Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Soil corrosion behavior of hot-dipped galvanized steel in infrastructure applications Padilla Perez, Victor Eduardo 2014

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

Item Metadata

Download

Media
24-ubc_2015_february_padilla_victor.pdf [ 4.98MB ]
Metadata
JSON: 24-1.0165584.json
JSON-LD: 24-1.0165584-ld.json
RDF/XML (Pretty): 24-1.0165584-rdf.xml
RDF/JSON: 24-1.0165584-rdf.json
Turtle: 24-1.0165584-turtle.txt
N-Triples: 24-1.0165584-rdf-ntriples.txt
Original Record: 24-1.0165584-source.json
Full Text
24-1.0165584-fulltext.txt
Citation
24-1.0165584.ris

Full Text

  iSoil Corrosion Behavior of Hot-Dipped Galvanized Steel in Infrastructure Applications    by  VICTOR EDUARDO PADILLA PEREZ      A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF   DOCTOR OF PHILOSOPHY  in   THE FACULTY OF GRADUATE STUDIES (Materials Engineering)        THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)         December 2014 © Victor Eduardo Padilla Perez, 2014    iiAbstract   Galvanized steel is one of the most common materials used in the construction industry for its relatively low cost paired with an acceptable corrosion resistance. Nevertheless, the early failure of a number of structures around the world that use galvanized steel has raised some controversy on the understanding of the corrosion behavior of zinc. This dissertation presents the results of several electrochemical studies and mathematical models done on zinc and galvanized steel as an attempt to fill in the gaps of current knowledge.   Results indicate an increase on the corrosion rate with increasing amounts of Na2SO4, as well as a potential difference between samples in oxygen saturated, aerated, and de-aerated conditions is large enough to promote macrocell formation under aggresive conditions. The presence of sulphate in the soil significantly increased the corrosion rates and, thus, it is important to consider the effect of sulphate in determining the type of de-icing salt. In sulphate-free solutions, potassium acetate appeared to be the best option; while in the presence of sulphate, MgCl2 and CaCl2 had the lowest corrosion rate. The improved performance was attributed to the formation of a more evenly distributed corrosion product with better protective properties. Furthermore, when measuring corrosion at temperatures ranging from -5°C to 25°C, the rate observed at sub-zero temperatures is still higher than the rate acceptable for galvanized steel reinforced structures. SEM pictures show that the corrosion products grows preferentially in the vicinity of zinc grain boundaries and that it is apt to cracking with increasing thickness.  A numerical model was developed to calculate the corrosion rate of galvanized steel in soil at three different stages of corrosion by considering key soil corrosion parameters. This thesis focuses on the   iiieffect of field conditions relevant to the Canadian climate on the corrosion performance of Mechanically Stabelized Earth (MSE) wall soil reinforcement and facings. Results indicate that the proposed model is suitable to be used for the service-life design and risk assessment of MSE walls and to determine the optimum zinc cover thickness.   ivPreface  The following journal papers and conference presentations have been published form the research work presented in this dissertation. Professor Akram Alfantazi extensively helped with all the aspects of the research work.  Articles published or accepted in refereed journal  1. Victor Padilla, Akram Alfantazi, (2013) Corrosion Performance of Galvanized Steel in Na2SO4 and NaCl solutions at Subfreezing Temperatures, NACE Corrosion Journal. Vol. 69, No. 2, pp. 174-185 2. Victor Padilla, Pouria Ghods, Akram Alfantazi, (2013), Practical Model for the Three-stage Corrosion Behavior of Galvanized Steel Reinforcement in Soil, NACE Corrosion Journal. Manuscript Vol. 69, No. 5, pp. 509-521. 3. Victor Padilla, Pouria Ghods, Akram Alfantazi, (2014), Parametric Studies and Application of a Practical Model for Corrosion of Galvanized Steel in Soil, In preparation to be submitted 4. Pouria Ghods; Victor Padilla; Akram, Alfantazi (2013), Evaluation of galvanized steel corrosion in mechanically stabilized earth walls, Materials Performance, v 52, n 11, p 29-33.  5. Victor Padilla, Akram Alfantazi, (2014), Corrosion Film Breakdown of Galvanized Steel in Sulphate-Chloride Solutions, Construction and Building Materials 66, Pages 447–457 6. Victor Padilla, Pouria Ghods, Akram Alfantazi, (2013), Effect of De-icing Salts on the Corrosion Performance of Galvanized Steel in Sulphate Contaminated Soil, Construction and Building Materials 40 908–918 7. Victor Padilla, Akram Alfantazi (2012), Effects of Oxygen and Sulphate Concentrations on the Corrosion Behavior of Zinc in NaCl Solutions, NACE Corrosion Journal. Vol. 68, No. 3, pp. 035005-1-035005-11   Conference Papers 8. Victor Padilla, Akram Alfantazi, (2011), Effect of Differential Oxygen Access on the Corrosion Behavior of Zinc, Proceedings of the TMS Annual Meeting, Vol. 3, pp. 395-403 9. Victor Padilla and A. Alfantazi (2011). Effect of Temperature and Thermal Cycling on the Corrosion Performance of Galvanized Steel Relevant to Mechanically Stabilized Earth Walls. Oral presentation at the 219th ECS Meeting in Montreal in the Corrosion General Session 10. Victor Padilla, Chemistry Physics Annex- Whole Building Life Cycle Assessment (2010). Published in the UBC Social, Ecological Economic Development Studies (SEEDS) Student Report 11. Victor Padilla (2010). Effect of Differential Oxygen Access of the Corrosion Performance of MSE walls and Facings. Oral presentation at the 2010 Hydrometallurgy Chair at UBC Materials Engineering Department    v12. Pouria Ghods, Victor Padila, Akram Alfantazi, (2013) Numerical Evaluation of Galvanized Steel Corrosion in Mechanically Stabilized Earth Walls, NACE - International Corrosion Conference Series, 2013, NACE International - Corrosion 2013 Conference and Expo 13. Victor Padilla, Pouria Ghods, Akram Alfantazi, (2013), A Novel Model to Predict the Corrosion of Mechanically Stabilized Earth Structures, Transportation Association of Canada Conference, Winnipeg, Canada.                viTable of Contents ABSTRACT ............................................................................................................................... ii PREFACE ................................................................................................................................ iv TABLE OF CONTENTS ............................................................................................................. vi LIST OF TABLES ..................................................................................................................... ix LIST OF FIGURES ..................................................................................................................... x LIST OF SYMBOLS ................................................................................................................ xvi ACKNOWLEDGEMENTS .......................................................................................................... xx DEDICATION ......................................................................................................................... xxi  1.  INTRODUCTION ................................................................................................................. 1 1.1 MSE WALLS IN BRITISH COLUMBIA ................................................................................ 6 2.  LITERATURE REVIEW ....................................................................................................... 8 2.1 GALVANIZED STEEL AS CORROSION PROTECTION METHOD FOR MSE WALLS ..................... 9 2.2 BASIC FEATURES ON THE CORROSION OF GALVANIZED STEEL ......................................... 15 2.3 CORROSION OF GALVANIZED STEEL IN TEMPERATURES RELEVANT TO FIELD APPLICATIONS (-5°C TO 25°C) ................................................................................................................... 23 2.4 EFFECT OF DE-ICERS ON THE CORROSION OF GALVANIZED STEEL .................................... 25 2.5 MATHEMATICAL MODELING OF THE CORROSION OF GALVANIZED STEEL .......................... 28 3.  OBJECTIVES ................................................................................................................... 32 3.1 KEY TECHNICAL OBJECTIVES ........................................................................................ 32 4.  APPROACH AND METHODOLOGY .................................................................................... 33 4.1 MATERIALS ................................................................................................................... 33 4.2 TEST METHODS ............................................................................................................. 34   vii4.3 MODELING .................................................................................................................... 39 5  EFFECTS OF OXYGEN AND SULPHATE CONCENTRATIONS ON THE CORROSION BEHAVIOR OF ZINC IN NACL SOLUTIONS ............................................................................................... 40 5.1 POTENTIODYNAMIC ELECTROCHEMICAL TESTING .......................................................... 42 5.2 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY TESTING ............................................. 49 5.3 SURFACE CHARACTERIZATION ...................................................................................... 56 5.4 SUMMARY ..................................................................................................................... 61 6  CORROSION PERFORMANCE OF GALVANIZED STEEL IN NA2SO4 AND NACL SOLUTIONS AT SUBFREEZING TEMPERATURES ............................................................................................. 63 6.1 EFFECT OF TEMPERATURE ON THE POTENTIODYNAMIC TESTING .................................... 63 6.2 EIS TESTING ................................................................................................................. 70 6.3 SURFACE CHARACTERIZATION ...................................................................................... 76 6.4 SUMMARY ..................................................................................................................... 81 7  EFFECT OF DE-ICING SALTS ON THE CORROSION PERFORMANCE OF GALVANIZED STEEL IN SULPHATE-CONTAMINATED SOIL ..................................................................................... 82 7.1 POTENTIODYNAMIC TESTING ......................................................................................... 82 7.2 LINEAR POLARIZATION RESISTANCE TESTING ............................................................... 89 7.3 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY RESULTS ............................................. 92 7.4 SURFACE CHARACTERIZATION .................................................................................... 100 7.5 SUMMARY ................................................................................................................... 104 8  CORROSION FILM BREAKDOWN OF GALVANIZED STEEL IN SULPHATE-CHLORIDE SOLUTIONS .......................................................................................................................... 106 8.1 CYCLIC VOLTAMMETRY AND CYCLIC POLARIZATION RESULTS .................................... 106   viii8.2 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY RESULTS ........................................... 116 8.3 SUMMARY ................................................................................................................... 125 9  MODEL FOR THE CORROSION BEHAVIOR OF GALVANIZED STEEL IN SOIL .................. 126 9.1 MODELING CONCEPT ................................................................................................... 126 9.1.1 STAGE 1: CORROSION OF ZINC .................................................................................................. 129 9.1.2 STAGE 2: COMBINED CORROSION OF STEEL AND ZINC ........................................................... 132 9.1.3 STAGE 3: CORROSION OF STEEL ................................................................................................ 134 9.2 TWO-DIMENSIONAL NUMERICAL SOLUTION ................................................................ 135 9.3 EFFECT OF SOIL PROPERTIES ....................................................................................... 140 9.4 SUMMARY ................................................................................................................... 155 10  VALIDATION AND APPLICATION OF THE PRACTICAL THREE - STAGE CORROSION BEHAVIOR OF GALVANIZED STEEL IN  SOIL ....................................................................... 156 10.1 COMPARISON TO MODELS ACCEPTED MODELS .......................................................... 156 10.2 SERVICE LIFE DESIGN ............................................................................................... 162 10.3 MODEL VALIDATION ................................................................................................. 165 10.4 SUMMARY ................................................................................................................. 167 11  CONCLUSIONS .............................................................................................................. 168 11.1 TECHNICAL CONTRIBUTIONS TO THE FIELD ............................................................... 170 12  RECOMMENDATIONS FOR FUTURE WORK .................................................................... 172 REFERENCES ....................................................................................................................... 174      ixList of Tables Table 4-1 –Electrochemical Properties of soil samples collected across British Columbia ............ 36 Table 5-1 - Different experimental conditions used on the zinc samples ........................................ 41 Table 5-2- Corrosion parameters of the zinc sample after immersion ............................................. 47 Table 5-3- EIS fitting data obtained for zinc during 24 hours immersion tests in three different concentrations of oxygen. ................................................................................................................ 55 Table 6-1 Corrosion parameters for the galvanized steel samples after immersion ........................ 70 Table 6-2 - EIS fitting data obtained for galvanized steel during 24 h immersion in 3.5wt% NaCl and 1wt% Na2SO4 at temperatures ranging from 25ºC to -5ºC ....................................................... 75 Table 7-1 The properties of the test solutions .................................................................................. 86 Table 7-2 Calculated corrosion parameters using the Butler–Volmer equation for samples immersed in different solutions ........................................................................................................................ 88 Table 7-3- Corrosion rate of galvanized steel from different electrochemical test methods ........... 92 Table 7-4 - Ranking of the effect of de-icing salts on the corrosion performance of galvanized steel........................................................................................................................................................ 103 Table 8-1- Calculated relative surface coverage for galvanized steel recorded after 20 cycles in 3.5wt% NaCl, MgCl2, CaCl2 and CH3CO2K. .............................................................................................. 113 Table 9-1-The values of the parameters used in the model ........................................................... 139 Table 10-1- Upper and lower-bond rate of metal loss obtained from the proposed model in comparison with the AASHTO model ........................................................................................... 158 Table 10-2- Average Monthly Temperature for fours selected cities ............................................ 162       xList of Figures Figure 1-1- Cross-sectional diagram of a MSE wall ......................................................................... 2 Figure 1-2 – North to South view of structure number 9488 located in the Sea to Sky Highway. Two MSE walls are substructures to this bridge, which supports the highway on top of a railroad (above). MSE wall providing support for a road on top of a galvanized steel culvert on Highway 1 ............. 3 Figure 1-3 Picture taken of a corroded MSE wall on Highway 1 near Revelstoke, BC (left). Picture of corroded sample of reinforcement taken from the same wall, showing the loss of cross section (right) ................................................................................................................................................. 5 Figure 1-4 -Total MSE Walls Count by Retaining Wall Type .......................................................... 6 Figure 1-5- Total MSE Walls Count by Construction Date .............................................................. 7 Figure 2-1 –Three stages of galvanized steel corrosion ................................................................... 12 Figure 2-2 – Comparison of potentiodynamic polarization test results for pure zinc, galvanized steel, and steel sample in 3.5 wt% NaCl, plus 1 wt% Na2SO4 in aerated conditions ............................... 13 Figure 2-3 - Potential-pH diagram for the iron-water system at 25°C, concentration of species 10-6.......................................................................................................................................................... 19 Figure 2-4 Potential-pH diagram for the zinc-water system at 25°C, concentration of species 10-619 Figure 2-5- Potential-pH diagram for the zinc-water system at 25°C showing ZnO as the stable corrosion product, concentration of species 10-6 ............................................................................. 20 Figure 3-1- Schematic of experimental set-up ................................................................................. 35 Figure 3-2- Selected results of three replicates of PDP tests performed on samples immersed in 3.5 wt% CH3CO2K................................................................................................................................. 39 Figure 5-1 a) OCP measurements for the control samples immersed in 3.5wt% of NaCl in oxygen-saturated conditions, aerated conditions, and de-aerated conditions, and b) OCP measurements for the samples immersed in oxygen-saturated solutions with 3.5wt% of NaCl, and different concentrations of Na2SO4 (1-4wt%) ................................................................................................ 43 Figure 5-2 Potentiodynamic polarization tests results for zinc samples immersed in 3.5wt% NaCl, and 0wt% Na2SO4 with different concentrations of oxygen ............................................................ 44 Figure 5-3 Potentiodynamic polarization tests results for zinc samples immersed in 3.5wt% NaCl and 1wt% Na2SO4 with different concentrations of oxygen ............................................................ 44 Figure 5-4 Potentiodynamic polarization tests results for zinc samples immersed in 3.5wt% NaCl and 3wt% Na2SO4 with different concentrations of oxygen ............................................................ 45   xiFigure 5-5  Potentiodynamic polarization tests results for zinc samples immersed in 3.5wt% NaCl and 4wt% Na2SO4 with different concentrations of oxygen ............................................................ 45 Figure 5-6 a) EIS results for zinc immersed in an oxygen-saturated solution with NaCl 3.5 wt% and Na2SO4 after 1h, 2h, 3h, 4h, 6h, 7h, and 24 h of immersion b) R(QR) equivalent circuit. ............. 51 Figure 5-7 a) EIS results for zinc immersed in de-aerated solution with NaCl 3.5 wt % and Na2SO4 after 1h, 2h, 3h, 4h, 6h, 7h, and 24 h of immersion, b) High Frequency region zoom for Figure 6a, c) R(Q(RW)) Equivalent circuit for immersion times > 1 h >24 h, d) R(CR(CW)) Equivalent circuit after 24 hours of immersion ............................................................................................................. 52 Figure 5-8 a) EIS results for zinc immersed in an aerated solution with NaCl 3.5wt% and 1wt% Na2SO4 after 1h, 2h, 3h, 4h, 6h, 7h, and 24 h of immersion, b) R(QR) Equivalent circuit after one hour of immersion, c) R(Q(RW) Equivalent circuit for immersion times > 1 h. ............................ 54 Figure 5-9- SEM picture of zinc immersed in de-aerated conditions with NaCl 3.5wt%, Na2SO4 3wt%.......................................................................................................................................................... 56 Figure 5-10  SEM picture of zinc immersed in aerated conditions in NaCl 3.5wt%, Na2SO4 1wt%.......................................................................................................................................................... 57 Figure 5-11- SEM picture of zinc immersed in a solution saturated with oxygen with NaCl 3.5wt%, Na2SO4 1wt% ................................................................................................................................... 58 Figure 5-12 Micro crystal deposits (5-20 µm) found on the surface of corroded zinc providing preferential corrosion sites ............................................................................................................... 58 Figure 5-13 Representative XRD diagrams for pure zinc immersed in a) de-aerated solution with NaCl 3.5wt%, b) oxygen-saturated solution with NaCl 3.5wt% - Na2SO4 1wt%, c) aerated solution with NaCl 3.5wt% - Na2SO4 1wt%, and d) de-aerated solution with NaCl 3.5wt% - Na2SO4 1wt%........................................................................................................................................................... 61 Figure 6-1 Effect of decreasing temperature in a 5 hour interval over a 25 hour period on the OCP reading of a galvanized steel sample immersed in 3.5wt% NaCl, and 1wt% Na2SO4. ................... 64 Figure 6-2 Potentiodynamic polarization curves for the galvanized steel samples immersed in 3.5wt% NaCl, and 1wt% Na2SO4 temperatures ranging from 25ºC to -5ºC ................................................. 65 Figure 6-3 Nyquist Plots for galvanized steel immersed during 24 h in aerated solutions with 3.5 wt % NaCl and 1wt % Na2SO4 at a) 25ºC, b) 15ºC, c) 5ºC, d) 0ºC, and e) -5ºC ...................................... 71   xiiFigure 6-4 a) Nyquist Plots for galvanized steel immersed in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 after 24 h of immersion at 25ºC, 15ºC, 5ºC, 0ºC and -5ºC. b) Insert of the high frequency region of Figure 5a. ......................................................................................................... 72 Figure 6-5 Schematic representation of the equivalent electrical circuit for an electrode protected by a porous RS(Qc(Rp(QctRdl))). ............................................................................................................ 73 Figure 6-6 a) Sulphate nest formation found on the surface of corroded galvanized steel after immersion in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 at 25ºC. b) Smaller sulphate nest formation found on the surface of corroded galvanized steel after immersion in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 at 15ºC. ............................................................................. 77 Figure 6-7 Blisters found on top of corroded galvanized steel after immersion in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 at 0ºC, suggesting early stages for the formation of sulphate nests.................................................................................................................................................. 77 Figure 6-8 SEM pictures showing corrosion products found on the surface of corroded galvanized steel after immersion in an aerated solution with 3.5 wt % NaCl and 1wt% Na2SO4. .................... 78 Figure 6-9 - Thin film found on the surface of corroded galvanized steel after immersion in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 at -5ºC. ............................................................... 78 Figure 6-10 - XRD patterns for galvanized steel immersed in NaCl 3.5wt% and Na2SO4 1wt% at a) 25ºC, b) 15ºC, c) 5ºC, d) 0ºC, e) -5ºC, and f) uncoated steel sample immersed in NaCl 3.5wt% and Na2SO4 1wt% at 25ºC. ..................................................................................................................... 81 Figure 7-1- PDP curves of samples immersed at 25ºC aerated condition in the solutions containing 3.5 wt% of the four distinct de-icing salts: CaCl2, CH3CO2K, MgCl2, and NaCl. .......................... 84 Figure 7-2- PDP curves of samples immersed at 25ºC aerated condition in solutions containing 1wt% Na2SO4 plus 3.5 wt% of the four distinct de-icing salts: CaCl2, CH3CO2K, MgCl2, and NaCl. ..... 87 Figure 7-3 - 24 hours Linear Polarization Resistance measurements for galvanized steel samples immersed in solutions containing 3.5 wt% of CH3CO2K, MgCl2, CaCl2, and NaCl at 25ºC .......... 90 Figure 7-4- 24 hours Linear Polarization Resistance measurements for galvanized steel samples immersed in solutions containing 3.5 wt% of CH3CO2K, MgCl2, CaCl2, and NaCl with 1wt% Na2SO4 added at 25ºC. ..................................................................................................................... 90 Figure 7-5-  Nyquist and Bode Phase plots recorded for samples immersed in aerated solutions with 3.5 wt% MgCl2 plus 1wt% Na2SO4 at immersion times of 1 h and 24 h ........................................ 93   xiiiFigure 7-6- Nyquist plots recorded for samples immersed in aerated solutions with 3.5 wt% NaCl, MgCl2, CaCl2, and CH3CO2K at 25ºC, at immersion times of a) 1 h, b) 2 h, c) 4h, d) 7h, and e) 24 h.......................................................................................................................................................... 95 Figure 7-7- Corrosion rate results obtained from the EIS fitting polarization results during 24 hour immersion in the 3.5wt% NaCl, MgCl2, CaCl2, and CH3CO2K solutions....................................... 96 Figure 7-8- Nyquist plots recorded for samples immersed in aerated solutions with 3.5 wt% NaCl, MgCl2, CaCl2, and CH3CO2K with 1wt% Na2SO4 added into the solutions at 25ºC, at immersion times of a) 1 h, b) 2 h, c) 4h, d) 7h, and e) 24 h ............................................................................... 98 Figure 7-9-  Corrosion rate results obtained from the EIS fitting polarization results during 24 hour immersion in the 3.5wt% NaCl, MgCl2, CaCl2, and CH3CO2K with 1wt% Na2SO4 added into the solutions ........................................................................................................................................... 99 Figure 7-10 - Surface morphology of galvanized steel obtained in a) 3.5wt% MgCl2 and b) 3.5wt% MgCl2 plus 1wt% Na2SO4 solutions after PDP testing in aerated conditions at 25ºC .................. 101 Figure 7-11-  Surface morphology of galvanized steel obtained in a) 3.5wt% CH3CO2K plus 1wt% Na2SO4 and b) 3.5wt% NaCl plus 1wt% Na2SO4 solutions after PDP testing in aerated conditions at 25ºC ................................................................................................................................................ 103 Figure 8-1- 20th cycle Cyclic Voltammogram for a galvanized steel sample in solution with 1wt% Na2SO4 and 3.5 wt% NaCl at 25ºC and a scan rate of 20 mV s-1 .................................................. 107 Figure 8-2-  Cyclic voltammograms recorded for samples immersed in solutions with 1wt% Na2SO4 and 3.5wt% a) CaCl2,  b) CH3CO2K, c) MgCl2, and d) NaCl at 25ºC and a scan rate of 20 mV s-1........................................................................................................................................................ 109 Figure 8-3-  Cyclic voltammograms recorded for samples immersed in solutions with 1wt% Na2SO4 and 3.5 wt% CaCl2, CH3CO2K, MgCl2, and NaCl after 20 cycles at 25ºC and a scan rate of 20 mV s-1 .................................................................................................................................................... 110 Figure 8-4- Cyclic Polarization curves for samples immersed in solutions with 1wt% Na2SO4 and 3.5 wt% a) CaCl2, b) CH3CO2K, c) MgCl2, and d) NaCl .............................................................. 114 Figure 8-5– Nyquist and Bode Phase plots recorded for samples immersed in aerated solutions with 3.5 wt% NaCl plus 1 wt% Na2SO4 at immersion times of 1 h, 2 h, 3 h, 4 h, 6 h, 7 h, and 24 h.... 117 Figure 8-6- Nyquist plots recorded for samples immersed in aerated solutions with 3.5wt% CaCl2, CH3CO2K, MgCl2, and NaCl solution with the addition of 1wt% Na2SO4 into all the solutions at 25ºC at immersion times of a) 1 h and b) 24 h .............................................................................. 119   xivFigure 8-7- Schematic representation of the two equivalent electrical circuit used for modelling an electrode protected by a porous layer ............................................................................................ 120 Figure 8-8- Calculated Rp values at 25ºC samples immersed in aerated solutions with 3.5wt% CaCl2, CH3CO2K, MgCl2, and NaCl solutions with the addition of 1wt% Na2SO4 in all the solutions ... 121 Figure 8-9- Calculated film thickness for samples immersed in aerated solutions with 3.5wt% CaCl2, CH3CO2K, MgCl2, and NaCl solutions with the addition of 1wt% Na2SO4 into all the solutions. At 25ºC, after 24 hours of immersion ................................................................................................. 123 Figure 8-10 -Selected Cross-sectional SEM images showing taken from galvanized steel obtained in 1 wt% Na2SO4 plus a) 3.5wt% CaCl2, and b) 3.5wt% NaCl after 24 hours of immersion EIS testing in aerated conditions at 25ºC ......................................................................................................... 124 Figure 9-1 The schematic illustration of microcell and macrocell corrosion of galvanized steel in soil........................................................................................................................................................ 127 Figure 9-2 The combined microcell and macrocell corrosion of galvanized steel ........................ 128 Figure 9-3 - Schematic illustration of the domain, corroding and non-corroding zones and corresponding ................................................................................................................................. 136 Figure 9-4- The effect of soil resistivity on the corrosion rate of galvanized steel at two levels of pH: a) pH=6, b) pH=10 ......................................................................................................................... 141 Figure 9-5- The effect of soil pH in the three stages of corrosion rate of galvanized steel ........... 143 Figure 9-6- The effect of resistivity on the three stages of corrosion rate of galvanized steel at three different pH, a) pH 6, b) pH 10, and c) pH 13 ............................................................................... 146 Figure 9-7 - The effect of temperature on the corrosion rate of galvanized steel at two different resistivities: (a) Resistivity = 3,000 Ω.cm, (b) Resistivity=7,500 Ω.cm ........................................ 148 Figure 9-8 - The effect of moisture content on the three stages of corrosion rate of galvanized steel........................................................................................................................................................ 151 Figure 9-9- The effect of limiting current density (i.e., oxygen concentration) on the three stages of corrosion rate of galvanized steel .................................................................................................. 154 Figure 10-1- Comparison of results from the AASHTO guidelines with results from the proposed model at mildly corrosive and highly corrosive conditions ........................................................... 158 Figure 10-2 -Comparison of the metal loss calculated from the proposed model and other models in the literature during the 75 year service life of a MSE wall .......................................................... 159   xvFigure 10-3 - Comparison of the metal loss calculated from the proposed model and other models in the literature during the first 20 years service life of an MSE wall ............................................... 161 Figure 10-4- The effect of temperature on the calculated corrosion rate for each stage of galvanized steel corrosion for four distinct temperature profiles at selected cities: a) Ottawa, b) Vancouver, c) Los Angeles, and d) Acapulco ....................................................................................................... 163 Figure 10-5- Calculated metal loss for the first 7 years of service life for theoretical structures, in each of the four different temperature profiles compared with the AASHTO model corrosion rates........................................................................................................................................................ 164 Figure 10-6 – Model Comparison with average corrosion rates of galvanized steel pipe specimens in soils for 10 years as reported by NBS........................................................................................ 166     xviList of Symbols Α Symmetry factor A Effective area (cm2) Acor Area at corroding site (cm2) Anon-cor Area at the non-corroding site (cm2) B Constant (V/dec) Βa Anodic Tafel Slope (V/dec) Βa Cathodic Tafel Slope (V/dec) ox  Oxidation Tafel Slope (V/dec) red  Reduction Tafel Slope (V/dec) Ceff Effective capacitance (F/cm2) df Film Thickness (µm) ∆G Gibbs Free Energy (KJ/mol) Ee,  Eo Equilibrium Potential (V) E(v) Measured Potential (V) ZnaE ,  Anodic corrosion potential of zinc (V) oZnaE ,  Equilibrium potential of the anodic reaction of zinc (V) OxcE ,  Cathodic corrosion potential (V) FeaE ,  Anodic corrosion potential of iron (V) oFeaE ,  Equilibrium potential of the anodic reaction of iron (V) pHOxcE ,  Modified cathodic potential of oxygen (V)   xviiEcorr Corrosion Potential (V) Eb Film Breakdown Potential (V) Er Film Healing Potential (V) Enon-cor,mic Microcell Potential at the non-corroding site (V) miccorE ,  Microcell Potential at the corroding site (V) Ecor,mac Macrocell Potential at the corroding site (V) Enon-cor,mac Macrocell Potential at the non-corroding site (V) o  permittivity of vacuum (8.854 ×10−12 F m-1)   dielectric constant F Faraday’s Constant (96500 C/mole) F Frequency (Hz) Φfilm Relative percentage of the film surface area covered by a corrosion layer )//(max xZnOZnoxi  maximum current density for the anodic peak for a “x” salt )//(max NaClZnOZnoki  maximum current density for the anodic peak in NaCl I Current (A) ia Anodic current density (A/cm2) ic Cathodic current density (A/cm2) io,x Exchange Current Density (A/cm2) or element “x” icor,mic Microcell corrosion current density at the corroding site (A/cm2)  icorr Corrosion Current Density  (A/cm2) inon-cor,mic Microcell corrosion current density  at the non-corroding site (A/cm2)   xviii icor,mac Macrocell corrosion current density at the corroding site (A/cm2)  inon-cor,mac Macrocell corrosion current density at the non-corroding site (A/cm2)  iL Limiting Current Density (A/cm2) ipass Passivation Plateau (A/cm2) Qx Constant Phase Element at oxide film/solution interface/element “x” Η Overpotential (V) θ Volumetric moisture content (%) T Temperature (oK ) ρsoil Soil resistivity (Ω.m) ρt Calculated soil resistivity at temperature To (Ω.m) ρm Calculated soil resistivity at  volumetric moisture content of  θ m ρr Measured soil resistivity at  volumetric moisture content of  θ r R Gas Constant (8.314 J/(mole.K)) Rx Resistance at oxide film/solution interface/element “x”  (Ω.m) Rs Solutions Resistance  (Ω.m) Rp Polarization Resistance  (Ω.m) Rf,Zn   Electrical resistance of the zinc corrosion product  (Ω.m) U Gravimetric moisture content W Warburg Element γ Bulk specific gravity of the soil Χ2 Chi-squared error zx Valance electron count involved in the reaction “x”   xix      Z´ Real Impedance Z´´ Imaginary Impedance   xxAcknowledgments  First and foremost, I would like to express my most sincere gratitude to Professor Akram Alfantazi. He was a constant source of inspiration for my PhD work, and constantly challenged me to learn and gain experience not only in the corrosion laboratory but also in other areas. Wihthout his guidance I would have never been able to accomplish this work.   Additionally, I want to show my appreciation to Kevin Baskin, Ian Sturrock, and Daniel Belisle from the Ministry of Transportation and Infrastructure for their help and feedback during this project. Furthermore, many thanks to Wayne Ford and Anna West from Atlantic Industries Limited. The input provided by people from the public and the private sector was crucial to make this work more relevant and useful to the industry.  I would also like to thank the members of the Materials Engineering staff at UBC for their help and patience during this work. In particular: Michelle Tierney, Glenn Smith, Ross McLeod, Fiona Webster, Mary Jansepar, Jacob Kabel, Carl Ng, David Torok, and Rudy Cardeno.   I also want to thank all the members of the Corrosion Group at UBC, in particular to Pouria Ghods, for their help and insights during this time.  Finally, I aknowledge the financial support for this work was provided by the British Columbia Ministry of Transportation and Infrastructure (MOTI), Atlantic Industries Limited (AIL), El Consejo Nacional de Ciencia y Tecnología (CONACYT), and by Natural Sciences and Engineering Research Council of Canada (NSERC).     xxi                A mis famlia,   To my family,   11. INTRODUCTION The main reason for premature failure in steel-reinforced structures is corrosion of the steel reinforcement (Aperador et al., 2009, Bastidas et al., 2008, Yadav et al., 2007). According to a study released by the National Association of Corrosion Engineers (NACE) International, the United States alone spends $276 billion annually in repair and maintenance of reinforced structures, of which 25-30% of this cost could be saved by using new corrosion management practices (Javaherdashti, 2000). Out of those costs, it has been estimated that the cost of corrosion directly associated with bridges and road infrastructure, in the US alone, ranges from $2.5 - $5 billion per year (Xianming et al., 2009, Vitaliano, 1992). Even though it is difficult to quantify, one needs to also take into account indirect costs such as delays due to building and road repairs, loss of productivity, and the environmental impact due to construction and production of the required raw materials. According to the Organization for Economic Cooperation and Development, in the next 20 years investments in the construction industry around the globe will amount to $1,626-$1,897 billion per year. This unprecedented projected growth makes corrosion in steel-reinforced structures an important area of concern that presents challenges and economic opportunities for both government and private construction industries around the world. This dissertation focuses on the corrosion of Mechanically Stabilized Earth (MSE) walls. These walls and facings have gained widespread acceptance in North America over the past 30 years (Elias, 2000). The first Reinforced Earth bridge abutment was built in France in 1969 by Terre Armée Internationale. Then, in 1971 the Federal Highway Administration (FHWA) of the US introduced this technology in their construction practices and the first structure in the US was built in 1974. The next year, the FHWA published the “Standard Specifications for Reinforced Earth Walls” to standardize the design methods and the selection of materials. However, these 2specifications were replaced in 1990 by the ones published by the American Association of State Highway and Transportation Officials (AASHTO). After 1990 the construction rate increased to approximately 600 abutments per year, i.e. 300 bridges (Anderson, Gladstone, & Sankey, 2012). A typical MSE wall system has the purpose of providing support for the backfill of roads and bridges in highways and consists of three major structural components: the vertical facing element, the level pad, and the soil reinforcement (Yohchia, 2000).  The backfill soil is typically reinforced by steel or geosynthetic mesh, which supports the vertical facing element of the system. Figure 1-1shows a cross-sectional diagram of the three major elements of an MSE wall.Figure 1-1- Cross-sectional diagram of a MSE wallFigure 1-2 depicts a picture with an example of an MSE wall installed in the Sea to Sky Highway. The picture shows the north to south view of structure number 9488, a bridge that supports the highway on top of the railroad, and has two MSE walls as sub-components. Additionally a second example is presented for a MSE wall providing support for a road constructed on top a stream on Highway 1.   3    Figure 1-2 – North to South view of structure number 9488 located in the Sea to Sky Highway. Two MSE walls are substructures to this bridge, which supports the highway on top of a railroad (above). MSE wall providing support for a road on top of a galvanized steel culvert on Highway 1      4The service life of an MSE structure is defined under the Load and Resistance Factor Design (LRFD) platform that is currently in use as “the period of time during which the factored tensile resistance of the soil reinforcements is greater than or equal to the factored tensile load”(Anderson et al., 2012), Y. Chen, 2000). These reinforced walls are designed to have a minimum service life of 75 years. However, the early failure of some galvanized steel-reinforced structures, such as bridges (Jiang et al., 2009), Mechanically Stabilized Earth (MSE) walls (Padilla et al., 2013, Armour et al., 2004), water pipes (Wu et al., 2010, Carbucicchio et al., 2008), and reinforced structures (Andrade & Alonso, 2004) to mention just a few, have raised concerns of the existing understanding of the factors that lead to early corrosion.  The integrity of the wall depends on two factors: external and internal stability. External stability refers to the global stability of the structure, the bearing capacity of the foundation soils and the appropriate settlement of the structure. Internal stability is closely related to the design of the system components, including checking the pullout and rupture reinforcements (Anderson et al., 2012).Corrosion of steel reinforcements causes internal instability because it slowly degrades the reinforcing material. But the stage of the degradation depends on the characteristics of the reinforcement, the electrochemical properties of the backfill soil and the age of the structure.   Early failure due to corrosion has been attributed to aggressive environments. An aggressive environment is defined by the AASHTO LRFD Bridge Design Specifications as when soil conditions exceed a chloride concentration of >100 ppm per weight, sulphate concentration of >200 ppm per weight, a soil resistivity lower than 3000 ohm per cm, a pH out of the 5 to 10 range, or if the organic or moisture contents exceed 1 wt% (Gladstone et al., 2006, Maslehuddin et al., 2007).   5Exposing the MSE wall facings to these conditions, which are commonly found in Canada, could lead to structural instability and early failure (Thornley et al., 2010).    Figure 1-3 shows an example of a corroded MSE wall located on Highway 1 about 10 km before reaching Revelstoke, BC. The picture depicts the heavily corroded facing of the wall, and a picture of a sample taken from the same wall (Figure 1-3 below), showing the severity of corrosion and loss of cross section. This wall had to be replaced, and a comprehensive investigation on the causes for early corrosion was carried out. The accelerated corrosion was attributed to the heavy use of de-icing salts in the region, coupled with a galvanic corrosion effect originated both by differential oxygen access throughout the structure, and insufficient drainage which caused moisture conditions on the backfill to vary significantly accross the structure.   Figure 1-3 Picture taken of a corroded MSE wall on Highway 1 near Revelstoke, BC (left). Picture of corroded sample of reinforcement taken from the same wall, showing the loss of cross section (right)      6It is important to note that the AASHTO LRFD Bridge Design Specifications were completed based on field data collected in the United States of America and most of the readings were done in places with different climate conditions compared to those commonly found in Canada. As of today, there have not been any comprehensive corrosion studies done taking into account the conditions found in Canada. The proposed project seeks to contribute to the development of a set of construction guidelines for MSE walls and similar structures in aggressive conditions. 1.1 MSE WALLS IN BRITISH COLUMBIA The British Columbia Ministry of Transportation and Infrastructure (BC-MOTI) has over 1100 retaining walls across the province. Figure 1-4 shows the number of retaining walls by wall type. Additionally, there are a large number of walls considered as “sub-structures” for bridges which are not currently listed as MSE walls and thus do not appear in Figure 1-4. British Columbia is likely the province with the largest number of MSE walls and bridges in Canada, and one of the largest in North America.   Figure 1-4 -Total MSE Walls Count by Retaining Wall Type * All wall types with less than 10 built structures in the province were grouped together in the “other” category  Numer of MSE Walls by Retaining Wall Type358308138 127976221 18 6 3 1Precast Concrete Rock SteelBinwallUnknown Wire Face Wood SteelOtherOther Shotcrete Sandbags  7Figure 1-5 shows the number of MSE walls by construction date. In this graph, only MSE walls have been considered and all other retaining walls (such as concrete, wood piles, steel gabion, rock, and sandbags) have been left out to get a better idea on the age of MSE walls in the province.   Figure 1-5- Total MSE Walls Count by Construction Date   MSE Walls by Construction Date0102030405060708090195019521954195619581960196219641966196919711973197519771980198319851987198919911993199519971999200120032005200720092011Year BuiltOther Concrete Precast Wire Face`Coquihalla HighwaySea to Sky Highway  82. LITERATURE REVIEW The corrosion performance of steel in soil and porous media has been extensively studied in the past (Alamilla et al., 2009b, Glass et al., 2000, Gupta & Gupta, 1979, Gardiner & Melchers, 2002), and many studies have been published on the performance of galvanized steel, both in field and laboratory experiments, under varying conditions (Elias, 2000, Gladstone et al., 2006, Yadav et al., 2004a, Macias & Andrade, 1990). Several paths are currently being followed to improve the lifetime of steel-reinforced structures: (i) the use of more corrosion-resistant alloys (Pérez-Quiroz et al., 2008, García-Alonso et al., 2007), (ii) the use of corrosion inhibitors (Söylev & Richardson, 2008, Garcés et al., 2008), (iii) the use of different backfill compositions (Al-Mehthel et al., 2009), (iv) cathodic protection (Parthiban et al., 2008, Hou & Chung, 1997), and (v) corrosion-resistant coatings (Jiang et al., 2011, Lindström et al., 2011, Hamlaoui et al., 2010, Le Manchet et al., 2010a).   From the different approaches to enhance corrosion resistance of the steel reinforcement, corrosion resistant coatings are probably the most cost-effective solution due to their relatively low cost paired with good corrosion resistance. Epoxy Coated Rebars (ECR) and galvanized coatings have been used for the last fifty years in the construction industry and, even though they are not perfect, it has been demonstrated both in studies done in laboratories and in field applications that, if correctly applied, they can slow down corrosion compared to uncoated steel reinforcements.   Epoxy coatings are polymer, resin-based coatings that form a physical barrier to protect the steel reinforcement against corrosion.  They are applied to the steel reinforcement by a simple “paint-like” process at relatively low temperatures and low cost. The main problems of using epoxy coatings are the defects generated in the coating due to improper handling during the installation of   9the rebar, the vast variation in thermal expansion between steel and epoxy coating which causes stress that creates cracks in the coating (Singh & Rita Ghos, 2006,  McHattie et al., 1996), and the decrease of bond strength between the coated steel and its environment (Jalili et al., 2009).  Galvanized steel is one of the most commonly used materials in the construction industry because of its relatively low cost and higher corrosion resistance compared to mild steel. This improved performance is provided by the combination of the barrier action of the zinc layer, the secondary barrier action of the zinc corrosion products, and the cathodic protection of zinc when steel is exposed (Yadav et al., 2004a, Macias & Andrade, 1990, El-Mahdy et al., 2000), so even if the coating is subjected to abrasion prior to or during installation, the overall corrosion resistance of the steel will not be compromised as much as when epoxy coated steel is damaged during installation.  2.1 GALVANIZED STEEL AS CORROSION PROTECTION METHOD FOR MSE WALLS  There have been many attempts to enhance the corrosion performance of galvanized steel. The studies mentioned below show different conclusions on the effectiveness of using rust converters versus normal galvanized coatings. Morris et al. (2000) reported that combining rust converters (a chemical compound containing tannic acid used to form a stable cover layer) with zinc coatings do not perform well under wet conditions where the steel reinforcements may be corroded due to chloride penetration. Singh et al. (2008) reported that in high alkaline environments, zinc gets damaged unless the coating quickly develops a passive film. At a pH below 13.3, the zinc coating successfully protects the steel, but at a pH above 13.3, the zinc coating dissolves continuously until the coating disappears (Manna et al., 2008), thus corrosion takes place at an unacceptable rate in   10highly alkaline environments. By the time the protective layer at the zinc surface is formed, enough damage has already been done, which causes the acceleration of the corrosion rate. Some authors have developed studies in new galvanizing compositions (Simescu & Idrissi, 2008, Ghosh &  Singh, 2007) that show promising results. However, the installation costs would increase, rendering these options impractical for large scale and field applications, such as one pertaining to road and bridge infrastructures.  The corrosion of galvanized steel is a complex process that involves several electrochemical and physical mechanisms. Studies on the corrosion of galvanized steel revealed that the corrosion behavior consists of three different stages (Figure 2-1) (Yadav et al., 2004a, El-Mahdy et al., 2000, Yuan et al., 2009, Yadav et al., 2004b): In Stage I, the corrosion rate increases and the corrosion potential shifts towards less noble values, which implies the acceleration of the anodic process. This stage is mainly related to the dissolution of the oxide layer (ZnO) which was formed in the air. In stage II, the surface of the zinc layer is covered with thick, white rust and the underlying steel begins to corrode. During this stage, the corrosion rate rapidly decreases, accompanied by a shift in the corrosion potential to more noble potentials. This indicates that the anodic dissolution of zinc is inhibited in this stage. Ideally, the zinc coating would act as a sacrificial anode, still protecting the surrounding steel. In Stage III, the amount of red rust on the coating surface rapidly increases. The corrosion rate remains constant and the corrosion potential continues to shift to more noble values. The galvanized steel shows almost the same corrosion potential as that of carbon steel, even though the zinc coating is still covering a few parts of the reinforcement. The underlying steel corrosion progresses by the dissolution of iron and, therefore, at this stage the zinc coating no longer acts as a sacrificial anode (El-Mahdy et al., 2000).   11 Yang et al. (2010) proposed that, during the first stage of corrosion, zinc-based corrosion products precipitated at the reaction sites and gradually expanded to the other recession places. During the second stage of corrosion, the zinc corrosion products continue to expand, depleting the zinc layer. Once the third stage of corrosion is reached, the depletion of the zinc coating, and the dissolution of Zn2+ results in a decrease in pH near the interface; this drop in pH contributes to the corrosion of the steel base once SO4-2 and Cl- diffuse through the barrier, reaching the interface area via the electric field force provided by the corrosion cell (Yang et al., 2010).  Another interesting feature on the corrosion of galvanized steel is the formation of sulphate nests. Weissenrieder et al. (2004) proposed that sulphate nests are blisters formed on the surface, containing high concentrations of corrosion products and electrolytes. These products are thought to be retained by a semi-permeable membrane of colloidal oxyhydroxides. It is believed that sulphate nests form when water diffuses into blisters, or semi-permeable membranes, formed on the surface of the metal. Diffusion causes the semi-permeable membranes to swell and burst, thus allowing for the expansion of the corrosion product on the surface of the corroding metal (Weissenrieder et al., 2004b).  12Figure 2-1 –Three stages of galvanized steel corrosionFigure 2-2 shows three polarization curves obtained from pure zinc, galvanized steel, and a steel sample. This set of experiments was performed to confirm the theory proposed in Figure 2-1. In order to test the electrochemical behavior on uncoated steel, the samples were polished until the galvanized layer was completely removed. SEM and EDX were used to confirm that the zinc had   13been totally polished off the surface. The anodic regions of the curves show that, as the potential is increased, the zinc coating starts to dissolve, and the anodic current density tends to become closer to that of the steel substrate moving away from that of pure zinc (Vagge et al., 2007).  i (A/cm2)1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2 1e-1 1e+0E(v) vs Ag/AgCl-1.4-1.2-1.0-0.8-0.6-0.4-0.20.00.20.40.60.8Pure Zinc SampleGalvanized SteelSteel Sample Figure 2-2 – Comparison of potentiodynamic polarization test results for pure zinc, galvanized steel, and steel sample in 3.5 wt% NaCl, plus 1 wt% Na2SO4 in aerated conditions  This “S” shape, observed in Figure 2-2, is commonly observed in potentiodynamic polarization tests performed on galvanized steel. The inflection point at which the galvanized steel sample begins to show reversal behavior is about -630 mV Ag/AgCl, which can be attributed to the transition between zinc dissolution and the dissolution of iron in steel`(Vagge et al., 2007, Yadav et   14al., 2007). In order to confirm that the reversal behavior can be attributed to the depletion of the zinc coating, the mass loss was calculated using Faraday`s law of electrolysis as shown in equation 2.1:    ݉ ൌ ቀொிቁ ቀெ௭ ቁ              (2.1) Where, m is the mass of the substance liberated at an electrode in grams, Q is the total electric charge passed through the substance, F has its conventional value of 96485 C mol−1, M is the molar mass of the substance (for Zinc 65.409 gmol-1), and z is the valency (z = 2). The value of the total charge transferred at the point where the reversal behaviour is noticed is 34 C. Using equation 2.1, the estimated mass lost at that point is 0.0117 grams. The coating mass was estimated by calculating the volume of the coating, using the tested area (1 cm2) and the coating thickness (20 μm), and multiplying the value with the density of the coating. The two alloying elements of the coating (Zn= 7.14 g/cm3, Al=2.7g/cm3) were taking into account when calcutaing the density value of the coating. The estimated mass of the coating of the tested samples is: 0.0113 grams. The total mass lost at the reversal behaviour is large enough to confirm the dissolution of the coatings, and thus it is safe to conclude that this behaviour can be attributed to the transition between stage 2 and stage 3 of the corrosion process of galvanized steel.   It is clear, based on previous studies performed on the corrosion performance of galvanized steel, that there are some knowledge gaps that are either due to complexity or because experimental difficulties have been left unexplored. The areas of opportunity for further research are: (i) the effect of oxygen access on the corrosion of galvanized steel, especially during the corrosion stages governed by the electrochemical properties of zinc; (ii) the effect of temperature, specifically the corrosion performance of galvanized steel at temperatures relevant to field applications in cold countries; (iii) the effect of an increased amount of soluble salt contents due to the use of de-icers   15and how these enhance corrosion when in the presence of sulphates; and finally, (iv) the development of a corrosion model, relevant to the environmental conditions of cold countries, capable of integrating the complex three step corrosion process of galvanized steel in soil.  2.2 BASIC FEATURES ON THE CORROSION OF GALVANIZED STEEL According to the zinc Pourbaix diagram (Beverskog & Puigdomenech, 1997) zinc should remain immune when the potential is below -1 V, then change to an active corrosion state due to either the dissolution of zinc or the formation of soluble corrosion products, namely Zn(OH)2(aq). According to Dafyyd et al. (2005), predominantly a 2e- reduction process occurs at potentials where zinc is covered with Zn(OH)2, and a predominantly 4e- reduction process occurring at potentials where zinc is bare metal. However, even when the oxygen reduction pathway follows a 2e- process, the formed peroxy species could go through further reduction to form OH-, so the oxygen reduction could follow equations 2.2 to 2.4 when oxygen is available and equation 2.5 when oxygen is not available.  O2 + 2H2O + 4e-  4OH-               (2.2) O2 + 2H2O + 2e-  HO2- + OH-      (2.3) HO2- + H2O + 2e-  3OH-              (2.4) H2O  1/2O2 + 2H+ + 2e-            (2.5) In alkaline and neutral conditions it is generally accepted that the cathodic reaction will follow equation 2.2. The formation of zinc hydroxide or zinc oxide (Type II) by direct oxidation could be explained by the reactions described in equations 2.6 to 2.8.  Zn  Zn2+ + 2e-                (2.6) Zn2+  + 2OH-    Zn(OH)2                            (2.7) Zn2+ + O2 + 2H+ + 2e-   ZnO  + H2O   (2.8)   16Moreover, ZnO could be formed directly from Zn(OH)2 when the pH range is between 7-9 (Feitknecht, 1959).  The process described by equations 2.5 and 2.8 will result in a more desirable corrosion product, ZnO, due its more protective nature when compared to highly soluble Zn(OH)2. Thomas et al. (2012) observed in the pH range from 7 to 10, which is relevant to field applications, acidity arising from anodic processes and metal hydrolysis undermine the passive layer formation favoring the reaction described in equation 2.7 over equation 2.8.  A second mechanism has been proposed (Baugh & Higginson, 1985), in which a porous type of zinc oxide (Type I zinc oxide) precipitates onto the electrode due to the saturation of Zn(OH)42− ions near the metal surface. The mechanism proposed by Baugh et al, is the following:    Zn  + 4OH-    Zn(OH)42− + 2e-                       (2.9)    Zn(OH)42  ZnO  + H2O+ 2OH-          (2.10)  And finally, a third mechanism consist of the formation of a passive oxide when adsorbed species like ZnOH(ads) were reported to reject protons at some critical potential. Additionally, depending on the exposure conditions, ZnO/Zn(OH)2 can further react to form more complex corrosion products. The two most commonly reported have been simonkolleite Zn5(OH)8Cl2, and hydrozincite Zn5(CO3)2(OH)6  (Zhang et al., 2013), but some athors have reported others such as hydroxysulphate Zn4SO4(OH)6 (Elsner et al., 2012). Regardless of the formation mechanism, either direct oxidation or precipitation, it is believed that formation of the type I film is a prerequisite for the type II film (Baugh & Higginson, 1985)    17Similarly, the availability of oxygen will also have an effect on the end product of the corrosion process of steel. Oxygen reduction will occur as given by equations 2.2-2.5, depending on the availability of oxygen. The formation of red rust is expected when oxygen and moisture are readily available and will form, for example, by the reaction described below in equations 2.11 through 2.13 (Broomfield, 1997).  Fe2+  + 2OH + 2e-   Fe(OH)2                (2.11) 4Fe(OH)2 + 2H2O + O2   4Fe(OH)3    (2.12) 2Fe(OH)3   2H2O + Fe2O3·H2O    (2.13) Usually steel corrosion products, such as Fe(OH)2 and Fe(OH)3, will tend to further oxidize to produce secondary corrosion products or even tertiary corrosion products. This mostly depends on the oxygen availability at the metal surface. If oxygen is limited and reduction takes place, as given by equation 2.4, the slow formation of a secondary compact layer could be expected, such as magnetite (Fe3O4) or haematite (Fe2O3), depending on the concentration of oxygen at the metal surface. If oxygen is available at very low concentrations, the corrosion product is likely to be Fe3O4. 3Fe2+  + 1/2O2 +3H2O   Fe3O4  + 6H+            (2.14) 2Fe2+  + 1/2O2 +3H2O    Fe2O3 + 4H+      (2.15)  Regardless of the end corrosion product at any of the corrosion stages of galvanized steel, the amount of oxygen that comes in contact with the galvanized steel reinforcement is difficult to control due the nature of the MSE walls and facing constructions. Thus, differential oxygen access will have a tremendous effect on the corrosion rate of galvanized steel, not only because of the corrosion product formed as previously discussed, but also due to the creation of corrosion macro-cells and the increase of the corrosion rate.    18 Figure 2-3 show a simplified potential-pH diagram for the iron-water system at 25°C done with HSC Chemistry 5 Software ©. Figure 2-3 is in agreement with a previously published diagram in the Atlas of Electrochemical Equilibria in Aqueous Solutions, Pourbaix 1974 (Pourbaix, 1974). The main iron corrosion products presented in the diagram are ferrous hydroxide Fe(OH)2 and iron(III) hydroxide  Fe(OH)3. Previous research supports that the formation of Akaganéite, a poor crystalline form of iron(III) hydroxide, is favored during the corrosion of galvanized steel (Autengruber et al., 2012, Tanaka et al., 2012).  Figure 2-4 show a simplified potential-pH diagram for the zinc-water system at 25°C  done with HSC Chemistry 5 Software ©. Figure 2-4 is also in agreement with previously published diagram in the Atlas of Electrochemical Equilibria in Aqueous Solutions, Pourbaix 1974 (Pourbaix 1974). Alternatively, Figure 2-5 shows the potential-pH diagram for the zinc-water system at 25°C with ZnO as the stable corrosion product in the system. Both ZnO and Zn(OH)2 are formed in the same pH range, and the different formation mechanism have been discussed above.      19Figure 2-3 - Potential-pH diagram for the iron-water system at 25°C, concentration of species 10-6Figure 2-4 Potential-pH diagram for the zinc-water system at 25°C, concentration of species 10-61614121086420-22.01.51.00.50.0-0.5-1.0-1.5-2.0Fe - H2O - System at 25.00 CEh (Volts)H2O LimitsFeFe(OH)2Fe(OH)3Fe(+3a)Fe(+2a) FeOH(+a) HFeO2(-a)1614121086420-22.01.51.00.50.0-0.5-1.0-1.5-2.0Zn - H2O - System at 25.00 CEh (Volts)H2O LimitsZnZn(OH)2Zn(+2a) ZnO2(-2a)pHpHSoil pH rangeSoil pH range20Figure 2-5- Potential-pH diagram for the zinc-water system at 25°C showing ZnO as the stable corrosion product, concentration of species 10-6As mentioned before, the corrosion process of galvanized steel can be summarized in three different stages. In stage 2, the zinc cover partially dissolves, and the underlying steel is exposed to the soil and, therefore, in addition to zinc, steel is also involved in the corrosion reaction. During this short second stage, Zn2+ dissolved from ZnO and Zn(OH)2 products suppress the crystallization and particle growth of β-FeOOH, thus providing an inhibitory effect (Tanaka et al., 2012). Relatively new techniques such as Localized Electrochemical Impedance Spectroscopy (LEIS) and Scanning Electrochemical Microscopy (SECM) have been useful in studying, in more detail, the processes involved during the second stage of corrosion, showing that zinc corrosion products successfully protect the exposed metallic surface against the corrosion (Mouanga et al., 2013), zinc undergoes a higher corrosion rate in the vicinity of exposed steel, and it has been confirmed that during Stage 2, 1614121086420-22.01.51.00.50.0-0.5-1.0-1.5-2.0Zn - H2O - System at 25.00 CEh (Volts)H2O LimitsZnZnOZn(+2a) ZnO2(-2a)pHSoil pH range  21while the anodic process still occurs mainly on the zinc surface, the oxygen reduction process takes place on the steel surface (Mouanga et al., 2013). 2.2.1 CORROSION IN SOIL  Corrosion in soil is a very complex process due to the complex nature of soils. Soil has a wider range of chemical and physical compositions when compared with other corrosive media. The pH of soil may vary significantly from acidic to alkaline conditions, and the soil resistivy also ranges from several tens to mega ohms (Romanoff, 1977).  There are very few comprehensive studies performend on the corrosion performance of galvanized steel in soil, one of them is the study performed by Darbin/Romanoff which was originated based on a very comprehensive database collected on a National Bureau of Standards (NBS) study. This study will be discussed more in depth in later sections of this document. Apart from differences in texture and porosity, soil differs from other porous media because a rangfe of chemical compunds exists in soils. The soluble base-forming elements are sodium, potassium, calcium and magnesium. The acid forming species are carbonate, bicarbonate, chlories, nitrates, and sulfates (Zhang, 1996). These compounds affect both the soil pH, and the ability to conduct electric current. Additionally, it has been recorded that poorly and very poorly aerated soils are more corrosive to zinc, and high corrosion rate is not always associated with pitting. On the other hand, soils with good aeration, but containing chlorides and sulfates, tend to induce pitting (Romanoff, 1977).  The main parameters affecting corrosion in soil are pH, soil resistivity, moisture content, and oxygen content. All these parameters have been accounted for in chapter 9 and 10 of this thesis when developing the corrosion model.     22The value of the electrical resistivity of the backfill soil depends on several factors:  a) Nature of solid constituents: It refers to the particle size and their mineralogy, i.e. chemistry and physical properties. Most soil minerals are insulators, but each solid particle possesses a certain amount of negative charge at the surface that attracts exchangeable cations (Kibria, 2011). For coarse materials, conductivity at the surface is negligible compared to conductivity of the pore fluid.  b) Degree of saturation: Measurement of the amount of dissolved ions from the pore fluid that are adsorbed by the solid surface. When the degree of saturation is higher than the minimum amount of water to maintain a continuous film of water surrounding the solid particles, an increase in the degree of saturation will result in an increase of resistivity. c) Size of particle: Soils with smaller particles have bigger specific surface area and therefore exhibit higher conductivity at the surface than soils with coarse particles of the same mineralogy. However, it has been concluded that, in general, surface conductance is negligible compared to the conductance of interstitial water, which contains dissolved electrolytes. Only in soils with high clay the decrease of the resistivity of the bulk would be appreciable.  d) Porosity and pore structure: The electrical resistivity in porous media depends on the movement of ions in the water in the void spaces. The void geometry (distribution and form) will influence the proportion of air and water than can be retained in the soil. This property depends on the pore size distribution and connectivity.  e) Water content and ionic concentration: The electrical resistivity in the pores depends on the amount of water, the viscosity, the concentration of dissolved salts in it and the ionic composition.  f) Temperature: An increase in temperature decreases the viscosity of water and increases the ion agitation resulting in a decrease of resistivity.    232.3  CORROSION OF GALVANIZED STEEL IN TEMPERATURES RELEVANT TO FIELD APPLICATIONS (-5°C TO 25°C) There is very limited information dealing with the effect of temperature on the corrosion performance of zinc and galvanized steel. A variation in temperature could have a direct impact on the electrical resistance of the soil even in the absence of soil salinity (Sagüés et al., 2000). Temperature also influences many other corrosion related parameters, such as oxygen solubility, diffusion rates, activity coefficients, enthalpies of reaction, compound solubility, corrosion rates, biological activity, and coating bonding strength. Additionally, since temperature is a non-controllable factor influenced by environmental conditions and natural climate cycles, it is imperative to quantify its effect on the corrosion performance of these structures so that it can be properly considered during the design process. Yamashita (1980) found that the corrosion current of zinc in the presence of sulphates and oxygen may increase with temperature, but decreases with temperature in the absence of oxygen.  The difference was due to the distinct formation of the corrosion products. Generally, sulphate and chloride ions caused a decrease in corrosion potential of zinc immersed in water with increasing temperature (Hoxeng, 1949, Hubbard & Shanahan, 1973)  Both field and laboratory studies on the electrical resistance of ice suggest that decreasing the temperature of water below the freezing point would result in a significant increase in the resistance of the environment, reaching magnitudes of megaohms·cm (Roman, 1938, Ostrem, 1967, Reynolds & Paren, 1984), thus a remarkable drop in the corrosion rate is expected. This is because, at freezing temperatures, a greater energy barrier needs to be overcome for ion migration due to a substantial drop in the total accumulation of ions and in the decrease in ionic permeability of ice and frozen soils (Chuvilin et al., 1998, McMillan et al., 1982a).    24 It is important to remember that, in the cathodic region, the limiting current density (iL) is the most important characteristic regarding diffusion-controlled processes (Ajeel & Ali, 2008), and it is expected to decrease the oxygen availability with decreasing temperature, thus causing a decrease in the limiting current density value. Nevertheless, in the presence of salts, and when the ice layer is relatively thin and the temperature is close to zero, corrosion still represents a threat for the structural integrity of MSE walls because of a drop in soil resistance. When studying the resistance of seawater with brine, Ingham, et al. (2008), found that, close to the ice–water interface, resistivity drops rapidly from 0.1 or 1 mega Ω · cm to less than 10,000 Ω cm when the temperature is above −5°C and the brine volume fractions exceed 8−10 wt% (Ingham et al., 2008).    The exchange current density of metals can be affected by the pH, temperature, and characteristics of the substrate (Pour-Ghaz et al., 2009). It could be assumed that the pH surrounding an MSE wall remains relatively constant due the use of soils meeting AASHTO criteria.  The US Transportation Research Board reported after several field measurements that no significant differences on the corrosion rate of galvanized reinforcements were observed in different climatic regions (Fishman & Withiam, 2011). The means of the corrosion rates were observed to be approximately 1 μm/yr to 2 μm/yr (Fishman & Withiam, 2011). However, the effect of temperature variation cannot be ignored because temperature is a non-controllable factor that directly affects the anodic and cathodic exchange current densities and, therefore, the corrosion rate, as well as its equilibrium potentials.     25Additionally, even if in cold regions one could expect corrosion to be negligible due the reasons discussed above, evidence of early MSE wall deterioration in cold regions seems to point to the opposite conclusion. If corrosion at freezing conditions should be expected to happen at a slow rate, why is this still an issue? It is then imperative to deepen the understanding of the corrosion performance of galvanized steel reinforcement in colder climates where, despite following AASHTO criteria, the MSE walls and similar structures are lacking in performance.  2.4  EFFECT OF DE-ICERS ON THE CORROSION OF GALVANIZED STEEL AASHTO backfill is the preferred material for construction because, when it is correctly processed, it helps in retarding the corrosion initiation of the reinforcement by providing an alkaline environment. This helps galvanized steel to form a more stable passive layer (Glass et al., 2008), even when the zinc coating has been consumed. Nevertheless, the breakdown of the passive film allows the corrosion process in the steel to occur. This process can be activated by a number of different factors, however, two are regarded as the most common: carbonation and chloride penetration (Glass et al., 2000, Page et al.,1982, Moreno et al., 2004, Etteyeba et al., 2007, Cramer et al., 2002).   Carbonation is a phenomenon observed when calcium hydroxide of the concrete matrix transforms to calcium carbonate (among other products) due the ingress of carbon dioxide from the environment. It is not a major issue when addressing corrosion performance of MSE walls because the concrete cover is not in direct contact with the galvanized steel reinforcement. In the case of MSE walls, the concrete cover, when used, is constructed only for aesthetic purposes.      26Chloride penetration on the other hand, is an important issue due the nature of the porous backfill used in MSE walls.  When a critical chloride content is reached, the pH of the porous environment drops and the corrosion process begins. The critical chloride is also referred to as the chloride threshold by some authors, and it is believed that 2% chloride concentration per weight of cement is enough to provoke the corrosion process in steel (Pérez-Quiroz et al., 2008, García-Alonso et al., 2007). Due to the nature of the chloride penetration process and the relatively high use of de-icing salts in cold regions, it is relevant for this research to look into the contribution of different de-icing salts to this process as well as their behavior when sulphates are present in the environment.  The use of de-icing salts is a complex problem that involves an enormous quantity of variables of interest, ranging from corrosion and maintenance costs to environmental and health damages associated with the use of different de-icers. However, due to the complexity of this issue, the proposed research will be narrowed to the effect of de-icing salts on the corrosion of the galvanized steel reinforcement used in structures such as MSE walls.  De-icing salts are used to lower the freezing point of water and to avoid the reduction of friction between the car tires and the pavement due to accumulation of ice on the road.  These salts applied onto highways often contain chlorides because of their cost-effectiveness. The main salts used are sodium chloride (NaCl), magnesium chloride (MgCl2), calcium chloride (CaCl2), and less commonly used acetates (Fay et al., 2008). Some believe that CaCl2 increases the corrosion rate as compared to other chloride-based de-icers such as NaCl, while acetate-based de-icers and other agricultural compounds do the least damage (Wang et al., 2006, Flis et al., 1998, Amrheln et al., 1992). Other sources, however, oppose the use of MgCl2 as a de-icer because of a reported increase   27in the corrosion rate when compared with NaCl, when used in a concrete environment (Mussato et al., 2004) due a higher chloride diffusion rate. The cation (Na+, Ca2+, or Mg2+) associated with Cl- also affects the chloride diffusion coefficient. The chloride diffusion coefficients for MgCl2 are typically two to three times greater than NaCl, which may significantly reduce the time-to-corrosion initiation for the steel reinforcement. The effective diffusion coefficient of CaCl2 was found to fall between that of NaCl and MgCl2.  Potassium acetates are gaining popularity as de-icers in the aircraft industry (Huttunen-Saarivirta et al., 2009, Corsi et al., 2009) due its alleged low corrosiveness compared to chloride-based de-icers and lower toxicity, resulting in a lower negative effect on the surrounding environment. However, caution has been advised because of secondary reactions, such as the alkali silica reaction which results in concrete deterioration when potassium acetate or similar chemicals are used as de-icers (Rangaraju et al., 2006). Acetate-based de-icers are considered to be non-corrosive to mild steel, but they are comparably corrosive as chloride-based de-icers to galvanized steel. For practical purposes, all chloride-based de-icers were ranked equally high in causing corrosion of the reinforcing steel in a recent NCHRP study, even though hygroscopic chlorides of magnesium and calcium can be more aggressive to the exposed metals than NaCl because of the longer time of wetness (Xianming et al., 2009). Ultimately, the choice of the de-icing salt used depends on the type of structure, whether concrete is present or concrete integrity is important, cost, and the temperature in the surrounding environments; the effective application temperatures for CaCl2, MgCl2, and NaCl are −25°C, −15°C, and −10°C, respectively (Yehia & Y. Tuan, 1998). Because of the lack of agreement in the existing literature on which de-icer is most suitable for road infrastructure, it is   28important to develop a comprehensive study on how these de-icers affect the corrosion performance of galvanized steel in the presence of sulphates. 2.5 MATHEMATICAL MODELING OF THE CORROSION OF GALVANIZED STEEL  Because of the complexity of the galvanized steel corrosion process, combined with the dynamic nature of soil as complex, non-uniform porous media, predicting the corrosion rate of galvanized reinforcements in soil as a function of time is quite challenging. At the same time, it can be very valuable due its potential application for the engineers who need to determine the service life of the MSE walls during the design process.   Weight-loss coupon tests, solution analysis and measurements of resistance, detection of galvanic current, electrochemical resistance probes, and electrochemical methods such as linear polarization resistance (LPR) and electrochemical noise techniques, are some of the traditional methods used in corrosion monitoring.   Coupon tests, used in both LPR and weight loss measurements, has been the most widely used method in providing baseline criteria in many corrosion monitoring programs and models (Elias, 2000, Zhang, 1996, Romanoff, 1957, Elias, 1990), however, the drawbacks associated with this technique are the inability to monitor the time dependency of the corrosion process and the long exposure time (Baboian, 2005), often limiting the applications of these methods. Potential mapping is another widely used and simple technique, but it only provides information on some thermodynamic corrosion parameters and it does not give any information on kinetics of corrosion (Elsener, 2001). Measurement of the galvanic current is often used to detect environmental corrosion in inaccessible parts of a structure, but the main limitation is that it heavily relies on empirical   29knowledge and, in reality, it provides information on the electrolyte corrosivity rather than the actual corrosion rate or corrosion mechanism (Tan, 2011). Electrical resistance probes have been found to be useful in detecting generalized corrosion but lack in their ability to detect other types of corrosion, such as localized corrosion (Legat, 2007). Electrochemical Noise techniques and other modified configurations such as coupled multi-electrode array sensors (CMASs) and wire beam electrodes (WBE), have shown promising results in detecting localized corrosion; however, the setups are relatively complicated and are not able to measure rapid electrode processes (Tan, 2009) and have even been deemed not reliable as real-time sensors (Yang et al., 2005). Each of these methods presents both advantages and disadvantages that are inherent to the theory behind the method and the application limitations; nevertheless, they all require the use of costly equipment and trained experts for their use and analysis, limiting the use and application of these techniques.   Corrosion prediction is a different and complementing approach to the methods mentioned above. The ability to influence decisions at the design stage is one of the main advantages that corrosion prediction has over corrosion monitoring.  Corrosion prediction is generally done through mathematical models used to calculate accumulated corrosion in a given period of time.  A number of models have been developed over the past sixty years to model the corrosion process of carbon steel as well as galvanized steel in reinforced structures. Although they offer different solution methods, such as the finite element method (Alamilla et al., 2009b, Thébault et al., 2007, Amleh & Ghosh, 2006, Munn & Devereux, 1991, Melchers, 2003, Pour-Ghaz et al., 2009a, Pour-Ghaz et al., 2009b, Doig & Flewitt, 1979), finite difference method (Doig & Flewitt, 1979, Brow & Barnard, 2006, Barnard & Brow, 2009), or the boundary element method (Abootalebi et al.,   302010, Farid Uddin et al., 2007), they all estimate corrosion rates by using a combination of Laplace’s equation and Ohm’s law.   Most of the models previously developed for galvanized steel in soil are able to predict only the average corrosion rate during the life of the structure and do not consider the effect of environmental conditions such as temperature and humidity variation or soil parameters such as the alkalinity, resistivity, and moisture content of the soil on the corrosion rate.  Currently, there are four widely accepted and used models: the Darbin/Romanoff Model, the Stuttgart Model, the Caltrans Model, and finally, the AASHTO Model (Elias, 2000, Fishman & Withiam, 2011, Romanoff, 1957, Rehm, 1980, Jackura et al., 1987, Darbin et al., 1988).  The Darbin/Romanoff model was originated based on a very comprehensive database collected on a National Bureau of Standards (NBS) study and, later, elaborated on and improved to better estimate the corrosion of MSE walls by Darbin et al. in 1988 and by Elias in 1990, and is applicable for the conditions generally found in MSE applications with a soil resistivity greater than 1,000 Ω-cm. The rest of the models are linearized forms of this model. The Stuttgart Model was proposed by Rehm in 1980, in which the change of the corrosion rate between the three different stages is taken into account by assuming that the corrosion of zinc is usually greater in the first 2 to 4 years followed by a reduced rate and, finally, by a corrosion rate governed by the bare steel consumption. Rehm proposed two different models depending on the soil conditions surrounding the MSE wall. The first model is used when the backfill in the MSE wall meets specification, and a second model when it does not. Both the Darbin/Romanoff and the Stuttgart Models contribute to the basis of the AASHTO Model, in which additional data was included from laboratory tests. This model is similar to the   31Stuttgart model with only two main differences: first, it requires the backfill to have a resistivity greater than 3,000 Ω·cm, and second, the corrosion rate of zinc after the second year of service life is twice the value considered in the Stuttgart model, thus, the AASHTO model is considered to be conservative. Finally, based on the work by Jackura et al. (1987), Caltrans proposed a model for a wider range of backfill conditions: a soil resistivity greater than 2,000 Ω.cm, a pH between 5.5 and 10, and chloride and sulphate concentrations of 250 ppm and 500 ppm, respectively, and considers that the zinc coating will be depleted after 10 years of service life as opposed to the 15 years proposed in the AASHTO model (Elias, 2000).  These models have two major shortcomings: they assume conditions will remain constant throughout the lifetime of the structure, and, the models can only predict corrosion performance within an established range that does not include aggressive conditions. Current models cannot successfully predict corrosion performance of galvanized steel in aggressive conditions. Because of these disadvantages, it is necessary to develop a flexible model that, considering all three corrosion stages, will be able to predict corrosion performance in conditions relevant to those commonly found in cold countries.      323. OBJECTIVES The goal of this work is to develop a comprehensive understanding of the corrosion behavior of hot-dip galvanized steel in aggressive soil conditions. 3.1 KEY TECHNICAL OBJECTIVES 1. Fill the knowledge gaps identified in the field of study of electrochemical  behavior of the soil corrosion performance of hot-dipped galvanized steel:  a.  Study the effect of variable oxygen concentration on the corrosion performance of MSE walls. Oxygen content will range from full oxygen saturation (loose soil compaction), ~10 ppm, to complete oxygen depletion (perfect soil compaction), ~ 0.01 ppm. b.  Study the corrosion performance of MSE walls in temperatures relevant to field applications. Temperate will range from -5ºC to 25ºC. c. To extend the understanding of the interaction of the commercially available de-icing salts with sulphates and how these affect corrosion of galvanized steel. A fixed concentration of 3.5 wt% of de-icings, NaCl, MgCl2, CaCl2, or CH3CO2K; with Na2SO4. 2. Develop a mathematical model to predict the corrosion of galvanized steel at the three different stages of corrosion and validate the model using available literature data, as well as field data and experimental data collected during this project. In practice, this work will also contribute towards the improvement of the current construction guide used to predict the design life of Mechanically Stabilized Earth (MSE) walls and facings. A better understanding of the factors that lead to early corrosion on MSE walls and facings, coupled with information on the optimum galvanized coating thickness, can increase the service life of road infrastructure, particularly MSE walls and facings.     334. APPROACH AND METHODOLOGY In order to achieve the proposed goals, this work will include a comprehensive set of experiments coupled with theoretical modeling. The corrosion studies were performed in a way that each knowledge gap or specific goal is being studied separately. Each set of experiments are performed using a combination of advanced electrochemical techniques such as: Potentiodynamic Polarization (PDP), Electrochemical Impedance Spectroscopy (EIS), Linear Polarization Resistance (LPR), and Cyclic Voltammetry (CV) measurements. PDP is used to obtain basic corrosion parameters; EIS and CV measurements are performed in order to understand the corrosion protection mechanism of the coatings and as supplementary techniques to PDP. X-Ray Diffraction (XRD) and Scanning Electron Microscope (SEM) are used to characterize the corroded samples the zinc oxide layer formed.  4.1 MATERIALS: In order to effectively assess the effect of the studied variables on the three different corrosion stages of galvanized steel, the electrochemical tests are performed using an as-received, galvanized, low carbon, cold-rolled galvanized steel sheet with an exposed area of 1/cm2 instead of the actual MSE reinforcements. The galvanized steel sheet used is fabricated under the ASTM A653 specification with a sheet thickness of 0.03 cm and a coating thickness of approximately 20 µm ± 2 µm, the bath composition for the hot-dip coating was 99.96 wt% zinc and 0.04 wt% Al, without a chromate finish. The presence of aluminum in the coating surface has been confirmed with Energy Dispersive X-ray (EDX). Chromium was not detected on the surface either by EDX or XRD. Each individual sample was cut out of the as-received sheet. All samples are cleaned with ethyl alcohol and dried prior to being immersed in the test solution.     34The coating thickness of 20 µm was selected to ensure that the corrosion tests performed on galvanized steel would successfully capture all three stages of corrosion. As mentioned in earlier sections, the typical thickness in field applications is approximately 86 µm. During the first stage of corrosion of galvanized steel, the electrochemical behaviour of the reinforcement is that of pure zinc coating. Because the corrosion during that first stage is uniform, it is expected that changing the thickness of the galvanized coating will not affect the significance of the results presented in this thesis. It will only shorten the duration of the first stage of corrosion in the accelerated tests.   For the experiments performed during this project, the presence of aluminum on the galvanizing composition was not a concern. A previous study demonstrated that zinc and high purity zinc coatings (Zn0.1Al and Zn4.3Al) show similar electrochemical behavior when studying the oxygen reduction pathway (Dafydd et al., 2005). It was concluded that the alloyed zinc coatings behave electrochemically as though it were pure zinc.  4.2 TEST METHODS: Potentiodynamic polarization tests are conducted in different solutions to study the effect of chloride-induced corrosion due the use of de-icing salts (NaCl, MgCl2, CaCl2, and CH3CO2K) in the presence of sulphate ions (Na2SO4) at temperatures ranging from -5ºC to 25ºC, and controlled oxygen concentrations as the experimental needs warrant. The temperature is monitored and controlled using a VWR cooling and heating bath with a programmable temperature controller. A magnetic stirrer is used at a rotating speed of 300 rpm to obtain a slight vortex at the electrode. The oxygen content is measured using an Omega DO Meter DOB21 oxygen sensor. The pH values were monitored both at the beginning and the end of the tests; however the values were not controlled 35during the tests. The pH values remain fairly constant during the duration of the tests, and did not change significantly after the tests. Figure 4.1 shows a schematic of the experimental set-up.Figure 4-1- Schematic of experimental set-upThe electrochemical corrosion tests were carried out using a conventional three-electrode electrochemical cell. The samples, either galvanized steel or zinc, are to be placed in a water-jacketed cell as a working electrode, along with a graphite counter electrode and an Ag/AgCl ([Cl-1]=4 M) reference electrode. Electrochemical measurements were performed with a Princeton Applied Research (PAR) VersaStat 4 potentiostat/galvanostat.  36Sample Org. Cont. (wt%) Conductivity (μS/cm) pH SO4-2 Cont. (wt%) Cl- Cont. (wt%) Moist. Cont. (wt%) 1 2.52 297 6.8 0.4 0.8 1.872 3.05 249 6.0 0.4 2.2 0.133 4.61 313 6.6 1.2 2.8 10.234 1.21 295 8.6 0.7 1.0 1.805 0.87 188 6.8 0.9 1.3 3.676 3.32 274 7.0 0.3 1.9 1.077 5.32 276 5.7 11.3 3.1 3.358 3.65 120 6.1 4.7 1.6 1.569 5.43 101 5.9 6.8 1.0 2.6610 1.24 114 7.5 0.4 0.6 0.1611 2.55 66 5.8 6.2 0.6 0.4412 0.53 207 6.2 6.0 0.3 2.1013 4.48 354 6.4 21.1 1.3 0.1914 1.88 212 6.7 3.7 1.8 0.1715 0.70 557 8.6 1.1 12.9 0.0416 11.08 243 6.0 8.3 4.7 6.6717 2.44 85 6.6 7.0 0.9 0.8518 2.34 123 6.5 7.8 1.0 0.1819 4.61 119 6.2 1.2 1.9 5.2920 5.85 175 5.5 0.2 2.2 3.2721 1.46 192 6.4 4.9 0.7 15.2922 4.79 151 5.3 0.6 1.4 1.1823 2.02 77 5.8 <0.1 0.5 0.1724 4.02 45 5.7 1.0 1.2 0.3725 3.52 119 5.9 0.1 1.4 0.3026 2.44 41 5.6 3.0 0.7 0.3727 2.06 83 6.0 <0.1 1.0 0.2228 2.20 92 5.9 <0.1 2.4 0.2129 7.52 179 6.5 0.6 2.0 1.4330 1.68 2375 7.5 19.6 1.8 0.6531 2.98 300 6.5 10.0 3.2 0.2832 3.47 156 5.7 0.5 2.3 1.2133 9.76 274 6.2 13.3 1.9 12.3034 1.61 2012 7.1 16.6 0.9 0.21Table 4-1 –Electrochemical Properties of soil samples collected across British Columbia  Table 4-1 shows the average electrochemical properties of soil samples collected across British Columbia. The samples were collected on 34 selected sites in different highways with distinct de-icing salt use (ranging from cold to mild weather conditions). The soil was collected at a superficial level (30 to 40 cm approximately below the surface) at both the top and the bottom of the structures. As evidenced by the data shown on the table, the sulphate and chloride content used on the electrochemical tests on this work might not reflect the concentrations we typically see in the field.   37Consequently, the laboratory results shown in this work might not be directly transferable to field applications; however it was decided to use higher concentrations in order to perform accelerated testing. The general trends and the conclusions on the electrochemical behaviour of galvanized steel in aggressive conditions presented in this work are stil valid.  Each set of tests are confirmed by at least two of the following techniques (parameters are described below):  Prior to each experiment, samples are held at Open Circuit Potential (OCP) for one hour in order to monitor the potential-time behavior. Most of the galvanized steel samples and zinc samples reach OCP within 100 seconds, however, OCP measurements are always done within one hour to ensure that a steady state potential was reached.  PDP testing is carried out between 0.250 V Ag/AgCl below the open circuit potential (OCP) to 1 V Ag/AgCl above the OCP using a 4M KCl saturated Ag/AgCl reference electrode. The scan rate used is 0.166 mV/s. The potentiodynamic polarization (PDP) tests were conducted three times on replicate samples to ensure reproducibility of the results. The reproducibility of the PDP curves was acceptable, as the variation between the repeating tests did not affect the performance ranking of the galvanized steel in various solutions: the variance of Ecorr values was less than 20 mV and the variance for the corrosion rate among the repeated tests was less than 11 µm/yr, or 0.8 µA/cm2. Figure 4.2 shows an example of the reproducibility among the three repeated samples testes in 3.5 wt% CH3CO2K. Temperature was monitored and controlled during each experiment as needed.  EIS is used to study the effect of the studied variables on the electrolyte/corrosion layer interaction. This is done to determine whether galvanized steel forms a protective corrosion   38layer under the studied conditions or if the corrosion products formed on the metal surface are able to protect the metal from further dissolution. EIS measurements are performed every hour over a 24 hour period at OCP in the frequency range of 0.01 to 10,000 Hz, amplitude of 10 mV, and a sampling rate of 10 points per decade.  The 24 hour LPR tests are performed between 20 mV Ag/AgCl below OCP to 20 mV Ag/AgCl above the OCP. The scan rate used is 0.166 mV/s. Short term LPR measurements are taken every hour over a 24 hour period, and long term LPR measurements will be performed daily.  Cyclic Voltammetry (CV) is performed at an optimum scan rate of 20 mV/s. The cyclic voltammograms are obtained over a voltage range of 275 to -500 mV Ag/AgCl at room temperature.  Cyclic Polarization (CP) was performed by starting the scanning electrode potential from an initial potential of 250 mV below the open circuit potential (OCP) up to 0 V. When the electrode potential reached vertex potential of 0 V, the potentials were scanned back to the initial potential.   39i (A cm-2)10-7 10-6 10-5 10-4 10-3E(V) vs Ag/AgCl-1.2-1.0-0.8-0.6-0.4-0.20.0 3.5 wt% CH3CO2K- Sample 13.5 wt% CH3CO2K- Sample 23.5 wt% CH3CO2K- Sample 3 Figure 4-2- Selected results of three replicates of PDP tests performed on samples immersed in 3.5 wt% CH3CO2K 4.3 MODELING: The modeling part of the work simplifies the corrosion behavior of galvanized steel into three stages. In order to consider the three-stage corrosion behavior of galvanized steel, a combined macrocell and microcell corrosion mechanism is assumed as a governing corrosion mechanism of the galvanized steel reinforcement in the soil. Due the complexity of this corrosion process, there is a need for a two-dimensional domain approach; therefore, it is not possible to define a closed-form solution. Thus, a nonlinear finite element solution technique is used in this study to solve the governing differential equation of the potential distribution in the domain of the problem (i.e. soil), and to calculate the corrosion current densities on the surface of the galvanized steel reinforcement. The analysis is carried out in a domain that is 300 mm long; the domain is discretized by triangle finite elements. Please see section 5.3 for more details on the modeling concept.    405 EFFECTS OF OXYGEN AND SULPHATE CONCENTRATIONS ON THE CORROSION BEHAVIOR OF ZINC IN NACL SOLUTIONS:  Any corrosion process is promoted by two independent variables: oxygen and moisture (Mangat & Molloy, 1992). When either the amount of oxygen or the corrosive medium conductivity increases, it is expected that the corrosion process will be enhanced. The effect of oxygen access in conjunction with other aggressive agents on the corrosion performance of bare steel has been well documented in the past (Glass et al., 2000, Mangat & Molloy, 1992, Gonzalez et al., 1993), but the effect of variable oxygen concentration on galvanized steel has not been as comprehensively researched and, due the nature of the MSE wall design and its components, this is a factor that cannot be overlooked and should be taken into account. The facing element is usually exposed to the environment, allowing a greater access of oxygen to the metal surface.  A number of different factors can promote the conditions to have a variable oxygen access on a structure: partial immersion of the structure in an aquatic/marine environment, partial coverage of the reinforcement by concrete, or simply a gradient in the soil compaction (Elias, 2000).  Variation in oxygen access to the corroding metal is expected to promote corrosion macrocells with the cathode on the site with higher oxygen content (Sagüés et al., 2000) and with possible corrosion enhancement near the high stress regions, such as the facing and load-bearing elements. The availability of oxygen can also affect the oxidation and corrosion pathway and, ultimately, will also have an impact on the expected corrosion product during any of the corrosion stages.   The objective of this section is to present the results on the performance of zinc as a protection method under aggressive environments, and to study the effect of differential oxygen access on the   41corrosion performance of zinc.  As a first step, we started with experiments with the aim to fill in the knowledge gaps identified in the literature review process. The effect of variable oxygen access in conjunction with other aggressive agents on the corrosion performance of steel has been well documented in the past (Glass et al., 2000, Mangat & Molloy, 1992, Gonzalez et al., 1993, Pettersson, 1996), but the effect on zinc has not been as thoroughly researched in the past. This chapter is based on a paper published as part of the work towards this PhD thesis. 1 The electrochemical tests were performed using zinc (99.6 wt%) with an exposed area of 1/cm2. Pure zinc was used during this first set of experiments to test whether or not the first stage of corrosion of galvanized steel was comparable to the corrosion of pure zinc. All the pure zinc samples were mounted, polished using up to a 1200 grit sand paper, and cleaned with ethyl alcohol prior to being immersed in the test solution. Each sample was immersed at room temperature in solutions with a fixed 3.5 wt% of NaCl, and concentrations of Na2SO4 ranging from 0% to 4 wt% to simulate the aggressive condition found in the field. Three different concentrations of oxygen: fully saturated with oxygen, aerated, or de-aerated, were used to simulate the differential oxygen access. Table 5-1 below shows the experimental conditions used. Sample Na2SO4 (wt%) Condition Solution pH A0 0% O2 saturated 8.2 B0 0% Aerated 8 C0 0% De-aerated 8.1 A1 1% O2 saturated 8.2 B1 1% Aerated 8 C1 1% De-aerated 8.1 A3 3% O2 saturated 8.4 B3 3% Aerated 8.3 C3 3% De-aerated 8.4 A4 4% O2 saturated 8.4 B4 4% Aerated 8.7 C4 4% De-aerated 8.4 Table 5-1 - Different experimental conditions used on the zinc samples                                                  1 Victor Padilla, Akram Alfantazi,  Corrosion, 68 (2012)   425.1 POTENTIODYNAMIC ELECTROCHEMICAL TESTING All samples reached a steady open circuit state within the first 100 seconds of the test. Sziráki et al., proposed that zinc forms an oxide/hydroxide spontaneously in the presence of aggressive ions (1998), therefore achieving a stable condition quickly.   Figure 5-1a shows three OCP measurements for the control samples immersed in 3.5wt% of NaCl; as the concentration of oxygen decreases on the solution OCP values were shifted to more negative values. In Figure 5-1b, the effect changing the concentration of Na2SO4 becomes evident. As the concentration of sulphate ions in the solution increases, OCP values decrease. The effect of changing the amount of oxygen on the open circuit readings appears to be slightly higher than varying the quantity of Na2SO4.    PDP testing was performed in order to assess the effect of oxygen and Na2SO4 concentration on zinc and understand how varying these two affect corrosion rates. Figure 5-2 shows PDP curves for zinc samples immersed in 3.5wt% NaCl, and 0wt% Na2SO4 with different concentrations of oxygen. Figure 5-3 shows zinc samples immersed in 3.5wt% NaCl, and 1wt% Na2SO4 with different concentrations of oxygen. Figure 5-4 depicts zinc samples immersed in 3.5wt% NaCl, and 3wt% Na2SO4 with different concentrations of oxygen. Finally Figure 5-5 shows zinc samples immersed in 3.5wt% NaCl, and 4wt% Na2SO4 with different concentrations of oxygen. In agreement with the OCP measurements, results of the  potentiodynamic polarization tests suggest that oxygen availability has a greater effect on the corrosion rate of the samples than increasing the amount of Na2SO4.      43                     Figure 5-1 a) OCP measurements for the control samples immersed in 3.5wt% of NaCl in oxygen-saturated conditions, aerated conditions, and de-aerated conditions, and b) OCP measurements for the samples immersed in oxygen-saturated solutions with 3.5wt% of NaCl, and different concentrations of Na2SO4 (1-4wt%)  Time (s)0 1000 2000 3000E(v) vs Ag/AgCl-1.3-1.2-1.1-1.0-0.9-0.8-0.7Oxygen Saturated AeratedDe-aerated a) Time (s)0 1000 2000 3000E(v) vs Ag/AgCl-1.1-1.0-0.9-0.8-0.70 wt% Na2SO4 1wt% Na2SO43wt% Na2SO44wt% Na2SO4b)   44i (A/cm2)1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2 1e-1 1e+0E(v) vs Ag/AgCl-1.5-1.0-0.50.00.51.01.53.5wt% NaCl, 0wt% Na2SO4 - O2 Saturated3.5wt% NaCl, 0wt% Na2SO4 - Aerated3.5wt% NaCl, 0wt% Na2SO4 - De-aerated Figure 5-2 Potentiodynamic polarization tests results for zinc samples immersed in 3.5wt% NaCl, and 0wt% Na2SO4 with different concentrations of oxygen i (A/cm2)1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2 1e-1 1e+0E(v) vs Ag/AgCl-1.5-1.0-0.50.00.51.01.53.5wt% NaCl, 1wt% Na2SO4 - O2 Saturated3.5wt% NaCl, 1wt% Na2SO4 - Aerated3.5wt% NaCl, 1wt% Na2SO4 - De-aerated Figure 5-3 Potentiodynamic polarization tests results for zinc samples immersed in 3.5wt% NaCl and 1wt% Na2SO4 with different concentrations of oxygen   45i (A/cm2)1e-7 1e-6 1e-5 1e-4 1e-3 1e-2 1e-1 1e+0E(v) vs Ag/AgCl-1.5-1.0-0.50.00.51.01.53.5wt% NaCl, 3wt% Na2SO4 - O2 Saturated3.5wt% NaCl, 3wt% Na2SO4 - Aerated3.5wt% NaCl, 3wt% Na2SO4 - De-aerated Figure 5-4 Potentiodynamic polarization tests results for zinc samples immersed in 3.5wt% NaCl and 3wt% Na2SO4 with different concentrations of oxygen i (A/cm2)1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2 1e-1 1e+0E(v) vs Ag/AgCl-1.5-1.0-0.50.00.51.01.53.5wt% NaCl, 4wt% Na2SO4 - O2 Saturated3.5wt% NaCl, 4wt% Na2SO4 - Aerated3.5wt% NaCl, 4wt% Na2SO4 - De-aerated Figure 5-5  Potentiodynamic polarization tests results for zinc samples immersed in 3.5wt% NaCl and 4wt% Na2SO4 with different concentrations of oxygen    46All the polarization curves show a limiting behavior on the cathodic branch. Yadav et al., when studying the cathodic polarization of zinc (Yadav et al. 2005), found two noticeable limiting plateaus: in the first limiting region, oxygen reduction took place on oxide-covered surface with 56% of oxygen being reduced through a 4-electron process, while in the second limiting region, the oxygen was exclusively reduced in a 4-electron reaction giving hydroxide on a quasi-uniformly active surface. They also found that when the zinc surface was corroded with thick zinc corrosion product layers, only the first limiting region was observed because the reduction of zinc corrosion products overlapped the second limiting region.  In previous work, it has been reported that zinc  dissolution  changes  from  activation  to  diffusion  control,  the  rate  determining  step  being  the  Zn2+ ion  diffusion  through  a  porous  corrosion  layer (Sziráki et al., 1998). However, the significant increase in anodic current density with potential increase suggests that zinc undergoes intense dissolution during anodic polarization and the formation of a porous corrosion layer has little effect on the dissolution of the materials. Baugh (1979) proposed that in the presence of a Cl- ion, the corrosion layer is non-passivating and therefore is likely porous. Culcasi et al. (2009) proposes that zinc dissolution is assisted by the chloride ions forming part of the electrolyte whose main role is believed to act as a catalyst in the corrosion process.  When trying to determine the corrosion parameters, it was noted that none of the curves showed Tafel behavior, so the Butler–Volmer equation was used instead to approximate the kinetic parameters using an optimization function.  To calculate the corrosion rate, a nonlinear least squares regression analysis was conducted between the recorded polarization plot using both the anodic and   47cathodic branches at ± 50mV from the recorded Ecorr value, and the Butler–Volmer equation to find the corrosion rate of each sample. The absolute squared error between the recorded results and theoretical results obtained from the Butler–Volmer equation was minimized to find the best fit. The modified variables during the regression analysis were io, E, and the cathodic and anodic slopes. Table 6-2 shows the calculated corrosion parameters.  Sample E(corr) (V) vs Ag/AgCl Βa V.dec-1 i(corr) (A/m2) Potential difference versus more noble potential V vs Ag/AgCl Corrosion rate (mm/y) A0 -0.846 0.08942 12.28768 - 18.40291B0 -0.901 0.09158 1.74869 -0.055 2.61896C0 -1.104 0.10685 0.20657 -0.258 0.30937A1 -0.858 0.01486 11.50931 - 17.23716B1 -1.014 0.01710 0.35320 -0.156 0.52897C1 -1.046 0.01714 0.00040 -0.188 0.00060A3 -0.870 0.01066 6.92635 - 10.37340B3 -1.022 0.01228 0.04099 -0.152 0.06139C3 -1.020 0.01494 0.00054 -0.150 0.00080A4 -0.830 0.06874 7.76924 - 11.63577B4 -0.955 0.06802 1.02419 -0.125 1.53391C4 -1.145 0.09139 0.00979 -0.315 0.01467Table 5-2- Corrosion parameters of the zinc sample after immersion   All the values for the cathodic slopes were close to 66.72 V.dec-1 and the values for the anodic slopes are shown in the table. An increase of Na2SO4 shows an increase in the corrosion rate relative to sulphate free solutions. When comparing a sample immersed in aerated conditions with 3.5 wt% NaCl and 1wt% of Na2SO4 to a sample in similar conditions but with 4wt% of Na2SO4 instead, both in aerated conditions with 3.5wt% of NaCl, for example, we see that by adding Na2SO4   48into the solution, icorr values shift from 0.3532 A/m2 to 1.024 A/m2. In previous studies, it is proposed that the presence of sulphates will lower the local pH on the surface of the metal (Qu et al., 2002), accelerating corrosion and further preventing the metal from reaching passivity. However, the effect of increasing the amount of oxygen in the solution seems to have a greater effect on the corrosion rate in comparison with that of increasing the Na2SO4 concentration. When comparing sample A3 with C3, we see that as we decrease the amount of dissolved oxygen in the solution, the icorr value changes from 6.9263 A/m2 to 0.0005 A/m2; a considerably higher change is observed. As the amount of oxygen decreases, both the corrosion current (icorr) and the corrosion potential (Ecorr) decrease, but at higher potentials, the zinc dissolution rates appear to be similar in spite of the oxygen concentration. This supports the theory that the corrosion layer formed is porous and lacks in its capacity to protect the material from further dissolution.   Previous work suggests that the initial linear slope of the anodic region indicates the formation of an oxide film on the surface of the zinc sample (Abd El Aal, 2000), which fails to protect the metal surface as indicated by the subsequent intense dissolution. All PDP tests indicate that zinc fails to passivate regardless of the concentration of oxygen under these conditions; increasing the amount of oxygen increases the corrosion. This is in agreement with work published by Boto & Williams, where they conclude that the zinc corrosion process is essentially controlled by oxygen availability on the surface of the material (Boto & Williams, 1977).  PDP curves indicate that the anodic process is controlled mainly by activation. Even though published work on the behavior of zinc under similar conditions is very limited, the PDP test results are in agreement with previous work published on galvanized steel (Culcasi  et al., 2009, Le Manchet et al., 2010b, Manov et al., 2000).    49 A previous study demonstrated that zinc and high purity zinc coatings (Zn0.1Al and Zn4.3Al) show similar electrochemical behavior when studying the oxygen reduction pathway (Dafydd et al., 2005). It was concluded that the zinc component of alloy coatings behaves electrochemically as though it were pure zinc. When building an MSE wall and facing following AASHTO LRFD design specifications, the galvanized steel needs to be in compliance with ASTM A641M for bar mat of grid-type reinforcements and the Standard Specification for Continuous Galvanizing Grade (CGG) Zinc Alloys for HotDip Galvanizing of Sheet Steel in ASTM B852 – 08. This standard dictates that high purity zinc is required and the aluminium content should be kept in the 0.22 to 1.1wt% range. This corroborates that performing tests on pure zinc is a viable option to study and develop understanding for the first corrosion stage on galvanized steel.   Finally, the potential differences between the oxygen purged samples and the rest are large enough to promote galvanic corrosion, and, consequently, the creation of corrosion macrocells. This means that under these conditions the reinforcement that is exposed to the air would corrode faster than the buried reinforcement.  The largest potential difference was found between sample C4 and sample A4 with a difference of ~313 mV. The combined effect of having a higher amount of sodium sulphate with the formation of macrocells due to different oxygen availability may lead to higher structural deterioration.   5.2 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY TESTING EIS is a useful technique used to study the effect of variable oxygen content on the corrosion layer formed on the zinc surface as an attempt to determine whether zinc forms a protective corrosion   50layer under these conditions or if the corrosion products formed on the metal surface fail to protect the metal from further dissolution.  Figure 5-6 show the EIS results for zinc in an oxygen-saturated solution, with the real impedance in x-axis (Z′) and imaginary impedance values (Z′′) on the y-axis and the corresponding equivalent circuit used to model the impedance results to shed light on the metal-oxide-environment interactions. As shown in Figure 5-6a, the gradual decrease of polarization resistance (Rp) after the second hour of immersion with time indicates that zinc does not form a passive layer. The first increase of polarization resistance after two hours of immersion is a sign of formation of a protective layer (Vagge et al., 2007, Sziráki et al., 1998), but the decrease of Rp in subsequent readings indicates either a breakdown of that layer or that the formed layer is porous and fails to fully protect the surface in the long term. Sulphur compounds lower the local pH on the surface of the metal (Culcasi et al., 2009), accelerate corrosion, and further prevent the metal from reaching passivity. The electrochemical impedance spectra was recorded at OCP and the resulting data fits almost perfectly with a simple equivalent circuit similar to the one reported by Cai & Park (1996). Cai proposed that ZnO can be found at an early stage of the dissolution process, but as the potential is increased further, a more porous Zn(OH)2 layer forms. This could also explain why zinc fails to fully passivate under these conditions.    51Figure 5-6 a) EIS results for zinc immersed in an oxygen-saturated solution with NaCl 3.5 wt% and Na2SO4after 1h, 2h, 3h, 4h, 6h, 7h, and 24 h of immersion b) R(QR) equivalent circuit.Figure 5-7 show the EIS results for zinc in de-aerated conditions and the corresponding equivalent circuits are shown in Figure 5-7c and Figure 5-7d, respectively. After the second hour of immersion, in contrast with the behavior observed in oxygen-saturated conditions, it is possible to observe a partially resolved semicircle at high frequencies with a Warburg response at lowfrequencies. A Warburg response is expected due the formation of thick porous corrosion products on the surface of the metal. In this case, the use of a Warburg element (W) is justified because under these conditions the diffusion at the metal/solution interface may determine the rate of reaction(Yadav et al., 2008, Sziraki et al., 2001). After 24 hours of immersion, the electrochemical behavior of zinc changes in comparison with earlier measurements. This change can be explained by the circuit displayed in Figure 5-7d.  52                 Figure 5-7 a) EIS results for zinc immersed in de-aerated solution with NaCl 3.5 wt % and Na2SO4 after 1h, 2h, 3h, 4h, 6h, 7h, and 24 h of immersion, b) High Frequency region zoom for Figure 6a, c) R(Q(RW)) Equivalent circuit for immersion times > 1 h >24 h, d) R(CR(CW)) Equivalent circuit after 24 hours of immersion  Given enough time, two different layers can be formed on top of the metal: a thick porous layer of Zn(OH)2 on top of the ZnO film and on top of the uncovered surface area. A compact form of zinc oxide has been reported to form underneath a porous oxide by direct oxidation, favored by a lower pH at the metal-electrolyte interface; the more compact form of zinc oxide serves as a diffusion QdRsRpQfWZ'0 500 1000 1500 2000 2500Z''05001000150020002500EIS After 1 hEIS after 2 h EIS after 3 hEIS After 4 hEIS After 6 hEIS After 7 hEIS After 24 h QRsRp Wac dzoomZ'0 100 200 300 400Z''0100200300400 b  53barrier to OH− ions (Thomas et al., 2013). Baril et al. (2007) and Sere et al. (1999) proposed the equivalent electrical circuit in Figure 5-7d, in which the faradaic impedance, Rp, is in parallel with the double-layer capacitance, Qd. In this case, the ZnO layer is characterized by a film capacitance, Qd, in parallel with film resistance Rp. The ZnO oxide film is assumed to be very protective and thus Rp is very high with respect to Qd (Sziraki et al., 2001). Zn(OH)2 is considered less protective and is characterized by film capacitance Qf, with a value of 1.142 x10-4, in parallel with a Warburg element.   Figure 5-8a shows the EIS results for a control sample of zinc immersed in naturally aerated conditions. Contrary to the sample immersed in oxygen-saturated conditions, it is possible to notice a gradual increase of polarization which could indicate a slow formation of a corrosion product on the metal surface. After one hour of immersion time, the equivalent circuit is best described by the circuit depicted in Figure 5-8b. But after the second hour, a diffusion behavior is observed at low frequencies, thus, the circuit in Figure 5-8c is used.   Similar behavior has been reported in the past for zinc immersed in an aerated solution containing NaCl. In the range of high and intermediate frequencies, Nyquist diagrams show a complex capacitive loop which can be related to the dynamic response of the active dissolution of zinc alloys. At low frequencies, the impedance behavior resembles the contribution of a diffusion process (Rosalbino et al., 2009). In aerated conditions, the rate of the reaction might be determined by the diffusion at the metal/solution interface (Yadav et al., 2008, Sziraki et al., 2001).     54               Figure 5-8 a) EIS results for zinc immersed in an aerated solution with NaCl 3.5wt% and 1wt% Na2SO4 after 1h, 2h, 3h, 4h, 6h, 7h, and 24 h of immersion, b) R(QR) Equivalent circuit after one hour of immersion, c) R(Q(RW) Equivalent circuit for immersion times > 1 h.   As the concentration of oxygen increases, the passive layer resistance formed after 24 hours of immersion seems to decrease from 3682 Ω cm2 in de-aerated conditions to 938 Ω cm2 in aerated conditions and to 575.9 Ω cm2 in oxygen-saturated conditions; this is in agreement with previously published results that show that zinc resists better against corrosion in de-aerated solutions (Hamlaoui et al., 2010).  b QRsRpQRsRp Wc Z'0 200 400 600 800 1000 1200Z''020040060080010001200EIS After 1 hEIS after 2 h EIS after 3 hEIS After 4 hEIS After 6 hEIS After 7 hEIS After 24 h a  55Table 5-3 shows a summary of the EIS parameters obtained in different concentrations of oxygen.  Conditions Exposed Timed Rs Ω Q1  S-sn Rp Ω W Χ2 De-aerated 1 hour 35.96 2.3 x10-4 67 0.07055 2.211 x10-4 2 hours 35.91 6.769 x10-5 288 0.00048 2.782 x10-4 3 hours 35.09 5.55 x10-5 324 0.04777 2.507 x10-4 4 hours 34.43 5.046 x10-5 420 0.00047 2.854 x10-4 6 hours 33.93 4.156 x10-5 567 0.05055 4.435 x10-4 7 hours 34.19 3.666 x10-5 650 0.05405 3.607 x10-4 24 hours 31.29 7.296 x10-6 3682 0.006191 5.74 x10-4 Oxygen Saturated 1 hour 45.18 9.81 x10-6 84 - 6.001 x10-4 2 hours 43.48 9.45 x10-6 791 - 5.64 x 10-4 3 hours 41.44 1.14 x10-5 787 - 5.42 x 10-4 4 hours 39.6 1.43 x 10-5 783 - 5.14 x10-4 6 hours 32.23 1.64 x 10-5 728 - 3.89 x10-4 7 hours 33.88 1.81 x10-5 722 - 3.58 x10-4 24 hours 325.4 3.26 x10-5 575.9 - 4.93 x10-4 Aerated 1 hour 46.13 2.77 x10-5 303.7 - 3.379 x10-4 2 hours 42.41 4.81 x10-5 620.7 0.003626 1.084 x10-3 3 hours 41.56 3.91 x10-5 761 0.003081 1.138 x10-3 4 hours 38.74 4.017 x10-5 763 0.01552 1.851 x10-3 6 hours 40.19 2.685 x10-5 913 0.2164 1.353 x10-3 7 hours 39.42 2.203 x10-5 610 0.8582 2.510 x10-3 24 hours 35.5 1.81 x10-5 938 0.00108 3.377 x10-3  Table 5-3- EIS fitting data obtained for zinc during 24 hours immersion tests in three different concentrations of oxygen.  In all cases, a constant phase element (CPE) is introduced in the circuit instead of a pure, double-layer capacitor to account for the deviation from an ideal capacitive behavior that could be attributed to surface non-homogeneity, surface roughness, porous layer formation, or impurities (Bommersbach et al., 2005). Values in the range of 0.5 ≤ n ≤ 1 are related to a non-ideal capacitive behavior and the appearance of a depressed semicircle in the Nyquist plot (Hassana et al., 2007). In   56this study, the values for the coefficient ‘n’ were all close to 0.8. In addition, values for the solution resistance (Rs) remained almost constant which supports the repeatability of the results. As mentioned above, only after 24 hours of immersion time in de-aerated conditions was it necessary to introduce a second constant phase element due the possible formation of two corrosion layers.  5.3 SURFACE CHARACTERIZATION SEM results show that zinc corroded in three very distinct ways, depending on the concentration of oxygen. As confirmed with EDX and XRD, the presence of sulphates did not appear to have a strong impact on the final corrosion products formed. This is in agreement with the PDP results which suggest oxygen is the main corrosion driver and not the presence of sulphates.        Figure 5-9- SEM picture of zinc immersed in de-aerated conditions with NaCl 3.5wt%, Na2SO4 3wt%  Figure 5-9 shows a representative surface morphology of a zinc sample in de-aerated conditions. Two distinct phases could be observed on its surface labeled as phase α and phase β, respectively. EDX analysis performed on phase α shows that the surface is composed of mainly zinc. On the other hand, phase β appears to be a corrosion layer with a complex composition (55wt% zinc, 33wt% oxygen, 1 wt% sulphur, 2 wt% chloride, and 9 wt% carbon). Cachet et al., when studying the electrochemical behavior of zinc coatings, found a similar black colored surface morphology and α β  57came to the conclusion that it was made of a non-hydrated, non-stoichiometric zinc oxide with wurtzite structure (Cachet et al., 2002).   Figure 5-10 shows a representative SEM picture of all samples corroded in aerated conditions. Two distinct phases could be observed on its surface as well, labeled as phase α and phase β, respectively. EDX confirms that phase α is composed primarily of zinc with traces of small oxygen (13 wt% Oxygen), while the amount of oxygen on phase β is considerably larger (36 wt% Oxygen). The corrosion layer covering phase β appears to be porous. In the presence of chlorides, the corrosion layer formed on the zinc is porous (Mouanga & Berçot, 2010)        Figure 5-10  SEM picture of zinc immersed in aerated conditions in NaCl 3.5wt%, Na2SO4 1wt%  Figure 5-11 shows the surface of a sample in oxygen-saturated conditions where an amorphous, porous layer of zinc oxide was formed. As the concentration of dissolved oxygen increased, the layer formed on the surface of the material appears to be thicker and becomes more uniform. Similar surface morphologies have been reported in the past when studying galvanized steel in oxygen-saturated solutions and described as the result of a skin pass process with hollowed areas surrounding the protruding ones (Olivier et al., 2010).  100 µmβα   58       Figure 5-11- SEM picture of zinc immersed in a solution saturated with oxygen with NaCl 3.5wt%, Na2SO4 1wt%   Micro-sized crystals, of sizes ranging from 5-20 µm, were found deposited on the surface of the corroded samples. Figure 5-12 shows an example of these crystals. These crystals appear to get deposited on the surface defects of the metal to later provide preferential corrosion sites where the formation and growth of the oxide layer begins. EDX analyses performed on these crystals suggest that they are some form of zinc oxide.          Figure 5-12 Micro crystal deposits (5-20 µm) found on the surface of corroded zinc providing preferential corrosion sites  100 µm100 µm  59 According to the zinc Pourbaix diagram (Beverskog & Puigdomenech, 1997), zinc should remain immune when the potential is below -1 V, then change to active corrosion state due to either the dissolution of zinc or the formation of soluble corrosion products, namely Zn(OH)2(aq). XRD analyses suggest the formation of three distinct corrosion products: zinc hydroxide (Zn(OH)2), zinc oxide (ZnO), and zinc chloride (ZnCl2). The corrosion mechanism of pure zinc under these conditions is dependant on the concentration of oxygen. According to Dafyyd et al. (2005), predominantly a 2e- reduction process occurs at potentials where zinc is Zn(OH)2 covered and a predominantly 4e- reduction process occurring at potentials where zinc is bare metal. However, even when the oxygen reduction pathway follows a 2e- process, the formed peroxy species could go through further reduction to form OH-. Having established that, it has been assumed that oxygen reduction could follow equations 5.1 to 5.3 when oxygen is available (Dafydd et al., 2005) and equation 5.4 when oxygen is not available.   O2 + 2H2O + 4e-  4OH-             (5.1) O2 + 2H2O + 2e-  HO2- + OH-    (5.2) HO2- + H2O + 2e-  3OH-            (5.3) H2O  1/2O2 + 2H+ + 2e-         (5.4)  On the other hand, the anodic reaction is more likely to follow the path summarized below (Sziraki et al., 2001, Cachet, 1992). The cations migrate towards the cathodic areas, while anions (Cl-, SO42- and OH-) migrate to the zinc dissolution sites.  NaCl  Na+ + Cl-     (5.5)   60Na2SO4  2Na+ + SO42-  (5.6) Zn  Zn2+ + 2e-         (5.7)  Formation of zinc hydroxide, or zinc oxide, could be explained by the reactions described in equations 8 and 9. Moreover, ZnO could be formed directly from Zn(OH)2 when the pH range is between 6 and 9, explaining the circuit described earlier in Figure 5-7d.            Zn2+ + 2OH   Zn(OH)2                       (5.8) Zn2+ + O2 + 2H+ + 2e-   ZnO  + H2O  (5.9)  Figure 5-13 shows the diffraction pattern for a control sample and three different samples with different concentrations of oxygen. According to the database in the the MDI Jade 7 software, ZnO peaks were detected at 2Ө = 31° and 36°. Peaks registered between approximately 2Ө = 20° and 47°, in Figure 5-13b and Figure 5-13c, indicate the presence of Zn(OH)2. It has been reported in the past that equation 5.4 leads to a local increase in pH values and, as a result, the accumulation of chloride ions in the pits and their neighbouring areas (Feitknecht, 1959). This justifies the presence of ZnCl2 which was detected in solutions with low concentrations of Na2SO4 as shown in Figure 5-13d with peaks at 2Ө = 26° and 29°.   61Figure 5-13 Representative XRD diagrams for pure zinc immersed in a) de-aerated solution with NaCl 3.5wt%, b) oxygen-saturated solution with NaCl 3.5wt% - Na2SO4 1wt%, c) aerated solution with NaCl 3.5wt% -Na2SO4 1wt%, and d) de-aerated solution with NaCl 3.5wt% - Na2SO4 1wt%. 5.4 SUMMARYThe purpose of this chapter was to study the effect of variable oxygen concentration on the corrosion performance of zinc. Zinc was immersed in solutions with varied concentration of Na2SO4 (1-4 wt %), a fixed concentration of NaCl (3.5 wt %), and different concentrations of oxygen to simulate the conditions found in soils and ground water. Differential oxygen access can promote corrosion macrocells that will compromise the performance of the protected structure. The electrochemical corrosion behavior of zinc was studied using potentiodynamic polarization testing and Electrochemical Impedance Spectroscopy (EIS). Scanning Electron Microscope (SEM) and X-Ray Diffraction (XRD) were used to characterize the corroded samples. The results revealed that increasing the concentration of oxygen in the solution appears to have a greater effect on icorr  62compared to the effect of adding more Na2SO4 to the solution. Potentiodynamic polarization data provides enough information to confirm that the potential difference between samples immersed in oxygen-saturated solutions, compared with samples immersed in de-aerated and aerated solutions, are large enough to confirm the creation of corrosion macrocells. SEM showed three distinct surface morphologies depending on the concentration of oxygen; as oxygen concentration increases, the corrosion layer becomes thicker and more porous.     636 CORROSION PERFORMANCE OF GALVANIZED STEEL IN NA2SO4 AND NACL SOLUTIONS AT SUBFREEZING TEMPERATURES  The US Transportation Research Board reported after several field measurements that no significant differences on the corrosion rate of galvanized reinforcements were observed in different climatic regions. Nevertheless, it is imperative to further develop an understanding of these structures in colder climates where, despite previous reports, corrosion is still a problem. The objective of this work is to study the effect of subfreezing temperatures on the corrosion performance of galvanized steel and, more specifically, to determine whether or not it is safe to assume that, under freezing temperatures, the corrosion rate decreases enough to not be a concern anymore. This chapter is based on a paper published as part of the work towards this PhD thesis.2  6.1 EFFECT OF TEMPERATURE ON THE POTENTIODYNAMIC TESTING The galvanized steel samples reached a steady open circuit state within the first 100 s of the test, which is in agreement with previous results (Padilla & Alfantazi, 2010). Additionally, Sziráki et al. reported that zinc forms an oxide/hydroxide spontaneously in the presence of sulphates in near-neutral pH solutions (pH=5) (Sziráki et al., 1998), therefore achieving a stable condition quickly. At this stage of the experimental procedure, the galvanized layer is still fully intact, consequently, the electrochemical behavior is governed by the properties of zinc. As the temperature drops, OCP values shift to less noble potentials, from -0.855 V Ag/AgCl at 25ºC to -1.009 V Ag/AgCl at -5ºC. This could be explained by the fact that a decrease in temperature causes oxygen diffusivity on the metal surface to also decrease, thus lowering OCP values.                                                  2 Victor Padilla, Akram Alfantazi,  Corrosion, 69 (2013)   64Figure 6-1 illustrates the OCP results when decreasing temperature every 5 hours over a 25 hour period on a galvanized steel sample immersed in 3.5wt% NaCl, and 1wt% Na2SO4. During the 25 hour immersion tests, OCP values dropped from an average of -0.953 V Ag/AgCl at 25ºC to approximately -0.974 V Ag/AgCl at -5ºC. This smaller drop is expected since the OCP increases with immersion time as the surface becomes more active (Tsai et al., 2010).   Figure 6-1 Effect of decreasing temperature in a 5 hour interval over a 25 hour period on the OCP reading of a galvanized steel sample immersed in 3.5wt% NaCl, and 1wt% Na2SO4.   Furthermore, Figure 6-2 shows potentiodynamic polarization (PDP) curves of samples immersed in 3.5 wt% NaCl and 1wt% Na2SO4 at different temperatures. In agreement with the OCP measurements, results of the potentiodynamic polarization tests suggest that, as temperature decreases, the corrosion rate also decreases.   -0.98-0.975-0.97-0.965-0.96-0.955-0.95-0.945-0.94-0.935-0.930 2 4 6 8 10 12 14 16 18 20 22 24t / hE vs. (Ag/AgCl) / V-10-5051015202530Temperature º CVoltage at OCP vs Ag/AgC Temperature º C  65i (A/cm2)10-8 10-7 10-6 10-5 10-4 10-3 10-2E vs. (Ag/AgCl) /V-1.2-1.0-0.8-0.6-0.4-0.20.0 3.5 wt% NaCl, 1 wt% Na2SO4 at 25 ºC3.5 wt% NaCl, 1 wt% Na2SO4 at 15 ºC3.5 wt% NaCl, 1 wt% Na2SO4 at 5 ºC3.5 wt% NaCl, 1 wt% Na2SO4 at 0 ºC3.5 wt% NaCl, 1 wt% Na2SO4 at -5 ºC Figure 6-2 Potentiodynamic polarization curves for the galvanized steel samples immersed in 3.5wt% NaCl, and 1wt% Na2SO4 temperatures ranging from 25ºC to -5ºC  The plotted polarization curves follow the typical polarization behavior of zinc immersed in corrosive alkaline solutions (Yadav et al., 2007, Hamlaoui et al., 2010, Culcasi et al., 2009). Mokaddem et al., when studying the corrosion behavior of zinc in 0.1M to 1M NaOH solutions, reported that under those conditions, galvanized zinc showed a passive-like behavior (Mokaddem et al., 2010). The difference could be explained in terms of the presence of aggressive anions, such as Cl- and SO4-2, which can disrupt the corrosion layer formed on galvanized steel. For comparison purposes, Figure 2-2 shows three polarization curves obtained from pure zinc, galvanized steel, and   66a steel sample. In order to test the electrochemical behavior on uncoated steel, the samples were polished until the galvanized layer was completely removed. SEM and EDX were used to confirm that the zinc had been totally polished off from the surface. The anodic regions of the curves show that, as the potential is increased, the zinc coating starts to dissolve and the anodic current density tends to become closer to that of the steel substrate moving away from that of pure zinc (Vagge et al., 2007). The Ecorr values for galvanized steel and pure zinc are nearly the same: -0.988 V Ag/AgCl for pure zinc and -0.984 V Ag/AgCl for galvanized steel. On the other hand, the Ecorr value for the steel sample is -0.499 V Ag/AgCl. Moreover, the inflection point at which the galvanized steel sample begins to show reversal behavior is at about -630 mV Ag/AgCl.  In fact, it is possible to appreciate this reversal behavior in all results from Figure 6-2. Regardless of the temperature, a reversal behavior at approximately -640 mV Ag/AgC is observed which can be attributed to the transition between zinc dissolution and the dissolution of iron in steel.  The polarization curves in Figure 6-2 reveal that the anodic process is controlled by activation and the cathodic process by diffusion. As reported by Short et al. (1989) and Alvarez et al. (1976), the relatively small slope observed in the anodic region, coupled with the small shift from the equilibrium potential and the strong dependency on temperature suggest that the anodic process is controlled by activation. This, coupled with the lack of one decade of linearity at the cathodic branch, implies that the corrosion rate cannot be determined by using the Tafel extrapolation method (Culcasi et al., 2009), hence, the Butler–Volmer equation was used instead to approximate the kinetic parameters using an optimization function. The calculated corrosion parameters are presented in Table 6-1.  As expected, by decreasing temperature, a decrease is observed both in the corrosion rate and in the magnitude of the anodic Tafel slope, except for the sample immersed at -5ºC. The sudden increase in the magnitude for the anodic slope can be explained by direct dependence of the   67Tafel slope upon temperature and the activation energy. When temperature reaches 0ºC the solution is still in a purely liquid state but, at -5ºC, the solution starts freezing until it becomes solid. At freezing temperatures, a greater energy barrier needs to be overcome for ion migration due to a substantial drop in the total accumulation of ions and in the ionic permeability of ice and frozen soils (Chuvilin et al., 1998, McMillan et al., 1982b).  In the cathodic region, the limiting current density (iL) is the most important characteristic regarding diffusion-controlled processes (Ajeel & Ali, 2008). The cathodic polarization curves at different temperatures showed a similar behavior with a limiting current density ranging from approximately 4.5 x 10-5 A/cm2 to 1.0 x 10-6 A/cm2 with decreasing temperature. Moreover, the significant increase in anodic current density with a potential increase suggest that zinc undergoes intense dissolution during anodic polarization and the formation of a porous corrosion layer has little effect on the dissolution of the materials. Culcasi et al. proposes that zinc dissolution is assisted by the chloride because chloride acts as a catalyst in the corrosion process (Culcasi et al., 2009). The corrosion current values show a steady decrease with temperature with the value dropping from 2.97 x10-5 A/cm2 (0.444 mm/year) at 25ºC to 1.08 x10-6 A/cm2 (0.016 mm/year) at -5ºC. The corrosion rates reported at room temperature are in agreement with previously reported values in both laboratory conditions and atmospheric tests (Le Manchet et al., 2010a, Culcasi et al., 2009, Manov et al., 2000, Chen et al., 2006, Berke & Sagüés, 2009). According to AASHTO, the accepted corrosion rates are 15 μm/year for the first two years, followed by 4 μm/year from year 2 until zinc depletion, and 12 μm/year after zinc depletion until the end of a 75-year service life (Berke & Sagüés, 2009). However, the lowest corrosion rate calculated in this study is 16 μm/year at -5ºC when the solution is completely frozen and solidified. This indicates that corrosion rates in aggressive environments are still too high even at temperatures below the freezing point when   68compared with those accepted by the AASHTO corrosion rate guidelines for galvanized steel, where it is stated that for the first 2 years after installation, the corrosion rate should not exceed 15 μm/yr (Fishman & Withiam, 2011). Even though published work on the behavior of zinc under similar conditions is very limited, the corrosion rates presented are in agreement with previous work published on galvanized steel (Le Manchet et al., 2010a, Culcasi et al., 2009). Furthermore, the corrosion rates reported by linear polarization resistance (LPR) are also in agreement. Fishman & Withiam reported mean corrosion rates ranging from 26 µm/year to 99 µm/year for structures in high humidity conditions (Fishman & Withiam, 2011). As an initial attempt in predicting the effect of temperature on the corrosion rate of galvanized steel, a simple model was developed. Temperature affects exchange current density as suggested by Tanaka and Tamamushi (1964), and later elaborated by Pour-Ghaz et al. (2009) in equation 6.1:   211  12,1, TTRzFEeeii   (6. 1) where i 1,   is the exchange current density (A/cm2) at temperature T1 (Kelvin), i 2,  is the exchange current density (A/cm2) at temperature T2 and Ee (V) is the equilibrium potential. When solving equation 6.1 for one of the current densities, it is possible to obtain a theoretical value for i 2,  taking into account a known current density at T1 and the studied temperature change at T2, as shown in equation 6.2:   211  11,2,TTRzFEeeii     (6. 2)   69In a study carried out by Dirkse (Dirkse, 1979) on the behavior of zinc in alkaline solutions, it is found that Zn(OH)2 is  the  electroactive  species. Therefore, the rate-determining step occurs early in the anodic  process before the dissolution of  Zn(OH)2 that leads to the formation of zincate ions. Moreover, since pH is constant, the equilibrium potentials of the anodic half-cell reactions can only be affected by temperature and, if assuming a two-electron transfer process (n = 2), the corrosion rate might be expressed by the low field approximation of the Butler-Volmer Equation:   RTEEnFii ocorrocorr )(   (6.3) where oE is the equilibrium potential. The exchange current density is inversely affected by a change of temperature.  When combining equation 6.2 and equation 6.3, it is possible to obtain an expression to directly calculate icorr when temperature changes from T1 to T2 based on an experimentally-obtained corrosion current value at T1.  211  11,22,11,2,)()(TTRnFEocorrocorrcorrcorr eeEETEETii   (6.4) where i 1,   is the exchange current density (A/cm2) at temperature T1, i 2,  is the exchange current density (A/cm2) at temperature T2, Ecorr1, and Ecorr2 are the corrosion potentials at T1 and T2 respectively, Ee is the equilibrium potential, z is the valance electron count, α is the symmetry factor which is assumed to remain constant (Hurlen & Eriksrud, 1973), and R, n, and F have their conventional meaning. Table 6-1 also presents the experimentally obtained corrosion rates and the calculated values using equation 6.4. Even though this is a preliminary model, the calculated values are in agreement with the experimentally-obtained values.    70Sample E(corr) vs. Ag/AgCl / V βa  V.dec-1βc  V.dec-1i(corr)   A/cm2 Corrosion rate  mmy-1 Calculated Corrosion rate  mmy-1 25º C -0.855 0.0207 0.078 2.97 x10-5 0.444 N/A 15º C -0.891 0.0109 0.079 1.65 x10-5 0.247 0.083 5º C -0.897 0.0109 0.015 1.88 x10-6 0.028 0.033 0º C -0.906 0.0088 0.848 1.93 x10-6 0.029 0.028 -5º C -1.009 0.0131 0.057 1.08 x10-6 0.016 0.017 Table 6-1 Corrosion parameters for the galvanized steel samples after immersion  6.2 EIS TESTING EIS measurements were carried out at different temperatures in the solution of 3.5 wt% NaCl, plus 1 wt% Na2SO4 in order to assess the effect of temperature on the corrosion performance of galvanized steel. Figure 6-3 shows the Nyquist diagrams recorded for samples immersed in aerated solutions with 3.5 wt% NaCl and 1 wt% Na2SO4 at the different tested temperatures, after 1 h, 2 h, 4 h, 6 h, and 24 h of immersion.  A decrease in total impedance and polarization resistance during immersion, coupled with an increase of the coating capacitance, indicates the degradation of the zinc coating on the steel substrate.  Similar behavior was observed at all tested temperatures. This is contrary to previously published results, in which an increase in polarization resistance was reported with time. That increase was attributed to a possible build-up of zinc corrosion products controlled by a charge transfer reaction (Hamlaoui et al., 2010). The difference in behavior can be attributed to the presence of sulphates, since sulphates lower the local pH on the surface of the metal (Qu et al., 2002), accelerating corrosion and preventing the metal from reaching passivity in the long term. Results indicate that the degradation of the zinc coating and the corrosion product with immersion time causes resistance values to decrease.    71                                                      Figure 6-3 Nyquist Plots for galvanized steel immersed during 24 h in aerated solutions with 3.5 wt % NaCl and 1wt % Na2SO4 at a) 25ºC, b) 15ºC, c) 5ºC, d) 0ºC, and e) -5ºC Z' cm2 0 5000 10000 15000 20000 25000Z''  cm2 0500010000150002000025000NaCl 3.5wt% Na2SO4 1wt% After 1 hNaCl 3.5wt% Na2SO4 1wt% After 2 hNaCl 3.5wt% Na2SO4 1wt% After 4 hNaCl 3.5wt% Na2SO4 1wt% After 6 hNaCl 3.5wt% Na2SO4 1wt% After 24 h12.5 Hz1 Hzd) 0篊Z' /cm2 0 2000 4000 6000Z'' /cm2  0200040006000NaCl 3.5wt% Na2SO4 1wt% After 1 hNaCl 3.5wt% Na2SO4 1wt% After 2 hNaCl 3.5wt% Na2SO4 1wt% After 4 hNaCl 3.5wt% Na2SO4 1wt% After 6 hNaCl 3.5wt% Na2SO4 1wt% After 24 h79.6 Hz79.6 Hzc) 5篊b 15癈Z'0 2000 4000 6000 8000 10000 12000Z''020004000600080001000012000NaCl 3.5wt% Na2SO4 1wt% After 1 hNaCl 3.5wt% Na2SO4 1wt% After 2 hNaCl 3.5wt% Na2SO4 1wt% After 4 hNaCl 3.5wt% Na2SO4 1wt% After 6 hNaCl 3.5wt% Na2SO4 1wt% After 24 h39.8 Hz79.4 Hza) 25 癈Z'0 1000 2000 3000Z''0100020003000NaCl 3.5wt% Na2SO4 1wt% After 1 hNaCl 3.5wt% Na2SO4 1wt% After 2 hNaCl 3.5wt% Na2SO4 1wt% After 4 hNaCl 3.5wt% Na2SO4 1wt% After 6 hNaCl 3.5wt% Na2SO4 1wt% After 24 h100 Hz125 HZe) -5 癈Z' cm2 0 2000 4000 6000Z'' cm2 0200040006000NaCl 3.5wt% Na2SO4 1wt% After 1 hNaCl 3.5wt% Na2SO4 1wt% After 2 hNaCl 3.5wt% Na2SO4 1wt% After 4 hNaCl 3.5wt% Na2SO4 1wt% After 6 hNaCl 3.5wt% Na2SO4 1wt% After 24 h42.3 Hz10.3 Hz  72Most Nyquist plots showed one high frequency capacitive loop (HFC) and, in some cases, the beginning of one low frequency inductive loop (LFI). The deformation of the left part of the HFC loop is often attributed to the existence of pores on the surface (Cachet, 1992). When looking at the values registered at -5ºC, it is easier to notice the formation of an LFI at low frequencies.                              Figure 6-4 a) Nyquist Plots for galvanized steel immersed in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 after 24 h of immersion at 25ºC, 15ºC, 5ºC, 0ºC and -5ºC. b) Insert of the high frequency region of Figure 5a. Z' cm2 0 200 400 600 800 1000Z''  cm2 02004006008001000a)Z' cm2 0 2000 4000 6000 8000 10000 12000Z''  cm2 020004000600080001000012000 NaCl 3.5wt% Na2SO4 1wt% at 25 篊 After 24 hNaCl 3.5wt% Na2SO4 1wt% at 15 篊 After 24 hNaCl 3.5wt% Na2SO4 1wt% at 5 篊 After 24 hNaCl 3.5wt% Na2SO4 1wt% at 0 篊 After 24 hNaCl 3.5wt% Na2SO4 1wt% at -5 篊 After 24 h  73 Figure 6-4 shows the Nyquist plot for galvanized steel immersed in an aerated solution after 24 hours of immersion at the different tested temperatures and a magnification into the respective high frequency region.  Baugh (1979) proposed that, in the presence of a Cl-, the corrosion layer is non-passivating and therefore, probably porous. The equivalent electrical circuit used to obtain the EIS fitting data has RS(Qc(Rf(QdlRct))). This is commonly used to model porous corrosion product layers, where RS accounts for the solution resistance, Qc and Rf for the film capacitance and resistance of the porous layer, respectively, Qdl and Rct account for the capacitance and transfer resistance/diffusion through the pores, respectively (Xing et al., 2010).           Figure 6-5 Schematic representation of the equivalent electrical circuit for an electrode protected by a porous RS(Qc(Rp(QctRdl))).  Figure 6-5 shows the schematic representation of the proposed circuit. During the first stage of corrosion, in the case of near neutral pH conditions, a semi-compact layer of precipitates, mainly Zn and ZnO with small amounts of Zn(OH)2, is present on the surface (Bonk et al., 2004). However, as immersion time increases, the layer grows, depleting the zinc coating and forming a thick, porous layer consisting mainly of a non-protective and soluble layer of Zn(OH)2. The decrease of RP values RsRp QcQdlRctPorous layer Partially dissolved Zinc coatingSteel  74can be explained by the dissolution of the porous layer into the solution. Rct behaved fairly consistently throughout the test. This suggests, as previously reported (Padilla & Alfantazi, 2010), that a thick, porous layer of Zn(OH)2 is formed on top of a thinner, more compact film of ZnO at the metal surface.   Table 6-2 presents the EIS fitting data calculated for the galvanized steel samples during 24 h immersion in 3.5 wt% NaCl and 1wt% Na2SO4 at temperatures ranging from 25ºC to -5ºC. A criterion for a coating system to take part in both charge transfer and the diffusion control process is that the ratio of Rct/Rf ranges between 0.2 and 5 (Rout, 2007).  Additionally, values for the coefficient n were all around 0.8. It is reported that a non-ideal capacitive behavior and the appearance of a depressed semi-circle in the Nyquist plot occurs when n is located between 0.5 and 1 (Hassana et al., 2007). RS increases from an average of 0.36 x102 Ω cm2 observed at higher temperatures to values larger than 1.10 x102 Ω cm2 at lower temperatures, attributable to the increase of water resistance at lower temperatures.  At 25ºC Rf decreased from 3.39 x103 Ω cm2 to 3.78 x103 Ω cm2 and Rct from 1.22x103 Ω cm2 to 0.434 x103 Ω cm2. When immersed at 15ºC, a decrease in total impedance is also noticed. However, Rf decreased from 3.25 x103 Ω cm2 to 2.10 x103 Ω cm2, indicating that, as temperature decreases, damage on the porous layer also decreases. Rct, after 1 hour of immersion, was relatively high compared to the rest of the measurements at this temperature: after 1 hour of immersion, Rct was 7.42 x103 Ω cm2 whereas throughout the rest of the experiment, it ranged from 0.11 x103 Ω cm2 to 0.29 x103 Ω cm2. For the sample immersed at 5ºC, Rf decreased from 4.17 x103 Ω cm2 to 0.80 x103 Ω cm2 and Rct from 2.86 x103 Ω cm2 to 0.32 x103 Ω cm2.     75T (ºC) time (h) Rs   Ω·cm2 Qc  S-sn n Rp  Ω·cm2 Qdl   S-sn n Rct  Ω·cm2 25 C 1 0.29 x102 2.95x10-6 0.73 3.39 x103 2.96 x10-4 0.80 1.22 x1032    0.27 x102  3.62 x10-6 0.71 1.40 x103 8.24 x10-8 0.86 0.63 x1034 0.29 x102 3.46 x10-6 0.72 0.96 x103 2.14 x10-5 0.78 0.215 x1036 0.18 x102 4.07 x10-6 0.71 0.87 x103 2.45 x10-5 0.93 0.16 x10324 0.37 x102 3.17 x10-4 0.77 0.38 x103 9.34 x10-5 0.60 0.43 x10315 C 1 0.51 x102 6.87 x10-7 0.83 3.25 x103 4.31 x10-7 0.80 7.42 x1032    0.49 x102  1.43 x10-6 0.77 3.63 x103 8.85 x10-6 0.82 0.18 x1034 0.23 x102 1.85 x10-6 0.76 3.88 x103 1.62 x10-6 0.84 0.28 x1036 0.27 x102 1.91 x10-6 0.76 3.29 x103 3.63 x10-5 0.95 0.11 x10324 0.17 x102 2.18 x10-6 0.76 2.09 x103 5.37 x10-5 0.96 0.29 x1035 C 1 0.69 x102 8.87 x10-7 0.80 4.17 x103 1.03 x10-6 0.92 0.28 x1032 0.15 x102 1.17 x10-6 0.79 2.57 x103 2.86 x10-6 0.83 0.72 x1034 0.45 x102 1.47 x10-6 0.80 1.92 x103 7.17 x10-6 0.90 0.38 x1036 0.49 x102 2.04 x10-6 0.78 2.26 x103 9.79 x10-5 0.86 0.35 x10324 0.64 x102 6.30 x10-6 0.80 0.80 x103 1.92 x10-5 0.80 0.32 x1030 C 1 1.12 x102 8.02 x10-7 0.83 18.36 x103 1.97 x10-6 0.88 6.06 x1032 1.17 x102 8.06 x10-7 0.83 14.11 x103 9.58 x10-1 0.96 5.12 x1034 1.12 x102 9.66 x10-7 0.82 10.11 x103 3.00 x10-6 0.78 6.43 x1036 1.12 x102 2.04 x10-6 0.80 1.54 x103 9.79 x10-5 0.86 0.35 x10324 1.11 x102 1.13 x10-6 0.80 6.40 x103 1.12 x10-5 0.80 7.28 x103-5 C 1 0.69 x102 8.87 x10-7 0.82 4.17 x103 1.03 x10-6 0.80 2.86 x1032 0.42 x102 1.02 x10-6 0.82 2.25 x103 8.97 x10-7 1.0 1.15 x1034 0.21 x102 3.79 x10-7 0.90 2.26 x103 3.38 x10-5 0.52 1.63 x1036 1.01 x102 5.61 x10-7 1.0 0.22 x103 2.39 x10-5 0.67 1.24 x10324 1.00 x102 5.14 x10-7 1.0 0.122 x103 1.41 x10-4 0.59 0.86 x103 Table 6-2 - EIS fitting data obtained for galvanized steel during 24 h immersion in 3.5wt% NaCl and 1wt% Na2SO4 at temperatures ranging from 25ºC to -5ºC  As expected, when reaching near freezing temperatures, Rs, Rf, and Rct dramatically increased. At 0ºC, the total resistance of the system increased, compared to the previously recorded resistance at higher temperatures. After 1 hour of immersion, Rf was 18.36 x103 Ω cm2 and even after 24 hours of immersion, a Rf value was 6.40 x103 Ω cm2 which is relatively high compared to   76the resistance values observed above 0ºC. Rct, at 0ºC, was also higher than those observed above 0ºC and constant during the period of the test. After 1 hour of immersion, Rct was 6.06 x103 Ω cm2 and after 24 hours, value was 7.30 x103 Ω cm2.   6.3 SURFACE CHARACTERIZATION Figure 6-6 depicts an SEM picture of a galvanized steel sample immersed at 25ºC. In agreement with EIS results, the corrosion products formed on top of the metal surface appear to be thick and porous forming, which is commonly referred as sulphate nests. SEM results show that the corrosion products found on the galvanized steel samples after 24 hours of immersion in 3.5 wt% NaCl and 1 wt% Na2SO4 are similar to those found on pure zinc.  Weissenrieder et al. (2004) proposed that sulphate nests are blisters, formed spatially heterogeneously on the surface, containing high concentrations of corrosion products and electrolyte (Weissenrieder et al., 2004a). These products are supposed to be retained by a semi-permeable membrane of colloidal oxyhydroxides. With respect to the current model for sulphate nests, water starts to diffuse into the nests under the influence of the high ionic strength of the solution. As a result, the semi-permeable membranes swell and eventually burst, spreading the electrolyte to adjacent areas to stimulate the corrosion process (Weissenrieder et al., 2004a). These sulphate nest-like formations were found also at 15ºC. Figure 6-6b shows smaller sulphate nests found on the surface of a sample exposed to 15ºC and it is also possible to see pseudo-hexagonal plane crystals which have been previously reported when studying corrosion of galvanized steel (Tanaka et al., 2011). These crystals, also referred to in the literature as platelets, are believed to have a preferential facing normal to the substrate plane. These crystals nucleate on surface defects and then grow, 77forming islands which spread over the entire surface as the exposure time increases (Asgari et al.,2009, Seré et al., 1999). Figure 6-6 a) Sulphate nest formation found on the surface of corroded galvanized steel after immersion in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 at 25ºC. b) Smaller sulphate nest formation found on the surface of corroded galvanized steel after immersion in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 at 15ºC.Figure 6-7 shows blisters found on a sample immersed at 0ºC, suggesting an early stage of formation of these nests was found at lower temperatures. SEM exploration also revealed hexagonal corrosion products. Figure 6-7 Blisters found on top of corroded galvanized steel after immersion in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 at 0ºC, suggesting early stages for the formation of sulphate nests.78Figure 6-8 shows two examples of clearly delimited hexagonal and quasi-hexagonal shaped corrosion products indicating that corrosion occurs at zinc grain boundaries on the surface (Muñozet al., 2002) and the presence of ZnO which has been reported to be hexagonal in shape (C. 2005).Figure 6-8 SEM pictures showing corrosion products found on the surface of corroded galvanized steel after immersion in an aerated solution with 3.5 wt % NaCl and 1wt% Na2SO4.As shown in Figure 6-9 at -5ºC, a thin film was found covering the whole surface of the sample. While it was impossible to determine the composition of the film because it was too thin, the shape of the film suggests that corrosion and cracking of the film initiated at the grain boundaries. Lu et al. proposed that the film grows more quickly in the vicinity of zinc grain boundaries and it is apt to crack, resulting in eventual warping and flaking off of the film with increasing film thickness (Lu et al., 2006).Figure 6-9 - Thin film found on the surface of corroded galvanized steel after immersion in aerated solution with 3.5 wt % NaCl and 1wt % Na2SO4 at -5ºC.a b  79As confirmed with PDP and EIS test results, the corrosion rate reduces with decreasing temperature. This was confirmed with the qualitative characterization of a series of SEM pictures taken from the surface of galvanized steel (not presented here due to the space limitation). SEM revealed corrosion products and sulphate nests in all of the samples, though size and thickness seemed to decrease as temperature decreased. This suggests that galvanized steel corrodes even at sub-freezing temperatures: a drop in temperature lowers the corrosion rate, but does not stop corrosion altogether. This might occur because water in adsorption layers suffers a lowering of the freezing point and remains liquid at temperatures considerably less than 0ºC (Graedel, 1989), thus allowing for the corrosion process to continue.   In the Zn and Fe Pourbaix diagrams (Beverskog & Puigdomenech, 1997), zinc remains immune when the potential is below -1.19 V Ag/AgCl, then changes to active corrosion state due to either the dissolution of zinc or the formation of soluble corrosion products, namely Zn(OH)2(aq). Once corrosion is initiated, zinc will preferentially corrode until fully depleted, soon followed by the corrosion of steel which becomes active at potentials slightly above -0.5 V Ag/AgCl. Under these conditions, the formation of either FeO(OH) or Fe3O4 could be expected (Pineau et al., 2008). Previous XRD analyses performed on pure zinc samples immersed in similar electrolytes confirmed the formation of Zn(OH)2 as the main corrosion product (Padilla & Alfantazi, 2010). However, XRD analyses performed on the galvanized steel samples after polarization testing showed that, under the tested conditions, the corroded surface was free of zinc corrosion products, confirming that they are soluble and non-protective in the long term.  As evidenced by the XRD pattern shown in Figure 6-10, the corrosion products found on all the samples were consistent. Figure 6-10 shows XRD patterns for galvanized steel immersed in NaCl   803.5 wt% and Na2SO4 1 wt% at a) 25ºC, b) 15ºC, c) 5ºC, d) 0ºC, and e) -5ºC. In addition, for comparison purposes, Figure 6-10 f) depicts results of an uncoated steel sample immersed in NaCl 3.5 wt% plus Na2SO4 1 wt% at 25ºC. Using the database of the MDI Jade 7 software, all the peaks in the XRD spectra were identified. Pure Fe peaks were detected at approximately 45° and 65°; which were attributed to the steel substrate. The main corrosion product detected was iron oxide (Fe3O4) with several peaks in the XRD spectra. It is important to bring up that, due to the nature of the samples, it is possible that excessive heat or drying of the samples after immersion could have induced the transformation of FeO(OH) into Fe3O4 (Deliyanni et al., 2001). A small peak attributed to pure Zn was detected around 44° in most of the samples. It is believed that the presence of Zn in the XRD spectra comes from the unimmersed area of the substrate. Previous work performed on atmospheric corrosion of galvanized steel suggest the formation of protective zinc corrosion products (Mokaddem et al., 2010, Graedel, 1989, Cole et al., 2010); however, the XRD analysis performed in this study confirmed that no zinc corrosion products were detected on the surface of the corroded metal under the studied conditions.     81            Figure 6-10 - XRD patterns for galvanized steel immersed in NaCl 3.5wt% and Na2SO4 1wt% at a) 25ºC, b) 15ºC, c) 5ºC, d) 0ºC, e) -5ºC, and f) uncoated steel sample immersed in NaCl 3.5wt% and Na2SO4 1wt% at 25ºC. 6.4 SUMMARY This chapter focused on the effect of temperature variation on the corrosion performance of galvanized steel. Samples were immersed in solutions containing Na2SO4 and NaCl to simulate the aggressive conditions found in soils and ground water. Corrosion rates were measured using potentiodynamic polarization tests at temperatures ranging from -5°C to 25°C. Electrochemical Impedance Spectroscopy (EIS) was used to study the performance of the zinc cover under the different conditions. Results showed that the corrosion rate observed at sub-zero temperatures is still higher than the rate acceptable for galvanized steel-reinforced structures. At -5°C, the calculated rate is 16 μm/yr, while the rate acceptable in the AASHTO model for galvanized steel reinforcements less than 2 years old is 15 μm/yr.  a2degrees 30 40 50 60Galvanized Steel in 3.5 wt% NaCl, 1 wt% Na2SO4 Galvanized Steel in 3.5 wt% NaCl, 1 wt% Na2SO4 Galvanized Steel in 3.5 wt% NaCl, 1 wt% Na2SO4 Galvanized Steel in 3.5 wt% NaCl, 1 wt% Na2SO4 Galvanized Steel in 3.5 wt% NaCl, 1 wt% Na2SO4 Steel in 3.5 wt% NaCl, 1 wt% Na2SO4bcdefFe3O4ZnFe  827 EFFECT OF DE-ICING SALTS ON THE CORROSION PERFORMANCE OF GALVANIZED STEEL IN SULPHATE-CONTAMINATED SOIL  Most of the studies which compare the effect of different de-icing salts that have been performed so far focused only on how each salt affected the corrosion performance of either mild steel or galvanized steel (Vitaliano, 1992, Macias & Andrade, 1990, Wang et al., 2006, Amrheln et al., 1992, Petkuviene & Paliulis, 2009).  However, the interaction of these salts with other known contaminants such as sulphates has not been thoroughly investigated and needs to be adequately addressed. It has been proven in the past that sulphur compounds have a great effect on corrosion (Qu et al., 2002, Dehwah et al., 2002), and thus their effect should not be disregarded when studying the effect of de-icing salts on the corrosion performance of galvanized steel. This chapter focuses on how these salts affect the corrosion performance of the galvanized steel reinforcement in the soil by adding Na2SO4 into the tested environment and to determine whether or not the presence of sulphates should be taken into account in the selection of de-icing salts. This chapter is based on a paper published as part of the work towards this PhD thesis.3  7.1 POTENTIODYNAMIC TESTING Most of the galvanized steel samples reached a steady open circuit state within the first 100 s of the test, which is in agreement with previous results (Padilla & Alfantazi, 2010), however, the samples immersed in MgCl2 took close to 2000 s to stabilize. Upon visual observation, at this stage of the experimental procedure, the galvanized layer is still fully intact, thus, the electrochemical                                                  3	Victor	Padilla,	Pouria	Ghods,	Akram	Alfantazi,	(2012),	Construction	and	Building	Materials	40	(2013)	908–918	   83behavior is governed by the properties of zinc. At first glance, it appears that the addition of different salts does not affect the recorded OCP, the difference between the values was about 40mV. The higher OCP value was observed for MgCl2 with a value of -905 mV Ag/AgCl and the lowest values was -935 mV Ag/AgCl for NaCl. This indicated that the specimens were in the first stage of galvanized steel corrosion (Stage I), as shown in Figure 2-1. In the presence of sodium sulphate (1wt% Na2SO4), the difference between the OCP measurements in four different solutions was larger than those in free-sulphate solutions. The highest recorded value was -915 mV Ag/AgCl for CaCl2 and the lowest was again observed for NaCl with a value of -1.04 V Ag/AgCl. Overall there was a slight decrease in the OCP values when Na2SO4 was added.   Figure 7-1 shows potentiodynamic polarization (PDP) curves of the samples immersed in the solutions containing 3.5 wt% of the four distinct tested de-icers: NaCl, MgCl2, CaCl2, and CH3CO2K. The plotted polarization curves follow the typical polarization behavior of zinc immersed in alkaline solutions. As expected with galvanized steel, the PDP curves reveal that the anodic process is controlled by activation and the cathodic process by diffusion. The relatively small slope observed in the anodic region, coupled with the small shift from the equilibrium potential, suggests that the anodic process is controlled by activation (Shortet al., 1989, Alvarez & Galvele, 1976). Due to the lack of Tafel behavior, the corrosion rate cannot be determined by using the Tafel extrapolation method, hence, the Butler–Volmer equation was used instead to approximate the kinetic parameters.   84i (A/cm2)10-8 10-7 10-6 10-5 10-4 10-3 10-2E(V) vs Ag/AgCl-1.2-1.0-0.8-0.6-0.4-0.20.0 3.5 wt% CaCl23.5 wt% CH3CO2K3.5 wt% MgCl23.5 wt% NaCl Figure 7-1- PDP curves of samples immersed at 25ºC aerated condition in the solutions containing 3.5 wt% of the four distinct de-icing salts: CaCl2, CH3CO2K, MgCl2, and NaCl.  The anodic branches of the curves show that, as the potential increases, the zinc coating starts to dissolve, and the anodic current density tends to become closer to that of the steel substrate moving away from that of pure zinc. The subsequent significant increase in the anodic current density with a potential increase suggests that zinc undergoes intense dissolution during anodic polarization and the formation of a porous corrosion layer has little effect on the dissolution of zinc.   The inflection point at which the galvanized steel sample began to show reversal behavior varied from -500 mV (Ag/AgCl) to -300 mV (Ag/AgCl). This reversal is attributed to the transition between zinc dissolution and the dissolution of iron in steel (Padilla & Alfantazi, 2012a), and the potential at which this reversal occurred appeared to vary depending on the salt used. However,   85when immersed in CH3CO2K, the reversal was not observed at all, and upon examination of the samples immersed in CH3CO2K, it was possible to detect some galvanizing left on the surface of the metal. This could be attributed to a change in the nature of the protective corrosion layer formed when immersed in distinct CH3CO2K, or in the lack of chloride. Culcasi et al. proposes that zinc dissolution is assisted by the chloride because chloride acts as a catalyst in the corrosion process (Culcasi et al., 2009).The subsequent significant increase in the anodic current density with a potential increase suggests that zinc undergoes intense dissolution during anodic polarization and the formation of a porous corrosion layer has little effect on the dissolution of zinc.   In the cathodic region, the limiting current density (iL) is the most important characteristic regarding diffusion-controlled processes (Ajeel & Ali, 2008). The cathodic branch of the polarization curves showed a similar behavior with a limiting current density ranging from approximately 8.2 x10-7 A/cm2 when immersed in MgCl2 solution to 4.7 x10-5A/cm2 in the CaCl2 solution. Variations in the cathodic parts of the PDP curves depend mainly on changes on the limiting current density. Since the oxygen concentration is similar in all the tested solutions and the stirring velocity and temperature are controlled, we believe that the perceived change in the CaCl2 solution might be caused by an increased oxygen availability at the metal interface due to a slower formation of corrosion products on the surface, and/or a greater solution conductivity (as indicated in the measurements shown in Table 7-1).      86Sample  pHConductivity (mS/cm) Ionic Strength (mol/l) Oxygen Concentration (ppm)* NaCl 6.5 51.9  0.060 8.67  MgCl2 6.4 26.8  0.110 8.57  CaCl2 8.9 49.8  0.095 7.85  CH3CO2K 7.4 26.4  0.036 7.02  NaCl + Na2SO4 6.5 63.6  0.081 8.88  MgCl2 + Na2SO4 6.2 34.8  0.131 8.69  CaCl2+ Na2SO4 8.6 56.9  0.116 9.30  CH3CO2K+ Na2SO4 7.3 37.2  0.057 7.43 *ppm = parts per million Table 7-1 The properties of the test solutions   Figure 7-2 shows the potentiodynamic polarization (PDP) curves of samples immersed in solutions containing 1wt% Na2SO4 plus 3.5 wt% of NaCl, MgCl2, CaCl2, and CH3CO2K, respectively. The shape of the cathodic and the anodic curves were similar to those observed in Figure 7-1. The lowest reversal behavior potential was recorded for samples immersed in 1wt% Na2SO4 plus 3.5wt% of NaCl at a potential of -640 mV (Ag/AgCl). In the presence of sulphates, it was possible to notice that CH3CO2K displays the reversal behavior at about the same potential as MgCl2 and CaCl2. The difference in behavior can be attributed to the presence of sulphates, since sulphates lower the local pH on the surface of the metal (Qu et al. 2002), accelerate corrosion, and prevent the metal from forming a protective layer.    87i (A/cm2)10-8 10-7 10-6 10-5 10-4 10-3 10-2E(V) vs Ag/AgCl-1.2-1.0-0.8-0.6-0.4-0.20.0 3.5 wt% CaCl2  + 1wt% Na2SO43.5 wt% CH3CO2K + 1wt% Na2SO43.5 wt% MgCl2 + 1wt% Na2SO43.5 wt% NaCl + 1wt% Na2SO4 Figure 7-2- PDP curves of samples immersed at 25ºC aerated condition in solutions containing 1wt% Na2SO4 plus 3.5 wt% of the four distinct de-icing salts: CaCl2, CH3CO2K, MgCl2, and NaCl.   The calculated corrosion parameters are presented in Table 7-2. The published work on the behavior of zinc under similar conditions is very limited, however, the PDP test results are in agreement with previous work published on galvanized steel (Le Manchet et al., 2010a, Zhang, 1996, Culcasi et al., 2009, Manov et al., 2000). The Ecorr values remained fairly constant in the different conditions, even when compared to the samples without sulphate added into the solution. The higher value, -855 mV Ag/AgCl, was recorded both for the sample immersed in 3.5wt% NaCl plus 1wt% Na2SO4, and the sample immersed in 3.5wt% CH3CO2K with 1wt% Na2SO4. The most negative value was recorded for the sample immersed in 3.5wt% MgCl2 plus 1wt% Na2SO4 with a value of -957 mV (Ag/AgCl).     88Sample  pH E(corr) (V/Ag/AgCl) i(corr) (µA/cm2)  iL (µA/cm2)  Error * (A/cm2) NaCl 6.5 -0.873 9.81 8.80  3.9 x10-09 MgCl2 6.4 -0.863 0.68  0.82  2.8 x10-11  CaCl2 8.9 -0.849 7.97  47.2  1.1 x10-07  CH3CO2K 7.4 -0.858 5.72  8.34 3.0 x10-09  NaCl + Na2SO4 6.5 -0.8554 29.1 43.6 5.0 x10-08  MgCl2 + Na2SO4 6.2 -0.9566 1.23  N/A 1.2 x10-11  CaCl2+ Na2SO4 8.6 -0.8664 12.9 7.78 4.1 x10-09  CH3CO2K+ Na2SO4 7.3 -0.8554 21.8 32.6 4.7 x10-08  * Nonlinear regression analysis absolute total error. Table 7-2 Calculated corrosion parameters using the Butler–Volmer equation for samples immersed in different solutions   In general, the corrosion rates of galvanized steel in the sulphate-added solutions were higher than those of the sulphate-free solutions after only 1 h of immersion. In agreement with the results observed in sulphate-free solutions, the PDP test results suggest that galvanized steel at early stage of corrosion (i.e., around 1 h after exposure to de-icing salts) is more resistant to MgCl2 salt, with the value of only 1.23 µA/cm2 (0.018 mm/year) in the sulphate-added solution and 0.68 µA/cm2 (0.010 mm/year) in the sulphate-free solution. However, it has the highest corrosion rate in the NaCl solution both in the absence and presence of sulphate with the corrosion rate value of 9.81µA/cm2 and 29.1µA/cm2, respectively. The corrosion rate in the other two solutions (i.e., CaCl2 and CH3CO2K) was ranked within these two ranges.    897.2 LINEAR POLARIZATION RESISTANCE TESTING The corrosion rates were obtained from the slopes of the LPR plots, using the following equations: pcorr RBi         (7.1)  redoxB113.21     (7.2)  where icorr stands for the corrosion current, B (V/dec) is a constant defined by equation 7.2, Rp (ohm.cm) is the polarization resistance, and ox  and red  are the oxidation and reduction Tafel slopes, respectively. In the corrosion rate calculation, it was assumed that the samples were still at the first corrosion stage and, thus, their electrochemical properties were those of pure zinc: molar mass of 65.39 g, density of 7.14 g/cm3,  a two-electron transfer process (n = 2), ox of 0.35 V/dec and red  of  -10 V/dec (El-Mahdy et al., 2000, Walter, 1991). This assumption was verified by the XRD and SEM results obtained after the LRP test. Figure 7-3 and Figure 7-4 show the results of a series of LPR measurements performed every hour over a 24 hour period. In Figure 7-3, the corrosion rate of the MgCl2 and CaCl2 solution after passing 5 hours became almost stable, whereas for the CH3CO2K and NaCl solutions, it started increasing after 10 hours of immersion and never became stable.  At 24 h, the lowest corrosion current values in the absence of sulphate were recorded in solution with 3.5wt% CaCl2, while the solution with 3.5wt% CH3CO2K rendered the highest value.    90Time (h)0 5 10 15 20 25corrosion rate (mm/yr)0.000.050.100.150.203.5wt% CH3CO2K3.5wt% MgCl23.5wt% CaCl23.5wt% NaCl Figure 7-3 - 24 hours Linear Polarization Resistance measurements for galvanized steel samples immersed in solutions containing 3.5 wt% of CH3CO2K, MgCl2, CaCl2, and NaCl at 25ºC            Figure 7-4- 24 hours Linear Polarization Resistance measurements for galvanized steel samples immersed in solutions containing 3.5 wt% of CH3CO2K, MgCl2, CaCl2, and NaCl with 1wt% Na2SO4 added at 25ºC.Time (h)0 5 10 15 20 25corrosion rate (mm/yr)0.00.10.20.30.4 3.5wt% CH3CO2K + 1wt% Na2S043.5wt% MgCl2 + 1wt% Na2S043.5wt% CaCl2 + 1wt% Na2S043.5wt% NaCl + 1wt% Na2S04  91In Figure 7-4, the corrosion rates of the NaCl, MgCl2, and CaCl2 followed a steady line and there were no remarkable changes over time. In addition, the values of the corrosion rate at 24 hours of testing were almost the same, which showed the performance of these three salts in the presence of sulphate were very similar. Comparing Figure 7-3 and Figure 7-4, the considerable change in corrosion behavior of galvanized steel is seen in the solution containing 3.5wt% CH3CO2K plus 1wt% Na2SO4 compared with the no-sulphate CH3CO2K solution. The corrosion values doubled in the presence of sulphate for the samples immersed in potassium acetate, while the samples immersed in MgCl2 and NaCl behaved fairly constant regardless of the presence of Na2SO4. The average corrosion rate of CaCl2 solution was around two times higher than the one in the sulphate-free solution. On the other hand, in sulphate-free environments, CH3CO2K seems to be a viable option to be used as a de-icer for galvanized steel in the soil due to its relatively low tendency to cause corrosion, however, when in the presence of sulphate, the possible protective layer formed on top of the metal surface is destroyed and the corrosion rate is almost four times higher.   The corrosion rates recorded after 1 hour of immersion are shown in Table 7-3. While some of the results are in agreement, it is possible to see important differences when calculating the corrosion rate for MgCl2; these differences could arise due to a change in OCP during the testing period. As reported earlier, when samples were immersed in MgCl2, they showed a higher instability in the OCP readings, which could have had an effect on the corrosion rate calculations.        92Sample PDP Corrosion rate (mm/year) LPR  Corrosion rate after 1h of immersion (mm/year) EIS Corrosion rate after 1h of immersion (mm/year) NaCl 0.146 0.336 0.204 MgCl2 0.010 0.693 0.217 CaCl2 0.119 0.356 0.334 CH3CO2K 0.085 0.312 0.085 NaCl + Na2SO4 0.435 0.698 0.225 MgCl2 + Na2SO4 0.018 0.124 0.212 CaCl2+ Na2SO4 0.192 0.475 0.720 CH3CO2K+ Na2SO4 0.326 0.605 0.265 Table 7-3- Corrosion rate of galvanized steel from different electrochemical test methods   7.3 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY RESULTS EIS was used in order to assess zinc coating degradation and the protectiveness of the corrosion layer formed on top of the galvanized steel surface when immersed in different de-icing agents. Figure 7-5 shows a typical Nyquist and Bode plot for the EIS results recorded in this work. Figure 7-5 shows that as immersion time increases, both the impedance and the phase angle decreases. An overall decrease in impedance, coupled with a decrease in the phase angle, indicate the deterioration of the coating over time.    93Z'0 2000 4000 6000 8000 10000 12000Z''020004000600080001000012000EIS in 3.5wt% MgCl2 + 1wt% Na2SO4 after 1 hEIS in 3.5wt% MgCl2 + 1wt% Na2SO4 after 24 h f / Hz1 10 100 1000 10000Phase / deg0204060803.5 wt% MgCl2, 1 wt% Na2SO4 after 1 h3.5 wt% MgCl, 1 wt% Na2SO4 after 24 h Figure 7-5-  Nyquist and Bode Phase plots recorded for samples immersed in aerated solutions with 3.5 wt% MgCl2 plus 1wt% Na2SO4 at immersion times of 1 h and 24 h    94The first set of EIS measurements was carried out in the solutions with 3.5wt% of each studied de-icing salt. Figure 7-6 shows Nyquist plots recorded for samples immersed in aerated solutions with 3.5 wt% NaCl, MgCl2, CaCl2, and CH3CO2K at 25ºC, at immersion times of a) 1 h, b) 4h, c) 7h, and d) 24 h. A decrease in the total impedance and polarization resistance during immersion along with a decrease in the coating capacitance indicates the degradation of the zinc coating on the steel substrate.  This behavior was observed at all de-icing salt solutions.  Only one semi-circle was observed in all Nyquist plots as shown above in Figure 7-6. Therefore, the equivalent electrical circuit of RS(QcRp) was used to obtain the EIS data of all the solutions. This is a commonly used circuit to interpret the electrochemical corrosion test results, where RS accounts for the solution resistance, Qdl and Rp for the double layer capacitance and polarization resistance of metal/solution interface, respectively (Xing et al., 2010).   The corrosion rate was calculated using the EIS polarization data similar to the procedure explained in section 7.2. The corrosion rates recorded after 1 h of immersion are shown in Table 7-3.  Figure 7-7 shows a plot with the corrosion rate results obtained from the EIS fitting polarization results during 24 h immersion in the 3.5wt% NaCl, MgCl2, CaCl2, and CH3CO2K solutions. The corrosion rate increased steadily in every tested solution. The increase in the corrosion rate indicated the deterioration of the zinc oxide layer on top of the underlying steel brought forward by the dissolution of the zinc layer, or more likely an increase in its porosity. Values for the coefficient n were all around 0.8. It has been reported that a non-ideal capacitive behavior and the appearance of a depressed semi-circle in the Nyquist plot occurs when n is located between 0.5 and 1 (Hassan et al., 2007). RS stayed relatively constant in each solution throughout the 24 h immersion test.   95                                                    Figure 7-6- Nyquist plots recorded for samples immersed in aerated solutions with 3.5 wt% NaCl, MgCl2, CaCl2, and CH3CO2K at 25ºC, at immersion times of a) 1 h, b) 2 h, c) 4h, d) 7h, and e) 24 h  Z'0 1000 2000 3000 4000 5000 6000Z''0100020003000400050006000 EIS  in 3.5wt% CaCl2EIS in 3.5wt% CH3CO2KEIS in 3.5wt% MgCl2 EIS in 3.5wt% NaCle) 24hZ'0 2000 4000 6000 8000Z''02000400060008000EIS  in 3.5wt% CaCl2EIS in 3.5wt% CH3CO2KEIS in 3.5wt% MgCl2 EIS in 3.5wt% NaClc) 4hZ'0 1000 2000 3000 4000 5000 6000Z''0100020003000400050006000 EIS  in 3.5wt% CaCl2EIS in 3.5wt% CH3CO2KEIS in 3.5wt% MgCl2 EIS in 3.5wt% NaCld) 7hZ'0 5000 10000 15000 20000 25000 30000Z''050001000015000200002500030000EIS  in 3.5wt% CaCl2EIS in 3.5wt% CH3CO2KEIS in 3.5wt% MgCl2 EIS in 3.5wt% NaCla) 1hZ'0 2000 4000 6000 8000 10000 12000 14000Z''02000400060008000100001200014000EIS  in 3.5wt% CaCl2EIS in 3.5wt% CH3CO2KEIS in 3.5wt% MgCl2 EIS in 3.5wt% NaClb) 2h  96The lowest recorded corrosion rate for CH3CO2K was 0.43 mm/year after 24 hours of immersion; while the highest value belonged to the CaCl2 solution with the corrosion rate of 2.08 mm/year. It appears that, when the galvanized steel was immersed in the CH3CO2K solution, the zinc cover is not affected as much as the one in NaCl, MgCl2, and CaCl2 solutions.  Time (h)0 5 10 15 20 25Corrosion rate (mm/yr)0.00.51.01.52.02.53.03.5wt% CH3CO2K3.5wt% MgCl23.5wt% CaCl23.5wt% NaCl Figure 7-7- Corrosion rate results obtained from the EIS fitting polarization results during 24 hour immersion in the 3.5wt% NaCl, MgCl2, CaCl2, and CH3CO2K solutions   To take into account the effect of sulphates, a second set of EIS measurements was carried out with the addition of 1wt% Na2SO4. Figure 7-8 shows the Nyquist plots recorded for samples immersed in aerated solutions with 3.5 wt% NaCl, MgCl2, CaCl2, and CH3CO2K plus 1wt% Na2SO4 at 25ºC, at immersion times of a) 1 h, b) 4h, c) 7h, and d) 24 h. The observed overall response was similar to the one reported above, with a general decrease in impedance and polarization resistance during immersion, coupled with a decrease in the coating capacitance, which indicated the degradation of the zinc coating on the steel substrate.  It has also seen a general drop in resistance compared to the samples immersed in sulphate-free solutions.    97 As shown in Figure 7-8a, the sample immersed in the solution with 3.5wt% MgCl2 plus 1wt% Na2SO4 showed a different behavior than the rests of the samples at lower frequencies. At low frequencies, it is possible to observe either the beginning of a second semi-circle or a Warburg-like behavior. As a consequence of this, one could use either the equivalent circuit RS(Qc(Rp(QdlRct)), or RS(Qc(RpW)) for the sample immersed in the 3.5wt% MgCl2 plus 1wt% Na2SO4.  Warburg responses indicate that the film is thick compared to the diffusion layer thickness, and the mass transport across this layer is similar to the semi-infinite model. (Mouanga & Berçot, 2010). Conversely, the RS(Qc(Rp(QdlRct)) equivalent circuit is commonly used to model porous corrosion product layers, where RS accounts for the solution resistance, Qc and Rp for the film capacitance and resistance of the porous layer respectively, Qdl and Rct account for the capacitance and transfer resistance/diffusion through the pores, respectively (Xing et al. 2010). However, in order to be able to reliably compare the results in the present work, the same equivalent above-mentioned circuit was used to find the EIS fitting data, so caution is advised with the results reported for the 3.5wt% MgCl2 plus 1wt% Na2SO4.           98       \                                Figure 7-8- Nyquist plots recorded for samples immersed in aerated solutions with 3.5 wt% NaCl, MgCl2, CaCl2, and CH3CO2K with 1wt% Na2SO4 added into the solutions at 25ºC, at immersion times of a) 1 h, b) 2 h, c) 4h, d) 7h, and e) 24 h Z'0 1000 2000 3000Z''0100020003000EIS  in 3.5wt% Ca2Cl + 1wt% Na2SO2SO4EIS in 3.5wt% CH3CO2K + 1wt% Na2SO2SO4EIS in 3.5wt% Mg2Cl + 1wt% Na2SO2SO4EIS in 3.5wt% NaCl + 1wt% Na2SO2SO4e) 24hZ'0 1000 2000 3000 4000Z''01000200030004000EIS  in 3.5wt% Ca2Cl + 1wt% Na2SO2SO4EIS in 3.5wt% CH3CO2K + 1wt% Na2SO2SO4EIS in 3.5wt% Mg2Cl  + 1wt% Na2SO2SO4EIS in 3.5wt% NaCl + 1wt% Na2SO2SO4d) 7hZ'0 1000 2000 3000 4000 5000 6000Z''0100020003000400050006000EIS  in 3.5wt% Ca2Cl + 1wt% Na2SO2SO4EIS in 3.5wt% CH3CO2K + 1wt% Na2SO2SO4EIS in 3.5wt% Mg2Cl  + 1wt% Na2SO2SO4EIS in 3.5wt% NaCl + 1wt% Na2SO2SO4c) 4hZ'0 2000 4000 6000 8000 10000Z''0200040006000800010000EIS  in 3.5wt% Ca2Cl + 1wt% Na2SO2SO4EIS in 3.5wt% CH3CO2K + 1wt% Na2SO2SO4EIS in 3.5wt% Mg2Cl  + 1wt% Na2SO2SO4EIS in 3.5wt% NaCl + 1wt% Na2SO2SO4b) 2hZ'0 2000 4000 6000 8000 10000 12000 14000Z''02000400060008000100001200014000 EIS  in 3.5wt% Ca2Cl + 1wt% Na2SO4EIS in 3.5wt% CH3CO2K + 1wt% Na2SO2SO4EIS in 3.5wt% Mg2Cl  + 1wt% Na2SO2SO4EIS in 3.5wt% NaCl + 1wt% Na2SO2SO4a) 1h  99Figure 7-9 shows the corrosion rates calculated from the EIS fitting data obtained for the total polarization resistance of galvanized steel during 24 h immersion in 3.5wt% NaCl, MgCl2, CaCl2, and CH3CO2K solution with the addition of 1wt% Na2SO4 into all the solutions. In the presence of Na2SO4, the sample immersed in the CaCl2 solution had the highest corrosion rate (1.41 mm/year) until 7 h of immersion, but it suddenly dropped likely due to the initiation of corrosion in the underlying steel. This shows that the corrosion mechanism of galvanized steel in the CaCl2 solution after 24 h of immersion mainly governed by the second stage of corrosion (stage 2) described in section 2. Apart from this abnormality, the corrosion mechanism of the galvanized steel in other solutions was generally attributed to the first stage of corrosion (stage 1). The sample immersed in the NaCl solution and CaCl2 solution had respectively the highest and lowest corrosion rate at 24 h among the solutions. Time (h)0 5 10 15 20 25Corrosion Rate (mm/yr)0.00.51.01.52.02.53.03.5wt% CH3CO2K + 1 wt% Na2SO43.5wt% MgCl2 + 1 wt% Na2SO43.5wt% CaCl2 + 1 wt% Na2SO43.5wt% NaCl + 1 wt% Na2SO4 Figure 7-9-  Corrosion rate results obtained from the EIS fitting polarization results during 24 hour immersion in the 3.5wt% NaCl, MgCl2, CaCl2, and CH3CO2K with 1wt% Na2SO4 added into the solutions   1007.4 SURFACE CHARACTERIZATION Previous work suggests that galvanized steel, when corroded, builds up corrosion products controlled by a charge transfer reaction (Hamlaoui et al., 2010); however, these products have been found to be not protective due to its very porous character, so this layer had a limited effect on the corrosion rate (Cachet et al., 2002,  Baugh, 1979). In the case of chloride-contained salts, zinc dissolution is assisted by chloride whose main role is believed to act as a catalyst in the corrosion process (Culcasi et al., 2009). The expected end corrosion product formed on top of the galvanized metal is Zn(OH)2. During the first stage of corrosion, in near neutral pH conditions, a semi-compact layer of precipitates, mainly Zn and ZnO with small amounts of Zn(OH)2, is present on the surface (Bonk et al., 2004). However, as immersion time increases, the layer grows, depleting the zinc coating and forming a thick, porous layer consisting mainly of a non-protective and soluble layer of Zn(OH)2 (Cachet et al., 2002). The reactions for the corrosion product formation have already been detailed in earlier work (Padilla & Alfantazi, 2010) so they will not be described in detail here. In the case of potassium acetate, where there is no chloride in the solution, the aforementioned mechanism is not applicable. This acetate species absorb on the zinc oxide surfac to form a surface zincacetate ligand (Johnson 2006). Acetate complexes absorb on zinc through the rupeture of O-H bonds (Bowker, 1983); which results in the formation of an absorbed acetate species and absorbed hydroxul groups. The following reaction mechanism has been proposed (Johnson 2006): 1. Formation of an adsorbed surface species, CH3COO(a), 2. Ligand exchange between CH3COO(a) and a surface hydroxyl group, 3. Surface coordination of CH3COO(a) with zinc, 4. CH3COO(a) induced dissolution of zinc, 5. Precipitation of zinc acetate, 6. Growth of three-dimensional zinc acetate with random orientation.   101Even in potassium acetate solutions, the end corrosion product of galvanized steel is composed of ZnO covered with a thick, porous layer of Zn(OH)2. Zinc acetate was not found after XRD nor EDX analysis. This indicates either that the amount of zinc acetate produced is minimal compared to zinc oxides, or a noncrystalline phase of the acetates has formed. Furthermore, zinc acetate reacts with the metal surface or corrosion products, consuming OH- or reducing the pH near the metal surface (Mahdavian, 2011). Acetic species adsorb on oxidized zinc to form zinc-acetate species. When formed, zinc acetates species are poor corrosion inhibitors, and zinc-acetate species are suggested to act as a precursor for the anodic zinc dissolution step.   Figure 7-10 - Surface morphology of galvanized steel obtained in a) 3.5wt% MgCl2 and b) 3.5wt% MgCl2 plus 1wt% Na2SO4 solutions after PDP testing in aerated conditions at 25ºC  Figure 7-10 shows the two distinct surface morphologies of galvanized steel obtained in a) 3.5wt% MgCl2 and b) 3.5wt% MgCl2 plus 1wt% Na2SO4 solutions after PDP testing in aerated conditions at 25ºC. In Figure 7-10(a), it is possible to appreciate the granular-shaped corrosion layer formed on top of the galvanized steel sample after PDP testing in 3.5wt% MgCl2. The film is formed by round, porous particles with a size ranging from 2 to 4 µm. Figure 7-10(b) shows similar shaped   102particles found on top of the galvanized steel sample after PDP testing in 3.5wt% MgCl2 plus 1wt% Na2SO4, the particles are the same size but no pores are visible.  Figure 7-11 shows two SEM pictures showing the surface morphologies of galvanized steel obtained in a) 3.5wt% CH3CO2K plus 1wt% Na2SO4 and b) 3.5wt% NaCl plus 1wt% Na2SO4. The predominant corrosion product detected in all of the samples immersed by the XRD analysis was Zn(OH)2. When comparing Figure 7-10(a) and Figure 7-11(a) it is possible to appreciate the distinct corrosion products formed on the metal surface depending on the de-icing salt used. Additionally, it is possible to see in Figure 7-11 (b) pseudo-hexagonal plane crystals known as platelets on the surface of the galvanized steel sample. These crystals are believed to have a preferential facing normal to the substrate plane. They nucleate on surface defects and then grow, forming islands which spread over the entire surface as the exposure time increases and are believed to be crystals of Simonkolleite (Zn5(OH)8Cl2.H2O) (Asgari et al., 2009, Seré et al., 1999). The presence of Simonkolleite was confirmed with XRD analysis by finding large diffraction peaks for Zn5(OH)8Cl2.H2O at angles 13° and 19° in the samples immersed in CaCl2 and MgCl2.  In  Table 7-4, based on the average of the EIS and LPR test results during 24 h exposure, the corrosion performance of different de-icing salts were ranked from the lowest corrosion rate (top) to the highest corrosion rate (bottom) for the two testing groups separately: sulphate-free solutions and sulphate-added solutions. This ranking clearly shows that the presence of sulphates should be taken into account in the selection of de-icing salts. Due to a higher concentration of chloride, one could intuitively expect that MgCl2 or CaCl2 could be one of the most damaging corrosion agents, which is in agreement with the results observed in the sulphate-free solutions in Table 7-4.  103Figure 7-11-  Surface morphology of galvanized steel obtained in a) 3.5wt% CH3CO2K plus 1wt% Na2SO4 and b) 3.5wt% NaCl plus 1wt% Na2SO4 solutions after PDP testing in aerated conditions at 25ºCPerformance Rank PDP test LPR test EIS testSulphate-free solutions (Lowest corrosion rate)(Highest corrosion rate)MgCl2 CH3CO2K CH3CO2KCH3CO2K NaCl NaClCaCl2 CaCl2 MgCl2NaCl MgCl2 CaCl2Sulphate-added solutions(Lowest corrosion rate)(Highest corrosion rate)MgCl2+Na2SO4MgCl2+Na2SO4CaCl2+ Na2SO4CaCl2+ Na2SO4CaCl2+ Na2SO4MgCl2 + Na2SO4CH3CO2K+ Na2SO4CH3CO2K+ Na2SO4CH3CO2K+ Na2SO4NaCl + Na2SO4NaCl+Na2SO4NaCl+Na2SO4Table 7-4 - Ranking of the effect of de-icing salts on the corrosion performance of galvanized steel  104In the presence of sulphates, the results suggest that galvanized steel is more resistant to corrosion when exposed to MgCl2 or CaCl2 de-icing salts compared with NaCl and CH3CO2K. In terms of how corrosive the chemicals are in the presence of sulphates, the tested de-icers can be arranged as follows: NaCl > CH3CO2K > CaCl2/MgCl2; and in the absence of sulphate, the de-icing salts can be classified as: CaCl2/MgCl2 > NaCl > CH3CO2K.   The difference in corrosion performance in the two distinct conditions (sulphate containing and sulphate free solutions) is noteworthy. The rate-determining step for zinc dissolution depends on the type of anions in the electrolyte. The dissolution of zinc in NaCl solution is diffusion-limited controlled, and the dissolution of zinc in presence of acetates appears to be dependant on the concentration of absorbed intermediate species (Zhang, 1996). Anions that enhance surface acidification (SO4-2), or form zinc salts more soluble than zinc oxide and hydroxydes (such as the case of zinc acetates) tend to reduce the stability of the corrosion products creating surface inhomogenieties (Zhang, 1996). If the corrosion products are not “self-repairing” then dissolution at these areas will intensify and expand. This could explain why corrosion behavior of galvanized steel in the presence of CH3CO2K seems to change significantly in the presence of sulphates.   It is important to note that the corrosion performance of CaCl2 and MgCl2 was very similar in this study. Therefore, more studies are required to determine the relative performance of these two salts as de-icing chemicals.   7.5 SUMMARY This chapter presented the effect of commonly used de-icing salts such as sodium chloride, magnesium chloride, calcium chloride, and potassium acetate on the corrosion performance of the   105galvanized steel in sulphate-contaminated soil. To achieve this goal, different electrochemical testing methods such as potentiodynamic polarization, linear polarization resistance, and electrical impedance spectroscopy were used to determine the corrosion rate of galvanized steel reinforcement in sodium-sulphate-containing solutions. Results showed that the presence of sulphate in the soil significantly increased the corrosion rates and, thus, it is important to consider the effect of sulphate in determining the type of de-icing salt. In sulphate-free solutions, potassium acetate appeared to be the best option; while in the presence of sulphate, MgCl2 and CaCl2 had the lowest corrosion rate. Electrochemical test results showed that the corrosion rate of galvanized steel in the presence of sulphate is considerably higher than the one in the sulphate-free soil condition.     1068 CORROSION FILM BREAKDOWN OF GALVANIZED STEEL IN SULPHATE-CHLORIDE SOLUTIONS  The previous chapter focused on the behavior of galvanized steel immersed in commonly used de-icing salts plus sulphates. The results suggest that both CaCl2 and MgCl2 are the least corrosive de-icing salts in sulphate-contaminated conditions, followed by CH3CO2K and finally NaCl. Nevertheless, the effect of these commonly used de-icing salts on the corrosion of galvanized steel has not been fully addressed and more in depth work was needed to understand how the corrosion products of galvanized steel are affected when exposed to these aggressive agents. This work focused on the destructive effect of that commonly used de-icers, combined with sulphates, have on the corrosion film formed on galvanized steel, destroying or avoiding passivity altogether. This chapter is based on a paper currently under review, submitted as part of the work towards this PhD thesis.4 8.1 CYCLIC VOLTAMMETRY AND CYCLIC POLARIZATION RESULTS  Cyclic voltammetry was used in order to study surface area changes of galvanized steel in the different tested salts. Figure 8-1 shows a representative cyclic voltammogram for the 20th cycle reading of a galvanized steel sample immersed in 3.5wt% NaCl plus 1wt% Na2SO4 at 25ºC with a scan rate of 20 mV s-1. The bold arrows show the direction of the scan. Two oxidation peaks (labelled A1 and A2) and two reduction peaks (labelled C1 and C2) were identified. The first peak, A1, was located between -960 mV and -910 mV (Ag/AgCl), and in agreement with the PDP graph shown in Figure 7-2, the A1 peak is noticeable at more noble potentials than the open circuit potential value, indicating the relatively fast activation of the metal surface at potentials close to the OCP. In the                                                  4 Victor Padilla, Akram Alfantazi,  Corrosion Science, CORSCI-S-13-01733   107nearly neutral pH ranges in which the samples were immersed, it is possible to attribute the A1 peak to the formation of a zinc corrosion product.  The A2 peak was attributed to the active anodic dissolution of zinc (Miao et al., 2007a), which was observed as a rapid increase of current in a small potential range, from -900 mV to -800 mV (Ag/AgCl) approximately. E(V) vs Ag/AgCl-1.4 -1.2 -1.0 -0.8(i) A/cm2-5.0e-40.05.0e-41.0e-31.5e-33.5 wt% NaCl + 1wt% Na2SO4A1C1C2EbErA2  Figure 8-1- 20th cycle Cyclic Voltammogram for a galvanized steel sample in solution with 1wt% Na2SO4 and 3.5 wt% NaCl at 25ºC and a scan rate of 20 mV s-1  In potentials following the formation of the first oxidation peak, A1, the surface is believed to be covered with the corrosion product, and thus the current drops slightly to what is commonly known as the ipass plateau. This ipass plateau is attributed to the protective nature of the formed corrosion product and, ideally, the bigger the plateau, the more desirable. However, as observed, this passive-like regime is very short. The current increases when the critical potential (Eb) is reached,   108which is attributed to the breakdown of the protective film. This abrupt increase is followed by a hysteretic loop.  This feature has been frequently reported when breakdown of the corrosion film is observed (Hamdy H, 2001). Some authors (Fletcher et al., 1983) believe that the current maximum of the reverse scan is associated with a nucleation and growth mechanism which has been described earlier as the direct reaction of zinc with hydroxyl groups. Thus, process A2 in Figure 8-1 could involve the breakdown of corrosion films (Miao et al., 2007a). Finally, the corrosion film healing potential can be identified when a hysteresis loop is observed in the cyclic voltammograms (Hamdy H, 2001) (Miao et al., 2007b), and it has been identified in the cyclic voltammogram as Er.  Both A1 and C1 show similar charge magnitude but the potential difference between the two peaks is approximately 100 mV, suggesting that these two processes are not reversible. In addition, the charge consumed for process A2 is larger than that for process C2; Miao et al. (2007b) suggested that this indicates that not all of the corrosion products can be reduced back to their initial species under these conditions.   Figure 8-2 shows the cyclic voltammograms recorded for samples immersed in solutions with 1wt% Na2SO4 and 3.5wt a) CaCl2, b) CH3CO2K, c) MgCl2, and d) NaCl at 25ºC and a scan rate of 20 mV s-1. Both the anodic and cathodic peaks showed only small changes in position and magnitude with following cycle scans.  However, it is possible to notice that Er shifts to more negative potentials as the cycle increases, which has been attributed to the formation of meta-stable and stable film defects (Franke, 1998). The same fundamental behavior was observed in all the samples.   109E(V) vs Ag/AgCl-1.4 -1.2 -1.0 -0.8i (A/cm2)-0.00050.00000.00050.00100.0015 Cycle 1Cycle 5Cycle 10Cycle 15Cycle 20d) 3.5 wt% NaCl + 1wt% Na2SO4E(V) vs Ag/AgCl-1.4 -1.2 -1.0 -0.8 -0.6-0.0010.0000.0010.0020.0030.004Cycle 1Cycle 5Cycle 10Cycle 15Cycle 20a) 3.5wt% CaCl2 + 1wt% Na2SO4E(V) vs Ag/AgCl-1.4 -1.2 -1.0 -0.8-0.00020.00000.00020.00040.00060.0008 Cycle 1Cycle 5Cycle 10Cycle 15Cycle 20c) 3.5wt% MgCl2 + 1wt% Na2SO4E(V) vs Ag/AgCl-1.4 -1.2 -1.0 -0.8 -0.6i (A/cm2 )-0.00010.00000.00010.00020.0003Cycle 1Cycle 5Cycle 10Cycle 15Cycle 20b) 3.5wt% CH3CO2K + 1wt% Na2SO4 i (A/cm2 )i (A/cm2 ) Figure 8-2-  Cyclic voltammograms recorded for samples immersed in solutions with 1wt% Na2SO4 and 3.5wt% a) CaCl2,  b) CH3CO2K, c) MgCl2, and d) NaCl at 25ºC and a scan rate of 20 mV s-1  Samples immersed in NaCl, MgCl2, and CaCl2 show a shift of Er values to more noble potentials as the number of cycles increased. For the sample immersed in CH3CO2K, Er shifted to more negative potentials as the number of cycles increased. The highest peak A1 current density recorded was that of the sample immersed in NaCl (Figure 8-2), while the lowest corresponds to the   110sample immersed CH3CO2K (Figure 8-2). The lower intensity of the peaks can be attributed to slower corrosion kinetics.  E(V) vs Ag/AgCl-1.4 -1.2 -1.0 -0.8i (A/cm2 )-0.0010.0000.0010.0023.5wt% CaCl2 + 1wt% Na2SO4 3.5wt% CH3CO2K + 1wt% Na2SO4 3.5wt% MgCl2 + 1wt% Na2SO4 3.5wt% NaCl+ 1wt% Na2SO4 After 20 cyclesA1C1C2A2 Figure 8-3-  Cyclic voltammograms recorded for samples immersed in solutions with 1wt% Na2SO4 and 3.5 wt% CaCl2, CH3CO2K, MgCl2, and NaCl after 20 cycles at 25ºC and a scan rate of 20 mV s-1  Figure 8-3 shows the cyclic voltammograms recorded for samples immersed in solutions 1wt% Na2SO4 and 3.5 wt% CaCl2, CH3CO2K, MgCl2, and NaCl after 20 cycles. When the galvanized steel is immersed in CH3CO2K, the two anodic peaks, A1 and A2, as well as the cathodic peaks, show the lowest intensity. Peak A1’s intensity is similar in samples immersed in MgCl2 and CH3CO2K, but peak A2 shows a larger intensity when immersed in MgCl2, indicating a higher activity for active zinc dissolution.  This could be explained by the slow kinetics observed for samples immersed in MgCl2 during PDP testing, delaying the formation of a protective layer and thus increasing zinc   111dissolution in shorter periods of time. The sample immersed in NaCl solution exhibited the shortest plateau, while the sample immersed in CH3CO2K, the longest. This confirms the strong corrosion film breakdown dependence in the presence of Cl- (Liu et al., 2012).   The amount of charge that is transferred to the electrolyte when the material is reduced or oxidized is directly proportional to the active surface area which is in contact with the electrolyte. Previous work done on galvanized steel has shown that when the surface is covered with a barrier film that cannot be oxidized or reduced, then the protected area can be estimated by the difference of the maximum oxidation peaks current on an unprotected and a partially-protected surface (Titz et al., 2010). A second viewpoint to this phenomenon is that the increase in the oxidation peaks current could arise either due to an increase of area or an increased number of defects on the surface (Miao et al., 2007b). This comparative analysis provides information on the protective nature of the corrosion film formed on the surface of the metal, either by a decrease in active area due to the formation of a protective corrosion product film, or by an increase in area due to a higher surface deterioration.  As evidenced by Figure 8-4, there is no evidence of pitting of galvanized steel under the studied conditions. Further supported by the evidence found on SEM pictures, there is no evidence of an increase of surface area due to an increased number of defects on the corrosion products.  Because of this, one can assume that the decrease in peak magnitude is related to an increase in protected surface area by an oxide barrier. However, at this moment it is not possible to determine with a high degree of certainty which of the viewpoints better reflect the reality of the nature of the change in magnitude of the anodic peaks, and more reaearch needs to be done. The changes in the active surface areas estimated by this method are relative to one another and the samples used as a baseline for the calculations were the samples immersed in NaCl solutions.   112Samples immersed in NaCl solutions were selected as baseline because, according to the PDP results, the samples immersed in solutions containing NaCl yielded the highest corrosion currents. This is consistent with the CV readings; the samples immersed in NaCl also showed the highest maximum oxidation peaks current after the 20th cycle in the CV tests. The increase of the maximum oxidation current density is associated with the increase of the active surface area and a decrease on the oxidation current density could be associated with a decrease of surface activity.   The cyclic voltammetry started with a cathodic pulse, so one can expect that the native oxide film formed on the surface of the galvanized steel is reduced with the first scan, and the subsequent reading will reflect the oxide formation in those particular conditions.  Titz et al. (2010) suggested the following equation to estimate the film surface area covered by a layer:   )//()//(maxmax1NaClZnOZnokxZnOZnoxiifilm     (8. 1)  where Φfilm is the relative percentage of the film surface area covered by a corrosion layer (%), )//(max xZnOZnoxi is the maximum current density for the anodic peaks recorded for a given salt, x, and )//(max NaClZnOZnoki  is the maximum current density for the anodic peaks recorded when the sample was immersed in NaCl solution. The calculated values for the percentage of the covered surface area relative to the sample immersed in NaCl are shown in Table 8-1. The highest calculated value was that of CH3CO2K with 90.2% less surface activity than the one observed when immersed in 3.5wt% NaCl. This difference in maximum peak currents is being attributed to less active surfaces, which could be caused by a higher surface coverage of a corrosion product film. Due to the decrease of surface activity, one could expect CH3CO2K to show the lowest corrosion current density in both   113sulphate-containing and sulphate-free solutions, however, that is not case. This could be explained by the absence of Cl- in the CH3CO2K solutions, which could favor the stability of the corrosion product or by the fundamental differences between PDP and CV testing.  The potential range used for the cyclic voltammetry scans, 275 to -500 mV above and below OCP, is large enough to change the surface of the material and potentially degrade the corrosion films formed. Caution is advised when comparing PDP and CV results.      Sample Er (V) vs Ag/AgCl Φfilm (%) NaCl + Na2SO4 -0.914 - MgCl2 + Na2SO4 -0.907 56.5 % ** CaCl2 + Na2SO4 -0.910 17.6%  ** CH3CO2K + Na2SO4 -0.918 90.2%  ** ** Relative to Max iox value recorded for 3.5wt% NaCl + 1 wt% Na2SO4   Table 8-1- Calculated relative surface coverage for galvanized steel recorded after 20 cycles in 3.5wt% NaCl, MgCl2, CaCl2 and CH3CO2K.    In order to better understand the susceptibility of film breakdown, cyclic polarization was done. The Cyclic Polarization curves for samples immersed in solutions with 1wt% Na2SO4 and 3.5 wt% a) CaCl2, b) CH3CO2K, c) MgCl2, and d) NaCl are shown in Figure 8-4. The bold arrows show the directions of the forward and the backward scans.     114                           Figure 8-4- Cyclic Polarization curves for samples immersed in solutions with 1wt% Na2SO4 and 3.5 wt% a) CaCl2, b) CH3CO2K, c) MgCl2, and d) NaCl    When the sample was immersed in MgCl2 (Figure 8-4c), the current densities of both the forward and the backward scan are very close; for the rest of the samples, the current densities of the backward scan were initially lower than those of the forward scan. This feature is commonly known as positive hysteresis and is being attributed to the formation of a corrosion product on top of the corrosion products, and the fact that the reverse anodic curve is shifted to lower currents suggests that the corrosion is taking place in a uniform manner as opposed to localized or pitting corrosion. The wider hysteric loop is observed for the sample immersed in NaCl (Figure 8-4) and also the backward scan crosses the forward scan at the lowest potential, indicating a higher i (A/cm2)10-8 10-7 10-6 10-5 10-4 10-3 10-2E(V) vs Ag/AgCl-1.2-1.0-0.8-0.6-0.4-0.20.0 3.5 wt% CH3CO2K + 1wt% Na2SO4b)Eri (A/cm2)10-8 10-7 10-6 10-5 10-4 10-3 10-2E(V) vs Ag/AgCl-1.2-1.0-0.8-0.6-0.4-0.20.0 3.5 wt% CaCl2 + 1wt% Na2SO4a)Eri (A/cm2)10-8 10-7 10-6 10-5 10-4 10-3 10-2E(V) vs Ag/AgCl-1.2-1.0-0.8-0.6-0.4-0.20.0 3.5 wt% MgCl2 + 1wt% Na2SO4c)Eri (A/cm2)10-8 10-7 10-6 10-5 10-4 10-3 10-2E(V) vs Ag/AgCl-1.2-1.0-0.8-0.6-0.4-0.20.0 3.5 wt% NaCl + 1wt% Na2SO4d)Er  115degradation of the corrosion film. The corrosion film healing potential, Er, has also been identified in the cyclic polarization curves. For the sample immersed in the CaCl2 solution, the observed Er was -803 mV (Ag/AgCl), -819 mV (Ag/AgCl),  for the sample immersed in CH3CO2K,  -880 mV (Ag/AgCl) for the sample immersed in MgCl2,  and finally – 905 mV (Ag/AgCl) for the sample immersed in NaCl.    Galvanized steel did not show a passive behavior in any of the samples and the results are in good agreement with PDP results. Liu et al. (2012), reported similar behavior when galvanized steel was immersed at pH=7, with current density increasing abruptly until it reaches the value of nearly 0.01 A/cm2, followed by a slow and steady increase as the potential applied increased. In all cases, the passivity breakdown potential is very close to the equilibrium potential of zinc, the same behavior was observed previously in experiments performed in zinc samples (Thomas et al., 2012).   When the samples were studied with CV, the best performing samples were those immersed in the CH3CO2K, closely followed by those immersed in MgCl2 containing solutions. All CV graphs showed two clear anodic peaks, the first peak (A1) was attributed to the formation of a zinc corrosion product followed by a short current plateau and a subsequent current increase at the second peak (A2) which was attributed to the active anodic dissolution of zinc. This supports that the initial current increase near the corrosion potential observed during PDP results can be attributed to the formation of a poorly protective corrosion. Cyclic Polarization results also confirm that the galvanized steel samples failed to form a passive film under the studied conditions and the positions of the reverse scans confirm that no pitting occurs, but rather that the corrosion film is not protective, and the metallic zinc coating dissolves through a film, forming soluble, less dense, and scarcely protective corrosion products (Elsner et al., 2012).    116  8.2 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY RESULTS  A second set of EIS tests were performed for this section. In this section, EIS was used in order to assess coating degradation, estimate the degree of protection provided by the corrosion layer formed on top of the galvanized steel surface when immersed in different sulphate-chloride solutions, and to provide theoretical information on the corrosion layer thickness (Wang et al., 2012, Mohammadi et al., 2011). Figure 8-5 shows the Nyquist and Bode Phase plots recorded for the sample immersed in 3.5 wt% NaCl plus 1 wt% Na2SO4 at immersion times of 1 h, 2 h, 3 h, 4 h, 6 h, 7 h, and 24 h. This is the typical Nyquist and Bode plot for the EIS results recorded in this work. Figure 8-5 shows that, as immersion time increases, both the impedance and the phase angle decrease. This same behavior, decrease in the total impedance and polarization resistance during immersion, was observed in all tested solutions. Two time constants became visible as immersion time increased and a low frequency inductive behavior was observed at early immersion times. The high frequency loop could be related to a charge transfer reaction of the coating (Li et al., 2013), while the low frequency inductive behavior has been attributed to the changes in the surface coverage by the adsorbed species involved in the zinc dissolution process (Cachet et al., 2001) or the diffusion process (Gervasi et al., 1994).   117f / Hz0.01 0.1 1 10 100 1000 10000Phase / deg-60-50-40-30-20-1001 h2 h 3 h4 h6 h 7 h24 h Z'0 1000 2000 3000 4000Z''010002000300040001 h2 h3 h4 h6 h7 h24 h Time Time Figure 8-5– Nyquist and Bode Phase plots recorded for samples immersed in aerated solutions with 3.5 wt% NaCl plus 1 wt% Na2SO4 at immersion times of 1 h, 2 h, 3 h, 4 h, 6 h, 7 h, and 24 h   118Figure 8-6 shows the Nyquist plots obtained for samples in aerated solutions immersed in 3.5wt% CaCl2, CH3CO2K, MgCl2, and NaCl solution with the addition of 1wt% Na2SO4 into all the solutions at 25ºC at immersion times of a) 1 h and b) 24 h. A decrease in the total impedance and polarization resistance during immersion along with a decrease in the coating capacitance indicates the degradation of the zinc coating on the steel substrate.  When immersed in CaCl2, galvanized steel shows the highest impedance values, indicating slower corrosion kinetics, or a more protective corrosion film formed on the surface of the corroding metal.  On the other hand, the sample immersed in the CH3CO2K solution had the lowest recorded impedance after 24 hours of immersion.    It is possible to observe a second semi-circle at low frequencies. This behavior is often attributed to the formation of porous corrosion products. The RS(Qct(Rct(Q1Rl))) equivalent circuit is commonly used to model porous corrosion product layers (Baril et al., 2007, Seré et al., 1999, Le Manchet et al., 2010c), where RS accounts for the solution resistance, Q1 and R1 for the film capacitance and resistance of the porous layer, respectively, and Qct and Rct account for the capacitance and transfer resistance/diffusion through the pores, respectively (Xing et al., 2010). Nevertheless, the RS(Qct(R1(Q1Rel)))Q2Ril) circuit had a better fit to the results yielded by the EIS tests presented in this work. The Q1Rel element represents the film capacitance and resistance of a porous corrosion film product (presumably Zn(OH)2). The Q2Ril element has been added in a series representing the film capacitance and resistance of an inner film. This inner layer could be formed by a dense and compact corrosion product under the porous layer. Given enough time, two different layers could be formed on top of the metal: a thick porous layer of Zn(OH)2 on top of the ZnO film (Baril et al., 2007).    119Z' / cm20 200 400 600 800 1000 1200 1400 1600Z'' /  cm202004006008001000120014001600b) 24hZ' / cm20 2000 4000 6000 8000 10000 12000 14000Z'' / cm202000400060008000100001200014000 EIS  in 3.5wt% CaCl2 + 1wt% Na2SO4EIS in 3.5wt% CH3CO2K + 1wt% Na2SO4EIS in 3.5wt% MgCl2  + 1wt% Na2SO4EIS in 3.5wt% NaCl + 1wt% Na2SO4a) 1hZ'0 500 1000 1500 2000 2500 3000Z''050010001500200025003000EIS  in 3.5wt% CaCl2 + 1wt% Na2SO4EIS in 3.5wt% CH3CO2K + 1wt% Na2SO4EIS in 3.5wt% MgCl2  + 1wt% Na2SO4EIS in 3.5wt% NaCl + 1wt% Na2SO4 Figure 8-6- Nyquist plots recorded for samples immersed in aerated solutions with 3.5wt% CaCl2, CH3CO2K, MgCl2, and NaCl solution with the addition of 1wt% Na2SO4 into all the solutions at 25ºC at immersion times of a) 1 h and b) 24 h    120Figure 8-7 shows a schematic representation of the equivalent electrical circuit. Due to the noticeably depressed semi-circles in the Nyquist plot, the proposed model in figure 8.7 assumes that constant phase elements (CPE) were present in the system. A constant phase element is an equivalent electrical circuit component imitating the behavior of an imperfect capacitor (Mohammadi et al., 2011). It has been previously reported that a non-ideal capacitive behavior and the appearance of a depressed semi-circle in the Nyquist plot occurs when n is located between 0.5 and 1 (Hassan et al., 2007).          Figure 8-7- Schematic representation of the two equivalent electrical circuit used for modelling an electrode protected by a porous layer   The decrease of total polarization resistance values can be explained by the dissolution of the porous layer into the solution. On the other hand, Ril behaved fairly consistently throughout the test, this suggests, as previously reported (Padilla & Alfantazi, 2012a), that a thick porous layer of Zn(OH)2 is formed on top of a thinner, more compact film of ZnO at the metal surface. During the first stage of corrosion, in the case of near neutral pH conditions, a semi-compact layer of precipitates, mainly Zn and ZnO with small amounts of Zn(OH)2, is present on the surface (Feitknecht, 1959, RsR1RelQ 3R il Q2Qc  121Bonk et al., 2004) . However, as immersion time increases, the layer grows, depleting the zinc coating and forming a thicker, porous layer consisting mainly of a poorly protective and soluble layer of Zn(OH)2. Klemm et al. (2011), when studying the dissolution of zinc in a similar pH range, described the reaction mechanisms for the chemical dissolution of the corrosion film as follows: ZnO + H+ [Zn(OH)] +      (8.2) Zn(OH)2 +H+ [Zn(OH)] + +H2 O     (8.3) The decrease of the resistance polarization indicated the deterioration of the zinc oxide layer on top of the underlying steel brought forward by the dissolution of the zinc layer or, more likely, an increase in its porosity. Values for the coefficient n were all close to the range of 0.77 to 0.81. RS values were relatively constant in each solution throughout the 24 h immersion test.  Total polarization values, the summation of R1 with Rel and Ril, are depicted in Figure 8-8. Time (h)1 2 3 4 6 7 24Total Rp cm2   1021031041053.5wt% CaCl2 + 1wt% Na2SO43.5wt% CH3CO2K + 1wt% Na2SO43.5wt% MgCl2  + 1wt% Na2SO43.5wt% NaCl + 1wt% Na2SO4b) Figure 8-8- Calculated Rp values at 25ºC samples immersed in aerated solutions with 3.5wt% CaCl2, CH3CO2K, MgCl2, and NaCl solutions with the addition of 1wt% Na2SO4 in all the solutions The calculated values, from the fitting EIS models to the experimental results, were used to estimate the thickness of the corrosion product formed on the samples. In order to calculate the film thickness, the simulated values for resistance and CPE were used. However, to estimate film   122thickness, it is imperative to first estimate the capacitance values from the CPE values. Hsu and Mansfeld proposed the following expression (Equation 8.4) to estimate effective capacitance from CPE parameters when assuming distribution of time constants in the presence of an Ohmic Resistance:                 (8.4)  where Ceff is the effective capacitance (F/cm2), Q is the CPE, and Rf is the resistivity of the film. The calculated true capacitance values were all in the range of 10-5 to 10-8 F/cm2. This suggests the presence of an oxide layer or a porous surface for values closer to 10-5 F/cm2. Using the calculated effective capacitance with the formula above, the film thickness (df) was calculated according to following formula (Equation 8.5):                (8.5)  where C is the capacitance, ɛo is the permittivity of vacuum (8.854 ×10−12 F m-1), is the dielectric constant and it was set to 30, A is the effective area, and d is the thickness of the oxide layer. The calculated film thickness is shown in Figure 8-9. The circuit elements QctR1 were not taken into account for these calculations since they do not represent any of the corrosion layers formed, the double layer capacitance may be negligible when the corrosion product film is very thin (Wallinder et al., 1998).  effof CAd nnfneff RQC /)1(/1   123Time (h)1 2 3 4 6 7 24Film Thickness ( m)10-310-210-11001013.5wt% CaCl2 + 1wt% Na2SO43.5wt% CH3CO2K + 1wt% Na2SO43.5wt% MgCl2  + 1wt% Na2SO43.5wt% NaCl + 1wt% Na2SO4b)   Figure 8-9- Calculated film thickness for samples immersed in aerated solutions with 3.5wt% CaCl2, CH3CO2K, MgCl2, and NaCl solutions with the addition of 1wt% Na2SO4 into all the solutions. At 25ºC, after 24 hours of immersion   Figure 8-10 shows selected SEM cross-sectional images of the corroded samples showing the thickness of the corrosion product layer. The samples were mounted in cold cure epoxy resin, cut, polished, and cleaned with ethanol prior to the SEM exploration. The composition of each layer was confirmed using EDX. The underestimation of the values could be explained due to the assumption of equation 9.5 that the corrosion product is evenly distributed across the exposed surface area. After visual inspection and SEM exploration, it was possible to confirm that a) not all the surface area is covered with a thick and porous corrosion product, and b) in areas where the   124galvanized steel was covered with a corrosion product, the corrosion product thickness ranged from 2 – 5 µm.       Figure 8-10 -Selected Cross-sectional SEM images showing taken from galvanized steel obtained in 1 wt% Na2SO4 plus a) 3.5wt% CaCl2, and b) 3.5wt% NaCl after 24 hours of immersion EIS testing in aerated conditions at 25ºC Steel SubstrateZinc Coating Corrosion ProductEpoxy Resin Steel SubstrateZinc Coating Corrosion ProductEpoxy Resin b) a)   125SEM pictures also revealed a decrease in thickness of the galvanized coating layer, which confirms that the corrosion product formed on top of the zinc layer is not protective and does not fully stop the dissolution process of zinc. The Hsu and Mansfeld formula used to calculate effective capacitance, and to later estimate film thickness, underestimated the values when compared to SEM cross-sectional pictures. This was attributed to the non-uniform nature of the film growing on the exposed area. The Hsu and Mansfeld formula is perhaps better suited for the calculation of a uniform passive film and not for rough, uneven porous corrosion products.  8.3 SUMMARY This chapter presented the corrosion performance of galvanized steel in sulphate-chloride solutions and studies the grown corrosion products films, looking at surface area changes, film breakdown susceptibility, and protectiveness. The electrochemical corrosion behavior of zinc was studied using Potentiodynamic Polarization, Cyclic Voltammetry and Cyclic Polarization, and Electrochemical Impedance Spectroscopy. Results showed that the formed corrosion layer is non-protective. MgCl2 had the lowest corrosion rate, while samples immersed in the NaCl-containing solutions showed the worst corrosion performance. The highest performance observed in MgCl2 can be attributed to slower corrosion kinetics, a good surface coverage of the corrosion product (surpassed only by samples immersed in CH3CO2K solutions), and higher polarization resistance as studied by EIS.      1269 MODEL FOR THE CORROSION BEHAVIOR OF GALVANIZED STEEL IN SOIL   This chapter introduces an advanced deterministic model for the corrosion rate determination of galvanized steel in soil. The developed model simulates all three stages of the galvanized steel corrosion process through the fundamental electrochemical reactions involved in the corrosion process. It considers the effect of environmental conditions such as temperature and moisture content of the soil as well as its physicochemical properties, including the pH and the electrical resistivity. This chapter is based on a paper published as part of the work towards this PhD thesis.5 9.1 MODELING CONCEPT In this study, the corrosion behavior of galvanized steel is simplified into three stages which are very similar to the ones explained in earlier chapters. In the first stage, the corrosion of the zinc layer is modeled so that in the contaminated region (i.e. corroding site) of the soil, the zinc layer is dissolved by anodic reaction and oxygen reduction takes place on the surface of the zinc layer as a cathodic reaction. In the second stage, once the zinc layer in the corroding site is removed by the anodic reaction, the underlying steel is exposed to the contaminated soil. As a result, the steel dissolution starts at the corroding site coupled by oxygen reaction in the non-corroding site (i.e. the uncontaminated region) of the soil on the surface of the remaining zinc layer. In the last stage of corrosion, when the large amount of the zinc layer is dissolved, the corrosion of underlying steel continues at the corroding site, however, at the non-corroding site, the oxygen reduction mainly occurs on the surface of the underlying steel instead of the zinc layer.                                                  5	Victor Padilla, Pouria Ghods, and Akram Alfantazi (2013). Corrosion: May 2013, Vol. 69, No. 5, pp. 509-521	 127Figure 9-1 The schematic illustration of microcell and macrocell corrosion of galvanized steel in soilIn order to consider the three-stage corrosion behavior of galvanized steel, a combined macrocell and microcell corrosion mechanism is assumed as a governing corrosion mechanism of galvanized steel reinforcement in the soil. Therefore, as shown in Figure 9-1, the microcell corrosion activities simultaneously occur in both the corroding site (contaminated soil) and the non-corroding site (uncontaminated soil) during the three stages of corrosion, however, the electrochemical potential imbalance between the corroding and non-corroding sites results in macrocell corrosion activity; hence, a macrocell corrosion current. Because of the fact that the microcell corrosion activities of galvanized steel at each stage of corrosion depends on the anodic and cathodic reactions in both corroding and non-corroding sites, the details of the governing electrochemical reactions and the corresponding equations are described separately for each stage as follows:    128  Figure 9-2 The combined microcell and macrocell corrosion of galvanized steel   Current density, i Current density, i                                ic,Ox  icor,mic        cor,mic                       ia,Zn  inon-cor,mic    IRsonon-cor,micicor,macinon-cor,mac (a) Stage I cor,mac non-cor,mac non-cor,mac -cor,mac=IRsoil Corroding Non-Corroding Current density, i Current density, i                         ic,Ox        icor,mic    ia,Fecor,mic                       ia,Zn  inon-cor,mic  IRsonon-cor,mac -cor,mac=IRsoil non-cor,micicor,macinon-cor, mac (b) Stage II cor,mac non-cor,mac Corroding Non-Corroding Current density, i Current density, i                           ic,Ox   icor,mic   cor,mic          ia,Fe  inon-cor,mic  ic,Ox IRsonon-cor,mac -cor,mac=IRsoilnon-cor,micicor,macinon-cor, (c) Stage III cor,manon-cor,mac Corroding Non-Corroding Potential,  Potential,  Potential,  Potential,  Potential,    1299.1.1 STAGE 1: CORROSION OF ZINC   At stage 1, the microcell corrosion current density of zinc in the corroding site, icor,mic (A/m2), as illustrated in Figure 9-2(a), can be easily determined from the exchange current density, Tafel slope, and the equilibrium potential of the electrochemical reactions using the polarization and the mix potential theories (Jones, 1995):   Anodes polarize through activation polarization mechanism as follows:  ZnaoaZnaoZnaZna iiEE,,,,, log        (9.1)  where ZnaE , (V) is the anodic corrosion potential of zinc, o ZnaE ,  (V) is the equilibrium potential of the anodic reaction of zinc, a,Zn  is the anodic Tafel slope (V/dec), and ioa,Zn (A/m2) is the exchange current density of zinc.   Polarization of cathodic sites are determined by activation and concentration polarization mechanism (due to the limiting effect of oxygen availability around the cathodic sites) such that  cLLcOxoccOxcoOxcOxc iiiFZRTiiEE log303.2log,,,,      (9.2)   where OxcE , (V) is the cathodic corrosion potential, o OxcE ,  (V) is the equilibrium potential of the cathodic reaction, c,Ox  is the cathodic Tafel slope (V/dec), ioc,Ox (A/m2) is the exchange current Activation Activation Concentration   130density of the cathodic reaction, iL (A/m2) is the limiting current density,  R (≈8.314 J/(mole.K)) is the universal gas constant, F (96500 C/mole) is the Faraday’s constant,  T (oK) is temperature, and Zc is number of electrons that are involved in the cathodic reaction. The limiting current density, iL (A/m2), is a measure of oxygen availability around the cathodic sites on the reinforcement surface (Bohni, 2005).  At each point on the reinforcement surface, the equilibrium condition is reached when the rates of anodic and cathodic reactions, ia (A/m2) and ic (A/m2), respectively, become equal to each other and to the microcell corrosion current density, icor,mic.  The effect of soil resistivity (i.e. IR drop) in microcell corrosion activity does not need to be considered because the distance between the anodic and cathodic sites on the steel surface is very small (Gulikers, 2005) and, consequently, the potentials of anode and cathode are equal. Therefore, the microcell corrosion current density in the corroding site, icor,mic , on the zinc surface in the soil can be calculated by numerically solving the following nonlinear equation (Figure 9-2a) obtained from Equations. 9.1 and 9.2: 0log303.2loglog,,,,,,,,, miccorLLcZnocmiccorZnaOxocmiccorOxcoZncoOxc iiiFZRTiiiiEE        (9.3) In the non-corroding site, the polarization curve of the anodic microcell reaction is not just the function of the activation polarization equation given in Equation 10.1 due to the formation of the zinc oxide products on the surface of steel but follows an s-shape behavior as shown in Figure 9-2a (Revie & Uhlig, 2008). This polarization behavior can be defined through the following modified anodic polarization equation (Revie & Uhlig, 2008): ZnfaZnaoaZnaoZnaZna RiiiEE ,,,,,, log                (9.4)   131where ZnaE ,  (V) is the anodic corrosion potential of zinc in the non-corroding site, Rf,Zn  (Ω.m2) is the electrical resistance of the zinc corrosion product, and ia (A/m2) is the anodic current density of zinc in the non-corroding site. The polarization curve of the cathodic microcell activity in the non-corroding site still follows the same equation as the cathodic polarization equation (Equation 9.2); therefore, the following equation can be written:      cLLcOxoccOxcoOxcOxc iiiFZRTiiEE log303.2log,,,,         (9.5)  where OxcE ,  (V) and ic (A/m2) are the cathodic corrosion potential and the cathodic microcell corrosion density in the non-corroding site, respectively.   At the equilibrium condition, the rates of anodic and cathodic reactions, ia (A/m2) and ic (A/m2), respectively, at any node on the surface of reinforcement are equal to each other and to the microcell corrosion current density, inon-cor,mic.  Thus, 0log303.2loglog ,,,,,,,,,,,  ZnfmiccornonmiccornonLLcZnoamiccornonZnaOxocmiccornonOxcoZncoOxc RiiiiFZRTiiiiEE   (9.6) The simultaneous solution of equations 9.3 and 9.6 provides the microcell corrosion current density in the non-corroding zone (inon-cor,mic) and corroding zone (icor,mic) of the reinforcement. By substituting these values in equations 9.1 and 9.4, the microcell corrosion potential in the non-corroding miccornonE , ) and the corroding site ( miccorE , ) are determined. Since the corrosion potential of the corroding site ( miccorE , ) is not the same as that of the non-corroding site ( miccornonE , ) (i.e., Activation Concentration   132miccorE , < miccornonE , ), the macrocell corrosion current is produced to stabilize the imbalanced situation such that soilmiccormiccornon IREE  ,,         (9.7)  where Enon-cor,mac (V) and Ecor,mac (V) are the macrocell corrosion potentials of the non-corroding site and the corroding site, respectively, Rsoil (Ω) is the soil resistance, and I (A) is macrocell corrosion current which is equal to (icor,mac xAcor) or (inon-cor,mac xAnon-cor) where Acor and Anon-cor are the area of the corroding and non-corroding sites, respectively.  9.1.2 STAGE 2: COMBINED CORROSION OF STEEL AND ZINC  At Stage 2, in the corroding zone, when the zinc cover is removed from the surface of the reinforcement, the underlying steel is exposed to the contaminated soil and, therefore, instead of zinc, steel is involved in the anodic activity. The microcell corrosion current density of iron in the corroding site, icor,mic (A/m2), as illustrated in Figure 9-2b, can be easily calculated using a similar procedure explained in section 9.1.1. :   Anodes polarize through activation polarization such that  FeaoaZnaoFeaFea iiEE,,,,, log      (9.8)  where FeaE ,  (V) is the anodic corrosion potential of iron, o FeaE ,  (V) is the equilibrium potential of the anodic reaction of iron, a,Fe  is the anodic Tafel slope (V/dec), and ioa,Fe (A/m2) is the exchange current density of iron.   Activation   133The cathode polarization equation is the same as the one defined in Equation 10.2. When the equilibrium is reached, the rates of anodic and cathodic reactions, ia (A/m2) and ic (A/m2), respectively, are equal to the microcell corrosion current density, icor,mic. Using Equations 9.2 and 9.8, the microcell corrosion current density, icor,mic, on the steel surface in the corroding zone of the soil is determined by numerically solving the following nonlinear equation (Figure 9-2 (b)):  0log303.2loglog,,,,,,,,, miccorLLcZnoamiccorZnaOxocmiccorOxcoZnaoOxc iiiFZRTiiiiEE        (9.9) In the non-corroding site, the polarization curve of the anodic microcell reaction is the same as the one introduced in Equation 9.4, as shown in Figure 9-2 (b) (Revie & Uhlig, 2008) and the polarization curve of the cathodic microcell activity still follows the same equation as the cathodic polarization equation (Equation 9.2). On the surface of reinforcement at equilibrium condition, the rates of anodic and cathodic reactions, ia (A/m2) and ic (A/m2), respectively, become identical to the microcell corrosion current density, inon-cor,mic.  Hence, 0log303.2loglog ,,,,,,,,,,,  ZnfmiccornonmiccornonLLcZnoamiccornonZnaOxocmiccornonOxcoZnaoOxc RiiiiFZRTiiiiEE    (9.10) solving Equations. 9.2, 9.4, 9.9, and 9.10, the microcell corrosion current density in the non-corroding zone (inon-cor,mic) and corroding zone (icor,mic), as well as the microcell corrosion potential of the non-corroding site (Enon-cor,mic) and corroding site (Ecor,mic) are calculated. Since the corrosion potential of the corroding site (Ecor,mic) is not equal to that of the non-corroding site (Enon-cor,mic) (i.e., Ecor,mic < Enon-cor,mic), the macrocell corrosion current is generated to balance the unequal situation (Equation 9.7).   1349.1.3 STAGE 3: CORROSION OF STEEL  At Stage 3, when the zinc cover is widely removed from the surface of the reinforcement in both the corroding zone and the non-corroding zone, the underlying steel becomes exposed to the soil and, therefore, instead of zinc, the underlying steel is involved in the anodic reaction at both the corroding and non-corroding sites.   The microcell corrosion current density of iron in the corroding site, icor,mic (A/m2), as illustrated in Figure 9-2c), can be easily calculated from the exchange current density following the polarization and the mix potential theories (Jones, 1995). Anode polarizes through the activation polarization equation as defined in equation 9.8. The cathode polarization equation is the same as the one defined in equation 9.2. When the equilibrium is obtained, the rates of anodic and cathodic reactions, ia (A/m2) and ic (A/m2) will be the same as the microcell corrosion current density, icor,mic , respectively.  Therefore, the microcell corrosion current density, icor,mic, on the steel surface in the soil can be obtained by numerically solving equation 9.9.  In the non-corroding site, the polarization curve of the anodic microcell reaction takes on an s-shape behavior, as shown in Figure 9-2c) (Revie & Uhlig, 2008). Therefore, the polarization reaction follows the modified anodic polarization equation as follows (Maslehuddin et al., 2007):  0log ,,,,,  FefaFeoaaFeaoFeaFea RiiiEE     (9.11) The polarization curve of the cathodic microcell activity still follows the same equation as the cathodic polarization equation (Equation 9.2). At the equilibrium condition, similar to the procedure   135described above, the microcell corrosion current density, inon-cor,mic , is identical to the rates of anodic (ia (A/m2)) and cathodic reactions (ic (A/m2)).  Thus, 0log303.2loglog ,,,,,,,,,,,  FefmiccornonmiccornonLLcFeoamiccornonFeaOxocmiccornonOxcoFeaoOxc RiiiiFZRTiiiiEE (9.12)  by solving Equations 9.9 and 9.12, the microcell corrosion current density in the non-corroding site (inon-cor,mic) and corroding site (icor,mic), as well as the microcell corrosion potential of the non-corroding site (Enon-cor,mic) and corroding site (Ecor,mic) are determined. Since the corrosion potential of the corroding and non-corroding site is different, the macrocell corrosion current is produced to balance the situation (Equation 9.7).  9.2 TWO-DIMENSIONAL NUMERICAL SOLUTION Although the one-dimensional simplification explained above is useful to theoretically describe the concept of microcell and macrocell corrosion of galvanized steel reinforcement in soil, the solution of the real problem is illustrated in Figure 9-1 needs to be conducted in the two-dimensional domain and it is not possible to define a closed-form solution. As a result, a nonlinear finite element technique is used in this study to solve the governing differential equation of the potential distribution in the domain of the problem (i.e., soil), as illustrated in Figure 9-3, and to calculate the corrosion current densities on the surface of galvanized steel reinforcement.         136                                  Figure 9-3 - Schematic illustration of the domain, corroding and non-corroding zones and corresponding  boundary conditions   The governing equation for electric potential distribution is in the form of a Laplacian differential equation as follows:  01   Esoil    (9.13) where E (V) is the electrical potential and  soil  Ω.m) is the soil resistivity. Once the nodal  potentials and the potential gradients are determined, the current density on the galvanized steel   Corroding site boundary icorcicoraicor EEE ,,   Contaminated d= Soil thickness Uncontaminated 0nECl - Non-corroding site boundary  icornoncicornonaicornon EEE   ,,Cl - Cl - Undefined node  137surface at node i in both the corroding and non-corroding sites (i.e., icor and inon-cor) is calculated using Ohm’s law such that  nEiinsoilicor 1      (9.14) And nEiinsoilicornon 1     (9.15) where  soil  Ω.m) is the soil resistivity and   n is the direction normal to the equipotential lines.  At each node of the boundary on the reinforcement surface, two conditions should be satisfied. In the non-corroding site, the anodic and the cathodic potentials need to be equal such that   Einon-cor = Eia,Zn= Eic,ox    (Stage I- Figure 9-2 Figure 9-2a), from Equations 9.4 and 9.5)      Einon-cor = Eia,Zn= Eic,ox     (Stage II- Figure 9-2b), from Equations 9.4 and 9.5)    Einon-cor = Eia,Fe= Eic,ox      (Stage III- Figure 9-2c), from Equations 9.11 and 9.5) These values are used as Dirichlet boundary conditions at each node on the surface of the reinforcement in the non-corroding site.     In the corroding site, the anodic and the cathodic potentials (given by equations. 9.7 and 9.8, respectively) are also identical to each other. Eicor = Eia,Zn= Eic,ox     (Stage I- Figure 9-2a), from Equations. 10.1 and 9.2)      Eicor = Eia,Fe= Eic,ox      (Stage II- Figure 9-2b), from Equations. 9.8 and 8.2)    Eicor = Eia,Fe= Eic,ox       (Stage III- Figure 9-2c) , from Equations. 9.8 and 9.2) These values are applied as Dirichlet boundary conditions to the nodes on the surface of the reinforcement in the corroding site.    138  In the non-corroding site, the current density, inon-cor (A/m2) at each node is equal to the difference of anodic and cathodic current densities of the node as defined by: , ,i i inon cor c Ox a Zni i i                          (Stage 1- Figure 9-2 a) , ,i i inon cor c Ox a Zni i i                (Stage 2- Figure 9-2 b) , ,i i inon cor c Ox a Fei i i                (Stage 3- Figure 9-2 c)  and in the corroding site the current density, icor (A/m2), is calculated as follows: , ,i i icor c Ox a Zni i i                (Stage 1- Figure 9-2 a) , ,i i icor c Ox a Fei i i                (Stage 2- Figure 9-2 b) , ,i i icor c Ox a Fei i i                (Stage 3- Figure 9-2c)  The corrosion current density at each node on the reinforcement in the corroding site, Ecor, and also in the non-corroding site, Enon-cor , is determined by simultaneously solving equations 9.13-9.19.  During the solution of equation 9.13, the electrical neutrality of the domain is enforced. Since the nature of the boundary conditions is nonlinear, equation 9.13 needs to be solved by a nonlinear solution algorithm with an appropriate iteration technique. In this study, the modified direct iteration technique (i.e. a relaxation algorithm) is implemented to reach convergence. Nevertheless, achieving the convergence by this method can be complicated when the resistivity of the soil is low.  The details of the nonlinear solution procedure, the finite element algorithm and the numerical difficulties associated with the convergence of the solution, can be obtained from previously published work (Pour-Ghaz et al., 2007, Ge, 2006, Ghods et al., 2007) therefore, they will not be described in this chapter.    139As illustrated in Figure 9-3, the analysis is carried out in a domain with a length of 300 mm; the domain is discretized by triangular finite elements. The mesh size is optimized after a preliminary sensitivity analysis carried out on the different mesh types. The optimal mesh is defined as a mesh that sets up a balance between accuracy and numerical efficiency. The values of the constant parameters of the simulations (e.g. standard potential, exchange current densities, Tafel slopes, and anode-to-cathode ratio) are presented in Table 9-1. These constants were selected to represent typical values reported in the literature for each stage of galvanized steel corrosion in soil (Zhang, 1996, Armstrong & Bell, 1974, Stankovic et al., 2003, Chung & Kim, 2011, Sakairi et al., 2008, Pilbáth & Sziráki, 2008, Li et al., 2008).                      Table 9-1-The values of the parameters used in the model   Parameter Value Cathodic exchange current density of oxygen (ioc,Ox) 0.00001 A/m2 Anodic exchange current density of iron (ioa,Fe) 0.0003 A/m2 Anodic exchange current density of zinc (ioa,Zn) 0.001 A/m2 Cathodic standard potential of oxygen ( OxcE , ) 0.16 V Anodic standard potential of iron ( o FeaE , ) -0.78 V Anodic standard potential of zinc ( o ZnaE , ) -1.007 V Passive film electrical resistance of iron (Rf,Fe) 500 Ω.m Passive film electrical resistance of zinc (Rf,Zn) 5  Ω.m Cathodic Tafel slope of oxygen (βc,Ox) -0.180 V/dec Anodic Tafel slope of iron (βa,Fe) 0.090 V/dec Anodic Tafel slope of zinc (βa,Zn) 0.120 V/dec Ratio of anode area to cathode area 0.1   1409.3 EFFECT OF SOIL PROPERTIES  9.3.1 EFFECT OF SOIL RESISTIVITY  The resistivity of soil is a critical parameter that significantly affects the corrosion rate of galvanized steel, mainly because ion transfer is directly related to the resistivity of the porous media. Figure 9-4 shows the sensitivity of the model to resistivity changes at two different pH levels: pH=6 and pH=10. In general, increase in soil resistivity reduces the corrosion rate at all three stages of galvanized steel corrosion at all levels of pH. For all resistivities, the corrosion rate of stage 3 is larger than that of stage 1 and stage 2 and the corrosion rate of stage 2 is the lowest one; i.e., i Stage 2 < i Stage 1 < i Stage 3.  This behavior is consistent with the experimental results reported by other researchers (Elias, 2000, Anonymous, Yadav et al., 2004a, El-Mahdy et al., 2000, Yadav et al., 2004b).   Resistivity appears to have a greater effect on the third stage of the corrosion process and the least effect on the second stage.  At the pH of 10, corrosion rates for stage 2 and stage 3 are close together with an average difference of 8%, while at the pH of 6 the average difference is 38% which is substantially larger than the one for pH=10.    141a) pH = 6Resistivity cm0 2000 4000 6000 8000 10000 12000 14000 16000i corr uA/cm20123456Stage IStage IIStage III b) pH = 10Resistivity cm0 2000 4000 6000 8000 10000 12000 14000 16000i corr uA/cm20.00.51.01.52.02.53.03.5Stage IStage IIStage III Figure 9-4- The effect of soil resistivity on the corrosion rate of galvanized steel at two levels of pH: a) pH=6, b) pH=10    142Additionally, Figure 9-5 shows the effect of changing soil resistivity from 2,500 Ω.cm to 15,000 Ω.cm on the corrosion rate at three different pH values of 6, 10, and 13 on each of the different corrosion stages separately. According to the AASHTO specification (Elias, 1990), the lower and upper values of the chosen resistivity respectively resemble an aggressive and non-aggressive environment condition for the soil around the reinforcement. The rest of the input parameters were fixed to study the effect of oxygen concentration; the limiting current density was set at 15 µA/cm2 and temperature at 25ºC. In this section, due to the fact that moisture content directly affects resistivity in the proposed model, both the reference moisture content and the soil moisture content were set at 3 wt%, at a temperature of 25ºC. For more details on how soil moisture content affects soil resistivity on the model, please refer to previous publications (Padilla et al., 2013). A low value for the limiting current density (i.e., iL = 15 µA /cm2) was selected to simulate the condition where the concentration of oxygen around the reinforcement is relatively low and the rate of the corrosion process is controlled by the oxygen diffusion which would be the case in soil reinforcements buried in compacted soil. The corrosion literature is somewhat ambiguous on whether correlations exist between the corrosion rate and soil resistivity, independent of variations in soil moisture (Cole & Marney, 2012), some previous experiment and field results support that the corrosion rate is directly related to soil resistivity. Kasahara and Kajimaya found a direct relationship between resistance and corrosion rate (Kasahara & Kajiyama, 1983), and Alamilla et al. also found a high correlation between soil resistivity and corrosion (Alamilla et al., 2009a); but on the other hand, Barbalat et al. (2012) concluded that the influence of soil resistivity is negligible when studying the corrosion behavior of steel coupons buried in soil, but that can perhaps be attributed to the specific soils used in the study.    143                      Figure 9-5- The effect of soil pH in the three stages of corrosion rate of galvanized steel Figure 9-5 suggests that there is a direct correlation between corrosion rate and soil resistivity in all three different stages of corrosion at the three different values for pH. As shown in Figure 9-pH5 6 7 8 9 10 11 12 13 14corrosion rate (m/year)1011141516 2500 穋m7500 穋m15000 穋ma) Stage 1pH5 6 7 8 9 10 11 12 13 14corrosion rate (m/year)56789 2500 穋m7500 穋m15000 穋mb) Stage 2pH5 6 7 8 9 10 11 12 13 14corrosion rate (m/year)68101214161820 2500 穋m7500 穋m15000 穋mc) Stage 3  1445c, the overall effect of soil resistivity is higher at stage 3 of corrosion, which corresponds to the corrosion after the zinc coating has been depleted. At stage 3, a decrease in corrosion rate from 34%-54% is observed when soil resistivity values are increased from 2,500 Ω.cm to 15,000 Ω.cm, compared to a decrease of 35% to 40% for stage 1 in Figure 9-5a, which corresponds to an intact zinc cover,  and 10% to 30% for stage 2 in Figure 9-5b.  In addition, at low pH values (pH of 6), the effect of soil resistivity is more pronounced than at higher pH values (pH of 13), which suggests that the overall effect of soil resistivity is a much more important factor in acidic soils than in alkaline soils. These results suggest that increasing the zinc cover is highly recommended at lower soil resistivity values and lower pH.  9.3.2 EFFECT OF SOIL PH  In the model, the pH effect is simply considered on the cathodic standard potential of oxygen by (Revie & Uhlig, 2008): )14(0592.0,,  pHEE oOxcpHOxc      (9.16) where pHOxcE , is the modified cathodic potential of oxygen, o OxcE ,  is the cathodic standard potential of oxygen, and pH indicates the soil alkalinity.  Figure 9-6 shows the effect of increasing pH from 6 to 13 on the corrosion rate at two different resistivity values. The rest of the input parameters were fixed to study the effect of oxygen concentration; the limiting current density was set at 15 µA/cm2, and temperature at 25ºC. Each of the figures presents the results of varying the pH at three distinct resistivity values: 2,500 Ω.cm, 7,500 Ω.cm, and 15,000 Ω.cm. The resistivity values were set at a reference moisture content of 20 wt% and at a temperature of 25ºC.    145 As pH increases, the corrosion rate values of all three stages of corrosion decreases. It appears that stage 1 has the least sensibility to changes in pH with an overall decrease in the corrosion rate of 4% when resistivity was set to 2,500 Ω.cm, and less than 1% when resistivity was set to 15,000 Ω.cm. Stage 3, which is believed to be mostly governed by the electrochemical properties of steel, on the other hand, appeared to be more sensitive to pH changes with an overall decrease in the corrosion rate of 33% when resistivity was set to 2,500 Ω.cm, and less than 26% when resistivity was set to 15,000 Ω.cm.   This behavior is in agreement with each of the published Potential-pH diagrams for the zinc-water system at 25°C and the steel-water system at 25°C (Pourbaix, 1974). According to the Potential-pH diagram, zinc forms a stable corrosion product from pH of around 6 to up to a pH of around 14, which allows zinc to behave stable in the studied pH ranges (Beverskog & Puigdomenech, 1997). The region of stability for steel, on the other hand, is highly dependent on the potential at the studied pH ranges and, according to Beverskog and Puigdomenech (1996), at 25°C reaches the stability region at a relatively high pH.           146                                                    Figure 9-6- The effect of resistivity on the three stages of corrosion rate of galvanized steel at three different pH, a) pH 6, b) pH 10, and c) pH 13 Resistivity cm220 40 60 80 100 120 140 160corrosion rate (m/year)102030405060 pH 6pH 8pH 13c) Stage 3Resistivity cm220 40 60 80 100 120 140 160corrosion rate (m/year)68101214pH 6pH 10pH 13b) Stage 2Resistivity cm220 40 60 80 100 120 140 160corrosion rate (m/year)101520253035 pH 6pH 10pH 13a) Stage 1  1479.3.3 EFFECT OF TEMPERATURE The relationship between the soil resistivity and temperature has been previously proposed in the form of (Chaker, 1990): 00248.5248.5t tTT      (9.17)where ρt (Ω.cm) is the calculated soil resistivity at temperature and Tt (K) and ρ0 (Ω.cm) is the measured soil resistivity at temperature, To (K). This simple equation is used in the model to consider the effect of temperature on the corrosion rate of galvanized steel.  Figure 9-7 shows the effect of temperature at two different resistivity values of 3,000 and 7,500 Ω.cm. The trend for all three different stages is similar: as temperature increases, the corrosion rate increases. When temperature increases from 0°C to 40°C, in general, the corrosion rate increases approximately two times for all three distinct corrosion stages. However, in both resistivities, the largest effect is on the stage 3 corrosion rate and the smallest one is on the stage 2 corrosion rate.    148a) 3000 cmTemperature oC0 10 20 30 40 50i corr uA/cm20123456Stage IStage IIStage III b) 7500 cmTemperature oC0 10 20 30 40 50i corr uA/cm20.00.51.01.52.02.53.0Stage IStage IIStage III Figure 9-7 - The effect of temperature on the corrosion rate of galvanized steel at two different resistivities: (a) Resistivity = 3,000 Ω.cm, (b) Resistivity=7,500 Ω.cm        1499.3.4 EFFECT OF MOISTURE CONTENT  The relationship between the soil resistivity and moisture content has been previously proposed in the form of (Samouëlian et al., 2005): 220.00143 0.01560.00143 0.0156rm rm        (9.18)where ρm (Ω.cm) is the calculated soil resistivity at the volumetric moisture content of ρm (%) and ρr (Ω.cm) is the measured soil resistivity at the volumetric moisture content of ρr (%). This equation is used in the model to consider the effect of soil moisture content on the corrosion rate of galvanized steel. The volumetric moisture content (θ can be calculated by multiplying the gravimetric moisture content (u) by the bulk specific gravity (γ) of the soil as follows: u    (9.19) Water content in soil is, by itself, a very complex area of study with competing approaches and ongoing research focused on measuring and estimating average water content in soil (Vachaud et al., 1985, Hendrickx et al., 2001). Soil water content is both spatially and temporally variable and there are many factor that influence water intake to soil such as: soil type and texture, initial water content, surface sealing, presence of cracks, temperature, terrain slope, and drainage to mention a few; however, in order to simplify this study, the corrosion rate of galvanized steel was calculated for set values of the moisture content ranging from 5 to 35 wt %. Figure 9-8 shows the effect of moisture content at three different soil resistivity values: 2,500 Ω·cm, 7,500 Ω·cm, and 15,000 Ω·cm. The rest of the input parameters were fixed to isolate the effect of moisture content; the pH was set at 7.5, and temperature at 25ºC, and the limiting current density at 15 µA/cm2. Moisture content has a direct effect on the medium resistivity so, as expected,   150as moisture content increases, so does the corrosion rate. By looking at Figure 9-8a, Figure 9-8b and Figure 9-8c, it is possible to notice that the effect of moisture content is similar for all the three different stages of corrosion, being higher for stage 3, followed by stage 1. Even though the corrosion rates reported for stage 2 are overall lower than in the other two stages, it is possible to notice that the increase from the corrosion rate when the moisture content ranges from 30% to 35% is considerable. It is also possible to notice in Figure 9-8 that the effect of moisture content is obvious in highly corrosive environments (soil resistivity at ≤ 2500 Ω·cm). When the soil resistivity is set at 7,500 Ω·cm and 15,000 Ω·cm the increase in corrosion rate as moisture content increases is less pronounced, with an average difference of 54% for 7,500 Ω·cm, and 44% for 15,000 Ω·cm; however when the soil resistivity was set at 2,500 Ω·cm, an average increase of 72% in corrosion rate was reported. According to these results, moisture content is a critical parameter to take into account when the conditions of the structure can be considered aggressive in terms of corrosiveness.   With respect to the cyclic change of moisture content (i.e. dry/wet cycling), it is important to note that this effect is not considered in the developed model. As observed in some experimental studies (Stephenson et al., 2009), the accumulation of corrosion products on the surface of metal due to wet/dry cycling may limit the access of oxygen/water which may lead to a reduction in corrosion rate. This effect is not directly incorporated into the model. However, the direct effect of oxygen availability on the surface of galvanized steel can be captured through the limiting current density parameter (iL) introduced in section 10.2.1.       151                      Figure 9-8 - The effect of moisture content on the three stages of corrosion rate of galvanized steel moisture content (wt %)0 10 20 30 40corrosion rate (m/year)010203040502500 穋m7500 穋m15000 穋ma) Stage 1moisture content (wt %)0 10 20 30 40corrosion rate (m/year)0510152025 2500 穋m7500 穋m15000 穋mb) Stage 2moisture content (wt %)0 10 20 30 40corrosion rate (m/year)010203040506070 2500 穋m7500 穋m15000 穋mc) Stage 3  152 When the soil moisture content is close to the saturation level (i.e. 100% flooded soil), iL becomes very small and the corrosion process is therefore controlled by an oxygen diffusion mechanism. There would be a peak corrosion rate for an optimum moisture content beyond which the corrosion rate starts to drop. Therefore, the effect of the fully saturated soil condition on the corrosion rate can be seen in this model by reducing the value of the limiting current density.  9.3.5 EFFECT OF OXYGEN CONTENT  When either the amount of oxygen or the corrosive medium conductivity increases, it is expected that the corrosion process will be enhanced. A number of different factors can promote the conditions where we have a variable oxygen access on a structure: partial immersion of the structure in an aquatic/marine environment; partial coverage of the reinforcement by concrete in MSE walls, for example, or simply a gradient in the soil compaction (Padilla & Alfantazi, 2012b). Previous work revealed that an increased concentration of oxygen appears to have a greater effect on the corrosion rate than the presence of corrosive agents in the environment (Padilla & Alfantazi, 2012b). Furthermore, differential oxygen access is expected to promote corrosion macrocells with the cathode on the site with the higher oxygen content (Sagüés et al., 2000) and with possible corrosion enhancement near the high stress regions. In this work, the application of the model in prediction of the oxygen access on the corrosion rate of galvanized steel is studied. The effect of oxygen concentration on the corrosion rate is only considered through the limiting current density (iL) and other types of aforementioned effects was neglected for simplicity. The general assumption is that greater oxygen availability will result in an increased limiting current density value, ultimately increasing the corrosion rate.    153Figure 9-9 shows the effect of limiting current density (i.e. oxygen concentration) at three different soil resistivity values: 2,500 Ω·cm, 7,500 Ω·cm, and 15,000 Ω·cm. The corrosion rate of galvanized steel at each stage of corrosion was calculated for various values of limiting current densities ranging from 20 to 70 µA/cm2.  The rest of the input parameters were fixed to study the effect of oxygen concentration; the pH was set to 7.5, moisture content to 20 wt%, and temperature to 25ºC. Comparing Figure 9-9a, Figure 9-9b, and Figure 9-9c, one can conclude that, according to the proposed model, the first stage of corrosion is slightly more sensitive to oxygen concentration variation that the other two stages. In both stage 1 and stage 2, the corrosion rate increases slightly until reaching a plateau at a limiting current density of 40 µA/cm2.    It is also possible to notice in Figure 9-9a that the effect of oxygen concentration seems to be more pronounced in highly corrosive environments (e.g. low soil resistivity). The differences in corrosion rate, at all different stages, when the soil resistivity is set at 7,500 Ω·cm, and 15,000 Ω·cm seem to be insignificant. Figure 9-9c shows that increase in the limiting current density values (i.e. increasing oxygen concentration) has little effect on the third stage of corrosion of galvanized steel. However, based on the results obtained from the proposed model, it seems that corrosion rates for stage 1 and stage 3 are in good agreement with the limits set by the AASHTO model (15µm/year, 4 µm/year, and 12 µm/year for stage 1, stage 2 and stage 3, respectively), being only higher for stage 2.        154                          Figure 9-9- The effect of limiting current density (i.e., oxygen concentration) on the three stages of corrosion rate of galvanized steel  iL (A/cm2)20 30 40 50 60 70corrosion rate (m/year)020406080 2500 穋m7500 穋m15000 穋mc) Stage 3iL (A/cm2)20 30 40 50 60 70corrosion rate (m/year)010203040506070 2500 穋m7500 穋m15000 穋mb) Stage 2iL (A/cm2)20 30 40 50 60 70corrosion rate (m/year)10203040506070 2500 穋m7500 穋m15000 穋ma) Stage 1  155 The presented model is a practical tool for engineers since it is able to estimate corrosion damage evolution with good approximation, while the variables can be easily adjusted to consider any specific soil environment and climatic conditions.  Therefore, the model can be practically used to determine optimum zinc cover thickness and to estimate the remaining service life of the existing MSE walls. In the next chapter, the model is compared to other widely used and accepted models in the industry. Additionally, an example of the model capacity to more accurately calculate the corrosion rate taking into account changing conditions is given.  9.4 SUMMARY In this chapter, a numerical model was introduced. The model can be used to calculate the corrosion rate of galvanized steel in soil at three different stages of corrosion by considering key soil corrosion parameters such as resistivity, temperature, moisture content, pH, and oxygen availability. Results showed that the proposed model is in agreement with widely accepted empirical models, and can thus be used for the service-life design and risk assessment of MSE walls and to determine the optimum zinc cover thickness. The presented model is quite practical for engineering applications since it is able to estimate corrosion damage evolution with good approximation and variables can be easily adjusted in the model to consider any specific soil environment.       15610 VALIDATION AND APPLICATION OF THE PRACTICAL THREE - STAGE CORROSION BEHAVIOR OF GALVANIZED STEEL IN  SOIL 10.1 COMPARISON TO MODELS ACCEPTED MODELS  In order to verify the results, the proposed model is compared with four previously published and widely accepted models. The models used for comparison are the Darbin/Romanoff Model, the Stuttgart Model, the Caltrans Model, and finally, the AASHTO Model. These four models were developed in different periods of time, reflecting empirical data obtained from extensive monitoring and sampling from as far back as 50 years ago. Regardless of when these models were developed, the methodology followed for their development is similar. All of them are empirical models with two main variables: zinc coating thickness and service life. The details on how these models were developed, as well as the formulas used, can be seen in the respective publications (Fishman & Withiam, 2011, Elias, 1990, Rehm, 1980, Jackura et al., 1987, Darbin et al., 1988).  The Darbin/Romanoff model was originated based on a very comprehensive database collected on a National Bureau of Standards (NBS) study and later elaborated and improved on to better estimate corrosion of MSE walls by Darbin et al.  (1988) and by Elias (1990), and is applicable for the conditions generally found in MSE applications with a soil resistivity greater than 1,000 Ω-cm. The rest of the models are linearized forms of this model. The Stuttgart Model was proposed by Rehm in 1980 (Rehm, 1980), in which the change of the corrosion rate between the three different stages is taken into account by assuming that the corrosion of zinc in usually greater in the first 2 to 4 years, followed by a reduced rate and, finally, by a corrosion rate governed by the bare steel consumption. Rehm proposed two different models depending on the soil conditions surrounding the MSE wall. The first model is used when the backfill in the MSE wall meets specification and a   157second model when it does not, thus, it is considered corrosive. Both the Darbin/Romanoff and the Stuttgart Models contribute to the basis of the AASHTO Model, in which additional data were included from laboratory tests. This model is similar to the Stuttgart model with only two main differences: first, it requires the backfill to have a resistivity greater than 3,000 Ω.cm, and second, the corrosion rate of zinc after the second year of service life is twice the value considered in the Stuttgart model, thus the AASHTO model is considered to be conservative. Finally, based on the work by Jackura et al. (1987), Caltrans proposed a model for a wider range of backfill conditions: a soil resistivity greater than 2,000 Ω.cm, a pH between 5.5 and 10, and chloride and sulphate concentrations of 250 ppm and 500 ppm, respectively, and considers that the zinc coating will be depleted after 10 years of service life as opposed to the 15 years proposed in the AASHTO model (Elias, 2000).  Figure 10-1 shows the comparison of results from the AASHTO guidelines with results from the proposed model at mildly corrosive and highly corrosive conditions, defined as the lower and upper bond, respectively. The conditions are described in Table 10-1. As seen in Figure 10-1, the results of the proposed model are in agreement, at the three different corrosion stages, with the results of the AASHTO model. Both the upper and lower bounds successfully encompass the results provided by the AASHTO model, which means that the proposed model is able to provide a service life estimate for MSE walls with galvanized steel reinforcement in accordance with the AASHTO model. Additionally, the proposed model is able to predict corrosion behavior outside the ranges considered in the AASHTO model, providing a greater degree of flexibility and control for field applications.   158Stage I Stage II Stage IIIi corr uA/cm20.00.51.01.52.02.53.0Upper Bond  - Highly CorrosiveAASHTO Model at specificationsLower Bond - Non corrosive Figure 10-1- Comparison of results from the AASHTO guidelines with results from the proposed model at mildly corrosive and highly corrosive conditions   pH ResistivityΩ-m iL A/m2Corrosion rate icorr µA/cm2 Stage 1 Stage 2 Stage 3 Upper Bound 6 5,000 0.2 1.68 0.83 2.68 AASHTO Model 5-10 >3,000 N/A 1.0 0.34 1.03 Lower Bound 10 15,000 0.2 0.58 0.21 0.67 Table 10-1- Upper and lower-bond rate of metal loss obtained from the proposed model in comparison with the AASHTO model  Figure 10-2 shows a cumulative total metal loss comparison of the proposed model with those obtained from the previously mentioned models during the first 75 years of service life for an MSE   159wall. The total metal loss includes the three stages of corrosion, therefore, the represented metal loss in Figure 10-2 includes that of the zinc coating and steel reinforcement.  Time (years)0 20 40 60Cumulative Metal Loss (µm)02004006008001000 Caltrans ModelDarbin ModelAASHTO ModelStuttgart ModelProposed Model Figure 10-2 -Comparison of the metal loss calculated from the proposed model and other models in the literature during the 75 year service life of a MSE wall   The corrosion rates used for each stage are as follows: the Stuttgart model for corrosive conditions (17µm/year, 2 µm/year, and 12 µm/year for Stage I, Stage II and Stage III, respectively), the AASHTO model (15µm/year, 4 µm/year, and 12 µm/year for Stage I, Stage II and Stage III, respectively), the Caltrans model (for selected-granular backfill environments with constant K defined as 13 µm/year and C as 30 years) (Fishman & Withiam, 2011), and the Darbin/Romanoff   160model for galvanized reinforcements (with a constant exponent equal to 0.65, a coefficient K of 25 µm/year for soils with a resistivity value greater than 1,000 Ω.cm, and a factor of 2 applied to consider strength loss) (Elias, 1990, Darbin et al., 1988). The corrosion rates obtained from the proposed model were 13.44 µm/year, 5.55 µm/year, and 10.55 µm/year for three stages of corrosion, respectively. These rates were calculated considering an average pH of 7.5, an average resistivity of 7,500 Ω.cm, a temperature of 25ºC, a moisture content of 3 wt% (dry conditions), and a limiting current density of 0.2 A/m2.  For each model, it is possible to see three different slopes in Figure 10-3 (the only exception being the Caltrans model, since the slope change takes place after the 30th year of service life); a closer look into the first 20 years of service life has been included to make these three slopes more evident. These slopes represent the three different corrosion stages. The general assumption in this comparison is that stage I of the corrosion process lasts 2 years, after which the corrosion rate of zinc decreases due to the accumulation of corrosion products and, thus, stage II begins. This assumption is generally accepted and followed in all of the models (Fishman & Withiam, 2011, Elias, 1990, Rehm, 1980, Jackura et al., 1987, Darbin et al., 1988). After the second year of service life, the slope slightly decreases until there is a sharp increase when stage III is reached, attributed to the complete depletion of the zinc coating.  As seen in Figure 10-2, the proposed model is also in agreement with other models, with the results being a little more conservative to those of the Caltrans and the Stuggart models, but less than those of the Darbin model. Of all the compared models, the Caltrans model seems to be the least conservative, yielding the lowest metal cumulative metal loss.   161Time (years)0 5 10 15 20Cumulative Metal Loss (µm)050100150200250300 Caltrans ModelDarbin ModelAASHTO ModelStuttgart ModelProposed Model Figure 10-3 - Comparison of the metal loss calculated from the proposed model and other models in the literature during the first 20 years service life of an MSE wall  The difference in the results may arise due to a fundamental distinction in the nature of the models: while the compared models are empirical, and the estimated corrosion rates given by these models are based on averages taken from a large number of samples with a large variation, the proposed model is theoretical and the estimated corrosion rate is based on a numerical approach. This approximately linear overall model was evaluated for up to 75 years and then successfully compared against four other published, long-term, empirical models. Nevertheless, it is important to consider that recognize that the soil corrosion process is quite complex and probably nonlinear with   162respect to time. Therefore, adding more complexity into this model is essential to enhance the accuracy of the calculated results.  10.2 SERVICE LIFE DESIGN One of the most interesting features of the proposed model is its ability to predict corrosion rates under variable conditions. The purpose of this section is to present an example of how this can be used to calculate the expected corrosion rates in different geo-climatic conditions, which provides an excellent opportunity for a more accurate service life design and important cost savings by using the right amount of zinc-coating thickness and sacrificial thickness depending of the design conditions of the structure. Table 10-2 shows the average monthly temperature for four selected cities. The cities selected for this study were Ottawa, Vancouver, Los Angeles, and Acapulco; each city has a unique temperature profile. City Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecOttawa -10.8 -8.7 -2.5 5.7 13.4 18.3 20.9 19.5 14.3 7.8 1 -7.1Vancouver 3.3 4.8 6.6 9.2 12.5 15.2 17.5 17.6 14.6 10.1 6 10.1Los Angeles 13.8 14.2 14.4 15.6 17.1 18.7 20.6 21.4 21.1 19.3 16.4 13.8Acapulco 26 26 26 26 28 28 28 28 28 28 27 26 Table 10-2- Average Monthly Temperature for fours selected cities   Figure 10-4 shows the effect of temperature on the calculated corrosion rate for each stage of galvanized steel corrosion for four distinct temperature profiles at selected cities: a) Ottawa, b) Vancouver, c) Los Angeles, and d) Acapulco. The rest of the input parameters were fixed to study   163only the effect of temperature; the pH was set to 7.5, moisture content to 10 wt%, and temperature to 25ºC.               Figure 10-4- The effect of temperature on the calculated corrosion rate for each stage of galvanized steel corrosion for four distinct temperature profiles at selected cities: a) Ottawa, b) Vancouver, c) Los Angeles, and d) Acapulco  As expected, temperature has a direct effect on the corrosion rate and the monthly corrosion rates matched the temperature profile of each city; i.e. corrosion rate increases as temperature increases and vice versa. Using this information it is possible to estimate the total yearly metal loss for each temperature profile. Figure 10-5 provides an example of the metal loss during the first 20 years of service life for theoretical structures, in each of the four different temperature profiles.    Montha) OttawaJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Deccorrosion rate (m/year)141819Temperature 癈-15-10-50510152025Stage 1Stage 2Stage 3Average Monthly Temperture     Monthb) VancouverJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Deccorrosion rate (m/year)141819Temperature 癈2468101214161820Stage 1Stage 2Stage 3Average Monthly Temperture       Monthc) Los AngelesJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Deccorrosion rate (m/year)13141819Temperature 癈121416182022Stage 1Stage 2Stage 3Average Monthly Temperture     Monthd) AcapulcoJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Deccorrosion rate (m/year)141819Temperature 癈252627282930Stage 1Stage 2Stage 3Average Monthly Temperture  164Time (years)0 5 10 15 20Cumulative Metal Loss (µm)020406080100120140OttawaVancouverLos AngelesAcapulcoAASHTO ModelStage 1 Stage 2 Stage 3 Figure 10-5- Calculated metal loss for the first 7 years of service life for theoretical structures, in each of the four different temperature profiles compared with the AASHTO model corrosion rates  The three different corrosion stages are delimited using a dotted line in the graph. For all different cases, the duration of the first stage of corrosion was set to two years in which the general assumption is that the electrochemical behavior of the galvanized steel is that of pure zinc since the coating is intact; the second stage of corrosion was then calculated from the second year of corrosion until depletion of the zinc coating, assuming an homogeneous galvanized coating of 86 µm, and the third stage follows from the depletion of the coating. As it is made evident from the Figure 10-5, all the theoretical structures reach the third stage of corrosion between 11 and 12 years. Using these results it is then possible to easily estimate the galvanized coating thickness required to protect the underlying steel after a certain amount of time. For example, for a desired 14 years of life time for   165the galvanized coating the thickness of the zinc required would be around 117 µm in Ottawa, 118 µm in Vancouver, 122 µm in Los Angeles and 128 in Acapulco.    The importance of having accurate corrosion rate estimations, which can be properly calculated according to specific environmental conditions, goes beyond the obvious safety advantages that come from proper sacrificial thickness calculations. The model not only allows for more accurate calculations for corrosion rate and metal loss, but it can also be potentially used for life cycle cost assessment of the structures.   10.3 MODEL VALIDATION There are very few published studies with experimental and field data that can be used to validate the current model. Either some of the studies that have been published do not disclose the data used for the development of the model or the information given is limited. The Darbin/Romanoff model, for example, was originated based on a very comprehensive database collected on a National Bureau of Standards (NBS) study and, later, elaborated on and improved to better estimate the corrosion of MSE walls by Darbin et al. in 1988 and by Elias in 1990 and is applicable for the conditions generally found in MSE applications with a soil resistivity greater than 1,000 Ω-cm (Fishman & Withiam, 2011, Romanoff, 1957, Elias, 1990, Rehm, 1980, Jackura et al., 1987, Darbin et al., 1988). The specific soil properties for each of the readings are not provided in the study.    166corrosion rate (m/year) NBS Field Data0 10 20 30 40corrosion rate ( m/year) Model Calculation010203040f(x)=xModel vs NBS Field Data Figure 10-6 – Model Comparison with average corrosion rates of galvanized steel pipe specimens in soils for 10 years as reported by NBS.   In his book, Corrosion and Electrochemistry of Zinc, Zhang (1996) presents a table containing results from an NBS investigation for the corrosion rates of zinc and galvanized steel in various soils. The information given on that study is the type of soil, soil resistivity, and soil pH. Figure 10-6 shows the comparison of the results presented in the study with results yielded by the NBS study. This study presents field corrosion measurements after 10 years of immersion. The assumptions taken for the comparison were: all specimens reported in the NBS study were on stage 2 of corrosion (which generally takes place between years 2 and 14), the reference moisture content was set at 20% wt, limiting current density a 15 µA/cm2, temperature at 25ºC, and soil moisture content at 10% wt. As made evident in Figure 10-6, the model is in good agreement with the field   167observations. The average absolute variability between the model results and the NBS field observation is 5.8µm/year; however it is important to mention that there are some field observations that present abnormally low corrosion rates when taking into account the reported conditions (i.e. very low soil resistivity).  10.4 SUMMARY In this chapter, a parametric study and an example of an application of the developed model was presented. The model  allows an accurate and dynamic estimation of corrosion damage at each of the corrosion stages of galvanized steel in soil, taking into account both typical parameters that are relevant in soil corrosion and environmental factors. The proposed model can be used to more accurately estimate the needed galvanized coating thickness in different environmental conditions. It is an easy-to-use, practical tool to estimate corrosion damage evolution with good approximation, while the variables can be easily adjusted to consider any specific soil environment and climatic conditions as needed on a case-by-case basis.     16811 CONCLUSIONS The electrochemical tests and analysis presented in this dissertation lead to the following conclusions:  Galvanized steel does not form a passive film under the studied conditions. The zinc cover of the galvanized steel fails to form a non-soluble and protective film under any conditions.  SEM pictures revealed that the corrosion product formed is porous. Additionally, SEM pictures revealed that sulphate nest corrosion products grow preferentially in the vicinity of zinc grain boundaries.  Increasing the concentration of oxygen in the solution appears to have a greater effect on icorr compared to the effect of adding more Na2SO4 to the solution. The impedance data show that zinc resists better against corrosion in de-aerated solutions; as the concentration of oxygen is increased, the corrosion film product resistance drops significantly.  Potentiodynamic polarization data provides enough information to confirm that the potential difference between samples immersed in oxygen-saturated solutions compared with samples immersed in de-aerated and aerated solutions are large enough to confirm the creation of corrosion macrocells.  Decreasing temperature has a considerable effect on the corrosion current of galvanized steel in 3.5 wt% NaCl plus 1 wt% Na2SO4 solutions. Nonetheless, the corrosion rate observed at sub-zero temperatures is still higher than the rate acceptable for galvanized steel reinforced structures. As evidenced both by the higher impedance recorded at 0ºC, and the lesser drop in resistance with time, the galvanized coating appears to perform best when immersed at this temperature.   169 Electrochemical test results showed that the corrosion rate of galvanized steel in the presence of sulphate is considerably higher than the one in the sulphate-free soil condition. When galvanized steel is exposed to MgCl2 or CaCl2 in the sulphate-contaminated soil, it yields the lowest average corrosion rate and thus appears to have the most superior corrosion performance. In a sulphate-free environment, CH3CO2K seems to be a viable de-icing salt option for galvanized steel reinforcement due to its low tendency to cause corrosion, but NaCl has the most aggressive corrosion effect on galvanized steel in the absence of sulphate. Based on the electrochemical test results conducted during the period of 24 hours, the effect of de-icing salts on the corrosion performance of galvanized steel can be ranked as: CH3CO2K (best performance) > NaCl > CaCl2/MgCl2  in the sulphate-free condition; MgCl2/CaCl2 (best performance)  > CH3CO2K > NaCl in the sulphate-contaminated condition.  The highest performance observed in MgCl2 can be attributed to slower corrosion kinetics, a good surface coverage of the corrosion product (surpassed only by samples immersed in CH3CO2K solutions), and higher polarization resistance as studied by EIS.  The numerical model presented in this thesis allows the estimation of corrosion damage at each of the corrosion stages of galvanized steel in soil, taking into account typical parameters that are measured in MSE walls in the field tests.  The model was compared with four of the most widely accepted and used corrosion models for MSE wall applications and the results show that the proposed model effectively predicts the results of the other empirical models.  The presented model is able to estimate corrosion damage evolution with good approximation and variables can be easily adjusted in the model to consider any specific soil environment. Additionally, the results of the model shed the following conclusions:   170o As pH increases, the corrosion rate values of all three stages of corrosion decreases. o The corrosion rate of galvanized steel at each stage of the corrosion process decrease by the increase of the soil resistivity.  o According to the results yielded by the model, limiting current density (oxygen availability) does not have a high impact on the corrosion of galvanized steel. Nonetheless, the first stage of corrosion is slightly more sensitive to oxygen concentration variation than the other two stages. o From all the different studied parameters, soil moisture appeared to have the greater impact on the corrosion rate. As moisture content increases, so does the corrosion rate.  11.1 TECHNICAL CONTRIBUTIONS TO THE FIELD The goal of this work was to develop a comprehensive understanding of the corrosion behavior of hot-dip galvanized steel in aggressive soil conditions and to ultimately contribute towards the improvement of the current construction guide used to predict the design life of Mechanically Stabilized Earth (MSE) walls and facings.   The results presented in chapter 5 to chapter 8 shed information on important variables that affect the corrosion behavior of galvanized steel that had been ignored so far. Chapter 5 showed that variable oxygen concentration can accelerate corrosion and thus should be taken into account in structural design when aggressive conditions are expected. Chapter 6 proved that the corrosion of galvanized steel is still a concern even in colder regions where temperature often drops below the freezing point, so that general assumption taken so far that the AASHTO corrosion rates were adequate is challenged. Chapters 7 and 8 presented, for the first time, an in- depth analysis of the   171combined effect of sulphates naturally found in soil with commonly used de-icing salts on the corrosion behavior of galvanized steel, and even though the choice of de-icer will ultimately depend on temperature and budget constraints, the conclusions presented can be used for future selection of de-icers and for a larger, long-term study of the effect of these de-icers in the field.   Chapters 9 and 10 presented a model that simulates all three stages of the galvanized steel corrosion process. The main advantage of the developed model is its ability to estimate corrosion damage evolution while the variables can be easily adjusted to consider any specific soil environment and climatic conditions. In contrast, other corrosion models can only predict corrosion performance within an established range and assume conditions will remain constant. This model can be used as a powerful engineering tool for the design of galvanized-steel-reinforced structures.        17212 RECOMMENDATIONS FOR FUTURE WORK  The corrosion of galvanized steel is a very complex process since it involves two distinct metals and each of the corrosion stages is very different from the other two. A lot of the research focuses on either the corrosion properties of zinc or steel, and there is little research published in the combination of both. This dissertation explored the corrosion properties of zinc and galvanized steel, but the corrosion properties of carbon steel in similar environments will be highly beneficial and will provide important information regarding the third stage of corrosion.    The techniques presented in this dissertation provide the results of the average electrochemical behavior over a small area of 1 cm2; however, as technology becomes more readily available, the use of Scanning Vibrating Electrode Technique (SVET), Scanning electrochemical microscopy (SECM) and Localized Electrochemical Impedance Spectroscopy (LEIS) can be used to study at a greater level of detail the processes involved during the second stage of corrosion in galvanized steel.   The work presented focused mainly on the effect of soluble salt content, soil compaction (oxygen availability), and temperature on the corrosion properties of zinc and galvanized steel. Resistivity and pH were also measured and considered during chapters 9 and 10. However, there are more factors that are known to affect the corrosion of hot-dipped galvanized steel in soil, and it would be beneficial to have in-depth studies of their effect on the corrosion of galvanized steel. These factors are organic content, moisture content, and stray currents.      173 Organic content is a very important factor affecting the corrosion behavior of steel reinforcements in soil. Organic content in earth is responsible or Microbial-Induced Corrosion (MIC). It is also expected that the presence of organic contant could form biofilms, which affect the corrosion behavior of metals. More importantly these biofilms are not uniform, affecting both anodic and cathodic reactions, and potentially leading to localized corrosion. More work needs to be done in this are to better understand the effects of organic content on the corrosion behavior of steel reinforcements in soil.   Chapters 9 and 10 introduced a novel model used to predict the corrosion of galvanized steel. This model is unique in its kind and has proven to yield results comparable to available data and widely used models; however, there is the need to have a long term study presenting the corrosion rates of galvanized steel in soil in conditions relevant to field applications. There are not many published attempts to do this, and the few that are available do not present a full set of data with all the relevant parameters. The recommended length of this study would be at least 15 to 20 years in order to capture the three corrosion stages of galvanized steel.      174REFERENCES "Canadian Standards Association, CAN/CSA-S6-06 Canadian Highway Bride Design Code, Published in November 2006, approved by Standards Council of Canada".  Abd El Aal, E.E. 2000, "Effect of Cl- anions on zinc passivity in borate solution", Corrosion Science, vol. 41, no. 1, pp. 1–16.  Abootalebi, O., Kermanpu, A., Shishesaz, M.R. & Golozar, M.A. 2010, "Optimizing the electrode position in sacrificial anode cathodic protection systems using boundary element method", Corrosion Science, vol. 52, no. 3, pp. 678-687 Alamilla, J.L., Espinosa-Medina, M.A. & Sosa, E. 2009a, "Modelling steel corrosion damage in soil environment" Corrosion Science, vol. 51, no. 11, pp. 2628-2638 Alamilla, J.L., Espinosa-Medina, M.A. & Sosa, E. 2009b, "Modelling steel corrosion damage in soil environment", Corrosion Science, vol. 51, no. 11, pp. 2628-2638.  Al-Mehthel, M., Al-Dulaijan, S., Al-Idi, S.H., Shameem, M., Ali, M.R. & Maslehuddin, M. 2009, "Performance of generic and proprietary corrosion inhibitors in chloride-contaminated silica fume cement concrete", Construction and Building Materials, vol. 23, no. 5, pp. 1768-1774.  Alvarez, M.G. & Galvele, J.R. 1976, "Pitting of high purity zinc and pining potential significance", Corrosion, vol. 32, pp. 285-294.  Amleh, L. & Ghosh, A. 2006, "Modeling the effect of corrosion on bond strength at the steel–concrete interface with finite-element analysis", Canadian Journal of Civil Engineering, vol. 33, no. 6, pp. 673-682.  Andrade, C. & Alonso, C. 2004, "Chapter 5 - Electrochemical Aspects of Galvanized Reinforcement Corrosion" in Galvanized Steel Reinforcement in Concrete, ed. Stephen R. Yeomans, Elsevier Science, Amsterdam, pp. 111-144.  Aperador, W., Mejia de Gutierrez, R. & Bastidas, D.M. 2009, "Steel corrosion behavior in carbonated alkali-activated slag concrete", Corrosion Science, vol. 51, no. 9, pp. 2027-2033.  Armour, T.A., Bickford, J. & Pfister, T. 2004, "Repair of failing MSE railroad bridge abutment", Geotech Spec Publ, vol. 124, pp. 380-394-08950563.  Armstrong, R.D. & Bell, M.F. 1974, "The Electrochemical Behavior of Zinc in Alkaline Solution", Electrochemistry in: H.R. Thirsk, Editor, The Chemical Society, , pp. 1-17.    175Asgari, H., Toroghinejad, M.R. & Golozar, M.A. 2009, "Effect of coating thickness on modifying the texture and corrosion performance of hot-dip galvanized coatings", Current Applied Physics, vol. 9, no. 1, pp. 59-66.  Autengruber, R., Luckeneder, G. & Hassel, A.W. 2012, "Corrosion of press-hardened galvanized steel", Corrosion Science, vol. 63, pp. 12-19.  B. Beverskog & I. Puigdomenech 1997, Revised Pourbaix Diagrams for Zinc at 25-300°C.  Barbalat, M., Lanarde, L., Caron, D., Meyer, M., Vittonato, J., Castillon, F., Fontaine, S. & Refait, P. 2012, "Electrochemical study of the corrosion rate of carbon steel in soil: Evolution with time and determination of residual corrosion rates under cathodic protection", Corrosion Science, vol. 55, no. 0, pp. 246-253.  Baril, G., Galicia, G., Deslouis, C., Pebere, N., Tribollet, B. & Vivier, V. 2007, "An Impedance Investigation of the Mechanism of Pure Magnesium Corrosion in Sodium Sulphate Solutions", Journal of the Electrochemical Society, vol. 152, no. 2, pp. 108-113. Barnard, N.C. & Brow, S.G.R. 2009, "Modelling the relationship between microstructure of Galfan-type coated steel and cut-edge corrosion resistance incorporating diffusion of multiple species", Corrosion Science, vol. 50, no. 10, pp. 2846-2857 Bastidas, D.M., Fernandez-Jimenez, A., Palomo, A. & Gonzalez, J.A. 2008, "A study on the passive state stability of steel embedded in activated fly ash mortars", Corrosion Science, vol. 50, no. 4, pp. 1058-1065.  Baugh, L.H. 1979, "Corrosion and polarization characteristics of zinc in neutral—acid media — I. Pure zinc in solutions of various sodium salts", Electrochimica Acta, vol. 24, no. 6, pp. 669–677.  Baugh, L.M. & Higginson, A. 1985, "Passivation of zinc in concentrated alkaline solution—I. Characteristics of active dissolution prior to passivation", Electrochimica Acta, vol. 30, no. 9, pp. 1163-1172.  Berke, B.S. & Sagüés, A. 2009, "Update on Condition of Reinforced Earthwall", Report No. BD544-32, , pp. 1-112.  Beverskog, B. & Puigdomenech, I. 1996, "Revised pourbaix diagrams for iron at 25–300 °C", Corrosion Science, vol. 38, no. 12, pp. 2121-2135.  Bohni, H. 2005, "Corrosion in Reinforced Concrete Structures", CRC Press, New York, .    176Bommersbach, P., Alemany-Dumont, C., Millet, J.P. & Normand, B. 2005, "Formation and behavior study of an environment-friendly corrosion inhibitor by electrochemical methods", Electrochimica Acta, vol. 51, no. 6, pp. 1076-1084.  Bonk, S., Wicinski, M., Hassel, A.W. & Stratmann, M. 2004, "Electrochemical characterizations of precipitates formed on zinc in alkaline sulphate solution with increasing pH values", Electrochemistry Communications, vol. 6, no. 8, pp. 800-804.  Boto, K.G. & Williams, L.F.G. 1977, "Rotating disc electrode studies of zinc corrosion", Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, vol. 77, no. 1, pp. 1-20.  Brent T. Mussato, Oliver K. Gepraegs, and Gary Farnden 2004, "Relative Effects of Sodium Chloride and Magnesium Chloride on Reinforced Concrete", Transportation Research Record: Journal of the Transportation Research Board, vol. 1866, pp. 59-66.  Broomfield, J.P. 1997, "Corrosion of steel in concrete: Understanding, investigation and repair", E & FN Spon Press, New York, NY, .  Brow, S.G.R. & Barnard, N.C. 2006, "3D computer simulation of the influence of microstructure on the cut edge corrosion behavior of a zinc aluminium alloy galvanized steel", Corrosion Science, vol. 48, no. 8, pp. 2291-2303 C. I. Elsner, P. R. Sere & A. R. Di Sarli 2012, "Atmospheric corrosion of painted galvanized and 55 % Al-Zn steel sheets Results of 12 years of exposure", International Journal of Corrosion, vol. Article ID 419640, 16 pages, 2012. doi:10.1155/2012/419640.  C., L. 2005, "Grain boundary structures in zinc oxide varistors", Acta Materialia, vol. 53, no. 2, pp. 237-245.  Cachet, C. 1992, "The Behavior of Zinc Electrode in Alkaline Electrolytes – II. A Kinetic Analysis of Anodic Dissolution", Journal of The Electrochemical Society, vol. 139, no. 3, pp. 644.  Cachet, C., Ganne, F., Joiret, S., Maurin, G., Petitjean, J., Vivier, V. & Wiart, R. 2002, "EIS investigation of zinc dissolution in aerated sulphate medium. Part II: zinc coatings", Electrochimica Acta, vol. 47, no. 21, pp. 3409-3422.  Cachet, C., Ganne, F., Maurin, G., Petitjean, J., Vivier, V. & Wiart, R. 2001, "EIS investigation of zinc dissolution in aerated sulphate medium. Part I: bulk zinc", Electrochimica Acta, vol. 47, no. 3, pp. 509-518.    177Cai, M. & Park, S. 1996, "Oxidation of Zinc in Alkaline Solutions Studied by Electrochemical Impedance Spectroscopy", Journal of the Electrochemical Society, vol. 143, no. 12, pp. 3895-3902. Carbucicchio, M., Ciprian, R., Ospitali, F. & Palombarini, G. 2008, "Morphology and phase composition of corrosion products formed at the zinc–iron interface of a galvanized steel", Corrosion Science, vol. 50, no. 9, pp. 2605-2613.  Chaker, V. 1990, "Corrosion Testing in Soils-Past, Present, and Future", Corrosion-Testing and Evaluation ASTM International, , pp. 95-111.  Chen, Y.Y., Chung, S.C. & Shih, H.C. 2006, "Studies on the initial stages of zinc atmospheric corrosion in the presence of chloride", Corrosion Science, vol. 48, no. 11, pp. 3547-3564.  Christopher Amrheln, James E. Strong, and Paul A. Mosher 1992, "Effect of De-icing Salts on Metal and Organic Matter Mobilization in Roadside Soils", Environmental Science and Technology, vol. 26, pp. 703-709.  Chung, S.J. & Kim, Y.K. 2011, "Hydrodynamic and Atmospheric Effects on Corrosion of Zinc in Borate Buffer Solution", Journal of the Korean Chemical Society, vol. 55, no. 4, pp. 575-580.  Chuvilin, E., Ershov, E. & Smirnova, O. 1998, "Ionic Migration in Frozen Soils and Ice", PERMAFROST Seventh International Conference (Proceedings), vol. 55, pp. 167-171.  Cole, I.S. & Marney, D. 2012, "The science of pipe corrosion: A review of the literature on the corrosion of ferrous metals in soils", Corrosion Science, vol. 56, no. 0, pp. 5-16.  Cole, I.S., Muster, T.H., Lau, D., Wright, N. & Azmat, N.S. 2010, "Products Formed during the Interaction of Seawater Droplets with Zinc Surfaces", Journal of the Electrochemical Society, vol. 157, no. 6, pp. C213-C222.  Cramer, S.D., Covino Jr., B. S., Bullard, S.J., Holcomb, G.R., Russell, J.H., Nelson, F.J., Laylor, H.M. & Soltesz, S.M. 2002, "Corrosion prevention and remediation strategies for reinforced concrete coastal bridges", Cement and Concrete Composites, vol. 24, pp. 101-117.  Culcasi, J.D., Elsner, C.I. & Di Sarli, A.R. 2009, "Effect of Zinc Crystals Size on Galvanized Steel Deformation and Electromechanical Behavior", Materials Research, vol. 12, pp. 273-279.  D.D.N. Singh & Rita Ghos 2006, "Electroless nickel–phosphorus coatings to protect steel reinforcement bars from chloride induced corrosion", Surface and Coatings Technology, vol. 201, no. 1-2, pp. 90-101    178Darbin, M., Jailloux, J.M. & Montuelle, J. 1988, "Durability of Reinforced Earth Structures: The Results of a Long-term Study Conducted on Galvanized Steel", Proc. Institution of Civil Engineers, vol. 84, pp. 1029-1057.  DDN Singh & Shyamjeet Y. 2008, "Role of tannic acid based rust converter on formation of passive film on zinc rich coating exposed in simulated concrete pore solution", Surface and Coatings Technology, vol. 202, no. 8, pp. 1526-1542   Dehwah, H.A.F., Maslehuddin, M. & Austin, S.A. 2002, "Long-term effect of sulphate ions and associated cation type on chloride-induced reinforcement corrosion in Portland cement concretes", Cement and Concrete Composites, vol. 24, no. 1, pp. 17-25.  Deliyanni, E.A., Bakoyannakis, D.N., Zouboulis, A.I., Matis, K.A. & Nalbandian, L. 2001, "Akaganéite-type β-FeO(OH) nanocrystals: preparation and characterization", Microporous and Mesoporous Materials, vol. 42, no. 1, pp. 49-57.  Doig, P. & Flewitt, P.E.J. 1979, "A Finite Difference Numerical Analysis of Galvanic Corrosion for Semi-Infinite Linear Coplanar Electrodes", Journal of the Electrochemical Society, vol. 126, no. 12, pp. 2057-2063..  Donald F. Vitaliano 1992, "An Economic Assessment of the Social Costs of Highway Salting and the Efficiency of Substituting a New De-icing Material", Journal of Policy Analysis and Management, vol. 1, no. 3, pp. 397-418.  Elias, V. 1990, "Durability/Corrosion of Soil Reinforced Structures", Report No. FHWA-RD-89-186, vol. Office of Engineering and Highway Operations, R&D, no. McLean, Virginia, pp. 173.  El-Mahdy, G.A., Nishikata, A. & Tsuru, T. 2000, "Electrochemical corrosion monitoring of galvanized steel under cyclic wet–dry conditions", Corrosion Science, vol. 42, no. 1, pp. 183-194.  Elsener, B. 2001, "Half-cell potential mapping to assess repair work on RC structures", Construction and Building Materials, vol. 15, no. 2–3, pp. 133-139.  Etteyeba N, Dhouibia L., Takenoutib, H., Alonso M.C. & Triki, E. 2007, "Corrosion inhibition of carbon steel in alkaline chloride media by Na3PO4", Electrochimica Acta, vol. 52, no. 27, pp. 7506–7512.    179F. Simescu & H. Idrissi 2008, "Effect of zinc phosphate chemical conversion coating on corrosion behavior of mild steel in alkaline medium: protection of rebars in reinforced concrete", Sci. Technol. Adv. Mater, vol. 9, pp. 1-10.  Farid Uddin, A.K.M., Ohtsu, M., Hossain, K.M.A. & Lachemi, M. 2007, "Simulation of reinforcement-corrosion-induced crack propagation in concrete by acoustic emission technique and boundary element method analysis", Canadian Journal of Civil Engineering, vol. 34, no. 10, pp. 1197-1207.  Feitknecht, W. 1959, "Studies on the influence of Chemicals Factors on the Corrosion of Metals", Chemistry and Industry, vol. 36, pp. 1102-1109.  Fishman, K., L & Withiam, J.L. 2011, "LRFD Metal Loss and Service-Life Strength Reduction Factors for Metal-Reinforced Systems", NCHRP REPORT 675, vol. Project 24-28, no. ISBN 978-0-309-15549-6.  Fletcher, S., Halliday, C.S., Gates, D., Westcott, M., Lwin, T. & Nelson, G. 1983, "The response of some nucleation/growth processes to triangular scans of potential", Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, vol. 159, no. 2, pp. 267-285.  Flis, J., Pickering, H.W. & Osseo-Asare, K. 1998, "Interpretation of impedance data for reinforcing steel in alkaline solution containing chlorides and acetates", Electrochimica Acta, vol. 43, no. 12-13, pp. 1921-1929.  G. S. Franke 1998, "Pitting corrosion of metals. A review of the critical factors", Journal of the Electrochemical Society, vol. 145, no. 6, pp. 186-2198.  Garcés, P., Saura, P., Méndez, A., Zornoza, E. & Andrade, C. 2008, "Effect of nitrite in corrosion of reinforcing steel in neutral and acid solutions simulating the electrolytic environments of micropores of concrete in the propagation period", Corrosion Science, vol. 50, no. 2, pp. 498-509.  García-Alonso M C, González J A, Miranda J, Escudero M L, Correia M L, Salta M & Bennani A 2007, "Corrosion behavior of innovative stainless steels in mortar", Cement and Concrete Research, vol. 37, no. 11, pp. 1562-1569.  Gardiner, C.P. & Melchers, R.E. 2002, "Corrosion of mild steel in porous media", Corrosion Science, vol. 44, no. 11, pp. 2459-2478.  Ge, J. 2006, "On the Numerical Solution of Laplace’s Equation with Nonlinear Boundary Condition for Corrosion of Steel in Concrete", M.A.Sc. Thesis, Carleton University, .    180Gervasi, C.A., Di Sarli, A.R., Cavalcanti, E., Ferraz, O., Bucharsky, E.C., Real, S.G. & Vilche, J.R. 1994, "The corrosion protection of steel in sea water using zinc-rich alkyd paints. An assessment of the pigment-content effect by EIS", Corrosion Science, vol. 36, no. 12, pp. 1963-1972.  Ghods, P., Isgor, O.B. & Pour-Ghaz, M. 2007, "A practical method for calculating the corrosion rate of uniformly depassivated reinforcing bars in concrete", Materials and Corrosion, vol. 58, no. 4, pp. 265-272 Glass, G.K., Reddy, B. & Buenfeld, N.R. 2000, "The participation of bound chloride in passive film breakdown on steel in concrete", Corrosion Science, vol. 42, no. 11, pp. 2013-2021 Gonzalez, J.A., Otero, E., Feliu, S. & Lopez, W. 1993, "Initial steps of corrosion in the steel/Ca(OH)ZCl system: The role of heterogeneities on the steel surface and oxygen supply", Cement and Concrete Research, vol. 23, no. 1, pp. 33.  Graedel, T.E. 1989, "Corrosion Mechanisms for Zinc Exposed to the Atmosphere", Journal of the Electrochemical Society, vol. 136, no. 4, pp. 193C-203C.  Gulikers, J. 2005, "Theoretical considerations on the supposed linear relationship between concrete resistivity and corrosion rate of steel reinforcement", Materials and Corrosion, vol. 56, no. 6, pp. 393-403.  Gupta, S.K. & Gupta, B.K. 1979, "The critical soil moisture content in the underground corrosion of mild steel", Corrosion Science, vol. 19, no. 3, pp. 171-178.  H. Dafydd, D.A. Worsley & H.N. McMurray 2005, "The kinetics and mechanism of cathodic oxygen reduction on zinc and zinc–aluminium alloy galvanized coatings" Corrosion Science, vol. 47, no. 12, pp. 3006-3018 H. Tanaka, T. Kitazawa, N. Hatanaka, T. Ishikawa & T. Nakayama 2012, "Simulating Study of Atmospheric Corrosion of Galvanized Steels: Influence of Zinc Rusts on the Formation of β-FeOOH Rust Particles by Aerial Oxidation of Aqueous FeCl2 Solution" Industrial and Engineering Chemistry Research, vol. 51, no. 1, pp. 248-254.  Hamdy H, H. 2001, "Corrosion behavior of zinc in sodium perchlorate solutions", Applied Surface Science, vol. 174, no. 3-4, pp. 201-209.  Hamlaoui, Y., Tifouti, L. & Pedraza, F. 2010, "On the corrosion resistance of porous electroplated zinc coatings in different corrosive media", Corrosion Science, vol. 52, no. 6, pp. 1883-1888.    181Hassan, H.H., Amin, M.A., Gubbala, S. & Sunkara, M.K. 2007, "Participation of the dissolved O2 in the passive layer formation on Zn surface in neutral media", Electrochimica Acta, vol. 52, no. 24, pp. 6929-6937.  Hassana, H.H., Amin, M.A., Gubbala, S. & Sunkara, M.K. 2007, "Participation of the dissolved O2 in the passive layer formation on Zn surface in neutral media", Electrochimica Acta, vol. 52, pp. 6929–693.  Hendrickx, J.M.H., Borchers, B., Woolslayer, J., Dekker, L.W., Ritsema, C. & Paton, S. 2001, "Spatial variability of dielectric properties in field soils", In Detection and Remediation Technologies for Mines and Minelike Targets, pp. 398-408 Hou, J. & Chung, D.D.L. 1997, "Cathodic protection of steel reinforced concrete facilitated by using carbon fiber reinforced mortar or concrete", Cement and Concrete Research, vol. 27, no. 5, pp. 649-656.  Hubbard, D.J. & Shanahan, C.E.A. 1973, "Corrosion of zinc and steel in dilute aqueous solutions" Corrosion Engineering Science and Technology, vol. 8, no. 6, pp. 270-274  Hurlen, T. & Eriksrud, E. 1973, "Kinetics of the Zn(Hg)/Zn(II) electrode in acid chloride solution", Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, vol. 45, no. 3, pp. 405-410.  Huttunen-Saarivirta, E., Kuokkala, V.-., Kokkonen, J. & Paajanen, H. 2009, "Corrosion behavior of aircraft coating systems in acetate- and formate-based de-icing chemicals", Materials and Corrosion, vol. 60, no. 3, pp. 173-191.  Ingham, M., Pringle, D. & Eicken, H. 2008, "Cross-borehole resistivity tomography of sea ice", Cold Regions Science and Technology, vol. 52, pp. 263-277.  J. S. McHattie, I. L. Perez & J. A. Kehr 1996, "Factors affecting cathodic disbondment of epoxy coatings for steel reinforcing bars", Cemente and Concrete Composites, vol. 18, no. 2, pp. 93-103  J.D. Thornley, R.V. Siddharthan, B. Luke, J.M. Salazar 2010, "Investigation and Implications of Mechanically Stabilized Earth Wall Corrosion in Nevada", Transportation Research Record: Journal of the Transportation Research Board, issue 2186. pp.154-160  Jackura, J.A., Garofalo, G. & Beddard, D. 1987, "Investigation of Corrosion at 14 Mechanically Stabilized Embankment Sites", Report No. CA/TL-87/12, vol. California Department of Transportation, Sacramento.    182Javaherdashti, R. 2000, "How corrosion affects industry and life", Corrosion Methods and Materials, vol. 47, no. 1, pp. 30-34.  Jiang, J.H., Ma, A.B., Weng, W.F., Fu, G.H., Zhang, Y.F., Liu, G.G. & Lu, F.M. 2009, "Corrosion fatigue performance of pre-split steel wires for high strength bridge cables", Fatigue & Fracture of Engineering Materials & Structures, vol. 32, no. 9, pp. 769-779.  Jiang, L., Wolpers, M., Volovitch, P. & Ogle, K. 2011, "The degradation of phosphate conversion coatings by electrochemically generated hydroxide", Corrosion Science, vol. 55, pp. 76-89.  Jones, D.A. 1995, "Principles and Prevention of Corrosion, second ed.", Prentice Hall.  Johnson, M.D., Leygraf, C. 2006, "Atmospheric corrosion of Zinc by Organic Constituents II. Reaction Routes for Zinc-Acetate Formation", Journal of The Electrochemical Society, vol. 153 (12), pp. B542-B546 José Daniel Culcasi, Cecilia Inés Elsner, Alejandro Ramón Di Sarli 2009, "Effect of zinc crystals size on galvanized steel deformation and electrochemical behavior", Materials Research, vol. 12, no. 3, pp. 273-279.  K.H. Pettersson 1996, "Factors influencing chloride induced corrosion of reinforcement in concrete", C. Sjostrom (Ed.), Durability of building materials and components, vol. 1 Chapman and Hall, pp. 334.  Kasahara. k & Kajiyama , F. 1983, "Determination of Underground Corrosion Rates from Polarization Resistance Measurements", NACE International.  Klemm, S.O., Schauer, J., Schuhmacher, B. & Hassel, A.W. 2011, "A microelectrochemical scanning flow cell with downstream analytics", Electrochimica Acta, vol. 56, no. 11, pp. 4315-4321.  L. Fay, K. Volkening, C. Gallaway, X. Shi 2008, "Performance and Impacts of Current De-icing and Anti-icing Products: User Perspective Versus Experimental Data", Proceedings 87th Annual Meeting of Transportation Research Board, , pp. 1-23.  L.D. Stephenson, A. Kumar, J. Hale & .N. Murray 2009, "System for measurement of corrosion under coating", Materials Performance, vol. 48, pp. 36-41.  L.H. Baugh 1979, "Corrosion and polarization characteristics of zinc in neutral—acid media — I. Pure zinc in solutions of various sodium salts", Electrochimica Acta, vol. 27, pp. 657-667.    183Le Manchet, S., Landoulsi, J., Richard, C. & Verchère, D. 2010a, "Study of a chromium-free treatment on Hot-Dip Galvanized steel : Electrochemical behavior and performance in a saline medium", Surface and Coatings Technology, vol. 205, no. 2, pp. 475–482.  Le Manchet, S., Landoulsi, J., Richard, C. & Verchère, D. 2010b, "Study of a chromium-free treatment on Hot-Dip Galvanized steel: Electrochemical behavior and performance in a saline medium", Surface and Coatings Technology, vol. 205, no. 2, pp. 475-482.  Le Manchet, S., Landoulsi, J., Richard, C. & Verchère, D. 2010c, "Study of a chromium-free treatment on Hot-Dip Galvanized steel: Electrochemical behavior and performance in a saline medium", Surface and Coatings Technology, vol. 205, no. 2, pp. 475-482.  Legat, A. 2007, "Monitoring of steel corrosion in concrete by electrode arrays and electrical resistance probes", Electrochimica Acta, vol. 52, no. 27, pp. 7590-7598.  Li, H.Y., Duan, J.Y. & Wei, D.D. 2013, "Comparison on corrosion behavior of arc sprayed and zinc-rich coatings", Surface and Coatings Technology, vol. 235, pp. 259-266.  Li, M.C., Royer, M., Stien, D., Lecante, A. & Roos, C. 2008, "Inhibitive effect of sodium eperuate on zinc corrosion in alkaline solutions", Corrosion Science, vol. 50, no. 7, pp. 1975-1981.  Lindström, D. & Odnevall Wallinder, I. 2011, "Long-term use of galvanized steel in external applications. Aspects of patina formation, zinc runoff, barrier properties of surface treatments, and coatings and environmental fate", Environmental Monitor and Assessment, vol. 173, no. 4, pp. 139-153. Liu, S., Sun, H., Sun, L. & Fan, H. 2012, "Effects of pH and Cl− concentration on corrosion behavior of the galvanized steel in simulated rust layer solution", Corrosion Science, vol. 65, pp. 520-527.  Lu, J., Wu, H., Kong, G., Che, C. and Xu, Q. 2006, "Growth and corrosion behavior of rare earth film on hot-dip galvanized steel", Transactions of Nonferrous Metals Society of China, vol. 16, no. 6, pp. 1397-1401.  M. Manna, G. Naid, N., Raniz, N. & Bandyopadhyay 2008, "Characterisation of coating on rebar surface using Hot-dip Zn and Zn-4.9Al-0.1 misch metal bath", Surface and Coatings Technologies, vol. 202, no. 8, pp. 1510-1516  M.M. Jalili, S. Moradian & D. Hosseinpour 2009, "The use of inorganic conversion coatings to enhance the corrosion resistance of reinforcement and the bond strength at the rebar/concrete", Construction and Building Materials, vol. 23, no. 1, pp. 233-238   184Macias, A. & Andrade, C. 1990, "The behavior of galvanized steel in chloride-containing alkaline solutions-I. The influence of the cation", Corrosion Science, vol. 30, no. 4-5, pp. 393-407.  Mahdavian, M., Naderi, R,. 2011, "Corrosion inhibition of mild steel in sodium chloride solution by some zinc complexes", Corrosion Science, vol. 42, pp. 1194-1200 Mangat, P. & Molloy, B. 1992, "Factors influencing chloride-induced corrosion of reinforcement in concrete", Materials and Structures, vol. 25, no. 7, pp.404-411  Manov, S., Lamazouere, A.M. & Aries, L. 2000, "Electrochemical study of the corrosion behavior of zinc treated with a new organic chelating inhibitor", Corrosion Science, vol. 42, no. 7, pp. 1235–1248.  Maslehuddin, M., Al-Zahrani, M.M., Ibrahim, M., Al-Mehthel, M.H. & Al-Idi, S.H. 2007, "Effect of chloride concentration in soil on reinforcement corrosion", Construction and Building Materials, vol. 21, no. 8, pp. 1825-1832.  McMillan, R.S., MacKinnon, D.J. & Dutrizac, J.E. 1982, "Anodic dissolution of n-type and p-type chalcopyrite", Journal of Applied Electrochemistry, vol. 12, no. 6, pp. 743-757  Melchers, R.E. 2003, "Mathematical modelling of the diffusion controlled phase in marine immersion corrosion of mild steel", Corrosion Science, vol. 45, pp. 923–940.  Miao, W., Cole, I.S., Neufeld, A.K. & Furman, S. 2007a, "Pitting Corrosion of Zn and Zn-Al Coated Steels in pH 2 to 12 NaCl Solutions", Journal of The Electrochemical Society, vol. 154, pp. 7-15.  Miao, W., Cole, I.S., Neufeld, A.K. & Furman, S. 2007b, "Pitting Corrosion of Zn and Zn-Al Coated Steels in pH 2 to 12 NaCl Solutions", The Journal of the Electrochemical Society, vol. 154, no. 1, pp. C7-C15.  Mohammadi, F., Nickchi, T., Attar, M.M. & Alfantazi, A. 2011, "EIS study of potentiostatically formed passive film on 304 stainless steel", Electrochimica Acta, vol. 56, no. 24, pp. 8727-8733.  Mokaddem, M., Volovitch, P. & Ogle, K. 2010, "The anodic dissolution of zinc and zinc alloys in alkaline solution. I. Oxide formation on electrogalvanized steel", Electrochimica Acta, vol. 55, no. 27, pp. 7867-7875.  Moreno, M., Morris, W., Alvarez, M.G. & Duffó, G.S. 2004, "Corrosion of reinforcing steel in simulated concrete pore solutions Effect of carbonation and chloride content", Corrosion Science, vol. 46, no. 11, pp. 2681–2699.    185Morris, W., Vázquez, M. & De Sánchez, S.R. 2000, "Efficiency of coatings applied on rebars in concrete", Journal of Materials Science, vol. 20, no. 8 pp. 1885-1890 Mouanga, M. & Berçot, P. 2010, "Comparison of corrosion behavior of zinc in NaCl and in NaOH solutions; Part II: Electrochemical analyses", Corrosion Science, vol. 52, no. 12, pp. 3993-4000.  Mouanga, M., Puiggali, M. & Devos, O. 2013, "EIS and LEIS investigation of aging low carbon steel with Zn–Ni coating", Electrochimica Acta, vol. 106, no. 0, pp. 82-90.  Mouanga, M., Puiggali, M., Tribollet, B., Vivier, V., Pébère, N. & Devos, O. 2013, "Galvanic corrosion between zinc and carbon steel investigated by local electrochemical impedance spectroscopy", Electrochimica Acta, vol. 88, no. 0, pp. 6-14.  Munn, R.S. & Devereux, O.F. 1991, "Numerical Modeling and Solution of Galvanic Corrosion Systems: Part ll. Finite-Element Formulation and Descriptive Examples", Corrosion NACE, vol. 47, pp. 618-634.  Muñoz, A.G., Saidman, S.B. & Bessone, J.B. 2002, "Corrosion of an Al–Zn–In alloy in chloride media", Corrosion Science, vol. 44, no. 10, pp. 2171-2182.  Olivier, M., Lanzutti, A., Motte, C. & Fedrizzi, L. 2010, "Influence of oxidizing ability of the medium on the growth of lanthanide layers on galvanized steel", Corrosion Science, vol. 54, no. 2, pp. 1428–1439.  Ostrem, G. 1967, "Laboratory measurements of the resistivity of ice", Journal of Glaciology, vol. 6, no. 47, pp. 463-469.  Padilla, V., Ghods, P., Alfantazi, A. 2013, "Practical Model for Three-Stage Corrosion Behavior of Galvanized Steel Reinforcement in Soil.", Corrosion, vol. 69(5), pp. 509-521.  Padilla, V. & Alfantazi, A. 2010, "Effects of Differential Oxygen Access on the Corrosion Behavior of Zinc Relevant to Mechanically Stabilized Earth Walls", Supplemental Proceedings: Volume 3: General Paper Selections, TMS, , pp. 395-403.  Padilla, V., Ghods, P. & Alfantazi, A. 2013, "Effect of de-icing salts on the corrosion performance of galvanized steel in sulphate contaminated soil", Construction and Building Materials, vol. 40, no. 0, pp. 908-918.  Page, C.L., K. W. J. Treadaway 1982, "Aspects of the electrochemistry of steel in concrete", Nature, vol. 297, pp. 109-115.    186Parthiban, G.T., Parthiban, T., Ravi, R., Saraswathy, V., Palaniswamy, N. & Sivan, V. 2008, "Cathodic protection of steel in concrete using magnesium alloy anode", Corrosion Science, vol. 50, no. 12, pp. 3329-3335.  Pérez-Quiroz, J.T., Terán, J., Herrera, M.J., Martínez, M. & Genescá, J. 2008, "Assessment of stainless steel reinforcement for concrete structures rehabilitation", Journal of Construction Steel Research, vol. 64, no. 11, pp. 1317–1324.  Petkuviene, J. & Paliulis, D. 2009, "Experimental research of road maintenance salts and molasses ("safecote") corrosive impact on metals", Journal of Environmental Engineering and Landscape Management, vol. 17, no. 4, pp. 236-243.  Pilbáth, Z. & Sziráki, L. 2008, "The electrochemical reduction of oxygen on zinc corrosion films in alkaline solutions", Electrochimica Acta, vol. 53, no. 7, pp. 3218-3230.  Pineau, S., Sabot, R., Quillet, L., Jeannin, M., Caplat, C., Dupont-Morral, I. & Refait, P. 2008, "Formation of the Fe(II–III) hydroxysulphate green rust during marine corrosion of steel associated to molecular detection of dissimilatory sulphite-reductase", Corrosion Science, vol. 50, no. 4, pp. 1099-1111.  Pourbaix, M. 1974, "Atlas of electrochemical equilibria in aqueous solutions", National Association of Corrosion Engineers, vol. 2nd ed. Houston:.  Pour-Ghaz, M., Isgor, O.B. & Ghods, P. 2007, "Virtual experiments to investigate steel corrosion in concrete", ASCE International Workshop on Computing in Civil Engineering, , pp. 542.  Pour-Ghaz, M., Isgor, O.B. & Ghods, P. 2009a, "The Effect of Temperature on the Corrosion of Steel in Concrete. Part 1: Virtual Polarization Resistance Tests and Model Development", Corrosion Science, vol. 51, pp. 415-425.  Pour-Ghaz, M., Isgor, O.B. & Ghods, P. 2009b, "The Effect of Temperature on the Corrosion of Steel in Concrete. Part 2: Model Verification and Parametric Study", Corrosion Science, vol. 51, pp. 426-433.  Pour-Ghaz. M & Burkan Isgor, O, Ghods, P 2009, "The effect of temperature on the corrosion of steel in concrete. Part 1: Simulated polarization resistance tests and model development", Corrosion Science, vol. 51, no. 2, pp. 415-425.  Qing Qu, Chuanwei Yan, Ye Wan & Chunan Cao 2002, "Effects of NaCl and SO2 on the initial atmospheric corrosion of zinc", Corrosion Science, vol. 44, pp. 2789–2803.    187R.A. Gladstone, P.L. Anderson, K.L. Fishman, J.L. Withiam 2006, "Durability of Galvanized Soil Reinforcement: More Than 30 Years of Experience with Mechanically Stabilized Earth", Transportation Research Record: Journal of the Transportation Research Board, vol. 1975, pp. 49-59.  R.B. Hoxeng, C.F.P. 1949, "Electrochemical behavior of zinc and steel in aqueous media, Part I", Corrosion, vol. 5, pp. 330-338.  Rangaraju, P.R., Sompura, K., Desai, J.1. & Olek, J. 2006, "Potential of potassium acetate de-icer to induce ASR in concrete, and its mitigation", Proceedings of the 2006 Airfield and Highway Pavement Specialty Conference, , pp. 683-694.  Rehm, G. 1980, "The Service Life of Reinforced Earth Structures from a Corrosion Technology Standpoint, Expert Report", The Reinforced Earth Company, vol. Vienna, Virginia.  Revie, R.W. & Uhlig, H.H. 2008, "Corrosion and Corrosion Control, fourth ed.", John Wiley & Sons, .  Reynolds, J.M. & Paren, J.G. 1984, "Electrical Resistivity of ice from the Antarctic peninsula", Journal of Glaciology, vol. 30, no. 106, pp. 289-295.  Rita Ghosh & D.D.N. Singh 2007, "Kinetics, mechanism and characterisation of passive film formed on hot-dip galvanized coating exposed in simulated concrete pore solution", Surface and Coatings Technology, vol. 201 no.16-17 pp. 7346-7359  Robert Baboian 2005, "Corrosion Tests And Standards: Application And Interpretation - Volume 20 of ASTM manual series: American Society for Testing and Materials", ASTM International, vol. 20, pp. 1-882.  Roman, I., 1938, "Electrical resistivity of snow and ice", International Association of Hydrological Sciences, vol. 23, pp. 483-491.  Romanoff, M. 1957, "Underground Corrosion", National Bureau of Standards, vol. Circular 579, no. U.S. Department of Commerce, pp. Washington, D.C.  Rosalbino, F., Angelini, E., Macciò, D., Saccone, A. & Delfino, S. 2009, "Application of EIS to assess the effect of rare earths small addition on the corrosion behavior of Zn–5% Al (Galfan) alloy in neutral aerated sodium chloride solution", Electrochimica Acta, vol. 54, no. 4, pp. 1204-1209.  Rout, T.K. 2007, "Electrochemical impedance spectroscopy study on multi-layered coated steel sheets", Corrosion Science, vol. 49, no. 2, pp. 794-817.    188S. A. Ajeel & G.A. Ali 2008, "Variable Conditions Effect On Polarization Parameters Of Impressed Current Cathodic Protection Of Low Carbon Steel Pipes", J. Med. Eng. Technol, vol. 26, pp. 636-647.  S. Thomas, N. Birbilis, M.S. Venkatraman, I.S. Cole 2012, "Corrosion of Zinc as a Function of pH", Corrosion, vol. 68:1, 015009-1-015009-9.  Sagüés, A., Scott1, R., Rossi, J., Peña, J. & Powers, R. 2000, "Corrosion of Galvanized Strips in Florida Reinforced Earth Walls", Journal of Materials in Civil Engineering, vol. 12, pp. 220-227.  Sakairi, M., Uchida, Y., Kikuchi, T. & Takahashi, H. 2008, "Influence of Zinc Ions on Initial Stage of Localized Corrosion of Zn and Zn–Al Alloy Coated Steels with Photon Rupture Method", ISIJ International, vol. 48, pp. 988–993.  Samouëlian, A., Cousin, I., Tabbagh, A., Bruand, A. & Richard, G. 2005, "Electrical resistivity survey in soil science: a review", Soil and Tillage Research, vol. 83, no. 2, pp. 173-193.  Sere, P.R., Culcasi, J.D., Elsner, C.I. & Di Sarli, A.R. 1999, "Relationship between texture and corrosion resistance in hot-dip galvanized steel sheet", Surface and Coatings Technology, vol. 122, no. 2, pp. 143–149.  Seré, P.R., Culcasi, J.D., Elsner, C.I. & Di Sarli, A.R. 1999, "Relationship between texture and corrosion resistance in hot-dip galvanized steel sheets", Surface and Coatings Technology, vol. 122, no. 2-3, pp. 143-149.  Short, N.R., Abibsi, A. & Dennis, J.K. 1989, "Corrosion resistance of electroplated zinc alloy coatings", Transactions of the Institute of Metal Finishing, vol. 67, no. 3, pp. 73-77.  Söylev, T.A. & Richardson, M.G. 2008, "Corrosion inhibitors for steel in concrete: State-of-the-art report", Construction and Building Materials, vol. 22, no. 4, pp. 609-622.  Stankovic, S., Grgur, B., Krstajic, N. & Vojnovic, M. 2003, "Kinetics of the zinc anodic dissolution reaction in near neutral EDTA solutions", Journal of the Serbian Chemical Society, vol. 68, pp. 207–218.  Steven R. Corsi, Steven W. Geis, George Bowman, Greg G. Faile & Troy D. Rutter 2009, Aquatic Toxicity of Airfield-Pavement De-icer Materials and Implications for Airport Runoff, - American Chemical Society.    189Sziraki, L., Szocs, E., Pilbath, Z., Papp, K. & Kalman, E. 2001, "Study of the initial stage of white rust formation on zinc single crystal by EIS, STM/AFM and SEM/EDS techniques", Electrochimica Acta, vol. 46, no. 24-25, pp. 3743–3754.  Sziráki, L., Cziráki, Á., Geröcs, I., Vértesy, Z. & Kiss, L. 1998, "A kinetic model of the spontaneous passivation and corrosion of zinc in near neutral Na2SO4 solutions", Electrochimica Acta, vol. 43, no. 1-2, pp. 175-186.  T. Yamashita 1980, "Electro-deposition and electro-dissolution of zinc in the vicinity of equilibrium potential in aqueous zinc sulphate solutions", Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, vol. 106, pp. 95-102  T.P. Dirkse 1979, "The Behavior of the Zinc Electrode in Alkaline Solutions II. Reaction Orders at the Equilibrium Potential ", J. Electrochem. Soc., vol. 126, pp. 784-786.  Tan, Y. 2011, "Experimental methods designed for measuring corrosion in highly resistive and inhomogeneous media", Corrosion Science, vol. 53, no. 4, pp. 1145-1155.  Tan, Y. 2009, "Sensing localised corrosion by means of electrochemical noise detection and analysis", Sensors and Actuators B: Chemical, vol. 139, no. 2, pp. 688-698.  Tanaka, H., Takeuchi, Y., Ishikawa, T. & Nakayama, T. 2011, "Influence of anions on the formation and structure of artificial zinc rusts", Corrosion Science, vol. 53, no. 8, pp. 2502-2507.  Tanaka, N. & Tamamushi, R. 1964, "Kinetic parameters of electrode reactions", Electrochimica Acta, vol. 9, no. 7, pp. 963-989.  Thébault, B., Vuillemin, R., Oltra, C., Allely, L., Dosdat, K. & Ogle 2007, "Predictive Model for Cut-Edge Corrosion of Galvanized Steels", ECS Transactions, vol. 31, pp. 343-353.  Thomas, S., Cole, I.S., Sridhar, M. & Birbilis, N. 2013, "Revisiting zinc passivation in alkaline solutions", Electrochimica Acta, vol. 97, no. 0, pp. 192-201.  Titz, T., Hörzenberger, F., Van den Bergh, K. & Grundmeier, G. 2010, "Correlation of interfacial electrode potential and corrosion resistance of plasma polymer coated galvanized steel. Part 2: Influence of forming induced defects", Corrosion Science, vol. 52, no. 2, pp. 378-386.  Tsai, Y.T., Hou, K.H., Bai, C.Y., Lee, J.L. & Ger, M.D. 2010, "The influence on immersion time of titanium conversion coatings on electrogalvanized steel", Thin Solid Films, vol. 518, no. 24, pp. 7541-7544.    190V. Elias 2000, "Corrosion/Degradation of Soil Reinforcements for Mechanically Stabilized Earth Walls and Reinforced Soil Slopes", vol. FHWA-NHI-00-044, pp. 1-94.  V.E. Padilla Perez & A. Alfantazi 2012a, "Corrosion Performance of Galvanized Steel in Na2SO4 and NaCl solutions at Subfreezing Temperatures", Corrosion NACE, vol. 69, no. 2, pp. 174-185.  V.E. Padilla Perez & A. Alfantazi 2012b, "Effects of Oxygen and Sulphate Concentrations on the Corrosion Behavior of Zinc in NaCl Solutions", Corrosion NACE, vol. 68, no. 3, pp. 035005-1-035005-11 Vachaud, G., Passerat de Silans, A., Balabanis, P. & Vauclin, M. 1985, "Temporal Stability of Spatially Measured Soil Water Probability Density Function", Soil. Sci. Soc. Am., vol. 49, pp. 822-828.  Vagge, S.T., Raja, V.S. & Narayanan, R.G. 2007, "Effect of deformation on the electrochemical behavior of hot-dip galvanized steel sheets", Applied Surface Science, vol. 253, no. 20, pp. 8415-8421.  W. J. Yang, P. Yang, X. M. Li & W. L. Feng 2010, "Influence of tensile stress on corrosion behavior of high-strength galvanized steel bridge wires in simulated acid rain", Materials and Corrosion, vol. 63, no. 5, pp. 401-407. Wallinder, D., Pan, J., Leygraf, C. & Delblanc-Bauer, A. 1998, "Eis and XPS study of surface modification of 316LVM stainless steel after passivation", Corrosion Science, vol. 41, no. 2, pp. 275-289.  Walter, G.W. 1991, "Laboratory simulation of atmospheric corrosion by SO2—I. Apparatus, electrochemical techniques, example results", Corrosion Science, vol. 32, no. 12, pp. 1331-1352.  Wang, K., Nelsen, D.E. & Nixon, W.A. 2006, "Damaging effects of de-icing chemicals on concrete materials", Cement and Concrete Composites, vol. 28, no. 2, pp. 173-188.  Wang, W., Mohammadi, F. & Alfantazi, A. 2012, "Corrosion behavior of niobium in phosphate buffered saline solutions with different concentrations of bovine serum albumin", Corrosion Science, vol. 57, no. 0, pp. 11-21.  Weissenrieder, J., Kleber, C., Schreiner, M. & Leygraf, C. 2004, "In Situ Studies of Sulphate Nest Formation on Iron", Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, vol. 151, no. 9, pp. B497-B504.    191Wildt, R. & Cook, M. 2002, "Zinc and Steel - Partners in sustainable Construction, Steel in Sustainable Construction", IISI World Conference 2002 Proceedings, pp. 321-328  Wu, K., Zhu, L., Li, W. & Liu, H. 2010, "Effect of Ca2+and Mg2+ on corrosion and scaling of galvanized steel pipe in simulated geothermal water", Corrosion Science, vol. 52, no. 7, pp. 2244-2249.  Xianming, S., Fay, L., Zhengxian, Y., Tuan, A.N. & Yajun, L. 2009, "Corrosion of de-icers to metals in transportation infrastructure: Introduction and recent developments", Corrosion reviews, vol. 27, no. 1, pp. 23.  Xing, S., Li, Y., Wu, J. & Yan, Y. 2010, "Electrochemical methods study on corrosion of 5% Al–Zn alloy-coated steel under thin electrolyte layers", Materials and Corrosion, vol. 61, no. 5, pp. 428-431.  Yadav, A.P., Nishikata, A. & Tsuru, T. 2008, "Evaluation of impedance spectra of zinc and galvanized steel corroding under atmospheric environments", Corrosion NACE, vol. 43, no. 1, pp. 23-29.  Yadav, A.P., Nishikata, A. & Tsuru, T. 2005, "Oxygen reduction mechanism on corroded zinc", Journal of Electroanalytical Chemistry, vol. 585, no. 1, pp. 142-149.  Yadav, A.P., Katayama, H., Noda, K., Masuda, H., Nishikata, A. & Tsuru, T. 2007, "Effect of Fe-Zn alloy layer on the corrosion resistance of galvanized steel in chloride containing environments", Corrosion Science, vol. 49, no. 9, pp. 3716-3731.  Yadav, A.P., Nishikata, A. & Tsuru, T. 2004a, "Electrochemical impedance study on galvanized steel corrosion under cyclic wet–dry conditions––influence of time of wetness", Corrosion Science, vol. 46, no. 1, pp. 169-181.  Yadav, A.P., Nishikata, A. & Tsuru, T. 2004b, "Degradation mechanism of galvanized steel in wet–dry cyclic environment containing chloride ions", Corrosion Science, vol. 46, no. 2, pp. 361-376.  Yang, L., Sridhar, N., Brossia, C.S. & Dunn, D.S. 2005, "Evaluation of the coupled multielectrode array sensor as a real-time corrosion monitor", Corrosion Science, vol. 47, no. 7, pp. 1794-1809.  Yehia, S., and Y. Tuan 1998, "Bridge Deck De-icing", Transportation Conference Proceedings, , pp. 51-57.    192Yohchia, C. 2000, "Practical analysis and design of mechanically-stabilized earth walls—I. Design philosophies and procedures", Engineering Structures, vol. 22, no. 7, pp. 793-808.  Yuan, Y., Ji, Y. & Jiang, J. 2009, "Effect of corrosion layer of steel bar in concrete on time-variant corrosion rate", Materials and Structures, vol. 42, no. 10, pp. 1443-1450 Zhang, J.Y. 2008, "Corrosion of Reinforcing Steel in Concrete Structures: Understanding the Mechanisms", Construction Business, vol. 5, pp. 2.  Zhang, X.G. 1996, "Corrosion and Electrochemistry of Zinc", Plenum Press, New York, .  Zhang, X., Leygraf, C. & Odnevall Wallinder, I. 2013, "Atmospheric corrosion of Galfan coatings on steel in chloride-rich environments", Corrosion Science, vol. 73, no. 0, pp. 62-71.   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0165584/manifest

Comment

Related Items