates on internal surfaces and notably different forms of impedance spectra [67]. On the inner surfaces, the metal behaved as a semi-infinite conducting porous electrode, where a CO32− layer at pore edges limited the cathodic reaction to occurrence exclusively on the bottom of the pores. On external surfaces, mass-transfer limited behavior manifests akin to that seen in [45], where it is believed that the cathodic reaction is controlled by diffusion through the non-conducting porous layer. Figure 2-5a is an example of the porous nature of calcium carbonate (CaCO3) corrosion product formed on X70 steel after 40 hour (h) immersion in 5% CO2 NS4 solution at 22 °C, with the corresponding corrosion current density (icorr) map showing a localized corrosion distribution dependent on the surface layer morphology [41]. The icorr map shown is obtained using the Scanning Vibrating Electrode Technique (SVET) on a Scanning Electrochemical Microscope (SECM). In this technique, icorr is proportional to the test solution conductivity, the vibrating amplitude (usually 30 m) of the scanning microelectrode, and the potential difference of this fine microelectrode (10 m tip) between the peak and valley during its vibration normal to the surface of the specimen.   Figure 2-5: (a) Surface morphology of CaCO3 layer formed on X70 surface after 40 h immersion in 5% CO2 NS4 solution at 22 °C; (b) icorr map measured on X70 specimen covered with CaCO3 [41] a b  17  Figure 2-6: (a) Ecorr (vs. SCE) of pretreated steel under anoxic conditions in nn-pH saline solutions, where square points show Rp values from LPR; (b) Schematic illustrating the film transition process within an acidified pore, assuming seperation of anode (Fe dissolution) and cathode (HCO3− discharge) [68] In CO32−–HCO3− environments, a main corrosion product causing passivation is iron carbonate (siderite, FeCO3), with a formation dependent on pH and HCO3− concentration according to {R-2.2} below [56], [67]. Not only do CO32−–HCO3− concentrations have a significant effect on passive/oxide layer development as shown in [67], the development of this layer over time greatly affects ensuing corrosion processes. Electrochemical studies contribute to the understanding of this time-dependent process in ways not readily possible from field exposures, where cumulative corrosion rate is normally only measured at the end of tests.  a b  18 Fe + HCO3− → FeCO3 + H+ +  2e−      {R-2.2}  A common trend in the growth of passive/oxide layers on corroding steel is the decrease of Icorr with time, coupled with an increase in Rct or Rp measured by EIS or LPR, respectively [34]. X60 steel exposed to a soil-simulating solution develops a corrosion layer which causes both an initial decrease in Rct and a continuous increase in the double-layer capacitance modelled with a constant phase element (CPE) in the three-element EEC [43]. Yet decreasing overall corrosion can lead to localization of anodic activity and an increase in Icorr as shown by Aung and Tan [69]. This is backed by the findings of Norin and Vinka which suggest differences in the factors controlling local and general corrosion rates of buried carbon steel [70]. In a study of nominally anoxic corrosion of carbon steel in nn-pH saline environments, Sherar et al. identified an upsurge in corrosion rates corresponding to a sharp decrease in Rp (see Figure 2-6a) during periodic LPR tests on specimens pretreated with an initial FeCO3 film [68]. It is postulated that a film transition process occurs within the acidified pore of iron oxide/oxyhydroxide formations developed during the 35 day exposure of these tests. The enhanced corrosion rate is suggested to be sustained by proton (H+) reduction from HCO3− on Fe3O4 surfaces, while accelerated corrosion occurs primarily at anodic sites at the base of acidified faults within the film as shown in Figure 2-6b.  The potential decrease observed in Figure 2-6a on pretreated carbon steel resembles that which was reported by Qin et al. on a millscale-covered pipeline surface [71]. Millscale is a highly oxidized deposit (O/Fe ≈ 2.4) containing iron oxidation products and aluminosilicate clay minerals commonly originating from furnace slag, and consisting of a dual-layer morphology. Whereas millscale-free specimens exhibit a constant OCP or Ecorr during immersion, millscale-covered specimen exhibit potential decreases with time in three distinct stages. After a first relatively stable Ecorr stage, potential falls rapidly in the second stage due to reductive oxide dissolution, wherein pore enlargement of the porous oxide morphology occurs. The resistance of  19 the pores (Rpore) in the EIS fit results falls markedly in this second stage of immersion, consistent with solution penetration into the pores of the millscale. Again, like in [68] the porous structure of the millscale augments the corrosion rate by separating the anodic and cathodic sites and keeping a high ratio of cathodic to anodic surface areas. The pores continue to enlarge with time and the establishment of a double-layer at the exposed metal becomes easier, corroborating the mechanism proposed by Linter and Burstein for pores in a Fe(OH)2-containing film [64] and by Belmorke et al. in an electrolyte-penetrated paint film [43].     In O2-containing actual and simulated environments, HSLA steels often develop multi-component corrosion products including complexes of Fe2+ and/or ferric (Fe3+) with hydroxide (Fe(OH)2), carbonate (FeCO3), oxide (Fe3O4 and Fe2O3), and oxyhydroxide (γ/α-FeOOH) [30], [42], [72]–[78]. Oxidation reactions of compounds found in these multi-layer corrosion products are viable at different O2 concentrations, such as the oxidation of FeCO3 in nominal O2 conditions [79] versus that of Fe(OH)2 in the complete absence of O2 [32], [80]. With increasing O2 content, reddish/orange oxide/oxyhydroxide tubercle formations appear during excavation as speckles on the white FeCO3 deposit [81]. Some suggest that this is the result of anoxic corrosion products air oxidizing during excavation and subsequent examination [82], whereas it could potentially be due to electrochemical reactions involving anoxic corrosion products, such as through {R-2.3} [73]. Furthermore, temperature plays a role in protective iron oxide formation; Nie et al. found that oxides form on metal surface at ambient temperatures but not at elevated temperatures at around 50 °C [45]. Conflicting causalities of corrosion product formations prompt more studies of corrosion and passivation steps in oxic environments, especially on new alloys like X100. O2 concentrations are variable in practical situations not limited simply to soils, hence understanding the different O2-dependent viabilities of these reactions is important.  3FeCO3 + 4H2O → Fe3O4 + 3HCO3− + 5H+ + 2e−   {R-2.3}  Underneath disbonded pipeline coatings, corroded surfaces not only affect the corrosion rate of the exposed steel, but also the rate of polarization achieved within the disbondment. The  20 effectiveness of CP in a highly insulating tape disbondment on a steel surface covered with γ-FeOOH and Fe2O3 was investigated by Perdomo et al. [83]. It was found that the rate at which the potential decreases in the disbondment in the corroded conditions is about half that of the bare steel condition (rate of decrease: bare steel > Fe2O3 > γ-FeOOH). In addition, OCPs for the rusted surfaces are more positive than the bare-steel condition. Low O2 levels recorded from samples extracted from within the crevice are suggested to be due to the reduction of O2 into OH-, a process which increases the acidity of the local solution with distance from the coating opening (the highest CP is found at the opening). This finding agrees with the periodic pH measurements of Eslami et al., shown previously in Figure 2-4b.  The anoxic local conditions at the coating opening mouth with the application of CP, and the corresponding increased alkalinity, are fundamental in reducing corrosion rates at exposed metal coordinates there. It appears from the findings presented in [83] that CP current does not need to flow directly to all coordinates of the exposed steel in a disbondment for protection to be realized. Since the ability of pits to remain active depends on the local composition and pH of the solution within the pit [55], the concentration cell effect with the disbondment has implications on pit initiation and growth. Concentrated solutions within the pit are often covered by the remaining passive oxide layer, which ruptures when the pit reaches a critical size [84]. Rupturing can also occur with applied stress and can be simulated through the forced electrode abrasion technique, with which Park et al. [52] identified increased pit repassivation on X65 pipeline steel in high pH solutions compared to low pH solutions. The sufficient passivity of the high pH conditions allows for selective dissolution at the grain boundaries, leading to intergranular SCC. In contrast, the pits in low pH conditions do not repassivate and manifest a pit-to-transgranular SCC mechanism.    21 2.3 Hydrogen evolution, absorption, and diffusion processes in steels exposed to simulated soil solutions Hydrogen induced cracking (HIC) and hydrogen embrittlement (HE), resulting from the entry of atomic H into metals, are predominant forms of failures to which high-strength steels are susceptible [85], [86]. Transgranular SCC in underground pipelines is characterized by wide cracks with quasi-cleavage morphology and little branching [28], [87]–[89]. The crack sides often suffer significant lateral corrosion resulting in the destruction of the original crack faces.   Figure 2-7: Variation of embrittlement index (using % reduction in area) with potential for X100 steel [90] The mechanism of transgranular SCC in underground pipelines is not yet fully understood, albeit a combination of anodic dissolution and HE is proposed to be responsible for the cracking [21], [91], [92]. Evidences supporting the contribution of H to the crack growth were reported in literature [19], [51], [93]–[96]. For instance, the morphology of the transgranular SCC fracture surface resembles that observed in HIC [21], [97]. A further indication of the entry of H into the steel is the manifestation of secondary cracks not connected to the surface of the specimen and predominantly nucleated on bands of pearlitic material [21], [28]. In addition, the  22 increase in the susceptibility of steels to transgranular SCC at more cathodic potentials also supports the proposed HE mechanism, as shown by Mustapha et al. [90] (Figure 2-7) and in [92], [95], [96]. It is generally acknowledged that higher strength reduces steel resistance to HIC, as found in the work of Hardie et al. [86], where tensile specimens of X60, X80, and X100 steels are pre-charged with H through cathodic polarization in a sulphuric acid solution. H content in metals increases with charging time [98], but reaches what can be considered a saturation plateau after a certain time. For example, in [93] the H content in X65 steel is observed to increase linearly with charging time up to 800 minutes, remaining nearly constant thereafter (Figure 2-8). Upon loading, steel specimens typically experience significant loss in ductility, the degree of which increases with charging current density [86]. This is confirmed by [19], where the SCC susceptibility of steel in the simulated soil solution was found to have the same order as the subsurface H concentration (Cs). At a H pre-charging current density of 0.10 mA mm-2, the degree of HE increased with steel strength [86], matching the findings of Capelle et al. [98] for X52, X70, and X100.  Electrochemical and corrosion studies of steels in nn-pH soil environments mainly focus on anodic processes [19], [97], [99]. The cathodic processes on steels, particularly the hydrogen evolution reactions (HER) which might play an important role in controlling transgranular SCC mechanisms in the nn-pH environment, are not thoroughly understood. This is because our understanding of the HER in CO32−–HCO3− solution is mainly obtained from studies of internal corrosion of pipelines exposed to CO2 solutions [36], [100]–[103], wherein corrosion environments usually contain a significant amount of dissolved salts (e.g. approximately 3% NaCl) and have a pH in the range of 4 to 5. These electrolytes are far more concentrated and more acidic than the corrosive electrolyte of soils surrounding pipelines at locations where transgranular SCC is found. Nevertheless, in these internal pipeline corrosion studies though, it was found that the presence of CO2 increases the corrosion rate of corrosion of Fe by increasing  23 the rate of the HER [64], [100], [103]. Many researchers have suggested that the cathodic current is the sum of the currents for H+ reduction and H2CO3 discharge, at least in nn-pH and alkaline environments. However, at higher pH the direct discharge of HCO3− becomes important at the corresponding Ecorr of steels [104].  Figure 2-8: (a) Hydrogen permeation curves of X65 specimens (600-grit finish) at -1200 mV with and without calcium carbonate coatings (b) Hydrogen permeation current vs. applied cathodic potential [93]  Understanding the effects of parameters such as solution chemistry, aeration condition, CO2, pH, and temperature on the kinetics of the cathodic reactions, particularly HER on HSLA steels, is important in studies of the transgranular SCC mechanism in nn-pH soil environments surrounding buried pipelines. H generation kinetics on the steel surface will affect the amount of H absorbed into the material and subsequently assist SCC crack growth. Following ingress, H in steel reduces the threshold stress intensity (Kth) as shown in {E-2.2} below. This equation describes the dependence of Kth on Cs, where α, α”, and β’ are constants, kIG is the intrinsic Griffin toughness for cleavage fracture without H, 𝜎𝑌𝑆 is the yield strength of the material, and CH,T is the H concentration at the crack tip [105]. CH,T is proportional to Cs through {E-2.3}, where 𝜎𝐻 is the crack-tip hydrostatic stress, 𝑉𝐻 is the molar volume of H in the steel, T is temperature, and R is the gas constant [106], [107]: 𝐾𝑡ℎ = 1𝛽′𝑒𝑥𝑝 [(𝑘𝐼𝐺−𝛼𝐶𝐻𝜎,𝑇)𝛼"𝜎𝑌𝑆]      {E-2.2}  24 𝐶𝐻𝜎,𝑇 = 𝐶𝑠𝑒𝑥𝑝 [𝜎𝐻𝑉𝐻R𝑇]      {E-2.3} Recent studies on H damage manifestations in API X100 line pipe steels, including HIC under severe cathodic charging in both acidic and alkaline environments [104], [108], [109] and HE under different potential ranges in bicarbonate solutions [90], emphasize the need to understand the H generation, absorption, and transport kinetics in this modern HSLA material. Depending on HSLA steel grade, a critical H concentration (𝐶𝐻∗ ) exists at which significant loss of local fracture resistance occurs. 𝐶𝐻∗  decreases with increasing 𝜎𝑌𝑆 and ultimate tensile strength (𝜎𝑈), approximately following an inverse quadratic relationship where 𝐶𝐻∗  is proportional to 1/𝜎𝑈2 [98]. This decrease can be explained by dislocations being “more pinned” by the precipitates found in higher strength steels. In other reports, a CP current density threshold (𝑖𝑡ℎ,𝐶𝑃) is identified instead of 𝐶𝐻∗  in the steel, since this is more applicable for creating safe CP design standards. In a report on the precautions necessary to eliminate microcracking damage to plain carbon steels subject to CP, Cialone and Asaro indicate that CP currents should be restricted to below 0.06 mA mm-2 [110], a value comparable to that reported by Hardie et al. in [90].  Steel alloy grade, strength, and heat treatment, in addition to specimen geometry, notches, defects, mode of H charging, and rate of loading upon charging are all factors which affect 𝐶𝐻∗  and 𝑖𝑡ℎ,𝐶𝑃 [98], [104], [108], [111], [112]. Researchers have focused on the specific aspects of steel grade to better understand its influence on HIC/HE. The number, distribution, and types of inclusions in each steel alloy influence H uptake, diffusion, and HIC. Inclusions, especially oxides and carbides, serve as effective sites for HIC crack initiation [104], [108], [111]. Dong et al. observe that cracks initiate preferentially at aluminum oxide, titanium oxide, and ferric carbide inclusions in X100 steels [104]. This suggests that different types of inclusions are more detrimental than others, an observation ratified by the findings of Jin et al. where H-induced cracks initiated mainly at Aluminum (Al) and Silicon (Si) enriched oxide inclusions (Figure 2-9) in X100 specimens [108]. No cracks were associated with the elongated Magnesium Sulfide  25 (MnS) inclusions or the inclusions with a mixed composition including Calcium (Ca), Al, O, and Sulphur (S).   Figure 2-9: SEM morphologies of the cross-section of X100 specimen and the chemical composition obtained at the individual inclusions (a) an Al-enriched inclusion; (b) a Si-enriched inclusion [108] Since steel alloy grade is only one of the factors affecting 𝐶𝐻∗  or 𝑖𝑡ℎ,𝐶𝑃, even HSLA steels of the same grade can exhibit different sensitivities to H. This can be due to different heat treatments, where for instance H appears uninfluential on the ductility of as-received and spray-cooled X70 specimens, yet affects quenched and quenched-&-tempered specimens considerably [112]. Environmental and surface conditions of steels also play a large role in H-induced damage mechanisms. Corrosion product deposits on the steel greatly impede H ingress. Reduced H flux is typically measured on specimens with a corrosion product deposits such as CaCO3, compared to those with bare surfaces [93]. Of the 3 steps controlling H permeation/ingress, namely {1} evolution of H through reduction, {2} H adsorption onto steel surface, and {3} H penetration into  26 the steel, H adsorption and penetration are greatly hindered by the presence of corrosion products [41], leading to improved resistance to SCC (shown for mill-scale covered vs. machined surfaces in [113]). It is well accepted that the HER of step {1} proceeds through three sub-steps [114]–[116]: {i} electrochemical reduction/evolution of H2O (in nn-pH or alkaline solutions) or H+ (in acidic solutions), {ii} electrochemical recombination of evolved H, and {iii} chemical desorption of H. Even though H evolution is greatly enhanced on the typically porous corrosion products (e.g. Figure 2-5a) due to the increased effective area for the HER as shown in [41], the decelerated steps {2} and {3} on electrodes covered with corrosion products decreases Cs.   Figure 2-10: Steady-state H amount released from an X100 steel specimen as a function of the charging current density, where the 𝐶𝐻∗  needed to initiate HIC is identified [108] It is also well documented that the uptake of H is increased by increasing temperature [20], [117]. A comparison of HIC findings in [86], [110] versus those in [108] further verifies this; in the latter, HIC only occurs in the X100 sample after 20 hours of charging at 𝑖𝑡ℎ,𝐶𝑃 = 30 mA cm-2 (0.3 mA mm-2), corresponding to 𝐶𝐻∗  = 3.24 parts per million (ppm) in Figure 2-10, found using the JIS Z3113 method [118], [119]. This value is 3 times that of [86], reported to be 0.10 mA mm-2 for the same X100 alloy, and also around 3 times the 𝐶𝐻∗  ≈ 1 ppm found for X65 quenched and tempered specimen [120]. The difference in results is suggested here to be due to 𝐶𝐻∗  = 3.24  27 dissimilar charging times between the reports and the 20 ºC temperature difference during pre-charging.   With respect to mechanical property degradation, H absorption, desorption, specimen geometry, and 𝐶𝐻∗  of a material are all interrelated. For example, even though resistance to H absorption is high for bare X100 steel in comparison to other HSLA grades, X100 still has relatively higher susceptibility to HE due to a lower 𝐶𝐻∗  (inversely proportional to 𝜎𝑈 as mentioned earlier). Moreover, the H desorption behavior of the steel will dictate the speed of recovery of ductility after H ingress. Recovery to air-charged levels of ductility can fully materialize if specimens are left for several days at ambient temperature after charging [121]. This H desorption can be quantified to identify the H concentration decay profile, which normally reaches a steady-state concentration after 72 hours [93]. These findings from desorption tests suggest that the primary cause of HE is the H trapped in the steel microstructure and not the microcracks resulting from the H stressing trap locations and inclusions, since microcracks do not heal after desorption.  2.4 Numerical modelling of corrosion, CP, and pipeline structural integrity    Models and simulations of steel corrosion with CP and gas transport Mitigation strategies for external pipeline corrosion revolve around two methods: CP and protective coatings [35], [122]. The corrosion of pipeline steel following coating failure is lessened by CP systems, but is heavily affected by physicochemical properties of the immediate surrounding soil as elaborated in section 2.1. In the bulk soil not immediately adjacent to a coating failure, physicochemical soil properties like soil moisture content, soil type, soil resistivity, soil pH, redox potential, and microbes in the soil are critical with regards to CP effectiveness [123], especially in sections between the CP anode and the pipeline [124]. These properties have influential impacts on gas diffusion and heat transfer phenomena. Complexities  28 arise from the compound dependence of electrical resistivity, gas diffusivity, or thermal conductivity parameters on one or more soil properties/conditions such as particle distribution, porosity, moisture, and temperature [42].  Figure 2-11: (a) Planar cross-section of buried tank and anode system modelled in [125], with meshing of the soil media; (b) Potential distribution at surface of tank for different anode positions (circle: bottom of pit, square: at half depth, triangle: critical potential) Empirical correlations for predicting soil corrosivity often intentionally overlook certain soil parameters or conditions for simplicity. Booth et al. report in an early study that simply resistivity and redox potential are sufficient predictors of corrosivity (better than moisture content) [126], whereas more recently, Tomlinson and Woodward reveal that soil type and a b  29 structure are the determining factors regardless of other properties including the structure’s position with respect to the water table [127]. Empirical limits separating corrosivity categories have been postulated, such as a minimum of 20 wt. % moisture content for the soil to be considered “non-aggressive” [126], or a soil resistivity of at least 1000 Ω-cm to avoid classification as “very corrosive” [128]. Empirical guidelines such as these simplify the complex reliance of gas transport, heat transfer, and ensuing reaction kinetics on many interdependent soil parameters, leading to erroneous predictions of corrosion rates and ultimately incorrect assessments of structural integrity. Modelling and numerical simulations can capture a larger spectrum of the fundamental processes occurring on cathodically protected pipelines with coating failures and provide more accurate data on unfavorable environmental conditions, perilous operating parameters, and critical corrosion locations on exposed surfaces [125], [129], [130].  In the work of Rabiot et al., a finite element model was used to compare the relative influence of coating quality, soil electric conductivity, and CP anode size, position, and type on the corrosion of buried steel tanks [125]. Although the soil’s electrical conductivity played a leading role in consequent corrosion, in addition to parameters related to the CP system like the anode position as shown in Figure 2-11b, the model did not consider other important soil-related parameters such as water content. The work of Miltiadou and Wrobel incorporated the influence of the limiting current density of O2 reduction with the electrolyte conductivity parameter, and used a semi-analytical solution of governing equations to compute the distribution of electrical potential within the electrolyte [129]. Yet, spatial and time dependent differences in electrolyte properties, specifically gas transport or electric conductivity, were not accounted for since the electrolyte considered was uniform and non-porous. Similarly, Martinez and Stern simulated the effectiveness of an impressed current CP system in a uniform and non-porous electrolyte media inside a cylindrical structure [130]. The wire anode running along the cylinder’s length supplied the CP current to the internal surfaces through the conductive path provided by the electrolyte within (Figure 2-12a). The spatially-dependent potential of the internal surfaces was simulated  30 based on input parameters such as anode-cathode distance and electrolyte conductivity (Figure 2-12b). When compared to experimental results, the model estimated the cathode potential distribution with good accuracy, provided the experimental conditions resembled those presumed by the model.        Figure 2-12: (a) Schematic view of a cross-section of the internal CP system investigated in [130], where ra is the wire anode radius, rc is the hollow cylinder (cathode) radius, and d is the offset distance from the central axis (b) Steady-state potential measured at 180° in (a) as a function of d in electrolytes of different conductivities (triangle: 1% NaCl, square: 0.6% NaCl, diamond: 0.3% NaCl, and star: 0.15% NaCl) In soil systems, current distribution in a structure under CP is a function of system geometry, relative magnitudes of O2 diffusivity and ionic conductivity of the bulk soil, and reaction kinetics at electrode-soil interfaces. However, since soil is porous, tortuous mass-transfer within it obeys multi-phase flow laws which have critical dependence on air-filled void porosity, moisture content, and temperature [131]–[135]. These parameters are interrelated and depend on parameters such as soil structure, for both gas [136] and charge [137] transport. For evaluation of gas transport within the soil, previous numerical models of CP and corrosion systems often make narrow use of experimental diffusivity measurements like those of [134]. Rather, the rate of transport is determined through use of empirical parameters based on limiting current density such as in [130], [138], [139]. Seldom do models present coupled gas transport (of O2) and electrolyte potential distribution like in [140], [141], or variable gas diffusivity in bulk porous media like in [142] (yet without CP presence). For evaluation of a b  31 electrical current within the soil, the most rigorous technique is to employ Nernst-Planck based on concentrations, diffusivities, and electric mobilities of ionic species within the soil. Difficulties in quantifying such parameters across a wide range of soil types (with different ionic species and concentrations) make it more practical to simplify the method to a governing equation with a single soil conductivity parameter, used by most researchers [125], [129], [130], [138]–[143].  Reaction kinetics at electrode-electrolyte interfaces is a key determiner of final results in any corrosion model. Formerly, problems in obtaining reliable tabulated kinetic parameters have led to a wide range of values being used in studies. Yet, dependable parameter values or models can be obtained from experiments run in conditions simulating a specific system. Such an approach was presented by Riemer and Orazem [138] using polarization curve parameters for reaction kinetics at the corroding buried structure and real coating properties for CP current leakage through disbonded coating sections. The polarization curve used to represent the contributions of the anodic Fe oxidation reaction and the cathodic O2 reduction and H evolution reactions took the form of {E-2.4}, where 𝑖 is the net current density, 𝑖𝑙𝑖𝑚,𝑂2 is the current density for O2 reduction limited by mass transfer, 𝛽 is the Tafel slope (specific to each reaction), 𝛷 is the off-potential of the pipe, and 𝐸 is the effective equilibrium potential specific to each reaction, taking into account the influence of exchange current density. 𝑖 = 10𝛷−𝐸𝐹𝑒𝛽𝐹𝑒 − (1𝑖𝑙𝑖𝑚,𝑂2+ 10𝛷−𝐸𝑂2𝛽𝑂2 ) − 10−(𝛷−𝐸𝐻2)𝛽𝐻2    {E-2.4} Similarly, Muehlenkamp et al. [33] reported using parameter values for their modelled system from measurements presented in [143]. Nonetheless, since almost all examples of this approach involve parameter values being extracted from the experimental results based on the model intended to govern reaction kinetics in the simulation (e.g. Tafel), incongruities between the governing kinetics model and the actual experimental data are inevitable.   32  Models of the residual strength and integrity of corroded pipelines The corrosion of load-bearing metals in engineering applications leads to reduction of their structural integrity and eventually complete failure of the overall system. In the pipeline industry, dependable criteria utilized to evaluate the burst pressures and residual strengths of corroded pipeline are very valuable. The most famous of such criteria developed in the past is the ASME B31G code [144], which was later revised for over-conservatism to yield the 0.85dL method [145]. The B31G model for predicting the burst pressure (𝑃𝑏) of a corroded pipeline is evaluated through {E-2.5}, where 𝑡 is the wall thickness, 𝐷 is the outer diameter (𝐷 = 2𝑅, 𝑅: outer radius), 𝑑 is the maximum defect depth, 𝐿 is the defect length, and 𝜎𝑦 is the material yield stress at an offset strain of 0.5%. Another established code is Det Norske Veristas (DNV), evaluated by {E-2.6}, where 𝛾𝑚 is the partial safety factor for model prediction, 𝜎𝑈𝑇𝑆 is the material ultimate tensile strength, 𝛾𝑑  is the partial safety factor for corrosion depth, (𝑑/𝑡)𝑚𝑒𝑎𝑠 is the measured relative corrosion depth, 𝜀𝑑 is the factor defining fractile value for corrosion depth, and 𝑆𝑡𝐷 ∙ [𝑑/𝑡] is the standard deviation of (𝑑/𝑡)𝑚𝑒𝑎𝑠 based on the specification of the inspection tool. 𝑃𝑏,𝐵31𝐺 =2𝑡𝐷(1.1𝑆𝑦) [1−(23)(𝑑𝑡)1−(23)(𝑑𝑡)(𝑀𝑇−1)]   ; 𝑀𝑇 = √1 + 0.8𝐿2𝐷𝑡  {E-2.5} 𝑃𝑏,𝐷𝑁𝑉 = 𝛾𝑚2𝑡𝜎𝑈𝑇𝑆(1−𝛾𝑑(𝑑𝑡)∗)(𝐷−𝑡)(1−𝛾𝑑(𝑑𝑡)∗𝑄)  ;  𝑄 = √1 + 0.31 (𝐿√𝐷𝑡)2;  (𝑑𝑡)∗= (𝑑𝑡)𝑚𝑒𝑎𝑠+ 𝜀𝑑𝑆𝑡𝐷 [𝑑𝑡] {E-2.6} In these two codes, only the 𝐿 and 𝑑 dimensions are needed to define the corrosion defect. Addressing the simplifications/assumptions of such codes and creating enhanced and more realistic models for corroded pipelines is an active area of research. Netto et al. evaluated the residual strength of pipelines with single longitudinal corrosion defects through small-scale  33 experiments and finite element simulations [146]. Real defects in the physical pipeline were produced using a spark erosion process, adjustable for different defect sizes. Amongst the geometric parameters, the experimental 𝑃𝑏 upon internal pressurization was most affected by the maximum d. 𝑃𝑏 values predicted by the computer model were lower than experimental counterparts; thus, they could be used to obtain lower bound estimates of the response of pipes with similar defect sizes and shapes. In numerical investigations, real corrosion defect geometries are hard to replicate, so most models simplify defect geometries to semi-elliptical, semi-spherical, or constant depth contours. Pipeline material models include elasto-plastic, finitely deforming solids with isotropic hardening [146], elastic-plastic simple power hardening [147], and substitutive Bilinear Isotropic Hardening (BISO) material curves [148] adopted from data for grades such as API X60 and API X65 ([149] and [150], respectively). A simulated 𝑃𝑏 is the pressure which causes the Von-Mises equivalent stress (𝜎𝑉𝑀) at the bottom of the defect (i.e. maximum d) to equal the 𝜎𝑈𝑇𝑆 of the material.   Figure 2-13: Comparisons between experiments, the linear fit (i.e. {E-2.7}), and predictions from DNV and B31G codes [146]  34 The 𝐿 dimension of corrosion defects does not influence the 𝑃𝑏 values significantly, whereas 𝑑 does [146], [148]. Circumferential length (or width, 𝑐) of the defect has a tangible but minor influence on 𝑃𝑏 [146], [151] which increases with the growth of the 𝑐/𝐿 ratio [148] but is always less of an influence than that of d. Netto et al. used their results to develop a simplified formula, {E-2.7}, to rapidly evaluate structural integrity diminishment due to external corrosion. In this expression, 𝑃𝑏𝑖 is the burst pressure of an intact pipe. This formula correlates well with their experimental test results, and confirms conservatism of previous B31G and DNV codes (Figure 2-13).  𝑃𝑏𝑃𝑏𝑖= 1 − 0.9435 (𝑑𝑡)1.6(𝑙𝐷)0.4;  𝑐𝐷≥ 0.0785, 0.1 ≤𝑑𝑡≤ 0.8, and 𝑙𝐷≤ 1.5    {E-2.7} Choi et al. presented a three-dimensional (3D) elastic-plastic Finite Element Model (FEM) with a stress-based failure criterion used to simulate pipeline burst tests [151]. 𝜎𝑉𝑀 values at the defect are compared to different factored reference strength (S) values (yield strength: 𝑆𝑌, flow strength: 𝑆𝑓 = 1.1 ∙ 𝑆𝑌, and 𝜎𝑈𝑇𝑆). The best correlation with the experimental data is achieved with the 𝜎𝑈𝑇𝑆 reference. Regression analysis of the results yield a limit load expression as a function of the non-dimensional parameters √𝑅𝑡, 𝑅/𝑡, and 𝑑/𝑡. This produces slightly conservative estimates of excellent overall agreement with experimental results. Recently, Xu and Cheng studied the structural performance of X65, X80, and X100 also using FEM, comparing their results to the established codes B31G, RSTRENG, and DNV [152]. A non-linear elasto-plastic material model was used for all the steel grades. The 𝜎𝑉𝑀-based yield criterion for 𝑃𝑏 was less conservative than the codes, for 𝑑 values up to 40% of the wall thickness 𝑡. As d increases, the model results became more conservative. Since pipelines frequently experience external rigid body loads like bending or soil-induced strains, the authors investigated the effect of coupling the presence of a defect with a longitudinal applied pre-strain on the structure. 𝑃𝑏 of the corroded  35 pipe was reduced with applied tensile or compressive pre-strain, though tensile pre-strain was notably more detrimental.  Bedairi et al. recently investigated the hybrid pit-crack defect known to occur during external corrosion of pipelines [53]. The crack initiates from the base of the corroded region (Figure 2-14) simulating the initial stages of SCC, and is termed a Crack-in-Corrosion (CIC) defect. After material characterization determining strength and toughness values (tensile and Charpy V-notch testing, respectively), an elastic-plastic material model is used. For basic corrosion defects, a typical 𝜎𝑈𝑇𝑆-based criterion is used like most other investigations, whereas for crack defects, the criterion was based on energy: if the J-integral value in the vicinity of the crack reaches the critical fracture toughness of the material (𝐽0.2), the failure is predicted to occur by fracture. Simulation results were, on average, 17% more conservative than the experimental results obtained by the authors in previous models [153]. This conservatism increased with d, demonstrating the impact of crack profiles and dimensions on 𝑃𝑏. CIC is modeled as a double-profile defect, where both the crack and the corrosion defect are modelled as flat bottom uniform depth profiles (Figure 2-14b) to avoid meshing problems. The authors suggest further studies investigating different pipe geometries, material properties, and crack profiles.  Figure 2-14: (a) Experimental and FEM CIC defect profiles; (b) 3D simulation results for 60% wall thickness CIC defect model, showing spatially-dependent 𝜎𝑉𝑀 from a 5.59 MPa internal pressure [53] a b  36 3 3. Objectives In the multi-faceted review of relevant previous literature presented above, it is apparent that some environmental, material, and surface/interfacial aspects of HSLA steel corrosion in nn-pH soil environments are understudied or completely unexplored. Due to the consequential influence of HSLA steel grade on corrosion and mechanical behaviors, these less studied topics require dedicated investigations on each steel grade. Exploring knowledge gaps therein and contributing to a more complete understanding of corrosion mechanisms in nn-pH soil environments thus has unique novelty when conducted on modern HSLA steels such as API X100. Likewise, developing and improving numerical models of X100 corrosion in nn-pH soil environments based on dedicated laboratory experiments is of particular benefit to the corrosion science and pipeline integrity fields. The goal of this work is to contribute to a comprehensive understanding and modelling capability of the corrosion of X100 pipeline steel in nn-pH soil environments. Practically, the fulfillment of the key technical objectives outlined in the proceeding section (3.1) contributes towards the improvement of external pipeline corrosion control methods and design guidelines, especially CP standards. The results of this work highlight the specificity of environmental, material, and surface conditions on the external corrosion of buried pipelines. Simple standard testing environments and corrosion control guidelines do not adequately address the full spectrum of integrity-threating situations present in practice. A better understanding of the fundamental corrosion processes occurring across the full nn-pH range, coupled with enhanced simulation ability of the corrosion processes, can therefore increase the safety and reliability of buried pipeline infrastructure.   37 3.1 Key technical objectives {1} Study the effects of pH-altering electrolyte constituents and properties (HCO3−, dissolved  CO2, and temperature) on X100 corrosion behavior, in concentrations and combinations  with resultant nn-pH values in the 5 – 9 range. {2} Investigate the impacts of HCO3−, Cl−, SO42−, nitrate (NO3−), and temperature on  anodic/cathodic and passivation processes on X100 in nn-pH HCO3− electrolytes,  including the semiconductive properties of the passive layer(s). {3} Identify the specific role of dissolved O2 in the formation and evolution of iron- oxides/oxyhydroxides on X100 surfaces in nn-pH aqueous electrolytes, and assess the  inhibition effectiveness of an O2-scavenger chemical treatment to mitigate corrosion in  O2-containing nn-pH soil environments. {4} Evaluate hydrogen evolution, permeation, and diffusion behaviors for X100 steel in nn- pH HCO3− electrolytes, considering the impacts of HCO3− and dissolved CO2  concentration, temperature, and surface condition. {5} Develop a numerical model of physiochemical soil/electrolyte phenomena (CP current,  O2 diffusion, and heat transfer) to simulate the corrosion and structural integrity of buried  pipelines at coating failure sites.          38 4 4. Approach and methodology In order to achieve the proposed objectives listed in section 3.1, this work includes a comprehensive set of experiments coupled with numerical multiscale models and simulations. The electrochemical tests of corrosion, passivation, and hydrogen-related processes are designed to achieve objectives {1} to {4} separately. Fulfillment of objective {5} is achieved through the incorporation of laboratory results from previous objectives into numerical models developed on a commercial Multiphysics software package, as described in section 4.3 and chapter 9.  Each set of laboratory experiments is performed using a combination of the following electrochemical techniques: OCP, Linear Polarization Resistance (LPR), Potentiodynamic Polarization (PDP), Potentiostatic Polarization (PSP), EIS under OCP or PDP conditions, and Mott-Schottky. Optical Microscopy and Scanning Electron Microscopy (SEM) are used to determine the microstructure of the X100 specimen and the morphologies of corrosion products and passive layers. Evaluations of the chemical composition of the steel or the corrosion products forming under different test conditions is conducted using one or more of the following techniques: Energy-dispersive X-ray Spectroscopy (EDX/EDS), X-ray Diffraction (XRD), X-ray Photoelectron Spectroscopy (XPS), Inductively Coupled Plasma Mass Spectrometry (ICP-MS), and Raman Spectroscopy. Hydrogen permeation and diffusion tests are performed in a hydrogen permeation electrochemical setup [154] (also known as a Devanathan-Stachurski cell [155]). 4.1 Material and specimen preparation API X100 pipeline steel is the material used throughout this work. All the specimens tested in the laboratory studies of this work are cut from a larger steel section removed from an oil pipeline manufactured by Evraz Inc. The chemical composition of this HSLA steel section is evaluated with ICP-MS, the results of which are listed in Table 4-1 along with the corresponding carbon equivalent (CE), which is calculating using: CE = %C + %Mn/6 + %Cr/5 + (%Mo+%V)/4 + (%Cu+%Ni)/15.   39 Table 4-1: Chemical composition and carbon equivalent (CE) of API X100 steel used in laboratory tests Composition [wt%] C Mo Mn Al Ni Cu Ti Cr V Nb CE 0.1 0.19 1.66 0.02 0.13 0.25 0.02 0.016 0.003 0.043 0.45   Microstructural evaluation Microstructural evaluation of the X100 steel is performed using a Nikon EPIPHOT 300 series Optical Micorscope and ImageJ analysis software. Polishing is performed with 6 and 1 µm diamond suspension abrasives, then samples are etched with freshly prepared 2% nital (2 mL nitric acid (HNO3) and 98 mL ethanol (C2H6O)) or LePera solution [156] (1:1 ratio of 4 g picric acid (C6H3N3O7) in 100 ml C2H6O and 1 g sodium metabisulfite (Na2S2O5) in 100 ml deionized H2O, mixed immediately beforehand) to reveal ferrite or martensite-retained austenite (M-A) phases, respectively. Samples are etched for ~15 s in 2% nital or for ~35 s in LePera solution.   The formation of different phases in HSLA steels is reported to be reliant on three main factors: prior austenite grain size before cooling, the cooling rate during processing, and the state of Nb (i.e. in solution or precipitated) [157]. Grain boundaries are preferential sites for new phase nucleation, and since the ferrite nucleating on the austenite grain boundaries grows into grains through a diffusional transformation, cooling rate influences the process. From the micrograph of the X100 steel sample etched in 2% nital (Figure 4-1a), the presence of ferrite structures is evident in different forms as identified by arrows. In general, the majority of ferrite observed in the steel sample is irregular; featureless grains with non-smooth borders or, in many instances, needle-like ferrite formations are seen. Phase quantification of the ferrite was performed based on color contrast with bainitic regions, which also contain M-A, yielding a content around 45%. This result depends on the color intensity and shape criteria employed in the image analysis. Phase quantification was therefore done with a range of realistic intensity and shape criteria, and conducted on micrographs from other regions. The results yielded were generally within an acceptable ±10% range of the value reported here. Indeed, this 45% ferrite content compares favorably with similar HSLA pipeline steels like X80 in the as-received condition [158].   40   Figure 4-1: Microstructure of API X100 steel sample etched with: (a) 2% nital, with corresponding differentiation of ferrite and bainite phases according to ASTM E562-08; and (b) LePera solution; (c) graphical representation of M-A phase identification using ImageJ analysis software It is established that prior austenite grain boundaries cannot be observed for ferritic structures, whereas they are noticeable for bainitic structures. Bainitic regions in the steel microstructure are observed in Figure 4-1a, yet it is difficult to identify specific forms of bainite (aside from the granular form) or M-A phases in these regions with a 2% nital etchant. Instead, LePera etchant was used, with which M-A phases appear white, bainite appears black, ferrite appears tan, and grain boundaries are not strongly etched [156]. Figure 4-1b shows the results of  41 LePera etching and the corresponding Figure 4-1c shows isolation of the martensite phase for fractional content measurement using ImageJ. Based on the analysis of several captured micrographs after LePera etching, the M-A content of the sample is deemed to be within the 4-5% range. The remaining phase (i.e. bainite of all forms) constitutes around 50%. Both increased prior austenite grain size (which decreases the concentration of preferential nucleation sites at grain boundaries) and Nb presence will necessitate a higher driving force for phase transformations to happen, through increasing the transformation temperature. Hence, the increased M-A% of as-received X100 compared to weaker grades such as X80 is closely related to the higher Nb presence (0.043 vs. 0.034 wt. %, respectively [158]).   Preparation of specimens Laboratory specimens for all electrochemical tests are connected to wires using conductive silver (Ag) paste and then mounted in hard cold-curing epoxy resins. High-temperature epoxy resins are used for any test which involved temperatures ≥ 50°C. Only a flat surface is exposed to solutions in all tests. ASTM G1-03 standards are followed for chemical cleaning [159]. Before each experiment, specimens are sequentially wet-ground with 320, 600, and 1200 grit silicon carbide (SiC) papers and subsequently degreased ultrasonically in acetone (C3H6O) for 10 minutes. Then, they are rinsed in deionized H2O and dried in a stream of cool air.  For H permeation and diffusion tests, the X100 pipeline section is cut from the half-thickness plane to produce thin square sheets, with a side length of 15 mm. Both sides of the thin steel samples used for these experiments are ground with 800 grit SiC paper to obtain a final thickness of 1 mm. One side of the samples is further ground and polished with 1 µm diamond suspension for Palladium (Pd) coating. The choice of this sample thickness ensures a diffusion controlled regime during permeation experiments [160], [161]. This behavior is validated through calculations on preliminary results, but the details are not presented here. After sonication in ethanol for 5 min and then drying in air, the detection side (also referred to as oxidation or exit  42 side) of each sample is immediately coated with 99.99% Pd using a vapor deposition Edwards Coating System E306A [162]. In this coating system, the solid Pd source cube to be evaporated is placed in a filament crucible through which a large current is passed. The crucible and Pd coating metal are placed at the base of a sealed dome which is kept at a vacuum pressure. On the top end of this dome is the steel sample, with the detection side facing the crucible. As the Pd melts in the crucible then evaporates, a very thin film (i.e on the order of 10-9 m) is deposited uniformly on the exposed surface of the steel sample. The coated samples are then carefully removed from the machine, cleaned, and stored appropriately in a desiccator until tested (see section 4.2.3). 4.2 Test environments and methods Laboratory tests in this study are conducted in nn-pH HCO3−-based solutions between pH 5 – 9, containing different concentrations of Cl−, SO42−, NO3−, Na+, K+, Ca2+, Mg2+, dissolved CO2, dissolved O2, and temperature. The specifics of each test environment depends on the corrosion and passivation aspects being investigated, and will be described in each corresponding chapter separately. Moreover, test routines are designed with various electrochemical procedures, microscopy methods, and chemical characterization techniques to achieve the key technical objectives outlined in section 3.1. Each particular investigation in this study will utilize a combination of at least three of these various methods, as will be shown in each chapter accordingly. A general overview of the parameters and devices used in the electrochemical, microscopy, and chemical characterization techniques is described in sections 4.2.2 – 4.2.4.   Standard NS4 solution and variants Generally, a baseline reference solution is used for all the environmental conditions studied here, namely NS4 at 25 °C purged with 5% CO2/95% N2. NS4 solution has the following ionic composition in M [162]: 5.75 x 10-3 HCO3−, 5.75 x 10-3 Na+, 4.10 x 10-3 Cl−, 1.64 x 10-3 K+, 1.23 x 10-3 Ca2+, 5.32 x 10-4 SO42−, and 5.32 x 10-4 Mg2+. In this standard form, NS4 is one simulation of the nn-pH electrolyte typically found under disbonded pipeline coatings, as  43 determined through chemical analysis of such electrolytes from real coating disbondments in the field [13]. From this baseline standard reference, more alkaline or more acidic HCO3−-based conditions are instigated by changing HCO3− content, purging gas, and temperature. Increased CO2 dissolves in the solution as H2CO3, with a concentration dependent on the partial pressure of CO2. This reduces solution pH by shifting the H2CO3/HCO3− equilibrium of {R-4.2} in the solution. HCO3− concentration has a direct influence on this equilibrium – added HCO3− increases the pH by pushing {R-4.2} to the left. CO2 (aq. ) + H2O ↔  H2CO3 (aq. )    {R-4.1} H2CO3 (aq. ) ↔  H+ (aq. ) + HCO3− (aq. )    {R-4.2} The varying of HCO3− content, purging gas, and temperature simulates the wide diversity of nn-pH electrolytes and addresses the need to study different environmental parameters within the nn-pH range, arising from the concentration cell effect in the disbondment crevice. Cl− and SO42− additions to solutions are also studied, although they do not change their corresponding solution’s pH. All environments throughout this work are created using double-distilled deionized H2O, analytic grade reagents, and Praxair® 99.99% purity purging gas or gas mixtures.  Electrochemical test methods for corrosion and passivation processes Electrochemical experiments conducted to investigate corrosion and passivation processes in this study are performed in a glass jacket test cell with a total volume of 0.6 L. Extended immersion experiments are performed either in the same glass jacket test cell or in individual glass containers of a smaller 0.2 ml volume. The temperature of the solution in the glass jacket test cell is regulated by connecting the cell to a circulation water heater equipped with an accurate digital controller. In immersion tests the proper solution temperature is achieved by placing the cells on a digitally controlled hot-plate. In both cases, temperatures are periodically verified using a thermometer inserted in the cells. The cells are sealed from the external atmosphere except for a vent in the gas purger to prevent pressure buildup. Experiments are left  44 open to air or purged with some form of gas or gas mixture before specimen immersion until test completion.  A conventional three-electrode setup is used for electrochemical measurements. The working electrode in the tests is the studied X100 specimen and the counter electrode or auxiliary electrode is a slender graphite rod or a platinum wire mesh. The reference electrode used is a SCE of +0.241 VSHE, isolated in a salt bridge which is electrochemically in contact with the working electrode through a Luggin capillary tube with a low leak-rate Vycor frit. The potentiostats used are either Princeton Applied Research Versastat machines controlled by the accompanying VersaStudio software, or Gamry Reference 600 machines controlled by the accompanying Gamry Electrochemical Suite software package. Figure 4-2 shows a schematic of the experimental setup used to acquire all OCP, LPR, PDP, PSP, EIS under OCP or PDP conditions, and Mott-Schottky results presented throughout this dissertation.  Figure 4-2: Schematic of experimental setup for electrochemical tests of corrosion and passivation [162] Electrochemical tests are performed after the proper solution temperature is verified (±1 ̊C) and the pH of the solution reaches a stable state under the effect of gas purging, if present (maximum change of ±0.2 pH during 15 minutes). To reach this condition, at least 30 minutes of  45 pre-immersion gas purging is normally required for all tests. Immediately following specimen immersion, a potentiostatic voltage of -1.5 VSCE is always applied for 1 minute in order to remove any air-oxide film on the steel surface. Each set of tests designed to achieve a specific objective in the overall thesis comprises of at least two electrochemical techniques, for confirmation of measured behaviors. The following is a brief description of the parameters used for each technique:  OCP testing immediately follows immersion and the short cathodic cleaning routine. Specimen are held at OCP until a stable profile is achieved, characterized by fluctuations no greater than ±0.1 mV/s. This is often achieved within 30 minutes to 1 h of immersion. For extended immersion tests, specimens are held at OCP for 24 h, within which changes in profiles are observed mainly due to corrosion product development on the surface.   PDP testing is carried out following stable early stages of OCP, between cathodic and anodic potentials chosen for the specific processes of interest, described in subsections of each chapter accordingly. Cathodic potentials in the PDP tests are not less than -2.0 VSCE, and anodic potentials are not greater than 1.6 VSCE. PDP sweeps are carried out at a standard scan rate of 1/6 mV/s [163], 0.5 mV/s, or 1 mV/s deemed suitable for revealing the vital corrosion kinetics features of the system, since previous studies of pipeline steel in HCO3− solutions have appropriately used them (1 mV/s upper limit [164]).  LPR tests are performed to extract Rp and corrosion rate information, within ±10 mV or ±20 mV of OCP depending on the specific test. These quick sweeps are always conducted at a standard scan rate of 1/6 mV/s [165] and repeated three times to ensure reproducibility – the corresponding results are averaged and confidence intervals are presented to ensure statistical relevance.   EIS is utilized extensively in this study to evaluate interfacial processes [166] and surface interactions on unpolarized specimen after 1 h immersions and during 24 h extended  46 immersions, and on polarized specimen in a dynamic EIS routine [167], [168] employed to study anodic and cathodic processes. For EIS tests in this study, the AC disturbance signal is 10 mV, the measurement frequency (ω) range is between 0.01 Hz and 20 kHz (or 0.1 Hz – 10 kHz for some tests), and the sampling rate is 10 points/decade. All EIS results are analyzed and fitted to electrochemical equivalent circuits (EEC) using the ZSimpWin software.  PSP is performed for the tests in section 6.3 to grow passive corrosion product layers for EIS and Mott-Schottky evaluations. Nonetheless, in this PSP or anodizing step conducted at 0.5 VSCE for 1 h, icorr decay behaviors are recorded and used to corroborate the results interpreted from other techniques.  Mott-Schottky tests are performed to study the influence of HCO3−, temperature, and pH on the semiconductive properties of passive corrosion product layers following anodizing. Mott-Schottky tests in this study are scanned between -0.5 VSCE and 1 VSCE at a frequency of 1 kHz and with a step height of 20 mV.  Electrochemical test method for hydrogen permeation and diffusion  Tests of H permeation and diffusion in the thin X100 specimens prepared specifically for this purpose (see section 4.1.2) are performed in a H permeation electrochemical setup [154] (also known as a Devanathan-Stachurski cell [155]). Figure 4-3 shows a general schematic of the setup. In the H permeation and diffusion experiments of this study however, the charging cell (entry compartment) and oxidation cell (exit compartment) are of the same volume, and both reference electrodes are inserted within Luggin capillary tubes with frits close to their corresponding sides of the specimen. The following points, as mentioned in the ASTM G148 standard [154], summarize the hydrogen permeation and diffusion procedure in this apparatus. Further details on the electrochemical apparatus and methods of studying H diffusion, permeation, and solubility in metals can be found in [154], [169].     47  The metal membrane (i.e. specimen) of interest is inserted between the hydrogen charging and oxidation cells. The charging cell contains the environment of interest. H atoms are generated on the membrane surface exposed to this environment. In this work, these atoms are the result of corrosion processes on the steel, namely cathodic reduction reactions.  Some of the H atoms generated on the charging side diffuse through the membrane and are then oxidized on exiting from the other side of the metal in the oxidation cell.  The conditions (for example, environment and the electrode potential) on the oxidation side of the membrane are controlled so that the metal surface is either passive or immune to corrosion. Additional material (usually Pd), either plated or sputter deposited/coated onto or clamped against the specimen on the oxidation side, may be used to achieve this provided that it is demonstrated that the introduction of this additional interface has no effect on the calculated diffusivity. The background current established under these conditions prior to H transport should be relatively constant and small compared to that of the H atom oxidation current.  The electrode potential of the specimen in the oxidation cell is controlled at a value sufficiently positive to ensure that the kinetics of oxidation of hydrogen atoms are limited by the flux of hydrogen atoms (i.e. the oxidation current density is diffusion limited).  The total oxidation current is monitored as a function of time. The total oxidation current comprises the background current and the current resulting from oxidation of hydrogen atoms. The latter is the permeation current. In the present study, the Pd coated X100 steel sample (as described in section 4.1.2) was mounted between the hydrogen charging and oxidation cells to expose a circular region of approximately 1 cm2 area to the solution on each side. The experiment is commenced by filling the Pd coated side (exit or oxidation) compartment with deaerated (100% N2) 0.01 M NaOH  48 solution (pH 12), followed by rapidly applying a potential of 140 mVSCE (1.09 V more positive than the H+/H2 reversible potential at pH 12) and recording the background current density (ibg). The applied potential on the Pd coated side ensures fast kinetics of the H oxidation reaction and effectively maintain a H atom concentration of near zero at the Pd coated surface. In each test, after the current density on the exit side reaches a stable value of approximately 70 nA/cm2, the other compartment (entry or charging) is filled with the deaerated NS4 solution being tested.    Figure 4-3: Electrochemical H permeation and diffusion cell: (a) components and assembly (b) measuring apparatus and settings (with two potentiostat instruments or two channels of one multistat)  [154] a b One of the following 3 settings:  Galvanostatic charging at  -500 µA/cm2  Potentiostatic charging at  -0.9 VSCE  Step-wise potentiostatic charging from -0.85 VSCE to  -1.15 VSCE at -0.1 V/4 h, then OCP, then finally a complete drain (open to air)   NS4 solution or variant   0.01 NaOH  (pH 12)   Setting: Potentiostatic oxidation at 0.14 VSCE   Potentiostat 1  or  Multistat channel 1  Potentiostat 2 or Multistat channel 2 Exit side   Entry side    49 In the entry compartment, H charging is performed under several different conditions. In the first set of experiments, a galvanostatic charging current density of -500 µA/cm2 is applied to the uncoated face of the sample for up to 10 hours. In the second set of experiments, a constant cathodic potential of -0.9 VSCE is applied during which the purging gas of the charging solution is switched from 100% N2 to 5% CO2, and then subsequently switched to 100% CO2. In the third set of experiments, the surface potential of the steel sample on the entry side is changed by first applying cathodic potentials varying from -0.85 to -1.15 VSCE with a step size of -0.1 V per step (4 h), then leaving the sample at the OCP, and finally draining the solution in the entry compartment to leave the sample exposed to air.   Microscopy and chemical characterization methods Morphologies of the corrosion products formed on specimens following laboratory tests are observed using SEM. The SEM images presented in this study are from either a Hitachi S570 model or a Hitachi S3000N Variable Pressure model, both of which use a conventional tungsten hairpin electron gun. All EDX/EDS spectra presented here are obtained using the EDX/EDS capability in the Hitachi 3000N Variable Pressure SEM. XRD characterization results presented throughout this study are obtained using a Rigaku MultiFlex machine with a 2kW x-ray generator, from start angles ≥ 3° to a stop angle of 90°. For some XRD routines, corrosion products are scraped off with a clean razor onto a 0-diffraction quartz plate. The x-ray tube Cu Kα settings are 40 kV and 20 mA, and the scan speed is a very slow 0.125˚ min-1 to minimize noise. XPS characterization is performed on an Omicron & Leybold MAX200 machine with a monochromatic Al Kα X-ray source at 10 kV and 20 mA. For the XPS measurements of this thesis, the system pressure is 2 x 10-7 Pascal (Pa). Raman Spectroscopy is conducted on a Horiba LabRAM HR Raman spectrometer with a 633 nm HeNe laser and a Pelletier-based cooled CCD Si detector. The Raman spectrometer is fitted with an Olympus microscope with a 100x magnification objective lens and has a 1200 line/mm diffraction grating capability.  50 4.3 Modelling geometries and meshes This study presents simulation results from two models of the external corrosion of buried pipelines. The first model is at a macroscale, on the order of meters, simulating physicochemical soil phenomena (i.e. heat transfer, CP, and gas diffusion) in soils of various structures and moisture contents, and their effect on corrosion processes at a coating failure site on a buried pipeline. The second model is at a mm-scale in the trapped water region beneath a disbonded external coating of a pipeline. The second model incorporates both a corrosion and a structural stress analysis module – corrosion defect sizes simulated by the former affect the structural integrity of the pipeline, the extent of which is simulated by the latter module. For the macro m-scale model of pipeline corrosion, and the corrosion module of the mm-scale model beneath a disbonded coating, Comsol Multiphysics® version 4.3a is the modelling software used. The stress analysis module of the second model is developed in ANSYS®. This dissertation will focus more on the first model and the corrosion module of the second model, whereas the development and results from the structural model will only be presented and discussed briefly. In all of the models/modules developed here, the Finite Element Method (FEM) is the numerical technique employed. The presence of spatially varying governing properties in the modelled systems (e.g. diffusivity in soil) is a key reason why FEM is favored over other techniques such as the Finite Difference Method (FDM) or the Boundary Element Method (BEM). FDM has inadequate resolution capabilities & difficulties in handling irregular meshes, moving meshes, and nonlinear effects, while BEM cannot handle the spatially-varying properties of the electrolyte media modelled here.       51  Figure 4-4: (a) 3D representation of buried pipeline with CP anodes (right), and axis of model cross-section (left); (b) 2D representation of model geometry and dimensions A 3D representation of a sectioned buried CP-pipeline system is shown in Figure 4-4a. A cross-section of this structure is taken at plane a-a resulting in the 2D section shown in Figure 4-4b, which is the basis of the m-scale model in this thesis. The reduction of the 3D geometry to a 2D one permits the use of 2D plane elements and significantly simplifies the model. Consequences of this simplification on simulation results are discussed in chapter 9. Dimensions in the model are the typical values for onshore transmission pipelines [13] in excavated ditches [170]. A 90° arc of the steel is exposed to the corrosive soil environment shown between A and B in Figure 4-4a, representing a site where the protective coating completely  52 deteriorated. Although the overall size of this defect is larger than would normally be present on a buried pipeline, it is modelled in this way here to identify the varying potential, anodic/cathodic current density, and O2 concentration distributions at different angles θ. In practice, a single simulation result from the present study can be used to evaluate the disparity between CP and corrosion at 0° ≤ θ ≤ 90°, instead of running separate simulations for each location. Although the influence of localized corrosion is diminished with this approach, the model maintains the ability to investigate localized effects if desired by reducing θ.  A cross-section of the 3D pipeline representation in Figure 4-4a taken on the longitudinal y-z plane results in the 2D section shown in Figure 4-5. The basis of the corrosion module of the mm-scale model is the zoomed portion of this figure, with the assigned dimensions of a suggested coating disbondment in section A-A. CP current availability at the disbondment opening is assumed to be sufficient enough for proper protection (i.e. -0.85 VCu-CuSO4 or -0.77 VSCE according to the NACE SP0169 standard), while simulating CP shielding and the resulting growth of a corrosion defect at the exposed steel surface. Pipeline wall thickness (t) and coating disbondment width (w) and length (L) dimensions simulated are normal values for standard 36” buried transmission pipelines [35].  Discretization versatility of the FEM is exploited by solving 5 sets of mesh resolutions for the m-scale model, and 3 sets of mesh resolutions for the mm-scale model for all simulated parameters. Field variables within each element are linear throughout this thesis. For the m-scale model, all the meshes are discretized more finely around electrode-soil interfaces and the ground surface boundary for enhanced identification of critical phenomena occurring between the CP anode, the ground surface, and the pipeline’s exposed surface. Similarly, all the meshes in the mm-scale model are made finer at the exposed steel surface (electrolyte-electrode interface) for more accurate simulation of the shape and growth of the corrosion defect. Mesh parameter values for the various mesh resolutions are listed in Table 4-2 for the m-scale model of Figure 4-4b. The equivalent mesh details for the mm-scale model can be found in Table E-1 in Appendix E. The  53 adequacy of the mesh sizes utilized for both models is ensured through suitable convergence in the mesh sensitivity analyses of the results (see section 9.5).  Figure 4-5: 2D cross-section of pipeline in longitudinal plane with coating defect (zoomed section), dimensions, and mesh of trapped water region Table 4-2: Maximum and minimum element size, maximum element growth rate, and resolution of curvature values for extra coarse to extra fine mesh resolutions, within the soil domain vs. at electrode/ground boundaries in Figure 4-4b Mesh parameters  Mesh resolution Extra coarse Coarse Normal Fine Extra fine   In general domain Maximum/minimum element size 0.402 m/ 1.8 x 10-3 m 0.318 m/ 1.8 x 10-3 m 0.222 m/ 7.5 x 10-4 m 0.120 m / 4.5 x 10-4 m 0.060 m/ 1.2 x 10-4 m Maximum element growth rate 1.30 m 1.30 m 1.25 m 1.20 m 1.10 m Resolution of curvature 0.30 0.30 0.25 0.25 0.20   At critical boundaries Maximum element size 0.318 m/ 1.8 x 10-3 m 0.222 m/ 7.5 x 10-4 m 0.120 m / 4.5 x 10-4 m 0.060 m/ 1.2 x 10-4 m  Same as general domain Maximum element growth rate 1.30 m 1.25 m 1.20 m 1.10 m Resolution of curvature 0.30 0.25 0.25 0.20  54 5 5. Electrochemical behavior of X100 pipeline steel in deaerated 𝐇𝐂𝐎𝟑− solutions of near-neutral, mildly acidic, or mildly alkaline pH2,3 External pipeline corrosion and SCC of the nn-pH form occurs in HCO3−-CO32− trapped water containing small quantities of Cl− and SO42−, with a pH in the range from 5.5 to 8.5 [171]. The role of this local environment on corrosion and the subsequent initiation of cracking in pipeline steels, especially newer HSLA grades, is not completely clear. In specific, an understanding of the effect of changes in pH-altering environmental variables such as H-based electrolyte constituents, %CO2 in dissolved gases, and temperature on the electrochemical behavior of pipelines steels is lacking. These parameters are coupled in a way that variations in one immediately cause changes in others according to the equilibrium reactions {R-4.1} and {R-4.2} shown in the previous chapter. In addition, the effect of these pH-altering variables on electrochemical behavior appears to be interrelated [64], [65].  The objective of this chapter is to evaluate the influence of combinations of the abovementioned environmental variables on the corrosion behavior of API X100 pipeline steel. The overall pH of the environments tested is kept within or close to the established nn-pH range of 5.5 to 8.5 [13]. A nn-pH solution (NS4 deaerated with a 5% CO2/95% N2 mixture) is used as a reference, and environmental variations are introduced to this condition by changing: ion content (HCO3−, Cl−, and SO42−), purging gas (5% CO2/95% N2, 100% CO2, and 100% N2), and temperature (25, 40, and 55 °C). These three parameters determine the pH of the condition being tested, as shown in Figure 5-1 for a subset of the presented results. An environmental condition matrix stems from different combinations of these variables. Electrochemical kinetics and passivation behavior of the steel are assessed for each condition therein using OCP, LPR, PDP,                                                  2  I. M. Gadala and A. Alfantazi, Corrosion Science, vol. 82, pp. 45–57, May 2014. 3  I. M. Gadala and A. Alfantazi, Metallurgical and Materials Transactions A, vol. 46, no. 7, pp.  3104–3116, Apr. 2015.  55 and EIS measurements. This chapter is based on two papers [74], [172] which were published as part of the research work leading towards this PhD thesis.  Figure 5-1: Effect of HCO3−, purging environment, and temperature on solution pH 5.1 OCP and LPR measurements The free corrosion potentials (OCP) of the specimens are observed up to about 5000 s to ensure fluctuations are negligible. The OCPs are all below -710 mVSCE and in the anaerobic range [24], consistent with fact that all solutions are deaerated by purging some form of gas or gas mixture. The OCP values for the highest pH solutions are the least noble. As shown in Figure 5-2, OCP values decrease linearly with pH at all temperatures. Each line in this plot represents a single purging environment and temperature. Data points plotted on each line represent the different [HCO3−] investigated (i.e. 48.3, 438, or 4830 mg/L NaHCO3). Based on the HCO3−, dissolved CO2, and pH relations discussed in 4.2.1, the trend in Figure 5-2 suggests an increase in the active corrosion of the steel with decreased HCO3− content and/or increased CO2 concentration. A consistent correlation between temperature and OCP is not clear. However, it is observed that in more acidic conditions within any single purging environment, higher temperatures increase OCP, and vice versa. This effect is noticed even though the pH of higher temperature solution becomes slightly more alkaline. The trend of decreasing OCP values with pH can be explained by the decrease in cathodic kinetics in {R-5.2} at higher pH.  56   Figure 5-2: Open circuit potentials vs. pH in all solutions free of Cl− and/or SO42− additions (varying HCO3−, purging gas, and temperature) 100% CO2 only 5% CO2/95% N2 only 100% N2  only Overlap region  57 Fe →  Fe2+ + 2e−      {R-5.1} 2H+ + 2e−  →  H2      {R-5.2} HCO3− + e−  →  CO32− +12H2     {R-5.3} 2H2O + 2e−  →  H2 +  2OH−     {R-5.4} In addition to {R-5.2}, the simultaneous discharge of HCO3− in {R-5.3} can be present with sufficient HCO3− content, though it is thermodynamically less favorable than H evolution at lower pH [100]. Larger drops in potential with increased HCO3− observed in mildly alkaline and nn-pH conditions compared to lower pH conditions is most likely due to {R-5.3} and {R-5.4} becoming more dominant.  Analysis of the “overlap region” between the 100% CO2 and 5% CO2/95% N2 purged data in Figure 5-2 advocates that corrosion activity of the steel cannot be determined qualitatively based on the solution pH only. Data points in this region exhibit diverse OCP values although their corresponding solutions possess nearly identical pH. Similarly, a number of points have identical OCPs, such as at the -740 mV or -735 mV mark, although the pH levels of these points vary significantly. The region where this discrepancy occurs in Figure 5-2 is the overlap of data from the HCO3−-rich solutions purged with 100% CO2 and the NS4- or below NS4-level HCO3− solutions purged with 5% CO2/95% N2. Based on this, it is deduced that quantitative comparisons of corrosion activity in different environments using the pH dependence approach should only be performed in the same purging environment. Other purging environments can be included only if there exists significantly large pH variations (±1.5 or more) in the solutions.   58  Figure 5-3: Corrosion activity vs. pH based on polarization resistance values of select data LPR results illustrated in Figure 5-3 are consistent with the OCP findings. The plotted data set in Figure 5-3 was selected from various purging and solution HCO3− conditions to observe the overall trend with respect to pH. Between the high pH solutions at 9 (100% N2 purged NS4 + 10x HCO3−) and the low pH solutions at 4.7 (low HCO3− content purged with 100% CO2), an overall decrease in Rp values is observed. Corrosion activity measured through icorr is inversely related to Rp, implicating that corrosion activity increases at lower pH. Yet, for small pH variations, specifically at neutral or nn-pH, such a tendency cannot be precisely discerned. This corroborates the large pH variation requirement (±1.5 pH or more) mentioned earlier. This central sensitivity to pH variations is extended even further with results presented in [74], which reveal that not only is the pH change itself important, but the cause of the change (e.g. different HCO3− content vs. different %CO2 purging environment) can actually yield opposing influences.  5.2 PDP testing and parameter variations based on E-pH  The PDP profiles for the environments shown in Figure 5-1 manifest different electrochemical behaviors, which are divided here into two groups according to active vs. passive  59 (and passive-like) responses. Only the latter are shown and discussed in detail in this chapter. Profiles manifesting active responses can be seen in Appendix A.  Passive and passive-like PDP responses All solutions with HCO3− levels 10x the NS4 baseline exhibit passive or passive-like behavior, regardless of the pH (Figure 5-4 to Figure 5-6). The pH of the conditions for the PDP profiles illustrated in Figure 5-4 to Figure 5-6 range from alkaline (8.9) for the 100% N2 purged NS4 + 10x HCO3−, to just below neutral (6.4) for the 100% CO2 purged NS4 + 10x HCO3−, both at 25 °C. This ascertains that passive film formation is possible on HSLA steel in nn-pH conditions, provided that sufficient HCO3− is present in the solution. This is contrary to what has been reported previously in [99] and [173] that no stable passive film formation is possible on carbon steel in nn-pH solutions. This is because the conditions investigated in [99] are both low in HCO3− (directly added or dissociated from purging gas) and possess a nn-pH. In contrast, here some of the conditions are nn in pH yet contain considerably more amounts of HCO3−, both added directly and dissociated from the purging gas. Backed by the findings in [174] and [175], it could be concluded that the increased HCO3− content is thus what causes passivation. In [174], addition of HCO3− was beneficial to both the formation of a passive film and the inhibition of pitting of X70 micro-alloyed steel. The results of [175] demonstrate that the anodic polarization curves of Fe in 0.01-1.0 mol L-1 HCO3− solutions have a typical active-passive-transpassive characteristic, where the form and number of the anodic peaks varies with HCO3− concentration. The polarization results of this investigation show that this remains true even for micro-alloyed X100 steel in the presence of small amounts of Cl− and SO42−.  It should be noted that the concentration of HCO3− ions originating from the dissociation of dissolved gas in the solution is not trivial. In the work of Xie et al. [176], this concentration was calculated and it affected the H adsorption and ductility of X-65 in stress tests. In this study, calculations of HCO3− concentrations from the dissociation of dissolved gas in the solution  60 (assuming Henry’s Law) reveal that purging 5% CO2/95% N2 adds 0.0017 mol HCO3− to the solution. This is the equivalent of 0.1037 g NaHCO3. Purging 100% CO2 is equivalent to upwards of 2 g NaHCO3 added. These HCO3− ion concentrations have a tangible effect on the polarization results, specifically the formation of passive films and their stability.   Figure 5-4: PDP in 5% CO2/95% N2 purged NS4 + 10x HCO3− at 25, 40, and 55 °C (pH 7.6, 7.9, and 8.1, respectively) The PDP profiles for 5% CO2/95% N2 purged NS4 + 10x HCO3− at 25, 40, and 55 °C are shown in Figure 5-4. The anodic branches in Figure 5-4 show clear signs of multi-step dissolution and passive film formation for the higher pH conditions. At potentials between around -500 and -650 mVSCE a decrease in current densities is observed in the 100% N2 purged condition at 25 °C and the 5% CO2/95% N2 purged condition at 40 and 55 °C. The anodic peak potentials at which this deceleration occurs increases with decreased solution pH, indicating the likely involvement of OH−. This involvement is in the form of a defective OH−-based hydrous film which decelerates the current densities transitorily [177]. This film has previously been shown to develop within a narrow potential range [178] and follows the ensuing formation sequence:  61 Fe + OH− ↔ FeOH−ads  →  FeOHads + e−   {R-5.6} FeOHads  →  FeOH+ads + e−      {R-5.7} FeOH+ads + OH−  →  Fe(OH)2     {R-5.8} The 25 °C profile exhibits subtle early signs of this iron hydroxide (Fe(OH)2) film formation at around -500 mVSCE, yet does not fully materialize like the others. It is highly suspected that this is due to insufficient OH− content or solution alkalinity. This suggests that the threshold pH for observable formation of hydrous Fe(OH)2 in the environments investigated herein is around 7.6. This threshold is dependent on other environmental parameters like temperature and would change for conditions different than the ones explored here. In the profiles which experienced significant Fe(OH)2 formation, an increase in current density is observed above the -500 and -650 mVSCE potential range where Fe(OH)2 formed. This is evidence of the Fe(OH)2 removal process described by {R-5.9} [99]: Fe(OH)2 + HCO3−  → CO32− + OH− + Fe2+ + H2O   {R-5.9} The initiation of this removal process is identified by the current density inflection point, which seems to occur fastest in the 5% CO2/95% N2 purged condition at 55 °C. This indicates an influence of HCO3− content, highest in this condition, on the removal process. Nonetheless, for all three cases this inflection point occurs at potentials 150 mV higher than the potential of first current density decrease (Fe(OH)2 formation). Alternatively, at this potential it has been reported that FeCO3 may start to form in a double-layered film with Fe(OH)2 [179]. It is doubted that this occurs in the conditions investigated here, since FeCO3 possesses better protection properties than Fe(OH)2 and would thus limit the accelerated current densities observed in this potential range.  Instead, at a potential of approximately -300 mVSCE, independent of purging gas and temperature, the formation of a single-layer FeCO3 film takes place. This is confirmed by upcoming SEM images and chemical characterization spectra (section 5.2.2). In the profiles which previously experienced the formation of Fe(OH)2, the FeCO3 film does not fully form  62 (100% N2 purged condition) or is unstable (5% CO2/95% N2 purged condition at 40 or 55 °C). In the 100% N2 purged condition, the FeCO3 film does not fully form due to the low HCO3− content from the lack of CO2 in the purging gas, greatly decelerating the reaction sequence of {R-5.10} and {R-5.11} [180]:  Fe + HCO3−  ↔ [FeHCO3−]ads → [FeHCO3]ads  + e−  {R-5.10} [FeHCO3]ads   + OH− → FeCO3 + H2O + e−      {R-5.11} The 5% CO2/95% N2 purged conditions at 40 or 55 °C have higher HCO3−  concentrations than the 100% N2 purged condition, yet these HCO3− concentrations are suspected to have reduced from the Fe(OH)2 removal process of {R-5.9} leaving less HCO3− for the FeCO3 formation process. As such, these two conditions exhibit passive-like behaviors which do not reach the level of stability or low current density of 5% CO2/95% N2 purged condition at 25 °C.   Figure 5-5: PDP profiles in 100% CO2 purged NS4 + 10x HCO3−  at 25, 40, and 55 °C (pH 6.4, 6.6, and 6.7, respectively)  63 In the 100% CO2 purged NS4 + 10x HCO3− conditions, it is clear that hydrous Fe(OH)2 does not form due to insufficient alkalinity or OH− involvement in all cases (Figure 5-5). A stable FeCO3 film is formed at all temperatures, with a passive current density (𝑖𝑝) relatively equal to one another in all cases. The 𝑖𝑝 in Figure 5-5 is the same as that of the 5% CO2/95% N2 purged condition at 25 °C in Figure 5-4. Between the different temperatures in Figure 5-5, evident delay in the onset of passivation is witnessed, signifying a role of temperature (or pH) in the formation of FeCO3 in nn-pH conditions where Fe(OH)2 does not form. This finding is corroborated in the results presented in the upcoming chapter 6.6 below  Figure 5-6: PDP in 5% CO2/95% N2 purged NS4 + 10x HCO3− at 25 °C with various chloride and/or sulfate ion content (pH 7.6 for all) The addition of Cl− and SO42− to conditions which are found to cause passivation induced visible kinetic changes. As shown in Figure 5-6, compared to the reference 5% CO2/95% N2 purged NS4 + 10x HCO3− condition at 25 °C, increased current densities are observed especially at higher potentials. The addition of 0.5 wt% Cl− prevents passivation altogether as its anodic profile appears active with no current density decreases. Sufficient Cl− concentration in solutions  64 is known to prevent passivation [181], [182] and pre-passivation steps on Fe and steels, as recently reported in [183]. Even in concentrations which do not completely prevent passivation, Cl− has been shown to accelerate anodic reactions on the corroding iron and steel surfaces [184]. This is observed in the 0.1 wt% Cl− profile when compared to the reference condition of Figure 5-6. The addition of an identical amount of SO42− instead has a similar effect, yet the severity is not as high as the Cl−. On the other hand, when Cl− and SO42− are added together, a compound effect is observed in which the anodic profile of the PDP profile becomes almost completely active and the passivation of the steel surface is practically eliminated. From Figure 5-6, it can be deduced that the aggressiveness ranking of anion attack on passive layers in nn-pH environments surrounding X100 pipeline steel is Cl− > SO42− > HCO3−. Also, a compound effect of Cl− and SO42− on passive layer degeneration in nn-pH conditions is established. In the cathodic region though, Cl− and/or SO42− additions garnered no discernible differences, and HCO3− retains its dominating influence on the cathodic processes. Similarly, added Cl− and/or SO42− has no practical effect on corrosion rates in the unpassivated states at OCP when extracted from Figure 5-6, although subtle effects on interfacial processes at OCP are detected as reported in the EIS results of section 5.3.   Anodic parameter relationships with E-pH Based on the consequential impacts of pH and HCO3− concentration on corrosion and passivation processes as shown above, it is deemed important to further investigate changes in the onset and the development of passive layer(s) on X100 as a function of pH. Since the electrochemical technique used here is PDP, this also involves electrochemical potential (E). Changes to the pH of the environments in this analysis are instigated through added HCO3−; unlike the approach of the previous section (5.2.1), temperature is maintained at 25 °C and the purging gas is kept consistent as 5% CO2/95% N2. The baseline NS4 environment of pH 6.7 and 0.00575 M HCO3− does not induce any clear passivation in laboratory PDP tests, hence the pH range for  65 the following analysis is between a mildly alkaline 7.5 and 8.9 instead. The influence of Cl− and SO42− presence, in the concentrations found in the standard NS4 condition, and PDP scan rate are explored.   Figure 5-7: 1 mV/s PDP profiles of specimen in NS4 Cl−- and SO42−-containing solutions with 7.19 < pH < 8.85 In Figure 5-7, E and current density (i) parameters used for analyzing PDP profiles are identified: Ep1 and Ep2 are the first and second potentials at which sustained current density decreases occur, respectively; Ebd1 and Ebd2 are the first and second potentials at which sustained breakdown or transpassivation occurs, respectively; and ip is the lowest current density reached in each passive region. As the X100 surface passivates, dissolution is hampered due to the reacting ion barrier covering anodic sites. Even though cathodic reaction rates at higher potentials are  66 greatly decreased, this should be counteracted by an accelerated dissolution half-cell reaction rate at these more anodic regions. Marked decreases in icorr and corrosion rate values are found in Cl−- and SO42−-free solutions, confirming the accelerated anodic dissolution and Fe2+ release of bare steel specimen [57] under the influence of these aggressive anions, not only their attack on passive layers formed. Corrosion rates at the lower and upper bounds of the pH range in this section (at 298 K or 25 °C, 6.7 - 8.9) increase from 0.055 to 0.326 mm per year (mmpy), for solutions with NS4 Cl− and SO42−. This is a significant difference for a relatively small pH change, and only occurs for a short period until dissolution sites are covered by corrosion product(s). Hence, this finding should not be generalized for long term behavior of X100 under these conditions since corrosion product deposition is only possible at higher pH, as discussed below.   Figure 5-8: Current density peaks Ep1 and Ep2 plotted vs. pH for Cl−- and SO42−-containing solutions at 1 mV/s and 0.5 mV/s, and Cl−- and SO42−-free solutions at 1 mV/s At potentials higher than the active region, passive layer formation is observed in all solutions with pH > 7.19 in Figure 5-7. In the pH 7.53 solution with the presence of Cl− and SO42−, Ep1 shifts downward by about -0.15 V compared to the pH 7.19 solution. Ep1 values for all  67 profiles are plotted in Figure 5-8, with error bars representing the range of values exhibited over the three scans conducted. Error bars extending in one direction only or not appearing at all are for parameter values which were the same for different scans. For the pH 7.53/7.56 solutions, further decrease in Ep1 occurs when Cl− and SO42− are removed. This is explainable by easier passive layer formation without the attack of these species. Sporadic corrosion product formation in this case is still expected to be FeCO3, at an E-pH still below that at which Fe3+ becomes viable [185] and well above the metal immunity point [186] (see Fe-H2O Pourbaix in Figure A-5 in Appendix A). In the presence of Cl− and SO42−, metastable FeCO3 formation is due to mediocre kinetics of {R-5.10} and {R-5.11} from HCO3−, FeHCO3, and pH levels which are low. Previous reports attribute broad current density peaks in this potential range to the intersection of the cathodic currents of O2 reduction with the passive current [187]. Here, this is shown to be unlikely due to its occurrence with nominal O2 levels.  By increasing HCO3− concentration of the solutions further, creating conditions with pH 7.78, 8.1, and 8.36, the FeCO3 formation potential (2nd peak in these cases, Ep2) is driven closer to the immunity-dissolution boundary potential governed by the Fe-H-C-O Pourbaix [186] (see Figure A-6 in Appendix A). Accelerated kinetics of {R-5.10} and {R-5.11}, and thus easier association of FeCO3 in more alkaline conditions richer in HCO3− causes this, provided the thermodynamic viability of Fe dissolution is still valid. Ep2 values are slightly dependent on scan rate and Cl− and SO42− at both pH 7.78 and 8.1 (Figure 5-8). The slower scan rate facilitates FeCO3 development at lower potentials due to increased [Fe2+] in the solution from increased time spent at active potentials. Yet, the absence of Cl− and SO42− appears to be more effective for stable FeCO3 formation, as Ep2 of HCO3−-only solutions are lower than for 0.5 mV s-1 cases (Figure 5-8).  In the mildly alkaline solutions of pH 7.78 - 8.36 in Figure 5-7, the same multi-step dissolution process of Figure 5-4 is observed in the -0.5 to -0.2 VSCE potential range. The  68 involvement of Fe(OH)2, interacting with FeCO3 forming at higher potentials in a multi-layered or intermixed morphology as discussed in section 5.2.1, is again likely here. Theoretically, this Fe(OH)2 formation cannot occur at a pH less than a mildly alkaline 7.6-8, depending on the presence of other anions in the solution and temperature [185]. The resulting passive layer Fe(OH)2 vs. FeCO3 constituent ratio depends on whether HCO3− or OH− dominantly drives the charge-transfer steps on which the growth of the corrosion products depend at higher potentials [187]. This mechanism is independent from O2 traces in the media as reported by Rangel et al. [188]. Thus, comparability of present results with those of previous tests, some of which containing O2 [175], [179], [188], is sustained.  The E-pH region of Fe(OH)2 dominance in the Fe-H-C-O Pourbaix has been superimposed onto Figure 5-8 to demonstrate this interpretation [178]. Ep1 data points for all pH > 7.5 lie within the Fe(OH)2 dominance region; yet, not necessarily implying the formation of Fe(OH)2 in all these cases. At the highest pH of 8.9 it is more likely that FeCO3 formation takes precedence over Fe(OH)2 due to the significantly high [HCO3−], making HCO3− the main driver of the charge transfer steps ({R-5.10} and {R-5.11}). This is corroborated by the absence of local current minima in this profile. The second stage of the multi-step dissolution indicates Fe(OH)2 removal under the influence of high [HCO3−] as described by Castro et al. (11) [179], and shown by {R-5.9} in the previous section. Beyond Ep2, the mildly alkaline (pH 7.78 - 8.36) profiles exhibit a region of constant or decreasing current density dependent on [HCO3−], scan rate, and the presence of Cl− and SO42−. Decreasing ip values in this region with increasing pH indicate the gradually enhancing protectiveness of FeCO3 at higher anodic potentials [188], [189], plotted vs. pH in Figure 5-9. The potential difference (∆E) parameter, between the most anodic passive layer development potential (Ep1 or Ep2) and the earliest breakdown potential (Ebd1), is plotted vs. pH in Figure 5-10. Fe2+ from Fe dissolution is involved in FeCO3 formation through direct association with CO32- or through FeHCO3 intermediary steps. Direct transformation of Fe(OH)2 into FeCO3 through {R- 69 5.12} is also possible. Although it is difficult to quantitatively specify the ratio of the latter FeCO3 formation path, it appears that Fe(OH)2 is always a part of the formation process in the pH 7.78 - 8.36 domain through multi-step passivation-dissolution behavior at these E-pH levels. Fe(OH)2 + HCO3−  → FeCO3 + H2O + OH−    {R-5.12}  Figure 5-9: Passive current density ip plotted vs. pH for Cl−- and SO42−-containing solutions at 1 mV/s and 0.5 mV/s, and Cl−- and SO42−-free solutions at 1 mV/s The protectiveness of the FeCO3 layer shows clear dependencies regardless of formation path(s). Prolonging formation times (with slower scan rates) and removing Cl− and SO42− results in lower ip (Figure 5-9), denoting better protectiveness. Although reducing the scan rate may sometimes have little or no effect on ip, removing Cl− and SO42− from the solutions always improves protectiveness (except in the pH 8.9 condition, where ip reaches a minimum). ∆E increases with pH, reduced scan rate, and the absence of Cl− and SO42−, signifying increased protectiveness with those changes. The ∆E trend vs. pH is inversely proportional to ip, and separations between different conditions are more pronounced. This indicates that although the corrosion rate of the passivated specimen might be the same in NaHCO3-only solutions or during  70 slower scan rates, the passive layer in these conditions will resist dissolution/breakdown more. The ∆E of the NaHCO3-only solutions are generally more than 3 times that of the solutions with Cl− and SO42− at the same scan rate, and almost double that of tests conducted at 0.5 mV s-1.  The tests presented in this section are all conducted at 25 °C, so temperature dependence of FeCO3 formation cannot be deduced from the results. It has been reported though that siderite’s solubility product decreases from 10-11 to 10-11.5 for a 50 degree increase above the temperatures tested here, implying slightly higher siderite precipitation with temperature [190]. However, this effect is offset and likely even overturned by decreased [HCO3−] with lower CO2 solubility at such higher electrolyte temperatures, due to the stronger influence of [HCO3−] on FeCO3 formation and robustness presented here and in previous studies.      Figure 5-10: Potential difference ∆E plotted vs. pH for Cl−- and SO42−-containing solutions at 1 mV/s and 0.5 mV/s, and Cl−- and SO42−-free solutions at 1 mV/s  71  Figure 5-11: Breakdown or transpassivation potentials Ebd1 and Ebd2 plotted vs. pH for Cl−- and SO42−-containing solutions at 1 mV/s and 0.5 mV/s, and Cl−- and SO42−-free solutions at 1 mV/s A passive-layer breakdown behavior dependent on pH and solution anion content is witnessed at potentials beyond the passive region in Figure 5-7. Here, excluding the lowest and highest pH profiles, an abrupt breakdown commences at around 0.25 VSCE, Ebd1 and Ebd2 extracted and plotted vs. pH for all conditions are shown in Figure 5-11. Uniquely, all Cl−- and SO42−-containing solutions of pH ≤ 8.4 experienced early breakdown at potentials ≤ 0.5 VSCE, well below the O2 evolution potential of around 1 VSCE. Ebd1 values are relatively independent of pH or [NaHCO3] for both the slow and fast scan rates, negating the sole connection of the behavior to weak passive layer formation. Rather, this behavior likely indicates Fe3O4 or γ-Fe2O3 presence, previously shown to be unaffected by the presence of HCO3− or CO32- as long as E remains in the passive region, since the oxidation current is practically independent of [HCO3−] or [CO32-] [189]. Corrosion product between Ebd1 and Ebd2 loses some but not all of its protective strength, suggesting formation of a more porous or less protective product/pseudomorph is occurring at these particular potentials. Superimposing the upper boundary of the Fe2O3 dominance region  72 from the Fe-H-C-O Pourbaix system [186] to Figure 5-11 further supports the impression that Fe3O4 or γ-Fe2O3 formation in alkaline media is happening around Ebd1. Voltammetric studies of carbon steel in deaerated NaHCO3 solutions by El-Naggar [191] have shown that former reaction complexes like Fe(OH)2 can be involved in Fe3O4 formation. Likewise, transformation from the more predominant FeCO3 has been reported by [192]:  Fe +  2Fe(OH)2 + 4OH− → Fe3O4 + 4H2O + 4e−   {R-5.13} 3FeCO3 +  5OH−  → Fe3O4 + 3HCO3− + H2O + 2e−  {R-5.14}  These reactions occur in the complete absence of O2, although its presence can equally enable the pure chemical oxidation of FeCO3 as shown in [79]. Here, current density accelerations are attributed to both oxidative electron release in {R-5.13}-{R-5.14} and the proven diminished protectiveness of Fe3O4 [82]. In contrast to the insulating properties of FeCO3, Fe3O4 is considered to be a good electrical conductor [191]. Intermediary current density levels between Ebd1 and Ebd2 indicate retentive protectiveness behavior. Possessing a similar structure to Fe3O4 but pronounced enhancement in passivity and diminishment in conductivity, γ-Fe2O3 forming from Fe3O4 [191] or the original FeCO3 [175] develops over Fe3O4 to create a Fe/FeCO3/Fe3O4/γ-Fe2O3/H2O system as confirmed by XRD results in Figure 5-12b [193]: 2Fe3O4 + H2O → 3γ − Fe2O3 + 2H+ + 2e−   {R-5.15} 3FeCO3 +  4OH−  → Fe2O3 + 2HCO3− + H2O + 2e−  {R-5.16} SEM imagery in this potential region identifies this proposed FeCO3 and Fe3O4/γ-Fe2O3 presence. Ex-situ images taken after the PDP scan on the specimen is stopped prematurely at a final E < Ebd2. FeCO3 precipitates of characteristic rhombohedric crystal structure form in a columnar-like fashion [194] with sizes ranging from 10 μm to 30 μm in the top right half of Figure 5-12a. They are covered by distinctive iron-oxide product in the lower left half. The cloud-like oxide morphology is encroaching on the surface of the FeCO3 formations, almost engulfing individual  73 crystals at the boundary. The lower magnification image reveals the varying porosity of Fe3O4/γ-Fe2O3 formations.   Figure 5-12: (a) SEM image (x150 magnification) of FeCO3-Fe3O4/γ-Fe2O3 discrete boundary: (zoom) x700 magnification of Fe3O4 growth over FeCO3 crystals; (b) XRD pattern of a specimen removed from a PDP scan in pH 8.36 solution (top profile), and control sample with no corrosion product (bottom profile)  a b  74 The exclusivity of this transformation process to environments with Cl− and SO42− is supported by the synergistic effect of both anions, as shown in the previous section [172]. As such, the process is more evident here than in investigations in which similar amounts of only Cl− was added [195]. Without Cl− and SO42−, breakdown occurs at much higher potentials, generally above the Fe2O3 formation region. Similarly, small Cl− and SO42− vs. HCO3− ratios in very high [NaHCO3] solutions are behind the absence of noticeable Fe3O4/γ-Fe2O3 formations in those conditions, since the FeCO3 passive layers present are more robust. Combining these findings with results which report the absence of Fe3O4/γ-Fe2O3 in mildly acidic, high temperature 80 °C, CO2-saturated conditions representing possible internal conditions of pipelines (Farelas et al. [196]) indicates that higher temperatures make FeCO3 to Fe3O4 transformation kinetics insignificant in the absence of O2, and hamper the pure chemical oxidation of FeCO3. It is noted that in [196], O2 levels were < 10 parts per billion (ppb) representing completely anoxic conditions inside a pipeline. External surfaces can certainly anticipate higher O2 amounts. Also, the flow of corrosive media within a pipeline as simulated in their setup will introduce mass-transfer effects not present in external corrosion of buried pipelines. 5.3 EIS tests at OCP in deaerated nn-pH conditions EIS tests are performed here to study the influence of pH, within the nn-pH range, on corrosion and auxiliary electrochemical processes on unpolarized X100 specimens. The tests were hence done at OCP following its stabilization to steady levels after 1 h immersion in solution. Changes in pH of the solutions in which EIS was conducted here are instigated in the same way as that illustrated in Figure 5-1, namely through different [HCO3−], purging environment, and temperature combinations. Due to similar responses for the reference NS4 and NS4 – 10x HCO3− conditions, plots and results are presented for the first condition only, in addition to NS4 + 10x HCO3− environments and solutions with added Cl− and/or SO42− in concentrations greater than the standard NS4 level.   75 In the standard NS4 solutions, the Nyquist plots of Figure 5-13a readily show the effect of purging gas environment and temperature on the impedance profiles. In the 100% CO2 gas purging condition, nearly-complete depressed semicircles decrease in size with temperature, and no induction effects are observed at lower frequencies. No positive inflections in the impedance profile are observed at low frequencies of the Nyquist plots either. This characteristic is revealed better by the small phase angle (𝜃𝐸𝐼𝑆) and nearly constant impedance moduli (|Z|) profiles at those frequencies in Figure A-7 and Figure A-8, respectively in Appendix A. This behavior suggests the absence of significant adsorption processes occurring outside the double layer. Combined with the absence of induction or diffusion characteristics to the plots, such a response advocates the triviality of any electrochemical developments occurring outside the double layer in the 100% CO2 gas purged NS4 solution at all temperatures. Since this reference NS4 [HCO3−] condition does not cause passivation even at potentials more anodic than OCP, the existence of a passive film layer is ruled out for this condition. Thus, a basic single time-constant equivalent circuit comprised of 𝑅𝑐𝑡, 𝑅𝑠, and a CPE double layer (𝑄𝑑𝑙) is used to model the data (Figure 5-13b). The CPE is used instead of ideal capacitor elements to account for frequency dispersion due to distributed time constants and surface heterogeneities. Capacitive idealness of the CPE is governed by the exponent ratio (n) in the CPE impedance (𝑍𝐶𝑃𝐸) relationship {E-5.1}, first proposed by Brug et al. [197]. The 𝑗 in {E-5.1} is the imaginary number, √−1, and 𝜔 is the frequency of the AC input.     𝑍𝐶𝑃𝐸 = [𝑄(𝑗𝜔)𝑛]−1      {E-5.1} The results of fits to the EEC correlate well with the experimental data as the 𝜒2 values of Table 5-1 reveal. 𝑅𝑐𝑡 decreases with temperature indicating increased corrosion rate of the steel, following well the LPR results presented in section 5.1. This same trend has previously been reported in other investigations for similar CO32−-HCO3− conditions [13]. The capacitive  76 character of the double layer is reflected by the reciprocal relationship between 𝑄𝑑𝑙 and 𝑅𝑐𝑡 . This typical CPE behavior [166] is observed in all the EIS results in this section.     Figure 5-13: EIS for NS4 solution at 25, 40, and 55 °C purged with 5% CO2/95% N2 (pH 6.7, 6.8, and 7.0, respectively) or 100% CO2 (pH 5.4, 5.5, and 5.6, respectively): (a) Nyquist impedance representation, (b) proposed EEC for 100% CO2 results         a b  77  Table 5-1: EIS component values for NS4 solution at 25, 40, or 55 °C purged with 5% CO2/95% N2, or 100% CO2  Contrary to the 100% CO2 case, the 5%CO2/95%N2 gas purged NS4 environment exhibits positive inflections in Z and 𝜃𝐸𝐼𝑆 at lower ω. Also, 𝜃𝐸𝐼𝑆 peaks are higher and more distinct as seen in Figure A-7 (Appendix A). This signifies the importance of processes occurring outside the double layer in this condition and suggests the presence of direct HCO3− adsorption. This behavior may also be due to relaxation of carbon carrying intermediate species [198]. The contribution of adsorption and/or relaxation of intermediate species external to the double layer is accounted for in the proposed EEC (Figure 5-14) with the addition of a nested parallel circuit to the previous case, comprised of an adsorption resistance 𝑅𝑎 and an adsorption CPE 𝑄𝑎. Other EECs have previously been demonstrated to be accommodating of adsorption processes in CO32−-HCO3− solutions [26, 34], yet since those reports involved passive layer formation, those circuits are unsuitable for the specific conditions addressed here.   Figure 5-14: Proposed EEC for 5% CO2/95% N2 purged NS4 solution at 25, 40, and 55 °C (pH 6.7, 6.8, and 7.0, respectively) and 100% CO2 purged NS4 + 10x HCO3− solution (pH 6.4, 6.6, and 6.7, respectively) Components 5% CO2 / 95% N2 100% CO2 25 °C   (pH 6.7) 40 °C    (pH 6.8) 55 °C    (pH 7.0) 25 °C    (pH 5.4) 40 °C    (pH 5.5) 55 °C    (pH 5.6) Rs [Ω cm2] 129.7 98.7 97.0 124.1 87.7 93.2 Qdl [Ω-1 sn] 2.22 x 10-4 2.28 x 10-4 2.92 x 10-4 3.52 x 10-4 4.71 x 10-4 5.45 x 10-4 ndl 0.81 0.82 0.78 0.84 0.80 0.79 Rct [Ω cm2] 856.3 811.9 761.3 291 188.9 138.3 Qa [Ω-1 sn] 1.85 x 10-3 4.74 x 10-3 1.06 x 10-2 \ \ \ na 0.61 0.65 0.81 \ \ \ Ra [Ω cm2] 500.8 371.3 287.5 \ \ \ 𝝌𝟐 1.36 x 10-5 7.36 x 10-5 1.55 x 10-4 6.59 x 10-4 4.54 x 10-4 1.07 x 10-3  78 The proposed EEC of Figure 5-14 achieves a good fit with the experimental data as shown in Table 5-1, wherein it is seen that resistance to adsorption (i.e. 𝑅𝑎) decreases with increased temperature. This behavior is observable graphically from the difference in impedance magnitude (|𝑍|) at low ω in Figure A-8. This perhaps explains the improved adsorption of hydrous Fe(OH)2 at higher temperatures (higher pH) witnessed in the polarization results of the previous sections. The 𝑅𝑐𝑡 values decrease with temperature indicating increased corrosion activity, matching previous polarization results. They are markedly higher than the 100% CO2 gas purged counterparts, confirming decreased corrosion severity in more alkaline solutions created by removal of dissolved CO2. The Nyquist plots for EIS in NS4 + 10x HCO3− solutions at 25, 40, and 50 °C with various purging gases are shown in Figure 5-15a and Figure 5-15b. Bode |𝑍| and 𝜃𝐸𝐼𝑆 plots for EIS in these conditions can be found in Figure A-9 to Figure A-11 in Appendix A. All these plots demonstrate the induced effects of increased solution alkalinity and added [HCO3−]. Nyquist plots for the 100% CO2 case at all temperatures exhibit a similar response to the mildly acidic 5% CO2/95% N2 purged NS4 condition discussed above. Adsorption effects are evident from the |𝑍| and 𝜃𝐸𝐼𝑆 values at lower ω. Good fit with the experimental data is achieved when modeled with the EEC of Figure 5-14, as seen in Table 5-2. The trend of decreasing 𝑅𝑎 with temperature continues to occur, suggesting increased adsorption due to temperature. Judging from the increased 𝑄𝑎 values with temperature, there appears to be a capacitive adsorption region near the specimen surface which behaves similar to the double layer. This adsorption region becomes more dominant at higher temperatures. The decreasing 𝑅𝑎 effect with temperature is apparently not due to the associated increase in pH with higher temperatures, since 𝑅𝑎 values increase markedly in the more alkaline solutions of 5% CO2/95% N2 and 100% N2 conditions in Table 5-2. 𝑅𝑎  only decreases with temperature within the 5% CO2/95% N2 test set, establishing the temperature-only dependence of adsorption even further.  79  Figure 5-15: NS4 + 10x HCO3− solution at 25 °C purged with 100% N2 (pH 8.9), and at 25, 40, and 55 °C purged with 100% CO2 (pH 6.4, 6.6, and 6.7, respectively), and 5% CO2/95% N2 (pH 7.6, 7.9, and 8.1, respectively) (a) Nyquist impedance representation for all cases, (b) enlarged Nyquist impedance representation for 5% CO2/95% N2 and 100% CO2 purged cases   a b  80 Table 5-2: EIS component values for NS4 + 10x HCO3− solution Components 5% CO2/95% N2 100% CO2 100% N2 25 °C   40 °C   55 °C   25 °C   40 °C   55 °C   25 °C   Rs [Ω cm2] 33.5 25.5 25.0 32.4 20.9 19.3 29.3 Qdl [Ω-1 sn] 2.33 x 10-4 5.98 x 10-3 1.10 x 10-2 1.32 x 10-4 3.53 x 10-4 4.50 x 10-4 2.75 x 10-4 ndl 0.83 0.66 0.66 0.96 0.85 0.86 0.85 Rct [Ω cm2] 534.3 449.6 422.7 300.24 208.2 142.4 930.4 Qa [Ω-1 sn] 1.64 x 10-4 2.58 x 10-4 3.08 x 10-4 3.43 x 10-2 0.10 0.44 6.11 x 10-4 na 0.30 0.81 0.81 0.74 1 1 0.87 Ra [Ω cm2] 527 347.4 291.3 261.6 37.65 24.7 1629 W [Ω-1 s0.5] \ 3.26 x 10-2 2.43 x 10-2 \ \ \ 2.14 x 10-3 𝝌𝟐 8.69 x 10-5 7.04 x 10-4 2.48 x 10-4 3.92 x 10-4 5.32 x 10-4 3.13 x 10-4 3.55 x 10-4 In environments with a pH of around 8 or greater, specifically 100% N2 and higher temperature 5%CO2/95%N2 conditions, a significant change in the low ω Nyquist profiles is observed. An incomplete semi-circle at high ω is followed by a nearly linear region at lower ω. The Bode plots for these conditions exhibit a constant 𝜃𝐸𝐼𝑆 for almost one decade or more of low ω, from 0.01 to 0.1 Hz or greater (see Appendix A). These |𝑍| and 𝜃𝐸𝐼𝑆 features are characteristic mass transfer controlled behaviors [158], typically modelled with a diffusion element in EECs. This suggests that in these conditions, diffusion of electro-active species to or from the specimen surface occurs, in agreement with findings of Linter and Burstein for similar mildly alkaline solutions [64].   Figure 5-16: Proposed EEC for 5% CO2/95% N2 purged NS4 + 10x HCO3− at 25, 40, and 55 °C (pH 7.6, 7.9, and 8.1, respectively), and 100% N2 purged NS4 + 10x HCO3− solution at 25°C (pH 8.9) An EEC with components to account for parallel diffusion and adsorption control at low ω is proposed in Figure 5-16, where 𝑊 represents the Warburg diffusion element. The experimental data fits the model well (Table 5-2), and a reciprocal relationship between the  81 diffusion and adsorption is seen at pH 8.9. Adsorption at this pH is greatly reduced yet not fully eradicated, and diffusion of complex ions at the surface greatly increases, hence contributes much more in controlling the corrosion rate of the specimen. This is witnessed in the 100% N2 purged column of Table 5-2, through the considerable increase in 𝑅𝑎 to 1629 Ω cm2 (indicating decreased adsorption) combined with the significant decrease in W to the order of 10-3 Ω-1 s0.5 (indicating increased diffusion). Conversely, in the 5% CO2/95% N2 conditions of Table 5-2, parallel adsorption and diffusion processes both slightly increase with temperature between 40 and 55 °C. The relationship between these two processes is thus not only directly pH-dependent, but also temperature-dependent as seen here and potential-dependent as shown further in section 6.2.1 of chapter 6. Here, the solution pH measured for the 25 °C condition with only adsorption effects versus the 40 °C condition with parallel adsorption and diffusion control suggests that the emergence of diffusive effects is pH dependent and occurs at a threshold of around 7.7. In addition, the small pH range between the 25 °C and 40 °C conditions reveals that this adsorption-diffusion transition is abrupt, not gradual.  Figure 5-17: Nyquist impedance representation plots for NS4 + 10x HCO3− solutions with added Cl− and/or SO42− at 25 °C, purged with 5% CO2/95% N2 (pH 7.6)   82 Table 5-3: EIS component values for NS4 + 10x HCO3− solutions with added Cl− and/or SO42− at 25 °C, purged with 5% CO2/95% N2 (pH 7.6) Components 5% CO2/95% N2 at 25 °C + 0.1 wt% 𝐒𝐎𝟒𝟐− + 0.1 wt% 𝐂𝐥− + 0.1 wt% 𝐒𝐎𝟒𝟐− + 0.1 wt% 𝐂𝐥− + 0.5 wt% 𝐂𝐥− Rs [Ω cm2] 36.7 26.2 24.2 18.0 Qdl [Ω-1 sn] 2.52 x 10-4 3.36 x 10-4 2.86 x 10-4 3.11 x 10-4 ndl 0.73 0.77 0.76 0.75 Rct [Ω cm2] 1722 1344 1617 1392 Qa [Ω-1 sn] 8.89 x 10-5 1.30 x 10-4 1.91 x 10-4 2.17 x 10-4 na 0.86 0.89 0.83 0.81 Ra [Ω cm2] 619.8 628.9 703.5 683.6 𝝌𝟐 5.46 x 10-5 1.56 x 10-4 1.11 x 10-4 7.87 x 10-5  Nyquist impedance plots for NS4 + 10x HCO3− solution with added Cl− and/or SO42− are shown in Figure 5-17. Bode |𝑍| and 𝜃𝐸𝐼𝑆 plots for EIS in these environments can be found in Figure A-12 in Appendix A. These plots show that Cl− and/or SO42− additions induce less adsorption than the reference solution free of such additions. The resemblance between features of these plots compared to the other nn-pH solution plots, such as NS4 + 10x HCO3− solution purged with 100% CO2, suggests the similarity of electrochemical processes occurring in both. Hence, the adsorption-controlled EEC of Figure 5-14 is used to model the data, resulting in the component values listed in Table 5-3. As seen here, reduction in adsorption propensity occurs, based on the increase in 𝑅𝑎 values versus the reference case. An increase in 𝑅𝑐𝑡 values compared to the reference solution is also observed, indicating a reduction in free corrosion activity due to Cl− and/or SO42−. However, this behavior is exclusive to OCP conditions, since the results are opposite to those of EIS tests conducted at E > OCP shown in section 6.2.1 of the forthcoming chapter 6. Electrode potential plays an important role in the corrosion severity of Cl− and/or SO42−, a proposition preluded by the PDP profiles in Figure 5-6. From the Table 5-3 results, no clear trend can be observed in 𝑅𝑐𝑡, 𝑅𝑎, or the related 𝑄𝑎 and 𝑄𝑑𝑙 values with respect to Cl− and SO42− combinations. The effect Cl− and/or SO42− additions to HCO3−/CO32- solutions have on the corrosion and passivation of X100 steel is hence an area which requires further investigation. Such a study is presented in chapter 6 of this thesis.  83  Subdivision of electrochemical response based on pH range  Examination of the EIS data presented in the previous section highlights three different regions of electrochemical processes which can occur on X100 steel during free corrosion (i.e. OCP), depending on pH of the environment. It is important to note that the nn-pH range of 5 to 8.5 overlaps each of these three pH regions, signifying the criticality of seemingly minor environmental differences on the progression of corrosion processes. In acidic solutions below a pH of 6, such as NS4 solution at 55 °C purged with 100% CO2, only a capacitive double layer is present at the steel specimen’s surface, with no signs of adsorption or diffusion effects beyond that. In mildly acidic (pH 6.5) and neutral pH solutions such as NS4 solution at 55 °C purged with 5% CO2/95% N2, adsorption of electroactive species outside the double-layer becomes evident. Adsorption dominance is decreased with the presence of Cl− and/or SO42− ions, even though the presence of these ions does not change the pH of the solution. In alkaline solutions with a pH ≥ 7.7, mass-transfer behavior manifests in the interfacial processes, and mainly the diffusion of complex ions controls the corrosion rate of the steel. 5.4 Summary In this chapter, X100 pipeline steel specimens are subject to a matrix of environmental conditions characterized by [HCO3−], purging gas, solution temperature, Cl− and/or SO42−, and resultant solution pH. The reference condition is standard NS4 solution at 25 °C, deaerated with 5% CO2/95% N2. OCP values decrease linearly with pH at all temperatures due to accelerated cathodic reactions at lower pH. When pH increase is created by decreasing %CO2 in the purging gas, the overall trend of the results implies increased corrosion at lower pH, as confirmed by LPR data. An opposing trend of increased corrosion rate with alkalinity manifests at pH > 7.6 in situations where only [HCO3−] is used to adjust pH, indicating not only the importance of the pH change but also the cause of the change. It is stressed that the corrosive aggressiveness of an environment cannot be determined qualitatively based on the solution pH only, for pH < 1.5.  84 Passive or passive-like behavior is exhibited only in solutions with above-reference [HCO3−], except for 0% CO2 situations where a defective hydrous Fe(OH)2 film forms due to the involvement of OH−. In solutions of pH ≥ 7.53, the passive layers formed also include Fe2+ with CO32- at medium range potentials as found through XRD characterization. At pH ≥ 7.78 this is followed by Fe3O4/γ-Fe2O3 formation at higher potentials under the influence of Cl− and SO42−. Passive layer protectiveness is enhanced at slower scan rates due to the greater time given for corrosion product formation and the increased availability of key ionic species (i.e. Fe2+) from increased time spent at active dissolution. Yet, passive layer protectiveness is more sensitive to Cl− and SO42− presence than scan rate. At solution pH levels > 8.5 (thus not within the nn-pH range), it seems HCO3−/CO32- drives FeCO3 passive layer reaction steps in a manner which takes precedence over Fe(OH)2 involvement, through direct association with Fe2+ or through FeHCO3 pre-passivation steps. Passive layer breakdown is highly dependent on the presence of Cl− and SO42−, and Fe3O4/γ-Fe2O3 formation over FeCO3 at high anodic potentials for specific conditions is observed under SEM. A synergistic effect for chloride and sulfate is established, even in tests conducted at OCP. EIS results reveal three varying surface interactions occurring on the steel based on the environmental conditions. Acidic solutions with a pH below 6 induce impedance responses representing the absence of any adsorptive and/or diffusive effects outside the capacitive double-layer. Solutions of nn-pH introduce a capacitive adsorption region at the steel surface whose dominance increases with solution temperature and decreases with Cl− and/or SO42− additions. Further increase in pH into the alkaline region brings about mass-transfer controlled behavior which, in parallel with adsorption effects, controls the corrosion rate of the steel. Cl− and/or SO42− additions of any amount completely eliminate this diffusion behavior on unpolarized specimen and are found to impede charge transfer at OCP or during free corrosion.   85 6 6. Quantitative studies of the properties & growth of corrosion products on X100 steel in mildly alkaline deaerated 𝐇𝐂𝐎𝟑− solutions using EIS and Mott-Schottky4   A main corrosion product causing passivation of steel in HCO3−-rich, deaerated, and mildly alkaline environments is FeCO3 as identified in Figure 5-12 of the previous chapter. A formation threshold pH for FeCO3 on X100 exists at around 7.6 in solutions with 10x the reference NS4 [HCO3−] (section 5.2.1). The dependence of FeCO3 formation on pH and [HCO3−] is clear according to {R-2.2} and the previously presented results. Aggressive anions like Cl− and SO42− also play an important role in passive layer development and performance. In HCO3−/CO32- electrolytes, Cl− positively shifts transpassivation potentials with CO32- at different temperatures [200], or eliminates the onset of passivation as shown by Alves et al. [192]. At anodic sites, Cl− accelerates anodic dissolution and Fe2+ release from Fe specimens, as discussed by Lorenz and Heusler in [57] and supported experimentally by Zhang et al. [58]. A synergistic effect of Cl− and SO42− on corrosion acceleration and the elimination of passivation on X100 has been shown earlier in chapter 5. However, quantitative studies of the synergistic influence of Cl− and SO42− on interfacial processes involved in HSLA steel corrosion/passivation in HCO3− environments are still needed. This can contribute to forming a mechanistic understanding of more advanced degradation processes in related environments like nn-pH SCC, which until now lacks such an understanding [201].   The nature of corrosion products and the corresponding necessary reaction steps of steel passivation in HCO3− solutions are potential dependent. Fundamental cyclic voltammetry studies by El-Naggar [191], [202] on carbon steel in deaerated NaHCO3 solutions have revealed the                                                  4  I. M. Gadala and A. Alfantazi, Applied Surface Science, vol. 357, Part A, pp. 356–368, Dec.  2015.  86 involvement of OH−, CO32-, and O2- complexing reactions with Fe2+ at low, medium, and high anodic potentials, respectively. Using EIS to study these potential dependent steps is beneficial. Yet, static EIS at OCP or any higher potential to which a bare sample is immediately polarized overlooks the effect of previous processes and surface states, leading to electrochemical responses which may misrepresent the actual chronological progression of corrosion and passivation. In contrast, EIS under PDP conditions can be used to determine changes in passive layers caused by variations in electrode potential. Linear or step-wise potential change of EIS conditions can capture and quantify the time and/or surface dependence of sequential electrochemical processes such as intermediate diffusion or pitting [167]. A theoretical formulation proving the suitability of this method in studying time-dependent events is outlined by Darowicki in [168].  Since the electronic properties of a passive layer play an essential role in corrosion processes of the underlying steel substrate [203], their analysis for passive layers formed in aqueous media with different environmental parameters such as [HCO3−],  temperature, and pH is valuable. It is acknowledged that the electrochemical and protective behavior of a passive layer is related to its electronic properties [204]. Mott-Schottky is commonly used to study the electronic properties of passive films by measuring electrode capacitance (C) as a function of potential [205], [206]. In this chapter, Mott-Schottky tests are performed on anodized specimens in nine different mildly alkaline HCO3− environments.  The main objective of this chapter is to further examine and quantify the interfacial and surface processes preceding, during, and following stable passivation in a HCO3−-based mildly alkaline solution (pH 7.8) using a dynamic step-wise anodizing-EIS routine. Passive layer protectiveness parameters are quantified for the effect of Cl−/SO42−. Furthermore, temperature dependent trends are identified using current density decay analysis of PSP tests, and Mott-Schottky is used to evaluate the semiconductive behavior of FeCO3 layers at normal to high  87 temperatures (25 – 75 °C). This chapter is based on a paper [75] which was published as part of the research work leading towards this PhD thesis. 6.1 Details of dynamic EIS, PSP, and Mott-Schottky test methods After reaching a stable OCP, either a step-wise anodizing-EIS routine or an extended anodizing, EIS, and Mott-Schottky routine is implemented in the tests here. The step-wise anodizing-EIS routine, performed in 0.1 M [HCO3−] solution with/without NS4-level Cl−/SO42− additions, is a six-stage process where the steel is sequentially anodized at a specific potential for 1200 s and during EIS. The anodizing period following each change ensures sufficient stability for the EIS at each stage. For such non-stationary potential conditions, either multi-sine wave [207] or single sine wave AC inputs could be used. In this study we opt for a single sine wave input (𝜔 range: 10,000 - 0.1 Hz; AC disturbance signal: 10 mV; and sampling frequency: 10 points/decade) to allow comparability with the proposed EECs and results of other studies, the majority of which also use single sine wave input.  Table 6-1: Anodizing potentials (Ean) in each potential range for step-wise anodizing-EIS routine  A comparative plot of the PDP profiles for 0.1 M [HCO3−] solution (pH 7.8) with/without NS4-level Cl−/SO42− additions is shown in Figure 6-1, displaying key multi-step dissolution, passive layer formation, and early passive layer breakdown or transformation features. This environment is selected for the step-wise anodizing-EIS tests due to its multifaceted response to accelerated corrosion/passivation through PDP, which gives opportunities for a more complete understanding of the chronological electrochemical step processes. The potential ranges of the main electrochemical processes of interest, for both the 0.5 mV s-1 and HCO3−-only cases are  Potential (E) range in Figure 6-1 Active (early/late) Transition Passive formation Passive Transpassive/Breakdown Anodizing potential  (Ean) [VSCE] Solution with  NS4  𝐶𝑙−/𝑆𝑂42−  -0.65/-0.5  -0.475  -0.1  0.175  0.65 Solution without  NS4  𝐶𝑙−/𝑆𝑂42−  -0.65/-0.5  -0.4  -0.25  0.5  1.1  88 labelled in Figure 6-1 for their importance in the step-wise anodizing-EIS routine. The specific anodizing potentials (Ean) at which the PSP and each subsequent EIS are performed within each range are listed in Table 6-1. In general, Ean are chosen to be in the middle of each E range to minimize any inaccuracies in determining the range boundaries and to isolate the specific processes occurring in each range for spectroscopic analysis.   Figure 6-1: Comparative PDP plot for 0.1 M [HCO3−] solution with and without NS4  Cl−/SO42− scanned at 0.5 mV/s or 1 mV/s, labelled for potential regions of step-wise anodizing-EIS routine The second independent test routine consists of a PSP or anodizing session at 0.5 VSCE for 1 h, followed by EIS and Mott-Schottky. The effects of higher alkalinity (pH 7.8 - 9.3) and temperature (25 - 75 °C) on the electronic and protective properties of the anodized passive layers are evaluated. Passive layer attack by Cl−/SO42− is eliminated in order to properly study the protective and electronic properties of well-formed FeCO3 under these environmental influences. Mott-Schottky tests are scanned between -0.5 VSCE and 1 VSCE at a frequency of 1 kHz and with a  89 step height of 20 mV. Figure 6-2 is a schematic of the two test routines of this study, and Figure B-1 in Appendix B illustrates the [HCO3−], temperature, and pH relationships of the solutions tested.  Figure 6-2: Schematic of the two independent electrochemical test routines conducted in this chapter  6.2 Dynamic step-wise anodizing EIS test  The following discussion of EIS results under PDP conditions follows the same order of the potential regions examined in each PDP scan of Figure 6-1 (i.e. from the active potential region to the transpassive/breakdown region, scanned in the anodic direction).   Active corrosion at E < Ep1 In the active potential region from OCP to just above -0.5 VSCE, nearly complete depressed semicircles were exhibited in the Nyquist profiles (Figure 6-3) of solutions with and without Cl−/SO42−. The existence of ancillary electrochemical processes is observable at lower ω. In the -0.65 VSCE profiles of Figure 6-3, slight positive inflections in the Nyquist profile at 1 Hz are the result minor impedance increases from a ω-dependent surface process on the otherwise bare steel. At this early active potential, well before the significant icorr decreases above Ep1, adsorption of pre-passive ferrous bicarbonates ([FeHCO3−]𝑎𝑑𝑠 and [FeHCO3]𝑎𝑑𝑠, [180]) occurs prior to formation of FeCO3. In addition, the adsorption of FeOHads is a necessary prerequisite to  90 the development of Fe(OH)2 as shown in [178]. These pre-passivation steps are involved with anodic dissolution of the substrate at anodic sites which cover the majority of the surface at this stage.   Figure 6-3: Nyquist impedance representation in active corrosion region (OCP < E ≤ Ep1) This parallel dissolution-adsorption process is adsorption-controlled and ω-dependent based on its influence beyond 1 Hz only. Modelling this process using an EEC with a capacitive-resistive element pair in a parallel orientation [208] can accommodate the less restricted movement of current through capacitive branches at higher ω, as the resistive branches are shorted. In situations with more compliant adsorption processes, the applicability of a nested parallel adsorption layer within the double layer of the proposed EEC has been shown before [208]. The configuration is {R(C(R(C(R))))} as illustrated earlier in Figure 5-14 on page 77. This proposed EEC orientation is used to model the present adsorption-controlled process, achieving a better fit than the {R(CR)(CR)} configuration in which the adsorption elements are outside of the  91 double layer. EEC elements are the same as with the previous usage of this model, where CPE double-layer and adsorption elements are again governed by {E-5.1} on page 75  Good correlation between the -0.65 VSCE experimental data and {R(C(R(C(R))))} is achieved, as seen from 𝜒2 values on the order of 1.0 x 10-4 and a maximum error in fit of 2.25% (Table 6-2). The Westing-Mertens method described in {E-6.1} is used to convert CPE amplitudes Qdl and Qa to corresponding equivalent capacitances Cdl and Ca, where 𝜔𝜃 is the inflection 𝜔 at which the absolute value of 𝜃𝐸𝐼𝑆 reaches a maximum [209]. This method is appropriate for multiple time constant nested RC circuits, and n > 0.8 for all instances in Table 6-2 meets CPE usage recommendations. Table 6-2: EIS component values for 0.1 M [HCO3−] solution with and without Cl−/SO42− at Ean = -0.65 VSCE (early active E region) Components Solution with 𝐂𝐥−/𝐒𝐎𝟒𝟐−  at -0.65 VSCE  Solution without 𝐂𝐥−/𝐒𝐎𝟒𝟐− at  -0.65 VSCE Rs [Ω cm2] 24.1 19.6 Qdl [Ω-1 sn] 2.03 x 10-4 2.38 x 10-4 ndl 0.87 0.84 Cdl [F cm-2] 1.32 x 10-4 1.41 x 10-4 Rct [Ω cm2] 158 181 Qa [Ω-1 sn] 4.01 x 10-3 3.76 x 10-2 na 0.82 0.87 Ca [F cm-2] 2.24 x 10-3 2.45 x 10-2 Ra [Ω cm2] 34.9 75.6 𝝌𝟐 5.5 x 10-4 4.05 x 10-4 % error in fit < 2.25 < 1.93  𝐶𝑖 =𝑄𝑖 ∙ (𝜔𝜃,𝑖)𝑛𝑖−1sin (𝜋𝑛𝑖2)        {E-6.1} In Table 6-2, Rct values increase in the absence of Cl−/SO42−, and it appears that Cl−/SO42− also promote adsorption on the exposed steel surface, judging from decreased Ra values in their presence. These effects are opposite to those induced by Cl−/SO42− additions at OCP seen in the previous chapter. The anodic polarization of the specimen proliferates anodic site coverage more so than when left at OCP, which attracts dissolved anions and is likely what increases Fe2+ release and anion adsorption. It is proposed that due to the adsorbed anions hampering the  92 establishment of a complete first [FeHCO3]𝑎𝑑𝑠 layer on the steel adsorbent, the attractive strength between the steel and the Cl−/SO42− adsorbates remain high. Once a first [FeHCO3]𝑎𝑑𝑠 layer is formed, further adsorption to this layer is greatly impeded as established by numerous adsorption models, such as Hill-Langmuir in {E-6.2} [210]. In this specific model, 𝜃𝑎𝑑𝑠 is the fractional coverage of the surface, 𝐶𝑎𝑑𝑠 is the concentration of the adsorbed electroactive species/compound, and 𝛼𝑎𝑑𝑠 is a constant. Impedance to further adsorption, represented by Ra in the present EIS results, is higher for surfaces with more filled particle sites.  𝜃𝑎𝑑𝑠 =𝛼𝑎𝑑𝑠∙𝐶𝑎𝑑𝑠1+𝛼𝑎𝑑𝑠∙𝐶𝑎𝑑𝑠        {E-6.2} At Ean = -0.5 VSCE, the Nyquist plots in Figure 6-3 reveal more noticeable supplemental electrochemical processes at low 𝜔 < 1 Hz, suggesting change in governing mechanism(s) [211]. The positive linearity of the Nyquist plots at 𝜔 < 1 Hz suggests a contribution of diffusion in the electrochemical process occurring. Mass-transfer controlled behavior, characterized especially by the linear slope segments in the Nyquist plots, give evidence for the necessary diffusion of electro-active species towards the specimen surface in support of the pre-passivation step reactions ensuing at this potential. In the mildly alkaline media here, these species are suggested to be HCO3− and OH− [212] diffusing due to a concentration gradient arising from depletion caused by earlier adsorption reactions. Due to continued association of the electrochemical process with Rct and Qdl, this diffusion is best accommodated in the proposed EEC by the W diffusion element. This element induces a 𝜔-dependent impedance (𝑍𝑊) which decreases at higher 𝜔 values as described by {E-6.3}, where 𝑊𝐴 is the Warburg constant: 𝑍𝑊 = 𝑊𝐴√𝜔+𝑊𝐴𝑗√𝜔       {E-6.3} The configuration with the diffusive element within the adsorption subset of the electrochemical circuit, namely {R(Q(R(Q(RW))))} as previously shown in Figure 5-16, achieves better fit than when the diffusive element is placed in series with Rs. Overall, very good  93 compatibility with the experimental results is attained, with the maximum error in fit not surpassing 2% as seen in Table 6-3.  Table 6-3: EIS component values for 0.1 M [HCO3−] solution with and without Cl−/SO42− at -0.5 VSCE (late active potential region), and 0.1 M [HCO3−] solution without Cl−/SO42− at -0.4 VSCE (transition potential region) Components Solution with NS4 𝐂𝐥−/𝐒𝐎𝟒𝟐− at -0.5 VSCE   Solution without NS4 𝐂𝐥−/𝐒𝐎𝟒𝟐− at -0.5 VSCE   Solution without  NS4 𝐂𝐥−/𝐒𝐎𝟒𝟐− at  -0.4 VSCE  Rs [Ω cm2] 20 24 24 Qdl [Ω-1 sn] 5.28 x 10-4 3.56 x 10-4 3.43 x 10-4 ndl 0.79 0.81 0.85 Cdl [F cm-2] 2.70 x 10-4 1.93 x 10-4 2.40 x 10-4 Rct [Ω cm2] 38.1 55.3 339 Qa [Ω-1 sn] 2.87 x 10-2 2.72 x 10-2 5.11 x 10-3 na 0.80 0.84 0.96 Ca [F cm-2] 1.51 x 10-2 1.61 x 10-4 4.62 x 10-3 Ra [Ω cm2] 91.7 115 84.9 W [Ω-1 s0.5] 4.42 x 10-2 3.28 x 10-2 2.53 x 10-2 𝝌𝟐 2.21 x 10-4 2.72 x 10-4 2.16 x 10-4 % error in fit < 1.52 < 1.63 < 1.51  The applicability of the Figure 5-16 configuration for nested parallel circuits with two time constants has previously been shown by Chen and Jepson in [213], albeit for a situation where the second time constant paired with the double layer is from a passive film (Rf  and Qf). Practically though, for the processes occurring in the present conditions adsorption and diffusion effects exist together due to the direct physical reliance of the former on the species transported by the latter. Thus, the proposed model configuration of Figure 5-16 is more representative of the suggested physical process. As such, increased diffusion W in Table 6-3 upholds well the accompanying increased adsorption (decreased Ra) at more anodic potentials. It is noted that increased diffusion is indicated by higher corresponding 𝑍𝑊 and that the W results in Table 6-3 are for ω-dependent units containing Ω-1. Compared to results for EIS at OCP in section 5.3, it is clear here that anodic polarization enhances the movement and adsorption of dissolved anions to an active corroding steel surface. Compared to the -0.65 VSCE results, decreased Rct values are anticipated in the -0.5 VSCE case due to higher dissolution with increased anodic polarization in  94 the active region. In both the -0.65 VSCE and -0.5 VSCE cases Cdl exhibits an inverse trend with Rct reflecting the capacitive character of the double layer [57].  Corrosion and passive-like behavior in the Ep1 < E < Ep2 transition In the active-passive transition potential region above -0.5 VSCE, the E at which the i minima occurs in Figure 6-1 most likely represents the change from Fe(OH)2 formation to its dissolution in HCO3− media [189]. The short E range where i decrease and subsequent acceleration occurs with the development and dissolution of this product substantiates the likelihood of it being Fe(OH)2. Fe(OH)2 development follows the reaction steps {R-5.6} – {R-5.8} shown before in section 5.2.1 [178], and its adherence to the surface in the final step {R-5.8} results in the intermittent current density decrease observed between Ep1 and Ep2 in Figure 6-1. The mechanism has been reported to be independent of O2 traces in the media [188]; hence, its occurrence in the deaerated solutions of the present investigation is not anomalous.  Since the E range of the Fe(OH)2 formation and dissolution region is very small, it is difficult to target the specific regions of decreasing or increasing i with a PSP. As such, polarization performed in this region might drive either Fe(OH)2 formation or dissolution, a result which can only be determined by observing i behavior during polarization or EIS analysis after it. From the Nyquist plot of 0.1 M HCO3− in the absence of Cl−/SO42− polarized at -0.4 VSCE (Figure 6-4a), it is more likely that the process being driven at this E is the formation of Fe(OH)2 below the current minima in Figure 6-1. The extended low ω portion of the Nyquist plot suggests that dissolution is not the only electrochemical process occurring. Rather, the diffusion of OH− and the adsorptive effect of FeOHads are still present. Continued Fe(OH)2 formation triggers OH− diffusion to counteract OH− depletion in early and late steps of the process (i.e. {R-5.6} and {R-5.8}, respectively). Adsorption, specifically of FeOH, is also inherently involved therein. Subsequent adherence of the final Fe(OH)2 product to the surface, as illustrated schematically in Figure 6-4b, leads to the observable increase of Zim and Zre values of the -0.4 VSCE case in  95 Figure 6-4a compared to the -0.5 VSCE case of Figure 6-3. The increased Z at the interface in the presence of this Fe(OH)2 product has the capability of causing noticeable declines in current density as reported in [177], a finding corroborated by the i minima between Ep1 and Ep2 exhibited in the PDP curves of Figure 6-1.   Figure 6-4: (a) Nyquist impedance representation in transition and passive layer formation region (Ep1 < E ≤ 0 VSCE); (b) Proposed physical occurrences for EIS at -0.4 VSCE in solution without NS4 Cl−/SO42− a b  96 Although Fe(OH)2 formation does partially cover the specimen surface, causing the increase in the impedance modulus of the -0.4 VSCE case, the continued applicability of the {R(Q(R(Q(RW))))} EEC configuration in modelling this stage is suggested. The Nyquist plot indicates the sustained involvement of adsorption-diffusion processes which precede and upkeep the Fe(OH)2 formation reactions (Figure 6-4b), requiring the inclusion of Ra, Qa, and W elements. The validity of including a third time constant with a parallel Rf and Qf configuration is considered to be poor. This is because the Fe(OH)2 is defective and does not completely passivate the surface [177], causing only transitory impediment of the dissolution process. In fact, even in situations where stable passive layers develop, Eliyan et al. [214] reported limited applicability of EECs with three time constant such as {R(QR)(QR)(QR)} and {R(QR)(QR)(Q(RW))} for situations with possible adsorption and/or insertion processes. Modelling the -0.4 VSCE case with the parallel diffusion-adsorption circuit illustrated of Figure 5-16 achieves a very good fit as seen by  𝜒2 and % error values in Table 6-3. The fitting results show sustained anion diffusion, mainly OH− based on the previous discussion, and increased adsorption compared to Ean = -0.5 VSCE in the late active E region. The increased adsorption represented by the decreased Ra values is attributed to the depletion of adsorbed OH− species and complexes (OH-ads, [FeOH-]ads, and FeOHads) due to the development of Fe(OH)2. The FeOHads oxidation product FeOH+ is a reactant in the Fe(OH)2 deposition mechanism, as shown schematically in Figure 6-4b. The depletion of adsorbed species/complexes decreases 𝜃𝑎𝑑𝑠 in {E-6.2}, or likewise in other models, thus causing more adsorption and supporting diffusion to sustain original concentrations.  Contrary to the -0.4 VSCE case, the complex plot for specimen polarized at -0.475 VSCE in Figure 6-4a lack any evident auxiliary effects at low ω. Even though the polarizing E is less anodic compared to -0.4 VSCE, it appears that the dissolution of exposed areas of the steel substrate is the controlling electrochemical process at this stage in these conditions. The strong influence of even small amounts of Cl−/SO42− is apparent here. These anions inhibit stable  97 adsorption of OH− species and OH− complexes, in addition to hampering the adherence of subsequently-formed products like Fe(OH)2. At this potential, any Fe(OH)2 deposit can also suffer from pure chemical breakdown into Fe2+, OH−, and CO32- ions, under the influence of HCO3− as seen in {R-5.9} [178], [215]. Combined with the accelerated Fe2+ release Cl− inflicts on the exposed steel [57], the chemical breakdown in {R-5.9} causes clear electrochemical dissolution-control at this E in the active-passive transition region. This process is electrochemically modelled by a single time-constant Randles EEC [216], namely{R(QR)}, resembling the active dissolution of stages at -0.65 and -0.5 VSCE except without adsorptive or diffusive effects. A CPE element is still used for the double layer, but the Brug method of {E-6.4} [197] is used instead of Westing-Mertens to convert its result to Cdl. The Brug method applies only to the single time constant Randles circuit and depends on resistances Rct and Rs, not ω: 𝐶𝑖 = [𝑄𝑖 (1𝑅𝑠+1𝑅𝑐𝑡)𝑛𝑖]1/𝑛𝑖     {E-6.4}  Table 6-4: EIS component values for 0.1 M [HCO3−] solution with Cl−/SO42−at -0.475 VSCE (transition potential region)       It is important to note that the proposed Randles circuit only models the active dissolution of the exposed sites of the steel substrate, not the Fe(OH)2 removal. The use of additional elements to represent any Fe(OH)2 presence on the surface is neglected for the same reasons outlined earlier in this section. Rct values can sufficiently gauge the impedance effect caused by the presence of any Fe(OH)2 deposits on the specimen. Indeed, the Rct value in  Components Solution with 𝐂𝐥−/𝐒𝐎𝟒𝟐−  at -0.475 VSCE Rs [Ω cm2] 19 Qdl [Ω-1 sn] 9.10 x 10-4 ndl 0.82 Cdl [F cm-2] 3.68 x 10-4 Rct [Ω cm2] 253 𝝌𝟐 6.04 x 10-4 % error in fit < 2.38  98 Table 6-4 agrees well with the PDP of Figure 6-1: the i of the solution containing Cl−/SO42− in the active-passive transition E region is more than the Cl−/SO42−-free solution in the same E region (Table 6-3), yet less than the same solution at a lower E = -0.65 VSCE (Table 6-2).   Corrosion and passivation processes at E > Ep2   With further increase of anodizing E into the passive-layer formation and passive regions above Ep2, passive layers develop stably as manifested by markedly higher impedance values in Figure 6-4a and Figure 6-5a (cross and star data points). Bode |𝑍| profiles for the star data points can be found in Figure B-2 in Appendix B. HCO3− in the electrolyte previously driving the complexing of pre-FeCO3 formation compounds at lower E [189] also contributes to the formation of FeCO3 at E > Ep2 as shown in the bottom half of Figure 6-5b. The degeneration of Fe(OH)2 deposits leads to increased FeCO3 either through direct transformation (shown in [217]) or through the increase of [CO32−] in the solution, which can later associate with Fe2+ from the substrate dissolution. Thus, the growth of the FeCO3 layer depends on HCO3− driving the charge-transfer steps, and forms over Fe(OH)2 in a multi-layer as illustrated in the top half of Figure 6-5b. The noticeable increase in impedance in the Nyquist plots in this E region is expected from increased Rct and the introduction of a robust barrier between the steel and the environment. This is modeled in the corresponding EEC with a second time-constant as proposed in [59] for steels in concentrated HCO3− solutions, with the configuration {R(QR)(QR)} in Figure 6-6a. Rf and Qf in this EEC are the passive film resistance and passive constant phase elements, respectively. This EEC is used to model Z spectra from both the passive-layer formation and passive regions below Ebd1 in the PDP plots, as only the growth of the FeCO3 layer and its level of protectiveness changes between these regions. The circuit achieves a good agreement with the experimental data (Table 6-5). Rf values are E-dependent and superior in the absence of Cl−/SO42−, indicating increased protectiveness. Rct values achieve agreeable compatibility with Rf  99 values due to decreased dissolution with greater FeCO3 growth, either at higher E or in Cl−/SO42−-free solutions. CPEs of both the double layer and the passive layer exhibit typical reciprocity with their corresponding resistances, again reflecting their capacitive character.   Figure 6-5: (a) Nyquist impedance representation in passive and transpassive region (E > 0 VSCE); (b) Proposed corresponding physical occurrences for EIS at -0.25 VSCE, -0.1 VSCE, 0.5 VSCE and 0.175 VSCE a b  100     Figure 6-6: (a) Proposed EEC and corresponding physical occurrences for EIS at -0.25 VSCE, -0.1 VSCE, 0.5 VSCE and 0.175 VSCE; (b) Proposed EEC for EIS at 1.1 VSCE in solution without Cl−/SO42−  Table 6-5: EIS component values for 0.1 M [HCO3−] solution with Cl−/SO42− at -0.1 VSCE (passive formation potential region) and 0.175 VSCE (passive potential region), and without Cl−/SO42− at -0.25 VSCE (passive formation potential region) and 0.5 VSCE (passive potential region) Components Solution with  NS4  𝐂𝐥−/𝐒𝐎𝟒𝟐−  at -0.1 VSCE  Solution without  NS4  𝐂𝐥−/𝐒𝐎𝟒𝟐−  at -0.25 VSCE  Solution with  NS4  𝐂𝐥−/𝐒𝐎𝟒𝟐−  at 0.175 VSCE  Solution without  NS4  𝐂𝐥−/𝐒𝐎𝟒𝟐−  at 0.5 VSCE  Rs [Ω cm2] 32 29 32 34 Qdl [Ω-1 sn] 9.14 x 10-3 7.40 x 10-3  7.54 x 10-3 3.35 x 10-5 ndl 0.84 0.81 0.94 0.99 Cdl [F cm-2] 7.85 x 10-3 6.22 x 10-3 6.97 x 10-3 3.29 x 10-5 Rct [Ω cm2] 1.00 x 103 1.40 x 103 4.42 x 103 8.82 x 103 Qf [Ω-1 sn] 9.97 x 10-4 6.54 x 10-5 5.24 x 10-5 2.09 x 10-5 nf 0.84 0.84 0.83 0.90 Cf [F cm-2] 8.56 x 10-4 5.62 x 10-5 4.29 x 10-5 1.76 x 10-5 Rf [Ω cm2] 1.54 x 103 3.71 x 103 1.07 x 103 2.26 x 103 𝝌𝟐 7.78 x 10-4 1.87 x 10-4 1.38 x 10-4 5.66 x 10-4 % error in fit < 2.76 < 1.44 < 1.33 < 2.29  At transpassive potentials above Ebd1, passive layer breakdown occurs as revealed by the sharp drop in |𝑍| values in Figure B-2 compared to spectra at passive potentials. The Nyquist profile of the 0.65 VSCE case was omitted from Figure B-2 due to its incoherence. The breakdown mechanism in Figure B-2 is dependent on Cl−/SO42−, manifested by the unstable fluctuation of the 0.65 VSCE case with Cl−/SO42− therein, compared to the orderly profiles of the higher applied E = 1.1 VSCE case without NS4 Cl−/SO42− in Figure 6-5a and Figure B-2. Here, Cl−/SO42− ions attack pre-existing deposits, increasing the density of anodic sites on the specimen exposed to electrolyte. Subsequently, Fe3O4/γ-Fe2O3 formation is likely occurring at this high anodic potential, resulting in a less protective more porous passive layer structure. Modeling the Fe3O4/γ-a b  101 Fe2O3 covered surface above Ebd1 involves an added time-constant to quantify pore impedance, but reliable fitting results were unachievable here. Separate EIS investigations dedicated to oxide and oxyhydroxides of Fe on X100 are presented in the next chapter (see section 7.4.2). Table 6-6: EIS component values for 0.1 M [HCO3−] solution without Cl−/SO42− at 1.1 VSCE (transpassivation potential region) Components Solution without 𝐂𝐥−/𝐒𝐎𝟒𝟐− at 1.1 VSCE  Rs [Ω cm2] 30 Qdl [Ω-1 sn] 2.22 x 10-3 ndl 0.82 Cdl [F cm-2] 1.88 x 10-3 Rct [Ω cm2] 1.28 x 103 Qf [Ω-1 sn] 3.80 x 10-4 nf 0.88 Cf [F cm-2] 3.37 x 10-4 Rf [Ω cm2] 1.90 x 103 W [Ω-1 s0.5] 2.58 x 10-2 𝝌𝟐 1.09 x 10-4 % error in fit < 1.27  The EIS response for the transpassive region in the solution free of Cl−/SO42− (1.1 VSCE) is stable, and diffusive effects appear once again after withdrawing during passive formation and passive regions. The lower |𝑍| values accompanying this suggests greater mobility of electroactive species driving the passive layer breakdown steps through increased transfer channels in the passive films. This resembles the interfacial process occurring on steel with weak passive layers created in more dilute HCO3− solutions in [218]. The increased transfer channels within the passive film require the diffusion element to be within the film section of the EEC, in a configuration similar to the Figure 5-14 model used for parallel diffusion-adsorption controlled processes at lower E. In principle, diffusion is also present within the electric double layer, including both the Gouy-Chapman layer below 200 Å and the thicker Nernst diffusion boundary. However, in the present proposed model any mass-transfer within the double-layer is neglected for simplicity. The parallel kinetics-diffusion controlled process observed here has been shown to be modeled well with the {R(QR)(Q(RW))} configuration of Figure 6-6b [219]. This circuit  102 achieves a good fit with the present experimental data at the transpassivation E in  Cl−/SO42−-free media (Table 6-6). Rf and Rct values drop significantly compared to passive E of Table 6-5, and it appears that species mobility is quite high, at a level almost comparable to pre-passivation active stages in Table 6-3.  6.3 Protective and semiconductive properties of anodized FeCO3  Current density decay and qualitative analysis of Bode |𝑍| results  The transient i decay profiles under a PSP voltage of 0.5 VSCE are illustrated in Figure 6-7.  The steepest decline is observed in the pH 8.5, 25 °C environment, reaching a stable i on the order of 1 µA cm-2 in less than 200 s of polarization. All profiles at 25 °C demonstrate rapid decay and stable steady-state i approaching 1 h PSP. Conversely, with increasing solution temperature, i reaches 5 µA cm-2 in approximately 800 s. Values of i shift higher with temperature and decreased pH due to less added [HCO3−], and changes due to [HCO3−] are more pronounced at higher temperatures. Transient profiles of tests conducted at 75 ºC have stability times greater than the time domain of the figure (i.e. > 1500 s) and still exhibit observable decrease after 1 h. Volatility in the decaying portion of the profiles alludes to unstable passivation development due to the thinner and less robust passive layer formed from sluggish film growth. The trend of Figure 6-7 confirms the development of a more robust and protective passive layer in colder, more alkaline, HCO3−-rich conditions. FeCO3 formation is generally accelerated with both increased alkalinity and [HCO3−] as described in [180]. Large Fe(OH)2 involvement in the formed passive layer is unlikely here due to thermodynamic unviability  at a direct PSP E of 0.5 VSCE, based on E-pH diagrams of the Fe-HCO3−-CO32−-H2O system. At any single [HCO3−], increasing solution temperature increases alkalinity due to decreased CO2 solubility and the resulting H2CO3 dissociation at higher temperatures. Yet, it is observed that this decreases passive layer protectiveness at any single [HCO3−]. This is due to decreased kinetics of pre-FeCO3 HCO3− complexation steps in {R-5.10} and {R-5.11} due to lower dissociated [HCO3−]  103 from the lower [H2CO3] at higher temperatures, hampering FeCO3 growth. Interestingly, this behavior offsets reduced FeCO3 solubility at higher temperatures, which would lead to increased precipitation. Previously, Benezeth et al. experimentally found that the solubility product (Ksp) of FeCO3 decreases from 10-11 to 10-11.5 for a temperature change from 25 to 75 ºC [220]. This influence is not enough to overcome sluggish FeCO3 growth at higher temperatures, leading to the observed reduced protectiveness at 50 and 75 ºC.   Figure 6-7: Current density decay profiles of specimen anodized at 0.5 VSCE in 0.1, 0.25, or 0.5 M NaHCO3 at 25, 50, or 75 ºC (pH 7.8 – 9.3) Identifying a passive layer protectiveness ranking based on temperature or [HCO3−] requires isolation of either variable to determine the individual effect of each. EIS is conducted on passivated specimen immediately following PSP to corroborate former interpretations. Only |Z| results are presented here for qualitative comparisons (Figure 6-8), and the EEC referred to herein is that which we used previously to model a well-developed passive layer which is not breaking down (Figure 6-6a). At ω < 100 Hz, |Z| are less influenced by the ω-dependent capacitive effects of Qf and Qct and hence the EIS spectra on the left half of Figure 6-8 follow the protectiveness trend of i decay results identically. In this ω range the resistances Rf and Rct, which are of physical  104 importance in identifying passive layer strength, are less skewed by the parallel CPE branches in the EEC. In contrast, at ω > 100 Hz, the capacitive branches experience reduced individual Z, decreasing the total |Z| and drawing current (I) from the resistive branches. Thus, |Z| at these ω represent less the combined resistivity of the passive and double layers, explaining the diminished separation of the data and the loss of the observed protectiveness trend. The reduced-slope region at 0.1 < ω < 100 Hz can be attributed to remnants of sustained mass-transfer, especially at high temperatures where full steady-state behavior is not achieved after 1 h of PSP. Comparing the |Z| spectra of Figure 6-8 to those of passivated surfaces in the step-wise anodizing-EIS tests (Figure B-2 in Appendix B) demonstrates that increasing temperature to 75 ºC has a detrimental effect on passive layer strength almost equivalent to the presence of aggressive Cl−/SO42−. In fact, the effect of higher temperature can be comparable to partial breakdown of the passive layer above Ep1, as |Z| approaches 103 Ω cm2 in both situations alike.   Figure 6-8: Bode |Z| profiles of specimen after 1 h PSP in at 0.5 VSCE in 0.1, 0.25, or 0.5 M NaHCO3 at 25, 50, or 75 ºC (pH 7.8 – 9.3)  105  Semiconductive properties analysis using Mott-Schottky  Mott-Schottky analysis relies on the assumption that the capacitance of the space charge layer (Csc) is much less than that of the Helmholtz layer. At high applied ω, such as 1 kHz used in the present investigation, the capacitance at the passive layer’s interface with the electrolyte mainly expresses Csc, since the contribution of Helmholtz capacitance to measured electrode capacitance is negligible [221]. As such, the 𝐶𝑠𝑐−2 versus 𝐸 plots describe the semiconductive behavior of the depletion region. Quantitatively, the charge distribution as a function of E can be determined using {E-6.5} for n-type semiconductors [221], where 𝑁𝑑, 𝐸𝑓𝑏, k, T, 𝜀, 𝜀0, and 𝑒 parameters hold the same definitions as those given earlier in section 2.1.      1𝐶𝑠𝑐2 =2𝜀𝜀0𝑒𝑁𝑑(𝐸 − 𝐸𝑓𝑏 −𝑘𝑇𝑒)      {E-6.5} 𝑁𝑑 is determined from the slope of the experimental 𝐶𝑠𝑐−2 vs. 𝐸 plots (i.e. the 2𝜀𝜀0𝑒𝑁𝑑 term), whereas 𝐸𝑓𝑏 is the potential intercept for 𝐶𝑠𝑐−2 = 0. The thickness of the space-charge layer (𝛿𝑠𝑐) for n-type semiconductors is evaluated using {E-2.1} in section 2.1. After passive layer formation through PSP at 0.5 VSCE for 1 h, all the Mott-Schottky tests are scanned from E < Ep to E > Ebd. The 𝐶𝑠𝑐−2 vs. 𝐸 plots exhibit positive slopes in the passive region of interest as shown in Figure B-3 and Figure B-4 in Appendix B, indicating n-type semiconductive behavior. These two figures illustrate examples of the temperature and [HCO3−] dependencies, respectively. The influence from [HCO3−] and temperature results in decreased 𝑁𝑑 with increased [HCO3−] at any fixed temperature, or increased 𝑁𝑑 with temperature at any fixed [HCO3−] (Figure 6-9), corresponding to the findings of [173]. Consistencies are found with respect to slope nonlinearity also, specifically at the critical potential (Ec) of approximately 0.3 VSCE in the 75 ºC data of Figure B-4, where a noticeable slope increase (resulting in a drop in 𝑁𝑑) is observed. Slope nonlinearity in Mott-Schottky plots has previously been attributed to the distribution of donor states over a broad range of energies, the presence of surface states altering  106 the ∆𝐸 across the Helmholtz layer, or surface roughness and nonlinear donor distribution [222], [223]. Since the system of the present study involves oxidized states of Fe, slope nonlinearity is attributed to the influence of iron (III) ions (Fe3+) on the electronic structure of passive layers developing at potentials beyond Ec [224]. This interpretation is thermodynamically supported by Ec, representing the E for the ionization of the deep level in the space charge layer, coinciding with the Pourbaix boundary between Fe2+ and Fe3+ valence states found in [225]. As such, a separate 𝑁𝑑 value is required for E > Ec in the 75 ºC data, represented by 𝑁𝑑2 and evaluated using the slope relationship in {E-6.6b}: 𝑆1 =2𝜀𝜀0𝑒𝑁𝑑1    for E < Ec    {E-6.6a} 𝑆2 =2𝜀𝜀0𝑒(𝑁𝑑1+𝑁𝑑2)   for E > Ec     {E-6.6b}  Figure 6-9: Dependence of Nd1, Nd2, and Efb of passive layer on [NaHCO3] (0.1, 0.25, or 0.5 M) and temperature (25, 50, or 75 ºC)   107 Passive layer charge neutrality and the higher tendency for Fe2+ to Fe3+ oxidation with temperature make slope nonlinearity in the Mott-Schottky plots highly temperature-dependent. However, as opposed to being almost equally influenced over a large range of temperatures as reported in [224], it appears from Figure B-3 that this dependence may be controlled by a minimum temperature threshold, since the change is almost non-existent at temperatures ≤ 50 ºC. It is likely that the higher temperature which impedes robust FeCO3 coverage of the steel as shown by i decay results leaves anodic sites where oxidation to Fe3O4/γ-Fe2O3 occurs. This results in decreased passive layer protectiveness. FeCO3 can also transform directly to Fe2+-Fe3+-oxide (FeO∙Fe2O3 or Fe3O4) at higher anodic E in anoxic conditions or with the influence of O2 in the solutions as shown in {R-6.2} and {R-6.3}, respectively [226]. It appears that the higher temperatures accelerate the kinetics of these transformations through higher alkalinity, supported by inflections becoming more pronounced in higher [HCO3−] solutions of pH 9.3 and 9.2 (Figure B-4). 3FeCO3 +  5OH−  → FeO ∙ Fe2O3 + 3HCO3− + H2O + 2e−  {R-6.2} 6FeCO3 + O2 + 6H2O → 2FeO ∙ Fe2O3 + 6HCO3− + 6H+  {R-6.3}  The larger reductions in 𝑁𝑑2 with [HCO3−] or pH witnessed in the 75 ºC plots in Figure 6-9 supports the suggested Fe3O4 involvement, since increased alkalinity reduces Fe2+ presence thereby resulting in a bigger influence of Fe3+ on the electronic behavior of the overall layer. Even though both Fe3O4 and Fe2O3 have similar crystallographic structures possessing an O2- sub-lattice which is cubic close-packed (ccp) [58], conversions therein have previously been shown to yield structural change and enhance transport pathways through passive layers [227].  The 𝑁𝑑 values of Figure 6-9 are comparable to those for passive layers on other alloy steels or Fe [228], signifying a highly disordered nature for the passive layer. Generally, the breakdown/pitting susceptibility of a passive layer increases with 𝑁𝑑, a correlation which is not violated in the present results; previous i and EIS results reveal that the weaker passive layers  108 formed in dilute solutions or higher temperatures experiences easier breakdown than those in colder conditions with greater [HCO3−]. 𝐸𝑓𝑏 values, critical in determining the positions of semiconductor energy bands as a function of the redox potentials of electroactive ions in the electrolyte, are extracted from the Mott-Schottky plots and reveal an inverse relationship with 𝑁𝑑. Since these positions are governed by the thermodynamic stability of the passive layer and the charge transfer across its interface with the electrolyte, 𝐸𝑓𝑏 is expected to increase for more stable, protective layers. This entails an increase at lower temperatures and higher [HCO3−]. It is noted that 𝐸𝑓𝑏 data for the 75 ºC case is calculated using the average of 𝑁𝑑1 and 𝑁𝑑2. The 𝐸𝑓𝑏 data in Figure 6-9, along with the corresponding increase of 𝛿𝑠𝑐 with [HCO3−], E, and reduced temperatures as shown in Figure 6-10, correspond well with these expectations. They are also in agreement with the steady-state i decay and EIS results presented earlier in section 6.3.1 above.  Figure 6-10: Dependence of (𝛿𝑠𝑐) with respect to E, [NaHCO3] (0.1, 0.25, or 0.5 M), pH, and temperature (25, 50, or 75 ºC) 6.4 Summary In this chapter, the properties and growth of corrosion product on an API X-100 pipeline steel sample in mildly alkaline HCO3− solutions are studied using EIS (under both PDP and PSP conditions), i decay analysis, and Mott-Schottky. Auxiliary electrochemical processes are  109 observed and quantified at lower applied ω in pre-passive, active-passive transition, and transpassive E regions. EECs with nested parallel subsets are used to numerically model the responses during dissolution-adsorption, adsorption-diffusion, and diffusion controlled processes. Adsorption of pre-passive complexes of Fe2+ with HCO3− and OH− is necessary prior to the formation of FeCO3 or Fe(OH)2, respectively. Steel dissolution at active sites is accelerated by Cl−/SO42− at applied E > OCP, contrary to the inhibitive effect these anions are seen to have in chapter 5. This is suggested to be because here, anodic polarization of the specimen proliferates anodic site coverage more so than during free corrosion or OCP, attracting these Cl−/SO42− and other anions more readily. This is also suggested to be what causes increased adsorption and Fe2+ release (charge transfer and hence, corrosion). Conversely, diffusion during pre-passive steps is inhibited by Cl−/SO42−. The applicability of film resistance and capacitance elements to model defective Fe(OH)2 presence on the steel surface is negated, and Rct is used instead to gauge the Z effects of such deposits. Stable passive layer formation and protectiveness is evaluated with a second time constant, revealing improved performance at higher potentials and in Cl−/SO42−-free solutions. Transpassive EIS response is modelled when stable, where diffusive effects reappear as increased transfer channels form within the passive films. Decay of i during PSP, in addition to EIS and Mott-Schottky tests after PSP, all indicate improved passive layer robustness in colder, more alkaline, and HCO3−-rich environments. Passive layer dependencies on [HCO3−] and temperature are supported by the changes observed in semiconductive behavior under the influence of these parameters: 𝑁𝑑 values decrease, 𝐸𝑓𝑏 values increase, and the 𝛿𝑠𝑐 decreases.       110 7 7. Extended immersion studies of the formation of oxide/oxyhydroxide corrosion products on X100 steel5 during exposure to nn-pH 𝐇𝐂𝐎𝟑− solutions with O2, N2H4, and/or 𝐍𝐎𝟑− Corrosion evaluations of HSLA steels in actual and simulated environments identify multi-component corrosion products forming, including complexes of Fe2+ and/or Fe3+ with OH− (Fe(OH)2), CO32− (FeCO3), oxide (Fe3O4 and Fe2O3), and oxyhydroxide (γ/α-FeOOH) as shown in [30], [42], [72]–[78] and the previous chapters above. This formation is repeatedly seen to depend on aspects such as solution constituents, dissolved CO2 or O2 concentrations, and polarization potentials. Oxidation reactions of compounds found in these multi-layer corrosion products are viable at different O2 concentrations ([O2]), such as the oxidation of FeCO3 in nominal O2 conditions [79] versus that of Fe(OH)2 in the complete absence of O2 [32], [80]. This O2 dependency raises concerns regarding the precise influence of [O2] in overall multi-layer corrosion product development. This is important since O2 presence in environments is variable in practical situations such as marine and soil conditions. Corrosion severity is affected by physicochemical properties, depths, temperatures, reduction/depletion rates, and natural cycles (e.g. drying-wetting) [39], [40], [80], [229]. The formation and evolution of corrosion products on HSLA steels upon exposure to O2-containing electrolytes is therefore still of particular interest.  The recognized detrimental influence which O2 has on external pipeline corrosion, combined with its foreseen presence in small quantities at exposed surfaces, is the motivation behind studying the inhibitive behavior of an O2 scavenger in this chapter. Mechanical steps for the removal of trace O2 amounts from critical industrial equipment (e.g. N2 purging or “scrubbing”, vacuum degasification, and heating [230]) are impractical in buried pipeline                                                  5  I. M. Gadala, H. M. Ha, P. Rostron, and A. Alfantazi, Corrosion, vol. 73, no. 3, pp. 221-237,  March 2017.  111 systems. However, chemical scavenging can be feasibly incorporated in such systems at predetermined critical locations. Hydrazine (N2H4) is a known effective O2 scavenger and reducing agent in the chemical, petroleum, and energy generation industries [231]–[233]. The kinetics of its reaction with O2 has formerly been proven in alkaline CO32−-HCO3− systems, exhibiting accelerated depletion at high temperatures (~ 80 °C) [232]. At 40 °C and within a 160 min time frame, its depletion behavior is nearly linear with respect to time. Although being an environmentally unfriendly substance, N2H4 deployment in minute ppm amounts at these temperatures will ensure its complete reaction with O2 through a process which yields simply H2O and gaseous N2. Additionally, N2H4 treatments can serve to remove nitrate (NO3−) from corrosive trapped water [233], a further beneficial outcome considering nitrate’s reported acceleration of carbon steel’s anodic dissolution in HCO3− solutions [183]. For these reasons, N2H4 treatments in concentrations ≤ 1 ppm are chosen for the investigative tests in this chapter. Previous chapters in this thesis have discussed in detail the strong pH influence on corrosion and passivation processes [74], [75], [172], [234]. Many of the nn-pH environments studied are low in electrical conductivity (on the order of 1 – 10 mS/cm) whilst containing the HCO3−, Cl−, and/or SO42− anions commonly found in natural environments. This combination of low electrolyte conductivity and nn-pH is prevalent in many corrosion situations, such as in potable or tap waters, freshwaters, closed-loop piping circuits, crevices, and infrastructure reinforcement of both buried and concrete-embedded types [29]–[33], [141]. The breadth of these applications has not been met with an understanding of a broad range of species which could be present therein. For example, NO3− facilitates carbon steel passivation in concentrated HCO3− solutions [202], yet its influence on HSLA steel corrosion in more dilute lower conductivity solutions has not been widely reported. NO3− could appear in natural environments from a variety of bacterial, industrial/agricultural (e.g. fertilizers), and mineralogical (e.g. naturally occurring Nitratine or NaNO3) sources [235], [236]. An objective of the present chapter is to study the  112 influence of NO3− on the evolution of corrosion rates and products on HSLA steel in nn-pH solutions with conductivities on the order of 1 – 10 mS/cm.  In addition to environmental influences, structural factors of corrosion products (e.g. compactness and porosity) forming on HSLA steel during corrosion are critical as reviewed in section 2.2 of chapter 2 [75], [77], [78], [237]. Films of uniform corrosion products containing few cracks tend to form on homogenous microstructures such as ferrite and bainite [237]. Microstructural effects arising from welding processes greatly differentiate corrosion product growth behaviors. Dense and fine corrosion product layers seen forming on base metals and weld metals with SEM yield the highest corrosion resistance, whereas coarse and porous layers on the heat-affected zones yield the lowest [158], [238]. This is supported by findings from microscopic studies using SVET and localized EIS [79], [239]. This chapter investigates the impact of structural factors of corrosion products on the corrosion behavior of HSLA steel in low conductivity nn-pH environments containing O2 and NO3−.  In this work, combinations of O2, N2H4, and/or NO3− concentrations constitute array environments in which samples are tested for various times and using various procedures (see below). Electrochemical techniques are periodically conducted to determine the evolution of corrosion processes during immersions. Mechanisms for the formation and evolution of corrosion products are proposed. The findings are corroborated by surface morphologies observed on specimens and by chemical compounds identified in the corrosion products. 7.1 Test environments and experimental procedures Similar to the tests in chapter 5, low conductivity test solutions in the present chapter are based on the standard NS4 solution which is deaerated with 5% CO2/95% N2 and has a nn-pH of 6.6 or 6.7. O2 in this reference condition is varied between around 0 ppm (anoxic), 6 ppm, and 20 ppm. The 6 ppm condition represents natural aeration around atmospheric air pressures, whereas the 20 ppm condition represents O2 levels which are significantly higher than expected in natural  113 environments. The latter is simulated here for mechanistic insight. The influence of NO3− is investigated at 0.005 and 0.015 M. In studies of corrosion inhibition through O2 scavenging, environments are treated with 0.5 ppm or 1 ppm N2H4 using a graduated syringe. Only environments with a 6 ppm initial O2 concentration (CO2_i) are treated. All O2 scavenging tests are conducted at both 25 and 50 °C to evaluate the influence of temperature. Table 7-1 summarizes the combinations of [O2] and [NO3−] studied in this chapter with corresponding measured pH and conductivities. Likewise, the temperature and [N2H4] combinations studied in corrosion inhibition are listed in Table 7-2 and Table 7-3 for short-term electrochemical studies (1 h OCP and PDP thereafter) and 168 h immersion tests, respectively. In all tests, the stable pH, CO2_i, and solution conductivities are measured beforehand using a bench top pH meter (±1% rated measurement precision), a new calibrated O2 probe cell, or a calibrated conductivity/TDS meter, respectively.   Table 7-1: List of [O2] and [NO3−] combinations studied in this chapter, based on the reference NS4 electrolyte of A, with corresponding measured CO2_i, pH, and conductivities of solutions Environment 𝐍𝐎𝟑− [M] CO2_i [ppm] pHa Conductivity[mS/cm] A 0 0.2 6.6  1.11 B 0 5.7 7.5 C 0 21.6 8.3 D 0.005 0.2 6.7  1.62 E 0.005 5.8 7.7 F 0.005 20.4 8.6 G 0.015 0.1 6.7  2.61 H 0.015 6.0 7.8 I 0.015 21.2 8.9 a: stable value measured at 25 °C over at least 30 minutes Mainly low polarization electrochemical techniques are used in the tests of this chapter, since many relevant low conductivity nn-pH environments suffer from restriction of CP due to aspects like the high Rs (e.g. tap waters), the shielding of CP current (e.g. under coating disbondments and crevices), or the lack of CP altogether (e.g. internal pipe surfaces). Two test routines are conducted for each test environment in Table 7-1. The first involved a short OCP period < 1 h followed by a PDP scan between -1.25 to 1.25 VSCE at a standard scan rate of 1/6 mV/s. This test routine is also conducted on each environment in Table 7-2. Scans are repeated at  114 least twice to ensure reproducibility. The second routine consists of 24 h immersion with continuous OCP monitoring, where LPR (±10 mV of OCP at 1/6 mV s-1) and static EIS (10,000 - 0.01 Hz, AC amplitude = 10 mV, and 10 points/decade sampling frequency) is conducted at prescribed 1 h intervals. All LPR and EIS measurements are conducted three times to ensure reproducibility of the results at each interval and for each environment.  For the N2H4 treatment tests of Table 7-3, specimens are fully immersed in their corresponding environment for 168 hours. Only environments with a non-zero CO2_i are tested here, as shown in Table 7-3. Treatments of 1 ppm N2H4 are periodically added to the environment every 24 h. After completion of either the first or second test routines, specimens are carefully removed from environments and dried in an inert gas stream of N2 or Argon (Ar) to avoid air oxidation, placed in previously deaerated sealed containers, and stored in a glass desiccator until ex-situ SEM, XPS, Raman Spectroscopy, and/or XRD examinations are conducted, where applicable. Details on the microscopy and chemical characterization equipment used in this chapter have previously been specified in section 4.2.4 on page 49. Table 7-2: List of temperature and [N2H4] combinations studied in short term corrosion inhibition tests of this chapter based on reference NS4 electrolyte of environment AA*, with corresponding measured CO2_i Environment  Temperature [°C] Aeration condition [N2H4]  in ppm CO2_i  [ppm] AA*  25 Deaerated 0 0.2 CC Naturally Aerated 0 5.8 EE 0.5 5.6 GG 1 5.8 BB  50 Deaerated 0 0.1 DD Naturally Aerated 0 4.9 FF 0.5 4.7 HH 1 4.9 Table 7-3: List of temperature and [N2H4] combinations studied in extended 168 h immersion corrosion inhibition tests of this chapter based on environment AA* in Table 7-2, with corresponding measured CO2_i Environment Temperature [°C] [N2H4] in ppm, added every 24 hours CO2_i  [ppm] CC  25 / 5.8 GG 1 5.7 DD  50 / 4.9 FF 1 4.8  115 7.2 Short-term electrochemical tests at immersion times ≤ 1 h  OCP < 1 h The OCP transients stabilize within 5 minutes of immersion, extending until the PDP scans are conducted in the first test routine (at 1 h). Figure C-1 in Appendix C illustrates these stable OCP profiles approaching the 1 h mark, and Figure 7-1 shows the final values at 1 h in terms of [O2]. Different redox conditions at the steel surface depending on [O2], specifically the reduction of O2 in {R-7.1} versus that of H2O previously shown in {R-5.4}, cause clear separation in the profiles.  O2  +  2H2O +  4e−  →   4OH−      {R-7.1}  Figure 7-1: Final OCPs reached after 1 h and 24 h immersion (separate tests) as a function of [O2], with nonlinear fit profiles overlaid OCP becomes more anodic with increased [O2] in the solution at any single [NO3−]. Values are close to those reported for carbon steel in similar pH and HCO3− environments  116 although the alloy and electrolyte compositions are different [40], [240]. In high [O2] conditions it is clear that O2 has the most consequential influence on OCP. In deaerated and 6 ppm O2 conditions though, the influence of HCO3−, other anions, and %CO2 in the purging gas become the controlling factors. In deaerated solutions specifically, alloy and electrolyte composition differences compared to previous reports reveal discrepancies in results. The deaerated OCP values between -0.71 and -0.75 VSCE here are 100 mV more anodic than those reported by Lee et al. [215] and Sherar et al. [68] in similar mixed anion solutions. Anion concentration and alloy composition differences (carbon steel vs. HSLA steel) between previous reports and the present work are influential but likely not greatly so. Instead, the most substantial influence is suggested to be a synergy of electrolyte pH and the surface condition of the specimen. Supporting this is the fact that OCP is less anodic when specimens have a pre-grown CO32− and O2- compound layer from a pre-treatment stage [68], which unmistakably protects the steel surface compared to the bare surface state here. Furthermore, the slightly more acidic pH of solutions here owing to the 5% CO2 content in the purging gas for the deaerated condition, causes increased OCP values compared to 0% CO2 conditions, a finding which has been established by results presented in previous chapters [172].  Using these results, it is beneficial to identify a quick method of determining the change in OCP from a benchmark condition based on the change in [O2] (∆O2). In the present conditions, a nearly linear response to ∆O2 is seen (Figure 7-1) based on the measured OCP in 6 ppm conditions compared to 0 ppm conditions. This is in agreement with OCP transients measured for other structural materials in soil environments [241] although different ion concentrations and temperatures alter [O2] in electrolytes (shown before for NaCl solutions [242]). Here, it can be said that preliminary predictions of X100 OCP upon exposure to nn-pH NS4 can be calculated using an approximate 3 mV anodic shift per unit ppm change in [O2] from the base-line deaerated condition. It should be noted that this 3 mV anodic shift value is only valid for the 0 – 6 ppm [O2]  117 range for times < 24 h. With higher [O2] and exposure times this number is expected to change, in addition to the linearity of the dependency. The effect of increased [NO3−] is generally similar to the influence of higher [O2], causing potential shifting in the anodic direction. It is suggested that the impact of NO3− reduction of {R-7.2} involving H+ in the media, as alluded to by El-Naggar [202], is causing this. However, a consistent trend is not observed here. In both 6 ppm and 20 ppm O2 solutions the 0.005 M profiles are more anodic than 0.015 M profiles upon reaching the 1 h mark, whereas in the deaerated case the 0.015 M profile is more anodic.  NO3− + 3H+ + 2e− → HNO2 + H2O    {R-7.2} The tendency for the reduction of NO2− is very low due to its more stable electronic structure. Thus, it is the single step reduction of NO3− which plays a role in intensifying corrosion in NO3−-containing environments (seen in 24 h OCP transients also). As NO3− reduction competes with O2 reduction here, the reaction of NO3− in lower [O2] environments is aided by its better kinetic feasibility versus that of O2 [202]. In the higher [O2] environments of B, C, E, F, H, and I in Figure C-1, this greater kinetic feasibility is less noticed, likely due to: (a) weakened NO3− reduction kinetics (based on {R-7.2}) in the more alkaline conditions of 6 ppm (pH ≈ 7.5) and 20 ppm O2 (pH ≈ 8.5) compared to the deaerated condition of pH 6.7, and (b) increased O2 reduction kinetics with more O2 availability. Also noted is that NO3− diminished in the {R-7.2} reaction is not resupplied with added NaNO3 during the immersion. In contrast, [O2] remains constant throughout the test due to continuous gas supply. This artifact of the experimental procedure can also contribute to the weakened influence of NO3− in 6 ppm and 20 ppm O2 conditions, since a lower NO3− is advantageous to O2 in the competitive reduction. Nonetheless, this latter factor is less important than pH or [O2], as NO3− will not deplete considerably during 1 h immersions. It becomes a more contributing influence during much longer exposures.    118  Figure 7-2: Influence of N2H4 treatment concentration on 1 h OCP values at 25 and 50 °C Figure 7-2 shows the influence of [N2H4] on the OCP values reached after 1 h of immersion in 6 ppm [O2] conditions at 25 and 50 °C. Figure C-2 in Appendix C shows the OCP transients approaching 1 h immersion in these same Table 7-2 environments investigating N2H4 treatment. It is clear here that the 50 C case with a 1 ppm treatment of N2H4 results in the most cathodic OCP, partly attributed to this condition having the lowest [O2] of all the conditions tested. Figure 7-2 also shows that the drop in OCP is larger with a 0.5 ppm N2H4 treatment compared to a rise in temperature from 25 to 50 C. Yet, the most important revelation from Figure 7-2 is that the OCP of conditions with 1 ppm N2H4 treatment (at either 25 to 50 C) and that of the 50 C condition with 0.5 ppm N2H4 addition stabilize at a value more cathodic than the deaerated case. This alludes to possible further protection of the steel surface with N2H4 treatment in a mechanism additional to the lessening of O2 levels in solution. This revelation is corroborated by findings in subsequent sections using other electrochemical and characterization techniques. N2H4 + O2  → N2 + 2H2O     {R-7.3}  119 The anodic shift between 6 and 0 ppm O2 concentration OCP values is much larger than that between the former and the N2H4 treated conditions of any concentration. This shows that although even a minute 0.5 ppm addition of N2H4 will react with the equivalent molar concentration of O2 present in the electrolyte, doubling this N2H4 amount to 1 ppm will not necessarily double the resulting effect on the redox conditions on the steel. The mechanism by which N2H4 additions lessen O2 presence in the solution is shown in {R-7.3}. Hydrazine is known to be a volatile compound, and its reaction with dissolved O2 is considered to occur quickly. However, based on results observed throughout this chapter, this reaction seems to not occur alone, rather in conjunction with other processes involving N2H4. The reasons for this postulation are alluded to in the following discussion.  A ratio of 1 ppm N2H4 to 1 ppm O2 corresponds to a 0.9375:1 molar concentration ratio. Hence, it is expected that even with a 1 ppm N2H4 addition to 6 ppm O2 conditions, a considerable amount of O2 will remain unreacted. This small change in [O2] may be the reason behind the small change in 1 h OCP behavior when [N2H4] is doubled (Figure 7-2). However, the exact reason why the N2H4 treated conditions with non-zero O2 concentrations exhibit improved corrosion performance over their completely deaerated counterparts is unknown. One possible explanation is that N2H4 additions  1 ppm do not react completely with the O2 in the solution; being a strong oxidizer, the remaining N2H4 might act as an anodic inhibitor by accelerating the formation of a protective corrosion (oxidation) product on the steel surface. This explanation is plausible based on the appearance of weak passivation characteristics in the PDP profiles of N2H4 treated tests after 1 h (see Figure 7-5). Also, film formations are observed on the sample surfaces during extended 168 h immersions with periodic N2H4 treatments (see section 7.3).   PDP at 1 h Figure 7-3 illustrates PDP profiles in nn-pH solutions of varying [O2] with 0.015 M NO3−. PDP profiles for 0 M and 0.005M NO3− solutions are shown in Figure C-3 and Figure C-4,  120 respectively in Appendix C. Also, Figure 7-4a and Figure 7-4b reveals the effect of changing [NO3−] in 6 ppm and 20 ppm O2 solutions, respectively. All PDP results are in good agreement with OCP profiles and support the greater influence of [O2] versus [NO3−]. In 6 ppm and 20 ppm O2 conditions, corrosion rate is not calculated from PDP results due to the absence of prerequisites for Tafel extrapolation such as the existence of well-defined Tafel regions [243]. Conversely, for deaerated environments Tafel extrapolation is appropriate and is conducted using the Tafel Curve Fitting Method [244]. The results are shown in Table C-1, where the trend of increased 𝑖𝑐𝑜𝑟𝑟 with [NO3−] supports the OCP findings. However, the quantitative results of the extrapolation are not intended for stand-alone use, even though anodic and cathodic Tafel constants (𝛽𝑎 and 𝛽𝑐, respectively) correlate well with previous reports on anaerobic corrosion of steel in similar nn-pH environments (e.g. 𝛽𝑎 = 86 mV and 𝛽𝑐 = 200 mV) [68]. These extrapolation results serve as a qualitative corroboration of the 24 h EIS and LPR tests on unpolarized samples in this chapter.  Figure 7-3: PDP in 0 ppm, 6 ppm, and 20 ppm O2 solutions with 0.015 M added NO3−  121  Figure 7-4: PDP in (a) 6 ppm O2 solutions with 0, 0.005, and 0.015 M added NO3−, and (b) 20 ppm O2 solutions with 0, 0.005, and 0.015 M added NO3−  Figure 7-4a indicates that even dilute [NO3−] will contribute to net anodic dissolution, much like Cl− causing the accelerated Fe2+ release from iron specimen at anodic sites [57]. Cathodic polarization branches below OCP for all non-zero [O2] tests show O2 diffusion behavior. At around -1 VSCE for 6 ppm O2 environments (i.e. B, E, and H) this behavior a b  122 subsequently withdraws completely (withdrawal 𝐸: 𝐸𝑤), as shown with the symbols 𝐸𝑤−𝐵, 𝐸𝑤−𝐸, and 𝐸𝑤−𝐻 in Figure 7-3, Figure C-3, and Figure C-4, respectively. Below 𝐸𝑤 the 6 ppm O2 profiles inflect to first follow the deaerated case with matching [NO3−], and then eventually the 20 ppm O2 profile with matching [NO3−]. The latter occurs at a potential where all three [O2] environments overlap (denoted 𝐸𝑜). Increased [NO3−] upwardly shifts this 𝐸𝑜 point, as seen by comparing 𝐸𝑜−𝐹 in Figure 7-3 with 𝐸𝑜−𝐼 in Figure C-4. Since increased [NO3−] accelerates {R-7.3} resulting in higher cathodic reaction rates, the associated i reach that of the shared 𝐸𝑜 faster. The cathodic region in Figure 7-4a supports this, because for equal 6 ppm O2 at different [NO3−], 𝐸𝑤 and cathodic current densities (𝑖𝑐) below 𝐸𝑤 both increase, as shown by the arrow in the figure. The strongest i inflection beyond this is manifested in the 0.015 M NO3− profile.  It is noted that in the NO3−-free conditions in Figure C-3, 𝐸𝑜 occurs beyond the measured cathodic potential range and thus was not captured. Its occurrence at a 𝐸𝑜 < -1.25 VSCE is consistent with the downwardly shift decreased [NO3−] has on 𝐸𝑜 values. This supports the attribution of changing 𝐸𝑜 values on the rate of {R-7.3}, since in NO3−-free environments there is no added i from NO3− and hence the profiles reach the shared 𝐸𝑜 the slowest (i.e. at the lowest potential). This interpretation relies on the fact that changing [NO3−] between 0 – 0.015 M in 20 ppm O2 environments does not influence 𝑖𝑐 values observably; this is shown in Figure 7-4b, where the cathodic profiles C, F, and I have no consistent difference. The dissimilar cathodic behaviors of Figure 7-4a and Figure 7-4b thus follow the results in the previous section and corroborate the suggested two main factors for decreased NO3− influence at higher [O2] (i.e. weakened NO3− reduction kinetics with alkalinity, and increased O2 reduction kinetics with [O2]).     Overall, the observable two-region cathodic profile in the 6 ppm O2 and 20 ppm O2 conditions is attributed to the domination of other cathodic reactions independent of [O2] below 𝐸𝑤. It is established that O2-free HCO3−-based solutions such as the baseline deaerated NS4 environment of Figure C-3 induce the discharge of HCO3−, {R-5.3}, along with H evolution in  123 acidic environments, {R-5.2}, or {R-5.4} in alkaline environments [64], [65], [74], [100]. The corresponding equilibrium E (Eeq) of {R-5.3}, denoted 𝐸𝑒𝑞_5.3, is a function of pH. It is calculated by combining the expressions for the hydration constant of CO2 with the ionization constants and reversible E of H2CO3 and HCO3− discharge reactions (at 20 °C) [245]. At pH 10.3, 𝐸𝑒𝑞_5.3 =−0.85 V𝑆𝐶𝐸, equaling the standard Eeq, hence validating its correctness since [HCO3−] = [CO32−] at this pH in H2CO3-HCO3−-CO32− speciation. The pH range of the present environments of Table 7-1 is between 6.6 and 8.9, thus the corresponding 𝐸𝑒𝑞_5.3 of -0.75 to -0.81 VSCE proves the viability of {E-7.1} involvement at the cathodic E of 𝐸𝑤  ~ -1 VSCE. In NO3−-containing conditions, the reduction reaction described by {R-7.3} ensues for E < 𝐸𝑤 at rates dependent on [NO3−] and pH. Cathodic reactions not concerning O2, namely those involving HCO3−, H+, H2O, and/or NO3−, do not significantly affect i at E > 𝐸𝑤 or E < OCP since O2 reduction is the dominant process in this range. Below 𝐸𝑤 the tendency for 6 ppm O2 and deaerated profiles to overlap in all presented figures, and then eventually include 20 ppm O2 at potentials below 𝐸𝑜, strongly verifies the absence of significant O2-dependent cathodic processes at 𝐸 <  𝐸𝑤. 𝐸𝑒𝑞_5.3 = −5.55 ∙ 10−1 − (2.9 ∙ 10−2) ∙ 𝑝𝐻 [VSCE]   {E-7.1} In the N2H4 treated environments of Table 7-2, it is clear from Figure 7-5 that within the 500 mV region above OCP where anodic processes would be important, N2H4 treatments result in lower i compared to untreated conditions. In fact, the results reveal a short-lived passive region in which i are around 2 orders of magnitude lower than what they would be if there were no N2H4 added. This passive region has boundaries specific to each [N2H4] and temperature, which are easily identified due to the sharp decrease in i to near zero levels. These are termed 𝐸𝑐𝑜1 and 𝐸𝑐𝑜2 based on their attribution to new couples in the anodic and cathodic reactions. These newly introduced couples are not related to [O2] due to their complete absence in aerated but untreated environments of Figure 7-3 and Figure 7-4. Furthermore, the introduction of a new anodic reaction not involving N2H4 or one of its products is highly unlikely, since such a reaction would  124 be Fe-based and Fe is not known to be in a reaction other than Fe/Fe2+ oxidation at these E-pH levels. These newly introduced couples are hence attributed to N2H4 in some way, where the anodic and cathodic current under the influence of N2H4 results in a net i of nearly zero at 𝐸𝑐𝑜1 and 𝐸𝑐𝑜2.   Figure 7-5: PDP profiles of aerated environments at 25 and 50 C, treated with N2H4 Table 7-4: Values of 𝐸𝑐𝑜1 and 𝐸𝑐𝑜2 extracted from PDP plots of N2H4 treated environments, with corresponding ranges of the low-current region Environment Temperature [°C] 𝑬𝒄𝒐𝟏 [mVSCE] 𝑬𝒄𝒐𝟐 [mVSCE] ∆𝑬𝒄𝒐 = 𝑬𝒄𝒐𝟐 − 𝑬𝒄𝒐𝟏 [mV] EE: NS4  + 0.5 ppm N2H4  25 -513 -434 79 GG: NS4  + 1 ppm N2H4 -558 -296 262 FF: NS4  + 0.5 ppm N2H4  50 -495 -385 110 HH: NS4  + 1 ppm N2H4 -443 -292 151  The magnitude of the potential range of the low i region between 𝐸𝑐𝑜1 and 𝐸𝑐𝑜2 is important, since it is a parameter which can describe the efficiency of the inhibition of corrosion  125 with N2H4 treatments. The values of this potential range, termed ∆𝐸𝑐𝑜, for all the relevant Table 7-2 environments are extracted from the PDP plots and are listed in Table 7-4. The results in this table indicate that the most persistent passive region of the conditions tested forms with 1 ppm N2H4 treatment at 25 C. Likewise at 50 C ∆𝐸𝑐𝑜 of the low i region is higher with 1 ppm N2H4 treatment than with the 0.5 ppm. This is expected, since the introduction of the low i region bounded by the 𝐸𝑐𝑜1 and 𝐸𝑐𝑜2 is itself due to N2H4 addition. However, it appears that increased temperature has a mixed effect on the persistence of this low i region depending on the [O2] in the environment. In lower [O2] due to 1 ppm treatment, higher temperature reduces ∆𝐸𝑐𝑜, whereas with 0.5 ppm treatment the range is increased in size. Upcoming EIS results do not offer a clarification to these mixed trends, which also appear in the Table 7-5 results of i below.   Table 7-5: Values of 𝑖𝑎1 and 𝑖𝑎2 extracted from PDP plots of N2H4 treated environments, with corresponding 𝑖𝑚𝑎𝑥  values in low i region and minimum percent decrease in i Environment  Temperature [°C] 𝒊𝒂𝟏 at  -0.2 VSCE  [A cm-2] 𝒊𝒂𝟐 at  0 VSCE  [A cm-2] 𝒊𝒎𝒂𝒙 in ∆𝑬𝒄𝒐 range  [A cm-2] Minimum decrease in 𝒊 [%] EE: NS4 + 0.5 ppm N2H4 25 5.85 x 10-5 1.25 x 10-3 1.02 x 10-3 18 GG: NS4 + 1 ppm N2H4 7.25 x 10-5 4.50 x 10-4 2.31 x 10-5   68 FF: NS4 + 0.5 ppm N2H4 50 1.50 x 10-4 3.66 x 10-3 9.45 x 10-5 37 HH: NS4 + 1 ppm N2H4 2.70 x 10-5 1.00 x 10-3 1.46 x 10-5 46  To further identify the performance of the two levels of N2H4 treatment in different temperatures, the i reached at E within the two active regions before and after the short-lived low i region are extracted from the profiles of Figure 7-5. The E within the active regions are denoted as 𝐸𝑎1 and 𝐸𝑎2 for the region before 𝐸𝑐𝑜1 and the region after 𝐸𝑐𝑜2 (and clear i acceleration) in the profiles, respectively. Their corresponding values are -0.7 VSCE and 0 VSCE, respectively. Coincidentally, the -0.7 VSCE case is a useful E to study since it coincides with the free corrosion potential or OCP of the typical deaerated nn-pH NS4 environment. The values of the i at these E for all the environments tested are extracted from the PDP plots and are listed in Table 7-5. In addition, the i in the low i region between 𝐸𝑐𝑜1 and 𝐸𝑐𝑜2 are studied to identify the maximum i in  126 the region, denoted 𝑖𝑚𝑎𝑥. The values of this parameter are used to evaluate the minimum decrease in i between the active region and the passive-like region of each profile, using {E-7.2}.  %𝑖𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒 = 𝑚𝑖𝑛 {(1 −𝑖𝑚𝑎𝑥_∆𝐸𝑐𝑜_𝑖𝑖𝑎1_𝑖) , (1 −𝑖𝑚𝑎𝑥_∆𝐸𝑐𝑜_𝑖𝑖𝑎2_𝑖)} ∙ 100 {E-7.2} Since the i of the active regions of all profiles are relatively similar to the untreated case, and it is the corresponding i from these regions which are used in the %𝑖𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒 calculation, the results of this procedure are a preliminary measure of the inhibition efficiency of N2H4 treatments. The results of this evaluation are in good correspondence with the trends revealed in the Table 7-4 interpretation, where the best performing situation from a corrosion control standpoint is the 1 ppm N2H4 treated environment at 25 C. This [N2H4] is significantly better than the 0.5 ppm counterpart at the same temperature. Similarly, at 50 C the higher [N2H4] offers more reduction in i. As mentioned earlier, for each [N2H4] pair, the influence of temperature is mixed between favorable and unfavorable to corrosion control, hence leaving the influence of temperature on the efficiency of N2H4 treatment unknown without further more dedicated studies on this subject. 7.3 Surface analysis following 24 h or 168 h immersions Specimens immersed in each environment of Table 7-2 are analyzed with SEM after 24 hours. Ex-situ characterization with XPS, Raman Spectroscopy, and XRD are only conducted on samples with visible corrosion products. NO3− did not affect characterization results greatly, so only NO3−-free spectra are shown. A reddish-orange corrosion “tubercle” forms on the specimen surface in 6 ppm and 20 ppm O2 environments, for every [NO3−] tested. In contrast, no significant corrosion product is observed on the specimens immersed in deaerated environments of any [NO3−]. The tubercles appearing in O2-containing environments cover approximately 50% of the specimen surface for 20 ppm O2 conditions. Coverage decreases in 6 ppm O2 conditions, yet does not change noticeably with [NO3−] for any [O2]. Surfaces of the specimen are imaged at the  127 microscopic scale using SEM, to identify differences in morphologies. The absence of observable corrosion product formations following immersion in deaerated conditions of any [NO3−] is confirmed, as seen in Figure C-5 in Appendix C for the 0.015 M NO3− (0.1 ppm O2, pH 6.7) case.  In the 6.0 ppm O2 environment of pH 7.8 and with 0.015 M NO3−, the tubercle formed (Figure 7-6) contains a mix of equally distributed large and small pores ranging in size from around 0.1 mm (largest pore size under 45x magnification) to < 10 µm (smallest pore sizes under higher magnification). Using image analysis software the pixels in the higher magnification image at and below a fixed brightness level are colored red, as shown in the bottom-right image in Figure 7-6. These red sections cover 30.4% of the total area, and the average size of the largest dimension of the coalesced pore openings they create is 9.0 µm (standard deviation: 2.6 µm). In the 20 ppm O2 environment of pH 8.9 and with 0.015 M NO3−, the tubercle formed is uniform as shown in Figure 7-7, but observably more compact than the 6 ppm O2 case of Figure 7-6. Higher magnification shows that the smallest pore size in the tubercle formed in 20 ppm O2 environments is < 5 µm (top-right corner). Using the same brightness level for image analysis as that employed in Figure 7-6, the red pores of Figure 7-7 cover only 11.7% of the total area, with an average coalesced pore size of 4.7 µm (standard deviation: 0.6 µm) in the largest dimension orientation. At the edges of the tubercles formed in both 6 ppm and 20 ppm O2 environments, the growth of corrosion product directly over the steel surface is seen in Figure C-6 of Appendix C, in a similar manner to that shown in Figure 5-12 of section 5.2.2 above. Here, two different corrosion product structures are identified, one of spherical or cylindrical large crystal structures, and the other of cloud-like formations of small globules. It is the latter structure which grows to form the tubercles observed on the surface in the images of Figure 7-6 and Figure 7-7.    128  Figure 7-6: SEM images of steel specimen surface following immersion in solution containing 6.0 ppm O2 (pH 7.8) and 0.015 M NO3−: (left) low magnification; (top-right) as imaged high magnification with 50 µm scale bar; and (bottom-right) high magnification with 50 µm scale bar analyzed by color threshold  129  Figure 7-7: SEM images of steel specimen surface following immersion in 21.2 ppm O2 (pH 8.9) and 0.015 M NO3−: (left) low magnification; (top-right) as imaged high magnification with 50 µm scale bar; and (bottom-right) high magnification with 50 µm scale bar analyzed by color threshold  130  Figure 7-8: Ex-situ XPS spectra of specimen immersed in 6 ppm and 20 ppm O2 solutions free of NO3−  From the peaks identified in Figure 7-8, ex-situ XPS reveals the involvement of Fe2+ and Fe3+ complexes with oxides and/or oxyhydroxides. Other peaks are artifacts of contamination during drying of the NaHCO3 electrolyte and the influence of the steel alloy substrate (Na 1s, Na KLL, and Fe LMM). The 2p3/2 peak shape strongly indicates the presence of high-spin Fe2+ and Fe3+ compounds due to its broadened appearance compared to Fe(0) or low-spin Fe2+ [246]. Electrostatic interactions, spin-orbit coupling between the 2p core hole and unpaired 3d electrons of the photoionized Fe cation, and crystal field interactions have been shown to cause this peak shape [247]. The 2p3/2 peak spectra for both [O2] cases shown resemble those attributed to Fe2O3 (of α and γ phases alike) and α-FeOOH in separate accounts [246],[248]. Based on the findings of Grovesnor et al. [246], the Fe 2p1/2 and Fe 2p3/2 peaks can identify specific compounds depending on the extrinsic loss as well as the intrinsic structures, both of which are distinct to γ/α-FeOOH, γ/α-Fe2O3, and Fe3O4 products. The intensities of α-FeOOH and γ-FeOOH reported in the high resolution photoelectron spectra are plotted over the zoomed portion of Figure 7-8 [246]. The  131 close match between the added lines and spectra of the products formed in the environments tested here strongly indicates the formation of both FeOOH phases on the steel surface.  Figure 7-9: Ex-situ Raman spectra of specimen immersed in 6 ppm and 20 ppm O2 solutions free of NO3− Surface analysis through Raman Spectroscopy further indicates the primarily Fe-oxide/oxyhydroxide constituents of the multi-layer corrosion product, as seen in Figure 7-9. By reference to standard spectra, the peaks at 390 cm-1 can be attributed to α-FeOOH (reference: 392 cm-1), whereas those at 298 and 527 cm-1 indicate Fe3O4 presence (reference: 297 and 523 cm-1, respectively) [249]. These two compounds are detected in the corrosion product of both 6 ppm and 20 ppm O2 conditions. Conversely, the peaks at 250 cm-1 and 670 cm-1 appear only in the 6 ppm O2 environment. Based on reference standard spectra, the 670 cm-1 peak can be attributed to either α-FeOOH again (reference: 674 cm-1), or could indicate the presence of γ-Fe2O3 (reference: 665 cm-1) [249]. Since the peak at 250 cm-1 is most corresponding to α-Fe2O3 (reference: 247 cm-1), it is evident that Fe2O3 forms in at least the α phase in 6 ppm O2 environments. The clear presence of γ/α-FeOOH in the corrosion product seen in the Raman spectra is further confirmed through distinct XRD peaks of Figure C-7 in Appendix C, appearing mainly around 20°, 30°, 35°, and 47.5°, and 52.5°. The existence of both phases is partially attributed to γ-FeOOH isomerizing  132 to α-FeOOH of a rhombic structure, a process which is dependent on water activity [80], [250]. Distinctly noticeable in the spectra of Figure 7-9 are the higher intensities of peaks associated with Fe2O3 in the 6 ppm O2 vs. the 20 ppm O2 environment, supporting the Raman findings. Peaks in the latter are much lower at ~40° or completely absent at ~33°.  In contrast to Figure 7-6 and Figure 7-7, the surfaces of the specimens immersed in N2H4 treated environments manifest much less corrosion product in total, if observed at the same magnification (Figure 7-10a and Figure 7-10c). At higher magnifications some minor crystal structures (likely FeCO3 based on rhombohedric crystal structure forming in columnar-like fashion [251]) and a uniform base product film appear on the surface, the latter being cracked, holed, or flaked all across the surface. Based on the present system, the four peaks seen in both XRD spectra are attributed to γ-FeOOH at around 29, FeCO3 at around 44 and 65, and substrate Fe at around 82. The base product film seen with SEM is suggested to be causing the weak passivation effect observed in the PDP results of Figure 7-5, which improves the corrosion performance of specimen in N2H4 treated environments to a level beyond that seen even in completely deaerated conditions. It is noted however that the surfaces illustrated in Figure 7-10a and Figure 7-10c are for the 168 h immersion tests which were treated with 7 ppm of N2H4 in total (i.e. 1 ppm every 24 h). Thus, in these conditions, even if all the stoichiometric [N2H4] needed to deplete the 6 ppm O2 is used, some N2H4 will remain. Then, with the O2 fully depleted and the test containers sealed from further O2 ingress, the remaining N2H4 will likely interact with the substrate in a manner similar to an anodic inhibitor, as previously proposed. Nonetheless, based on the 1 h N2H4 treatment results presented earlier in this chapter, it seems that even without a larger stoichiometric concentration ratio of N2H4:O2, some N2H4 will not participate in {R-7.3}, and will instead interact with or oxidize the Fe substrate. Though not a specific objective of this work, further analysis by future studies to discern the chemical characteristics of surface layers forming with N2H4 treatments would be particularly beneficial to support this hypothesis.  133   Figure 7-10: SEM images and corresponding ex-situ XRD spectra of X100 specimen surface following 168 h immersion in environments periodically treated with N2H4 as described in Table 7-3: (a), (b) 25 °C; (c), (d) 50 °C  134 7.4 Periodic electrochemical tests during 24 h immersions in 0 – 20 ppm [O2] and 0 – 0.015 M [𝐍𝐎𝟑−] conditions  OCP OCP transients during extended immersion until 24 h in the environments of Table 7-1 are shown in Figure 7-11 (same legend as Figure C-1). In deaerated conditions, a slight increase in OCP occurs over the 24 h period due to HCO3− presence accelerating active dissolution of the substrate as reported before [73]. Normally, in the absence of O2, ensuing cathodic reduction can only be of either H+ or of reducible corrosion products on the surface. The absence of corrosion product following immersion in deaerated conditions of any [NO3−] (Figure C-5) rules out the latter case, as well as the near impossibility of such reduction occurring in the present conditions, unaided by cathodic polarization or high temperatures [252]. Conversely, acceleration of corrosion in the presence of HCO3− stems from its role as an alternative H+ donor to pure H2O reduction in anaerobic nn-pH; the sluggish nature of the latter is considered to be rate-limiting [253]. H evolution as the cathodic half-cell in deaerated conditions is supported by OCP being noticeably lower than O2-containing conditions, dismissing O2 as the predominant depolarizer. OCPs are also below the (𝐸𝑒𝑞) for H evolution assuming a H2 partial pressure of 1 atm (-0.736 VSCE for pH 8.33) [73]. Corrosion product growth on the steel samples immersed in 20 ppm O2 conditions decreases OCP transients as shown in Figure 7-11. The final OCP values reached in both 1 h and 24 h immersion tests were shown previously in Figure 7-1 above, as a function of [O2]. Results of regression fits on this data are shown in Table C-2 of Appendix C. In a recent study, Lu et al. have also shown that OCP drops occur with greater coverage of α-FeOOH following initial formation stages [73]. The 6 and 20 ppm O2 profiles in Figure 7-11 appear to be converging to the -0.5 to -0.6 VSCE plateau reached after month long immersions in solution of similar constituents [68]. During aerobic stages of cyclic aeration-deaeration junctures, OCPs settle  135 habitually to between -0.5 to -0.6 VSCE shown in the figure with a shaded band [39], [40]. In such studies, the variable causing the OCP change at the boundaries of the cyclic aeration-deaeration intervals is [O2], directly altered with different purging gases. The growth of corrosion product also influences [O2] levels at preferential reduction sites and the kinetics of this reduction compared to that on a bare surface. In the present study, [O2] in the bulk solution is constant throughout the immersion, yet the movement of OCP transients still coincides with corrosion product growth. These results suggest that the presence of these oxide- and oxyhydroxide-based corrosion products on the surface, as shown with surface analyses, influences the [O2] at the steel-environment interface (below the tubercle), hence affecting redox conditions and OCPs. Both γ-FeOOH and α-FeOOH are the main tubercle products formed here in the presence of O2. The tubercle growth is driven by Fe2+ reacting with O2 not reduced to OH− (e.g. {R-7.4}) [80]. Similar tubercle formations containing γ/α-FeOOH have been observed following anaerobic-aerobic cycling of steel in solutions with Cl− [254].  Figure 7-11: Full spectrum OCP transients Table 7-1 conditions during 24 h (same legend as Figure C-1)  136 2Fe2+ +12O2 + 3H2O → 2γ − FeOOH + 4H+   {R-7.4} The growth of this oxyhydroxide corrosion product on the surface gradually decreases the surface area of the exposed steel at which O2 reduction formerly occurred during OCP < 1 h. Preferential O2 reduction occurs on covered areas, not at the interface of the exposed steel with the electrolyte [255], resulting in the intensification of corrosion through separation of anode and cathode sites. Based on mixed potential theory, when the rate of the anodic half-cell reaction branch increases its intersection with the cathodic branch sum occurs at a lower corrosion potential (lower OCP) and a higher 𝑖𝑐𝑜𝑟𝑟. OCP can also go lower from a decrease in the total rate of the cathodic half-cell reactions, yet at a lower 𝑖𝑐𝑜𝑟𝑟. The unmistaken drop in 𝑅𝑃 (indicating increasing 𝑖𝑐𝑜𝑟𝑟) during the 24 h test in 20 ppm O2 conditions (see LPR and EIS results below) is evidence that 𝑖𝑐𝑜𝑟𝑟 increases during this period, hence the associated OCP decrease in Figure 7-11 can be attributed to increased steel oxidation.  Amongst typical oxide/oxyhydroxide products known to form in O2-containing conditions, such as γ-FeOOH, Fe3O4, and γ/α-Fe2O3, O2 reduction occurs preferentially on Fe3O4, not γ-FeOOH or insulating Fe2O3 [202]. The clear detection of Fe3O4 in the corrosion products forming in 20 ppm O2 conditions could therefore partially explain the increased redox reactions occurring at OCP, simultaneously increasing 𝑖𝑐𝑜𝑟𝑟 and moving the OCP in the cathodic direction. The preferential O2 reduction on Fe3O4 (cathodic sites) accelerating oxidation of the steel substrate at separate anodic sites has previously been shown to occur in similar conditions by Sherar et al. [68], supporting the present interpretations. Fe3O4 develops from γ-FeOOH through various system-dependent reactions [80], [256]–[259], accelerated by Fe2+ presence. Specifically, the reduction of γ-FeOOH to Fe3O4 occurs through one of a set of virtually identical reactions either involving Fe directly {R-7.5a} [256], [257], Fe2+ chemically {R-7.5b} [258], or Fe2+ electrochemically {R-7.5c} [80] in low or zero [O2] environments. It is noted that although being  137 influenced by temperature, all these reactions are reported to be viable at an ambient temperature of 25 °C. 8γ − FeOOH + Fe →  3Fe3O4 + 4H2O    {R-7.5a} 2γ − FeOOH + Fe2+ → Fe3O4 + 2H+    {R-7.5b} 8γ − FeOOH +  Fe2+ + 2e− →  3Fe3O4 + 4H2O   {R-7.5c} The strong pH and Fe2+ dependence of the {R-7.5b} reaction corroborates the likelihood it being the reaction which occurs in the present conditions. In experiments conducted by Tamaura et al. [258], it was found that the transformation only occurs at a pH above 7.3, triggered by the adsorption of Fe2+ on the γ-FeOOH. The environments in which Fe3O4 forms here are all above this threshold pH. Furthermore, Fe2+ plays a critical role in lowering the electron transfer barrier of the corrosion product layer [256], or likewise the facilitation of γ-FeOOH “autoreduction” to Fe3O4 from Fe [257]. Either of these effects explains the trend of increasing corrosion rate with immersion observed for the full length of the tests in 20 ppm O2 conditions and until 7 h in 6 ppm O2 conditions (see LPR and EIS sections). γ-FeOOH has a band gap of 2.06 eV between conduction and valence whereas Fe3O4 has a band gap on the order of 0.1 eV [259], representing a stark reduction in electron transfer barrier strength, hence the augment of corrosion rate found here.  Even with the definite O2 presence in 6 ppm O2 conditions, OCP slowly moves in the anodic direction in contradiction with behavior in 20 ppm O2 conditions (shown in Figure 7-1 also). Three main differences are present between the two environments and the resultant corrosion product formations: (1) the [O2] and pH of the bulk solution, (2) the morphology of the γ/α-FeOOH tubercle forming during the immersion as seen in section 7.3, and (3) the presence of Fe2O3 exclusively in 6 ppm O2 conditions as revealed by both Raman Spectroscopy and XRD. It is therefore reasonable to attribute the different OCP transient behavior in 6 ppm O2 versus 20 ppm O2 conditions to one or more of these factors. Through the reciprocal of the methodology  138 used to explain the opposite OCP movement in 20 ppm O2 environments, when the rate of the anodic branch sum decreases its intersection with the cathodic branch sum occurs at a higher corrosion potential (higher OCP) and a lower 𝑖𝑐𝑜𝑟𝑟. This justification is consistent with the increased 𝑅𝑃 values measured in LPR and EIS tests during the same 7 – 24 h period (as shown below), and is the basis for the proposed corrosion evolution mechanism in 6 ppm O2 environments.     EIS Results from the periodic EIS scans during immersion reveal clear [O2] and time dependencies. Responses for different [NO3−] at any single [O2] are similar, so only 0.005 M NO3− results are presented.  The Nyquist plots of Figure 7-12a and Figure 7-12b readily show the stability of Z during 24 h of immersion in deaerated conditions. Nearly complete semicircles which do not change in size significantly with time reveal steady corrosion with no notable diffusion or induction effects at lower ω. For the same reasons outlined in the previous EIS analyses of sections 5.3 and 6.2 (i.e. slight positive inflections in 𝑍 and 𝜃𝐸𝐼𝑆 (not shown) at lower ω indicating adsorption of electroactive species or the relaxation of carbon carrying intermediate species [198]), the same EEC of Figure 5-14 with a nested parallel adsorption section shown in the corner of Figure 7-12a is used to model the results, where component symbols are the same as in earlier uses. This EEC once again achieves very good correlation with measured spectra as seen in the select fits (2 and 8 h) superimposed on the figures, and by 𝜒2 values on the order of 10-4 in Table C-3 of Appendix C. The inflections in Zre caused by adsorption effects are readily observed by comparing the fit curves from the proposed EEC with those from a basic Randles circuit (no nested parallel adsorption circuit) at 10 and 20 h immersions in Figure 7-12a and Figure 7-12b, respectively. The Westing-Mertens method of conversion from Q to C compatible with the two time constant of the EEC is used, the results of which are shown in Table C-3 [209].   139  Figure 7-12: Nyquist impedance representation and fit profiles for deaerated solution containing 0.005 M NO3− (0.2 ppm O2, pH 6.6): (a) at 2, 4, 6, 8, 10, and 12 h immersion times, with proposed EEC; (b) at 14, 16, 18, 20, 22, and 24 h immersion times a b  140  Figure 7-13: Nyquist impedance representation and fit profiles for 5.8 ppm O2 solution (pH 7.7) containing 0.005 M NO3−: (a) at 2, 4, 6, 8, 10, and 12 h immersion times, with proposed EEC; (b) at 14, 16, 18, 20, 22, and 24 h immersion times a b  141 The steady 𝑅𝑐𝑡 values for the deaerated condition are higher than their 6 ppm and 20 ppm O2 counterparts (Table C-4 and Table C-5, respectively) throughout the 24 h immersion, indicating lower corrosion. Increased 𝑅𝑎 with time demonstrates the decreasing attractive strength between the steel and the adsorbate(s), a finding which is once more explainable through basic dilute solute adsorption theory valid here, in which increasing 𝜃𝑎𝑑𝑠 of the surface with the growth of an adsorbate layer decreases the number of sites capable of binding the adsorbate. In section 6.2.1 the Hill-Langmuir model for adsorption was shown as an example; nonetheless, competing adsorption of dissolved electroactive species can also be represented by the classic Everett isotherm {E-7.3}, where 𝜃𝑖𝑠 is the fractional coverage of adsorbate 𝑖 on the solid (single component steel surface), 𝑥𝑖𝑙 is the molar fraction of adsorbate 𝑖 (∑ 𝑥𝑖𝑙 = 1𝑘𝑖=1 ), and 𝐾𝑒𝑞𝑖  is the adsorbate equilibrium constant [210], [260]. Monolayer capacity 𝑎𝑚 is assumed to be the same as surface phase capacity 𝑛𝑚 (i.e. monolayer adsorption analogous to Langmuir isotherm for gas adsorption), thus more filled binding sites on the adsorbent hinders further binding (increases 𝑅𝑎).    𝜃𝑖𝑠 =𝐾𝑒𝑞𝑖 [𝑥𝑖𝑙/(1−𝑥𝑖𝑙)]1+𝐾𝑒𝑞𝑖 [𝑥𝑖𝑙/(1−𝑥𝑖𝑙)]  assuming 𝑛𝑖/𝑛𝑚 ≈ 𝑎𝑖 𝑎𝑚⁄ = 𝜃𝑖𝑠  {E-7.3} Nyquist plots for 6 ppm and 20 ppm O2 nn-pH solutions containing 0.005 M NO3− are shown in Figure 7-13a-b and Figure 7-14a-b, respectively. The plots demonstrate the effects induced by above-nominal [O2] and mildly alkaline pH values. The 6 ppm O2 results during the first half of the immersion (Figure 7-13a) immediately reveal higher corrosion activity compared to the deaerated case, judging from the drop in maximum 𝑍𝑟𝑒 and 𝑍𝑖𝑚 values, respectively, at the same 𝜔 identified. Simultaneous to this reduced 𝑍 is a noticeable change in the shape of the Nyquist profile compared to Figure 7-12a and Figure 7-12b, characterized by: a smaller rise (lower slope) in the high ω region above 1 Hz, an extension in the 𝑍𝑟𝑒 direction at low ω < 0.1 Hz, and an elevated 𝑍𝑖𝑚 at the lowest applied ω of 0.01 Hz. The two latter characteristics are relative to overall 𝑍, so even though quantitatively the values are similar to the deaerated case, since overall 𝑍 values are less their effect is magnified. These 𝑍 features signify mass transfer  142 behaviors, which manifest in increased 𝑍𝑟𝑒 and 𝑍𝑖𝑚 at lower ω and could be accommodated by the EEC shown in Figure 7-13a. The 𝑊 diffusion element has previously been described in {E-6.3} of section 6.2.1. The new 𝑅𝑠𝑙 and 𝑄𝑠𝑙 elements in the EEC of Figure 7-13a represent total surface layer 𝑅 and 𝐶, respectively. These elements are included to gauge the protectiveness of the multi-layer developed, especially at immersion times ≥ 8 h as revealed by OCP behaviors in Figure 7-1 and Figure 7-11. 𝑅𝑠𝑙 and 𝑄𝑠𝑙 are like the adsorption subset in Figure 7-12a, but with the influence of corrosion product presence added to adsorption effects. The 𝑊 in this nested circuit represents the proposed parallel diffusion-adsorption mechanism, and the ensuing corrosion is modelled by the 𝑅𝑐𝑡-𝑄𝑑𝑙 parallel. A better fit is achieved with this configuration than when 𝑊 is placed in series with 𝑅𝑠.  Goodness of fits for 6 ppm O2 condition spectra are highly favorable, with 𝜒2 values on the order of 10-5 and errors in fit below 1% in general. 𝑅𝑠𝑙 of Table C-4 and Table C-5 is considerably higher in magnitude than 𝑅𝑎 of Table C-3, indicating the protectiveness of the corrosion product layers. The value of this element increases with immersion time, an expected result of tubercle and sub-tubercle growth. This is naturally accompanied by decreasing 𝑊 (lower diffusion in the product) as revealed by the results. 𝑄𝑠𝑙 (or 𝐶𝑠𝑙) and corresponding high 𝑛 values demonstrate what appears to be a capacitive region near the specimen surface stemming from the surface layer and adsorptive effects, which behaves similar to the double layer. This region becomes less dominant with immersion time akin to the overall trend in 𝑄𝑑𝑙 or 𝐶𝑑𝑙. The opposite trend is exhibited by 𝑅𝑐𝑡 which yields values lowest at 8 h and noticeably higher thereafter (shown graphically with arrows in Figure 7-13a and Figure 7-13b), a behavior which aligns with the OCP results in section 7.4.1 and the LPR results below. It is suggested that Fe2O3, found in the corrosion product of 6 ppm O2 conditions as shown by XRD and Raman Spectroscopy, is a main cause of this behavior. The insulating nature of γ-Fe2O3 in comparison with Fe3O4 (the latter has a very low energy band gap on the order of 0.1 eV) impedes charge transfer resulting in a  143 decreasing corrosion rate coinciding with its development [259]. Furthermore, the pseudo-morphing of γ-FeOOH to γ-Fe2O3, as seen in {R-7.6}, has previously been shown to be supported by [O2], and occurs in relatively low corrosion rates, two features which do not contradict with the present findings: (1) diffusion processes are occurring in the FeOOH tubercle as identified by 𝑊 in Table C-4, and (2) 𝑅𝑐𝑡 values are near those in deaerated conditions [80], [250].  2γ − FeOOH → Fe2O3 + H2O     {R-7.6} In contrast, the evolution of 20 ppm O2 spectra in Figure 7-14a-b is void of a 𝑍 inversion, manifesting profiles continually decreasing in size. The same characteristics of mass-transfer influence found in 6 ppm O2 conditions are seen here, where the low 𝜔 𝑍𝑟𝑒 extension is unmistaken in the context of the much smaller profile sizes. This region traverses nearly horizontally in many cases, especially at longer immersion times. Despite the difference in |𝑍| evolution with time between 20 ppm and 6 ppm O2 EIS results, the same proposed EEC is suitable; the processes leading up to and causing multi-layer corrosion product growth are the same although magnitudes of these processes and the characteristics of resulting developments are different.   144  Figure 7-14: Nyquist impedance representation and fit profiles for 5.8 ppm O2 solution (pH 7.7) containing 0.005 M NO3−: (a) at 2, 4, 6, 8, 10, and 12 h immersion times, with proposed EEC; (b) at 14, 16, 18, 20, 22, and 24 h immersion timesa b  145  Figure 7-15: Diagram illustrating the influence of ∆[O2] between: (1) 6 ppm O2 and (2) 20 ppm O2 conditions on the evolution mechanism of the Fe-oxyhydroxide tubercle and sub-tubercle Fe-oxide layer(s), where ?̃?𝑝 represents the average pore size in the tubercle  146  Quantitatively, this difference is identified through the results of the three key parameters: 𝑅𝑐𝑡, 𝑅𝑠𝑙, and 𝑊 in Table C-5. 𝑅𝑐𝑡 is initially large and then decreases substantially to end at a level an order of magnitude less than 6 ppm O2 and deaerated condition counterparts. Conversely, 𝑅𝑠𝑙 of the 20 ppm O2 conditions is generally higher than that of 6 ppm O2 conditions, considered to be the influence of the less porous tubercle layer of the former. The development and growth of the γ/α-FeOOH tubercle with this morphology limits diffusion, as verified by the lower 𝑊 values relative to 6 ppm O2 results (except at 2 h). Hence, the values and evolution of the electrochemical parameters are corroborative of the OCP transients and surface analyses discussed above. The absence of the 𝑍 inversion in the 20 ppm O2 conditions further supports the critical role of charge-transfer behavior of the intermediary corrosion products developing in different environments. Only Fe3O4 is detected in 20 ppm O2 conditions, on which preferential charge-transfer continually increases corrosion as opposed to insulating Fe2O3 in 6 ppm O2 conditions [202]. Figure 7-15 is a cumulative schematic summarizing the main differences identified and proposed in this chapter between 6 ppm O2 and 20 ppm O2 conditions, as supported by morphological, product characterization, and electrochemical evidences from experiments.  LPR The 𝑅𝑃 results of the specimen immersed in deaerated and 6 ppm O2 conditions are shown in Figure 7-16a, whereas Figure 7-16b shows the results for immersions in 20 ppm O2 conditions. Error bars show the range of values obtained for the three repetitions conducted for each test. The results agree well with the higher 𝑅𝑐𝑡 (lower corrosion rate) values of deaerated and 6 ppm O2 versus 20 ppm O2 conditions from EIS. Quantitatively, the sums of EIS resistance values in the deaerated environments (𝑅𝑃∗  = 𝑅𝑠 + 𝑅𝑐𝑡 + 𝑅𝑎) compare very well with measured 𝑅𝑝 values in the LPR tests, shown correspondingly in Table 7-6. This is also true for the non-zero [O2] tests in the same table, where 𝑅𝑃∗∗ (sum of 𝜔-independent resistances plus 𝑊, or 𝑅𝑃∗ + 𝑊) is  147 similar to 𝑅𝑝 for each corresponding time and environment. LPR results also indicate increased anodic dissolution with [NO3−], a finding discussed earlier. In all 6 ppm O2 conditions 𝑅𝑃 reaches a minimum at around 7 h followed by increasing corrosion protectiveness of the product layers until 24 h. At times < 7 h the formation of mainly γ/α-FeOOH and Fe3O4 underneath is conducive to corrosion, and diffusivity of species occurs through the porous tubercle as suggested by surface analysis and EIS findings. Yet, at times > 7 h the proposed localized influence of γ-Fe2O3 forming from γ-FeOOH diminishes the overall reduction rate of O2 due to its insulating nature, thereby reducing the corrosion of coupled anodic sites on the unpolarized steel specimen. If formed over Fe3O4, γ-Fe2O3 can also steadily lessen the significance of O2 reduction on the dual-layer with time [39]. The time at which 𝑅𝑃 upsurges in Figure 7-16a coincides with the 𝑍 inversion in the EIS results. In contrast, deaerated conditions exhibit nearly flat 𝑅𝑃 evolution with time and 20 ppm O2 conditions show continually decreasing profiles, both trends matching their corresponding [O2] case behavior in the OCP transients and EIS spectra fits.    Table 7-6: Comparison of Rp values from LPR with Rp* or Rp** from EIS component sums, in solutions with 0, 6, and 20 ppm O2 at 2, 8, 16, and 24 h immersion Condition Resistance parameter 2 h 8 h 16 h 24 h 0.2 ppm O2 pH 6.7 0.005 M NO3−  Rp* [kΩ cm2] 2.22 2.32 2.20 2.37 Rp (LPR) [kΩ cm2] 2.20 ± 0.05 2.15 ± 0.01 2.14 ± 0.04 2.23 ± 0.03       5.8 ppm O2 pH 7.7 0.005 M NO3−  Rp** [kΩ cm2] 1.8 1.6 1.8 2.0 Rp (LPR) [kΩ cm2] 1.54 ± 0.02 1.45 ± 0.03 1.72 ± 0.01 1.88 ± 0.04       20.4 ppm O2 pH 8.6 0.005 M NO3−  Rp** [kΩ cm2] 1.9 1.1 0.9 0.6 Rp (LPR) [kΩ cm2] 1.80 ± 0.05 0.98 ± 0.03 0.59 ± 0.13 0.46 ± 0.05  148  Figure 7-16: Polarization resistance (Rp) vs. immersion time, calculated from periodic LPR scans conducted in: (a) deaerated and 6 ppm O2 solutions containing 0, 0.005, or 0.015 M NO3−; (b) 20 ppm O2 solutions containing 0, 0.005, or 0.015 M NO3−  a b  149 7.5 Summary In this chapter, the electrochemical corrosion behavior of X100 steel specimens is tested in nn-pH NS4 solutions of 0.1 ppm ≤ [O2] ≤ 21.6 ppm, 0 M ≤ [NO3−] ≤ 0.015 M, 0 ppm ≤ [N2H4] ≤ 1, and temperatures of 25 °C and 50 °C. [O2] and [NO3−] are studied at 1 h ≤ immersion times ≤ 24 h, whereas temperature and N2H2 treatment combinations are studied during immersion times ≤ 1 h and following a 168 h extended immersion session. During the first hour of immersions, stable OCP profiles are more anodic with increased O2 and increased [NO3−] owing to the latter’s single stage reduction leading to intensified corrosion. The influences of higher [O2] and [NO3−] are clear in the cathodic region of PDP profiles, driving both diffusion-controlled i and overall cathodic i higher. NO3− imposes an upward shift on the E where H evolution dominates over O2 reduction. Where Tafel extrapolation is appropriate, NO3− is also found to increase icorr as corroborated by all electrochemical tests. N2H4 treatments in ppm concentrations impacts X100 corrosion by shifting the OCP and i of the steel to values less than those exhibited in the 5% CO2 deaerated environment. Increasing the concentration of N2H4 treatment always improves the creation of a weak passivation effect appearing at low anodic polarizations, enhancing corrosion control of the steel through partial covering of the surface (seen in SEM). Upon extended immersion, X100 in deaerated conditions exhibits the least noble OCPs governed cathodically by H evolution, and no corrosion products form. Both 6 ppm and 20 ppm O2 conditions appear to converge to an OCP between -0.5 to -0.6 VSCE through movement in either anodic or cathodic directions, respectively. Tubercles form in all above-nominal [O2] conditions and are characterized as FeOOH of both γ and α phases, with markedly higher porosities in conditions with lower [O2] in solution. Fe3O4, FeCO3, and Fe2O3 are also identified in the corrosion product, the latter being noticeably more extensive with higher porosity tubercles of the 6 ppm O2 conditions. N2H4 treatments in 0.5 ppm – 1 ppm concentrations virtually eliminate the formation of Fe-oxides and Fe-oxyhydroxides during immersion in O2-containing  150 nn-pH NS4 conditions. In cases where it forms, the insulating properties of Fe2O3 causes an upsurge in 𝑅𝑝 and 𝑅𝑐𝑡 values at around 7 h immersion, which otherwise continually decreases due to more enhanced cathodic kinetics (of O2 reduction) occurring preferentially on Fe3O4 beneath the γ/α-FeOOH tubercle. The growth of the multi-layered corrosion product formations including the porous tubercle is modelled with surface layer EEC elements, which also gauge adsorption effects. Lower diffusivity of O2 in the more compact tubercles formed in 20 ppm O2 conditions is revealed by EIS modelling results.               151 8 8. Evaluation of hydrogen evolution, absorption, and diffusion in X100 steel exposed to nn-pH 𝐇𝐂𝐎𝟑− solutions of various ion constituents and temperatures6 Transgranular SCC of underground pipelines occurs in nn-pH environments and is distinct from the classic intergranular (IG) SCC often found in high pH environments [28], [87], [261]–[263]. Experiments in dilute aqueous solutions concluded that transgranular SCC occurs with or without the presence of O2, HCO3−-CO32−, or dissolved CO2 [264]. Nevertheless, it is widely established that transgranular SCC in nn-pH simulated soil environments is greatly enhanced by the presence of dissolved CO2 [21], [50], [265]. Transgranular SCC in underground pipelines is characterized by wide cracks with quasi-cleavage morphology and little branching [28], [87], [88]. The crack sides often suffer significant lateral corrosion resulting in the destruction of the original crack faces. The mechanism of transgranular SCC in underground pipelines is not yet fully understood, albeit a combination of anodic dissolution and HE is proposed to be responsible for the cracking [21], [91], [92], [266]. Evidences supporting the contribution of H to the crack growth are reported in literature [51], [93], [94]; the morphology of the fracture surface resembles those observed in HIC [21], [97], and secondary cracks manifest in isolation of the surface of the specimen, whilst nucleating on bands of pearlitic material [21], [28]. In addition, the increase in the susceptibility of steels to transgranular SCC at more cathodic E also supports the proposed HE mechanism [92], [95], [96].  Electrochemical and corrosion studies of steels in nn-pH environments mainly focus on the anodic process [97], [172], [217], [267]. The cathodic processes on steels, particularly the hydrogen evolution reactions (HER) which might play an important role in controlling transgranular SCC mechanisms in the nn-pH environment, are not thoroughly understood. Knowledge regarding the HER in HCO3−-CO32− solution is mainly obtained from studies of                                                  6  H. M. Ha, I. M. Gadala, and A. Alfantazi, Electrochimica Acta, vol. 204, pp. 18–30, June 2016.  152 internal corrosion of pipelines exposed to CO2 solutions [36], [64], [100], [101], [268], [269]. Such electrolytes are far more concentrated and more acidic than any corrosive electrolyte surrounding pipelines at locations where transgranular SCC is found [21], [24], [265]. In these internal pipeline corrosion studies though, the presence of CO2 increases the corrosion rate of Fe by increasing the rate of the HER [36], [64], [100], [268]. It is widely suggested that the ic is the sum of the i for H+ reduction and H2CO3 discharge, at least in nn-pH and alkaline environments. However, at higher pH the direct discharge of HCO3− becomes important near the corresponding OCP of steels [65]. Understanding the effects of parameters such as solution chemistry, aeration condition, CO2, pH, and temperature on the kinetics of the cathodic reactions, particularly HER on steels, is important in studies of the nn-pH environments and transgranular SCC.  The kinetics of H generation on the steel surface will affect the amount of H ingress into the material and subsequently assist SCC crack growth. Recent studies on H damage manifestations in X100 steels, including HIC under severe cathodic charging in both acidic and alkaline environments [104], [108], [109] and HE under different E ranges in HCO3− solutions [90], emphasize the need to understand the H generation/evolution, absorption, and diffusion kinetics in this modern HSLA material in specific. This chapter addresses the shortage of studies investigating cathodic reaction kinetics particularly on X100 steels in nn-pH electrolytes relevant to underground pipeline corrosion. The effects of environmental parameters, specifically ion constituents, %CO2, temperature, and pH, in addition to surface states of the steel (i.e. bare versus covered with corrosion product) are examined. The significance of H uptake and H diffusion kinetics in X100 exposed to these environments is also studied. This chapter is based on a paper [245] which was published as part of the research work leading towards this PhD thesis. 8.1 Environments and hydrogen permeation conditions HSLA X100 steel samples are used for all the corrosion, permeation, and diffusion tests conducted in this chapter. The microstructural features of this steel have been discussed earlier in  153 this thesis (see section 4.1.1). The standard NS4 solution used in all previous chapters is the baseline test electrolyte for the tests conducted here. Reagent-grade chemicals and ultra-pure deionized water are used to synthesize this reference condition and all other modifications of it. Modified NS4 solutions with varying anion and cation contents are prepared to investigate the effect of these ion species on the corrosion and corresponding H-related phenomena. This is done by increasing the amount of KCl, NaHCO3, CaCl2∙2H2O, or MgSO4∙7H2O in the test environments by 10 times. Similarly, in pursuing a better understanding of CO2 and O2 influences on the overall degradation mechanism, solutions are either left open to air (natural aeration) or purged with pure N2, pure CO2, or the mixture of 5% CO2/95% N2 of the reference NS4 condition. Although not practically expected in service, 100% CO2 purging environments were tested here for mechanistic insight. All purging gases are of a purity ≥ 99.99% (Praxair® Inc.). Generally, purging is conducted for at least 30 minutes in order for environments to achieve their natural pH; yet, in some experiments the pH of the solution is adjusted by adding 1 M NaOH or 1 M HCl.  The influence of temperature is included in selected environments, simulating levels normally encountered in industrial applications in practice, specifically those of buried pipelines. The conductivity of the base-line NS4 solution is 1.11 mS/cm and varied in the range of 1.11 to 7.74 mS/cm in the modified NS4 solutions of this chapter. Table D-1 in Appendix D lists details of the test conditions studied here. Although some of the environments therein do not represent typical field conditions (specifically the pure CO2 purging situations), their inclusion is important to support a thorough understanding of H-related corrosion and transport phenomena on X100 steel sought here. Moreover, actual environments are irregular in nature as shown in section 2.1, depending on a wide variety of physicochemical soil variables; hence, simulations thereof should evaluate individual parameter changes built on an established base-line situation as presented in this study. Modifications from this central situation are amplified to properly magnify and identify the effects caused by the changes involved.  154 In this chapter, PDP experiments are commenced by leaving the specimens to freely corrode at OCP for 1 h. Then, E is swept from 50 mV > OCP to more negative E (cathodic polarization) at a 2 mV/s scan rate. As in all the test polarization tests of this thesis, iR-drop in low conductivity test solutions is compensated by the current interruption technique within the potentiostats used (Gamry Reference 600 is used in the experiments of this chapter). Anodic behavior of X100 in the environments of Table D-1 has been discussed in detail in previous chapters, hence will not be presented here.  Hydrogen permeation conditions Under the diffusion controlled regime, the permeation flux will follow Fickian behavior. The solution for the permeation transient is solved assuming no mobile H exists inside the sample at the beginning of the permeation experiment and the concentrations of H at the entry side and the exit side during the permeation experiment are fixed at a constant value of Co and zero, respectively [160], [270]. The analytical solution gives the relationship between the permeation current density and the effective hydrogen diffusivity as follows [160], [271]:  log[(𝑖𝑝𝑒𝑟𝑚 − 𝑖𝑏𝑔)𝑡0.5] = 𝐶 −𝐿2 log𝑒4𝐷𝑒𝑓𝑓∙1𝑡    {E-8.1}  where iperm is the permeation current density (A/cm2) at time t (s), ibg the background permeation current densities prior to charging (A/cm2), C is a constant, Deff is the effective hydrogen diffusivity (cm2/s), and L is the specimen thickness (cm). At steady state, the permeation current density is expressed by [155], [272]: eff sssnFD CiL         {E-8.2}  where iss is the H permeation current density at steady state (A/cm2), n is the charge of proton (n = 1 equivalent), F is the Faraday constant, and Cs is the subsurface H concentration at the entry side (mol/cm3).   155 The effective hydrogen diffusivity, Deff, could be determined from the plot of log[(i-ibg)t0.5] vs. 1/t according to {E-8.1}. The Deff could also be determined from the breakthrough time tb, which is the time when the permeation transient starts to rise, or from the lagging time tlag, which is the time when the permeation current achieves 63% of the steady state current [155]: 215.3effbLDt         {E-8.3}  26efflagLDt         {E-8.4}  8.2 Results of OCP and cathodic polarization (E < -1.2 VSCE )  The OCP of API X100 steel in different solutions at 20 oC is shown in Figure 8-1. The figure shows the average values of OCPs and the standard deviation determined from 3 measurements. Samples in solutions left open to air exhibited the highest OCP while those in solutions purged with 100% N2 gas exhibited the lowest OCP. Higher concentrations of CO2 in the mixed gas resulted in increased OCP values for the steel. Adding NaHCO3 to NS4 solution reduced the OCP of the steel in all gas atmospheres, whereas adding KCl, MgSO4, and CaCl2 did not significantly alter the steel’s OCP.  Figure 8-1: Open circuit potential of X100 steel in different test solutions of Table D-1 at 20 °C  156  Environmental effects on cathodic polarization results Cathodic polarization curves of the X100 steel sample in test solutions at 20 oC and deaerated with 100% N2 are shown in Figure 8-2a. In these solutions, the cathodic reactions on the steel start at approximately -760 mVSCE and exhibit a Tafel slope of approximately 120 mV/decade in the whole range of E down to -1.2 VSCE. However, the cathodic behavior of the steel in NS4 + 10x NaHCO3 is an exception in that its cathodic polarization curve exhibits higher ic compared to those in other 100% N2 purged solutions, suggesting additional cathodic reactions. The behavior exhibited by the X100 steel sample is similar for all open air solutions (Figure 8-2b), showing a diffusion controlled regime at OCP > E > -1.1 VSCE and approaching a linear behavior at E < -1.1 VSCE. The limiting current density (ilim) in the diffusion controlled regime is approximately 10-4 A/cm2. The Tafel slope of the linear segment at E < -1.1 VSCE is approximately 120 mV/decade. Compared to the ic in 100% N2 NS4 solution, the ic in open air solutions is significantly higher in the E range near OCP but gradually approaches the 100% N2 curve at more negative E.  Similarities are also observed in the cathodic polarization behavior of the X100 steel sample in all 100% CO2 solutions (Figure D-1 in Appendix D). The ic of the steel in this condition is 2 to 10 times higher than that in 100% N2 solutions. Figure 8-3a shows the cathodic polarization curves in NS4 solution with different purging gas, plotted in the same graph. An increase in ic is observed at all cathodic E when the 5% CO2 gas mixture is used (vs. 100% N2), and ic values further increase as the concentration of CO2 in the purging gas becomes 100%. The effect of CO2 on the cathodic reaction kinetics is clearly seen one more time in Figure 8-3b when the gas atmosphere is switched during PSP experiments at -0.8, -0.9 and -1.0 VSCE. In 100% N2 NS4 solution, ic at -0.8, -0.9 and -1.0 VSCE reaches stable values of approximately 1, 20 and 60 µA/cm2, respectively. The stable ic increases and establishes new, higher stable values as the gas atmosphere is changed from 100% N2 to 5% CO2 to 100% CO2.   157  Figure 8-2: Cathodic behavior of X100 steel in: (a) deaerated (100% N2) NS4 and modified NS4 solutions at 20 °C; (b) open air NS4 and modified NS4 solutions at 20 °C  a b  158      Figure 8-3: Cathodic behavior of X100 steel in: (a) NS4 solution under different gas atmospheres; (b) NS4 solution under different gas atmospheres during constant potential holds at -0.8, -0.9 and -1.0 VSCE a b  159   Figure 8-4: Cathodic polarization curves of X100 steel in: (a) 100% CO2 NS4 solution at 20, 40, and 60 °C, natural pHs; (b) NS4 solution with adjusted pH at 20 C a b  160 The cathodic behavior of the steel in 100% CO2 NS4 and NaHCO3-added NS4 solutions at 20, 40 and 60 °C are presented in Figure 8-4a and Figure D-2a in Appendix D, respectively. In general, cathodic kinetics increase with temperature, indicated by a shift in the polarization curves to higher ic levels. The polarization curves of X100 steel in NS4 solution at different pH levels is shown in Figure 8-4b. In 100% N2 solution, the curves are slightly shifted in the direction of higher ic as the pH is increased from 6.5 to 8.4. A similar behavior occurs in 100% CO2 solution as the pH is increased from 5.7 to 6.5. This behavior is reproducible in replicated experiments. In 100% N2 NaHCO3-added NS4 solution (Figure D-2b in Appendix D), the cathodic polarization curve in the natural pH 8.4 solution shows significantly higher ic compared to the curve in NS4 pH 8.4 solution. However, this increased ic behavior disappears when the pH of the NaHCO3-added NS4 is adjusted to 5.7.   Surface deposit effects on cathodic polarization results The nature of the deposit on samples after a 1 h PSP at -1.2 VSCE in 100% CO2 NS4 solutions is examined by SEM and EDX/EDS. Small spherical particles approximately 5 µm in diameter are observed on the sample surface as shown in Figure 8-5a. After 10 h, the sample surface is completely covered by a continuous layer. EDX/EDS analysis indicates the existence of C, O, Mg, and Ca in the composition of the deposit, which is in agreement with the chemical nature of the commonly found CaxMg1-xCO3 scale on steels in this environment [273]–[275]. The cathodic behavior of the scale-covered X100 steel in NS4 solution is shown in Figure 8-5b. The cathodic polarization curve of the sample after a 1 h PSP session at -1.2 VSCE is shifted to ic lower than those of the freshly polished sample. In addition, the extent of the shift in the cathodic polarization curve increases as the duration of the PSP at -1.2 VSCE increases from 1 to 10 h.  161  Figure 8-5: Characterization of the deposit on X100 steel held at -1.2 VSCE in 100% CO2 NS4 solution: (a) SEM photo after 1 h; (b) EDX/EDS spectrum of a particle formed 8.3 Results of hydrogen permeation experiments  The H flux transient in the X100 thin sheet sample when a charging current density (icharge) of -500 µA/cm2 is applied on the entry side is shown in Figure 8-6a. After a breakthrough time, the iperm increases to a steady state value of iss = 0.4 µA/cm2 in approximately 9000 s. The tb and the tlag of the permeation transient are 1594 s and 3398 s, respectively. Hence, Deff in X100 steel is determined to be 4.1x10-7 and 4.9x10-7 cm2/s using {E-8.3} and {E-8.4}, respectively. Hydrogen diffusivity could also be determined from the slope of log[(iperm-ibg)t0.5] vs. 1/t which returns a value of 4.3x10-7 cm2/s for Deff (Figure 8-6b). The average Deff calculated by these three different methods yields a value of 4.4x10-7 ± 4x10-8 cm2/s. a b  162  Figure 8-6: (a) Permeation current transient of X100 steel exposed to 100% CO2 NS4 solution (20 oC) at a hydrogen charging current density of -500 µA/cm2; (b) determination of Deff of the X100 steel from permeation current transient using {E-8.1}, where the slope of the log((iperm-ibg)×t0.5) vs. 1/t  plot is obtained by linear fitting  The response of icharge and iperm when the purging gas in the entry compartment is switched from 100% N2 to 5% CO2 to 100% CO2 is plotted in Figure 8-7. Here, icharge values are kept negative to show that they are a measure of cathodic/reduction reaction rates versus that of a b  163 the anodic/oxidation reaction rates measured on the exit side (i.e. iperm). The magnitude or absolute value of icharge on the entry side (i.e. |icharge|) increases from approximately 10 to 20 to 95 µA/cm2 as the entry compartment is purged with 100% N2, 5% CO2, and 100% CO2, respectively. These values are within around 10 µA/cm2 of the corresponding cathodic current densities of separate tests in the same conditions (i.e. NS4 solution and -0.9 VSCE charging potential, shown in Figure 8-3b). These are overlaid on Figure 8-7 with dashed lines. Similarly, iperm on the exit side increases from approximately 0.03 to 0.05 to 0.1 µA/cm2 corresponding to the switching of the purging gas in the entry compartment from 100% N2 to 5% CO2 to 100% CO2, respectively.  Several remarks could be made about this data. Regarding the increased amount of H generated on the steel sample (ic at the entry side) at higher %CO2, this is primarily due to the increased acidity of the solution (i.e. increased [H+]). At decreased pH levels the kinetics of H evolution through H+ reduction increases, partly due to the increased [H+] in the solution. Perhaps more important than the increased [H+] though is the increase in Eeq of the HERs with acidity. It is shown in {E-8.8}, {E-8.9}, {E-8.12}, and {E-8.13} of section 8.4.2 below that in more acidic solutions, the Eeq of H evolution increases for all the available cathodic reaction paths. Thus, even at a constant charging potential of -0.9 VSCE, the overpotential driving the cathodic reactions will be larger at lower pH due to the corresponding increases in Eeq of the HERs involved. Inherently, due to additional HERs involving H2CO3 or HCO3− becoming viable through increased Eeq at lower pH, and since the charging potential remains constant at a level more negative than these reversible potentials, the ic values measured at the entry side is augmented by the additional involvement of these reactions. The influence of CO2 on the kinetics of HERs will be elaborated further in subsequent sections below. The results in Figure 8-7 also reveal that the amount of H absorbed and transported through the steel sample in NS4 solutions is only 0.1 to 0.3% of the H generated at the charging side. Hence, most of H is harmless to the steel and bubbles out. Finally, it takes some time for iperm on the exit side to respond to the change in icharge on the entry side as the result of the delay associated with the diffusion of H through the sample.  164  Figure 8-7: The current response in the H entry and exit sides (cathodic icharge and anodic iperm, respectively) during the permeation experiment in NS4 solution. The purging gas in the charging compartment is switched from 100% N2 to 5% CO2 to 100% CO2 Figure 8-8 shows the transients of both icharge and iperm as the applied cathodic E on the entry side is changed from -0.85 to -0.95, -1.05, and finally -1.15 VSCE. The magnitude of the H icharge density increases from approximately 70 to 200 µA/cm2 for the full spectrum of the applied E change (i.e. from -0.85 to -1.15 VSCE). This indicates an increase in H generation kinetics on the entry side of the steel sample at more negative cathodic E. On the exit side, iperm also increases from 110 to 310 nA/cm2 correspondingly. However, after H charging on the entry side is halted (i.e. left unpolarized), iperm does not decay to zero but only decreases to approximately 60 nA/cm2. Only when all the solution in the entry compartment is drained out and the sample is left exposed Overlaid from Figure 8-3b: Cathodic current density of X100 steel in NS4 solution under corresponding gas atmospheres during constant potential hold at -0.9 VSCE  165 to air does iperm decay to zero. This indicates H entry in the sample continues to occur even at OCP.  Figure 8-8: The current response at the H entry and exit sides during the permeation experiment in 100% CO2 NS4 solution at 20 °C. The surface potential of the steel sample at the entry side is first switched from -0.85 to -1.15 VSCE with 100 mV/step/4 h then is left at unpolarized (i.e. OCP) before the solution in the entry compartment is drained out to leave the sample open to air 8.4 Discussion of OCP, cathodic PDP, and permeation results  Hydration of CO2 and [HCO3−-CO32−] in CO2-H2O system Gaseous CO2 dissolves in water to form CO2(aq) of {R-4.1} at a mole fraction solubility of 𝑥𝐶𝑂2(𝑎𝑞)  = 6.15x10-4 at 25 °C [276], based on a Henry’s constant for the pCO2/𝑥𝐶𝑂2(𝑎𝑞) ratio of 1.63 x103 atm ∙ molH2OmolCO2(g) (see Table D-2 in Appendix D) [277], [278]. The dissolved CO2 reacts with  166 water to form H2CO3, which dissociates to form HCO3− as shown previously in {R-4.1} and {R-4.2}, respectively. HCO3− dissociates to form CO32− according to {R-8.1} below: HCO3− ↔ H+  +  CO32−       {R-8.1} The temperature dependence of the hydration constant of CO2 forming H2CO3 in reaction {R-4.1}, K0, the ionization constant of H2CO3 forming HCO3− in {R-4.2}, K1, and the equilibrium constant of HCO3− forming CO32− in {R-8.1}, K2, are expressed by [279]: 𝑝𝐾0  = −𝑙𝑜𝑔𝐾0 = −2385.73𝑇− 0.0152642 ∙ 𝑇 + 14.0184  {E-8.5} 𝑝𝐾1  = −𝑙𝑜𝑔𝐾1 = 3404.71𝑇+ 0.032786 ∙ 𝑇 − 14.8435  {E-8.6} 𝑝𝐾2  =  −𝑙𝑜𝑔𝐾2 =2902.39𝑇+ 0.02379 ∙ 𝑇 +  6.4980  {E-8.7} The above equations are practically equivalent to the equations reported by other researchers [280], [281]. The equilibrium concentrations of CO2, H2CO3, HCO3−, and CO32− in the CO2-H2O system could thus be fully determined from the pressure of CO2 gas and the system temperature, with the latter defining the values of the necessary equilibrium constants listed above. As an example, the speciation diagram of the CO2-H2O system at 25 °C calculated based on thermodynamics using Medusa® software is plotted in Figure 8-9. The diagram shows that pH < 4, H2CO3 is the dominant species. As the solution pH increases from 4 to 8, the fraction of H2CO3 to the combined concentrations of H2CO3, HCO3−, and CO32− decreases to approximately zero while the corresponding fraction of HCO3− increases and approaches a maximum at pH 8.35. At pH < 8 the concentration of CO32− is several orders of magnitude less than either H2CO3 or HCO3−, hence its fraction to the overall sum of concentrations is practically zero. In more alkaline solutions, the fraction of HCO3− decreases and CO32− gradually becomes dominant, with the concentration of both species equaling one another at around pH 10. It should be noted that although the concentration of HCO3− increases at pH levels more alkaline than 8.35, the fraction of HCO3− to the overall sum of concentrations decreases with further alkalinity since the  167 concentration of CO32− increases at a faster rate, and the concentration of H2CO3 remains constant. This behavior is apparent in the concentration plots of Figure 8-10a to Figure 8-10c.  Figure 8-9: Speciation diagram of the CO2-H2O system calculated at 25 C using Medusa software  The concentration of each carbonate species in the solution at 20, 40, and 60 °C in a 1 atm (~100 kPa) CO2 atmosphere determined from the aforementioned equilibria is shown in Figure 8-10. The concentration of H2CO3 and dissolved CO2 are independent of pH, as expected due to the absence of H+ in the dissolving of CO2(g) and its hydration forming H2CO3, {R-4.1}. Conversely, the concentrations of HCO3− and CO32− increase as pH increases. It is important to note that this increase is in fact exponential, not linear, due to the log scale of the y-axis. In acidic and nn-pH conditions, dissolved CO2 accounts for most of the total concentration of carbonate species whereas in alkaline solutions HCO3− and CO32− are dominant. The concentration of H+ is also plotted in the same graph for comparison. At low pH, [H+] is much higher than [HCO3−] and [CO32−] but at nn-pH and higher, the later species are more abundant in the solution. Using these concentration plots, specifically Figure 8-10a at 20 C, it becomes apparent for Figure 8-9 how the fractions of H2CO3 and HCO3− concentrations are equal at around pH 6. Likewise, it can be seen how the fractions of H2CO3 and CO32− are equal at around pH 8, corresponding to the HCO3−   168 fraction being almost 1 due to [HCO3−] >> [CO32−] and [HCO3−] >> [H2CO3]. Finally, the fractions of HCO3− and CO32− are equal at around pH 10, and since the concentration of either specie is much larger than [H2CO3] at that pH, the fraction of H2CO3 is negligible there. To illustrate these relationships numerically, the concentrations of H2CO3, HCO3−, and CO32− are extracted from Figure 8-10a at four pH values: 4, 6.3, 8.3, and 10.3. Then, the fraction of each specie at each pH is calculated through the division of the concentration of that respective specie by the total sum of [H2CO3], [HCO3−], and [CO32−]. The results are shown in Table 8-1 below, with the fractions matrix being numerically consistent with the graphical speciation in Figure 8-9.   Table 8-1: Approximate concentrations and % fractions of H2CO3, HCO3−, and CO32− in the CO2-H2O system for a temperature of 20 C and 1 atm CO2  pH Concentration [M] Fraction H2CO3 HCO3− CO32− Total H2CO3 HCO3− CO32− Total 4 1 x 10-6 5 x 10-9 2 x 10-15 1x 10-6 99.5% 0.5% 0.0% 100.0%  6.3 1 x 10-6 1 x 10-6 5 x 10-11 2 x 10-6 50.0% 50.0% 0.0% 8.3 1 x 10-6 1 x 10-4 1 x 10-6 1 x 10-4 1.0% 98.0% 1.0% 10.3 1 x 10-6 4 x 10-3 4 x 10-3 8 x 10-3 0.0% 50.0% 50.0%  In addition to the pressure of CO2, the temperature of the CO2-H2O system determines the equilibrium concentrations of the carbonate species. This mainly is due to the temperature dependent solubility of CO2; when temperature increases, the solubility of CO2 in aqueous solutions decreases significantly. For instance the mole fractions of CO2 in the solution at 20, 40, and 60 °C in a 100 kPa CO2 atmosphere are 0.70, 0.43, and 0.30, respectively [276]. Meanwhile pK1 and pK2 only slightly decrease and pK0 slightly increases as the temperature increases. Overall, these changes result in a shift of all concentration distributions vs. pH to lower concentration as temperature increases (Figure 8-10). This means that the concentrations of H2CO3 and carbonate species in higher temperature solutions are lower than those in lower temperature solutions at the same pH. As seen in the next sections, the concentrations of H2CO3 and carbonate species in solution play a crucial role in the kinetics of HER.   169   Figure 8-10: Concentration of dissolved CO2 and carbonate species in solution with respect to pH, with pCO2 = 1 atm at: (a) 20 °C; (b) 40 °C; and (c) 60 °C a b c  170  HER on X100 at OCP and cathodic potentials In acidic CO32−-free and O2-free solutions, H+ reduction shown earlier in {R-5.2} is the dominant cathodic reaction, with an Eeq at 20 °C expressed by [124]: 𝐸1  =  −241 –  58𝑝𝐻 (𝑚𝑉𝑆𝐶𝐸)     {E-8.8} At higher pHs, H evolution by direct reduction of water, as shown previously in {R-5.4}, becomes more important as the concentration of H+ is negligible compared to the concentration of OH− [124], [268]. The Eeq of this reaction at 20 °C is expressed by [124], [268]: 𝐸2 = –  58𝑝𝐻 (𝑚𝑉𝑆𝐶𝐸)      {E-8.9} In solutions containing dissolved CO2 and HCO3−, additional cathodic reactions might occur by the direct discharge of these species according to {R-8.2} below and {R-5.3} from chapter 5  (reproduced here as {R-8.3}) [36], [64], [100], [268], [282]: H2CO3 + e− →  12H2 + HCO3−      {R-8.2} HCO3− + e−  →  CO32− +12H2      {R-8.3} The Eeq corresponding respectively to {R-8.2} and {R-8.3}: 𝐸3 = −622 − 58𝑙𝑜𝑔[𝐻𝐶𝑂3−][𝐻2𝐶𝑂3](𝑚𝑉𝑆𝐶𝐸)    {E-8.10} 𝐸4 = −856 − 29𝑙𝑜𝑔[𝐶𝑂32−][𝐻𝐶𝑂3−](𝑚𝑉𝑆𝐶𝐸)     {E-8.11} Combining {E-8.5}, {E-8.6}, {E-8.10}, and {E-8.11} gives the Eeq at 20 °C: 𝐸3 = −251 − 58𝑝𝐻 (𝑚𝑉𝑆𝐶𝐸)     {E-8.12} 𝐸4 = −555 − 29𝑝𝐻 (𝑚𝑉𝑆𝐶𝐸)     {E-8.13} Figure 8-11 illustrates the OCP values of API X100 steel in NS4 and modified NS4 solutions with respect to pH, under different purging gas atmospheres at 20 °C. The Eeq for HER are superimposed onto the plot. Figure 8-11 indicates that H evolution is feasible even in aerated solutions at the OCP or free corrosion condition because the Eeq of direct water reduction, {R- 171 5.4}, is more positive than the OCPs of the steel. It is consistent with the result of the H permeation experiment which shows a H flux at OCP (Figure 8-8). In O2-free (100% N2) solutions, in addition to the direct water reduction, H+ reduction through {R-5.2} may also occur at the freely corroding surface. However, because the pH of the O2-free (100% N2) solution is 8.8, the [H+] is low and so is the kinetics of H+ reduction. Therefore, the direct water reduction reaction is the dominant cathodic reaction in this condition. In O2-free (5% and 100% CO2) solutions, H2CO3 discharge through {R-8.2} is possible over the whole pH range while the discharge of HCO3− is only possible in solutions with pH below 5.4. However at this low pH, the concentration of HCO3− is negligible as indicated in the speciation diagram and the carbonate species concentration plot (Figure 8-9 and Figure 8-10, respectively). Therefore H evolution due to HCO3− discharge in NS4 solution at OCP is unlikely.   Figure 8-11: OCP of X100 steel in NS4 and modified NS4 solutions under different purging gas atmospheres at 20 C {R-8.3} {R-5.2} {R-5.4} {R-8.2}  172 It is noticed that adding HCO3− to the solution significantly reduces the OCP of the steel, particularly in open air and deaerated (100% N2) NS4 solutions (Figure 8-1). This is mainly due to the increase in the anodic dissolution kinetics in more concentrated HCO3− environments as can be seen from the shift of the anodic polarization curves towards higher i in the figures of previous chapters. Another reason for the OCP decrease in HCO3−-added solutions is the lower [H+] in these solutions as a result of the buffering effect of HCO3−. Solution pH increases by almost a full unit when NaHCO3 is added to the NS4 solution, as shown in Table D-1. In practice, for a buried pipeline application the E of the pipe surface is affected by CP and might be polarized to much more negative values. In this case, all four HERs will be possible, but each of them will have a different contribution to the overall kinetics of the cathodic process. During cathodic polarization in the results shown above, E is swept from OCP to -1.2 VSCE, which makes all four cathodic reactions in {R-5.2}, {R-5.4}, {R-8.2}, and {R-8.3} possible. In O2-free (100% N2) NS4 solution (Figure 8-2a), the cathodic polarization curve exhibits a Tafel slope of 120 mV/decade. As discussed earlier, H+ discharge is not significant in this solution due to its alkalinity (pH 8.8). Therefore, this Tafel behavior is attributed mainly to the HER of direct water reduction, {R-5.4}. Similar Tafel slopes for the cathodic polarization curve of steel in HCO3−-CO32− solutions have been observed by other researchers [268]. In open air solutions (Figure 8-2b), the acceleration of cathodic kinetics near the OCP is attributed to the reduction reactions of dissolved O2 in the open air solutions. Here, the Tafel behavior at the more negative potential range, which overlaps with the polarization curve in deaerated (100% N2) conditions, is again attributed to the HER of direct water reduction. In O2-free, 100% CO2 solution (Figure D-1 in Appendix D), the increase in the cathodic process observed could be attributed to the HER of H2CO3 and HCO3− discharge ({R-8.2} and {R-8.3}, respectively).  The effect of pH on the cathodic polarization provides some interesting insight into the nature of the cathodic reactions in the potential range from OCP to -1.2 VSCE. Cathodic kinetics in  173 all solutions is suppressed when the pH of the solution decreases from 8.4 to 5.7 given the same purging gas is used (Figure 8-4b). The effect is particularly pronounced when the pH of NaHCO3-added solution is adjusted from 8.4 to 5.7 (Figure D-2b). The cathodic polarization curve in the NaHCO3-added solution after the pH is adjusted is similar to the polarization curve in HCO3−-free solution. This indicates that the additional HER reactions in CO32−-containing solutions are suppressed by decreasing the pH. Because the [H2CO3] is independent of pH (Figure 8-10), the suppression of the HER reaction when pH decreases from 8.4 to 5.7 is attributed to the decrease in the [HCO3−] in the solution. This suggests the HER due to HCO3− discharge, {R-8.3}, is probably the dominant cathodic reaction over the discharge of H2CO3, {R-8.2}, in the OCP > E > -1.2 VSCE range, so a change in the [HCO3−] due to pH directly affects the cathodic kinetics. The shift of the cathodic polarization curve to higher ic when the temperature increases from 20 to 60 °C (Figure 8-4a) indicates a strong temperature dependence of the HER kinetics on X100 steel. It is shown in Figure 8-10 that the solubility of CO2 in aqueous solution decreases when the temperature increases. As a result, the concentration of H2CO3 and carbonate species also decreases when the temperature increases. Therefore, the temperature dependence of the HER kinetics cannot be explained based on the reduction/discharge reactions of carbonate species. This points to the role of the H+ reduction reaction which is facilitated by lower pH as the result of increased temperature. It is noted that the pH of the solution decreases from 6.5 to 5.1 in 100% CO2 NaHCO3-added NS4 solution and from 5.7 to 4.7 in 100% CO2 NS4 solution when temperature increases from 20 to 60 °C. It is the increase in the [H+] that causes the increase in the HER kinetics observed in Figure 8-4a. In addition, Figure 8-10 also shows that in the pH range between 5.7 and 4.7, H+ is dominant over HCO3−.  The role of CO2 in the cathodic process is demonstrated in the gas switching experiments as well as in the PDP experiments. It is clearly shown that ic significantly increases as the purging gas atmosphere is switched from 100% N2 to 5% CO2 to 100% CO2 (Figure 8-3b and Figure 8-7).  174 This is in agreement with studies of CO2 corrosion of pipeline steels which conclude that corrosion of steels is accelerated in the presence of CO2 [64], [100], [268]. Also, the role of calcareous deposits on the H evolution kinetics on steels can be seen when comparing the cathodic polarization of a freshly-polished sample and samples after cathodic holding at -1.2 VSCE (Figure 8-12). The cause of the shift to lower ic in the cathodic polarization curve of samples after cathodic PSP appears to be due to the deposition of a CO32− scale on the sample surface, which results in the inhibition of the cathodic reactions. Such scale is confirmed to be present in SEM images shown in Figure 8-5. The deposition of a CaCO3 layer might block the cathodic reaction sites and therefore reduce the HER kinetics.  Figure 8-12: Cathodic polarization curves of scale-covered X100 steel in NS4 solution at 20 C   Hydrogen permeation in X100 exposed to nn-pH HCO3− solutions From the permeation experiments, Deff in the X100 steel is determined to be 4.4x10-7 ± 4x10-8 cm2/s (Figure 8-6a and Figure 8-6b). This value is comparable to the H diffusivity normally measured in pipeline steels of difference grades [104], [161], [261], [262], [283], [284]. The amount of H absorbed in X100 steel at OCP in 100% CO2 NS4 solution can be quantified from iss using {E-8.14} which is derived from {E-8.2}:  175 ssseffi LCnFD        {E-8.14} Using the average value of Deff calculated from the range mentioned above, the sample thickness of 1 mm, and iss = 60 nA/cm2 determined from Figure 8-8, the diffusible hydrogen concentration in X100 steel exposed to 100% CO2 NS4 solution in the freely corroding condition is 1.4x10-7 mol/cm3 or 1 atomic ppm (appm).  Table 8-2: Summary of steady state permeation current densities and diffusible hydrogen concentrations in the X100 steel exposed to the NS4 solution under different charging conditions Charging condition iss (µA/cm2) Cs (mol/cm3) Cs (appm) -0.9VSCE, 100% N2 0.03 7.0x10-8 0.5 -0.9 VSCE, 95% N2 0.05 1.17x10-7 1.3 -0.9 VSCE, 100% CO2 0.1 2.33x10-7 1.7 OCP, 100% CO2 0.06 1.4x10-7 1 -0.85 VSCE, 100% CO2 0.11 2.57x10-7 1.8 -0.95 VSCE, 100% CO2 0.19 4.43x10-7 3.2 -1.05 VSCE, 100% CO2 0.25 5.83x10-7 4.2 -1.15 VSCE, 100% CO2 0.29 6.44x10-7 4.8  The values for diffusible hydrogen concentration in X100 steel exposed to NS4 solution at different applied cathodic E and under different purging gas atmospheres are calculated in the same way. A summary of the resulting diffusible hydrogen concentrations is presented in Table 8-2. As the purging gas is switched from 100% N2 to 5% CO2 to 100% CO2, the diffusible H concentrations change from 7x10-8 to 1.2x10-7 to 2.3x10-7 mol/cm3 (or 0.5, 0.8, and 1.7 appm), respectively. Certainly the increase in the diffusible H concentration in the steel is associated with an increase in the kinetics of the HER at the charging side of the sample when the purging is switched from N2 to CO2. This again confirms the role of dissolved CO2 in facilitating the ingress of H in steels exposed to the nn-pH HCO3− environments. The concentration of diffusible H also increases from 2.6x10-7 to 6.8x10-7 mol/cm3 (1.8 to 4.8 appm) as the applied cathodic E is augmented from -0.85 to -1.15 VSCE.   176 Hydrogen in steel could reduce the threshold stress intensity and therefore cause cracking in materials. The dependence of the threshold stress intensity (𝐾𝑡ℎ) on H concentration is described by {E-2.2} (reproduced as {E-8.15} below), with the parameters given earlier in section 2.3 [70]: 𝐾𝑡ℎ = 1𝛽′𝑒𝑥𝑝 [(𝑘𝐼𝐺−𝛼𝐶𝐻𝜎,𝑇)𝛼"𝜎𝑌𝑆]     {E-8.15} 𝐶𝐻𝜎,𝑇 = 𝐶𝑠𝑒𝑥𝑝 [𝜎𝐻𝑉𝐻R𝑇]      {E-8.16} The H concentration at the crack tip (CH,T) is proportional to the diffusible hydrogen concentration (𝐶𝑠) in the material through {E-8.16} (see section 2.3 again for parameter descriptions). Based on this relationship, an increase in the diffusible H concentration in the steel reduces Kth and hence the toughness of the material (i.e. resistance to fracture). The H concentrations in X100 steel exposed to nn-pH HCO3− environments here are relatively low compared to the diffusible H concentration in pipeline steels exposed to environments containing sulfide (S2-) [161], [285]–[287]. This is because S2--containing environments are known to promote H ingress into steels due to the poisoning effect of S2- on the H combination reaction. However, such low diffusible H concentrations in the nn-pH HCO3− environments shown here can still affect material toughness and cause HE and HIC issues in pipeline steels [19], [93]–[96], [264], [288]. 8.5 Summary In this chapter, the H generation kinetics on HSLA X100 steel in nn-pH HCO3− solutions simulating external pipeline corrosion environments are studied. Dissolved CO2 in the solution accelerates H evolution kinetics through the H2CO3 discharge and HCO3− discharge reactions. At OCP, H evolution due to direct water reduction is the dominant reaction in O2-free and CO32−-free solutions; however, the H2CO3 discharge becomes the dominant reaction in solutions with  177 dissolved CO2. At E < OCP, H evolution due to HCO3− discharge is proven to be thermodynamically inevitable and gradually dominates the other reactions occurring at the OCP. Through permeation tests in a H permeation setup and subsequent analysis of permeation currents, the H diffusivity in X100 steel is found to be approximately 4.4x10-7 cm2/s. The concentration of diffusible H in X100 exposed to external nn-pH HCO3− environments is hence relatively low compared to that in S2--containing environments. However, a measurable amount of diffusible hydrogen of approximately 1 appm is detected in the steel at the freely corroding condition. This H concentration in the steel increases at more negative applied E which could arise from CP of buried pipeline infrastructure. The H concentration also increases with higher CO2 partial pressure in the local electrolyte environment. The evidence shown in this chapter therefore supports the proposed role of H (through H embrittlement) in the transgranular SCC mechanism of pipeline steels exposed to dilute HCO3− environments of nn-pH.         178 9 9. FEM of the external corrosion and structural integrity of buried pipelines under the influences of CP, gas diffusion, and environment physicochemistry7  Strategies which mitigate the external corrosion of underground pipelines revolve around two methods: CP and protective coatings [35], [122]. Chemical degradation of pipeline steels at coating failure sites is lessened by CP systems, in which performance is heavily affected by physicochemical soil properties. As summarized in various sections of the literature review presented earlier, spatial- and time-dependent soil corrosivity is related to numerous physicochemical properties including: position of water table, soil moisture content, soil type, soil resistivity, soil pH, soluble salt content, structure-to-soil potential, redox potential, microbes in the soil, and stray currents [123]. Complexities arise from the interdependence of electrical resistivity, O2 diffusivity, or heat transfer on one or more soil properties such as particle distribution, porosity, moisture, and temperature [289]. Simplification of complex soil parameter interdependencies is often done when developing empirical correlations of soil corrosivity. Merely soil’s electrical resistivity is considered by some to be a sufficient predictor of corrosivity [126], whilst others consider soil type and structure as the determining factors regardless of position with respect to the water table [127]. Simply the wt% moisture content of the soil is also deemed sufficient in determining the aggressiveness category of soils [126]. Such empirical guidelines unduly simplify gas transport, heat transfer, and reaction kinetics phenomena (and complex interdependencies therein), making them generally inadequate for detailed evaluations. Instead, numerical simulations can capture a larger spectrum of the fundamental processes occurring on corroding pipelines and better predict perilous operating parameters and critical corrosion locations on exposed surfaces.                                                   7  I. M. Gadala, M. Abdel Wahab, and A. Alfantazi, Materials and Design, vol. 97, pp. 287–299,  May 2016.  179 Finite element modelling has been used to compare the relative influence of coating quality, soil electrical conductivity, and CP anode size, position, and type on the corrosion of buried steel tanks [125]. By incorporating the influence of diffusion-controlled O2 reduction with electrolyte conductivity, the distribution of electrical potential within the electrolyte can be computed [129]. Yet, spatial- and time-dependent differences in electrolyte properties are normally not accounted for in previous reports. Similarly, available numerical models of CP and corrosion systems often make narrow use of experimental diffusivity measurements. Rather, the rate of transport is determined with empirical parameters [130], [138]. Seldom do models present coupled gas transport and electrolyte potential distribution like in [140] or variable gas diffusivity in bulk porous media like in [141]. In porous soil, tortuous mass-transfer obeys multi-phase flow laws which depend on air-filled void porosity, moisture content, and temperature. These parameters are interrelated and depend on other conditions such as soil type and structure, for both O2 and CP. The macro meter-scaled model (m-model) presented in this chapter uniquely simulates the variable diffusivity of O2 within soil based on established soil science models. For CP, the Nernst-Planck formulation relies on concentrations, diffusivities, and electric mobilities of ionic species within the soil. This method is simplified to a governing equation with a single soil conductivity parameter (spatially variable, based on experimental measurements) due to difficulties in quantifying Nernst-Planck variables for irregular soils [125], [129], [130], [138], [140], [141]. The conductivity is adjusted for the temperature and moisture content of the soil in each simulation of the m-model, or simply the temperature of the trapped water in the millimeter-scaled model (mm-model) presented in the second part of this chapter.  Reaction kinetics at electrode-electrolyte interfaces are vital in any corrosion model. Formerly, problems in obtaining reliable kinetic parameters have led to a wide range of values being used. Difficulties in obtaining accurate tabulated kinetic parameters to simulate corrosion rates at electrode-electrolyte interfaces are often overcome by instead extracting them from  180 dedicated laboratory experiments simulating the system [138][33]. Muehlenkamp et al. [33] reported using parameter values for their modelled system from measurements. In almost all such previous works though, a prescribed number of parameter values are extracted from the experimental results based on the type of formulation intended to govern reaction kinetics. Incongruities between the kinetics model and the actual experimental data are inevitable. In the models presented in this chapter, such discrepancies are eliminated by integrating the full spectrum temperature-controlled polarization results into the governing equations. Critical temperature influences on corrosion behavior, shown in earlier chapters of this thesis, are carried through from the first step heat transfer simulations in the present chapter. This feature of the models presented here enables more accurate predictions of CP effectiveness and material corrosion in practical operating temperature ranges.  The objective of the m-scale model developed in the first part of this chapter is to allow for the study of synergistic interactions between heat transfer, CP, O2 diffusion, and corrosion of buried pipelines. This helps elucidate the strong implications of physiochemical soil parameters on CP, corrosion, and pipeline system design. The objective of the mm-scale model in the second part of this chapter is to simulate transient corrosion defect growth in X100 pipeline steel exposed to nn-pH trapped water under disbonded external coatings, and the resulting diminishment of the pipeline’s structural integrity. To date, this latter time-dependent corrosion model is the first to specifically simulate defect growth in trapped water relevant to the nn-pH SCC phenomena of buried pipeline applications.  Both models are intended for simulations of buried high-pressure gas transmission pipelines, and present opportunities for rapid scalability to different system dimensions or adjustment for different pipeline applications. Furthermore, they can be integrated into a single multi-scale model with additional work. This chapter is based on a paper [290] which was published as part of the research work leading towards this PhD thesis.  181 9.1 Modelling details: governing phenomena and equations Comsol Multiphysics® version 4.3a, in which the established FEM is the numerical technique employed, is the modelling software used in all the simulations of this chapter. The visual representations of both models and the associated geometries have been detailed earlier in section 4.3. The mesh parameters of the m-model were given in Table 4-2 on page 53, whereas the mesh parameters of the mm-model can be found in Table E-1 of Appendix E.    Heat transfer, CP, and O2 diffusion phenomena The m-scale model of this chapter includes heat transfer, CP, and O2 diffusion modules. Conversely, the mm-scale model does not simulate heat transfer since the small disbondment region is considered to be uniform at the temperature of the pipeline. Significant O2 diffusion within the disbondment is disregarded and only deaerated (standard 5% CO2 and nonstandard 0% CO2) environments are simulated. Sand, clay, and peat soil structures are simulated in the m-scale model of this chapter, at moisture contents (herein termed volumetric wetness, ψ) ranging from 0.1 to 0.6 depending on air porosity ratios (ϕ). The ψ of a soil is 0 if it contains absolutely no water (i.e. completely dry) and cannot be 1 unless the media is completely liquid phase. The latter case would constitute a different modelling situation, simulating phenomena related to submerged or underwater structures (not addressed here). Likewise, the ϕ of the soil is 1 if it contains absolutely no air pockets/pores and cannot be 0 unless the media is completely gas phase. These boundary values for both ψ and ϕ are not addressed in this work. Table 9-1a: Governing transport equations in each module of the m-model [290] Module Number Equation  Condition Heat transfer {E-9.1} 𝐶𝑣 𝜕𝑇 𝜕𝑡⁄ = 𝛻(𝜆𝛻𝑇) + 𝑄  \ CP {E-9.2} 𝒊𝒍 = −𝜎𝑖𝛻𝜑𝑖   𝛻 ∙ 𝒊𝒍 = 𝑄𝑙 = 0  O2 diffusion {E-9.4} 𝛻 ∙ (−𝐷𝑂2𝛻𝑐𝑂2) = 𝑅𝑂2  \  1a. Heat transfer: The thermal parameters studied in the m-model are listed in Table E-2 in Appendix E [291] and have been experimentally validated in the work of Abu-Hamdeh and Reeder [131]. The governing equation for heat flow in the soil ({E-9.1}, Table 9-1a) is a function  182 of thermal conductivity (λ) and the volumetric heat capacity (Cv), where T is temperature in °C, t is time in s, 𝛻 is the gradient in the 2D spatial domain (i.e. 𝜕𝜕𝑥 and 𝜕𝜕𝑦), and Q is a constant heat flux density in W/m2 (for the 2D case) in the presence of heat sources or sinks. Air’s contribution to Cv can generally be neglected since Cair = 0.0012 << Cwater = 4.18 MJ/m3K [292].  Table 9-2b: Values or expressions for the thermal, electrical, and diffusion properties of the soil media in the m-model [290]  Module Number Material property values/equations Heat transfer / See Table E-2 in Appendix E for 𝜆 and 𝐶𝑣 values CP {E-9.3} 𝜎𝑖 = 𝜎298𝐾(1 + 0.02(𝑇𝑖 − 298)) ∙ ψ O2 diffusion {E-9.5a} 𝐷𝑂2_𝑠𝑎𝑛𝑑 = 𝐷𝑂2_𝑇 ∙ (ϕ - ψ)5/3 {E-9.5b} 𝐷𝑂2_𝑐𝑙𝑎𝑦 = 𝐷𝑂2_𝑇 ∙ (ϕ - ψ)2.51 ∙ ϕ-2 {E-9.5c} 𝐷𝑂2_𝑝𝑒𝑎𝑡 = 0.66 𝐷𝑂2_𝑇  ∙ (ϕ - ψ) ∙ [(ϕ - ψ) / ϕ]2  1b. CP: In both the m-model and the mm-model, a single electrical conductivity (σ) parameter simulates the complex process of ionic transport within the electrolyte. In the m-model, the electric field between the CP anode and the exposed pipeline surface drives the transport of dissolved ions in the soil moisture (liquid phase). The gradient of the potential (φ) multiplied by σ at a point in the soil domain yields the current density vector 𝒊𝒍 at that point ({E-9.2} in Table 9-1a and identical {E-E.2} in Table E-3 in Appendix E). Upon reaching the steady state of the simulations in the m-model, the charge flux (𝑄𝑙) in the soil should equal zero. On the other hand, the mm-model simulations and analyses are transient, hence the 𝑄𝑙 in the trapped water electrolyte therein is consistently non-zero. In both models the influence of temperature on σ [137] is simulated with an experimentally validated 2% per degree Kelvin linear compensation ({E-9.3} in Table 9-1b). A nominal σ value at 298 K (25 °C) has been experimentally measured for the representative NS4 simulated soil solution (𝜎298𝐾 = 0.11 S m-1) using a calibrated Jenway® 4510 benchtop conductivity meter, as mentioned earlier in this thesis.  1c. O2 diffusion: In the m-model, O2 diffusion occurs due to the gradient in the soil O2 concentration (𝐶𝑂2), denoted 𝛻𝑐𝑂2, between the ground-atmosphere interface and deeper soil coordinates. Depletion mainly occurs due to O2 reduction at the exposed pipeline surface. Charge  183 neutrality of the molecular O2 nullifies migration effects due to 𝛻𝜑. In {E-9.4} shown in Table 9-1a, 𝐷𝑂2 is the effective O2 diffusivity and 𝑅𝑂2 is a flux term accounting for systematic O2 depletion from reduction at the exposed pipeline surface, the kinetics of which is defined in section 9.1.3 below. 𝐷𝑂2 is dependent on temperature, ϕ, ψ, tortuosity, capillary effects, and other factors within the soil. A nominal O2 diffusivity parameter at a reference temperature of 298 K (𝐷𝑂2_0 = 1 x 10-6 m2/s [293]) is adjusted for the actual soil temperature based on the heat transfer module solution through 𝐷𝑂2_𝑇 = 𝐷𝑂2_0 ∙ (T/Tref)1.5, then is used to evaluate the actual effective diffusivity through empirical correlations derived by soil researchers: Lai et al. for sand in {E-9.5a} of Table 9-1b, Xu et al. for clay in {E-9.5b}, and Moldrup et al. for peat in {E-9.5c}. These correlations capture complex ϕ, ψ, tortuosity, and capillary effects on the diffusion, and have been experimentally validated in dedicated works such as [134].  Initial conditions and boundary conditions for governing phenomena 2a. Heat transfer: In the m-scale model of this chapter, λ and Cv of the soil are considered to be homogenous and depth-independent. Natural annual periodicity in atmospheric conditions changes the soil’s thermal regime. Ground surface temperatures of 303 K (30 °C, maximum) in the summer and 263 K (-10 °C, minimum) in the winter are imposed, while the pipeline wall temperature (Twall) fluctuates between 323 K (50 °C) and 303 K (30 °C) inclusive during both those seasons, respectively. Initial soil temperature profiles of the complete 2D model section (Figure 9-1) are evaluated using {E-9.6} below [292], where z is depth below surface in m, t is ratio of annual cycle (0 ≤ t ≤ )], ?̅? is average annual temperature in K, 𝐴0 is amplitude of the annual temperature cycle in K, d is the characteristic annual damping depth in m (𝑑 =√365𝜆/𝜋𝐶𝑣), and 𝜔 is angular frequency in 1/s. To solve for induced temperature profiles, side boundaries are set to the boundary condition described by {E-9.7a}. 𝑇(𝑧, 𝑡) = ?̅? + 𝐴0𝑒−𝑧𝑑 𝑠𝑖𝑛 [𝜔(𝑡 − 8) −𝑧𝑑]  0 ≤ 𝑧 ≤ 𝑑 {E-9.6} −∇𝑇 ∙ ?⃗? = 0 where ?⃗?  is a unit vector normal to the surface {E-9.7a}  184 −𝜎𝑖∇𝜑i ∙ ?⃗? = 0       {E-9.7b} −𝐷𝑂2∇𝑐𝑂2 ∙ ?⃗? = 0       {E-9.7c}  Figure 9-1: Pipeline wall and initial (ambient) soil temperature fluctuations with respect to soil structure, depth, and time in m-model 2b. CP: In the CP module of the m-model, initial potentials are imposed onto the boundary of the CP anode (represented as Eanode) and the pipeline electrode (represented as Eapp) in Figure 4-4b presented in section 4.3 above. It is assumed that: (1) the CP anode has negligible resistivity; (2) the anode-soil interface has a constant interfacial ∆𝜑 due to very fast kinetics; and (3) aging effects such as deposition of corrosion products on the anode are insignificant. Eapp is the potential applied between the pipeline and the CP anode, with the anode treated as ground. This is set to – 1 V for the majority of simulations, yet -0.75 V and -1.25 V are also simulated. Eanode for all simulations is set to -2.37 VSCE [124], a parameter which could be easily modified to evaluate the performance of specific anode materials such as ferrosilicon. Boundaries which are neither the CP anode nor the exposed surface are set to the electrically insulated boundary condition of {E-9.7b}, with complete insulation assumed. Initial conditions in the mm-model of the disbondment region are imposed to simulate the three CP region situation shown to exist under coating disbondments [26], [27], [294]: (1) CP  185 potential sufficient region at the opening, (2) CP potential insufficient region in the middle, and (3) CP potential absent region at the leading edge. The potential applied to the pipeline is again denoted as Eapp, considered constant based on an uninterrupted CP power supply, and set to either –1 V or –0.75 V to investigate the impact of applied CP voltages. The coated and detached coating boundaries of Figure 4-5 are set to the electrically insulated boundary condition of {E-9.7b}, also assuming perfect insulation from charge/ion transport. Conversely, the exposed steel surface is modelled as a moving boundary at which the mesh is periodically regenerated. The boundary’s movement is correlated to the anodic corrosion rate, governed by the spatially-dependent reaction kinetics on the exposed surface as shown in section 9.1.3 below. 2c. O2 diffusion: In the O2 diffusion module of the m-model, 𝐶𝑂2 is constantly equal to a reference atmospheric 𝐶𝑂2_𝑟𝑒𝑓 of 9 mol/m3 at the ground-atmosphere interface. All boundaries and surfaces excluding the ground-atmosphere interface and the exposed pipeline steel section (at which O2 reduction occurs) have the no-flux diffusion boundary condition of {E-9.7c} imposed.  Reaction kinetics at exposed steel surface The three electrochemical reactions considered at the exposed pipeline surface interface with the soil in the m-model are Fe oxidation (Fe  →   Fe2+ +  2e−, {R-5.1}), O2 reduction (O2  +  2H2O +  4e−  →   4OH−, {R-7.1}), and water reduction/H evolution (2H2O +  2e−  →  H2  +  2OH−, {R-5.4}) for nn-pH conditions. In the mm-scale model, since no O2 presence is simulated in the disbondment region, water reduction through {R-5.4} is the only cathodic reaction considered. In both models, the effect of corrosion product buildup on the steel surface is not simulated. The following two subsections describe the reaction kinetics equations for the two models developed. 3a. m-model: In this model the total current density (𝑖𝑡𝑜𝑡𝑎𝑙) at any Twall and any point along the exposed arc angle θ (see Figure 4-4b) must equal the total i passing from the bulk soil evaluated at that same θ, normal to the surface ({E-9.8}, Table 9-3). No voltage accumulation can  186 occur in the complete electrical circuit of the CP system. As shown in {E-9.9a}, this includes Eanode (equilibrium potential of the anode-soil half-cell), ∆𝜑 (ohmic potential drop in soil between anode and cathode), EFe (potential difference at pipeline-soil interface), and Eapp (applied voltage between pipeline and anode). The relationships can be simplified and rearranged to {E-9.9b} yielding EFe, since the anode-soil interface potential (𝜑|𝑎𝑛𝑜𝑑𝑒) is equal to Eanode and since Eapp takes Eanode as ground. As shown in {E-9.10}, the flux associated with the depletion of O2 (𝑅𝑂2) is evaluated in the same way as 𝑖𝑡𝑜𝑡𝑎𝑙. As exhibited in this balance, 𝑅𝑂2 is undoubtedly a function of the stoichiometric coefficient of O2 reduction in {R-5.4} (𝜐𝑂2, negative for depletion), the number of electrons exchanged (𝑛𝑂2: 4e−/equivalent), the Faraday constant (F: 96,500 C/mol per equivalent), and the O2 reduction current density (𝑖𝑂2 in A/m2) as a function of 𝜃 and Twall. Table 9-3 also shows i parameters evaluated using Tafel kinetics for the “uncoupled” simulations. Parameter values used here are within ranges obtained from previous modelling studies relevant to this system [33], [123], [138], [141], [142], and are averaged where applicable. The Fe oxidation i, 𝑖𝐹𝑒, is spatially-dependent (function of θ) and temperature-dependent (function of Twall), where 𝑖𝐹𝑒°_𝑟𝑒𝑓 is the exchange Fe oxidation i (= 7 x 10-5 A/m2), 𝜂𝐹𝑒 is the overpotential for Fe oxidation in V, and 𝛽𝐹𝑒 is the Tafel slope for Fe oxidation (= 0.3 V/decade). The overpotential 𝜂𝐹𝑒 is that which is between 𝐸𝐹𝑒(𝜃) and the temperature-dependent equilibrium potential (𝐸𝐹𝑒𝑒𝑞) for Fe corrosion. Hence, the only outstanding relationship to evaluate 𝑖𝐹𝑒 is that of the temperature dependence of 𝐸𝐹𝑒𝑒𝑞 using the Nernst equation in {E-9.11c}. Values of -0.833 VSCE and 2 are used for the standard Eeq for Fe corrosion (𝐸𝐹𝑒𝑒𝑞_𝑠𝑡𝑑) and 𝑛𝐹𝑒, respectively. The dissolved [Fe2+] in the aqueous layer is assumed as a constant at 10-6 M [124]. Uncoupled O2 reduction and H evolution kinetics governed by Tafel ({E-9.12} and {E-9.13}, respectively) are the equivalents of {E-9.11} for Fe corrosion. O2 reduction at any 𝜃 is considered proportional to the steady-state 𝐶𝑂2 at that location through the 𝐶𝑂2(𝜃)/𝐶𝑂2_𝑟𝑒𝑓 ratio.  187 Values of -0.2 V/decade, 0.166 VSCE, 4e−/equivalent, and 7.7 x 10-7 A/m2 are used for 𝛽𝑂2, 𝐸𝑂2𝑒𝑞_𝑠𝑡𝑑, 𝑛𝑂2, and 𝑖𝑂2°_𝑟𝑒𝑓, respectively. Considering that the measured pH of aerated NS4 simulated soil solution is 7.5 at 303 K (30 ºC) and 8.1 at 323 K (50 ºC), the pOH used to find [OH−] in {E-9.12c} is 6.5 and 5.2, respectively (water self-ionization constant of 14 and 13.3, respectively). For H evolution kinetics, values of -0.15 V/decade, -0.241 VSCE, 2e−/equivalent, and 7 x 10-7 A/m2 are used for 𝛽𝐻2, 𝐸𝐻2𝑒𝑞_𝑠𝑡𝑑, 𝑛𝐻2, and 𝑖𝐻2°_𝑟𝑒𝑓, respectively. The same pOH methodology for calculating [OH−] in {E-9.12c} is extended to {E-9.13c}. Equations {E-9.11} to {E-9.13} fully define the Tafel-based reaction kinetics of the m-model developed in this chapter, for uncoupled simulations. Table 9-3: Overall formulations and reaction-specific equations governing reaction kinetics in the uncoupled m-model Reaction type Number Equation  Overall formulations {E-9.8}† 𝑖𝑡𝑜𝑡𝑎𝑙(𝜃, 𝑇𝑤𝑎𝑙𝑙) = 𝑖𝐹𝑒(𝜃, 𝑇𝑤𝑎𝑙𝑙) − 𝑖𝑂2(𝜃, 𝑇𝑤𝑎𝑙𝑙) − 𝑖𝐻2(𝜃, 𝑇𝑤𝑎𝑙𝑙)= (−𝜎𝛻𝜑 ∙ ?⃗? )|𝜃,𝑇𝑤𝑎𝑙𝑙  {E-9.9a} 𝐸𝑎𝑛𝑜𝑑𝑒 − ∆𝜑 − 𝐸𝐹𝑒(𝜃) + 𝐸𝑎𝑝𝑝 = 0     where ∆𝜑 = 𝜑(𝜃) − 𝜑|𝑎𝑛𝑜𝑑𝑒  {E-9.9b} 𝐸𝐹𝑒(𝜃) = 𝐸𝑎𝑝𝑝 − 𝜑(𝜃) + 𝐸𝑎𝑛𝑜𝑑𝑒 O2 depletion {E-9.10} 𝑅𝑂2(𝜃, 𝑇𝑤𝑎𝑙𝑙) =𝜐𝑂2 ∙  𝑖𝑂2(𝜃, 𝑇𝑤𝑎𝑙𝑙)𝑛𝑂2𝐹= (−𝐷𝑂2𝛻𝑐𝑂2 ∙ ?⃗? )|𝜃,𝑇𝑤𝑎𝑙𝑙   {A} 𝒊  {B} 𝜼  {C} 𝑬𝒆𝒒  Fe corrosion*  {E-9.11}† 𝑖𝐹𝑒(𝜃, 𝑇𝑤𝑎𝑙𝑙) = 𝑖𝐹𝑒°_𝑟𝑒𝑓𝑒{2.3𝜂𝐹𝑒(𝜃)𝛽𝐹𝑒} 𝜂𝐹𝑒(𝜃, 𝑇𝑤𝑎𝑙𝑙)= 𝐸𝐹𝑒(𝜃)− 𝐸𝐹𝑒𝑒𝑞(𝑇𝑤𝑎𝑙𝑙) 𝐸𝐹𝑒𝑒𝑞(𝑇𝑤𝑎𝑙𝑙) = 𝐸𝐹𝑒𝑒𝑞𝑠𝑡𝑑 +𝑅𝑇𝑤𝑎𝑙𝑙𝑛𝐹𝑒𝐹𝑙𝑛 ([Fe2+]) O2 reduction*  {E-9.12}† 𝑖𝑂2(𝜃, 𝑇𝑤𝑎𝑙𝑙) = 𝐶𝑂2(𝜃)𝐶𝑂2_𝑟𝑒𝑓𝑖𝑂2°_𝑟𝑒𝑓𝑒−{2.3𝜂𝑂2(𝜃,𝑇𝑤𝑎𝑙𝑙)𝛽𝑂2} 𝜂𝑂2(𝜃, 𝑇𝑤𝑎𝑙𝑙)= 𝐸𝐹𝑒(𝜃)− 𝐸𝑂2𝑒𝑞(𝑇𝑤𝑎𝑙𝑙) 𝐸𝑂2𝑒𝑞(𝑇𝑤𝑎𝑙𝑙) = 𝐸𝑂2𝑒𝑞_𝑠𝑡𝑑−𝑅𝑇𝑤𝑎𝑙𝑙𝑛𝑂2𝐹𝑙𝑛 ([OH−]4) H2O reduction/H evolution*  {E-9.13}† 𝑖𝐻2(𝜃, 𝑇𝑤𝑎𝑙𝑙) = 𝑖𝐻2°_𝑟𝑒𝑓𝑒−{2.3𝜂𝐻2(𝜃,𝑇𝑤𝑎𝑙𝑙)𝛽𝐻2} 𝜂𝐻2(𝜃, 𝑇𝑤𝑎𝑙𝑙)= 𝐸𝐹𝑒(𝜃)− 𝐸𝐻2𝑒𝑞(𝑇𝑤𝑎𝑙𝑙) 𝐸𝐻2𝑒𝑞(𝑇𝑤𝑎𝑙𝑙) = 𝐸𝐻2𝑒𝑞_𝑠𝑡𝑑−𝑅𝑇𝑤𝑎𝑙𝑙𝑛𝐻2𝐹𝑙𝑛 ([OH−]2) *: represents uncoupled case based on theoretical parameters provided in sub-section 3a above †: the convention for current density ensures that cathodic current density always has an opposite sign to that of anodic current density  3b. mm-model: The reaction kinetics formulations in the mm-model are similar to the m-model presented above. Here, the two main electrochemical reactions considered to occur at the trapped water-electrode interface in Figure 4-5b are the anodic oxidation reaction of Fe and the  188 cathodic H evolution reduction reaction for nn-pH conditions. It is assumed the availability of water does not limit the reduction reaction. Again, no voltage accumulation occurs, thus the sum of Ebulk (electrolyte potential in bulk solution outside the disbondment, top left corner in Figure 4-5b), ∆𝜑 (ohmic potential drop, within the disbondment in the mm-model), Es(𝑥) (potential difference at interface of exposed steel electrode with trapped water), and Eapp all equals zero (mathematically shown as {E-E.3} in Table E-3 in Appendix E). With 𝜑|𝑏𝑢𝑙𝑘 being equivalent to Ebulk, and Ebulk being the reference for Eapp, the voltage sum of {E-E.3} can be rearranged and reduced to {E-E.4}. The total i (𝑖𝑡𝑜𝑡𝑎𝑙(𝑥, 𝑇𝑤𝑎𝑙𝑙)) from the anodic and cathodic reactions at any Twall, summed at any point x along the exposed surface, must equal the total current density passing from the electrolyte evaluated at that same x, normal to the surface ((−𝜎𝛻𝜑 ∙ ?⃗? )|𝑥,𝑇𝑤𝑎𝑙𝑙). The convention in {E-E.5} is negative for cathodic i (𝑖𝑐(𝑥, 𝑇𝑤𝑎𝑙𝑙)) and positive for anodic i (𝑖𝑎(𝑥, 𝑇𝑤𝑎𝑙𝑙)).  The kinetics of 𝑖𝑎 and 𝑖𝑐 at the exposed steel surface depend on 𝐸𝑠(𝑥), Twall, and pH (through %CO2) based on experimental results. For the full spectrum of 𝐸𝑠(𝑥) potentials, 𝑖𝑐 is evaluated with an exponential function following the format of {E-E.6}, where C1 and C2 are 𝑇𝑤𝑎𝑙𝑙- and %CO2-dependent fitting parameters. Since %CO2 diminishes at locations deeper within the disbondment, the expression is designed to equal 𝑖𝑐(Twall, 5% CO2) at x = 0 cm and 𝑖𝑐(0% CO2) at x = L, based on a linear (x/L) ratio. {E-E.6} evaluates 𝑖𝑐 for 𝐸𝑠(𝑥) values above and below OCP, since experimental profiles follow Tafel behavior as shown in section 9.2.2. On the other hand, 𝑖𝑎 evaluation methodology is divided into two forms separating the 𝐸𝑠(𝑥) values above OCP from those below OCP. Since anodic branches (i.e. 𝐸𝑠(𝑥) > OCP) of PDP plots represent the total anodic current density, 𝑖𝑎(𝐸𝑠(𝑥) > OCP) is directly evaluated from experimental results, again with linear x/L proportionality for CO2. Below OCP, total 𝑖𝑐 controls the polarization plots, so the 𝑖𝑎 must be determined differently; LPR test results here determine the critical parameters for 𝑖𝑎(𝐸𝑠(𝑥) < OCP) in {E-E.7}, where 𝑖𝑎°  is the exchange Fe oxidation i  189 (calculated from a reference value of 𝑖𝑎°_𝑟𝑒𝑓 = 7 x 10-5 A/m2 using {E-E.11}), 𝜂𝑠 is the overpotential for Fe oxidation (evaluated through {E-E.8}), and 𝛽𝑎 is the Tafel slope for Fe oxidation (calculated from a standard value of 𝛽𝑎  = 0.4 V per decade using {E-E.12}). The formulation of {E-E.7} is adjusted for CO2 depletion in the same way as {E-E.6}, using an x/L ratio. The temperature dependence of 𝐸𝑠𝑒𝑞 is expressed using the form of the Nernst equation in {E-E.9}, where a va