"Applied Science, Faculty of"@en . "Materials Engineering, Department of"@en . "DSpace"@en . "UBCV"@en . "Asselin, Edouard"@en . "2011-01-20T00:32:29Z"@en . "2007"@en . "Doctor of Philosophy - PhD"@en . "University of British Columbia"@en . "The corrosion characteristics of Alloy 625 (UNS 00625, Ni - 22 Cr - 10 Mo) in\r\noxygenated ammoniacal sulphate environments are determined at room temperature and\r\npressure and up to high temperatures and pressures (673 K, 250 bar) commensurate with\r\nthe process of supercritical water oxidation (SCWO). Electrochemical methods such as\r\nlinear polarization, potentiodynamic polarization and impedance spectroscopy are used.\r\nIt is found that the electrochemical and morphological response is dictated by the\r\nalloying element Cr and the formation of a Cr(III) oxide. Mo and, to a lesser extent Ni,\r\nare found to dissolve readily. Thermodynamic analysis of the Ni-NH\u00E2\u0082\u0083-H\u00E2\u0082\u0082O system,\r\nincluding new Pourbaix diagrams at temperatures as high as 653 K , has shown that Ni -\r\nammine formation is possible at moderate temperatures but that the stability of these\r\ncomplexes decreases substantially with temperature. According to one of the models\r\ninvestigated, which is based on the only available high temperature equilibrium constant\r\ndata, Ni-ammines become unstable above approximately 473 K .\r\nImpedance spectroscopy has shown that transpassive dissolution of the alloy's ptype,\r\ncation conducting, Cr(III) oxide occurs at temperatures as low as 373 K and total\r\npressure (oxygen saturated) as low as 40 bar. As temperature and pressure are increased\r\nthe corrosion process is increasingly diffusion controlled. Transpassive dissolution\r\nresults in the thinning and eventual total removal of the alloy's protective semiconductor\r\nbarrier layer. Cation ejection from the barrier layer into the solution and porous outer\r\nlayer phase results in precipitation of a Cr(III) scale (oxide or hydroxyl-oxide) at the\r\nalloy surface which acts as a diffusion barrier. It is hypothesized that the outer layer is\r\neither physically removed at supercritical conditions due to rapid dissolution and grain boundary attack of the alloy or chemically removed by solution acidification due to the\r\nformation of sulphuric acid at high density supercritical conditions.\r\nAlloys 625, 316 L, Ni - 20 Cr and pure Nb are tested at SCWO conditions and it\r\nis found that the corrosion resistance increases with Cr content and Nb is found to\r\nperform well in sulphate containing SCWO solutions at oxygen concentrations up to 4 m.\r\nIt is also confirmed in this work that maximum material loss occurs in the high-density\r\nsupercritical region of the reactor.\r\nNew Pourbaix diagrams for Nb at elevated temperatures (348 and 368 K) are\r\ncalculated and compared to electrochemical and weight loss measurements performed in\r\nconcentrated acids. Through electrochemical experiments in concentrated sulphuric and\r\nhydrochloric acids, Nb is found to be an ideal candidate for the high-density supercritical\r\nsections of SCWO reactors."@en . "https://circle.library.ubc.ca/rest/handle/2429/30709?expand=metadata"@en . "H I G H T E M P E R A T U R E AND H I G H PRESSURE CORROSION O F Ni-BASED A L L O Y S AND STAINLESS S T E E L S IN A M M O N I A C A L S U L P H A T E SOLUTION by Edouard Asselin B . A . S . c , University of Ottawa, 2002 THESIS S U B M I T T E D I N P A R T I A L F U L F I L L M E N T OF T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F D O C T O R OF P H I L O S O P H Y in The Faculty of Graduate Studies (Materials Engineering) U N I V E R S I T Y OF B R I T I S H C O L U M B I A February 2007 \u00C2\u00A9 Edouard Asselin 2007 ABSTRACT The corrosion characteristics of A l loy 625 (UNS 00625, N i - 22 Cr - 10 Mo) in oxygenated ammoniacal sulphate environments are determined at room temperature and pressure and up to high temperatures and pressures (673 K , 250 bar) commensurate with the process of supercritical water oxidation (SCWO). Electrochemical methods such as linear polarization, potentiodynamic polarization and impedance spectroscopy are used. It is found that the electrochemical and morphological response is dictated by the alloying element Cr and the formation of a Cr(III) oxide. M o and, to a lesser extent N i , are found to dissolve readily. Thermodynamic analysis of the Ni-NH3-H20 system, including new Pourbaix diagrams at temperatures as high as 653 K , has shown that N i -ammine formation is possible at moderate temperatures but that the stability of these complexes decreases substantially with temperature. According to one of the models investigated, which is based on the only available high temperature equilibrium constant data, Ni-ammines become unstable above approximately 473 K . Impedance spectroscopy has shown that transpassive dissolution of the alloy's p-type, cation conducting, Cr(III) oxide occurs at temperatures as low as 373 K and total pressure (oxygen saturated) as low as 40 bar. As temperature and pressure are increased the corrosion process is increasingly diffusion controlled. Transpassive dissolution results in the thinning and eventual total removal of the alloy's protective semiconductor barrier layer. Cation ejection from the barrier layer into the solution and porous outer layer phase results in precipitation of a Cr(III) scale (oxide or hydroxyl-oxide) at the alloy surface which acts as a diffusion barrier. It is hypothesized that the outer layer is either physically removed at supercritical conditions due to rapid dissolution and grain i i boundary attack of the alloy or chemically removed by solution acidification due to the formation of sulphuric acid at high density supercritical conditions. Al loys 625, 316 L , N i - 20 Cr and pure Nb are tested at S C W O conditions and it is found that the corrosion resistance increases with Cr content and Nb is found to perform well in sulphate containing S C W O solutions at oxygen concentrations up to 4 m. It is also confirmed in this work that maximum material loss occurs in the high-density supercritical region of the reactor. New Pourbaix diagrams for Nb at elevated temperatures (348 and 368 K ) are calculated and compared to electrochemical and weight loss measurements performed in concentrated acids. Through electrochemical experiments in concentrated sulphuric and hydrochloric acids, Nb is found to be an ideal candidate for the high-density supercritical sections of S C W O reactors. in TABLE OF CONTENTS A B S T R A C T i i T A B L E OF C O N T E N T S iv LIST O F T A B L E S v i LIST O F F I G U R E S v i i i A C K N O W L E D G E M E N T S xv i D E D I C A T I O N xv i i 1 I N T R O D U C T I O N 1 2 L I T E R A T U R E R E V I E W 9 2.1 The Process of S C W O 9 2.1.1 The U B C reactor 9 2.1.2 Advantages of S C W O 10 2.1.3 Ammonia persistency 11 2.2 Properties of supercritical water 14 2.3 Thermodynamics and passivation 17 2.4 The point defect model (PDM) and Transpassivity 25 2.5 Alternatives to Ni-based alloys or stainless steels 32 2.6 The issue of ammonia and Ni-based alloys 34 3 O B J E C T I V E S 36 4 P O L A R I Z A T I O N S T U D Y OF A L L O Y 625, N i , Cr A N D M o I N A M M O N I A C A L S U L P H A T E S O L U T I O N S 37 4.1 Experimental 37 4.2 Results and discussion 40 4.2.1 Pourbaix diagrams 40 4.2.2 Effect of oxidant 42 4.2.3 Effect of p H 45 4.2.4 Effect of alloying elements 50 4.2.5 Steady State behavior and corrosion rate.. 53 4.3 Summary 55 5 E F F E C T O F O X Y G E N O N T H E C O R R O S I O N B E H A V I O U R O F A L L O Y 625 A T E L E V A T E D T E M P E R A T U R E S 57 5.1 Experimental 58 5.1.1 Safety considerations 58 5.1.2 General considerations. 59 5.1.3 Oxygen diffusivity and concentration calculations 62 5.2 Results and discussion 65 5.2.1 Potentiodynamic polarization experiments 65 5.2.2 Linear polarization experiments 78 5.2.3 Electrochemical impedance measurements 84 5.2.4 Corrosion rate and mixed control 99 5.2.5 Oxygen solubility at S C W O conditions 104 5.2.6 Effective diffusion layer thickness 109 5.3 Summary 113 6 T H E R M O D Y N A M I C S OF T H E N i - H 2 0 - N H 3 S Y S T E M 116 iv 6.1 Calculations 116 6.1.1 Condensed N i phases 122 6.1.2 Ionic N i species 122 6.1.3 Ammonia 128 6.2 Pourbaix diagrams 129 6.3 Summary 145 7 C O R R O S I O N A T S C W O C O N D I T I O N S 146 7.1 Failure analysis 146 7.2 316 L Stainless steel reactor experiment 153 7.2.1 Experimental 153 7.2.2 Results and discussion 156 7.3 Summary 160 8 C O R R O S I O N O F N I O B I U M I N C O N C E N T R A T E D S U L F U R I C A N D H Y D R O C H L O R I C A C I D A T E L E V A T E D T E M P E R A T U R E S 161 8.1 Experimental 164 8.1.1 Immersion tests 164 8.1.2 Polarization experiments 166 8.1.3 Pourbaix diagrams and thermodynamic calculations 167 8.2 Results and discussion 172 8.2.1 Pourbaix diagrams 172 8.2.2 Immersion tests 174 8.2.3 Polarization experiments 177 8.3 Summary 186 9 C O N C L U S I O N S A N D R E C O M M E N D A T I O N S 188 9.1 Conclusions 188 9.2 Tentative findings 189 9.3 Recommendations 190 R E F E R E N C E S 192 A P P E N D I C E S 207 Appendix A - High temperature electrochemical apparatus 207 Appendix B - Concentrations and diffusivities of O2 in the autoclave 215 Appendix C - Examples of EIS curve fitting results 216 Appendix D - Oxygen concentration and fugacity at S C W O temperatures and at constant pressure: PT0TAL - 250 bar 220 Appendix E - Matlab code used to generate R K S results 221 LIST OF TABLES Table 2-1 A l l o y compositions (wt%) 12 Table 5-1 Theoretical reduction potential E/ at constant pH, equilibrium constant kj for reaction 2, estimated p H as a function of temperature and E\ =f(pH(T)) 70 Table 5-2 Potential and kinetic data for the oxygen reduction reaction 74 Table 5-3 Calculated values of 8 based on measured iL data 76 Table 5-4 Model fit parameters for the simplified circuit model assumed (Figure 5-13) to represent the passive fi lm and interfacial behaviour 90 Table 5-5 Parameters derived from EIS measurements, according to equations 5-13 and 5-19(5=0.026, \u00C2\u00A3 = 1 .10x l0 6 V.cm~\ a = 0.878). S is the standard deviation for Rp. 92 Table 6-1 Thermodynamic data used for Pourbaix diagram calculations^ 118 Table 6-2 Some of the reactions considered and their calculated standard electrode potentials 120 Table 7-1 Metals content in feed, in sampling ports throughout the reactor (distance from failure) and in the effluent (Eff) , in ug/g [Vera-Perez 2002] 147 Table 7-2 Concentrations of dissolved metal after S C W O experiment in 316 L tube. 157 Table 7-3 Coupon corrosion rates after exposure to ammoniacal sulphate solution at 653-673 K , C 0 ; ~ 3 m 160 Table 8-1 Estimated activities of H 2 S 0 4 solutions at 25, 75 and 95\u00C2\u00B0C 170 Table 8-2 Estimated activities of HC1 solutions at 25, 75 and 95\u00C2\u00B0C 171 Table 8-3 Chemical potentials, of species at 25, 75 and 95\u00C2\u00B0C 172 v i Table 8-4 Corrosion rates for N b in urn/yr from immersion tests 176 Table 8-5 Potentiodynamic polarization data in sulphuric acid solutions 181 Table 8-6 Potentiodynamic polarization data in hydrochloric acid solutions 181 Table 8-7 Cathodic Tafel slopes, fic, in H 2 S 0 4 and HC1 solutions (mV/dec) 184 Table C - l Calculated data at 298 K 216 Table C-2 Calculated data at 448 K 216 vi i LIST OF FIGURES Figure 2-1 S C W O flowsheet as implemented at the U B C / N O R A M reactor 13 Figure 2-2 Properties of water at 240 bar. The property transition point is shown by the dashed line while the region of dense supercritical water is highlighted by the grey zone. After Kritzer 2001 14 Figure 2-3 Schematic Evans diagrams (a) thin film passivation, (b) thick film passivation 18 Figure 2-4 Schematic of bi-layered film structure on metal surface. Modified from Macdonald 1992 '. 20 Figure 2-5 The P D M : (1) Injection of a lattice cation with elimination of a cation vacancy (2) Injection of a lattice cation and creation of an oxygen vacancy (3) Oxidative dissolution and creation of a cation vacancy (4) Elimination of an oxygen vacancy, anion injection (5) Chemical oxide dissolution. After Macdonald 1992 27 Figure 2-6 Stability fields for N i O and Cr203 in S C W O . Transpassive dissolution (upwards arrow) only results in severe loss when p H is acidic. Modified from Kritzer 2000A 32 Figure 4-1 Pourbaix diagram for the Ni-NH3-Fi20 system at 25\u00C2\u00B0C 40 Figure 4-2 Pourbaix diagram for the Cr -B^O system at 25\u00C2\u00B0C 42 Figure 4-3 Effect of oxidant on alloy 625 at p H 10 in 0.2 m SO2' I 1.6 m NH3 43 Figure 4-4 Potentiodynamic polarization curves in selected de-aerated environments. Effect of p H . .... 45 viii Figure 4-5 Experimental points and linear fit lines for p H versus corrosion potential and oxidation peaks 1 and 2 48 Figure 4-6 Passive current densities for alloy 625 in solutions of ammonium sulphate and solutions of sodium sulphate 50 Figure 4-7 Polarization curves for alloy 625, N i , Cr and M o in de-aerated solution at p H 10 52 Figure 4-8 Polarization curves for alloy 625, N i , Cr and M o in naturally aerated solution atpFf 10 53 Figure 4-9 Polarization curves for alloy 625 under direct scan and steady state scan conditions - naturally aerated 55 Figure 5-1 Polarization curves for alloy 625 in ammoniacal solution from 298 to 473 K at absolute initial autoclave loading of 1 A T M or PT0TAL 2 g s K = 0 bar 67 Figure 5-2 Polarization curves for alloy 625 in ammoniacal solution from 298 to 473 K at initial autoclave l o a d i n g f r o r / j i 2 g g / , = 13.8 bar 68 Figure 5-3 Corrosion potentials ( E c o r r ) , theoretical (E,) and measured (E t ) transpassive potential as a function of temperature at initial oxygen loading PT0TAL^%K = 0 bar 69 Figure 5-4 Corrosion potentials ( E c o r r ) , theoretical (E,) and measured ( Et) transpassive potential as a function of temperature at initial oxygen loading PT0TAL 29SK =13.8 bar ... 71 Figure 5-5 Exchange current density as a function of temperature, pressure and sparge time. Circled values are considered most accurate due to smaller Tafel slopes 77 Figure 5-6 Linear polarization as a function of temperature at initial oxygen loading PIOTAL,79SK =6.89 bar 79 ix Figure 5-7 Linear polarization as a function of temperature at initial oxygen loading PTOTAL,29BK = 13.79 bar 80 Figure 5-8 Linear polarization as a function of temperature at initial oxygen loading PTOTAL,29BK =20.68 bar 81 Figure 5-9 Corrosion rates as measured by various methods (PD - Potentiodynamic, L P - Linear Polarization, EIS - Electrochemical Impedance Spectroscopy) and at various Figure 5-10 Nyquist plots taken at peak-to-peak voltage excitation amplitude of 10 m V vs. Ecorr (100 kHz , left most point - 10 mHz, right most point), and at various temperatures. PT0TAL 29SK = 27.58 bar 86 Figure 5-11 Bode plots at Ecorr, PT0TAL 29SK = 27.58 bar. (A) Magnitude and (B) Phase angle 87 Figure 5-12 Randies plot showing the real impedance as a function of a>~]'2 and temperature - the Warburg coefficient for cation vacancy transport was calculated by linear regression and is shown in the legend 88 Figure 5-13 Circuit scheme used for fit to experimental data. Interfaces 1 and 2 represent the m/bl and bl/s interfaces respectively 89 Figure 5-14 Calculated values of the steady-state barrier layer thickness (Lss) and film (bl + ol) resistance (R b ) as a function of temperature. Lss was calculated using the parallel plate formula, equation 4-17 92 Figure 5-15 Mott-Schottky plot for alloy 625 as a function of temperature at various ECorr values which were varied according to oxygen sparge time at 298 K 94 x corr Figure 5-16 Dependence of Ecorr on total pressure for all EIS data- decreasing EC: values above ~ 40 bar substantiate the assumption of diffusion control at high temperature 96 Figure 5-17 Predicted effect of pressure and temperature on Ecorr at PT0TALJ > 40 bar -T3 > T2 > T, and P3 > P2 > P, - it is expected that transpassive corrosion occurs readily at this pressure 97 Figure 5-18 S E M micrograph showing corrosion morphology after EIS testing through the temperature range 298 to 473 K , PT0TALag%K = 27.89 bar. I G A is apparent 99 Figure 5-19 Normalized corrosion rate as a function of temperature, pressure and method of measurement 102 Figure 5-20 Liquid and gas phase densities at 250 bar as calculated by the R K S for a 95 mol% H 2 0 , 5 mo l% 0 2 solution 106 Figure 5-21 Oxygen solubility - Henry's Law constant as a function of temperature and method of calculation 108 Figure 5-22 S E M micrograph of outer chromium oxide layer (precipitated) on the surface of alloy 625 exposed to ammoniacal sulphate solution at 250 bar, -633 K 112 Figure 5-23 S E M micrograph of inner surface of alloy 625 tubing exposed to ammoniacal sulphate solution at 250 bar, ~ 653 K 113 Figure 6-1 Aqueous N i speciation at 373 K , using Osseo-Asare N i model. [ M 2 + ] r = l x l 0 - 6 m , [NH3]T=1.5m 130 Figure 6-2 Aqueous N i speciation at 473 K , using Osseo-Asare N i model. [Ni2+]T =\x\0'6m, [NH3]T =1.5m 131 xi Figure 6-3 Aqueous N i speciation at 573 K , using Osseo-Asare N i model.. [Ni2+]T = l x l 0 ~ 6 m , [NH3]T =l.5m 132 Figure 6-4 Aqueous N i speciation at 653 K , using Osseo-Asare N i model. [Ni2+]T = l x l ( T 6 m , [NH3]T =l.5m 133 Figure 6-5 Aqueous N i speciation at 653 K , using original Helgeson model (eqn.6-9). [Ni2+]T = l x l ( T 6 m , [NH3]T = l.5m 134 Figure 6-6 Pourbaix diagram for the N i - H 2 0 - N H 3 system, 373 K . [Ni2+ ]T - 1 x 10\"6 m , [NH3]T =1.5m. Line c represents the neutral pH for water while line d represents the NH3 INHl equilibrium '. 135 Figure 6-7 Pourbaix diagram for the M-H2O-NH3 system, 473 K . . [Ni2+ ]r = 1 x 10~6 m , [NH3 ]T = 1.5m. Line c represents the neutral pH for water while line d represents the NH31NH* equilibrium 136 Figure 6-8 Pourbaix diagram for the N i - H 2 0 - N H 3 system, 573 K . [ M 2 + ] T = 1 x 10\"6 m , [NH3]r = 1.5m . Line c represents the neutral pH for water while line d represents the NH3 /NH+4 equilibrium 137 Figure 6-9 Pourbaix diagram for the N i - H 2 0 - N H 3 system, 653 K . [ M 2 + ] T = 1 x 10\"6 m , [NH3 ]T =l.5m. Line c represents the neutral pH for water while line d represents the NH31NHl equilibrium 138 Figure 6-10 Stability constants as a function of temperature (LogK, T ) and method of calculation. Letowski's data has been superimposed: LogKlT = - 5 5 8 . 4 2 / 7 + 9.6953-0.01757 140 x i i Figure 6-11 Effect of ammonia partial pressure at 653 K , 250 bar. Partial pressures are in bar and Hm =5 bar.kg.mof 1. Equations 6-13 through 6-16 143 Figure 7-1 Wal l thickness measurements and temperature profile near failed tubing location for P H I [Vera-Perez 2002] 147 Figure 7-2 (a) S E M micrograph of epoxy mounted alloy 625 tube section exposed to ammoniacal sulphate solution at ~ 635 K , 250 bar (b) E D X map showing alloy (pink) and Cr enriched ol (blue) and (c) Cr enriched ol (d) line scan showing N i depletion and Cr enrichment in ol , slight M o enrichment is also apparent 148 Figure 7-3 (a) E D X map overlaid on an S E M micrograph of the oxide layer on the inside surface of alloy 625 tubing exposed to ammoniacal sulphate solution at ~ 640 K , 250 bar. The oxide is visibly cracked, (b) ~ 642 K (c) ~ 645 K (d) ~ 645 K (magnified) showing intergranular attack and M o enrichment at the ol/s interface (yellow colour) 151 Figure 7-4 (a) Macroscopically visible surface roughening and only patches of oxide (circled area) are visible at ~ 650 K . (b) 20X optical microscope picture of the bakelite mounted cross section ~ 650 K . Wal l thinning is obvious, (c) Approximately 658 K : point of failure. Oxide is either completely removed (not shown here) or deep I G A is apparent, (d) ~ 658 K obvious wall thinning and ductile elongation are visible 152 Figure 7-5 Schematic of bench-top S C W O apparatus 153 Figure 7-6 Sectional view of test cell coupon assembly 155 Figure 7-7 High magnification morphology of deposits on the surface of coupons exposed to 653-673 K S C W O environment. E D X spectra reveal that the porous scale is mostly Fe oxide and Cr oxide removed from the 316 L reactor tube 158 Figure 7-8 Morphology of porous scale\":.. 159 xin Figure 8-1 Corrosion rates in um/yr for Nb in H2SO4 solutions at temperatures below the boiling point. Circle diameter reflects corrosion rate value rather than an area on the graph i.e. each corrosion rate value sits on a discreet point in the concentration-temperature plane 162 Figure 8-2 Corrosion rates in um/yr for N b in HC1 solutions at temperatures below the boiling 100\u00C2\u00B0C 163 Figure 8-3 Optical micrograph of the Nb microstructure after etching 165 Figure 8-4 Pourbaix diagram for the ND -H2O system at 25, 75 and 95\u00C2\u00B0C. \u00E2\u0080\u00A2 represents the corrosion potential for the 75\u00C2\u00B0C polarization tests and represents 95\u00C2\u00B0C tests. Lines a and b represent hydrogen evolution and oxygen reduction, respectively 173 Figure 8-5 Nb after two weeks immersion at (a) 95\u00C2\u00B0C in 20% H 2 S 0 4 , (b) 95\u00C2\u00B0C, 80% H2SO4, (c) 75\u00C2\u00B0C, 40% H2SO4, (d) 95\u00C2\u00B0C, 40% H 2 S 0 4 , (e) 95\u00C2\u00B0C, 20% HC1 and (f) 95\u00C2\u00B0C, 38% HC1 178 Figure 8-6 Polarization curves for Nb in 20, 40 and 80 % H 2 S 0 4 at 75\u00C2\u00B0C 180 Figure 8-7 Polarization curves for Nb in 20, 40 and 80 % H 2 S 0 4 at 95\u00C2\u00B0C 180 Figure 8-8 Polarization curves for Nb in 20 and 38 % HC1 at 75 and 95\u00C2\u00B0C 183 Figure A - l Working electrode 207 Figure A - 2 Working electrode sample detail 208 Figure A-3 Autoclave and working electrode assembly 209 Figure A - 4 Autoclave head and detail o f electrodes 210 Figure A-5 Autoclave Assembly 211 Figure A - 6 Working electrode sample showing relative size (Staedler felt tip pen) and internal thread (small hole) 212 xiv Figure A - 7 Working electrode showing used P T F E seal, connecting rod and sample. 212 Figure A-8 External pressure balanced working electrode assembly 213 Figure A - 9 Assembled autoclave at work 214 Figure C - l Complex capacitance plot showing measured and calculated capacitance at 298 K 217 Figure C-2 Bode plot showing the measured and model calculated data for the EIS spectrum at 373 K 218 Figure C-3 Nyquist plot showing the measured and model calculated data for the EIS spectrum at 448 K ;.. 219 xv ACKNOWLEDGEMENTS For their trust, encouragement, support and technical input throughout the course of this study, I would like to express gratitude to my supervisors, Dr. Akram Alfantazi and Dr. Steven Rogak. I would also like to acknowledge Dr. Desmond Tromans, Dr. David Dreisinger, Dr. Richard Branion, Dr. Berrend Wassink and Dr Jianming L u for many helpful discussions. Colleagues such as Dr. Andy Jung, Dr. Tawfik Ahmed, James Vaughan and John Skrovan have also helped enormously in flushing-out some of the ideas and experimental techniques presented below. I am indebted to Ross M c L e o d for his design input with regards to the experimental apparatuses used in this work. Other technical staff who have helped in various ways include Carl N g , Dave Torrok, Mary Mager and Serge Milaire. Finally, this work would not have been possible without the financial support of the Natural Sciences and Engineering Research Council o f Canada. xv i DEDICATION Pour mon pere qui me manquera toujours. Pour ma mere a qui je dois tout. Pour mon frere qui poursuit son propre chemin. A n d for my wife who is an inspiration. xv i i 1 INTRODUCTION 760 kg of iron has been corroded as you read this. This sentence was presented as a poster at a corrosion exhibition in Brussels in 1937 [Pourbaix 1973]. Undoubtedly, the rapid pace of the post Industrial Revolution development of corporate and national infrastructure required that attention be placed on the issue of corrosion and that a concerted effort be put towards understanding and mitigating it. The problem of corrosion is, however, not a modern preoccupation. Humans have been interested to varying degrees with the prevention of corrosion since the application of metals to practical purposes. The Greek Historian Herodotus (fifth century B.C. ) and the Roman naturalist Pliny the Elder (first century B.C. ) mention the use of tin for the protection of iron [Tomashov 1966]. It is likely that we have been studying this problem to an increasing degree since the Bronze Age, the earliest stages of which can be traced to southern Russia circa 3000 B C . Indeed, Chinese bronzes (weapons and mirrors) from the Shang and Chou dynasties (1600-221 B.C.) are known to have been intentionally coated with a glassy silicate which has resulted in corrosion protection despite three millennia of exposure to soil [Soto 1983]. Despite this impressive achievement in materials science, anti-corrosion technology remained entirely empirical until the 1920s [Crolet 2003]. The earliest documented corrosion experiments of a scientific nature date back to the mid 18 t h century. According to Tomashov, Lomosonov was the first to notice the passivation of iron in nitric acid (1743-1750). Later, in what is generally acknowledged as the first series of experiments on passivation (or \"curious appearances\"), Kei r 1 observed that when he put \"some pieces of clean fresh iron wire into a concentrated and red fuming nitrous acid (...) no apparent action ensued\" and that \"it was altered\" and \"rendered incapable of being attacked either by a phlogisticated solution of silver, or by dephlogisticated nitrous acid\". This altered iron was \"acted on by a dilute phlogisticated nitrous acid; but not by a red concentrated acid, which is known to be highly phlogisticated\" [Keir, 1790]. Simply stated, and without the benefit of phlogiston, Kei r found that when immersed in concentrated nitric acid, iron does not readily corrode but that when immersed in dilute nitric acid, corrosion occurs quite readily. Ke i r stated at the end of his paper that \"the explanation of these phenomena w i l l be attempted in the subsequent Papers which I propose to present on this subject to the Society.\" [Keir, 1790] Uh l ig notes that this was never done and that \"early attempts at explanation by others were vague and vascillating\" [Uhlig 1978]. In a letter to Faraday in 1836, Professor Schoenbein from the University of Basel reported several experiments dealing with the passivation of iron and the galvanic effects between active and passive iron. In fact, the first use of the term \"passive\" is attributed to Schoenbein [Uhlig 1978]. At Schoenbein's behest, Faraday performed some of the most interesting, perspicacious and now famous experiments on the passivation of iron. Aside from confirming the remarks by Kei r with regards to strong and dilute nitric acid, he also noted that while Fe in 1.3 to 1.35 sg nitric acid corroded readily, i f made to contact Pt wire immersed in the same solution, the corrosion stopped immediately [Faraday 1836]. Moreover, Faraday describes the most \"valuable result\" of this work as being the \"proof that voltaic electricity is due to chemical action not to contact\". Through measurement of induced current in a Passive Fe || Nitric A c i d || Pt | Active Fe || Nitric A c i d || Pt cell he was able to demonstrate that the passive film was not an insulator [Faraday 1836]. With these 2 experiments and the associated analysis, the oxide film theory upon which we have based the most modern aspects of corrosion science was born [Faraday 1836]: \" M y strong impression is that the surface of the iron is oxidized, or that the superficial particles of the metal are in such relation to the oxygen of the electrolyte as to be equivalent to an oxidation; and that having thus their affinity for oxygen satisfied, and not being dissolved by the acid under the circumstances, there is no renewal of the metallic surface, no reiteration of the attraction of successive particles of the iron on the elements of successive portions of the electrolyte, and therefore not those successive chemical actions by which the electric current (which is definite in its production as well as its action) can be continued\" It is only recently through the work of Pourbaix in the 1960s, that science and the study of corrosion have enabled a better understanding of the 18 t h and 19 t h century observations [Macdonald 1999]. Through the use of electrochemical thermodynamic calculations it is now known that the relative immunity of iron to concentrated nitric acid is due to a metastable oxide phase forming on the surface of the metal, resulting in a barrier layer, or passive film, between the bare metal and the solution [Macdonald 1999]. This explanation, on the surface is quite adequate but the reader w i l l notice that it does not add much to the discussion insofar as it very much resembles the Faraday analysis. There is then a certain degree of obscurity which surrounds such an analysis and serves to highlight the complexity of corrosion phenomena. For example, one might wonder what the real stability limits for this oxide are (distinction made here between the thermodynamic stability and the real observed meta-stability). Alternatively, the student might be interested in the chemical versus electrochemical stability of the film. Or perhaps, from a kinetic point of view: how fast does this passive film form? There is also the question of steady-state behaviour - does the fi lm grow infinitely? What about the 3 nature of the film? What exactly is it made of? What is its stoichiometry? How conductive is it? Is it porous? How does it form? How does it break down and how does it interact with the environment? Is it susceptible to certain species? What about the thermal and physical stability of the film? These are only some of the questions that have occupied the corrosion engineer and scientist for the last two centuries and they are of the greatest importance in designing and understanding the materials and processes we so heavily rely upon. They are extremely complicated questions which are highly dependent on the entire metal-environment system to which answers are more often than not unavailable. The consequence of these questions and of the phenomenon of passivation cannot be understated. Indeed, Macdonald has written that \"humankind has been able to develop a metals-based civilization primarily because the reactive metals (Fe, N i , Cr, A l , T i , Zr, ...) exhibit extraordinary kinetic stabilities in oxidizing environments\" [Macdonald 1999]. Crolet has said that \"corrosion is often qualified alternatively as an emerging science or a stabilized technology, which is quite obviously incompatible, but also equally wrong in each case\" [Crolet 2003]. It is my opinion that these opposing views stem from two facts: (1) the science of corrosion is still largely empirical in nature and (2) corrosion is an obvious Second Law process. To illustrate this, Speller notes in 1951 in his introduction to the subject, that \"the underlying facts which govern corrosion are now fairly well established\" and that \"al l metals when exposed to the elements have a certain inherent tendency to revert back to one or another of the forms in which they were originally found in the earth\"; but he goes on further to say that \"much more remains to be learned as to the operation of many of the factors\" [Speller 1951]. This is still the dilemma facing the corrosion community. Everyone understands corrosion to 4 the extent that everyone understands ageing, for example, but the underlying factors which drive the process are still not clear. Nevertheless, the costs involved in corrosion losses cannot be ignored and warrant an increased effort into the development of corrosion science. A 1998 congressional study into corrosion costs and preventive strategies in the U . S . estimated that the impact of corrosion on the U.S economy was on the order of 3.1% of G D P or some $276 bil l ion dollars [Koch 2001]. Estimates for the cost in Canada in 2004 were on the order of C A D $30 bil l ion [Shipilov 2005]. While cost reduction is the largest driving factor for studying corrosion, from a scientific and socio-economic standpoint it is perhaps not the most important. In Speller's book, a mining engineer turned depression-era President is quoted as saying: \"It is only through the elimination of waste and the increase in our national efficiency that we can hope to lower the cost of l iving, on the one hand, and raise our standards of living, on the other. The elimination of waste is a total asset. It has no liabilities.\" Herbert Hoover, 1925. Considering the energetic, environmental and human costs expended to make the materials that we use at ever increasing rates, the loss of these materials to corrosion is an inexcusable waste. With the impending issues related to climate change, resource and oi l scarcity and a growing population, any improvements in the efficiency of our materials consumption is a positive step. This thesis does not purport to offer any substantive assuagement to the global problem of corrosion. Instead, I hope it might offer some insight into how best to preserve certain, industrially relevant materials under very specific conditions related to a technology called Supercritical Water Oxidation (SCWO). 5 S C W O [Modell 1989, 1993] uses supercritical water (T > 374\u00C2\u00B0C, p > 221 bar) as a reaction medium to destroy many types of military and civilian waste which are environmentally harmful and typically difficult to dispose-of. Under S C W O conditions organics become entirely miscible with the solvent supercritical water which acts as a non polar dense gas. With the addition of oxygen or hydrogen peroxide the oxidation reaction proceeds rapidly. Products of the oxidation of hydrocarbons are H2O and CO2 while hetero-atoms are converted to inorganic compounds such as acids and salts. Specifically, compounds containing sulfur, chlorine or phosphorus, for example, w i l l oxidize to sulfuric acid (H2SO4), hydrochloric acid (HC1) or phosphoric acid (H3PO4) within less than a minute. Under these conditions of high temperature and high pressure and low p H (often chemical feeds are acidic even before treatment), corrosion has become a major concern in the development of S C W O technology. S C W O has earned a reputation for an almost insurmountable materials challenge. It has been said that no one material currently exists which meets all the requirements of the different conditions under which a S C W O reactor might function [Mitton 2000]. From preheat to high subcritical, to supercritical, the aggressive nature of the S C W O environment differs quite dramatically. Traditionally N i - C r alloys, particularly alloy 625 (UNS N06625) , have seen widespread service in S C W O reactors due to their demonstrated corrosion resistance and mechanical properties at high temperature and pressure in other successful processes. A s such, research has been mainly focused on the problem of corrosion of these materials. Furthermore, the overwhelming majority of work has focused on assessing the corrosion characteristics of S C W O systems under aggressive acidic feeds (HC1 and H 2 S 0 4 ) [Mitton 2000], [Mitton 2002], [Gonzalez], [Boukis 1997], [Downey 1995], [Konys 1999], [Kritzer 2000], [Kritzer 1999], [Fodi 1997], [Kritzer 1998], 6 [Watanabe 2002]. This is of course in large part due to the phenomenon of acid formation but also because the few studies performed under neutral to alkaline conditions, have shown little or no corrosion (in Na2S04 for instance [Kritzer 1998]). However, recently, the failure of two sections of alloy 625 tubing at the U B C / N O R A M S C W O pilot plant has shown that severe material degradation can be caused by mildly alkaline (pH29SK =10) ammoniacal sulphate solutions which were not anticipated to be corrosive. This solution has therefore been used as the main testing medium throughout the work presented below. A s there is very little information directly pertinent to the corrosion of Ni-based alloys or stainless steels in ammoniacal solutions, the following pages w i l l endeavour to present some aspects of this problem with a particular emphasis on alloy 625. For reasons to be discussed later, Ni-based alloys are generally preferred at high temperatures and pressures, but there was some indication that stainless steels might outperform them in ammoniacal solutions. While corrosion studies under supercritical water conditions w i l l eventually be presented, due to the paucity of information regarding the behaviour of these alloys in ammoniacal systems, this study w i l l begin by characterizing their electrochemical response at room temperature and then follow with results from successively higher temperatures. The effects of oxygen were also identified as important so a parallel emphasis is placed on the effect of oxidant concentration on the corrosion characteristics of alloy 625. The following section (Chapter 2) w i l l discuss some of the pertinent literature and serve to frame the discussion as well as set out the objectives of the work. Chapter 3 w i l l present the objectives of this work. Chapter 4 discusses the results of studies performed 7 at room temperature on alloy 625 and its constitutive elements at varying oxidative and p H conditions. Chapter 5 discusses the results of high temperature electrochemical work performed on alloy 625 to characterize the passive film and its response to oxygen concentration. Chapter 6 presents the results of thermodynamic calculations for the N i -NH3-H2O system at high temperatures and Chapter 7 relates the results of S C W O corrosion experiments. Chapter 8 presents the results of work aimed at showing that Nb may constitute a viable alternative to N i - C r alloys under oxidative high temperature conditions. Finally, Chapter 9 presents the conclusions and recommendations for future work in this area. There is no doubt that this contribution constitutes a \"drop in the proverbial bucket\" of engineering challenges which face the corrosion community. Notwithstanding, I have had the opportunity, through the course of the research presented here, to explore not only some aspects of corrosion science as it relates to high temperature, high pressure aqueous corrosion but also to apply this analysis to S C W O which provides considerable environmental benefits over traditional incineration for organic waste treatment. In so doing I hope to have taken up, in some small way, the challenge most aptly presented by Hoover some 81 years ago. 8 2 LITERATURE REVIEW 2.1 The Process of SCWO 2.1.1 The U B C reactor The process of S C W O , as used in the U B C reactor, is roughly schematized in Figure 2-1, p. 13. The waste feed stream is added to water and pumped into the reactor by a high pressure piston pump. The effluent end of the reactor is fitted with a back-pressure regulator which modulates pressure as required, usually between 220 and 300 bar. Oxygen is added to the mix by means of an oxygen pressure booster. Typically oxygen is added in large excess with respect to the organic to be degraded (ex: 200%) to ensure full oxidation of the influent. Hydrogen peroxide (H2O2) injection is an alternate route to adding oxidizing agent but some authors have reported that this method requires the use of a catalyst (platinum is one option) to decompose the peroxide as the kinetics of the hydrogen peroxide breakdown are slow. In general, the use of H2O2 is limited to laboratory scale units as it is more portable than direct oxygen injection but largely more expensive. Initially the feed passes through a recuperative heat exchanger and is then preheated up to the supercritical temperature (two preheater sections are employed to this end at U B C , P H I and PH2). In the reactor, the spontaneous oxidation of the organic feed stream raises the temperature to levels well in excess of 773 K . This reaction takes place very quickly with typical residence times of one minute or less. Salts in the stream precipitate because their solubility above the supercritical temperature decreases 9 substantially. Problems with respect to reactor fouling ensue but that is an issue beyond the scope of this work. A t the outlet of the reactor the effluent passes once again through the heat exchanger ( R H X ) and is then cooled to approximately 313 K in the process cooler. A gas separator is the last component of the system in Figure 2-1. Other reactor schemes have been employed such as tank or transpiring wall designs. Both these designs reportedly have the advantage of reducing salt deposition. In the case of the tank reactor a cool zone is maintained at the bottom where salts are dissolved and then removed as brine. The transpiring wall reactors use a porous pipe which is rinsed with water to prevent salts from depositing and plugging the reactor [Schmieder 1999]. 2.1.2 Advantages of SCWO With temperatures in the reactor section between 773 and 873 K , residence times for 98% conversion efficiencies are on the order of less than 1 min [Dinjus 2004]. Unlike traditional incineration which is known to release dioxins and N O x , the temperatures used in S C W O are low enough to ensure that organic-bonded N is converted to mostly N2 and minor amounts of N2O [Kritzer 2001]. One of the other major advantages to S C W O is that it can operate as a closed system such that reactants and products are easily isolated. With reported destruction efficiencies as high as 99.9999% it is then ideally suited for treating such wastes as radioactive sludge or military chemical agents such as V X or Sarin gas. The U S Department of Defense has reportedly spent a significant amount of resources investigating the application of S C W O to such purposes through General Atomics and Sandia National Laboratory. Other investigated systems with high destruction efficiencies include ethanol, toluene, trichloroethylene and phenols. Particularly 10 refractory compounds include pyridine, urea, ammonia and acetic acid [Schmieder 1999]. The issue of ammonia destruction is relevant to this thesis and w i l l be discussed in the following section. 2.1.3 Ammonia persistency Many of the wastes potentially treatable by S C W O either include ammonia at the onset (municipal waste or organic sludge which could depend on ammonia for liquefaction), or result in its formation as a refractory and oxidation rate-determining step. For instance Goto et al. have found that ammonia decomposition in sewage sludge after 4 hours of reaction at 723 K at 300 bar was 30 % [Goto 1999]. Ding et al. have reported approximately 30-40 % NH3 conversion over a Mn02/Ce02 catalyst at 723 K at 280 bar. Del l 'Orco et al. have reported conversion of less than 10 % after exposure to S C W O environments at 723 K , 300 bar and 10 sec residence times [Dell 'Orco 1997]. Hell ing and Tester have said that \"ammonia is very difficult to oxidize below 813 K at residence times of 6-13s\" and have reported conversions o f - 6 % under such conditions [Helling 1988]. Perhaps the most cited work in this area is that of Webley et al. who have summarized by saying that the ammonia oxidation rate was mainly dependent on reactor wall catalysis and that ammonia oxidation occurred to a maximum of ~ 40 % at 973 K , 246 bar, 16 sec residence times over a packed-bed reactor with a high surface to volume ratio (S/V) of - 600. Conversions of - 10% were reported for a tubular reactor with S/V - 20 under the same P and T conditions but with 10s residence times [Webley 1991]. Segond et al. found similar results with regards to the S/V ratio. Generally conversion rates of > 50% were obtained at residence times of - 30s at 873 K , 245 bar. Up to 95 % conversion was obtained at 1 min residence times for S/V = 4 [Segond 11 2002]. For purposes of comparison, the S/V ratio for the U B C reactor is roughly 6.5. A s such, oxidation of N H 3 is not expected to be fast until approximately 873 K . It should also be noted, and this is important, that none of these authors have reported corrosion damage to their 316 L , Inconel 625 or Hastelloy C-276 reactors (cf. Table 2-1). This is one of the reasons why experiments performed at U B C in the presence of ammonia were not expected to result in significant corrosion. Discussion on this point w i l l follow. Table 2-1 A l l o y compositions (wt%) Cr Mo Fe Nb W Ni 625 22.34 8.48 4.53 3.51 60.20 (20.0-23.0) (8.0-10.0) (5.0 max) (3.15-4.15) Ba l . C-276 16.11 15.99 5.76 3.85 56.42 (14.5-16.5) (15.0-17.0) (4.0-7.0) (3.0-4.5) Bal . C-22 21.31 13.55 3.52 3.17 57.52 (20.0-22.5) (12.5-14.5) (2.0-6.0) (2.5-3.5) Bal . 316L - SS (16.0-18.0) (2.0-3.0) Bal . (10.0-14.0) 12 Pressure regulator Oxygen booster Oxygen tank Steam Water Waste To drain Orifice meter Vent Si\u00E2\u0080\u0094 j Pulsation damper 1 H 20 High-pressure pump H 20 Backpressure regulator Wate? o f f \" 9 a s RHX Process cooler Vent -A-Pressure regulator A Liquid Gas-liquid separator Preheaterl Preheater2 Test section| Tubular reactor r\ r\ r\ r\ r\ w w w w Quenched sampling port High-pressure nitrogen Figure 2-1 S C W O flowsheet as implemented at the U B C / N O R A M reactor 2.2 Properties of supercritical water Upon heating water i n a closed system, vapour density increases while liquid density decreases. If the pressure is allowed to reach P > 221 bar at a temperature of T > 647 K , the vapour density is equal to the liquid density and the result is supercritical water. This phase exhibits entirely different properties from the solid, liquid and gas phases. Supercritical water shows high gas-like diffusion rates along with high liquid-like collision rates. In a S C W O reactor, pressure is maintained constant from ambient temperature to supercritical temperatures. The change from high temperature water to supercritical water occurs quite rapidly and markedly at temperatures close to 647 K and pressures in excess of 221 bar. Figure 2-2 illustrates the variation in density with respect to temperature at 240 bar. 1400 1200 o ^ 1000 o ^ 800 v. 1 \u00E2\u0080\u0094 1 jj> ~ 600 Q \u00C2\u00A3 400 200 0 0 385\u00C2\u00B0C 800 Temperature [\u00C2\u00B0C] Figure 2-2 Properties of water at 240 bar. The property transition point is shown by the dashed line while the region of dense supercritical water is highlighted by the grey zone. After Kritzer 2001. 14 The temperature at which the properties of water make the sub- to supercritical transition is greatly affected by pressure [Kritzer 1999A]. For instance, while water at 647 K , 240 bar has clearly exceeded the critical point, it has not yet undergone the distinguishing drop in density associated with lower pressure supercritical water -conversely, water at 658 K , 240 bar has made what one might describe as the \"property transition\". This distinction is often referred-to in terms of high-density supercritical water, before the property transition, and low-density supercritical water, after the transition. A s pressure is increased the property transition point of water moves to higher temperatures (at 250 bar it is closer to 673 K ) . Fluid density at S C W O conditions is in the region of 0.1g/cm 3. Similarly, the dielectric constant Of water drops from about 80 at room temperature to 1 or 2 at 250 bar and 723 K . The ion product of water also decreases from a value of 10\"1 4 to 10\"2 3 [Marshall 1981]. The result is that low density supercritical water resembles a non-polar dense gas in that it has a high solvency for hydrocarbons and organics whereas inorganics tend to precipitate [Mitton 2001]. In contrast, high density supercritical water displays less solvency for organics but a high solvency for inorganics. Most S C W O applications have operating pressures between 250 and 350 bar such that there are regions in the reactor exposed to high-density supercritical water which, due to its high solvency for salts, has been shown to be extremely corrosive [Macdonald 1993], [Kriksunov 1995]. In fact, it is generally acknowledged that corrosion rates reach a maximum in the high density supercritical region - between ~ 647 and 673 K from 221 to 250 bar [Liu 1994]. This maximum has been related to the competing effects of temperature on the corrosion rate constant and on the properties of the aqueous medium 15 such as density and the dielectric constant [Macdonald 2001]. Thus, while reaction kinetics increase with temperature, the solvent water is less able to dissociate acids or accommodate inorganic reaction products. A s a result, above the property transition temperature corrosion rates decrease substantially to the extent that they are generally acceptable in low-density supercritical water. A n interesting discussion on this topic is presented by Macdonald who has shown that as temperatures increase to the critical point, the contribution of the activation energy to the total reaction rate loses in relative importance to the contributions from solution density and acid dissociation (dielectric) [Macdonald 2004]. High concentrations of oxygen are required for high waste destruction efficiencies in S C W O reactors. In general, oxygen solubility decreases in water to a minimum somewhere in the region of 363 K . Above this temperature there is an exponential increase in the gas' solubility. A n y corrosion process requires both an anodic and cathodic reaction to proceed. The cathodic reaction consumes the electrons generated at the anode (presumably corroding tubing) and is considered the driving force in corroding systems. In the absence of a stronger oxidant, oxygen reduction is the cathodic reaction most likely to occur in S C W O systems. The amount of available oxygen and its solubility in the reactor environment is then a very important consideration in S C W O corrosion investigations. In high density supercritical water oxygen is almost fully soluble and this is a further aggravator to the increased malignancy of this medium [Kritzer 2000B]. A n y corrosion reaction can be limited by a build up of corrosion products such as salt films, oxides, or what some refer to as scales. In the case of N i - C r - M o alloys a mixed oxide layer containing chromium oxide and nickel oxide is possible. The 16 thermodynamic stability of these \"passive\" layers is very much dependent on p H and temperature. However, thermodynamic stability is not the only factor which controls passivation. In low temperature solutions an oxide may not be thermodynamically stable but the kinetics of dissolution are such that the transformation from oxide to soluble species is slow - the corrosion process is then under kinetic control. With higher temperatures, chemical reaction rates increase and the importance of kinetic control decreases [Kritzer 1999A]. Under such circumstances diffusion control of the corrosion process is expected to take over through rate controlling delivery of oxygen to the reacting surface, particularly i f oxides, salt films or scales build up on the surface as precipitates. Diffusion control is more easily modeled than kinetic control so that corrosion rates in the absence of oxides or scales might be predictable. This is one avenue which w i l l be examined in this thesis because it reveals the importance of salt film/oxide/scale solubility to the study of S C W O materials degradation. The apparent competition between kinetic control and thermodynamic stability is one of the limitations of Pourbaix diagrams and w i l l be discussed in the following section which deals with the thermodynamics and passivation of N i - C r - M o alloys. 2.3 Thermodynamics and passivation Since the inception of Pourbaix diagrams corrosion scientists and engineers have had access to a phenomenal tool for the interpretation and prediction of metal stability. There are 3 major categories delineated by Pourbaix diagrams, 1- active corrosion, 2-immunity, 3- passivation [Pourbaix 1966]. This last category is the focus of the following discussion and indeed a very important theme in this work. 17 Perhaps the most persuasive description of passivation is that which is inferred from a look at the Evans diagram (potential-log current) for a passive metal, as seen in Figure 2-3 (a). A definition for this type of passivity is as follows: \" A metal is passive if, on increasing its potential to more positive values, the rate of dissolution decreases, exhibiting low rates at high potentials (also known as thin film passivity).\"[Kelly 2003] With reference to Figure 2-3, there is clearly some process by which the corrosion rate of the material is slowed. Typically, N i , Cr or M o would exhibit this type of behaviour in solutions of H2SO4. There is, however, another type of passivity in the electrochemical sense, which is referred to as \"thick film passivity\" by K e l l y et al.(Figure 2-3(b)). The definition of which is: 18 \" A metal is passive i f it substantially resists corrosion in an environment where there is a large thermodynamic driving force for its oxidation.\"[Kelly 2003] M g or A l in water, or Pb in H2SO4 would behave according to this second definition. This issue of \"thin film\" or \"thick film\" is a bit misleading in some respect as it obfuscates the true underlying reason for the corrosion resistance which is l ikely due to the relative importance of one of the two sections which are known to constitute the passive film. Indeed, passive films are known to occur as bi-layer structures with a thin (1-3 nm) and homogeneous semiconductor-like film (barrier layer - bl) immediately adjacent to the metal (m) surface and a thicker precipitated, hydrated and porous mixed oxide-hydroxide layer (outer layer - ol) on the solution side [Macdonald 1992]. There is a substantial amount of experimental evidence to support this for the particular case of N i - C r alloys and also for other systems ([Bockris 1971] for iron, [Sato 1974] for nickel) . L loyd et al. and others ([Jabs 1997], [Maurice 1998] and [Machet 2002]) have effectively communicated that in N i - C r alloys the passive layers \"can be roughly divided into two sub-layers: the inner oxide layer consisting of Ni(II) oxide, Cr(III) oxide and O2', and the outer hydroxide layer consisting of Ni(II) hydroxide, Cr(III) hydroxide, Cr(VI), OFF, and H 2 0 \" [Jabs 1997]. A schematic of the bi-layered structure is shown in Figure 2-4. The ejection of metal cations through the bl to the solution side results in precipitation of the ol which, i f thick enough, can lead to the type of response shown in Figure 2-3 (b), often also referred to as \"spontaneous passivation\". 19 OUTER LAYER Porous SOLI TION Hyd rated Precipitated X = L X = 0 < Film growth Figure 2-4 Schematic of bi-layered fi lm structure on metal surface. Modified from Macdonald 1992. Whether passivation is \"thick f i lm\" or \"thin f i lm\", there is no doubt that it is the basis for the utility of most of today's technologically useful materials. Indeed, and quite paradoxically, we rely heavily on inherently reactive materials. For instance, N i , A l and T i w i l l spontaneously ignite in air i f ground into fine enough particles. Similarly, in a notable review on the subject of passivity entitled \"Passivity - the key to our metals-based civilization\", Macdonald has remarked that A l has an energy density of 22 kW.h.kg\" 1 whereas decane, a simulant for gasoline, has an energy density of 10 kW.h.kg\" 1 [Macdonald 1999]. N i and Cr form barrier layers ~ 1 to 3 nm in thickness. A t room temperature the electric field strength of these layers is somewhere in the region of 1 0 6 V .cm\" enabling ion migration through the layer. This migration corresponds to current densities of some 20 fxA/cm 2 or less at room temperatures [Strehblow 1995]. The chemical structure of oxide layers is not as straightforward or stoichiometric as \" C ^ C V ' or \" N i O \" . Barrier layers are highly doped semiconductors with multiple crystallographic defects while outer layers are a complicated configuration of oxides, hydroxides and hydroxy-oxides such that the stoichiometry usually used to describe these films is rarely observed. This is one of the many limitations of Pourbaix diagrams in that they require assuming the existence of a compound which, in reality, may differ from its ideal and assumed configuration. Moreover, as discussed above, reaction kinetics are not a factor in Pourbaix calculations though they may greatly affect the real behaviour of the metal. For instance, in acidic electrolytes the passive layer on metals such as N i , Fe and Cr is far from its domain of thermodynamic stability (well documented: [Pourbaix 1966], [Glasstone 1942]) and passivation in these cases is due to the slow dissolution kinetics of the metastable oxide. Furthermore, the issues of outer layer porosity and/or barrier layer adherence are not addressed in these diagrams. To illustrate the ambiguity associated with the label \"passive\" a quote from Pourbaix's Atlas of Electrochemical Equilibria in Aqueous Solutions may prove to be helpful: \" . . .the metal then tends to become coated with this oxide, hydroxide, hydride or salt which can, according to the circumstances, form on the metal either a non-porous film practically preventing all direct contact between the metal itself and the solution (in which case protection against corrosion is perfect), or a porous deposit which only partially prevents contact between the metal and the solution (in which case the protection is only imperfect). Understood in this way, passivation thus does not necessarily imply the absence of corrosion, \"(p. 71) 21 This final emphasis is not this author's but rather Pourbaix's - this point is paramount in understanding the discussion which follows because the surface of a metal may be both passive and corroding. For all the reasons stated, interpretation of electrochemical phenomena through reference to Pourbaix diagrams, especially for the N i - H 2 0 system, is a notoriously challenging endeavour. The addition of high temperature uncertainties related to imperfect thermodynamic data for both solid and, more so, ionic species complicates the problem. However, these interpretations certainly give a direction toward which to pursue further study and it is here that we shall start with a look at the passivation in the N1-H2O and Cr-H20 systems at high temperature. Through the last 40 years there has been a large body of work presented on the subject of Pourbaix diagrams for both the Ni-H20 and C r - H 2 0 systems up to ~ 573 K . The largest problem surrounding this body of work is that the ionic chemical potential data at T > 298 K is estimated from a variety of sources and using a variety of methods. Generally, the work can be summarized in this way: \u00E2\u0080\u00A2 There is a common movement of the diagrams to lower p H as temperature is increased, \u00E2\u0080\u00A2 and reactions are depressed to more negative potentials as temperature is increased With regards to the stability of ions, in 1972 Brook published the Pourbaix diagrams for N i and Cr up to 423 K and stated that \"since the domains of stability of (...) the lower valence oxides decrease with increasing temperature the domains for the oxyanions show a greatly enlarged range\" [Brook 1972]. Macdonald came to a similar finding in 1979 in his work on the nickel-water system in geothermal brines (up to 523 K ) : \"the stability 22 regions for cations at low p H values become more restricted as the temperature increases, whereas those for the anions (...) increase\" [Macdonald 1979A]. It should be noted that Brook used some estimated entropy data and his high temperature data was calculated using temperature coefficients derived from these same estimated ionic entropies. Macdonald's calculations used the linear ionic heat capacity approximation ( L I H C A ) based on Criss and Cobble's correspondence principle, which has been shown to be remarkably accurate up to ~ 573 K [Criss 1964 A , B ] , [Taylor 1978]; although the general applicability of the L I H C A past 473 K is not without controversy [Liu 2005]. Lee has presented Pourbaix diagrams for Cr and M o up to 573 K using the L I H C A and the main findings of pertinence to this work were (1) that the domain of stabilities of the oxides were generally reduced as temperature was increased, (2) that the domain of stability of chromium hydroxide ( C r O O H or Cr(OH)3) is wider than that of the anhydrous oxide (Cr203) and (3) that molybdenum dioxide is unstable above 523 K [Lee 1981]. A s temperatures are increased near the critical point of water, the L I H C A becomes unreliable so equations of state for the prediction of ionic standard partial molal thermodynamic properties have been developed by Helgeson, Kirkham and Flowers -known as the H K F model [Helgeson 1981]. Beverskog and Puigdomenech have calculated the Pourbaix diagram for N i from 298 to 573 K and used the H K F model for the aqueous species. They concluded that N i O is the most stable solid compound formed above 473 K [Beverskog 1997]. Kriksunov and Macdonald have also published Pourbaix diagrams at temperatures up to 723 K , 500 bar for N i which were generated using a computer program called Supcrt 92 and which employs the H K F model for its calculations. At 573 K , 500 bar their results are similar to those presented by Beverskog 23 although Kriksunov and Macdonald have chosen not to include \"hydrated forms of oxides and complex oxyanions, because of the great uncertainty in the extrapolated thermodynamic functions for those species in the supercritical region, and because low water densities in near-critical and supercritical systems at pressures below 500 bars favors decomposition of hydrated oxides to oxides and also greatly decreases the stability of ionic species\" [Kriksunov 1993]. Mitton et al. have also developed a recent E - p H diagram for N i - C r - M o alloy systems without the oxyanions (such as H N i C V ) [Mitton 2000] but based on experimental results from S C W O alloy 625 tubing exposed at 573 K . Their data is consistent with the previously cited literature and supports the following conclusions with regard to the thermodynamic stabilities of N i , Cr and M o at high subcritical temperatures: \u00E2\u0080\u00A2 A t neutral to high p H (approx. 5 to 14), N i O is stable. \u00E2\u0080\u00A2 A t neutral to low p H (approx. 2 to 5) some form of chromium oxide ( C r 2 0 3 or CrOOH) is stable. \u00E2\u0080\u00A2 M o is unstable. Based on these conclusions there is then a very wide p H range over which N i - C r and N i -C r - M o alloys such as alloy 625 are expected to be passive. However, when in the presence of strong oxidant, Cr(III) oxide has a tendency to corrode transpassively which effectively removes any protection in acidic environments. This phenomenon is discussed in the next section. 24 2.4 The point defect modei (PDM) and Transpassivity Transpassive corrosion has been accepted as the leading contributor to severe and general degradation of N i - C r alloys and stainless steels in oxidizing environments such as S C W O [Kritzer 1998, 1999A, 2000AJ3], [Hara 2002], [Laycock 1995], [Son 2002], [Wilson 2000]. Both classes of alloys rely on the stability of Cr(III) oxide for corrosion protection, particularly in the presence of acidic environments where N i O is not stable. The reaction proposed by Pourbaix is the following [Pourbaix 1966]: ICrO1; +\0H+ +6e\" = Cr203 +5H20 2-1 While this equation is adequate from an overall electrochemical standpoint, it does little to explain the mechanisms which lead to transpassive corrosion, particularly within the framework of the bi-layered structure mentioned earlier. These mechanisms could shed light on the expected corrosion morphology and electrochemical response of N i - C r alloys that are corroding in the transpassive region. This is.one area of fundamental corrosion research which has the potential for a very direct impact on service life and process design. Bojinov has said that \"transpassive dissolution is a complicated process including electron and ion transport through a surface film, multistep electrochemical reactions at the film/solution interface and transport of reaction products in the bulk solution\" [Bojinov 2000]. To gain a more fundamental perspective it is necessary to deviate from the discussion a bit and introduce the point defect model (PDM) . The P D M is the leading model to describe the kinetics of passivity. It was developed by Macdonald and co-workers in the early 1980s. There are three fundamental, well established, experimental observations which have framed the P D M and they are, very generally [Macdonald 1992]: 25 A . A s already mentioned, passive films are bi-layered and passivity is attributed to the inner b l which is a highly defective, defect semiconductor in which vacancies act as the electronic dopants [Macdonald 1999]. These vacancies are annihilated and generated at the interfaces - the metal (m)/barrier layer (bl) interface and the bl/solution (s) interface which are separated by a few nanometers. B . Under steady state conditions, the b l thickness and the logarithm of the currents both vary linearly with applied voltage C. The bl grows into the metal while the ol grows out from the metal surface Figure 2-5 is a schematic of the process of film growth and dissolution as described by the P D M where m = metal atom, MM = metal cation in a cation site, 00 = oxygen ion in anion site, Vfi = cation vacancy, V0 = anion vacancy, VM = vacancy in metal phase. 26 Metal Barrier layer Outer layer / solution 1 m + V, MM->Ms+(aq) + V*' +{S-Z)e< 3 2 m -+MM+ \u00C2\u00B1 V0+Ze' V0+H2O^Q0 +2H+ 4 MOx/2+%H+ ^Ms++ \u00C2\u00B1 H20+(S-z)e' 5 x = L x = 0 Figure 2-5 The P D M : (1) Injection of a lattice cation with elimination of a cation vacancy (2) Injection of a lattice cation and creation of an oxygen vacancy (3) Oxidative dissolution and creation of a cation vacancy (4) Elimination of an oxygen vacancy, anion injection (5) Chemical oxide dissolution. After Macdonald 1992. A t the m/bl interface, oxygen vacancies are generated and flow towards the bl/s interface. A t the bl/s interface cation vacancies are generated and flow towards the m/bl interface. These statements are equivalent to saying that metal cations diffuse from the m/bl to the bl/s interface and that the diffusion of oxygen anion occurs in the opposite direction. After studying Figure 2-5, it becomes clear that the diffusion of V0 results in oxide growth whereas the diffusion of V\u00C2\u00A3 is solely the result of cation ejection. There are two equations, (2) and (5), that determine growth and dissolution of the barrier layer - these 27 are termed lattice non-conservative processes as they result in the movement of the barrier layer boundaries relative to a fixed reference frame. Thus when reaction rates for (2) and (5) are equal, a steady-state film thickness is reached. Reactions (1), (3) and (4) are termed lattice conservative as they do not result in barrier layer boundary shift. Because the reactions presented in Figure 2-5 are not assumed but are in fact mandated by the widely accepted experimental observations (A through C), the P D M has the advantage of providing a clear interpretation of the mechanisms which determine the corrosion characteristics of a given film. One such mechanism is transpassive dissolution. Thus, according to Macdonald, i f reaction (3) is so fast as to overwhelm the capacity for cation vacancy annihilation through reaction (1) then cation vacancies w i l l condense at the m/bl interface. This in turn could lead to macroscopic de-cohesion of the bl from the metal and subsequent replacement of the b l by a salt film (or outer layer) whose protective properties are generally thought to be inferior [Macdonald 1999]. Bojinov et al. have shown, through extensive impedance studies of N i - C r alloys, that transpassive dissolution is in fact due to the oxidative ejection of hexavalent Cr from the bl to the ol through a two step process [Bojinov 2000, 2002, 2003], [Betova 2004]. To model the process they have employed a variant on the P D M called the \"mixed conductor model\" or M C M . Furthermore, Macdonald has shown that \"the transpassive state is induced by the oxidative emission of cations from, and by the oxidative dissolution of, the barrier layer, with the latter resulting in a decrease in the barrier layer thickness as the voltage is increased\" [Macdonald 2004B]. The issue of applied voltage is important because, from a thermodynamic standpoint, the equilibrium potential for equation 2-1 is ~ 0.23 V with respect to the standard hydrogen electrode (SHE - all further potentials w i l l be referred to this scale) at 28 298 K , p H 10. When compared to the oxidation of Cr to C r 2 0 3 which occurs at ~ -1.18 V and the oxidation of N i to N i O at ~ -0.46 V , under the same conditions, the transpassive potential for N i - C r alloys is high (cf. E, - transpassive potential and Ecorr -corrosion potential in Figure 2-3). Thus in the laboratory, it is necessary (undermost circumstances where oxidants have not been added) to apply high anodic overpotentials to study transpassive corrosion. From a practical engineering perspective, it is more interesting to know under what natural and un-simulated conditions one could expect to see transpassive corrosion because in a S C W O reactor the applied voltage is set by the chemical composition of the solution. Thus, i f the equilibrium potential of the oxygen reduction reaction is high enough, transpassive dissolution could be expected to occur. Such a case could arise when the solubility of oxygen is high such as in a S C W O reactor. But the concentration of oxygen, C0i, (and thus temperature) at which the transpassive corrosion begins is largely unknown as it depends on the kinetics of both the anodic and cathodic reactions. This temperature which results in the transition from the passive state to the transpassive state is known as the \"inversion temperature\" [Kritzer 2000A]. Attempts have been made to determine the inversion temperature by morphological analysis of corroded tubing exposed to S C W O conditions. Kritzer et al. have extensively studied the issue and have come to the conclusion that the inversion temperature is ~ 523 K for alloy 625 when exposed to acidic solutions with ~ 0.48 m C0i \u00E2\u0080\u00A2 Schroer et al. have proposed a similar result (523 to 573 K ) for various N i - C r alloys in 0.12 m HC1, 0.06 m C0i. The importance of this point is related to the interpretation of the corrosion process. Macdonald has argued that dissolution of the bl controls the process even up to high sub-critical temperatures [Macdonald 2001]. Kritzer 29 does not differentiate between a bl and ol but simply states that the corrosion rate depends strongly on the solubility of the oxides [Kritzer 1998]. These are the questions which arise: \u00E2\u0080\u00A2 If at all, when does the bl fail i.e. when does the solubility of the ol control the corrosion process? \u00E2\u0080\u00A2 What is the mechanism leading to failure? Is it dissolution of the bl first, then dissolution of the ol, or both at the same time? \u00E2\u0080\u00A2 Can morphological observations alone account for a complex electrochemical process such as transpassive corrosion? \u00E2\u0080\u00A2 If the dissolution of the ol controls the corrosion process, what are the morphological and electrochemical implications? Simply put, these questions highlight the fact that the issue of transpassive corrosion and its relationship to oxygen concentration is largely unknown. With regards to the implementation of SCWO, it is clear that the increased solubility of oxygen at high temperatures combined with increased reaction kinetics is expected to lead to high corrosion rates. All the more so that oxidant feed concentrations are by process necessity high in SCWO reactors. Corrosion in SCWO reactors is generally found to reach a maximum in the high density sub- to supercritical region (623 < T < 660 K) where it is believed that near perfect oxygen solubility leads to high rates of transpassive dissolution [Kritzer 2000A, B]. While this observation is indicative of some measure of oxygen control over the corrosion process, it is unclear to what extent charge transfer of the cathodic reaction (oxygen reduction) plays a role in determining corrosion rates, particularly as temperatures are varied. Further to this, it is also unclear at which 30 temperature a diffusion controlled process might be expected to determine the rate of corrosion. There have been some studies pertaining to the kinetics of oxygen reduction on stainless steels [Babic 1993], [LeBozec 2001] and on alloy C-22 [Davydov 2005] at room temperature. In general, the surface condition of the alloys is found to play a large role on the cathodic process and, in particular, the passive film was found to suppress both the diffusion limited current density and the charge transfer kinetics of the oxygen reduction reaction. Most corrosion studies tend to focus on the anodic processes and their kinetics. From a corrosion rate standpoint it may be more practical to study the rate of cathodic reaction as there is available data on the solubility of oxygen and its diffusivity. Moreover, given that the limiting diffusion current can be modeled by the flux of oxygen to the corroding surface, it may be possible to extrapolate room temperature corrosion rates up to high temperatures and pressures with an adequate degree of certainty. Indeed, such an analysis has been performed by Tromans for the corrosion of steels and the leaching of sulphide concentrates [Tromans 1999, 2000]. However, besides the work done by Speller [Speller 1951] there is little published data of an experimental nature with which to validate the oxygen diffusion controlled model and, to this author's knowledge, no data on the mixed control scenario which may occur at low to moderate temperatures. In particular, there does not appear to have been much work done of a quantitative nature in regards to the effect of oxygen on the corrosion behaviour of N i -C r - M o alloys which were developed for- and are frequently employed in oxidative high temperature media. Within the context of this discussion, the P D M is important as it provides a framework and a powerful tool through which to investigate and interpret these issues. 31 Finally, it should be reiterated that while transpassive dissolution leads to rapid corrosion, in acidic solutions, it does not affect N i - C r alloys exposed to neutral solutions at S C W O conditions. This is because under such conditions N i O is able to passivate the alloy. A schematic E - p H diagram showing the regions of stability for N i O and Cr203 as well as their chemical or electrochemical dissolution is presented in Figure 2-6. pH Figure 2-6 Stability fields for N i O and Cr203 in S C W O . Transpassive dissolution (upwards arrow) only results in severe loss when p H is acidic. Modified from Kritzer 2000A. 2.5 Alternatives to Ni-based alloys or stainless steels Due to their tendency towards transpassive dissolution, alternatives to N i - C r alloys are required in the high-density region of S C W O reactors. Two of the more 32 reasonably priced valve metals which are often used in the chemical industry for aggressive acidic environments are Nb and T i . It is generally accepted that these materials do not corrode transpassively and in fact remain passive over a wide range of oxidizing conditions. In fact, the breakdown potential of Nb is reportedly in the vicinity of 100V! Some conflicting work on their respective corrosion resistance has been published. For corrosion testing purposes researchers at General Atomics built a Pt lined S C W O reactor in which they placed coupons of such materials as Pt/Ir and Pt/Rh alloys, Hf, T i , Zr, M o , Nb, Ta, A 1 2 0 3 , A1N, Sapphire, Z r 0 2 and for good measure, alloys C22, C276 and 625. For various reasons, none of these materials performed well in all three of the tested solutions which were highly aggressive: 3.1wt% H 2 S 0 4 , 2.3% HC1 and 7.0% F f 2 0 2 is one example [Downey 1995]. T i was found to perform adequately, and better than Nb, and has since been adopted as a sacrificial reactor liner for use in the S C W O testing programs at General Atomics. However, the use of T i is generally not recommended in the presence of sulphates [Kritzer 1999A]. Moreover, Kritzer et al. have found that Nb was un-attacked (gained weight) up to 623 K , 240 bar in 0.05 m H 2 S 0 4 and 0.05 m HC1 [Krtizer 1999C]. The next temperature increment for Kritzer's work was 773 K so that the corrosion resistance in the 623 to 673 K range is unknown. Also , from a general corrosion rate perspective it is well documented that Nb outperforms T i in both HC1 and H 2 S 0 4 solutions up to at least 373 K [Brubaker 1987], [Degnan 1987], [Yau 1987]. Finally, there is the issue of Nb as an alloying element. A l l o y 625 is alloyed up to ~ 4% (Table 2-1) with Nb to tie up carbon and prevent grain boundary carbide precipitation as wel l as to promote strength through solid solution and precipitation hardening mechanisms. This is unfortunately not enough to form a protective N b 2 0 5 film. It is conceivable that 33 adding Nb to N i - C r alloys could result in better resistance to transpassive attack. The study of Nb corrosion at high potentials and in oxidizing environments is not complete so it was deemed important to contribute in this area with a view to employing this metal as an alloying element in N i - C r alloys or unalloyed as a section of a S C W O reactor. . . 2.6 The issue of ammonia and Ni-based alloys In the department of materials engineering at U B C there is sometimes a look of puzzlement on the faces of colleagues as I explain to them that I have undertaken the study of corrosion of Ni-based alloys in ammoniacal sulphate solutions in high temperature oxidative media. Perhaps this is due to the fact that the Sherritt process for the oxidative recovery of N i from sulphide ores at ~ 423 K was invented by Frank Forward, the first head of our department and our building's namesake [Forward 1951, 1953, 1955]. This process is based on the thermodynamic stability of some transition metal ammine complexes such as N i , Cu, Z n and Co. For the case of N i it is expected that Ni(NH3)l+, where v is 1 to 6, would form in the presence of ammoniacal solution. Furthermore the leaching of oxide ores using ammonia was patented by Lance in 1904 [Lance 1904]. Hydrometallurgy is an expertise in the department such that most are not surprised that N i corrodes readily under such exposure. In fact, the study of corrosion in ammoniacal solutions has historically been undertaken at U B C . For instance, Ffalpern published three papers on the dissolution kinetics of C u and C u - A u alloys in the presence of ammoniacal sulphate solution at 0 2 pressures up to 7 bar [Ffalpern 1953, 1959], [Fisher 1956]. O f relevance to this work is the fact that oxygen availability was rate determining at low oxygen pressures and that this limitation was ascribed to diffusion 34 limited oxygen transport through the reaction product film [Fisher 1956]. Birley and Tromans have also studied the brass-NH 3 system with an emphasis on stress corrosion cracking [Birley 1971]. Generally these corrosion studies have shown that copper alloys exhibit unsatisfactory behaviour when in the presence of ammonia. Copper ammine complexes readily form and stress corrosion cracking is possible. The behaviour of nickel in ammoniacal solutions is less predictable. Corrosion studies of copper-nickel alloys in ammonia containing environments have shown that increasing the alloying content of nickel increases corrosion resistance [Caruso 1981], [Rajagopalan 1981]. Similarly, Rice found that atmospheric corrosion of nickel was inhibited by the presence of ammonia [Rice 1980]. However, Uhl ig 's Corrosion Handbook makes a brief reference to the fact that nickel may be unsuitable for use in ammonium salt solutions [Revie 2000]. Electrodeposition studies have shown that nickel deposition increases by some 40 to 60% in high ammonia concentration baths at pH=10 (as compared to ammonium baths at pH=5). The reason for this was attributed to the predominance of a thermodynamically stable nickel ammine - Hexamminenickel (N^NH^ )26+) and the complexing nature of ammonia. Indeed, Rodriguez-Torres developed speciation diagrams at 298 K for the N i ( I I ) - N H 3 - H 2 0 system and found a large region of stability for N^NHj)2*, even at low partial pressures of ammonia [Rodriguez-Torres 1999]. At 298 K , p H 10 it appears as though nickel should readily corrode in ammoniacal solutions. Whether this tendency is translatable to a N i - C r - M o alloy such as alloy 625 at temperatures in excess of 647 K was one of the veins of pursuit for the study that follows. 35 3 OBJECTIVES In view of the preceding discussion the following objectives were set-out. A. To determine the corrosion resistance of Ni-Cr alloys, exemplified by alloy 625, to ammoniacal sulphate solution in the approximate temperature range of 298 to 673 K. B. To determine the effect of pH on the corrosion characteristics of alloy 625 when exposed to ammoniacal sulphate solution. C. To measure the effects of oxygen on the corrosion behaviour of alloy 625 in the same temperature range. D . To identify the expected electrochemical and morphological response to naturally induced transpassive corrosion (i.e. not by applied potential) E. To investigate the issue of diffusion control at high sub-critical temperatures with a view to possibly predicting corrosion rates. F. To calculate the phase diagram for the Ni-NH3-H20 system throughout the temperature range of interest. G. To explore the possibility of using Nb for the high density supercritical sections of the reactor. H. To determine the corrosion rate of other related alloys under SCWO conditions. 36 4 POLARIZATION STUDY OF ALLOY 625, Ni, Cr AND Mo IN AMMONIACAL SULPHATE SOLUTIONS The effect of ammoniacal salt solutions on the corrosion of N i - C r alloys is of significant fundamental interest but has of yet not been studied. Given that these alloys are known to be protected by N i O in moderately alkaline solutions, any tendency toward the selective removal of N i or instability of N i O due to Ni-ammine formation could result in unexpected degradation of the material. Since this is a baseline study into the corrosion characteristics of alloy 625 and its main constituents (Ni , Cr, M o ) in solutions of ammonium sulphate, all tests were performed at room temperature. High temperature studies w i l l follow. Polarization results are related to Pourbaix diagrams and the effect of p H , oxidant and alloying elements on the overall alloy behavior is investigated. 4.1 Experimental Electrochemical tests were performed using coupon samples lathe cut from 3/4\" diameter alloy 625 (UNS N06625) bar stock. High purity (-99%) molybdenum, nickel and chromium were purchased as commercially available samples. Coupons were 16mm in diameter and 3 mm thick so as to fit in a pre-fabricated, commercially available, nylon sample holder complete with crevice free washers. A standard A S T M corrosion cell with two high-density graphite counter electrodes and saturated calomel (SCE, E = 0.241 V vs. SHE) reference electrode was used. The reference electrode was kept in contact with the solution through a salt bridge terminating in a Luggin-Haber probe. Solutions were prepared in distilled water using reagent grade H2SO4 and N H 4 O H . To distinguish the 37 effect of ammonia from that of sulphate, some tests were also performed in solutions of Na 2 S04 or H2SO4 at adjusted pH. The results presented below which examine the effect of oxidant and alloying elements are from tests performed at p H 10 (0.2 m H2SO4 /1 .6 m NH4OH ) because, as demonstrated by the failure described in the introduction, alloy 625 has been shown to be susceptible to ammoniacal solutions at that p H . The test samples were wet polished to 600 S iC , degreased in ethanol and ultrasonically cleaned in acetone: They were then mounted in the sample holder, immersed in solution and cathodically pre-treated at -2.46 V for 600 seconds to remove any air formed oxide and create a consistent starting point. Potentiodynamic scans were not performed subsequent to reaching steady-state. If allowed to sit at open circuit the alloy 625/ammonium sulphate system would take roughly 8 hours to reach steady-state (defined here as a change of less than 1 m V over a 10 minute period). Instead, most scans were performed immediately after the cathodic pre-treatment, starting in the cathodic range at -1.26 V and ending at 0.942 V . The potential scan rate was 0.8 mV/s. The result of one polarization test performed at steady-state conditions is presented below for comparison. To assess the influence of oxygen on alloy 625 some tests were conducted under de-aerated or naturally aerated (open to the atmosphere) conditions while one graph presents results from an oxygen saturated solution. In this last case oxygen was sparged into a rapidly stirred solution for the duration of the experiment. In the case where de-aeration was required, the solution was argon sparged and stirred one hour prior to testing. A t the onset of scans following argon sparging, stirring was terminated while purging continued. Naturally aerated experiments were not stirred. Experiments were controlled through the use of a potentiostat run by a computer. 38 The corrosion potential, Ecorr, was taken as the point at which the cathodic current was equal to the anodic current on the polarization curves. The corrosion current density at Ecorr is icorr. Points referred to as E] and E2 are the potentials corresponding to the two anodic peaks ( i i and i2) observed on the polarization curves of alloy 625. Ex is the lower potential point. The passive current density for all scans, * , is taken as the current density corresponding to the potential which is 250 mV above E2. This potential was chosen as it lies within the passive range between E2 and Et (transpassive potential) and also because it roughly coincides with the corrosion potential achieved after 8 hours of immersion under open circuit. In other words it reflects the corrosion potential at steady state. The reader is referred to Figure 4-3 for an illustration of these points. In order to relate polarization results presented in this work to theoretical, thermodynamically stable species, Pourbaix diagrams for the N1-NH3-H2O and Cr-H 20 systems were calculated. Aside from the nickel ammines which were considered here, all species considered were those originally suggested by Pourbaix. To simplify the Ni-NH3-H2O diagram, complexation of Ni was considered only for the diammine, tetrammine and hexammine complexation groups. Free energy of formation values used to calculate the Pourbaix diagrams are largely from the Atlas [Pourbaix 1966] with the . exception of the ammines which are derived from equilibrium constants presented in Osseo-Asare's work on the Ni-NH3 -H 2 0 system [Osseo-Asare 1981]. 39 4.2 Results and discussion 4.2.1 Pourbaix diagrams The concentration of total ammonia was set at 1.6 molal (m) and dissolved metals were set at 10\"6 m. Figure 4-1 and Figure 4-2 are the diagrams for the N i - N H 3 - H 2 0 and Cr-FbO systems, respectively. Both diagrams are presented at 25\u00C2\u00B0C. The addition of ammonia to the N i - H ^ O system removes any predicted field of stability for N i O or Ni(OFf)2. Nickel is therefore not expected to passivate in solutions of ammonia. Cr, however, w i l l passivate between p H 4 and 12.6 under reducing and mildly oxidizing conditions. The transpassive oxidation of Cr(III) oxides to Cr(VI) soluble species is expected to occur above 0.25 V at p H 10. P H Figure 4-1 Pourbaix diagram for the N i - N H 3 - H 2 0 system at 25\u00C2\u00B0C. 40 The application of Pourbaix diagrams to the interpretation of electrochemical results is notoriously difficult. For example, under acidic conditions devoid of oxidant, such as in de-aerated I M H2SO4, G i l l i et al. reported passivation whereas the Pourbaix diagram for nickel predicts general corrosion [Gi l l i 1969]. Pourbaix's Atlas also presents some contradictions between the predicted N1-H2O system and experimental results. There are two approaches to explain conflicting passive behavior and thermodynamic predictions: salt film precipitation or metastable oxide formation. At high concentrations of sulphate, passive-like behavior has been linked to salt film precipitation. Melendres and Tani have presented laser Raman spectroscopy results identifying a corrosion film in the passive region on nickel in 14 N H2SO4 as N i S 0 4 * S 0 4 [Melendres 1986]. Similarly, under conditions of concentrated H2SO4, G i l l i et al. were able to identify B-NiS04*6H20 by X-ray diffraction powder patterns [Gi l l i 1969]. However, the experiments performed here were in solutions of 0.2 m S 0 4 2 \" and concentrations of N i were assumed to be 10\"6 m. The solubility of NiSC-4 is reported to be roughly 2.6 m at 25\u00C2\u00B0C [Linke 1965]. It is then unlikely that M S O 4 precipitation could have occurred during the'experiments conducted in this work. Metastable oxide formation is therefore a likely explanation for any passivation presented in this work which occurs outside the predicted fields of stability for N i O , N i ( O H ) 2 , or C r 2 0 3 [Macdonald 1999]. Recent surface analysis performed on N i - C r [Jabs 1997] and N i - C r - M o [Lloyd 2003] alloys, in solutions of 0.5 M H 2 S 0 4 and 0.1 M H 2 S 0 4 + 1 M N a C l respectively, have shown that the passive film is almost entirely comprised of Cr(III) ions. Given this information and the thermodynamic predictions herein, it was expected that alloy 625 would show a similar polarization curve to that of elemental Cr as its oxide has a wider region of stability. 41 HCr0 4 ' Cr 3 + Cr0 4 2 ' ** *\u00C2\u00B0 \u00E2\u0084\u00A2 *\u00E2\u0080\u00A2 Cr 2 0 3 Cr 2 + Cr0 2\" - Cr 2 4 6 8 10 12 1 pH Figure 4-2 Pourbaix diagram for the C r - H 2 0 system at 25\u00C2\u00B0C 4.2.2 Effect of oxidant Polarization curves of alloy 625 in 0.2 m H 2 S 0 4 /1 .6 m N H 4 O H , p H 10, under de-aerated (argon sparged), aerated (naturally aerated) and oxygenated ( 0 2 sparged) conditions are shown in Figure 4-3. In de-aerated solutions, two small oxidation peaks appear just below the passive region, the first is at E, = -0.539 V and the second is at E2 = -0.213 V . Given the Pourbaix diagrams displayed in Figure 4-1 and Figure 4-2 and the good agreement between these peaks and those demonstrated by N i and Cr (Figure 4-7) under the same 42 polarization conditions, it is hypothesized that Ex is associated with the onset of chromium oxidation to chromium oxide (C^Cb) and that E2 corresponds to the solubilization of nickel as the hexavalent nickel ammine N i ( N H 3 ) 6 2 + . It is likely that at E2 both reactions are occurring. Under aerated conditions, alloy 625 displays cathodic loop behavior between -0.33 and -0.13 V. This type of polarization curve is a result of the cathodic reaction rate exceeding the passive current density at a given potential range [Kelly 2003]. This range, which follows a \"second\" Econ, exhibits a net cathodic current, or what is conventionally displayed as a negative current on a non-log graph. 43 Under oxygenated conditions the increased mass transfer of 0 2 to the surface of the alloy caused a very high rate of oxygen reduction which exceeded the net anodic current and effectively raised the corrosion potential to approximately -100 m V . In fact, there is a much higher rate of oxygen reduction than one might expect for the oxygenated solution. Indeed, the bulk oxygen concentration of naturally aerated solutions is approximately 0.25 m M / L [Kelly 2003] whereas the solubility limit of O2 is somewhere in the region of 0.7 m M / L in solutions of ammonium sulphate at 298K - a difference of only a factor of 2.8 [Tromans 2000]. In contrast, Figure 3 shows that the limiting current density of the oxygen reduction reaction (ORR) for the oxygenated solution is roughly 2 orders of magnitude larger than the current registered at the cathodic loop in the naturally aerated solution. This high rate of oxygen reduction can be explained by the vigorous stirring which took place throughout the experiment. This was done intentionally to maximize the corrosion potential. Despite a very high rate of oxygen reduction alloy'625 maintains a relatively low corrosion current in the presence of ammonia/ammonium sulphate and the corrosion potential remains within the passive region. However, the addition of one part 30% H2O2 to five parts of ammonium sulphate solution at p H 10 (a concentration equivalent of 1.95 M H2O2) results in a much more aggressive environment. In this case the potential is raised to such an extent as to preclude passivation; the alloy then exhibits active behavior. Moreover, the corrosion current is an order of magnitude higher than the passive current density seen under less oxidizing conditions. 44 4.2.3 Effect of pH The polarization curves for alloy 625 in selected de-aerated solutions are shown in Figure 4-4. There is a strong agreement between thermodynamic predictions and the results o f the polarization experiments. The relationship between Ecorr and p H is linear and coincides almost exactly with the hydrogen evolution reaction (HER) (Figure 4-5). The slope of the line relating the first oxidation peak (E,) to p H is also similar to that of the H E R , only with a positive shift in ordinate of 70 m V . The data points obtained for E2 are more scattered than Ecorr or E,, particularly at p H 3.35 and 5.54, and for this reason the relationship between E 2 and p H is not obviously linear. 1500 -1500 I\u00E2\u0080\u0094'\u00E2\u0080\u0094\u00E2\u0080\u0094*\u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2! 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 Current density, A/cm2 Figure 4-4 Potentiodynamic polarization curves in selected de-aerated environments. Effect o f p H . 45 Following the progression of these potentials as p H is varied across the Pourbaix diagrams for nickel and chromium, it is possible to identify the major reactions occurring at each oxidation peak as p H is modified. Thus, at higher p H (6.5 and above) nickel w i l l dissolve at E2 through a complexation process to form an ammineNi(NH 3) 2 n +. The formation of ammines is important as it appears to remove one of the distinct advantages to the use of nickel in high p H solutions, particularly i f the solution is mildly oxidizing. A s p H is lowered and the free ammonia concentration drops, nickel dissolves as a divalent ion without forming an ammine. It is likely that at all p H , alloy 625 's surface is depleted in N i . This would result in a local enrichment in Cr and thus the formation of Cr203 can be identified as the main mode of passivation for alloy 625. However, and with reference to Figure 4-2, Cr2C>3 is not thermodynamically stable above p H 12.6 and below p H 4. Furthermore, i2 is larger with respect to i at p H levels lying outside or at the limit o f the thermodynamic stability of Cr203 (for example 0.05 and 11.91). This indicates a higher rate of N i dissolution and in turn a less protective surface film. Passivation at p H 0.05, as seen in Figure 4-4, may then be explained by the formation of a metastable oxide as outlined earlier. According to Shimakage and Morioka, nickel dissolves in ammonia through the following reaction, M + Y2 02 + nNH3 + H20-> Ni(NH3)2+ + 20H\" 4-1 and in order for the reaction to continue, the removal of OH\" ions is necessary, [Shimakage 1971] i.e. OH' + NH* -> NH,OH -> NH3 +H20 4-2 46 Then, the dissolution of nickel in ammonia/ammonium salt solutions proceeds as shown below: Ni + y202 +(n-2)NH3 + 2M/4+ -> Ni(NH3)2n+ +H20 4-3 In view of equations 4-2 and 4-3, the ratio of ammonium ions to dissolved ammonia is an important consideration when looking at the corrosion of nickel (or Ni-base alloys) in ammoniacal solutions. This ratio is linked to p H due to the following equilibria: NHl <^(NH3\q+H+ 4-4 (NH3\q +H20<^ NHl + OH~ 4-5 We can calculate the relative concentrations of ammonium and ammonia using the equilibrium constant for reaction 4-4 NHl {NH3) +H+ 4-4: kNHi =[(NH3)ai/][H + ]/[NHl] = 5.62SxlO-]0 4-6 where, for the purposes of this work, the square brackets [] correspond to molalities, m, in mol.kgEkO\" 1 . 47 400 Figure 4-5 Experimental points and linear fit lines for p H versus corrosion potential and oxidation peaks 1 and 2 We can calculate the ratio, R = njN, of ammonium to total ammonia in the system, according to the following modification of equation 4-6: n = [N-n][H+]/kNHj = N[H + ]/([H + ] +kNHJ 4-7 where n = [NH+4 ]and N = [NH+4 ] + [(NH3)aq] [Tromans 2000]. At 298 K the buffer point of ammonia (the point at which ammonium becomes more prevalent than ammonia as p H decreases) is p H 9.27. At equilibrium, the value of R for the p H 10 solution which was predominantly used in this work is 15.7% i.e. the available ammonia is largely in the form of dissolved ammonia. A n effort was made to determine a correlation between R and the passive current density for various solutions between p H 11.91 and 0.05. N o relationship was readily apparent. Instead, the 48 overriding variable appeared to be p H . This is another indication that Cr2C>3 determines the polarization characteristics of the alloy as one would expect more fidelity to R i f N i was dominating the corrosion process. Passive current densities are plotted with respect to p H in Figure 4-6. There is clearly a trend of increased current densities with decreasing pH. Current densities obtained from polarization tests of alloy 625 in solutions of similar p H and at the same sulphate concentration (H2SO4 or Na2SC>4 adjusted with NaOH) as in the ammonia experiments are also shown. Passive current densities obtained in solutions of sulphate alone are higher than in the ammoniacal solutions. This is probably related to the buffering action of the ammonium solution which would, according to reaction 3-7, resist local acidification at the bl-s interface during passivation. This is in fact a well understood phenomena which is linked to proton generation during film formation. If buffers are not present the enhanced proton concentration results in higher corrosion rates. The subject has been elegantly discussed by Tromans and Sun [Tromans 1992]. 49 1.4e-5 1.2e-5 h 04 E 1.0e-5|-u (7) W Q. 8.0e-6 6.0e-6 4.0e-6 _i i i i L_ Ammonium sulfate Analogous solutions of sodium sulfate 6 PH 10 12 Figure 4-6 Passive current densities for alloy 625 in solutions of ammonium sulphate and solutions of sodium sulphate 4.2.4 Effect of al loying elements Figure 4-7 presents the polarization curves for N i , Cr, M o and alloy 625 in de-aerated solution at p H 10. A l l o y 625 adopts a very similar general polarization profile to that of chromium especially with respect to the passive region. Chromium spontaneously passivates and has the lowest corrosion potential of all three metals seen in Figure 4-7. Chromium also has a very similar passive current density to that observed on alloy 625. Nickel displays a prominent anodic nose (~ 5.5 m A at ~ -0.180 V ) while its passive current density is twice that of the alloy and Cr. This is further weight to the argument 50 that the alloy's behavior is dictated by the formation of chromium oxide and that N i is selectively removed from this oxide. M o does not passivate and simply dissolves above Ecorr = -0.590 V as molybdate (Mo0 4 2 \" ) . Under aerated conditions N i , Cr and 625 show a decrease of roughly 30% in their passive current densities as compared to those recorded under de-aerated conditions (Fig. 8). It is believed that this decrease is due to a higher rate of oxygen reduction which acts to suppress the total measured current. The increased rate of the O R R is also observed in both the positive shift in E c o r r for nickel, chromium and molybdenum as well as the clear cathodic transition between the H E R and O R R on these same metals. The curve for the transpassive region of 625 is remarkably similar in shape to the anodic dissolution of M o , particularly in its limiting current behavior above 0.6 V . It is believed that although delayed to higher potential, the transpassive dissolution of 625 is in fact closely related to the dissolution of M o . This finding is not unexpected as Sugimoto and Sawada have shown that M o containing steels resist pitting attack through adsorption of M0O4 \" (molybdate ion) in pits at onset of pitting [Sugimoto 1976]. B y analogy, alloy 625 may resist pitting attack through selective adsorption of M0O4 2 \". The theory surrounding the role played by M o in the protection of corrosion resistant alloys is twofold. Firstly, it has been reported that Mo(VI) , located at the outer layers of the oxide film, is stabilized as M0O4 2 \" by the high electric field at that location. Such stabilization would lead to the de-protonation of OH\" within the oxide film, in turn supplying O 2 \" for the formation of C r 2 0 3 [Brooks 1986], [Clayton 1986]. Secondly, M o 0 4 2 \" is also believed to confer a cation selective quality to the outer regions of the oxide. This would result in higher resistance to chloride induced pitting [Sakashita 1977]. 51 1000 1e-9 1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2 1e-1 Current density, A / c m 2 Figure 4-7 Polarization curves for alloy 625, N i , Cr and M o in de-aerated solution at pH 10 52 1000 1e-9 1e-8 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2 1e-1 1e+0 Current density, A/cm2 Figure 4 - 8 Polarization curves for alloy 625, N i , Cr and M o in naturally aerated solution a t p H 10 4.2.5 Steady State behavior and corrosion rate It is customary to present data for polarization curves under steady-state conditions. However, the nature of alloy 625 in ammonium sulphate solution is such that steady state conditions require an impractical amount of time to be achieved. Furthermore, the free corroding potential is very difficult to duplicate given its dependency on the reproducibility of a passive film, particularly when in the presence of oxygen. Moreover, the real nature of the polarization curve is masked under these conditions. However, for the purposes of establishing a general idea of the steady state 53 behavior and corrosion rate of alloy 625 at p H 10, the alloy was allowed to rest in naturally aerated solution for 8 hours prior to running a polarization scan from -0.250 vs. open circuit to a maximum of 1.2 V . The result of this experiment is seen in Figure 4-9. The alloy exhibits Tafel behavior and active dissolution of the passive film is evident. ECOrr n e s within the passive range of the aerated scan and it is clear that the steady state corrosion current density, icorr = 1 . lxl0\" 7 A / c m 2 , is less than the passive current density Opass = 6x l0\" 6 A /cm 2 ) recorded under direct scan conditions. Using the passive current density as an approximation to the corrosion rate, [Kelly 2003] the highest rate of attack occurs at the lowest p H values (Figure 4-4). B y the same assumption N i would corrode roughly twice as fast as the alloy itself. Thus, it is reasonable to expect selective dissolution of N i from alloy 625 at p H 10. A t p H 10 in ammonium sulphate solution, the steady state corrosion rate for alloy 625 is calculated from icorr to be ~ 1.2x10\" mmpy. This corrosion rate is acceptable and one can conclude that alloy 625 does not corrode to any appreciable extent in this solution. Even with a high rate of oxidant delivery, as in the case of the oxygenated solution, the alloy remains passive. The corrosion rate in the presence of hydrogen peroxide is however much higher (0.15 mmpy) and highlights this alloy's susceptibility to transpassive corrosion. 54 1 0 0 0 5 0 0 LU X (/> > 0 E LU - 5 0 0 - 1 0 0 0 1 . E - 0 9 Direct scan Steady state 1 . E - 0 7 1 . E - 0 5 Current density, (A/cmA2) / 1 . E - 0 3 Figure 4-9 Polarization curves for alloy 625 under direct scan and steady state scan conditions - naturally aerated. 4.3 Summary A l l o y 625, N i , Cr and M o were tested in various solutions o f ammonium sulphate at different pH. The effect of oxidant on the polarization behaviour was also established. A l l o y polarization curves were similar to curves obtained for Cr indicating that the alloy's behavior was dominated by formation of a passive C r ( I I I ) film on its surface. N i displayed the highest rate of passive dissolution and a large anodic nose. M o readily dissolved. 55 Polarization curves were compared to Pourbaix diagrams. Passivation observed at acidic p H for N i and Cr, and at p H 10 for N i , could not be predicted thermodynamically. In fact N i is not thermodynamically stable at any p H in the presence of ammonia at 298 K . It is believed that metastable oxides are responsible for the passive behavior observed. Passive current densities increased with p H , pointing to the fact that proton concentration was the main determinant to the corrosion rate rather than ammine formation. Moreover, sulphate solutions alone were found to result in higher passive current densities than ammonium sulphate solutions at the same p H . The buffering effect of the ammonia system may explain this occurrence. In general, corrosion rates for alloy 625 were found to be acceptable, even at high rates of 0 2 mass transfer. However, under heavily oxidizing conditions, such as in the presence of H 2 O 2 , alloy 625 is susceptible to transpassive dissolution in the presence of ammonium sulphate. This is particularly relevant in the context of S C W O where solutions are heavily oxidizing. A s discussed throughout this work, while not necessarily an issue in the low temperature sections of a reactor, transpassive corrosion is a significant problem in S C W O reactors at higher temperatures. Moreover, there is ample evidence from these results that N i would preferentially dissolve in solutions of ammonia. It is l ikely that at higher temperatures this selective dissolution would be aggravated. The extent to which both the issue of de-alloying and transpassive dissolution affect the behaviour of alloy 625 at higher temperatures is addressed in the following pages. 56 5 EFFECT OF OXYGEN ON THE CORROSION BEHAVIOUR OF ALLOY 625 AT ELEVATED TEMPERATURES While the presence o f ammonium sulphate probably results in N i de-alloying through ammine complexation (alkaline ph) or dissolution as N i 2 + (acidic pH), it has been shown that corrosion rates at room temperature are low unless high concentrations of oxidant (H2O2) are present, at which point the transpassive potential of the alloy is exceeded. To determine the effect of oxygen solubility and partial pressure on the corrosion behaviour of alloy 625 in ammoniacal sulphate solution, a series of potentiodynamic, linear polarization and impedance measurements were performed between 298 and 473 K at varying oxygen partial pressures. Potentiodynamic polarization data was used to glean information about the kinetics of the oxygen reduction reaction as well as to shed light on the transpassive potential of the alloy as a function of temperature. Impedance data gathered at open circuit was used to determine the corrosion process and nature of the oxide film. Lastly, the corrosion rates obtained by electrochemical testing were compared to an oxygen diffusion model similar to that proposed by Tromans. 57 5.1 Experimental 5.1.1 Safety considerat ions It should first be noted that the work described below was performed in a titanium autoclave at high oxygen partial pressures. It is well known that T i can auto ignite in such environments and that industrial autoclave fires have resulted from such occurrences of auto ignition. A good review is provided by Costa [Costa 2005]. Generally these fires have resulted from either frictional heating, impact or promoted combustion (nearby burning component). There have also been instances of oxygen sparger/feed line ignition due to rapid, sudden and high pressure injection of oxygen. In all instances the Ti-oxide which protected the underlying metal was caused to fail. T i w i l l not spontaneously ignite at temperatures up to 250\u00C2\u00B0C without having a freshly fractured, abraded or scratched surface exposed. In an abundance of caution the following precautions were followed: \u00E2\u0080\u00A2 Injection lines were made from 316 SS \u00E2\u0080\u00A2 No dynamic motion occurred during the experiments (i.e. no stirring) \u00E2\u0080\u00A2 Injection of 0 2 was always performed cold \u00E2\u0080\u00A2 The autoclave area was cordoned off during runs \u00E2\u0080\u00A2 The partial pressure of O2 and the total pressure were kept just below the \"static\" ignition curves as determined by Littman and Church [Littman 1961] Static ignition curves generally describe a situation in which a freshly fractured T i surface is exposed to static oxygenated environments. Dynamic ignition curves relate to motion of O2 across the fractured surface and for this reason dynamic ignition curves delineate a more conservative area on the T and P plane. The reader w i l l note that there 58 was never the possibility of a fractured T i surface being exposed to the oxygenated environment in the work detailed below because no mechanical motion occurred during the experiments. Finally it might be mentioned that N H 3 is a fairly good fuel and as such any similar work undertaken in the future should be done very carefully. 5.1.2 General considerations A 1.8 L glass-lined titanium autoclave was used for the electrochemical experiments. The alloy 625 working electrode (WE) was a cylindrical specimen and the holder was of the Stern-Makrides variation [Kelly 2003] - please refer to Appendix A for drawings and pictures. The counter electrode (CE) consisted of a Pt wire with approxiamtely 3 times the surface area of the W E . A n external pressure balanced saturated A g / A g C l reference electrode was used ( E P B R E ) and was bridged to the solution through a fiberglass wicked P T F E tube which contained the test solution. A l l potentials have been converted from A g / A g C l to the S H E i.e. ESHE(mV) = EAg!AgCI +197'mV and all further potentials are reported with respect to S H E . The thermal liquid junction potential (TLJP) across the electrolyte bridge was ignored in this work. This potential is difficult to accurately estimate and is only available in the literature for HC1, N a C l and KC1 solutions [Lvov 1996] [Macdonald 1979B]. A s thermal diffusion phenomena are determinant in calculating the T L J P , it is generally expected that larger, less mobile ions w i l l result in lower potential drops across the bridge. Thus a very qualitative estimate for the anticipated T L J P in the NH^ , OH' and SO]' containing bridge used here could be between 0 m V at 298 K and 30 m V at 473 K [Lvov 1996]. 59 A 0.2 m sulphate and 1.6 m total ammonia solution at ~ p H 10 was prepared using de-ionized water and reagent grades of ammonium hydroxide and sulphuric acid. Prior to immersion in solution, the W E was wet ground to 600 S iC , rinsed in de-ionized water and ultrasonically cleaned in acetone. The solution volume poured into the autoclave at 298 K was 1.1 L allowing for a head space of 0.7 L . After assembly, the autoclave head space and solution were sparged with oxygen for at least 25 minutes and then, i f required, oxygen was added to the system to increase the total pressure (PTOTAL 29&K = p o 2 + PH2O + PNH, )\u00E2\u0080\u00A2 Varying total gauge pressures at 298 K were used, from atmospheric (0 bar g) up to 27.58 bar g (400 PSIG). It should be noted here that unless otherwise specified, all further pressures in this section w i l l be reported as gauge pressures. When the autoclave was fully sealed and therefore constituted a closed system at constant volume, it was brought incrementally to temperature (298, 323, 348,..., 473 K ) with electrochemical measurements performed at each increment. Three main types of electrochemical experiments were performed separately: potentiodynamic Tafel polarization, linear polarization and electrochemical impedance spectroscopy. The entire set-up procedure was started anew for each set, or type of experiment i.e. the data for the potentiodynamic sets, linear polarization sets and EIS sets of experiments was collected separately with fresh solutions and fresh samples. However, within each set of experiments, spanning the entire temperature range, neither the sample nor the solution was changed as the behaviour of a single, consistent sample/system interface was of interest. For example, a scan having been run at 298 K , the same system was then brought to 323 K and a subsequent scan was performed, and so on, until the last scan performed at 473 K . To attenuate concerns pertaining to the steady-state condition 60 and electrode stability, following the 20 minute heat-up period a 30 minute rest at open circuit was performed for each temperature increment. Furthermore, the alloy's oxide film was never overly perturbed as potentials did not exceed \u00C2\u00B1 250 m V O C . The autoclave temperature was controlled to within 5 degrees during any experiment and pressures were recorded from a pressure transducer to \u00C2\u00B1 3 bar. The error band in the temperatures and pressures are limitations of the equipment. During an electrochemical experiment the temperature controller was turned off to ensure that cold water passed through the autoclave's cooling coil did not induce convection currents in the autoclave. The pressure transducer was separated from the autoclave head by an oi l filled isolator/piston assembly to protect it from exposure to ammonia and oxygen which resulted in very little sensitivity at low pressures. The potentiodynamic polarization experiments (PD) were performed in the anodic direction from -250 m V versus open circuit (OC) to +250 m V O C at 1 mV/s. These experiments were performed at PT0TAL 2 g s K = 0 and at PT0TAL 2 9 i K = 13.79 bar (0 and 200 psig) and the entire sets were repeated at least twice. The results of these tests were used to calculate the cathodic Tafel slopes at each temperature and to estimate the corrosion current density by extrapolation of the linear portions of the cathodic curves to the corrosion potential. Linear polarization (LP) experiments were performed at PmTAL 29SK = 6.89, 13.79 and 20.68 bar (100, 200 and 300 PSIG). Scan rate was.0.5 mV/s in the anodic direction through the range of \u00C2\u00B120 m V O C . The polarization resistance (R p ) was determined by computer assisted linear least-squares fitting around the corrosion potential (Ecorr) and also manually by calculating the slope of the tangent to the E-i curves at Ecorr. 61 For the EIS measurements, the autoclave was loaded with oxygen to PTOTAL^K = 27.58 bar (400 PSIG). The W E was polarized at \u00C2\u00B1 10 m V O C in the frequency range of 100 k H z to 10 mHz. Similarly to the linear polarization experiments, the goal of this work was to estimate the polarization resistance but with the added advantage of shedding light on the metal/oxide/solution interfaces. These experiments were repeated twice. 5.1.3 Oxygen diffusivity and concentration calculations In order to fully explore the impact of oxygen on the corrosion characteristics of alloy 625, both its diffusivity and concentration are required. Tromans has developed a theoretical model for oxygen solubility which allows for the calculation of its concentration ([Tromans 1998, 2000B]). Generally, the relationship between P 0 I (partial pressure of oxygen in atmospheres) and C0 (dissolved oxygen in m o l . k g ^ O \" 1 ) is: C = k ^ p r0l 5-1 Where k is the inverse of the Henry's Law constant and, for pressures below 60 bar (-880 PSI), is given by: ( exp 0.046r2 +203.357Tn f j \ \29Sj \u00E2\u0080\u00A2 (299.378 + 0.0927X7/ - 298) - (20.591 x 10 3) 8.31447 ...5-2 62 The variable (/> represents the fraction of water which is available to dissolve O 2 ; (1 - (j>) being relegated to the interactions with dissolved ionic components. For the solution used here^ =0.96401 [Tromans 2000]. The variables y0 and 298 K ) . The calculated equilibrium potentials, corrosion potentials and extrapolated exchange current density are also shown in Table 5-2. Due to the differences in the effective diffusion layer thickness depending on the test conditions, the calculated values of i0 at a given temperature varied sometimes up to 4 orders of magnitude. The oxygen sparge time was also found to play a role here. In Figure 5-5, i0 is plotted against the pressure and sparge time conditions. The longer sparge times correlated well with larger \u00E2\u0080\u00A2 76 Tafel slope values: implying smaller limiting currents; this is an indication that the increased time of heightened mass transfer to the electrode resulted in larger effective diffusion layer thicknesses (probably because Ecorr was raised nearer to the transpassive range for longer periods of time). Other authors have also found that the oxygen reduction reaction has a very high dependence on alloy pre-treatment [LeBozec 2 0 0 1 ] . 1e-6 t 1 1 1 ' I 1 CM \"E o < 1e-7 r 1e-8 r 1e-9 r 1e-10r 1e-11r 1e-12r 1e-13r 1e-14r 1e-15r 1e-16r 1e-17-i * 1 1 * i * ' * ' i -1\u00E2\u0080\u0094)\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094r-\u00E2\u0080\u00A2 p = r TOTAL, 298K 13.79 bar, 35 min V P = TOTAL, 298K 13.79 bar, 25 min \u00E2\u0080\u00A2 P = r TOTAL. 298K 0 bar, 50 min O P = r TOTAL, 298K 0 bar, 35 min o \u00C2\u00A9 (J) o _L_i i i_i I i i i i I i_ _ J I I I I I I t I I I I I I I l _ _ l I I I I I I I I I I I f 280 300 320 340 360 380 400 420 440 460 480 500 T , K Figure 5-5 Exchange current density as a function of temperature, pressure and sparge time. Circled values are considered most accurate due to smaller Tafel slopes. 77 5.2.2 Linear polarization experiments The linear polarization experiments were performed to measure the effect of oxygen pressure on the corrosion rate without overly polarizing the alloy. The results of these experiments conducted at initial 298 K total pressures of 6.89, 13.79 and 20.68 bar are shown in Figure 5-6, Figure 5-7 and Figure 5-8, respectively. The overpotential, 77, is plotted in abscissa and is given by the formula: t] = E - Ecorr where E is the applied potential. It is clear that the anodic sections of the curves, in particular, have progressively increasing slopes as temperature is increased which indicates a progressively smaller polarization resistance (R p ). Linearity is restricted to AE ~10 m V , however some degree of curvature is expected away from O C [Kelly 2003]. 78 Overpotential, mV Figure 5-6 Linear polarization as a function of temperature at initial oxygen loading PTOTAL,29&K ~~ 6-89- bar 79 1.0 -10 0.5 E o < 0.0 -0.5 -1.0 -5 0 5 - | 1 1 1 1 1 1 1 1 1 f\u00E2\u0080\u0094 -S3* t0C^ A A A A A A A -10 0\u00C2\u00B0 o 10 \u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2MI:;; 298 K w V 323 K \u00E2\u0080\u00A2 348 K 0 373 K A . 398 K o 423 K 0 448 K , r 473 K V 1.0 0.5 0.0 -0.5 -1.0 10 Overpotential, mV Figure 5-7 Linear polarization as a function of temperature at initial oxygen loading PTOTAL,29SK 13.79 bar 80 298 K w v \u00E2\u0080\u0094 323 K \u00E2\u0080\u00A2 348 K O 373 K A 398 K 0 423 K \u00E2\u0080\u00A2 448 K , , 473 K V Overpotential, mV Figure 5-8 Linear polarization as a function of temperature at initial oxygen loading PTOTAL,29$K ~ 20.68 bar 81 There were three main issues of concern in this work with regards to the applicability of the linear polarization resistance method for corrosion rate measurement. The first was whether or not AEI B \u00C2\u00AB 1 to ensure that the Stern-Geary equation would hold: i = \u00E2\u0080\u0094 5-13 corr D B - - L 2.3 f PaPc 5-14 where f3n and fic are the anodic and cathodic Tafel slopes, respectively, and B is a proportionality constant known as the Stern-Geary coefficient. With Ba and fic values in the region of 90 and 180 mV/decade respectively, it was deemed safe to apply equation 5-14. Also , it was necessary to assume that B would not change significantly over the temperature range tested. Due to the increased oxygen partial pressure used for the L P experiments it was found that B did not, in fact, vary substantially (\u00C2\u00B1 3 mV) from the normal Stern-Geary coefficient of 26 m V i.e. both the anodic and cathodic curves were sufficiently removed from pure diffusion control to allow for charge transfer based estimations of R . Finally, due to experimental necessity, the hold-time at temperature (30 min) and voltage scan rate (0.5 mV/s) were respectively shorter and faster than is recommended for this type of experiment. This can lead to large capacitive effects and thus underestimates of R [Kelly 2003], particularly when mass transfer such as oxygen diffusion is playing a role in the corrosion process. To verify that the scan rate was adequate a few small amplitude cyclic voltametry experiments were performed at 298, 398 and 448 K to check for hysteresis. N o significant capacitive response was detected. 82 The corrosion rates obtained at various PT0TAL 29SK and by various methods of measurement are shown in Figure 5-9. There is little discernible effect of pressure on the corrosion rate between ~ 298 to 398 K . There does appear to be some separation in the data as temperatures increased above 398 K suggesting that the corrosion rates were increased by higher pressures. For example, the corrosion rate at 473 K , PT0TAL^%K = 20.68 bar is roughly 2.5 times the corrosion rate at PT0TAL 2 n K = 0 bar. This is l ikely the result of Ecorr approaching Et at the higher oxygen partial pressures and is also a sign of an increasingly diffusion controlled corrosion process. 1e-4 rr\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094|\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094|\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094|\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094|\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094|\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094|\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094r\u00E2\u0080\u0094|\u00E2\u0080\u0094r\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094|\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094\\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094|\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094r -1e-5 E u < ; \u00E2\u0080\u00A2 PD- P = TOTAL.298K 0 bar 0 LP- P = TOTAL,298K 6.89 bar T PD- P = TOTAL.298K 13.79 bar V LP- ^TOTAL,298K = 13.79 bar : \u00E2\u0080\u00A2 LP- P = TOTAL,298K 20.68 bar \u00E2\u0080\u00A2 EIS - P = TOTAL.298K = 27.58 bar 1e-6 5 0 9 \u00E2\u0080\u00A2 I T \u00E2\u0080\u00A2 8 V 1e-7 _ i i I i i i _ \u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094I\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i\u00E2\u0080\u0094i I i i _ J \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 i _ J i i i i I i i i i I i i i i I i i i i _ 280 300 320 340 360 380 400 420 440 460 480 500 T . K Figure 5-9 Corrosion rates as measured by various methods (PD - Potentiodynamic, L P - Linear Polarization, EIS - Electrochemical Impedance Spectroscopy) and at various TOTAL,298/C \u00E2\u0080\u00A2 83 5.2.3 Electrochemical impedance measurements Nyquist and Bode plots generated at temperatures up to 448 K at PTOTAL^%K = 27.58 bar are shown in Figure 5-10 and Figure 5-11, respectively. From these figures it is clear that the total impedance and thus Rp decreases with temperature as was seen in the L P experiments. The Nyquist plots show a partially resolved semicircle at high frequencies and a Warburg-type response at low frequencies. Warburg impedance is normally associated with a semi-infinite diffusion controlled process (such as through the solution) and is characterized by linear 45\u00C2\u00B0 sections on a Nyquist plot. Normally semi-infinite diffusion is measured at very low frequencies (<1 mHz) [Macak 2006]. The frequency range tested in this work is not low enough to capture solution phase diffusion. The general shape of the curve at 298 K is very similar to the one found by Chao et al. on passivated N i in a phosphate buffer solution [Chao 1982] and Macdonald and Smedley for the same system [Macdonald 1990A, B ] . In the case of the N i electrode and under the assumption that most of the current was carried by oxygen vacancy transport, Chao et al. attributed the observed Warburg-like diffusion to oxygen vacancy transport across the bl in accordance with the P D M [Chao 1982], [Macdonald 1990C, 1992A, B , 1999, 2004, 2005]. The P D M describes film growth in terms of oxygen anion or metal cation vacancy transport within the thin (1 to 3 nm) semiconductor-like bl . Cation vacancies diffuse from the bl/ol interface to the m/bl interface while oxygen anion vacancies diffuse in the opposite direction. A consequence of this model is that the \"primary passive film (i.e. that which is not precipitated from solution) can only grow by the diffusion of oxygen anion vacancies\" (cf. Figure 2-5) [Chao 1982]. The movement of cation vacancies does not result in film growth but does result in cation ejection from the 84 bl [Macdonald 1990C]. For most practical purposes this results in overall impedance (ZT) o f the form [Macdonald 2005]: ZT =o-Mco-v\\-j) 5-15 This is formally identical to Warburg impedance, where the Warburg coefficient is (JM and co is the angular frequency in rad.s\"1. It is assumed in this work that the current is due to cation vacancy transport rather than anion (oxygen) vacancy transport. Discussion of this point w i l l follow. To analyze the effect of temperature on the Warburg coefficient for alloy 625 a Randies plot is shown in Figure 5-12 where dZ' and Z ' represents the real impedance. The decreasing slope of the Randies plots in Figure 5-12 with respect to temperature and thus the decreasing Warburg coefficient, implies that icorr should increase with temperature because according to the P D M , aM is inversely proportional to the measured current. Moreover it is predicted in the P D M that the product, icorr \u00E2\u0080\u00A2 oM , should be constant. This was confirmed here. 85 E o d LU O < Q LU Q. < < 30 20 10 REAL IMPEDANCE, k\u00C2\u00A3J.cmJ fTD\" 5 * O 298 K o 323 K T 348 K v 373 K \u00E2\u0080\u00A2 398 K \u00E2\u0080\u00A2 423 K \u00E2\u0080\u00A2 448 K 0 5 10 15 20 25 30 REAL IMPEDANCE, k Q . c m 2 Figure 5-10 Nyquist plots taken at peak-to-peak voltage excitation amplitude of 10 m V vs. Ecorr (100 kHz , left most point - 10 mHz, right most point), and at various temperatures. P 7 TOTAL,29SK 27.58 bar. 86 (A) 30000 25000 CM E o CJ 20000 uf O z g 15000 UJ Q. < LLI 01 10000 5000 \u00E2\u0080\u00A2 298 K, a = 7880 Q.cm2.s\"1'2 V 323 K, a = 5140 Q.cm2.s\"1 / 2 \u00E2\u0080\u00A2 348 K, a = 3660 Q.cm2.s\"1'2 O 373 K, a = 3010Q.cm2.s\"1'2 \u00E2\u0080\u00A2 398 K, a = 2160 Q.cm2.s\"1 / 2 0 423 K, a = 490 Q . c m V \" 2 \u00E2\u0080\u00A2 448 K, a = 350 Q.cm2.s~1'2 T \u00E2\u0080\u0094 ' \u00E2\u0080\u0094 r ~ J I I I I I I I I I 1 I t I I I I I 1 I 1 _ 4.5 -1/2 01/2 j-1/2 oo , s .raa Figure 5-12 Randies plot showing the real impedance as a function of a>~U2 and temperature - the Warburg coefficient for cation vacancy transport was calculated by linear regression and is shown in the legend 88 To estimate icorr equation 5-14 was used where R was obtained by fitting a circuit model to the EIS results. O f course, the assumption that the Stern-Geary coefficient did not vary between the temperatures within the data set was again made. The model used was a simplified interfacial circuit model suggested by Priyantha et al. for alloy 22, which is presented in Figure 5-13 and where Rp = Rt + Rb + R2 [Priyantha 2004]. To account for capacitance dispersion across a rough barrier layer/solution interface, the model includes a constant phase element: Q2 where n can be treated as a coefficient of roughness for the bl/s interface. Representative model fit parameters are shown in Table 5-4. Agreement between the model and the experimental data was judged to be good and the reader is referred to Appendix C for examples. The corrosion rates obtained in this way are presented in Figure 5-9 and are consistent with those obtained by other methods. Metal Ii II M i l 1 l \ U Oll l l ' l 1 . I M 1 Solution m/bl interface bl/s interface - Ri R B R 2 \" c, c b Q2 R S Figure 5-13 Circuit scheme used for fit to experimental data. Interfaces 1 and 2 represent the m/bl and bl/s interfaces respectively. 89 Table 5-4 Model fit parameters for the simplified circuit model assumed (Figure 5-13) to represent the passive film and interfacial behaviour. T C, x l O5 Rb x l O \" 2 Q x l O 5 R2 xlO'2 n 0 2 x l O5 Rpx\0~2 K Q.cm2 F .cm'2 Q.cm2 F.cm~2 Q.cm 2 F.cm'2 Q.cm2 Q.cm2 298 1.853 0.007 712.84 27.49 141.65 0.86 9.52 5.57 854.50 323 0.006 0.004 424.72 24.82 107.28 0.85 10.20 4.46 532.00 348 298.62 53.634 357.56 23.21 163.73 0.84 12.06 3.85 524.27 373 184.24 23.944 157.05 82.70 89.11 0.84 13.56 3.26 248.00 398 145.62 96.704 59.95 23.49 51.81 0.84 16.24 2.93 113.22 423 1896.41 23.803 40.87 62.90 5.92 0.83 21.32 2.74 65.76 448 78.17 172.535 8.32 33.56 17.76 0.84 21.62 2.60 26.86 It was also of interest to estimate both the barrier layer thickness and electric field strength as a function of temperature. The former can be estimated by applying the formula for the parallel plate capacitor [Macak 2006] at high frequency: L . - ? & \u00C2\u00B1 5-17 SS s~i where Lss is the steady-state barrier layer thickness, sr is the relative permittivity of the oxide and s0 = 8 . 8 5 x l 0 \" 1 4 F .cm 1 is the vacuum permittivity and A is the electrode area which was 0.71 cm in this work. Macdonald et al. have estimated thatf-,. ~ 30 for alloy 22. This value was retained here as the composition of the oxide formed on alloy 22 is known to be mostly Cr203 and therefore similar to that formed on alloy 625. The geometric capacitance, Cm, is a parameter usually estimated from the intercept of the linearly extrapolated low frequency capacitive response and the real axis of the complex capacitance [Macak 2006]. To eliminate the error involved in line fitting and extrapolation the measured capacitance at 6 k H z was used here. This practice has been employed by other authors [Macdonald 2004B, 2005], [Raja 2006]. The calculated data 90 for Lss and Rb (the barrier layer resistance) is shown in Figure 5-14. The calculations suggest that that the barrier layer is being destroyed as temperature is increased: it thins linearly by roughly one half between 298 (1.4 nm) and 473 K (0.7 nm) and Rb decreases exponentially over the same temperature range. The values of Lss calculated by Macdonald et al. for alloy C-22 are similar: ~ 0.8 to 2.0 nm depending on applied voltage at 353 K . L loyd et al. have measured oxide thickness on alloys 22 and C-276 by time of flight secondary ion mass spectroscopy after up to 60 hours of potentiostatic treatment at temperatures ranging from 298 to 358 K and found values ranging from 2.0 to 2.8 and 1.5 to 3.2 nm, respectively with N i O concentrations in the region of 10 to 40 at% [Lloyd 2003]. Also , From X P S measurements after exposure to sulphuric acid at 338 K , Wallinder et al. found that the passive film on stainless steel 316 L was ~ 2nm [Wallinder 1999]. Both Macdonald and L loyd found that the oxides generally thickened up to ~ 450 m V . Above 450 m V , Macdonald et al. found that the oxide thinned linearly with increasing potential and it was concluded that this corresponded to transpassive dissolution [Macdonald 2004B]. While voltage was not varied explicitly in this work it was indirectly affected by the increasing temperature and oxygen partial pressures. Considering the fact that Ecorr was close to the transpassive potential for alloy 625 at PT0TAL 29SK = 13.79 bar and 473 K it seems reasonable to assume that within the tested temperature range at PT0TAL 2 9 i K = 27.58 bar one might encounter transpassive dissolution. 91 1.8 0.8 0.6 0.4 i 1 1 1 1 i 1 1 1 1 i 1 1 1 1 i 1 1 1 1 i 1 1 1 1 i - 1 \u00E2\u0080\u0094 I \u00E2\u0080\u0094 I \u00E2\u0080\u0094 I \u00E2\u0080\u0094 I \u00E2\u0080\u0094 I \u00E2\u0080\u0094 I \u00E2\u0080\u0094 I \u00E2\u0080\u0094 I \u00E2\u0080\u0094 I \u00E2\u0080\u0094 I \u00E2\u0080\u0094 i \u00E2\u0080\u0094 i \u00E2\u0080\u0094 r H ' Rb - J I I I I I I I I 1 t I I I I I I I ' ' I I I I I I I I L _ _ l I I I I I 1 I I I I L_ 1e+6 1e+5 E o 1e+4 \u00C2\u00A3 1e+3 1e+2 280 300 320 340 360 380 400 420 440 460 480 Figure 5-14 Calculated values of the steady-state barrier layer thickness {Lss) and film (bl + ol) resistance (Rb) as a function of temperature. Lss was calculated using the parallel plate formula, equation 4-17. Table 5-5 Parameters derived from EIS measurements, according to equations 5-13 and 5-19(5=0.026, \u00C2\u00A3 = 1.10xl06 V.cm~\ \u00C2\u00AB = 0.878). Z is the standard deviation for Rp. T 10-2 x^ 10\"2xE 106 X ! corr 106x^ K Q .cm2 Q.cm2 sU2.rad-U2 A.cm~2 2 -1 cm .s 298 854.50 185.94 7880 0.30 1.90 323 532.00 637.96 5140 0.49 1.95 348 524.27 66.36 3660 0.50 2.73 373 248.00 72.45 3010 1.05 4.17 398 113.22 46.06 2160 2.30 4.07 423 65.76 1.47 490 3.95 5.49 448 26.86 10.60 350 9.68 6.76 92 The advantage gained by the variance in initial oxygen sparge time is that it allowed for a spectrum of E c o r r data at a given temperature. This advantage was used to generate a Mott-Schottky plot for alloy 625 as a function of temperature and the result is shown in Figure 5-15. It appears as though the passive film formed under the conditions tested here is a p-type semiconductor. The slope for the Mott-Schottky relationship ( l / C 2 = f(v)) for a p-type semiconductor is given by, dC'2 - 2 ^=\u00E2\u0080\u0094 = \u00E2\u0080\u0094 5-18 3V s0ereNA where the electronic charge e is 1.6 x 10~19 C. Assuming a good estimate of sr ~ 30, we can calculate the film's acceptor (cation vacancy) concentration, NA as a function of temperature. This yields values of ~ 2.26xl02 0cm~3 at 298 K, 2.96x 1020cm~3 at 348 K and 1.69 x 1021 cm\"3 at 448 K. Macdonald et al. have calculated values of ~3 x 1020cm~3 for alloy 22 at 353 K in the transpassive range (>0.6 V SHE). At potentials below the transpassive region the barrier layer on alloy 22 is apparently n-type. The increased acceptor concentration and transition to a p-type film indicates, according to Macdonald et al., a high rate of cation vacancy generation at the bl/s interface due to ejection of high valence Cr(VI) cations through transpassive dissolution [Macdonald 2004B]. These vacancies diffuse to the m/bl interface and if they are not annihilated eventually condense leading to bl separation and eventual film breakdown. Such an analysis, applied here, would indicate that even at 298 K alloy 625 was corroding transpassively. However, while in the case of alloy 22 the transition from n-type to p-type is clearly an indication of transpassive corrosion, the case presented here is less clear. 93 10 CM LL. \u00E2\u0080\u00A2* E O TOTAL.298 K 1 \u00E2\u0080\u0094 ' \u00E2\u0080\u0094 ' \u00E2\u0080\u0094 ' \u00E2\u0080\u0094 i \u00E2\u0080\u0094 I \u00E2\u0080\u0094 r -28 bar \u00E2\u0080\u00A2 298 K o 323 K T 348 K V 373 K \u00E2\u0080\u00A2 398 K \u00E2\u0080\u00A2 423 K \u00E2\u0080\u00A2 448 K Increasing T _ i i i i i _ j i i i i i _ -100 -50 50 100 E, mV (SHE) 150 200 250 Figure 5-15 Mott-Schottky plot for alloy 625 as a function of temperature at various ^corr values which were varied according to oxygen sparge time at 298 K It is possible that the barrier layer on alloy 625 is a cation conductor throughout its passive range. Films on N i in borate buffer solutions, for example, are p-type in both the passive and transpassive range [Sikora 2002], [Macdonald 1992A]. Furthermore, it is well known that C r 2 0 3 is a p-type semiconductor [Hakiki 1995, 1998], [Kofstad 1980], [Sunseri 1990] and Hakiki et al. have found that pure chromium oxide is expressly p-type over the range ~ 0 to 0.5 V which is close to the observed range here (-0.050 to 0.25 V ) . The fact that alloy 625 is a highly doped cation conductor in ammoniacal solution may 94 also be a reflection of the fact that high rates of N i expulsion from the b l are occurring within the passive range and as a result of the instability of N i O which could be associated with ammine formation or solution acidification. Interestingly, Table 5-5 shows that the corrosion current densities for these experiments were low and not commensurate with what one would normally expect for transpassive corrosion i.e. icorr values larger than ~ 10 5 A.cm\" 2 at 298 K , for example (cf. Figure 4-4). Though, L loyd et al. have found that despite expulsion of Cr(VI) species from the passive film on alloy 22 at high potentials (up to 0.9 V ) at 353 K , this film was able to concentrate Cr within its oxide and only corroded at ~ l x l O \" 6 A.cm\" 2 [Lloyd 2003]; so the corrosion current is not necessarily a good diagnostic criterion for transpassive corrosion either. Perhaps the best way to make the determination of passive versus transpassive corrosion is the Ecorr data. During the course of the EIS experiments, Ecorr was generally higher than for all other experimental conditions - a result of the elevated equilibrium potential for the oxygen reduction reaction due to the higher pressures used (cf. Appendix B) . However, Figure 5-16 shows that above ~ 40 bar (373 K ) , Ecorr actually slightly decreased (125 to 57 m V ) as pressure and temperature were increased. A simplified Evans diagram is shown in Figure 5-17 and it graphically represents the scenario believed to have taken place. Figure 5-1 and Figure 5-2 show that the anodic Tafel slopes tend to zero as Et is approached and the effect of higher pressure and temperature on the oxygen reduction reaction has already been discussed. In Figure 5-17 it is assumed that iL continues to increase with temperature. This is perhaps not the case due to a thickening diffusion layer, as shown in Table 5-3 (iL 47JK ~ 0.9iL 44SK for the 95 PT0TAL 29&K =13.79 bar experiments), however, a decrease in iL would only result in the observed decrease inEcorr and concomitant increase in corrosion rate i f Et had been reached. Thus either explanation leads to the conclusion that the corrosion reaction is approaching limiting cathodic currents at ~ 373 K , 40 bar and that this temperature/pressure combination leads to transpassive corrosion. Therefore, it is very likely that at 40 bar and, in an oxygen saturated ammoniacal environment, the inversion temperature for alloy 625 is ~ 373 K . 400 300 h 200 W 100 co > E o h .8 -100 LU -200 h -300 -400 T = 373 K o All data Average for EIS measurements 20 40 60 TOTAL T ' b a l \" Figure 5-16 Dependence of Ecorr on total pressure for all EIS data- decreasing Ec values above ~ 40 bar substantiate the assumption of diffusion control at high temperature. 96 logi Figure 5-17 Predicted effect of pressure and temperature on Ecorr at PT0TAL T > 40 bar -T3 >T2> 7| and P3 > P2 > P} - it is expected that transpassive corrosion occurs readily at this pressure One final corroborating factor that the inversion temperature is 373 K is the general shape of the Nyquist plots shown in Figure 5-10. For T < 373 K the curves are nearly linear and the imaginary and real components are almost equal. A t T > 373 K the imaginary component is suppressed and the curve takes the shape of a semi-circle. This behaviour has also been observed by Sikora and Macdonald for N i in the transpassive region [Sikora 2002]. The preceding discussions regarding the inversion temperature 97 have been emphasized because for significantly harsher conditions (P = 240 bar, of ~ 523 K could be justified through morphological analysis of S C W O reactor tubing which showed generalized pitting [Kritzer 1998]. Given the findings presented here, it seems reasonable to suggest that morphological examination alone may not adequately estimate the inversion temperature; particularly as the corrosion rates found here were still quite low despite onset transpassive attack, meaning that the morphology may not be well developed. Figure 5-18 shows the surface of alloy 625 after the EIS experiments. There is a definite appearance of intergranular attack but it should be emphasized that this coupon was subjected to temperatures of up to 473 K therefore it is not clear at which temperature the morphology was developed. With the film thickness and voltage data (Figure 5-14 and Figure 5-15) the electric field strengths were calculated as a function of temperature for a cation conducting film: increase from a value o f - 2 x l 0 6 V.cm\" 1 at 298 K to ~ 7 x l 0 6 V.cm\" 1 at 448 K (Table 5-5). Both the value for a = 0.5 which was used for the calculations and the values obtained for the field strengths themselves are typical of those found by others [Macdonald 1990C]. Moreover, Sato and Noda have found that the field strength increases with decreasing p H . This is consistent with the results presented here which show the p H decrease as temperature is increased. [H2SO4]\" = 0.1 m, C0 = 0.48 m), Kritzer et al. have argued that an inversion temperature fc \u00E2\u0080\u0094 / _ \u00C2\u00BB / \ 5-19 where a is the polarizability of the bl/ol interface. The field strength was found to 98 Figure 5-18 SEM micrograph showing corrosion morphology after EIS testing through the temperature range 298 to 473 K, PTOTAL29SK ~ 27.89 bar. IGA is apparent. 5.2.4 Corrosion rate and mixed control It has been have shown that the pressure and temperature conditions examined in this work lead to a complex scenario of mixed control. Generally, there are a few conclusions which are consistent with all of the data: the corrosion rate increases with temperature, the effective diffusion layer thickness increases with temperature, the bl film is p-type in nature and the ejection of cations through the bl leads to the \"lengthening\" of this diffusion layer due to ol precipitation and, finally, as temperature and pressure are increased Ecorr tends to Et which results in the rate control scenario 99 being more diffusional in nature. It is also clear that the oxygen partial pressure and thus solubility are linked (at least in part) to the degree of mixed control. Qualitatively this observation is confirmed by the isothermal Tafel slopes observed at different pressures. Macak et al. have observed oxygen diffusion controlled corrosion for stainless steel in high temperature water but the concentrations of O2 used in their work was on the order of 5 - 30 ppb [Macak 2006]. In this work O2 concentrations are on the order of 40 ppm (0 bar, 298 K ) to 2000 ppm (62.5 bar, 448 K ) meaning that the limiting current densities are theoretically much higher (assuming no change in 8 cf. equation 5-13). Moreover, Macak et al.'s findings were also supported by semi-infinite Warburg diffusion at low frequency (< 1 mHz). The Warburg behaviour observed here is more consistent with vacancy diffusion through the bl . It is therefore of interest to explore the equations governing the corrosion rates as they relate to oxygen concentration, diffusivity and diffusion layer thickness. Under steady state conditions all steps in a consecutive corrosion reaction are proceeding at the same rate regardless of whether they are charge transfer or transport related. Therefore, according to Kirchhoff s first law we have: horr ~ Kx ~~ hrans 5-20 which can be restated as the flux equality condition [Bockris 1970], h ^ = h , _ = T = I \ ( C B CbUol \ 5 _ 2 1 nF nF 8 V \u00C2\u00B02 \u00C2\u00B02 ' where r is the diffusion flux, CQ is the oxygen concentration in the bulk solution, C\u00C2\u00A3 ' / o ' i s the oxygen concentration at the bl/ol interface, D0i is the oxygen diffusivity and 8 is the diffusion layer thickness. It is convenient now, to express the corrosion rate as a 100 function of temperature in terms of the normalized corrosion rate ((icorr)r/{icorr)29S )\u00E2\u0080\u00A2 Such a procedure yields: 5-22 horr J298 l A ? 2 j 2 9 8 lQ> 2 j 2 9 8 IQ>2 j 2 9 8 V\"V7 The first term on the right-hand side of equation 5-22 can be estimated and is known to increase exponentially with temperature (cf Table Al) . The third term has been observed to generally decrease monotonically with temperature (Table 5-3). The second term in equation 5-22 has the most potential to affect the normalized corrosion rate because we have shown that at 298 K the corrosion reaction is mostly under charge transfer control which implies that ~ 0. In Figure 5-19 the normalized corrosion rate is plotted as a function of temperature for each type of experiment. The purely diffusion controlled scenario is also plotted in Figure 5-19 and was obtained through calculations performed according to a method outlined elsewhere and in detail by Tromans. Central to this method is of course the assumption that the corrosion rate is under oxygen control. The principal equations for determining the maximum limiting diffusion flux of oxygen at temperature T ((TL )T) are [Tromans 2000A]: ^corr )r _ ( T L \ 'T 5-23 corr 7298 V / . 298 5-24 101 Where the solution density, (p)r , is in kg.cm\" 3 and can be calculated with the volumetric data generated by equations 5-1 through 5-8. (s)T for this model was assumed to be constant. 221 bar. Under transpassive conditions the solubility of the precipitated ol is then the determining factor of corrosion rate. Studies have found that the precipitated corrosion products (ol) or scales at S C W O conditions are mostly Cr203 or C r O O H [Schroer 2002], [Watanabe 2002]. Kritzer et al. found that scales formed in the transpassive region (T > 523 K ) had high atomic oxygen to chromium ratios of ~ 2.2, indicating that higher valence Cr-oxides could be present [Kritzer 1998]. Indeed, as alloy 625 corrodes in the transpassive range it is expected to release Cr(VI), Cr(III), Ni(II) and possibly Cr(IV) ions (other alloying elements such as Nb and M o are ignored here but are also factors). 103 Because the corrosion process occurs within the transpassive range, the ol can only be characterized as thermodynamically unstable but kinetically protected. This brings up the question: what prompts a sudden dissolution or removal of the ol and consequently high rates of corrosion ca. 650 K ? It could be postulated that the precipitated Cr(III) oxide is ultimately removed by solution acidification, i.e. chemically rather than electrochemically. While the data with regards to solution p H is not fully born-out [Kriksunov 1995], this scenario seems probable and w i l l be discussed later. Should the ol become entirely soluble, alloy 625 would be expected to corrode at its limiting current density. Because the corrosion rates under limiting currents are so high, this analysis suggests that Pourbaix diagrams at S C W O conditions are even more relevant than those at room temperature. It may be useful for future reference and corrosion rate estimation to obtain some data for the effective diffusion layer thickness in a S C W O reactor at temperatures where (a) the ol is present and (b) the ol is soluble. The concentration of 0 2 at temperatures near the critical point must first be obtained. This is done in the following section. 5.2.5 Oxygen solubility at S C W O conditions S C W O reactors typically operate at P > 250 bar and the temperature region of particular interest to corrosion studies is that which is near the critical point (623 < T < 660 K ) . Ideal gas assumptions are no longer valid under these conditions as the compressibility of the gas phase deviates substantially from unity. Non-ideal volumetric and phase data can be modeled using cubic equations of state (equation 5-25), such as the Redlich-Kwong-Soave (RKS) equation [Soave 1984]: 104 P = \u00E2\u0080\u0094 \u00E2\u0080\u00942 r 5 \" 2 5 V-b V2+ubV + wb2 T . , \u00E2\u0084\u00A2 \u00E2\u0084\u00A2 , ~ , 0.08664icT c 0.42748i?2rr2 r R L \u00E2\u0080\u009E 1 / 2 \ i 2 F o r t h e R K S u = l , w = 0,6 = - , a = ^ [ l + /ty(l-7/,. 1 / 2)j Pc Pc and fcv = 0.48 + 1.574 300 bar) and this data is re-plotted in Figure 5-21 along with two estimates near the critical point which they have calculated using second virial coefficients to obtain the fugacity coefficient [Japas 1985]. Japas and Franck have also presented data similar to the Himmelblau data for T < 573 K . The R K S model is reasonably accurate near the critical point and reproduces the experimental data to within 2% - a value which matches the uncertainty associated with the data itself. There is a remarkably good agreement between the Tromans model and the experimental data up to about 625 K - Tromans himself has stated that his model is valid up to 616 K but not above 61 bar [Tromans 1998]. Clearly this model is appropriate to at least 300 bar, ~ 616 K . Oxygen concentrations as well as other data calculated by the R K S are listed in Appendix D . A n 107 interesting observation from this data (p.216) is that the concentration of oxygen actually decreases between 643 and 673 K and then dramatically increases at 673 K . This decrease is. associated with the falling fugacity of oxygen above 473 K . The dramatic increase of oxygen concentration above 673 K is due to the supercritical transition where oxygen becomes entirely soluble. The local maximum in the oxygen concentration generally agrees well with peak corrosion rates being measured between 623 and 660 K , which is a further indication that the corrosion rate is oxygen diffusion controlled in this temperature range. This analysis also provides an explanation as to why experimental data overwhelmingly points to increased corrosion rates just below the critical point of water. 3r-S. 2 O a 1 -2 -3 Experimental D .'Estimated RKS model Tromans model Q Japas and Franck _j 1 t 1 i t_ 1.5 2 2.5 T\"1 x 103, K \" 1 3.5 Figure 5-21 Oxygen solubility - Henry's Law constant as a function of temperature and method of calculation 108 5.2.6 Effective diffusion layer thickness During experiments into the feasibility of treating 2,4 dinitrophenolate with ammonium sulphate, the University of British Columbia's S C W O reactor corroded to failure at ~ 655 K , 250 bar in about 1 hour [Vera-Perez 2002]. Under the assumption that the corrosion process was diffusion controlled and with the corrosion rates measured from S C W O reactor tubing penetration depths, the [C0i) data as calculated above was used to estimate (<5)7 through equation 5-12. S E M micrographs of the corroded sections of alloy 625 tubing subjected to ammoniacal solution (same concentrations as those used in this work) at PT0TAL = 250 bar are shown in Figure 5-22 and Figure 5-23. Figure 5-22 presents the case where the ol was present and the temperature was ~ 633 K while Figure 5-23 represents an area immediately adjacent to the locus of tubing failure where it is apparent that no ol is present, -653 K . After the failure, tube wall thicknesses were measured using ultrasound. A t 653 K the normally 1.7 mm thick tube wall (3/8\" tube with 0.065\" wall thickness) was reduced to approximately 0.3 mm in thickness. The ultrasound was not sensitive enough to measure a difference in wall thickness at 633 K -thus it was estimated using Figure 5-22 to be approximately 15 um. Using equation 5-29 the penetration depths were converted to current densities and are equivalent to corrosion \u00E2\u0080\u00942 2 2 rates of 1.63 x 10 A.cm\" at 633 K and 1.52 A.cm\" at 653 K i f nis set to an average (based on alloy composition - Table 2-1) and approximate value of 3 considering N i dissolution as N i 2 + and Cr dissolution as a Cr(VI) species. The effective diffusion layer thickness at 633 K was then 14.32 cm at 633 K (!!!) and 1.5 mm at 653 K . , npM 3.06x10~ 4 rate(A..cm ) = rate{mm I yr) x 5-29 ^allay 109 A t 633 K , an effective diffusion length of 14.32 cm points to the formation of a low void fraction and high tortuosity precipitate. This can be explained by introducing the effective inter-diffusivity, D0i eff, which accounts for the void fraction in the material, co (here the porous outer layer) and the tortuosity, r [Poirier 1994]: D02,eff ~ Do2 5-30 With D0i = 9.13 x 10\"4 cmls\" 1 (Appendix D) at 633 K , co = 0.001 - approximately one one hundredth the porosity given for piled glass spheres - and r = 8, a value often used for compacted particles, D0 ff\u00E2\u0080\u0094 1.14xl0~ 7 cm2 .s~x. If equation 5-12 is once again used to calculate the diffusion layer thickness with the new value for the effective inter-diffusivity, one obtains (s)T ~ 15 um which is the thickness of the precipitated ol as measured from the micrograph. In other words, the available area for oxygen diffusion can be accounted for by either a large effective diffusion length or a decreased effective inter-diffusivity. The value obtained at 653 K can be compared to an estimated value of the mass transfer coefficient obtained from analogy to the heat transfer coefficient as proposed by Rogak and Teshima [Rogak 1999] and based on the correlation proposed by Swenson et al. for the Nusselt number in supercritical water [Swenson 1965]: 2V r \u00E2\u0080\u00942- = 0.00459 / \ 0.923 f2Gr^ ^0.613 , N 0.231 p D, '2 J \pbj 5-31 where VD = DQi jS is the mass transfer coefficient in m.s\"1, r is the tube radius in m, G is the mass flow rate per unit area in kg.s\"\m\" 2 (270 kg.s _ 1.m\" 2), juw is the viscosity of 110 water at the wall of the tube in Pa.s (5.24 x 10 5Pa.s) and, pw and pb are the density of water at the wall and in the bulk in kg.m\" 3 (340 and 380 kg.m\"3), respectively. For the purposes of the calculations performed here it is assumed that the difference between the bulk and wall temperatures is ~ 10 K in agreement with findings by Rogak and Teshima [Rogak 1999]. Solving equation 5-30 for 8 = D0i jVD yields a value of ~ 0.08 mm, or a factor of 18 smaller than the value (1.5 mm) calculated according to the diffusion limited corrosion rate. This observed difference is likely due to the fact that at 653 K the corrosion reaction is still somewhat impeded by a very thin oxide film precipitate which was either lost during sample preparation or undetectable by S E M . However, factors such as viscosity and density which were used in equation 5-30 can only be considered rough estimates in that they do not account for dissolved metals, precipitates in solution, organic phase solubility etc. Thus the difference between the two values is judged to be acceptable and certainly points to a range of diffusion layer thicknesses which might be used for further corrosion rate estimations: ~ 0.1 to 1.5 mm. I l l x l . O k 0000 20kV 50um Figure 5-22 S E M micrograph of outer chromium oxide layer (precipitated) on the surface of alloy 625 exposed to ammoniacal sulphate solution at 250 bar, -633 K . Figure 5-23 S E M micrograph of inner surface of alloy 625 tubing exposed to ammoniacal sulphate solution at 250 bar, ~ 653 K . 5.3 Summary The results of the electrochemical work into the effect of oxygen concentration on the corrosion of alloy 625 in ammoniacal solution between 298 and 473 K can be summarized as follows: 1. A s temperature and pressure are increased, Ecorr and icorr both also increase while the passive region for alloy 625 gets smaller. 2. Et does not vary much in the tested temperature range because p H is seen to decrease with increasing temperature. 113 3. C-2 sparging time played a significant role in determining Ecorr and Et by increasing the equilibrium potential for the oxygen reduction reaction and by displacing dissolved ammonia and acidifying the solution. 4. Limit ing current measurements confirmed that O2 diffusion lengths were considerably affected by the precipitation of the ol which is in turn believed to be thickened by increased corrosion rates (cation ejection) resulting from Ecorr values near Er. 5. Exchange current densities for the oxygen reduction reaction were found to increase from ~ l x l 0 \" 1 6 at 298 K to ~ l x l O \" 9 A.cm\" 2 at 473 K . 6. A t 398 K and above, corrosion rates increased with increasing pressure due to the proximity of Ecorr to E,, and in one case ( PT0TAL > 40 bar, T ~ 373 K ) it is believed that the inversion temperature was actually exceeded. 7. Mott-Schottky analysis showed that the bl on alloy 625 is p-type i.e. cation conducting. Field strengths were estimated at between 2 x l 0 6 and 7 x l 0 6 V.cm\" 1 and this increase was commensurate with decreasing solution p H as temperature was increased. 8. The corrosion rate was shown to be increasingly diffusion controlled as temperature and pressure were increased such that corrosion rates under S C W O conditions are probably controlled by the precipitated ol and can be safely modeled by limiting currents due to oxygen diffusion. The barrier layer is likely not present on alloy 625 in a S C W O reactor. 9. Oxygen concentrations at 250 bar and at high-subcritical temperatures were calculated in order to estimate diffusion lengths near the sub- to supercritical 114 transition. At 633 K we have estimated a diffusion length of ~ 14 cm due to the presence of a thick and tortuous ol while at 653 K the diffusion length was found to be in the range of ~ 0.1 to 1.5 mm. 115 6 THERMODYNAMICS OF THE Ni-H20-NH3 SYSTEM Under circumstances where the Cr203 passive film (bl or ol) is not expected to be stable N i - C r alloys rely on the formation of a N i O film. In the presence of ammonia, alloy 625 appears to be susceptible to preferential N i dissolution, even at neutral p H . It is also of considerable interest to explore the influence of sulphate on the N i - H 2 0 - N H 3 system as it may affect the solution p H at near critical temperatures. Lastly, as previously discussed, Pourbaix diagrams are increasingly important as the critical point is approached due to the onset of diffusion limited corrosion. This section w i l l address the issue of N i O stability at high temperatures and delve in to the issue of p H . 6.1 Calculations Speciation and Pourbaix diagrams were constructed for the Ni -NHb -EbO system at 373, 473, 573 and 653 K . The method used for calculating high temperature Pourbaix diagrams is described in detail elsewhere [Cowan 1971], [Lee 1981], [Taylor 1978]. Generally, the standard chemical potential of species i at temperature T, // . r , was determined through the use of equation 6-1: 6-1 116 Where / / ; 2 9 8 is the absolute chemical potential of species / at 298 K , Cp is the average partial molal heat capacity over the temperature increment T - 298 K , and S\u00C2\u00B09S is the standard partial molal entropy at 298 K . A literature search was undertaken for each species considered in this work to find the best ways of estimating /uj T either directly, or through eqn. 6-1, in which case the estimation of Cp was required. Table 6-1 lists all of the relevant Cp equations, //, 2 9 g and S\u00C2\u00B09S used in this work while the reactions identified as relevant through the speciation study are listed in Table 6-2. The following sections w i l l describe the methodology used in the calculation of the Pourbaix diagrams. 117 Table 6-1 Thermodynamic data used for Pourbaix diagram calculations Species ju\u00C2\u00B029i, kJ.gmot1 CP(T) \u00C2\u00B0i',298 J.K1 J.(gmoLk)'1 H 2 0 22.496 +17.044 x 1 (T 3 7 + 0.365 x 105 T'2 +11.122 x 10\"6 7 2 130.68 o2 0 22.06 + 20.887 x 10\"3 T +1.621 x 10s 7~2 -8.207 x 10\"6 T2 273-495^ 186.88 - 464.25 x 10\"3 7 -19.57 x 105 T - 2 + 548.63 x 10\"6 T2 205.15 H 2 0 237.14 540-600A: 22605.05 - 54991.07 x 10\"3 T -11271.83 x 105 T\" 2 +37883.51 x 10 - 6 T 2 C \u00E2\u0080\u009E i 6 5 3 = 144.60 69.95 O H 157.40 - -10.71 H + 0 2 9 8 - 4 5 0 ^ 1 6 2 7 + 3 6 7 5 8 x l ( ) - 3 7 , + 0 1 6 7 x l 0 5 r - 2 _ J g J 54 x 1 fj\"6 T 2 0 Ni 0 450-631 K j 042.74 _ 2642.18 x 10\"3 T - 442.41 x 105 T2 +1946.97 x 10\"6 T2 63i-voo^_131 9 6 + 3 6 1 . 39x l0\" 3 7 , + 200.02xl0 57 , _ 2 -267.44xl0\" 6 r 2 29.80 NiO 211.75 d - 37.99 f N i 3 0 4 . -713.36e a140.838 + 17.177xl0\" 3r-24.38xl0 57;- 2 +54.02xl0\" 6r 2 114.27b N i 2 0 3 -470.51 e a143.566-36.323xl0\" 37 ,-31.433xl0 5r\" 2 +71.792xl0\"67;2 87.84b N i 0 2 -215.75* a 69.454 + 10.209 x 10\"3 T -16.234 x\0sT~2 + 0x 10\"6 T2 58.73 b N i 2 + -45.72 f L L H C A : Cp = OMT -128.9 f H N i 0 2 -343.23 d - -150.72 h Ni(NH 3 ) 2 + -85.90g L L H C A : Cp = 0.617/ and Ni model: Cp = 0.237 11.31c Ni(NH 3 ) 2 2 + -127.90f L I H C A : Cp =0.497 and Ni model: Cp =0.127 85.40 f Ni(NH 3 ) 3 2 + -160.708 L I H C A : Cp = 0.307 and Ni model: Cp = -0.047 199.07c Ni(NH 3 ) 4 2 + .-193.618 L I H C A : Cp =0.207 and Ni model: Cp =-0.137 258.60 f Ni(NH 3 ) 5 2 + -223.90 g L I H C A : Cp = 0.077 and Ni model: Cp = -0.247 335.50 g Ni(NH 3 ) 6 2 + -255.70 f L L H C A : Cp = -0.037 and Ni model: Cp = -0.337 394.64f * From the database assembled by Roine unless otherwise specified [Roine 2002]; aCp(T) functions for M 3 O 4 , M 2 O 3 and N i 0 2 were not available so they were assigned the function for C03O4, Fe2C>3 and M n 0 2 , respectively; b S \u00C2\u00B0 2 9 8 values for M 3 O 4 , M 2 O 3 and M O 2 were not available so they were estimated using equation 6-3 and the entropy data for [Roine 2002] C03O4, Fe203 and M n 0 2 , respectively; c ,S\u00C2\u00B0 2 9 8 for Ni(NH3) 2 + and N i ( N H 3 ) 3 2 + was not available so it was estimated using equation 8-2 and the entropy value 3 8 for Cu(NFf3)2+ and C u ( N H 3 ) 3 2 + ; dCalculated from Tremaine and Leblanc [Tremaine 1980]; e Data taken from the Pourbaix Atlas; f Data taken from the N B S Tables [Wagman 1982]; g Calculated from Smith and Martell, eqns. 10-11 [Smith 1977]; h From Shock [Shock 1997]. Table 6-2 Some of the reactions considered and their calculated standard electrode potentials. ET, VsHE No Reaction 298 K 373 K 473 K 573 K 653 K (a) 2H++2e'=H2 0 0 0 0 0 (b) 02 + 4H+ + Ae = 2H20 1.23 1.21 1.15 1.05 0.83 1 M 2 + + 2e~ = Ni -0.23 -0.17 -0.13 -0.16 -0.28 2 Ni(NH3)2+ +H+ + 2e~ = Ni + NH4 -0.03 0.01 0.08 0.13 0.16 3 Ni{NH3)22 + 2H+ + 2e~ = Ni + 2NH4 0.16 0.23 0.33 0.44 0.17 4 Ni(NH3)24+ + 4H+ + 2e~ = Ni + 4NH4 0.64 0.73 0.91 1.13 1.02 5 Ni(NH3)2+ + 6H+ + 2e\" = Ni + 6NH+A 1.14 1.27 1.54 1.88 2.42 6 Ni(NH3 )\+ + 2e~ = Ni + 2NH3 -0.39 -0.33 -0.21 -0.09 -0.31 7 Ni(NH3)26+ + 2e~ = Ni + 6NH3 -0.5 -0.38 -0.09 0.30 0.98 8 NiO + 2H+ + 2e' =Ni + 2H20 0.13 0.15 0.12 0.02 -0.09 9 HNiO; +3H+ +2e' = Ni + 2H20 0.68 0.77 0.86 0.90 0.53 10 Ni02 + H+ + 2e~ = HNiO' 0.66 0.60 0.44 0.22 0.35 11 Ni203 + H20 + 2e~ = 2HNi0'2 -0.11 -0.24 -0.52 -0.87 -0.49 12 Ni30, + 2H20 + 2e' = 3HNi02 + H+ -0.81 -1.02 -1.40 -1.88 -1.22 13 Ni30, +ISNH3 + 8 / T + 2e\" = 3Ni(NH3)26+ + 4H20 2.73 2.42 1.47 -0.08 -2.55 14 Ni30, + 6NH3 + SH+ + 2e\" = 3Ni{NH3)22 + 4H20 2.39 2.26 1.82 1.08 1.33 15 Ni304 + 2H+ + 2e' = 3NiO + H20 0.83 0.85 0.83 0.77 0.63 16 Ni304 + 6NH4 + 2H+ + 2e = 3Ni(NH3)22 + 4H20 0.75 0.60 0.19 -0.50 -0.11 r t V O C N O rt *t ~ V O o vo C N O O r- oo O N l O \u00C2\u00A9 ~ - 9 oo oo o ro \u00E2\u0080\u0094 ; C N C N V O v\u00C2\u00A3> O V O O O C N oo c^ ' ' rt \u00E2\u0080\u0094 1 O rt ro O N P rt rt \u00E2\u0080\u0094 O ro O C N O in o o r* in oo ro O N r- rt O rt O N V O r-o O N o rt rt CN ro o o \u00E2\u0080\u0094 rt C N \u00E2\u0080\u0094 ro r--3-oo V*l rt O - H ^ C N r-oo C N o Q ^ \u00C2\u00B0 S ^ ^ CN. + + rr CN o cN CN + + VO + CN + + ^ CN \u00E2\u0080\u00A2 CN ^ + + CN ^ + + <^ vo ^ + CN + CN + o CN + + . fN + I v\J CN + ^ v o + CN + o + ^ ^ II II < tc tc ^ S 7 II ^ \u00C2\u00A7 ^ + ^ CN + + tC5 CN + CN fN \" T T tN VO II + rr CN | CO + CO + VO + CN + O II II o o fN fN + CN + CN + CN + II o fN + r - - c x o O N O ' \u00E2\u0080\u0094 i f N t o - ^ i n v o r ^ \u00E2\u0080\u00A2 r t r t i r \u00E2\u0080\u0094 i c N C N C N C N C N C N C N C N oo ON O *\u00E2\u0080\u0094< CN co ^J\" CN CN CO CO CO CO CO 121 6.1.1 Condensed Ni phases The experimentally determined solubility constants, K , by Tremaine and LeBlanc [Tremaine 1980] were used to determine the chemical potentials of N i O at all temperatures, juNi0 T , according to equation 6-2: AG\u00C2\u00B0 = -RT ln Ksp = pNj2+j + MHIOJ ~ Mmo,r ~ 2 / V , r 6-2 Standard chemical potentials at 298 K for M 3 O 4 , N12O3 and MO2 were found in the Pourbaix Atlas. Because the heat capacity as a function of temperature was not readily available for these species, Kel ley functions [Kelley 1934] for species with similar molecular weights and chemical formulas (Table 6-1) were relied on to obtain the average heat capacity. Similarly, where 5 \u00C2\u00B0 9 8 data were not available, they were estimated using a method suggested by Latimer [Latimer 1938]: \ 1 \ + co V S Pr ' T r - 1 + coKJY (T - Tr) ...6-7 where AG\u00C2\u00B0P T = /u\u00C2\u00B0'7abs, AG\u00C2\u00B0 = / / , \u00C2\u00B0 2 9 8 and Tr and Pr are the reference temperature and pressure (298 K , 1 bar). The pressure independent parameters a],a2,ai,a4, the temperature independent parameters cx, c2 and the reference Born coefficient cop T are available in the literature for a great many ionic species [Shock 1988, 1997]. The conventional species dependent Born coefficient co must be calculated at the desired temperature and pressure and is based on the effective electrostatic radius of the ionic species under those conditions [Shock 1988]. The solvent-dependent parameters y/ and 6 are equal to 2600 bar and 228 K , respectively, for water. The dielectric constant of water, e , is available in the literature. In this work, the tabulated data by Archer and 124 if dins} Wang was used tors [Archer 1990]. Y is a Born function given by \u00E2\u0080\u0094 and is H dT Jp equal to - 5.8 x 10~5 K\" 1 at Tr and Pr. The Tremaine and Leblanc data for N i 2 + is available up to 573 K and was considered the most accurate so it was used to calculate the absolute (or standard) chemical potential of this species up to 573 K . It should be noted that because all reaction potentials (or electrode potentials: E\u00C2\u00B0) at a given temperature Tare usually reported with respect to the hydrogen evolution reaction (itself set to zero), equation 6-8 was employed to convert partial molal chemical potentials obtained from Tremaine and Leblanc from the conventional scale to the \"absolute\" scale: . ,0 _ ,.0,abs y 0 (\u00E2\u0080\u00A2 o Where //\u00C2\u00B0 + refers to the absolute standard partial molal chemical potential of the hydrogen ion at temperature T, and Z ( represents the charge of the ith aqueous ion [Tanger 1988]. It might also be mentioned that this equation is valid for all conventional standard partial molal properties such as entropy (eqn. 6-6) or enthalpy, for example. For the calculations at 653 K it was necessary to use the FfKF model to estimate u\u00C2\u00B0 \u00E2\u0080\u009E. In Ni ,653 A this instance the pressure term in eqn. 6-7 was set to 250 bar. Sensitivity trials were run to evaluate the effect of pressure between 250 and 300 bar on the calculated ju\u00C2\u00B0 1+ W K and it was found that this pressure variation resulted in an approximately 5 kJ difference. Translated to the Pourbaix diagram at 653 K this is equivalent to a p H shift for reaction 33 (Table 6-2) of 0.2. 125 Unti l recently it has been standard procedure to use the Criss-Cobble method or its variant, the L I H C A , to calculate the heat capacities of the ammines of N i to high temperatures. Unfortunately, the use of the original L I H C A for this purpose-is predicated on the assumption that these complex ions behave as simple cations. The L I H C A constants for simple cations are then used in the heat capacity calculations. There is a good degree of validity to this argument as ammonia is entirely miscible with water in all proportions and is similarly a protonated solvent. Thus Ni(NH3)2v+ is analogous to Ni(H20)2/ . However, to circumvent this assumption the so-called \" N i model\" was developed by Osseo-Asare and Asihene and it has fit the equilibrium constants (K) measured by Letowski [Letowski 1970] to the L I H C A , thus generating a separate set of constants, cf. Table 6-1 [Osseo-Asare 1981, 2004]. It should be emphasized that this model is solely based on the only available experimental high temperature equilibrium constant data for Ni-ammines and that the temperature of the experimental work was limited to 453 K [Letowski 1970]. Furthermore, the experimental method used in Letowski's work was E M F measurement. Briefly described, this method used an induced potential difference between two identical \u00E2\u0080\u00A2 2+ \u00E2\u0080\u00A2 electrochemical cells containing ammonium nitrate when a known amount of N i ions were introduced into one of the cells. However, the use of E M F measurements to calculate K data relies on an accurate knowledge of the concentration of free NH3 at high temperature. Indeed, as Muendel et al. have pointed out, considerable uncertainty about the state of ammonia in solution exists and some means of distinguishing between ammonia (NH3) and ammonium ( N H / ) should be found when trying to determine equilibrium constants based on E M F measurements [Muendel 1961]. To this end, the 126 ammoniacal solution from the liquid phase of the autoclave employed by Letowski was sampled at high temperature, condensed to room temperature and then titrated to measure the free NH3 concentration. Ammonia losses to evaporation are likely. Therefore, due to the sampling procedure, it is probable that the free NH3 concentration was underestimated. A s a result, the N i model probably underestimates K values. Thus, while this model is still considered to be the most accurate for the prediction of Ni-ammine heat capacity up to 573 K (it has been used in this work), the stability fields for the Ni-ammines, shown in Figure 6-6 through Figure 6-9 are likely smaller than they ought to be. According to calculations using the N i model, ammine stability is basically ni l at temperatures above -523 K . The results of the calculations from the N i model were compared to results using the simple cation constants from the L I H C A and the original Helgeson formula for complex dissociation (Figure 5, eqn. 9) [Helgeson 1967]. This last method was employed because the revised H K F parameters are not available for ammines. T 2303RT { r co 1-exp T-T exp(6 + aT) -c + \u00E2\u0080\u0094~^-L- 2.303RT ...6-9 The standard reaction enthalpy change, AH2\u00C2\u00B09g , and the standard entropy reaction change , AS\u00C2\u00B0gg, were taken from the data collected by Smith and Martell at zero ionic strength [Smith 1977]. LogKT data calculated in this way was then used to calculate M\u00C2\u00B0Ni(NH T (eqn. 6-11). Unfortunately ju\u00C2\u00B029S data for some of the ammines, necessary \"for eqn. 6-1, were not available so they were calculated from LogK29S values, also published by Smith and Martell , also through the use of equations 6-10 and 6-11. 127 Ni(NH3)2v+ = Ni2+ + vNH3 6-10 AG? = -2.3RTLogKT = (^>r + vju^) - f i \ N H ^ 6-11 The chemical potential of the dinickelite ion, //,\u00E2\u0080\u009E,.\u00E2\u0080\u009E- \u00E2\u0080\u009E , was calculated at all temperatures through the use of the revised H K F model [Shock 1997]. The dinickelite ion (HNiOl) is the traditional notation for the hydrolyzed anionic Ni(II) complex and is generally used in Pourbaix diagrams; this is the notation we have conserved here. It should be mentioned that Beverskog and Puigdomenech, Beas and Mesmer and Tremaine and Leblanc have preferred to use Ni(OH)~, which results in smaller stability regions for this species [Beverskog 1997], [Beas 1976]. Furthermore, this species was omitted from the diagram at 653 K . This was deemed prudent as there is a great degree of uncertainty involved in the prediction of complex oxyanions at supercritical conditions [Kriksunov 1993, 1995]. 6.1.3 Ammonia The chemical potentials of ammonia (pNHjT) and ammonium (pNH+ T) were calculated using the H K F [Plyasunov 2001]. Because the Born coefficient, co, is independent of temperature and pressure for aqueous neutrals, it can be replaced in eqn. 6-7 by \u00C2\u00AE-. \ . i * \u00E2\u0080\u00A2 -* * * \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 NiO 2 ' 1 1 \u00E2\u0080\u00A2\u00C2\u00BB \u00E2\u0080\u0094 i \u00E2\u0080\u0094 ^ 1 \u00C2\u00AE-. Ni2* j NT -\u00E2\u0080\u00A2 - \u00E2\u0080\u009E \ \" \" \u00E2\u0080\u00A2 - . * * \u00E2\u0080\u00A2 % 1 NiO 1 1 1 Ni j -2 -1 4 P H s s 3 K 10 Figure 6-9 Pourbaix diagram for the N i - H 2 0 - N H 3 system, 653 K . [Ni2+ ]T = 1 x 10~b m [ NH3 ]T =l.5m. Line c represents the neutral pH for water while line d represents the NH31 NHl equilibrium. It is important to note that the diagrams presented here are valid for ideal systems of pure substances and take no account of the kinetics of reaction (for instance rate of oxide growth versus dissolution), activity coefficients, and localized phenomena which may affect the corrosion morphology and behaviour. Furthermore, the potential/pH behaviour of alloy 625 is certainly different from that of nickel. These diagrams do, however, offer a good guideline for the study of alloy 625 as nickel is its main constituent. 138 There is a marked shrinking of the region of immunity for N i as temperature is increased; this enlargement of the metal cation stability field has also been noticed by Kriksunov and Macdonald for N i and Fe [Kriksunov 1993, 1997]. In contrast, the regions of stability for N i O and complex oxyanion increase with temperature. A large region of stability exists for nickel ammines between approximately p H 5 and 11 at 373 K and p H 4 and 7 at 473 K (Figure 6-6 and Figure 6-7). A t the temperature increments considered in this work, and according to the Ni-model calculations, the stability field for the ammines disappears between 473 and 573 K . This is a finding which is consistent with the current published literature as ammine stability constants are known to decrease in aqueous solutions [Isaev, 1990], [Letowski 1970], [Osseo-Asare 1981, 2004]. Figure 6-10 presents the stability constants of the ammine-nickel species as a function of temperature and method of calculation. Clearly there is a wide variation in the predicted stability depending on the method used to extrapolate chemical potentials. Generally the calculations are in trend agreement with those presented by Osseo-Asare, in that the Helgeson complex dissociation method (equation 6-9) predicts higher stability for ammines than both the L I H C A and the Ni-model [Osseo-Asare 2004]. There is, however, a surprising disagreement between the Ni-model as calculated here and the original Letowski stability constants on which this model is based. This is clearly a reflection on the differences between the thermodynamic data which was used to derive the Ni-model and that used in this work. For instance, the Ni-model is based on Letowski's experimental reference data (A/7 2\u00C2\u00B0 9 8 and A S 2 9 8 ) whereas the data provided by Smith and Martell was used here. It is interesting to note that even Letowski's LogK29S data is consistently lower than the Smith'and Martell data - which lends further weight to 139 the argument that the Ni-model represents a lower bound on the stability limits of the N i ammines. A s such, the Pourbaix diagrams presented here can be considered a conservative estimate for the stability of these species. Referring to Figure 6-4 and Figure 6-5 it is clear that the use of the original Helgeson model (eqn. 6-9) for complex dissociation would lead to regions of stability on the Pourbaix diagram at 653 K for mono- and diammine-nickel between approximately p H 3 and 5. o 4.00 3.00 2.00 1.00 0.00 -1.00 -2.00 -3.00 -4.00 \u00E2\u0080\u00A2 Helgeson \u00E2\u0080\u00A2 LIHCA \u00E2\u0080\u00A2 Ni-model x Letowski 290 340 390 440 490 540 Temperature, K 590 640 690 Figure 6-10 Stability constants as a function of temperature (LogK] T ) and method of calculation. Letowski's data has been superimposed: LogKx T = -558.42 / T + 9.6953 - 0.01757 Upon closer inspection of Figure 6-9, the issue of ammine stability at 653 K becomes irrelevant. Indeed, the ammonia/ammonium hydrolysis line (line d) lies well within the boundary of the Ni2+ stability field. If one assumes that the p H at 653 K is 140 buffered by the ammonia/ammonium equilibrium in the same way as the p H at 298 K then a solution at pH 2 98K =10 should be at roughly pH653K = 4, indicating that N i would not passivate. This assumption is in fact not baseless. The issue of the significance of p H under S C W O conditions is somewhat arcane and directly measuring it is difficult [Macdonald 2001], [Lvov 1998]. There are three interrelated components which must be considered when estimating p H at high temperature in the N H 3 - H 2 O - O 2 system encountered in S C W O reactors: chemical equilibria, gas solubility and phase equilibria i.e. partial pressures. The three equations determining the chemical equilibria are the ammonia/ammonium stability (eqn. 6-13), the electroneutrality condition (eqn. 6-14) and the ion product of water (eqn. 6-15). In the particular case of the experiments conducted for this work there is also dissolved sulphate which has been included in eqn. 6-14. [ N H , ] = [NH2\JJ\u00C2\u00A3] 6 1 3 ([H+] + Ka) [NH+4] + [H+] = [OH-] + 2[SO; ] 6-14 Kw=[OH~][H+] 6-15 Where K\u00E2\u0080\u009E is the ammonia/ammonium equilibrium constant and [ ] designates molal concentration. Activities have not been considered because they were not considered in the Pourbaix diagrams. Solving equations 6-13 through 6-15 at 653 K using the calculated value for Ka=\ .92xl0\" 4 , the ion product of water at 250 bar Kw=\.67xl0\"13 from Marshall and Franck, the concentration of sulphate [S04~] = 0.25 m and [NH3]lol = 1.5 m, one gets a pH653K of 4.04. The solubility and partial pressures of ammonia affect the available total aqueous ammonia concentration,[NH2] tot as determined by Henry's Law: 141 A s seen previously, S C W O reactors operating at 250 bar maintain a dual phase regime well into the supercritical temperature range (Figure 5-20) because oxygen is only fully miscible with low density/low polarity supercritical water (-0.15 g.cm\"3). A t 250 bar / 653 K water/oxygen mixture density is approximately 0.4 g.cm\"3 and only drops to the process required densities at temperatures close to 873 K . It is wel l known that ammonia solubility decreases with increasing temperature [Kawazuishi 1987]. A s such, ammonia evaporation is possible well into the supercritical region. Using a volumetric equation of state to estimate the infinite dilution Gibbs energy of hydration, Plyasunov et al. have estimated Henry's constant for NH3 in the binary N H 3 - H 2 O system up to the critical point and it was found to be - 5 bar.kg.mol\"1 [Plyasunov 2000]. A thorough discussion of the phase equilibria in the N H 3 - H 2 O - O 2 system is well beyond the scope of this work, however, to get an idea of the effect of varying PNH and [SO^~] at 653 K , equations 6-13 to 6-16 were simultaneously solved for pH using HNH \u00E2\u0080\u0094 5 bar.kg.mol\" 1. The results are presented in Figure 6-11 where the vertical line represents the p H above which N i O precipitates. In order to keep N i from corroding (i.e. p H > 4.69) in the presence of NH3 and SO]' at 653 K , high values of PNH are required. With the concentration of SOl~ used in the S C W O tests at U B C , even PNHj = 20 bar ([NH3]l0, - 4 m) is not high enough to ensure N i passivity. T h e P ^ which led to corrosion in the U B C S C W O reactor was definitely in the range of - 0 to 7.5 bar because [NH3]lol could not exceed 1.5 m. 142 Figure 6-11 Effect of ammonia partial pressure at 653 K , 250 bar. Partial pressures are in bar a n d / / w / ] =5 bar.kg.mor 1. Equations 6-13 through 6-16. Clearly, from a charge balance perspective, i f enough ammonia evaporates such that the concentration of ammonium is lower than twice the concentration of sulphate, ~ 0.5 m, then sulphuric acid would likely form. This is analogous to one of the perennial problems which is known to occur in S C W O reactors when halogenated organics are destroyed: the halogens combine with water to form the corresponding acid. This issue is, however, somewhat besides the point as Figure 6-11 shows that even i f all o f the available ammonia was in solution, the p H would still be too low to ensure passivation. Alternatively, the scenario represented by PNH = 0 could be the result of NH3 conversion to N2 however, as mentioned earlier (p. 11), the literature indicates that this is 143 impossible at 653 K and that even after high residency times at 873 K , NH3 is difficult to convert. It appears as though the presence of both NH3 at low enough concentrations and SO]' at high enough concentrations wi l l in most cases result in high rates of N i corrosion in and around the high density supercritical region. These findings are new but entirely in keeping with the literature which has never reported corrosion problems in S C W O experiments containing ammonia alone and has also found that Na2S04 solutions, for example, result in negligible corrosion rates. In other words, solutions containing NH3 alone w i l l maintain a high p H at 653 K (~ 8.5 atPNH^ =7.5 bar) due to N H 4 O H stability and thus w i l l ensure that the N i is kept in its passive range. Similarly, solutions containing S04~ alone in the form of a non-volatile inorganic salt w i l l maintain a neutral p H . It is clear from the literature and the work shown here that oxygenated S C W O solutions w i l l invariably lead to transpassive corrosion of N i - C r alloys. A resolution to this problem could be to inject oxygen in the low-density region of the reactor, thus removing the oxidative conditions to temperatures where oxides are insoluble. Maintaining the stability of N i O would also help in maintaining reactor integrity and it seems as though p H adjustments would be required to this end. Ammonia at high concentrations appears to be somewhat benign from a corrosion standpoint and could easily be recovered at the effluent end of the reactor. Ironically, the injection of NH3 to high partial pressures (PNH > 20 bar, figure 7) could therefore be an ideal buffer for mitigating the acidifying effect of SO]' originating from ammoniacal salts or even organic feed stock. 144 Finally, it should be reiterated here that the definitive stability of Ni-ammines (as discussed on page 128) at ~ 653 K has not been determined as the equilibrium constants employed here may in fact underestimate them. A s a result any attempt to regulate p H by ammonia should be preceded by the determination of newer more reliable (at least in this author's estimation) equilibrium data. 6.3 Summary \u00E2\u0080\u00A2 Pourbaix diagrams confirm that ammoniacal sulphate solutions can lead to preferential N i dissolution through the formation of ammines up to approximately 473 K . The Osseo-Assare Ni-model, which is the only published model based on high temperature equilibrium data, predicts that Ni-ammines should become unstable between 473 and 573 K . \u00E2\u0080\u00A2 Analysis of Pourbaix diagrams showed that the low ammonia concentration and partial pressures (~ 1.5 m) led to acidification of the feed solution and the dissolution of N i as N i 2 + , thus obviating the formation of N i O . \u00E2\u0080\u00A2 The low solubility of ammonia at low supercritical temperatures could also lead to the formation of sulphuric acid when sulphates are present. \u00E2\u0080\u00A2 Higher concentrations of ammonia (PNH > 20 bar) may result in better materials performance through p H adjustment. 145 7 CORROSION AT SCWO CONDITIONS 7.1 Failure analysis During the warm up phase.of a test following a run simulating \"red water\" discharge from a dinitro benzene plant (2,4DNP with ammonium sulphate and free ammonia at pH 2 98K ~ 10, 250 bar), a failure occurred in the first preheater (PHI) of the U B C S C W O reactor at temperatures not exceeding 570 K . On a later cold static pressure test at 500 bar the tube side of the recuperative heat exchanger ( R H X ) burst. Solutions collected at the sampling ports during the treatment of only 58 litres of the red water solution were blackish in colour and contained a visible particulate which could only be metal precipitates [Vera-Perez 2002]. The highest levels of metals were collected from the PH2 sampling port, immediately down-stream of P H I . In all cases high contents of dissolved nickel, chromium, molybdenum and iron were found (Table 7-1). A wall thickness and S E M / E D X analysis of the failed portions of tubing follows. Wal l thickness measurements were taken using an ultrasound at locations adjacent to the corroded section as well as at the corroded section itself. Four sets of measurements 90\u00C2\u00B0 apart were taken every one inch along the tube. Figure 7-1 shows the tube wall thickness and the estimated bulk temperature profiles for P H I [Vera-Perez 2002]. Material loss was highest in the high density supercritical region, approximately 650 to 655K. Moreover, little to no change in wall thickness occurred within a maximum distance of 0.8m from the failure site. Corrosion below temperatures of 640K and above temperatures of 660K was then not significant. 146 o 3 ro c/> w 3 3 1.6 1.8 2 Axial Position, PH1 [m] Figure 7-1 Wal l thickness measurements and temperature profile near failed tubing location for P H I [Vera-Perez 2002] Table 7-1 Metals content in feed, in sampling ports throughout the reactor (distance from failure) and in the effluent (Eff) , in ug/g [Vera-Perez 2002] Metal Feed ng/g 5 in 10 m 30 m 55 m Eff. Total % total dissolved Chromium <0.05 2640 891 437 836 5 4809 14.4 Iron <0.2 1020 351 134 1190 <0.2 2695 8.1 Molybdenum <0.1 1400 800 344 1120 231 3895 11.7 Nickel <0.05 9520 4710 2230 5350 81 21891 65.8 Total <0.4 14580 6752 3145 8496 317 33290 100 147 (c) (d) Figure 7-2 (a) S E M micrograph of epoxy mounted alloy 625 tube section exposed to ammoniacal sulphate solution at ~ 635 K , 250 bar (b) E D X map showing alloy (pink) and Cr enriched ol (blue) and (c) Cr enriched ol (d) line scan showing N i depletion and Cr enrichment in o l , slight M o enrichment is also apparent. 148 Cross sections of the tubing were analyzed using S E M and E D X . Selected samples were etched using a mixture of hydrochloric, acetic and nitric acid to check for sensitization. One of the main reasons alloy 625 is routinely used in high temperature processes is that it is known to have excellent sensitization resistance. Precipitates (niobium carbide and titanium nitride) found at the grain boundaries in the corroded material were similar both quantitatively and qualitatively to the ones found in the drawn condition. Figure 7-2 shows S E M micrographs and E D X maps of a section of alloy 625 taken from P H I . A l l the samples shown here were carefully cut from the tubing, epoxy mounted and polished prior to study under the S E M . The sample shown in Figure 7-2 was exposed to temperatures of approximately 635 K in the S C W O reactor i.e. just before the temperature at which the wall thickness measurements started to show thinning (640 K ) . There is clearly Cr enrichment in the oxide (\"oxide\" is used here to describe the outer layer, \"scale\" may also be used) which is some 15 urn thick. Perhaps of most interest is the fact that almost no N i was found in the oxide. According to these observations and to the calculations presented in Chapter 6, this is indicative of an acidic p H . This finding also echoes, from a qualitative standpoint, the morphology and surface characteristics found by other authors [Schroer2002], [Kritzer 1998]. Moreover, the oxide appears to be defective and the metal/oxide (outer-layer) interface is rough. This is consistent with the idea that the outer layer is formed by precipitation of ejected cations and that it is therefore inherently porous. In Figure 7-3 and Figure 7-4 morphological evidence of the outer layer's dissolution is shown. A s temperature was increased by only ~ 5 degrees between Figure 7-2 at ~ 635 K and Figure 7-3 at ~ 640 K , oxide heterogeneity including cracks and 149 macro-porosity were developed. A t 645 K (Figure 7-3 (c)) the oxide was even more degraded and magnification reveals what appears to be inter-granular attack (IGA)(Figure 7-3 (d)). Figure 7-4 shows macroscopically visible surface roughening and wall thinning and confirms that the outer layer was removed between 645 and 660 K . A s discussed in Chapter 5, p. 102, it is possible that chemical dissolution of the oxides occurs near ~ 650 K . However, given the current observations with regard to the precipitation of the outer layer and obvious stability of Fe and Cr oxides at S C W O conditions, it seems more likely that a mechanical process occurs to remove the oxide. It is hypothesized that the rate of metal corrosion and grain boundary attack at ~ 658 K (Figure 7-4 (c)) is such that the porous and agglomerated scale (or outer layer) is undermined by the removal of its support (the alloy below it). This would result in precipitates being \"let loose\" in the reactor and is a probable explanation for the \"murky\" blackish water observed during the experiment. 150 (c) (d) Figure 7-3 (a) E D X map overlaid on an S E M micrograph of the oxide layer on the inside surface of alloy 625 tubing exposed to ammoniacal sulphate solution at ~ 640 K , 250 bar. The oxide is visibly cracked, (b) ~ 642 K (c) ~ 645 K (d) ~ 645 K (magnified) showing intergranular attack and M o enrichment at the ol/s interface (yellow colour) 151 Figure 7-4 (a) Macroscopically visible surface roughening and only patches of oxide (circled area) are visible at ~ 650 K . (b) 20X optical microscope picture o f the bakelite mounted cross section ~ 650 K . Wal l thinning is obvious, (c) Approximately 658 K : point of failure. Oxide is either completely removed (not shown here) or deep I G A is apparent, (d) ~ 658 K obvious wall thinning and ductile elongation are visible. 152 7.2 316 L Stainless steel reactor experiment 7.2.1 Experimental To determine the effect of ammoniacal sulphate solution on the corrosion characteristics of a commonly used austenitic stainless steel (316 L - Table 2-1) at S C W O conditions, a bench-top S C W O reactor was built. The reactor consisted of a 20 ft (6 m) length of seamless V\" (6.35 mm) 316 L tubing which was fed by two high pressure piston pumps and heated throughout its length by ceramic heaters. The system was also fitted with an in-line test cell which housed coupons for weight loss measurements on two alloys and two metals (Nb, N i , N i - 20 Cr, 625) of interest to this study. A schematic of the apparatus is provided in Figure 7-5 where the test conditions used for the experiment are also shown. Pulsation dampener Pressure / \" \ transducer 5000 PSI Pressure valve relief HP pump 50 vol% H 2 0 2 (35wt% sol.) 5 mL/min 8 (A L\" HP pump 8 wt% NH dOH (30 wt% sol.) 2 wt% H 2SO, pH = 9.9 15 mL/min Test Cell 1 L\" 1 L J L\" h e a t e r s 1,2,3,4 n c 4 - A -3 L~ \"J c -a-Double pipe heat exchanger a Controlled or alarmed 2 -a -1 C J L Back pressure regulator G/L separator Figure 7-5 Schematic of bench-top S C W O apparatus 153 The solution used for this experiment was the standard 1.6 m NH3, 0.2 m SO2 , p H ~ 10 solution used throughout most of this work. Oxidant was delivered in the form of hydrogen peroxide (H2O2). Several H2O2 decomposition runs were made prior to the corrosion experiment to ensure that it decomposed to O2 in an adequately fast manor, so as to provide the desired O2 concentration in the temperature range o f interest ~ 653 to 673 K and in the test cell. Decomposition rates of up to 50 % were measured at 583 K in distilled water, and a total flow rate (fast) of 170 ml/min. Considering that peroxide decomposition is catalyzed by the reactor tube wall and that the flow rate for the corrosion experiment presented here was roughly 25 mL/min and also that temperatures reached 753 K at the outlet of heater bank 3, it is qualitatively estimated that all o f the peroxide was decomposed prior to reaching the test cell which contained the coupons (Figure 7-5). This would yield a concentration of roughly C0i = 3 m in the supercritical region. This concentration is similar to the one employed in the larger S C W O reactor and which led to failure (~C0i = 4 m, section 7.1, p. 146). The test cell consisted of a 316 L cylindrical vessel mounted in-line with the direction of solution flow. The solution was made to pass through tubular glass sample holders that were designed to stack (horizontally) in the test cell. The sample holders had an inlet diameter which was larger than the outlet diameter. Test coupons were rectangular and approximately 1 cm x 0.5 cm x 1.5 mm so as to fit inside the larger diameter section of the glass holders, but also so as not to fit through the smaller diameter section. A schematic of the test cell is given in Figure 7-6. The glass holders were necessary to remove any possibility of galvanic interaction. 154 Figure 7-6 Sectional view of test cell coupon assembly. The tested coupons were ground to 600 S iC, cleaned, dried and weighed prior to being put in the test cell. The bench-top system was brought to temperature with distilled water over a period of approximately 1 hour such that the temperature at the exit of heater bank 4 was ~ 673 K . The temperature profile was also arranged so that high-density supercritical fluid was obtained within the 316 L tubing, i.e. the fluid was low- density supercritical in bank 4 and high-density supercritical in bank 3 and in the test cell (Figure 7-5). Temperature was measured by thermocouples on the outside of the tubing and the whole system was well insulated. The temperature inside the tube was assumed to be roughly the same as that of the thermocouples. A s the hot fluid exited heater bank 4 it went directly into the test cell (Figure 7-5). Because the test cell was insulated but not heated, the incoming fluid was cooled as it passed the samples. The test cell was thermocoupled in six locations and temperatures measured on the surface of the cell and at the solution exit did not go below 653 K . It can then be safely assumed that the temperature in the cell was between 653 and 673 K - exactly the temperature range of interest. 155 7.2.2 Results and discussion After the ~ 1 hour heat-up period, the influent feed stream was switched from distilled water to the ammoniacal sulphate solution and hydrogen peroxide mixture. Within approximately 10 minutes oxygen bubbles were visible in the G / L separator (made from transparent acrylic). Approximately 35 minutes later the reactor effluent became dark brown in colour and contained suspended solids. Moreover, the system became unstable: temperatures decreased and pressure rose. The experiment was terminated and the system was flushed with distilled water and allowed to cool for 40 minutes. The effluent was collected and filtered to remove suspended solids. The filtrate was sent for inductively coupled plasma analysis to determine dissolved metals. The filtered cake was dried and weighed and it was found that 11.1 grams of solids had been removed from the reactor. The solids were analyzed by E D X and found to contain approximately 33 wt% 0 2, 26 wt% Fe, 22 wt% Cr, 8 wt% N i , 5 wt% S and 4.3 wt% M o . It was thus determined that roughly 6.9 grams of stainless alloy had been corroded in the 45 minute time period for which it was exposed to the ammoniacal solution. If the entire surface area of the 6 m tube had corroded evenly this would mean a corrosion rate of approximately 140 mm.yr\"1 or complete tube dissolution in roughly 78 hours. If the 1.5 m length of tube exposed to the high-density supercritical solution is considered, the corrosion rate becomes 560 mm.yr\"1, or wall thickness penetration in 20 hours. O f course, in reality the length of tube affected is l ikely smaller than either of these 156 calculations would suggest such that the corrosion rate is obviously completely unacceptable. The dissolved metals concentrations are presented in Table 7-2. Table 7-2 Concentrations of dissolved metal after SCWO experiment in 316 L tube Fe Cr Ni Mo Nb~ mg.L\"1 0.5 \u00C2\u00B1 0 . 1 1 0.17 \u00C2\u00B1 0 . 0 5 1.18 \u00C2\u00B1 0 . 0 3 1.28 \u00C2\u00B1 0 . 0 1 < 0.05 The order of solubility was Mo > Ni > Fe > Cr > Nb. This is similar to the order for the effluent concentrations collected from the tubing failure presented in section 7.1, Mo > Ni > Cr. This finding is also consistent with the room temperature polarization experiments which showed that Mo corroded readily in ammoniacal sulphate solution while the passive current on Ni was higher than that on Cr (Figure 4-7). The high rate of Ni and Mo dissolution (mostly as compared to Fe) is surprising given that the alloy concentrations are so minor: for 316 L Ni ~ 12 wt%, Mo ~ 2.5 wt%. The alloy 625, Ni-20Cr, Nb and Ni samples were removed from the test cell with great difficulty. It was found that the glass sample holders had expanded within the cell and fused themselves to its wall. Also, it was observed that the glass had either fragmented under the expansion pressure or reacted with the solutions to form compact and sand-like powder. To extract the coupons the test cell was carefully face milled until the exposed glass holders were visible. The sand like product was then removed with an acid brush and the samples were collected. The exposed coupons were all covered with a black scale (Figure 7-7 and Figure 7-8). They were first weighed after exposure without cleaning, and then were cleaned using a dilute HC1 solution with an inhibitor to remove the scale and subsequently re-157 weighed. The scale morphology is interesting as it appears to be the result of an agglomeration of spherical deposits. These deposits were analyzed with E D X and found to be roughly o f the same composition as the dried solids filtered out from the effluent. Thus it would appear as though the Fe and Cr oxides, while insoluble, were removed from the 316L alloy surface by fluid motion. The implication is that they do not form a solid and adherent oxide, rather that the reaction products are loosely released into the solution. This substantiates the idea that rapid dissolution at the m/ol interface leads to the complete removal of the ol through physical means rather than chemical means. This also adds further weight to the diffusion limited corrosion theory proposed in this work. Figure 7-7 High magnification morphology of deposits on the surface o f coupons exposed to 653-673 K S C W O environment. E D X spectra reveal that the porous scale is mostly Fe oxide and Cr oxide removed from the 316 L reactor tube. 158 Figure 7-8 Morphology of porous scale The measured corrosion rates for the cleaned, de-scaled coupons are presented in Table 7-3 and they reveal a very consistent picture of the corrosion behaviour of N i - C r alloys at S C W O conditions. The corrosion rates decreased with increasing Cr content. The decrease in corrosion rate can be attributed to the precipitation of an insoluble Cr(III) oxide. Furthermore, and perhaps most interestingly, Nb showed no attack up to at least 653 K . This is an entirely new result which provides extra incentive for examining the possibility of utilizing Nb in S C W O reactors. 159 Table 7-3 Coupon corrosion rates after exposure to ammoniacal sulphate solution at 653-673 K , C0 ~ 3 m. 625 Ni,20wt%Cr Ni Nb mm/yi\" .18 (3) 29(1) 35 (2) no change (2) (# coupons) 7.3 Summary \u00E2\u0080\u00A2 N i - C r alloys experience unacceptable corrosion rates when exposed to heavily oxygenated ammoniacal sulphate solutions in the high density supercritical region. The morphological evidence suggests that a porous agglomerate of spherical Cr(III) oxide particles is formed on the surface of these alloys. N i and M o (for N i - C r - M o alloys) dissolve readily. This morphology is consistent with the electrochemical measurements performed at 473 K which predict the formation of a diffusion limiting outer layer. \u00E2\u0080\u00A2 Deep inter-granular attack is likely the mode of failure for alloy 625. \u00E2\u0080\u00A2 Higher Cr content results in lower corrosion rates. \u00E2\u0080\u00A2 Nb was not attacked under the conditions tested here. 160 8 CORROSION OF NIOBIUM IN CONCENTRATED SULFURIC AND HYDROCHLORIC ACID AT ELEVATED TEMPERATURES Niobium (Nb) has received a great deal of attention in the. chemical industry owing to its corrosion resistance in low concentration acidic solutions and at moderate temperatures. Many corrosion studies have been undertaken on pure niobium in sulphuric acid and hydrochloric acid solutions, however most have focused on corrosion rate measurements by gravimetric means only and have not included thermodynamic considerations, micro-structural corrosion morphology or polarization data. Moreover, there has been an emphasis on test conditions at temperatures at or above the boiling point of the acid solutions. For instance, Lupton et al. exposed Nb to boiling 70 and 75 wt% (%) H 2 S 0 4 (boiling points of 165 and 185\u00C2\u00B0C respectively) [Lupton 1981]. Burns et al. have published corrosion rates for Nb in boiling 10 to 70 % H2SO4 and 10 to 36 % hydrochloric acid (HC1) with various salt additives (up to 15 % A I C I 3 , 10 % N a C l and 0.01 % FeCfj) [Burns 1984]. Bishop tested several refractory metals including niobium in H2SO4 and HC1 solutions at and above their boiling point (notably at temperatures up to 250\u00C2\u00B0C) [Bishop 1963]. Robin and Rosa also determined iso-corrosion curves for Nb and Nb-Ta alloys in HC1 solutions with concentrations ranging from 5 to 20 % HC1 and at solution boiling point, 150\u00C2\u00B0C and 200\u00C2\u00B0C [Robin 2000]. Exposure time was between 3 and 14 days. MacLeary presented corrosion rates for Nb under a variety of conditions and acid type/concentration also mostly at the boiling point [MacLeary 1962]. 161 On the lower end of the temperature scale, Covino et al. have presented results for the corrosion rate of Nb in HC1 at temperature ranging from 35 to 100\u00C2\u00B0C and concentrations of HC1 of 1, 5 and I O N (equivalent to roughly 4.8, 21.2 and 38.1 %) [Covino 1980]. Robin et al. have tested Nb and Nb-Ta alloys over a wide range of sulphuric acid concentrations (10 to 98 %), between room temperature and 175\u00C2\u00B0C and at exposure times varying from 1 to 19 days [Robin 1991, 1997]. Bulhoes and Joanni have also studied the effect of temperature up to 40\u00C2\u00B0C on Nb in a 96 % solution of H2SO4 and corrosion rates were given as current densities [Bulhoes 1984]. These low temperature data are summarized in Figure 8-1 and Figure 8-2 for H2SO4 and HC1 solutions, respectively. 150 125 o 1 0 0 0> I 75 a. 50 25 \u00C2\u00A9 Robin, de Almeida and Nunes -1991 1 Bulhoes and Joanni -1984 OMacleary -1962 \ > 1000 \ > 1 0 0 \u00C2\u00B0 (100-256. \ > 1000 0<25 <25 <25 Q < 2 5 <25 0<25 Q < 2 5 <25 20 40 60 H2S04 Concentration (wt%) 80 100 Figure 8-1 Corrosion rates in um/yr for Nb in H2SO4 solutions at temperatures below the boiling point. Circle diameter reflects corrosion rate value rather than an area on the graph i.e. each corrosion rate value sits on a discreet point in the concentration-temperature plane 162 100 75 o IS 50 25 0 <25 <25 G <25 <25 (3<25 G < 2 5 O Covino, Carter and Cramer -1980 10 20 HCI Concentration (wt%) 30 25-100 I 25-100) ^ y \ <25 40 Figure 8-2 Corrosion rates in um/yr for Nb in HCI solutions at temperatures below the boiling 100\u00C2\u00B0C. The main objective of the present work was to investigate the corrosion behaviours of Nb in solutions of H2SO4 and HCI at 75 and 95 \u00C2\u00B0C by evaluating its corrosion resistance, corrosion morphology, thermodynamic tendencies and polarization behaviour. It was hoped that investigation of the polarization response, in particular, may shed light on its resistance to high temperature oxidative media. Furthermore, resistance to acidic conditions is one of the important criteria for selecting SCWO materials. A metal's resistance to oxidative media can be tested by exposure (high temperature with high oxidant concentration) or by electrochemical polarization at potentials sufficiently removed from Ecorr. The exposure route was investigated in the last section and the results were encouraging though the sample size was small. The 163 autoclave used in section 5 for electrochemical measurements is currently not capable of exceeding temperatures o f -498 K ; nor is it advisable to expose it to strong acids and high, SCWO-l ike , oxygen concentrations. Therefore, in an effort to simulate electrochemical measurements at high temperature and to a lesser degree a S C W O environment, a balance was struck by exposing the Nb samples to exceedingly high concentrations of acid but at moderate temperatures. 8.1 Experimental 8.1.1 Immersion tests Niobium ingots weighing about 0.9 kg each were cut into halves and cold worked to about 85 % rolling reduction and a thickness of 0.5 mm, suitable for immersion testing. They were then annealed at 1100\u00C2\u00B0C for 1 hr under vacuum. The average grain size after annealing was -77 urn as measured by the mean intercept method. Figure 8-3 shows an optical micrograph of the microstructure after etching. Immersion tests were conducted in 20, 40 and 80 % H2SO4 and in 20 and 38 % H C I at 75 and 95\u00C2\u00B0C. Reagent grade chemicals were used to make the solutions. Rectangular specimens approximately 25 mm x 50 mm x 0.5 mm were cut from the rolled sheets, and cleaned and degreased ultrasonically in acetone. After cleaning, the samples were thoroughly washed with de-ionized water, rinsed in ethanol, dried and weighed to an accuracy of 0.1 mg. In the case of the H2SO4 solutions, the specimens were immersed in 300 m L beakers. The beakers were then sealed with rubber stoppers and placed in a water bath. 164 In the case of the HC1 solutions, which had low boiling points of ~ 106\u00C2\u00B0C for the 20 % HC1 and ~ 48\u00C2\u00B0C for the 38 % HC1 [Degnan 1987], the samples were immersed in a 2000 ml beaker connected to a glass condenser to reflux vapors back into the solution. The specimens were physically separated and hung using a Teflon-coated wire. The beaker was then placed on a hot plate. Testing was carried out for a period of two weeks. Two or three coupons were immersed for some temperature and concentration conditions (cf. Table 8-4) and the reproducibility was within 20%. Figure 8-3 Optical micrograph of the Nb microstructure after etching. Upon completion of each test, the samples were rinsed with water then ethanol and then dried with warm air. Specimens were re-weighed to determine the corrosion and the weight losses were converted into an average corrosion rate (r) in pm/yr using the relationship; r = 8 .76x l0 7 KATPj 8-1 165 2 Where Wis the weight loss in grams, A is specimen area (cm ), Tis the exposure time in hours, and p is the density of Nb (8.57 g/cm ). 8.1.2 Polarization experiments Potentiodynamic polarization tests were carried out in a conventional three-electrode cell using an electrochemical interface. Test solution and temperature conditions were the same as those used for the immersion tests. The electrochemical cell consisted of a commercially available one-liter glass cell, a graphite counter electrode, an external Ag/AgCl reference electrode (saturated KC1 fill solution), and the working electrode (WE). Determination of the corrosion current density, icorr, was done by back-extrapolation of the cathodic Tafel slope to the corrosion potential using a non-linear least squares fitting routine. The Tafel extrapolation was always initiated at least 0.1 V away from the corrosion potential, Ecorr. Chi-squared values were generally less than 10 and always less than 50, indicating a good fit to the polarization data. The passive current, i , at +1 V versus Ecorr was also registered for comparison. Experiments were reproducible to within 30 mV of EC0IT, 7 uA of icorr and 50 uA of i . The corrosion 1 9 rate (r) in um.yr\" can be related to icorr (um.cm\") through Faraday's law: M x i r = 3.28 ^ 8-2 np Where Mis the atomic weight of the metal (for Nb, M\u00E2\u0080\u0094 92.91 g/mole), n is the number of electrons exchanged for the reaction ( n = 5 here, discussion follows in the results section), and p is the density of the metal (for Nb, p = 8.57 g/cm ). 166 Tests were performed after 1 hour of exposure to the solution at a sweep rate of 0.166 mV/s between -0.25 V vs. Ecorr and 1.8 V. Because of the concentrated solutions used in these experiments the liquid junction potential (UP) between the reference electrode and the test solution was calculated at higher temperatures in a method described elsewhere, by Vaughan et al. [Vaughan 2003], and which was based on the Henderson equation [Newman 1973]. 8.1.3 Pourbaix diagrams and thermodynamic calculations New Pourbaix diagrams were constructed using the standard chemical potentials, fj\u00C2\u00B0 for the relevant species from the Bard, Parsons and Jordan text [Bard 1985]. Kelley equation coefficients for condensed phase species were used to calculate heat capacities at 75 and 95\u00C2\u00B0C and were obtained from the HSC thermo-chemical database [Roine 2002]. For all ionic species considered in this work, the chemical potentials at 75 and 95\u00C2\u00B0C were calculated using the Criss and Cobble entropy correspondence principle and ionic heat capacity constants [Criss 1964A, B ] . Estimated activities of species (represented by [ ]) in the H2SO4 solutions at 25, 75 and 95\u00C2\u00B0C were calculated using the first and second dissociation equilibrium constants according to a method described by Tromans [Tromans 1998B]. This was done to situate pH values for use with the Pourbaix diagrams and to calculate the liquid junction potential between the reference electrode and the solution for the potentiodynamic experiments. Briefly, the standard chemical potentials for aqueous H2SO4, hydrogen sulphate (HSOV) and sulphate (SO4\") at 25\u00C2\u00B0C were obtained from 167 published data [Bard 1985]. The dissociation constant was calculated by evaluating the reaction free energy change: AG\u00C2\u00B0 = ^(vM\u00C2\u00B0)products - 2 > / 0 \u00E2\u0080\u009E s =-RT\nkt 8-3 Where v is the stoichiometric coefficient of the corresponding reactant or product species and kt is the equilibrium constant for the following reactions: H2S04 HSO~4, kx = {[H+][HSO;]}l[H 2S04] = 99 at 25\u00C2\u00B0C 8-4 HSO; <^H+ + S024~, k2 ={[H+][S04-]}/[HSO 4-] = lxl0-2 at 25\u00C2\u00B0C 8-5 Given a total molal concentration N then: N = [H2S04] + [HSO; ] + [SO2' ] 8-6 2N = 2[H2 S04 ] + [HSO; ] + [H+] 8-7 B y simultaneously solving equations 8-4 through 8-7 it was possible to obtain approximate values for proton activity in the sulphuric acid solutions. Unit activity coefficients were assumed throughout this process which undoubtedly resulted in activity errors. Nevertheless, Tromans indicates that the results obtained in this way are adequate for order of magnitude estimates [Tromans 1998B]. Mean activity coefficients, y\u00C2\u00B1, for the hydrochloric acid solutions were extrapolated from data published by Marsh and McEl roy [Marsh 1985]. Estimates for the individual ionic activity coefficients (yH+ and y ) were made according to the Bates equations [Bates 1970]. These equations rely on the fact that salt hydration numbers are dominated by the Lewis -Acid type interaction with cations whilst hydrogen bonding with anions is typically weak [Muir 1985]. Thus it is a reasonable thermodynamic assumption 168 that the CI\" ion is not hydrated in concentrated solutions of HC1 and that measured values of y\u00C2\u00B1 can be separated as such: \ogyH+ = logy\u00C2\u00B1 +0.00782/w^ 8-8 logya- = l o g ^ \u00C2\u00B1 -0.00782/zra^ 8-9 Where h is the hydration number for HC1 {h = 8) [Robinson 1959], m is the molarity of the solution and ,m, i The osmotic coefficient is simply another way of expressing the activity of water, [H2O], while v. is the stoichiometric coefficient and mi is the molality of the dissociated species i. Activities of water were calculated using Raoult's law: water vapour pressure over a solution of sulphuric (or hydrochloric) acid divided by water vapour pressure over pure water. The former are tabulated in Perry's Handbook for Chemical Engineers [Perry 1997]. Table 8-lpresents the concentrations, estimated activities, p H and calculated equilibrium constants ( and k2) for the sulphuric acid solutions used in this work while Table 8-2 presents the concentrations, estimated activities and p H for the hydrochloric acid solutions. 169 Table 8-1 Estimated activities of H2SO4 solutions at 25, 75 and 95\u00C2\u00B0C. H2SO4 concentration p, kg/L [it] [HS04] [H\u00C2\u00A304] [S042] pH [H20] kx k2 Wt% M m 20 2~32 2~55 U 4 2~50 2~48 0~06 OM ^OAO 0.912 25\u00C2\u00B0C 40 5.31 6.80 1.30 6.40 6.38 0.41 0.01 -0.81 0.585 98.789 1.0123xl0\" 2 80 14.09 40.78 1.72 31.04 31.02 9.75 0.01 -1.49 0.004 20 232 2~55 U 0 2~37 2~36 0A9 0~00 ^037 0.897 75\u00C2\u00B0C 40 5.31 6.80 1.26 5.72 5.71 1.08 0.00 -0.76 0.619 30.197 2.6111xl0\" 3 80 14.09 40.78 1.67 23.11 23.10 17.68 0.00 -1.36 0.014 20 2~32 ~5 i~09 231 230 025 0~00 5^~36 0.892 95\u00C2\u00B0C 40 5.31 6.80 1.24 5.43 5.43 1.37 0.00 -0.74 0.632 21.589 1.5831xl0\" 3 80 14.09 40.78 1.65 20.78 20.78 20.00 0.00 -1.32 0.018 o Table 8-2 Estimated activities of HCI solutions at 25, 75 and 95\u00C2\u00B0C. HCI concentration P , kG/L R + r c r [C1] rH+ &i PH [H2O] Wt% M m 20 6.02 6.85 1.098 4.15 0.52 3.56 ' 33.16 227.14 -2.36 0.59 25\u00C2\u00B0C 38 12.39 16.81 1.189 47.82 0.03 0.43 8.95xl04 1.51xl06 -6.18 0.15 20 5.87 6.85 1.071 2.59 0.47 3.19 \u00E2\u0080\u00A2 14.47 99.11 -2.00 0.65 75\u00C2\u00B0C 38 12.04 16.81 1.156 14.62 0.03 0.56 6.44xl03 1.08xl05 -5.03 0.22 20 5.81 6.85 1.060 2.15 0.48 3.28 9.66 66.18 -1.82 0.69 95\u00C2\u00B0C 38 11.91 16.81 1.143 9.10 0.04 0.61 2.27xl03 3.82xl04 -4.5 8 0.25 8.2 Results and discussion 8.2.1 Pourbaix diagrams Figure 8-4 presents all three Pourbaix diagrams for the Nb-FbO system at the temperatures considered in this work i.e. 25, 75 and 95\u00C2\u00B0C. The chemical potentials used to calculate the reactions' Gibbs free energy are presented Table 8-3. The original Pourbaix diagram was updated to include the stability regions for N b C V (metaniobate) and N b ( O H ) 4 + (niobium tetrahydroxide). Both species were not considered in the original Pourbaix Atlas as at the time, the thermodynamic data was not available. Table 8-3 Chemical potentials, p., of species at 25, 75 and 95\u00C2\u00B0C Species State\" n (kJ/mol), 25\u00C2\u00B0C fi (kJ/mol), 75\u00C2\u00B0C ft (kJ/mol), 95\u00C2\u00B0C H 2 0 l -237.18 -240.97 -242.65 o2 g 0 H 2 g 0 H + aq 0 -0.52 -1.05 Nb0 3 \" aq -932.10 -930.25 -928.51 Nb(OH) 4 + aq -1208.6 -1211.63 -1215.27 Nb s 0 -1.92 -2.74 NbO s -378.60 . -381.17 -382.29 N b 0 2 s -740.90 ' -743.86 ' \u00E2\u0080\u00A2-745.17 N b 2 0 5 s -1768.10 -1775.50 -1778.75 a s = solid, 1 = liquid, g = gas, aq = aqueous It is unlikely that metaniobate would actually form under most conditions but there is evidence that niobium pentoxide (ISID2O5) is soluble in strongly alkaline solutions [Bard 1985]. A s temperature is increased, the stability plane of metaniobate increases in size and the chemical equilibrium between NbiOs and N b C V shifts to more acidic values. 172 In his work on the corrosion of Nb in sodium hydroxide solutions, Robin states that the corrosion of Nb in alkaline solutions could occurthrough a chemical dissolution-precipitation mechanism: Nb dissolves as ND6O19 \" (hexaniobate) that precipitates as a sodium niobium oxide such as NagNbeOig [Robin 2004A, B ] . Unfortunately a thorough literature search did not result in any thermodynamic data for hexaniobate. -2.0 -2 -1 1 2 3 4 5 6 7 8 9 10 11 12 PH Figure 8-4 Pourbaix diagram for the Nb-Ii^O system at 25, 75 and 95\u00C2\u00B0C. \u00E2\u0080\u00A2 represents the corrosion potential for the 75\u00C2\u00B0C polarization tests and represents 95\u00C2\u00B0C tests. Lines a and b represent hydrogen evolution and oxygen reduction, respectively Niobium tetrahydroxide has not seen much attention in the literature. In fact, the only text in which I was able to find reference to it was that of Bard, Parsons and Jordan [Bard 1985]. Upon study of the Pourbaix diagram presented here, the formation of N b ( O H ) 4 + seems a likely explanation for the dissolution of M ^ O s in strongly acidic 173 solutions. Robin has proposed that steady state corrosion, which in the case of Nb in concentrated acid solutions is observed after roughly 200 hours, is due to film dissolution [Robin 1991]: Nb205 +10H+ =2Nb5+ +5H20 8-11 There are little thermodynamic data to suggest that pentavalent Nb ions exist as hydrated ions, however there is at least some data to suggest that N b ( O H ) 4 + could form [Bard 1985]. The dissolution reaction in acidic solutions (below p H 0.5 at 25\u00C2\u00B0C), could then proceed as either a chemical dissolution of metastable niobium pentoxide or an electrochemical dissolution of niobium dioxide. Nb2Os + 3H20 + 2H+ = 2Nb(OH)\ 8-12 Nb02+2H20 = Nb(OH)+4+e' 8-13 8.2.2 Immersion tests Corrosion rates for the Nb samples after two-weeks of exposure to the H2SO4 and HC1 solutions are shown in Table 8-4. For the sulphuric acid solutions, corrosion rates were the lowest at 75\u00C2\u00B0C (below ~ 34 um/yr). In fact, specimens exposed to the 20 and 80 % concentrations did not show any macroscopically visible signs of attack after two weeks of immersion. The surfaces of the coupons exposed to the 40 % concentration were visibly pitted after exposure and corrosion rates were highest at this concentration for both temperatures. This was a surprising finding as it is generally found that corrosion rates increase with temperature and acid concentration. However, it is well documented that certain alloys perform well under concentrated sulphuric acid solutions. 174 Carbon steel, for example, is successfully used to handle concentrated H2SO4 (65 to 100 %) at ambient temperature and austenitic stainless steels can be used at concentrations above 93 % at ambient temperature and above 98.5 % at 70\u00C2\u00B0C [Brubaker 1987]. While the mechanism of protection for carbon steel is believed to be by sulphate film, the use of austenitic stainless steels relies on passivation [Brubaker 1987]. The analogy to Nb is therefore not far fetched. In fact, alloying stainless steels with molybdenum broadens the applicable concentration range to 90 % and above at ambient temperature [Brubaker 1987]. It is therefore conceivable that Nb be well suited to applications in the 80 % range. Indeed, with reference to equation 8-14 below, it is clear that the rate of niobium oxidation should increase with increasing H2O in solution. The cathodic reaction rate and therefore the overall corrosion rate w i l l depend highly on the concentration of H + (assuming that hydrogen evolution is the dominant cathodic reaction). Thus, the corrosion rate is determined by both the available water in solution and the number of H + ions at the electrode interface, shows that while proton activity is increased on the order of 4x between 40 and 80 % H2SO4, water activity is actually decreased by roughly 44x. The lack of available water may limit the solubility of Nb at the 80 % concentration. A similar finding was observed by Smart and Bockris for iron in solutions of high ionic strength [Smart 1992]. It should however be noted that Robin et al. have reported corrosion rates for Nb at 100\u00C2\u00B0C in 20, 50 and 80 % H 2 S 0 4 at 40, 95 and 250 um /yr , respectively [Robin 1991]. Clearly, these observations require more study; however, the gravimetric results are corroborated by the polarization data discussed below. Raising the temperature to 95\u00C2\u00B0C for the H2SO4 solutions resulted in larger corrosion rate increases as concentration was increased i.e. the corrosion rate in the 20 wt 175 % solution was ~ 1.5 times higher at 95\u00C2\u00B0C than at 75\u00C2\u00B0C while this multiple was ~ 2.5 and -3.75 at 40 and 80 %, respectively. Table 8-4 Corrosion rates for Nb in um/yr from immersion tests T wt% H2S04 wt% HCI 20 40 80 20 38 75\u00C2\u00B0C 10.4 33.6 \u00C2\u00B16.3 b 12.9 9.0 \u00C2\u00B1 0 . 8 \" 43.7 95\u00C2\u00B0C 15.2 a 87.4 48.4 c 25.1 306.2 a mean of 13.6 and 16.8 um/yr b mean and standard deviation for 3 coupons c mean of 46.5 and 50.3 um/yr For the H C I solutions, corrosion rates at 75\u00C2\u00B0C were relatively low, particularly for the 20% solutions. A s was expected, increasing the concentration generally led to higher corrosion rates. A significant increase in corrosion rates was observed upon increasing the temperature from 75 \u00C2\u00B0C to 95 \u00C2\u00B0C. Corrosion rates in the 38% concentration HCI at 95\u00C2\u00B0C were seven times higher than those observed at 75 \u00C2\u00B0C. Aside from the disagreement with Robin's data at the 80 % H2SO4 concentration, the results presented here are generally consistent with those presented by other authors. There is one other small point of disagreement (less than a factor of 4): at 95\u00C2\u00B0C and 38% HCI our sample corroded at 306 (xm/yr whereas Covino et al. report 90.5 um/yr [Covino 1980]. Scanning electron micrographs of niobium samples after two weeks of immersion are shown in Figure 8-5. Generalized pitting is evident on the samples exposed to 20 % H2SO4 (95\u00C2\u00B0C) and 40 % H 2 S 0 4 (75 and 95\u00C2\u00B0C) (Figure 8-5, (a), (c) and (d)). In contrast, 176 only traces of localized attack are visible on the surfaces of the samples exposed to 80 % H 2 S 0 4 and 20 and 38 % HC1 (Figure 8-5 (b), (e) and (f)). These latter samples showed more regular surface removal with patches of roughness in a preferred crystallization direction normal to the surface. This morphology was observed by Lupton et al. on Nb in boiling 75% H 2 S 0 4 [Lupton 1981] and by Vehlow on tantalum in 98% H 2 S 0 4 at 230\u00C2\u00B0C [Vehlow 1981]. According to Vehlow this effect is due to the different binding energies between different lattice planes in the B C C lattice. 8.2.3 Polarization experiments Figure 8-6 and Figure 8-7 show the polarization curves for Nb in H2SO4 solutions at 75 and 95\u00C2\u00B0C, respectively. Figure 8-8 shows the polarization curves for Nb in the HC1 solutions at 75 and 95\u00C2\u00B0C. Corrected corrosion potential (corrected for the liquid junction potential i.e. ECORR = EMEASURED + EUP), corrosion current density, passive current density, corrosion rate in um/yr, liquid junction potential correction (EUP) and the immersion test corrosion rates are listed in Table 8-5 and Table 8-6 for the H 2 S 0 4 and HC1 solutions, respectively. The potentiodynamic (PD) corrosion rates were calculated with equation 8-2 where n was assumed to be equal to 5. This is consistent with the oxidation of niobium to niobium pentoxide: 2Nb + 5H20 = Nb205 +10H+ +10e\" 8-14 177 Figure 8-5 Nb after two weeks immersion at (a) 95\u00C2\u00B0C in 20% H 2 S 0 4 , (b) 95\u00C2\u00B0C, 80% H 2 S 0 4 , (c) 75\u00C2\u00B0C, 40% H2SO4, (d) 95\u00C2\u00B0C, 40% H 2 S 0 4 , (e) 95\u00C2\u00B0C, 20% H C I and (f) 95\u00C2\u00B0C, 38% HCI . 178 Estimated test solution p H and corrosion potential for the P D tests in H2SO4 solutions are reported on the Pourbaix diagrams (Figure 8-4) where \u00E2\u0080\u00A2 represents the 75\u00C2\u00B0C tests a n d U represents the 95\u00C2\u00B0C tests. While the potentials in the 20 and 40% solutions are, from a strictly theoretical standpoint, more representative of N b 0 2 to Nb(OH )4 + oxidation, it is unlikely that N b 0 2 (or for that matter NbO) would form as they are known to be thermodynamically unstable species. Furthermore, given that the P D behaviour is indicative of at least a metastable oxide formation, it is likely that N b 2 05 formation (equation 8-14) and dissolution (equation 8-12) are the competing processes giving rise to the measured corrosion rates. E - p H data for the HC1 solutions is not plotted on the Pourbaix diagram, however Table 8-2 (pH) and Table 8-6 (E) list the data and it is apparent they would also lie along the N b 0 2 to N b ( O H ) 4 + stability line. It is interesting to note that in the H2SC\"4 solutions the corrosion potentials for the 20 and 40 % solutions are within 20 m V of each other at both temperatures while the corrosion potential in the 80 % solution is generally ~ 120 to 170 m V higher, also at both temperatures. In contrast the corrosion potential in hydrochloric acid solutions decreases (~ 200 mV) upon an increase in concentration and does not seem to be overly affected by temperature. From a qualitative standpoint this is explained by the mildly oxidizing nature of sulphuric acid solutions whilst hydrochloric acid solutions are reducing in nature. Thus an increase in the former's concentration would result in higher corrosion potentials and an increase in the latter's would yield the opposite effect. 179 _0.5 1 1 \u00E2\u0080\u0094 \u00E2\u0080\u00A2 \u00E2\u0080\u0094 . \" ' \u00E2\u0080\u00A2 \u00E2\u0080\u0094 ' \u00E2\u0080\u00A2 ' I 1e-7 1e-6 1e-5 1e-4 1e-3 /, A.cm\" 2 Figure 8-7 Polarization curves for Nb in 20, 40 and 80 % H 2 S 0 4 at 95\u00C2\u00B0C 1 8 0 Table 8-5 Potentiodynamic polarization data in sulphuric acid solutions (corrosion potential includes correction for liquid junction) 20 wt% H\u00C2\u00A304 40 wt% H2S04 80 wt% HjS04 75\u00C2\u00B0C 95\u00C2\u00B0C 75\u00C2\u00B0C 95\u00C2\u00B0C 75\u00C2\u00B0C 95\u00C2\u00B0C icorr(pA.cm 2) 17.40 50.46 26.92 81.86 8.03 12.68 ipass(nA.cm~2) 9.13 40.42 9.62 87.31 11.11 9.65 Ecorr(mV,SHE) -92 -140 -108 -121 29 31 Eup(mV) -27 -28' -39 -40 -62 -62 Corr. Rate (pm/yr) - Potentiodynamic 123.8 358.9 191.5 582.2 57.1 90.2 Corr. Rate (pm/yr) - Immersion 10.4 15.2 33.6 87.4 12.9 48.4 Table 8-6 Potentiodynamic polarization data in hydrochloric acid solutions (corrosion potential includes correction for liquid junction) 20 wt% HCl 38 wt% HCl 75\u00C2\u00B0C 95\u00C2\u00B0C 75\u00C2\u00B0C 95\u00C2\u00B0C icorr (pA.Ctn 2) 35.39 39.11 40.34 76.51 ipass (pA.cm~2) 28.25 87.89 185.52 289.09 Ecorr(mV, SHE) -216.30 -257.43 -435.54 -446.56 Eup(mV) -115 -107 -310 -289 Corr. Rate (pm/yr) - Potentiodynamic 251.7 278.2 286.9 544.1 Corr. Rate (pm/yr) - Immersion 9.0 25.1 43.7 306.2 Nb polarization curves exhibited typical passive behaviour over a wide potential range in all solutions. However, current fluctuations were observed in the passive region 181 for the samples exposed to the 20 % H 2 S 0 4 and 40 % H 2 S 0 4 solutions at 95\u00C2\u00B0C and the 20 % H C I solution at 75 and 95\u00C2\u00B0C. It is assumed that these fluctuations are due to the initiation of metastable pits followed by repassivation. This finding is substantiated by the generalized pitting observed in the 20 and 40 % H 2 S 0 4 solutions after immersion testing (Figure 8-5 (a)-(d)), and to a lesser extent by the few pits observed on the samples exposed to the 20 % H C I solution. It is expected that potentiodynamic polarization experiments are a better gauge of metastable pit initiation than simple observation of the corrosion morphology after immersion testing. However, it should be noted that polarization experiments performed under the conditions used for this work are difficult. The primary difficulty lies in the stability of the working electrode holder. Most experiments were performed using an epoxy resin mount and insulating paint was used to cover the epoxy/Nb interface. It was noticed, however, that in the H 2 S 0 4 solutions at 95\u00C2\u00B0C the epoxy became unstable, and so flag electrodes were used at this temperature. While we were careful to inspect all electrodes for stability and to repeat all experiments at least once, it is within the realm of possibility that crevice effects for the epoxy held samples or increased localized attack at the water line for the flag electrodes played a role in the measured fluctuations. 182 1e-7 1e-6 1e-5 1e-4 1e-3 1e-2 /, A.cm' Figure 8-8 Polarization curves for Nb in 20 and 38 % HCI at 75 and 95\u00C2\u00B0C Corrosion current densities were estimated by Tafel extrapolation to the corrosion potential and Tafel slopes ( Bc) are reported in Table 8-7. There was no apparent pattern to the Tafel slopes other than a general decrease as temperature was increased. Under the high field conditions for which Tafel kinetics are valid (\rj\ \u00C2\u00BB RT/F , where rj is overpotential and R and F have their usual meaning), Bc is given by equation 8-15. acnF where ac is the cathodic transfer coefficient and is usually assumed to be ~ 0.5 [Bockris 1970], [Vetter 1967] and n is the number of electrons taking place in the charge transfer limited reaction. Clearly, and in contradiction to the results presented in Table 8-7, one would expect an increase in Tafel slopes as temperature is increased. Furthermore, aside 183 from the values for the 38 % H C I solution, the fic values measured in this work are very high despite the elevated temperatures, usually a sign of concentration polarization. It is then likely that the observed Tafel slopes are a result of the natural aeration conditions under which the experiments were performed. Such conditions would lead to two cathodic processes occurring on the W E : oxygen reduction and hydrogen evolution, shown in equations 14 and 15, respectively. 02 +4e\" +4H+ = 2H20 8-16 2H++2e~=H2 8-17 Table 8-7 Cathodic Tafel slopes, Bc, in H2SO4 and H C I solutions (mV/dec). T H2SO4 HCI 20 40 80 20 38 75 291 287 295 255 151 95 226 276 268 125 136 A s the corrosion potentials for all species are below the hydrogen evolution line (line a in Figure 8-4), the oxygen reduction reaction was almost certainly diffusion controlled [Stansbury 2000]. This was likely exacerbated by the elevated viscosities of the test solutions. Therefore, the cathodic Tafel slopes are representative of the sum of oxygen diffusion and hydrogen charge transfer polarization. A s the temperature was increased from 75 to 95 \u00C2\u00B0C, the solubility of oxygen was reduced [Tromans 1998A] and this resulted in lower corrosion potentials (Table 8-5 andTable 8-6) and a more thermodynamically stable hydrogen evolution reaction. This is consistent with the 184 corrosion potential and Tafel slopes observed at 95\u00C2\u00B0C in both H C l solutions (125, 20% and 136, 38%) where the values are closer to what one might expect for pure charge transfer control over the hydrogen evolution reaction. While these findings do not invalidate the corrosion rates obtained from the polarization experiments, they do suggest that de-aeration prior to polarization would yield more valid cathodic Tafel information. In this study however, a comparison of corrosion rates measured by both immersion and potentiodynamic polarization was of interest and this is why the solutions were not de-aerated. Passive current densities, i (Table 8-5 andTable 8-6) are also reported for comparison and are generally trend consistent whTn' c o n. and within the same order of magnitude. In the H2SO4 solutions i is an average of 81% lower thanz'cor(. while in the H C l solutions, i is an average of 286% higher than icorr. This is an indication that Nb is better suited to service in sulphuric acid, and in particular, that oxidizing conditions are to be avoided in H C l solutions. The corrosion rates from the immersion tests are, of course, lower. This is normal as corrosion current estimates based on P D tests are generally higher than those collected through weight loss experiments. In this work, samples were held at open circuit in solution for one hour prior to polarization and potential sweep rates were low (0.166 mV/s). Notwithstanding, the accelerated nature of the P D tests still did not account for the time necessary to achieve steady state corrosion. It is known that corrosion rates for Nb in acid solutions decrease with time. Such a decrease is typical of the thickening oxide film formed on the metal's surface and the increasingly protective nature of the film as time elapses. Robin et al. have presented gravimetric data for corrosion rates in boiling (~100\u00C2\u00B0C) 20 % H2SO4 as a function of time at 1, 2, 4, 10 and 185 18 day intervals [Robin 1991]. Their data were fitted to a logarithmic function giving the corrosion rate as a function of time: r = -19.8051nf + 225.64 8-18 Where r is the corrosion rate in um/yr and t is the time in minutes. Considering the one hour immersion at open circuit, and 0.25 V at 0.166 mV/s for the potentiodynamic polarization between the start potential and corrosion potential, t is roughly 85 min for our polarization experiments. According to equation 8-16, r should then be approximately 138 um/yr which is within the same order of magnitude as that obtained in this work by potentiodynamic means (358.85 um/yr) and differs only by a factor of - 2.6. Similarly, Covino et al. have presented gravimetric corrosion rates as a function of time in 38 % H C I at 85\u00C2\u00B0C and the data was fitted to r(t) given by equation 8-17. r = -40.563 ln7 + 671.39 - \ ' : ' 8-19 According to equation 8-17, r = 491 um/yr after 85 minutes which lies well between the values obtained in this work for the 38 % HCI solutions at 75\u00C2\u00B0C (286.87 um/yr) and 95\u00C2\u00B0C (544.12 um/yr). 8.3 Summary Nb forms a passive film in concentrated sulphuric and hydrochloric acid and at temperatures below 95\u00C2\u00B0C. New Pourbaix diagrams for the N b - H 2 0 system, weight loss experiments and potentiodynamic polarization have shown that Nb exposed to concentrated HCI or H2SO4 solutions likely corrodes through a competing process involving the oxidation of Nb to NH^Os and the chemical dissolution of this oxide to N b ( O H ) 4 + . Corrosion rates for Nb in 20 and 38 % HCI at 75 and 95\u00C2\u00B0C increase with H C I 186 concentration and temperature. In the H2SO4 solutions, corrosion rates increased from the 20 to 40 % concentrations but decreased between the 40 and 80 % concentrations at both temperatures. More study is required to determine an exact corrosion rate/concentration profile for Nb in H2SO4 solutions. S E M micrographs revealed generalized pitting at the 20 (95\u00C2\u00B0C) and 40 % (75 and 95\u00C2\u00B0C) H 2 S 0 4 concentrations while the remaining test conditions resulted in preferential dissolution along certain crystallographic planes. Results of the polarization tests are consistent with the timescale involved in oxide thickening. The corrosion rates presented here are generally in good agreement with the literature. With regards to S C W O , it is apparent that Nb is better suited to service in sulphuric acid than in hydrochloric acid solutions. Also, despite metastable pitting at high anodic potentials, wholesale fi lm breakdown did not occur. Most of the S C W O research to date involving H2SO4 has tested solutions in the 0.2-0.5 m range. If this is representative of the concentration under S C W O conditions then the concentrations tested her are approximately an order of magnitude higher. Implications of this research are threefold: \u00E2\u0080\u00A2 Nb is likely to outperform T i in acidic sulphate bearing solutions \u00E2\u0080\u00A2 Nb is likely to outperform N i - C r alloys in high density solutions due to superior resistance to corrosion at high anodic potentials \u00E2\u0080\u00A2 Nb as a major alloying element to N i (~ 12% range) should be investigated as it may provide superior resistance to transpassive corrosion. 187 9 CONCLUSIONS AND RECOMMENDATIONS The electrochemical and thermodynamic analysis undertaken in this work to study the effects of oxygenated ammoniacal sulphate solution on the corrosion of N i - C r and stainless alloys has generated the following conclusions, tentative findings and recommendations. 9.1 Conclusions \u00E2\u0080\u00A2 Potentiodynamic, morphological and solution analyses have shown that the corrosion behaviour of alloy 625 is dominated by Cr in ammoniacal sulphate solutions between 298 and 673 K . Cr(III) oxides are responsible for passivating the alloy, while N i and M o dissolve readily. \u00E2\u0080\u00A2 Corrosion potential and impedance data have shown that transpassive corrosion of the cation conducting barrier layer on alloy 625 occurs at T > 373 K , P > 40 bar in oxygen saturated solutions. Furthermore, linear polarization, potentiodynamic and impedance experiments have shown that corrosion rates increase exponentially with temperature and become increasingly diffusion controlled, particularly as the barrier layer is transpassively attacked. This transpassive attack of the barrier layer results in cation emission to the barrier layer / outer layer interface and subsequent precipitation of mostly Cr(III) oxides which then reduce the ingress of oxygen to the barrier layer surface. The removal of this precipitate appears to occur in the 653 to 673 K range at which point the alloy becomes completely unprotected. 188 \u00E2\u0080\u00A2 Due to transpassive corrosion and solution acidification, N i - C r , N i - C r - M o and stainless 316 L are not adequate materials for use in S C W O under high-density supercritical conditions. Nb is a good candidate material for this temperature range, particularly in the presence of sulphates. 9.2 Tentative findings This work has also produced some results which require further investigation: \u00E2\u0080\u00A2 The process leading to final removal of the Cr(III) oxide precipitates from the alloy surface at T > 653K is not clear. It is hypothesized that as temperature is increased near the critical point, failure is induced by either solution acidification leading to chemical dissolution of Cr(III) oxide precipitates or physical undermining of these precipitates due to rapid intergranular attack occurring at the metal / outer layer interface. To estimate solution p H at high density supercritical temperatures and rule out the former, a thorough set of H2O-NH3-O2 phase equilibrium calculations should be performed. \u00E2\u0080\u00A2 Thermodynamic analysis of the N i - N H 3 - H 2 0 system based on the only available experimental equilibrium constants measured by Letowski in 1970 has shown that nickel-ammines become unstable at T > 473 K which means that N i - C r alloys are probably not directly affected by ammonia under S C W O conditions. This is a tentative finding in that the equilibrium constants employed appear to underestimate the stability of the ammines. New empirical equilibrium constant data is required to substantiate this claim. A new series of carefully performed E M F measurements should be undertaken to this end. 189 9.3 Recommendations The preceding discussions have also highlighted further areas of interest for future work: \u00E2\u0080\u00A2 Time. The effect of time is always an issue for electrochemical corrosion measurements as they have been conceived specifically to accelerate data collection and estimate long term corrosion rates. We have seen in Chapter 4 that \"steady-state\" conditions are difficult to attain from a practical standpoint as they require ~ 8 hours to develop. A s a result, the effect of time should be investigated on the isothermal and isobaric response of N i - C r alloys to oxygenated solutions. \u00E2\u0080\u00A2 473 - 623 K . The temperature ranges investigated in this work did not include the 473 to 623 K range. From a S C W O standpoint this may not be the most interesting temperature range but it may yield very interesting scientific data, particularly i f it is investigated electrochemically. \u00E2\u0080\u00A2 Ammine stability constants. These should be re-measured as discussed in section 9.2. \u00E2\u0080\u00A2 Critical concentration. The critical concentration of ammonium sulphate which leads to severe corrosion at a given total ammonia concentration should be investigated to confirm the predictions with regard to N i 2 + / N i O stability and solution acidity at high temperatures. In this regard, the effect of excess ammonia addition could be an effective way of mitigating corrosion by maintaining a constant concentration of ammonium ions to \"tie up\" anions such as sulphate or chloride. 190 Phase equilibria. Ternary phase equilibria for the H 2 O - O 2 - N H 3 system should be calculated to confirm N H 3 solubility. This would also shed light on the failure mechanism. SCWO experiments. The experiments conducted here at S C W O conditions are certainly not exhaustive and more work should be performed in this area. Electrochemical analysis would be preferable. Niobium. Al loy ing Nb with N i at levels higher than 5 % should be investigated. More data for the electrochemical response of Nb is required, particularly at high temperatures. 191 REFERENCES Archer, D.G. , and Wang, P., 1990. The dielectric constant of water and Debye-Huckel limiting law slopes. J. Phys. Chem. Ref. Data 19, 371-411. Asselin, E . , Alfantazi, A., Vera-Perez, I., Rogak, S., and Branion R., 2003. Case Study of the Failure of Supercritical Water Reactor Tubing during the treatment of 2,4 D N P with Ammonium Sulphate, Proceedings of the International Symposium on Environmental Degradation of Materials and Corrosion Control in Metals, 2nd, Vancouver, B C , Canada, Aug. 24-27, 307-319. Asselin, E . , Alfantazi, A., Rogak, S., 2005. Polarization Study of A l l o y 625, N i , Cr and M o in Ammoniated Sulphate Solutions, Corrosion 61, 579-586. Babic, R., and Metikos-Hukovic, M . , 1993. Oxygen reduction on stainless steel. Journal of Applied Electrochemistry 23, 352. Bard, A.J . , Parsons, R., Jordan, J . , 1985. Standard Chemical Potentials in Aqueous Solution. Marcel Dekker, New York, N Y , U S A . 527-531. Bates, R.G. , Staples, B.R., and Robinson, R.A., 1970. Ionic hydration and single ion activities in unassociated chlorides at high ionic strengths. Analytical Chemistry 42, 867-871. Beas, C.F. , and Mesmer, R.E. , 1976. The Hydrolysis of Cations. John Wiley, New York, N Y , U S A . Betova, I., Bojinov, M . , et al., 2002. Mechanism of transpassive dissolution of nickel-based alloys studied by impedance spectroscopy and rotating ring-disc voltammetry. Electrochimica Acta 47, 2093-2107. Beverskog, B. and Puigdomenech, I., 1997. Revised Pourbaix diagrams for nickel at 25-300\u00C2\u00B0C. Corrosion Science 39, 969-980. Birley, S.S. and Tromans, D., 1971. Ammoniacal stress corrosion cracking of alpha brass. Corrosion 27, 297-311. Bishop, C.R., 1963. Corrosion Tests at Elevated Temperatures arid Pressures. Corrosion 19, 308t-314t. Bjerrum, J . , 1957. Metal ammine formation in aqueous solution, P. Haase and Son, Copenhagen. Bockris, J . O ' M . , and Reddy, A .K. , 1970. Modern Electrochemistry V o l . 2. Plenum Press, Newy York, N Y , U S A . 1039. 192 Bockris, J . O ' M . , Genshaw, M . A . , Brusic, V. and Wroblowa, H . , 1971. Mechanism of the passivation of iron in neutral solutions. Ellipsometric and coulometric investigation. Electrochimica Acta 16, 1859. Bojinov, M . , et al., 2000. The Mechanism of Transpassive Dissolution of N i - C r alloys in Sulphate Solutions. Electrochimica Acta 45, 2791-2802. Bojinov, M . , et al., 2002. Transpassive dissolution of N i - C r alloys in sulphate solutions - comparison between a model alloy and two industrial alloys. Electrochimica Acta 47, 1697-1712. Bojinov, M . and Tzvetkoff, T., 2003. The influence of solution anion on the mechanism of transpassive dissolution of ferrous- and nickel-based alloys. Journal of Physical Chemistry B 107, 5101-5112. Boukis, N . , et al., 1997. Corrosion screening tests in supercritical water containing hydrochloric acid and oxygen. E U R O C O R R ' 9 7 , Euro. Fed. Corrosion, 1, 617-622. Brook, P .A., 1972. Potential-pH diagrams at elevated temperatures. Corrosion Science 12, 297-306. Brooks, A .R. , et al., 1986. On the role of Cr in the passivity of stainless steel.. Journal of the Electrochemical Society 133, 2459-2464. Brubaker, S.K., 1987. Corrosion by sulphuric acid. In: A S M Corrosion Handbook (Vol . 13), Davis, J.R., and Destefani, J.D., (Eds.), A S M International. 1148-1160. Bulhoes, L.O.S and Joanni, E . , 1984. Characteristics of corrosion of niobium and anodized niobium in sulfuric acid solution. In: Proceedings of the International Congress on Metallic Corrosion, Toronto, O N , Canada, V o l . 3, N R C , 437-439. Burns, R.H. , Shuker, F.S., and Manning, P.E., 1984. Industrial applications of corrosion Resistant tantalum, niobium and their alloys. In: R . E . Smallwood (Ed.), A S T M Special Technical Publication 849, A S T M . 50-69. Caruso, L . , and Michels, H.T. , 1981. The resistance of copper-nickel alloys to ammonia corrosion in simulated steam condenser environments. Materials Performance, 35-39. Chao, C.Y. , Lin, L . F . , and Macdonald, D . D . , 1982. A point defect model for anodic passive films, III. Impedance response. Journal of the Electrochemical Society 129, 1874-1879. Clayton, C.R., L u , Y.C.,1986. A bipolar model of the passivity of stainless steel: The role of M o addition. Journal of the Electrochemical Society 133, 2465-2473. 193 Corbett R.A., and Morrison, W.S., 1989. Comparative Corrosion Resistance of Some High-Nickel , Chromium-Molybdenum Al loys , Materials Performance, February 1989, 56-59. Costa, R., Weyandt, N.C., Defreyne, J . , Banker, J.R., Stoiber, R., and McMaster J.A., 2005. Titanium clad ignition studies in simulated pressure oxidation environments. In: Proceedings of the Corrosion Solutions Conference 2005. Covino, B.S., Carter, J.S., and Cramer, S.D., 1980. The corrosion behavior of niobium in hydrochloric acid solutions. Corrosion 36, 554-558. Cowan, R.L. , and Staehle, R.W., 1971. The thermodynamics and electrode kinetic behaviour of nickel in acid solution in the temperature range 25\u00C2\u00B0 to 300\u00C2\u00B0C. Journal of the Electrochemical Society 118, 557-567. Criss, C M . , Cobble, J.W., 1964A. The thermodynamic properties of high temperature aqueous solutions. IV. Entropies of the ions up to 200\u00C2\u00B0 and the correspondence principle. Journal of the American Chemical Society 86, 5385-5390. Criss, C M . , Cobble, J.W., 1964B. The thermodynamic properties of high temperature aqueous solutions. V . The calculation of ionic heat capacities up to 200\u00C2\u00B0. Entropies and heat capacities above 200\u00C2\u00B0C. Journal of the American Chemical Society 86, 5390-5393. Crolet, J .L . , 2003. L a corrosion face a son histoire: role omnipresent et insaissisable du temps sur les concepts et les besoins d'hier et d'aujourd'hui. Materiaux et Techniques 1-2, 3-8. Davydov, A., Rybalka, K . V . , Beketaeva, G.R., et al., 2005. Corrosion Science 47, 195. Degnan, T.F. , 1987. Corrosion by hydrogen chloride and hydrochloric acid In: A S M Corrosion Handbook (Vol . 13), Davis, J.R., and Destefani, J.D., (Eds.), A S M International. 1160-1166. Dell'Orco, P . C , Gloyna, E.F. and Buelow, S.J., 1997. Reactions of nitrate salts with ammonia in supercritical water\", Ind. Eng. Chem. Res. 36, 2547-2557. Ding, Z.Y. , L i , L . , Wade, D. and Gloyna, E.F. 1998. Supercritical water oxidation of N H 3 over a M n 0 2 / C e 0 2 catalyst. Ind. Eng. Chem. Res. 37, 1707-1716. Dinjus, E . and Kruse, A., 2004. Hot compressed water-a suitable and sustainable solvent reaction medium? Journal of Physics: Condensed Matter, 16, SI 161-SI 169. Downey, K.W., Snow, R.H. , Hazlebeck, D.A. and Roberts, A .J . , 1995. Corrosion and chemical agent destruction: research on supercritical water oxidation of hazardous military wastes. In: Hutcheson, K . W . , Foster, N . R . (Eds.), A C S Symposium Series 608, A C S , Washington, D C , 313-326. 194 Faraday, M . , 1836. On a peculiar voltaic condition of iron, by Professor Schoenbein. In: Experimental Researches in Electricity, V o l . 2. Dover Publications, New York, N Y , U S A . Reprinted 1965. 234-250. Fisher, I. and Halpern, J . , 1956. Corrosion of copper-gold alloys by oxygen-containing solutions of ammonia and ammonium salts. Journal of the Electrochemical Society 103, 282-286. Fodi, S., et al., 1997. Corrosion of high temperature alloys in a supercritical water oxidation process. EuroCorr, Euro. Fed. Corrosion, 1, 629-633. Forward, F . A . , 1951. Extracting of nickel values from nickeliferous sulfide material, U S Patent No.2576314. Forward, F . A . , 1953. Ammonia pressure leach process for the recovery of nickel, copper, and cobalt from Sherritt Gordon nickel sulphide concentrate. Transactions - the Canadian Institute of Min ing and Metallurgy and the Min ing Society of Nova Scotia 56, 645-648. Forward, F . A . , and Mackiw, V . N . , 1955. Chemistry of the ammonia pressure process for leaching N i , Cu, and Co from Sherritt Gordon sulphide concentrates. Transaction AEVIE, 457-463. Gilli, G. , Borea, P., Zucchi, F., et al., 1969. Corrosion Science 9, 673. Glasstone, S., 1942. A n introduction to electrochemistry. Van Nostrand, New York, 491. Gonzalez, V . , et al., Corrosion of nickel-base alloys under the conditions of supercritical water oxidation (SCWO). Goto, M . , Shriramizu, D., Kodama, A . and Hirose, T., 1999. Kinetic analysis for ammonia decomposition in supercritical water oxidation of sewage sludge. Ind. Eng. Chem. Res. 38, 4500-4503. Greeley, R.S., Smith, W.T. , Lietzke, M . H . , and Stoughton, R.W., 1960. Electromotive force series studies in aqueous solutions at elevated temperatures. I. The standard potential of the silver-silver chloride electrode. Journal of Physical Chemistry 64, 652-657. Hakiki, N . E . , Boudin, S., Rondot, B., and Da Cunha Belo, M . , 1995. The electronic structure of passive films formed on stainless steels. Corrosion Science 37, 1809-1822. Hakiki, N . E . , Da Cunha Belo, M . , Simoes, A . M . P . , and Ferreira, M.G.S., 1998. Semiconducting properties of passive films formed on stainless steels. Influence of alloying elements. Journal of the Electrochemical Society 145, 3821-3829. 195 Halpern, J . , 1953. Kinetics of the dissolution of copper in aqueous ammonia. Journal of the electrochemical society 100, 421-428. Halpern, J . , Milants, H . and Wiles, D.R., 1959. Kinetics of the dissolution of copper in oxygen-containing solutions of various chelating agents. Journal of the Electrochemical Society 106, 647-650. Hara N., Tanaka, S., Soma, S., Sugimoto, K. 2002. Corrosion resistance of Fe-Cr and N i - C r alloys in oxidizing supercritical H C l solution. Paper 02358 Corrosion 2002. N A C E International. Helgeson, H.C. , 1967. The thermodynamics of complex dissociation in aqueous solution at elevated temperatures. The Journal of Physical Chemistry 71,3121-3136. Helgreson, H.C. , Kirkham, D.H., and Flowers, G.C. , 1981. Theoretical prediction of the thermodynamic behaviour of aqueous electrolytes: IV. Calculation of activity coefficients, osmotic coefficients, and apparent molal and standard and relative partial molal properties to 600\u00C2\u00B0C and 5 kb. American Journal of Science 281, 1249-1516. Helling, R.K. and Tester, J.W., 1988. Oxidation of simple compounds and mixtures in supercritical water: carbon monoxide, ammonia, and ethanol. Environ. Sci . Technol. 22, 1319-1324. Himmelblau, D.M. , 1960. Solubilities of inert gases in H2O: 0\u00C2\u00B0 to near the critical point o f H 2 0 . Journal of Chemical and Engineering Data 5, 10-15. Isaev, I.D., Tverdokhlebov, S .V., Leont'ev, V . M . , et al., 1990. The influence of temperature and ionic strength on the stability constants of copper(II) ammines in aqueous solutions. Russian Journal of Inorganic Chemistry 35, 1159-1161. Jabs, T., Borthen, P., and Strehblow, H.H. , 1997. X - R a y Photoelectron Spectroscopic Examinations of Electrochemically Formed Passive Layers on N i - C r Al loys . Journal of the Electrochemical Society 144, 1231-1243. Japas, M . L . , and Franck, E.U. , 1985. High pressure phase equilibria and PVT-data of the water-oxygen system including water-air to 673 K and 250MPa. Berichte der Bunsen-Gesellschaft 89, 1268-1275. Kawazuishi, K. and Prausnitz, J . M . , 1987. Correlation of vapour-liquid equilibria for the system ammonia-carbon dioxide-water. Ind. Eng. Chem. Res. 26, 1482-1485. Keir, J . , 1790. Experiments and observations on the dissolution of metals in acids, and their precipitations; with an account of a new compound acid menstruum, useful in some technical operations of parting metals. Philosophical Transactions of the Royal Society of London 80, 359-384. 196 Kelley, K . K . , 1934. Theoretical metallurgy . II. High-temperature specific-heat equations for inorganic substances. U S Bureau of Mines Bulletin 371. Kelly, R.G. , Scully, J.R., Shoesmith, D.W. and Bucheit, R.G. , 2003. Electrochemical techniques in corrosion science and engineering. Marcel Dekker, New York. Kofstad, P., and Lillerud, K.P., 1980. On high temperature oxidation of chromium. II. Properties of chromium trioxide and the oxidation mechanism of chromium. Journal of the Electrochemical Society 127, 2410-2419. Konys, J . , Fodi, S., Hausselt, J . , et al., 1999. Corrosion of high-temperature alloys in chloride-containing supercritical water oxidation systems. Corrosion 55. 45-51. Kriksunov, L .B . and Macdonald, D.D., 1993. Development of Pourbaix diagrams for metals in supercritical aqueous media. Workshop on Corrosion in Supercritical Water Oxidation Systems, M.I .T. , Session XII : Applied R & D . Kriksunov, L .B . and Macdonald, D.D., 1995. Corrosion in supercritical water oxidation systems: a phenomenological analysis. Journal of the Electrochemical Society 142, 4069-4073. Kriksunov, L .B . and Macdonald, D.D., 1997. Potential-pH diagrams for iron in supercritical water. Corrosion 53, 605-611. Kritzer, P., et al., 1998. Corrosion of alloy 625 in high-temperature, high pressure sulphate solutions. Corrosion, 54. Kritzer, P., Boukis, N., Dinjus, E . , 1999A. Factors controlling corrosion in high-temperature aqueous solutions: a contribution to the dissociation and solubility data influencing corrosion processes. Journal of Supercritical Fluids 15, 205-227. Kritzer, P., Schacht, M . , Dinjus, E . , 1999B. The corrosion behaviour of nickel-base alloy 625 (NiCr22Mo9Nb; 2.4856) and ceria stabilized tetragonal zirconia polycrystal (Ce-TZP) against oxidizing aqueous solutions of hydrofluoric acid (HF), hydrobromic acid (HBr), and hydriodic acid (HI) at sub- and supercritical temperatures. Materials and Corrosion 50. Kritzer, P., Boukis, N., Franz, G. and Dinjus, E . , 1999C. The corrosion of niobium in oxidizing sub- and supercritical aqueous solutions of H C I and H2SO4. Journal of Materials Science Letters 18, 25-27. Kritzer, P., Boukis, N., Dinjus, E . , 2000A. Review of the corrosion of nickel-based alloys and stainless steels in strongly oxidizing pressurized high-temperature solutions at subcritical and supercritical temperatures. Corrosion 56, 1093-1104. 197 Kritzer, P., Boukis, N., Dinjus, E . , 2000B. Transpassive dissolution of alloy 625, chromium, nickel, and molybdenum in high temperature solutions containing hydrochloric acid and oxygen. Corrosion 56, 3. Kritzer, P., Dinjus, E . , 2001. A n assessment of supercritical water oxidation (SCWO) existing problems, possible solutions and new reactor concepts. Chemical Engineering Journal 83, 207-214. Lance, D., 1904. French Patent 342865 (1904); see Moniteur Scientifique 64-65 Patent Section 53 (1906). Latimer, W . M . , 1938. Oxidation states of the elements and their potentials in aqueous solutions. Prentice Hal l , New York, N Y , U S A . 328. Laycock, N.J., Newman, R.C. , Stewart, J . , 1995. Corrosion Science 37, 1637. LeBozec, N., Compere, M . , L'Her, M . , et al., 2001. Influence of stainless steel surface treatment on the oxygen reduction reaction in seawater. Corrosion Science 43, 765-786. Lee, J.B., 1981. Elevated temperature potential-pH diagrams for the C r - H 2 0 , T i - H 2 0 , M o - H 2 0 , and P t -H20 systems. Corrosion 37, 467-481. Letowski, F., 1970. Etudes thermodynamiques des solutions aqueuses des ammoniacates de metaux aux temperatures elevees. II. Constantes de stabilite des ammoniacates de nickel en temperatures jusqu'a 180\u00C2\u00B0C. Roczniki Chemii 44, 1665-1673. Lewis, D., 1970. Studies of redox equilibria at elevated temperatures. I. The estimation of equilibrium constants and standard potentials for aqueous systems up to 374\u00C2\u00B0C. Ark iv for K e m i 32, 385-404. Linke, W.F. and Seidell, A . , 1965. Solubilities of inorganic and metal organic compounds. American Chemical Society, Washington, D C , V o l . II. P. 1219. Liu, C , Macdonald, D.D., Medina, E . , Villa, J . and Bueno, J . , 1994. Probing corrosion activity in high subcritical and supercritical water through electrochemical noise analysis. Corrosion 50, 687-694. Liu, H . , Papangelakis, V . G . , 2005. Chemical modeling of high temperature aqueous processes. Hydrometallurgy, 79, 48-61. Littman, F .E . , and Church, F . M . , 1961. A study of metal ignitions, the spontaneous ignition of titanium. The Journal of Less Common Metals 3, 367-378 Lloyd, A . C . , Shoesmith, D.W., Mclntyre, N.S., et al., 2003. Effects of temperature and potential on the passive corrosion properties of alloys C22 and C276. Journal of the Electrochemical Society 150, J3120-B130. 198 Lupton, D., Aldinger, F., and Schulze, K., 1981. Niobium in corrosive environments. In: Proceedings of the International Symposium Niobium 1981. H . Stuart (Ed.). San Francisco. 533-560. Lupton, D., Schiffmann, W., Schreiber, F., and Heitz, E . , 1981. Corrosion behaviour of tantalum and possible substitute materials under extreme conditions. In: Proceedings of the International Congress on Metallic Corrosion, V o l . 2, Frankfurt-am-Main, Germany: Dechema. 1441. Lvov, S.N. and Macdonald, D.D., 1996. Estimation of the thermal liquid junction potential of an external pressure balanced reference electrode. Journal of Electroanalytical Chemistry 403, 25-30. Lvov, S.N., Gao, H . , Kouznetsov, D., et al., 1998. Potentiometric p H measurements in high subcritical and supercritical aqueous solutions. Fluid Phase Equilibria 150-151, 515-523. Macak, J . , Sajdl, P., Kucera, P. et al., 2006. In situ electrochemical impedance and noise measurements of corroding stainless steel in high temperature water. Electrochimica Acta 51, 3566-3577. Machet, A. , et al., 2002. X P S study of oxides formed on nickel-base alloys in high temperature and high-pressure water. Surface and Interface Analysis 34, 197-200. Macdonald, D.D. and Syrett, B.C. , 1979A. Potential-pH diagrams for iron and nickel in high salinity geothermal brine containing low concentrations of hydrogen sulphide. Corrosion 35, 471-475. Macdonald, D.D., Scott, A . C . , and Wentreck, P.J., 1979B. Silver-silver chloride thermocells and thermal liquid junction potentials for potassium chloride solutions at elevated temperatures. Journal of the Electorchemical Society 126, 1618-1624. Macdonald, D.D. and Smedley, S.I., 1990A. A n electrochemical impedance analysis of passive films on nickel(l 11) in phosphate buffer solutions. Electrochimica Acta 35. 1949-1956. Macdonald, D.D. and Smedley, S.I., 1990B. Characterization of vacancy transport in passive films using low frequency electrochemical impedance spectroscopy. Corrosion Science 31, 667-672. Macdonald, D.D., Urquidi-Macdonald, M . , 1990C. Theory of steady-state passive films. Journal of the Electrochemical Society 137, 2395-2402. Macdonald, D.D., Biaggio, S.R., and Song, H . , 1992A. Steady-state passive films: interfacial kinetic effects and diagnostic criteria. Journal of the Electrochemical Society 139, 170-177. 199 Macdonald, D.D., 1992B. The point defect model for the passive state. Journal of the Electrochemical Society 139, 3434-3449. Macdonald, D.D. and Kriksunov, L.B. , 1993. Phenomenological analysis of corrosion phenomena in supercritical water oxidation systems. Workshop on Corrosion in Supercritical Water Oxidation Systems, M.I .T. , Session XII : Applied R & D . Macdonald, D.D., 1999. Passivity - the key to our metals-based civilization. Pure App l . Chem. 71, 951-978. Macdonald, D.D. and Kriksunov, L .B. , 2001. Probing the chemical and electrochemical properties of S C W O systems. Electrochimica Acta 47, 775-790. Macdonald, D.D., 2004A. Effect of pressure on the rate of corrosion of metals in high subcritical and supercritical aqueous systems. The Journal of Supercritical Fluids 30, 375-382. Macdonald, D.D., Sun, A . , Priyantha, N. and Jayaweera, P., 2004B. A n electrochemical impedance study of alloy-22 in N a C l brine at elevated temperature: II. Reaction mechanism analysis. Journal of Electroanalytical Chemistry 572, 421-431. Macdonald, D.D. and McKubre, M . C . H . , 2005. Corrosion of materials. In: Impedance Spectroscopy, Theory, Experiment, and Applications. Barsoukov, E . , and Macdonald J.R., (Eds.). Wiley & Sons, Hoboken, N J , U S A . 400. Macleary, D.L. , 1962. Testing of columbium and columbium alloys. Corrosion 18, 67t-69t. Marsh, A .R .W. , and McElroy, W.J. , 1985. The dissociation constant and Henry's Law constant of H C l in aqueous solution. Atmospheric Environment 19, 1075-1080. Marshall, W.L. , and Franck, E.U.,1981. Ion product of water substance, 0-1000\u00C2\u00B0C, 1-10,000 bars. New international formulation and its background. J. Phys. Chem. Ref. Data 10, 295-304. Maurice, V. , Yang, W.P., and Marcus, P. 1998. X-ray photoelectron spectroscopy and scanning tunnelling microscopy study of passive films formed on (100) Fe-18Cr-13Ni single-crystal surfaces. Journal of the Electrochemical Society 145, 909-920. Melendres, C A . and Tani, B.S., 1986. Journal of the Electrochemical Society 133, 1059. Mitton, D.B., et al., 1996. Interpretation of the rationale for feed modification in S C W O systems. Journal of the Electrochemical Society 143, L59-L61. 200 Mitton, D.B., et al., 2000. Corrosion behavior of nickel-based alloys in supercritical water oxidation systems. Ind. Eng. Chem. Res. 39. Mitton, D.B., et al., 2001. A n overview of the current understanding of corrosion in S C W O systems for the destruction of hazardous waste product. Materials Technology, 16,1. Mitton, D.B., et al., 2002. A n examination of the corrosion phenomena of potential constructional materials for S C W O system fabrication. Paper No . 02353 N A C E Corrosion. Modell, M . , 1989. Super critical water oxidation. Chp 8.12 in Standard Handbook of Hazardous Waste Treatment and Disposal, Ed. Freeman, H . M . , McGraw H i l l , New York. Modell, M . , Kuharich, E.F. and Rooney, M.R., 1993. Supercritical water oxidation process of organics with inorganics. U S Patent 5252224. Muendel, C .H. , Linford, H.B., and Selke, W.A., 1961. Vapor-liquid equilibrium in ammonia complex systems. A. I .Ch.E. Journal 7, 133-137. Muir, D .M. , and Senanayake, G. , 1985. Principles and applications of strong salt solutions to mineral chemistry. Extraction Metallurgy'85. Institution of Min ing and Metallurgy, London. 65-91. Newman, J . , 1973. Electrochemical Systems. Prentice-Hall, Englewood Cliffs, N . J , U S A . Osseo-Asare, K. , 1981. Application of activity-activity diagrams to ammonia hydrometallurgy II. - The copper-, nickel-, cobalt-ammonia-water systems at elevated temperatures. In: Process and Fundemental Considerations of Selected Hydrometallurgical Systems. Kuhn, M . C . , (Ed.). Society of Min ing Engineers of the American Institute of Mining , Metallurgical and Petroleum Engineers [ A I M E ] . New York, N Y . Osseo-Asare, K. , and Asihene, S.W., 2004. Ammonia pressure hydrometallurgy: estimation of high temperature stability constants for metal ammine complexes. In: Pressure Hydrometallurgy 2004. Collins, M . J . , and Papangelakis, V . G . (Eds.). Perry, H.R., Gree, D.W. and Maloney, J.O. 1997. Perry's Chemical Engineer's Handbook, 6 t h Edition. McGraw H i l l , New York, N Y , U S A . 3-71, 3-72, 2-10, 2-12. Plyasunov, A.V. , O'Connell, J.P., Wood, R.H. , and Shock, E . L . , 2000. Infinite partial molar properties of aqueous solutions of nonelectrolytes. II. Equations for the standard thermodynamic functions of hydration of volatile nonelectrolytes over wide ranges of conditions including subcritical temperature. Geochimica et Cosmochimica Acta 64, 2779-2795. 201 Plyasunov, A.V. , and Shock, E . L . , 2001. Correlation strategy for determining the parameters of the revised Helgeson-Kirkham-Flowers model for aqueous nonelectrolytes. Geochimica et Cosmochimica Acta 65, 3879-3900. Poirier, D.R., and Geiger, G.H. , 1994. Transport Phenomena in Materials Processing. T M S , Warrendale Pennsylvania. P. 457. Pourbaix, M . , 1966. Atlas of Electrochemical Equilibria in Aqueous Solutions. Pergamon Press, London. Pourbaix, M . , 1973. Lectures on Electrochemical Corrosion. Plenum Press, New York. P. 334 Priyantha, N., Jayaweera, P., Macdonald, D.D. and Sun, A., 2004. A n electrochemical study of alloy 22 in N a C l brine at elevated temperature. I. Corrosion behavior. Journal of Electroanalytical Chemistry 572, 409-419. Raja, R.S., and Jones, D.A., 2006. Effects of dissolved oxygen on passive behavior of stainless alloys. Corrosion Science 48, 1623-1638. Rajagopalan, K.S., Raghavan, M . , Rengaswamy, N.S., Balasubramanian, T . M . and Muralidharan, V.S., 1981. Corrosion studies on certain copper based alloys in synthetic sea water containing traces of ammonia. Materials Performance, January, 19-27. Reid, H.R., Green, D.W., and Poling, B.E. , 1987. The properties of gases and liquids, 4 t h Edition. McGraw H i l l , New York, N Y , U S A . 207, 256, 350. Revie, R.W., Edt., 2000. Uhl ig 's Corrosion Handbook. John Wiley and Sons, New York, N Y , 831-851. Rice, D.W., Phipps, P.B.P. and Tremoureux R., 1980. Atmospheric Corrosion of Nickel . Journal of the Electrochemical Society 127, 563-568. Robin, A., de Almeida, M . E . , and Nunes, C.A., 1991. Corrosion behavior of niobium and Nb-25wt% Ta alloy in sulfuric acid solutions. Corrosion 47, 443-448. Robin, A., 1997. Corrosion behavior of niobium, tantalum and their alloys in boiling sulphuric acid solutions. International Journal of Refractory Metals & Hard Materials 15,317-323. Robin, A., Rosa, J .L . , 2000. Corrosion behavior of niobium, tantalum and their alloys in hot hydrochloric and phosphoric acid solutions. International Journal of Refractory Metals & Hard Materials 18, 13-21. Robin, A., 2004A. Corrosion behaviour of niobium in sodium hydroxide solutions. Journal of Applied Electrochemistry 34, 623-629. 202 Robin, A., 2004B. Comparative study of Nb , N b - l O W , and Nb-16Ta-12W corrosion behavior in sodium hydroxide solutions. Electrochimica Acta 49, 1915-1923. Robinson, R.A., and Stokes, R.H. , 1959. Electrolyte Solutions. Butterworths, London, U K . Rodriguez-Torres, I., Valentin, G. and Lapicque, F., 1999. Electrodeposition of zinc-nickel alloys from ammonia-containing baths. Journal of Applied Electrochemistry 29, 1035-1044. Roine, A., 2002. Outokumpu H S C Chemistry Thermochemical Database, Version 5.0. Outokumpu Research Oy, Finland. Sato, N., and Kudo, K. , 1974. Ellipsometric study of anodic passivation of nickel in borate buffer solution. Electrochimica Acta 19, 461-470. Sato, N., and Noda, T., 1977. Ion migration in anodic barrier oxide films on iron in acidic phosphate solutions. Electrochimica Acta 22, 839-843. Sakashita, M . , Sato, N., 1977. The effect of molybdate anion on the ion-selectivity of hydrous ferric oxide films in chloride solutions. Corrosion Science 17, 473-486. Schmieder, H . , Abeln, J . , 1999. Supercritical water oxidation: state of the art. Chemical Engineering and Technology 22, 903-908. Schroer, C , Konys, J . , Novotny, J . , and Hausselt, J . , 2002. Corrosion investigations on binary and ternary N i - C r alloys in S C W O systems. Corrosion 2002, N A C E , Paper 02359. Segond, N., Matsumura, Y. and Yamamoto, K., 2002. Determination of ammonia oxidation rate in sub- and supercritical water. Ind. Eng. Chem. Res. 41, 6020-6027. Shimakage, K., and Morioka, S., 1971. Fundamental studies on the ammonia pressure leaching of metallic nickel powder. Transactions of the Institution of Min ing and Metallurgy, Section C: Mineral Processing and Extractive Metallurgy 80(Dec), 228-234. Shock, E . L . , and Helgeson, H.C. , 1988. Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures: correlation algorithms for ionic species and equation of state predictions to 5 kb and 1000\u00C2\u00B0C. Geochimica et Cosmochimica Acta 52, 2009-2036. Shock, E . L . , Sassani, D.C., Willis, M . , and Sverjensky, D.A., 1997. Inorganic species in geologic fluids: correlations among standard molal thermodynamic properties of aqueous ions and hydroxide complexes. Geochimica et Cosmochimica Acta 61, 907-950. 203 Shreir, L . L . , Edt., 1977. Corrosion, Newnes-Butterworth, Toronto. Sikora, E . , and Macdonald, D.D., 2002. Nature of the passive film on nickel. Electrochimica Acta 48, 69-77. Smart, N.G., and Bockris, J . , O ' M . , 1992. Effect of water activity on corrosion. Corrosion 48, 277- 280. Smith, R . M . , and Martell, A . E . , Critical Stability Constants, V o l . 4, Inorganic Complexes. Plenum, New York, N Y , U S A . 40-42. Soave, G. , 1984. Improvement of the Van Der Waals equation of state. Chemical Engineering Science 39, 357. Son M . , Kurata, Y . and Ikushima, Y . , 2002. Corrosion behaviour of metals in S C W environments containing salts and oxygen. Paper 02357 Corrosion 2002. N A C E International. Soto, L . , Franey, J.P., Graedel, T .E . , and Kammlott, G.W., 1983. On the corrosion resistance of certain ancient Chinese bronze artifacts. Corrosion Science 23, 241-250. Speller, F.N., 1951. Corrosion: Causes and prevention. M c G r a w - H i l l , New York, N Y , U S A . 12, 168. Stansbury, E . E . , and Buchanan, R .A . , 2000. Fundamentals of Electrochemical Corrosion. A S M International. Materials Park, O H , U S A . 118-119. Stellwag, B., 1998. The mechanism of oxide film formation on austenitic stainless steels in high temperature water. Corrosion Science 40, 337-370. Strehblow, H.H. , 1995. Mechanisms of pitting corrosion. Chp. 7 in Corrosion Mechanisms in Theory and Practice, Eds. Marcus, P., Oudar, J., Marcel Dekker, New York, 201-237. Sunseri, C , Piazza, S., and Di Quarto, F., 1990. Photocurrent spectroscopic investigations of passive films on chromium. Journal of the Electrochemical Society 137, 2411-2417. Tanger, J .C. , and Helgeson, H.C. , 1988. Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures: revised equations of state for the standard partial molal properties of ions and electrolytes. American Journal o f Science 288, 19-98. Taylor, D.F., 1978. Thermodynamic properties of metal-water systems at elevated temperatures. Journal of the Electrochemical Society 125, 808-812. 204 Tomashov, N . D . , 1966. Theory of Corrosion and Protection of Metals. Macmil lan company, New York, N Y , U S A . Tremaine, P.R., and LeBlanc, J . C , 1980. The solubility of nickel oxide and hydrolysis of N i 2 + in water to 573 K . J. Chem. Thermodynamics 12, 521-538. Tromans, D . , and Sun, R.H. , 1992. Anodic behaviour of copper in weakly alkaline solutions. Journal of the Electrochemical Society 139, 1945-1951. Tromans, D . , 1998A. Temperature and pressure dependent solubility of oxygen in water: a thermodynamic analysis. Hydrometallurgy 48, 327-342. Tromans, D. , and Ahmed, T., 1998B. Active/passive behaviour of copper in strong sulphuric acid. Journal of the Electrochemical Society 142, 601-608. Tromans D. , 1999. Modeling oxygen-controlled corrosion of steels in hot water. Corrosion 55, 942-947. Tromans, D. , 2000A. Oxygen solubility modelling in ammoniacal leaching solutions: leaching of sulphide concentrates. Mineral Engineering 13, 497-515. Tromans, D . , 2000B. Modeling oxygen solubility in water and electrolyte solutions. Ind. Eng. Chem. Res. 39, 805-812. Uhlig, H .H. , 1978. Passivity of metals: history of passivity, experimenst and theories. In: Passivity o f Metals: Proceedings of the fourth International Symposium on Passivity. Kruger, J., (Ed.). The corrosion monograph series, Electochemical Society. 1-25. Vaughan, J . , Reid, P., Alfantazi, A., et al., 2003. Corrosion of titanium and ti-Alloys at high temperatures and pressures. Environmental Degradation of Materials and Corrosion Control in Metals, Proceedings of the International Symposium on Environmental Degradation of Materials and Corrosion Control in Metals, 2nd, Vancouver, B C , Canada, Aug . 24-27. Vehlow, J . , 1981. Corrosion of tantalum in sulfuric acid between 150 and 270\u00C2\u00B0C with nitric acid and hydrochloric acid metered in. In: Proceedings of the International Congress on Metallic Corrosion, V o l . 2. Frankfurt-am-Main, Germany. Dechema. 1436. Vera-Perez, I., 2002. Supercritical water oxidation of phenol and 2,4 dinitrophenol. Masters Thesis. The University of British Columbia. Vetter, K .J . , 1967. Electrochemical Kinetics. Academic Press, New York, N Y , U S A . 537-539. 205 Wallinder, D., Pan, J . , Leygraf, C , and Delblanc-Bauer, A . , 1999. EIS and X P S study of the surface modification of 316 L V M stainless steel after passivation. Corrosion science 41. 275-289. Wagman, D.D., Evans, W.H. , Parker, V.B. , et al., 1982. N B S Tables. J. Phys. Chem. Ref. Data 11, Suppl. 2, 2-64 to 2-168. Watanabe, Y. , et al., 2002. Effects of oxygen concentration on corrosion behavior of alloys in acidic supercritical water. Corrosion 2002, N A C E , Paper No. 2355. Webley, P .A. , Tester, J.W. and Holgate, H.R., 1991. Oxidation kinetics of ammonia and ammonia-methanol mixtures in supercritical water in the temperature range 530-700\u00C2\u00B0C at 246 bar. Ind. Eng. Chem. Res. 30, 1745-1754. Wilson, P., Laycock, N. J . , Ryan, M . , et al., 2000. Transpassive Corrosion of Stainless Steels and Nickel-Base Al loys . Paper 047, Corrosion & Prevention 2000, Auckland, New Zealand. Yau, T . L . and Webster, R.T., 1987. Corrosion of niobium and niobium alloys. In: A S M Corrosion Handbook (Vol . 13), Davis, J.R., and Destefani, J.D., (Eds.), A S M International. 722-724. 206 APPENDICES Appendix A - High temperature electrochemical apparatus C6M P R E S I O N N U T 3/B* C A R R I E R S W A G E L \u00C2\u00A9 K F I T T I N G P A R R I N C . P A R A L L E L T H R E A D C O M P R E S S I O N N L I T P A R R I N C . M E T A L G A S K E T Figure A-1 Working electrode 207 G \u00E2\u0080\u0094 B H Figure A - 2 Working electrode sample detail 208 fCALE 1 :$ to o VO Figure A-3 Autoclave and working electrode assembly. SECTION A-A DETAIL B DETAIL C SCALE 1:4 SCALE 3 :5 SCALE 1. : 1 tO \u00E2\u0080\u0094 - \u00E2\u0080\u0094 \u00E2\u0080\u0094\u00E2\u0080\u0094\u00E2\u0080\u00A2 o Figure A-4 Autoclave head and detail o f electrodes. Figure A-5 Autoclave Assembly 2 1 1 Figure A -6 Working electrode sample showing relative size (Staedler felt tip pen) and internal thread (small hole). Figure A-7 Working electrode showing used P T F E seal, connecting rod and sample. Ag wire Ag wire chloridized at this end Vycor frit Saturated KCI fill solution in contact with chloridized Ag wire Bridge tube to autoclave containing fiberglass wick Figure A-8 Ex te rna l pressure balanced w o r k i n g electrode assembly. 213 Figure A-9 Assembled autoclave at work. 214 Appendix B - Concentrations and diffusivities of 02 in the autoclave. p barg T K D o 2 cm2_s~' aNH2 a H 2 0 p barabs P 1 H20 bar\u00E2\u0080\u009Ebs P 1 TOTAL.T barabs P o 2 bar,lhs C o 2 mol. kgH20-' V gas cm3 298 3.38xl0\"5 0.086 0.999 0.02 0.03 1.01 0.96 0.0012 699.98 323 5.95 xlO' 5 0.102 0.999 0.09 0.12 1.27 1.06 0.001 690.04 348 9.25 x lO - 5 0.119 0.999 0.18 0.38 1.72 1.16 0.0009 675.18 0 373 1.33xl0\"4 0.135 0.999 0.38 0.99 2.65 1.28 0.001 656.16 398 1.80xl0\"4 0.151 0.999 0.75 2.26 4.43 1.42 0.0012 633.18 423 2.35 x l O - 4 0.167 0.999 1.13 4.63 7.34 1.57 0.0015 606.04 448 2.96 xlO\"4 0.183 0.999 1.95 8.69 12.39 1.76 0.0019 574.36 473 3.64 xlO\"4 0.198 0.999 2.98 15.14 20.10 1.98 0.0027 537.12 298 7.90 7.85 0.01 700.33 323 8.84 8.63 0.008 690.41 348 10.06 9.50 0.0076 675.61 6.89 373 398 423 448 473 a 11.84 14.59 18.61 24.98 34.28 10.47 11.58 12.85 14.34 16.17 0.0081 0.0095 0.0119 0.0159 0.0222 656.67 633.83 606.87 575.48 538.70 298 14.79 14.74 0.0187 700.67 323 16.41 16.20 0.015 690.78 348 18.40 17.84 0.0142 676.03 13.79 373 398 423 448 473 CC 21.03 24.75 29.87 37.53 48.41 19.67 21.73 24.11 26.90 30.30 0.0152 0.0178 0.0224 0.0298 0.0417 657.18 634.47 607.70 576.59 540.26 298 21.66 21.61 0.0274 701.01 323 23.97 23.76 0.0221 691.15 348 26.71 26.15 0.0209 676.46 20.68 373 398 423 448 473 CC 30.20 34.87 41.09 50.03 62.44 28.83 31.85 35.32 39.39 44.32 0.0223 0.0261 0.0328 0.0437 0.061 657.69 635.10 608.52 577.69 541.79 298 28.54 28.49 0.04 701.35 323 31.53 31.32 0.0291 691.51 348 35.03 34.47 0.0275 676.88 27.58 373 398 423 448 CC 39.36 44.98 52.29 62.49 37.99 41.97 46.52 51.86 0.0294 0.0344 0.0432 0.0575 658.20 635.73 609.33 578.77 215 Appendix C - Examples of EIS curve fitting results Table C - l Calculated data at 298 K 3.198e-03 = chi squared index fixed symbol start end % error 1 0 R 10.15 10.15 2.157 2 0 Q 6.305E-5 6.306E-5 4.192 3 0 n 0.8381 0.8381 0.8782 4 0 R 3.586E4 3.587E4 14.66 5 0 C 0.001316 0.0001316 25.18 6 0 R 316.7 316.9 39.32 7 0 C 0.0002119 0.0002119 11.46 8 0 R 1.652E5 1.652E5 18.41 Table C-2 Calculated data at 448 K 2.549e-03 = chi squared index fixed symbol start end % error 1 0 R 4.493 4.493 2.169 2 0 Q 0.000111 0.000111 4.126 3 0 n 0.8015 0.8016 0.7977 4 0 R 1.117E4 1.117E4 9.441 5 0 C 0.0001004 0.0001004.' 12.57 6 0 R 251 251.1 19.48 7 0 C 0.0007555 0.0007556 13.68 8 0 R 2.04E4 2.039E4 9.488 216 1.25E-04^ 1.20E-04 1.15E-04 1.10E-04 1.05E-04 1.00E-04 9.50E-05 9.00E-05 8.50E-05 8.00E-05 7.50E-05 7.00E-05 iL. . - 6.50E-05 O 6.00E-05^ 5.50E-05^ 5.00E-05-4.50E-05-4.00E-05-3.50E-05-3.00E-05 2.50E-05 2.00E-05-1.50E-05-1.00E-05-5.00E-06-3 1.19E-20 unknown.txt Model : R(QR)(CR)(CR) Wgt: Modulus -c-5 0 a . 8 e t a ft O *.. . \u00E2\u0080\u00A2 C , Msd. a C , Calc. 2.03E-20 2.00E-05 4.00E-05 6.00E-05 8.00E-05 1.00E-04 1.20E-04 1.40E-04 1.60E-04 1.80E-04 C ', F to Figure C-10 Complex capacitance plot showing measured and calculated capacitance at 298 K . 100,000-10,0004 1,000-1004-10-0.01 unknown.txt Mode l : R(QR)(CR)(CR) Wgt: Modulus A A O 6 o 6 o O o 6 A O A 0.1 10 100 Frequency, Hz 1,000 10,000 70 65 60 55 K50 45 > |-40 to. P C Q -25 20 15 10 5 100,000 e |Z|, Msd. \u00E2\u0080\u00A2 |Z|, Calc. A Angle, Msd. o Angle, Calc. ,o Figure C - l 1 Bode plot showing the measured and model calculated data for the EIS spectrum at 373 K O O E .n o N 1,200 1,150 1,100 1,050 1,000 950 900 850 800 750 700 650 600 550 500 450 400 350 300 250 200 150 100 50 0 -50 -100 -150 unknown.txt Model: R(QR)(CR)(CR) Wgt: Modulus , . L ~ ~~ \u00E2\u0080\u0094 \" -o \u00C2\u00A9 ! - -\u00E2\u0080\u00A2 \u00E2\u0080\u00A2 i - -- o \u00E2\u0080\u0094 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 o o \u00E2\u0080\u00A2 \u00E2\u0080\u0094 o \u00E2\u0080\u00A2 - ~ \u00E2\u0080\u0094 - -e ... I o , o , \u00E2\u0080\u00A2 \u00C2\u00AB 9 \u00E2\u0080\u00A2 I e e \u00E2\u0080\u00A2 I | I > ,\u00E2\u0080\u009E I \u00E2\u0080\u009E \u00E2\u0080\u009E \u00E2\u0080\u009E \u00E2\u0080\u00A2 z, Msd. \u00C2\u00BB z , Calc. 200 400 600 800 1,000 1,200 Z', ohm 1,400 1,600 1,800 2,000 2,200 t o Figure C - 1 2 Nyquist plot showing the measured and model calculated data for the E I S spectrum at 4 4 8 K VO Appendix D - Oxygen concentration and fugacity at SCWO temperatures and at constant pressure: PT0TAL = 250 bar T K fo2 Yo2 PL kg.L~l Co2 mol. kgH2a' Do2 2 -1 cm .s 473 249 1.07 0.66 0.31 3.64E-04 483 246 1.07 0.65 0.36 3.92E-04 493 243 1.07 0.64 0.41 4.22E-04 503 239 1.08 0.63 0.47 4.53E-04 513 234 1.08 0.62 0.53 4.84E-04 523 228 1.09 0.61 0.60 5.16E-04 533 222 1.10 0.60 0.67 5.49E-04 543 214 1.11 0.59 0.75 5.82E-04 553 206 1.12 0.58 0.83 6.17E-04 563 197 1.13 0.57 0.92 6.52E-04 573 187 1.15 0.55 1.01 6.87E-04 583 176 1.17 0.54 1.10 7.24E-04 593 164 1.20 0.52 ^ 1.19 7.61E-04 603 151 1.23 0.51 1.28 7.98E-04 613 137 1.28 0.49 1.37 8.36E-04 623 122 1.34 0.47 1.44 8.74E-04 633 106 1.42 0.44 1.50 9.13E-04 643 89 1.53 0.42 1.53 9.52E-04 653 71 1.68 0.38 1.52 9.92E-04 663 52 1.92 0.34 1.45 1.03E-03 673* 31 2.40 0.17 2.92 1.07E-03 220 Appendix E - Matlab code used to generate RKS results c l e a r %T = i n p u t ( ' T i n C e l c i u s ')+273.15; T = 4 7 3 . 1 5 ; P = 250; R = 83 .144; L = 0 . 9 5 ; % mole p e r c e n t Water component 1 z l = 0 .95 ; % mole p e r c e n t Oxygen component 2 z2 = 0 .05 ; % L i q u i d phase i n i t i a l guesses x l = 0 .95 ; x2 = 1 - x l ; % Vapour phase y l = 0 . 0 5 ; y2 = 1 - y l ; dont = 100; c n t = 1; sumxy = 1; t o t a l = x2 * 31 .25 + x l * 5 5 . 5 6 ; molesH20 = 5 5 . 5 6 ; w h i l e T <= 698 sumxy = 1; w h i l e abs(sumxy) >= l e - 9 % C o rap o n e n t 1, water T e l = 64.7. 3 ;%K, c r.). t i. c: a .1 p o i n t P e l = 2 2 1 . 2;%;Bar, c r i wl = 0 .344; f l = 0.48 + 1.574 . * wl - 0.176 . * w l / 2 ; T r l = T / T e l ; %Y1 - 1 - T r l . A 0.5; % i f T r l > 1 \u00E2\u0080\u00A2i f l - ( i + i .'0873' .* Y l ) , A 2; % e l s e % f l =--\u00E2\u0080\u00A2{! + 1.0873 . * Y l - 0.6377 . * Y l , A2 + 0.634 5 . .\"2; %end a l = 0 . 4 2 7 4 8 * R \" 2 * T c l \" 2 . / P e l . * (1 + f l . * (1 - T r l . A 0.5) b l = 0.08664 * R * T e l / P e l ; '^Component 2, oxygen Tc2 = 154.6;%K, c r i t i c a l p o i n t Pc2 = 5 0 . 4 ; % ;Bar, c r i w2 = 0 .025; f2 = 0.48 + 1.574 * w2 - 0.176 * w2 A 2; Tr2 = T / Tc2; %Y2 1 - Tr2 . A 0.5; * i f Tr2 > 1 % f 2 - (1 + 0. 8252 .* Y2) .' 2; %el.se % 1:2 - { .1 + 0.8252 . * Y2 + 0.25.12 . * Y2 . A 2 - 1.7039 . * Y2 . 3) .^2; tend a2 = 0.42748 * R * 2 * Tc2 \" 2 ./ Pc2 .* (1 + f2 .* (1 - Tr2 0.5) ) . A 2; b2 = 0.08664 * R * Tc2 / Pc2 ; % m i x i n g r u l e f o r c u b i c EOS; x i s f o r component 1 k i j = -0.11264; % 0 . 0 2 6 6 % e s t i m a t e d u s i n g Y c l and Vc.2-0.11264 L i j = - 0 . 0 0 6 6 ;%-0.00120;-0.0066 i f dont > 1 % L i q u i d c a l c u l a t i o n a = x l \"2 .* a l .* (1-kij) + 2 * x l * x2 .* (al .* a2) . A 0.5 * t i -k i j ) + x2 * 2 .* a2 .* ( 1 - k i j ) ; %b = x l * b l + x2 * b2; b = b l .* x l .* 2 .* (1-Lij) + x l .* x2 .* (bl + b2) .* (1-Lij) + b2 .* x2 . A 2 .* ( 1 - L i j ) ; A = a . * P / (R *2 . * T 2) ; B = b . * P . / (R . * T) ; r = 1; S = -1; t = A - B - B. A2; U = -A * B; Z l = l./6/r.*(36.*t.*s.*r-108.*u.*r. A2-8.*s. A3 + 1 2 . * 3 . A ( l . / 2 ) . * ( 4 . * t . A 3 . * r - t . A 2 . * s . A 2 -18.*t.*s.*r.*u+2 7.*u. A2.*r. A2+4.*u.*s. A3).\"(1/2).*r).\"(l/3)+2./3.*(-3.*t.*r+s.\"2)./r./(36.*t.*s.*r-108.*u.*r. A2-8.*s.\"3+12.*3.^(l/2).*(4.*t. A3.*r-t.*2.*s.*2-18.*t.*s.*r.*u+2 7.*u.*2.*r.*2+4.*u.*s. A3). A(1/2).*r). A(1/3)-1/3.*s./r; %Z.1.1 - -1 / 1 2 / r * {36*t*s*:r-108*u*r- N2-S*s A3 + .1.2*3'- (1/2) * (4 *t.'\3 *'r-t A 2 * s A 2 - 1 8 * t * s * r * u + 2 7 * u A 2 * r A 2 - f 4 * u * s A 3 ) A (1/2) + r ) A (1/3) -1/3* (-3 * t * r + s A 2 ) Irl (3 6*t * s * r - . l . 08 '* u*r A 2 - 8 * s A 3 + 12 * 3 A (1/2) * ( 4 * t A 3 * r - t A 2 * s A 2 -1 3 * t * s * r * u + 2 7 * u A 2 * r A 2 + 4 * u * s A 3 ) A ( 1 / 2 ) * r ) A ( 1 / 3 ) -l / 3 * s / r + l / 2 * i * 3 A ( 1 / 2 ) * ( 1 / 6 / r * ( 3 6 * t * s * r 1 0 8 * u * r A 2 -8*s A3+12*3 A (1/2) * ( 4 i t A 3 * . r - t A 2 # ' s A 2 -1 8 M : * s * r * u + 2 7 * u A 2 * r A 2 + 4 * u * s A 3 ) A ( 1 1 2 ) * r ) A ( 1 / 3 ) - 2 / 3 * ( 3 * t * r + s A 2 ) Irl ( 3 6 * t * s * r - 1 0 8 * u * r A 2 - 8 * s A 3 + 1 2 * 3 A (1/2) * (4 * t A 3 * r - t A 2 * s A 2 -18*t*s*r*u+27*u A2*r A2+4*u*s A3) A(1 1 2 ) * r ) A ( 1 / 3 ) } %Z12 = - 1 / 1 2 / r * ( 3 6 * t * s * r - 1 0 8 * u * r A 2 - 8 * s A 3 + 1 2 * 3 A ( 1 / 2 ) * ( 4 * t A 3 * r -t\"2*.s A2-18*t'*s*r*u+27*-u A2'-r A2 + 4*u' i rs A3) A (1/2) * r ) A (1/3) -1/3* (-3 * t * r + s A 2 ) I x I ( 3 6 * t * s * r - 1 0 8 * u * r A 2 - 8 * s A 3 + 1 2 + 3 A ( 1 / 2 ) * ( 4 * t A 3 * r t A 2 * s A 2 18'M::*s*r*u + 2 7 * u A 2 ' t r A 2 + 4*u*s A3) A (112.) *r) A (1/3) - l / 3 * s / r -l / 2 * i * 3 A (1/2) * ( 1 / 6 / r * ( 3 6 * t * s * r - 1 0 8 * u * r A 2 - 8 * s A 3 + 1 2 * 3 A (1/2) * ( 4 * t A 3 * r -t A 2 * s A 2 - 1 8 * t . * s * r * u + 2 7 * u A 2 * r A 2 + 4 * u * s A 3 ) A ( 1 / 2 ) * r ) A ( 1 / 3 ) 2/3* (-222 1 8 * t * s * r * u + 27*u A2 + rA2-!-4 *u*s A3) A (1/2) *r) A (1/3) ) d e l t a l l = (2.*al.*0.5) ./a .* ( ( x l . * a i . * 0 . 5 .* (1 - k i j ) ) + (x2.*a2.^0.5 .* (1 - k i j ) ) ); d e l t a l 2 = (2.*a2.*0.5)./a .* ( ( x l . * a l . A 0 . 5 .* (1 - k i j ) ) + (x2 . *a2 . A0 . 5 . * (1 - k i j ) ) ) ; p h i l l = exp( b l . / b .* ( r e a l ( Z l ) - 1) - l o g ( r e a l ( Z l ) - B) + A./B .* ( b l . / b - d e l t a l l ) .* l o g ( ( 2 . * r e a l ( Z l ) + 2.*B) . / ( 2 . * r e a l ( Z l ) ) ) ) ; p h i l 2 = exp( b2./b .* ( r e a l ( Z l ) - 1) - l o g ( r e a l ( Z l ) - B) + A./B .* (b2./b - d e l t a l 2 ) .* l o g ( ( 2 . * r e a l ( Z l ) + 2.*B) . / ( 2 . * r e a l ( Z l ) ) ) ) ; end a = y l .\"2 .* a l .* ( 1 - k i j ) + 2 .* y l .* y2 .* ( a l .* a2) .* 0.5 .* ( 1 - k i j ) + y2 . x 2 .* a2 .* ( 1 - k i j ) ; %b = y l \u00E2\u0080\u00A2 . * b l + y2 .* b2; b = b l .* y l 2 .* ( 1 - L i j ) + y l .* y2 .* ( b l + b2) .* ( 1 - L i j ) + b2 .* y2 . A 2 . * (1-Li'j ) ; A = a . * P . / (R^2 . * T. A 2 ) ; B = b . * P.- . / (R . * T) ; r = 1; S = -1; t = A - B - B . ^ 2 ; U = -A .* B; d e l t a v l = ( 2 . * a l . * 0 . 5 ) . / a .* ( ( y l . * a l . A 0 . 5 \u00E2\u0080\u00A2.* (1 - k i j ) ) + (y2.*a2. A0.5 .* (1 - k i j ) ) ); d e l t a v 2 = (2.*a2. * 0 .5)./a .* ( ( y l . * a l . \" 0 . 5 .* (1 - k i j ) ) + (y2.*a2 . ^ 0 . 5 . * (1 - k i j ) ) ); Zv = 1 . / 6 . / r . * ( 3 6 . * t . * s . * r - 1 0 8 . * u . * r . ^ 2 -8 . * s . \" 3 + 1 2 . * 3 . \" ( l . / 2 ) . * ( 4 . * t . x 3 . * r - t . x 2 . * s . * 2 -18.*t.*s.*r.*u+2 7.*u.\"2.*r.\"2+4.*u.*s.^3) .*(1/2) .*r) . A ( l / 3 ) + 2 . / 3 . * ( -3 . * t . * r + s . A 2 ) . / r . / ( 3 6 . * t . * s . * r - 1 0 8 . * u . * r . A 2 -8.*s. A3 + 12 .*3. A(1/2) . * ( 4 . * t . * 3 . * r - t . A 2 . * s . A 2 -18.*t.*s.*r.*u+2 7 . * u . A 2 . * r . x 2 + 4 . * u . * s . A 3 ) . A ( 1 / 2 ) . * r ) . x ( 1 / 3 ) - 1 / 3 . * s . / r ; p h i v l = exp( b l . / b .* (Zv - 1) - l o g ( Z v - B) + A./B .* ( b l . / b -d e l t a v l ) .* l o g ( ( 2 . * Z v + 2.*B) . / (2 . *Zv)) ) ; phiv2 = exp( b2./b .* (Zv - 1) - l o g ( Z v - B) + A./B .* (b2./b -d e l t a v 2 ) .* l o g ( ( 2 . * Z v + 2.*B) . / (2 . *Zv)) ) ; i f dont <= 1 KI = 1 ./ p h i v l ; K2 = 1 / p h i v 2 ; e l s e KI = p h i l l ./ p h i v l ; K2 = p h i l 2 ./ p h i v 2 ; x l = z l ./ (KI + L .* (1-K1)); y l = KI .* X l ; x2 = z2 ./ (K2 + L .* (1-K2)); y2 = K2 .* x2; sumxy = ( x l - y l ) + (x2 - y 2 ) ; end i f sumxy - 0; 223 L = L 1))./(K2 + (1-K2) (z2 . * (K2 - 1) . A2) end i f L < 0 dont L = 0 y l = z l ; y2 = z2; X l = 0; x2 = 0 ; sumxy = b r e a k end ( ( z l . * ( K l - 1)./(K1 + (1-K1).*L) + (z2.*(K2 *L) ) / (K2 1; 0 ; ./ ( ( z l . * ( K I - 1) + ( 1 - K 2 ) . * L ) . A 2 ) ) k 2 ) . / ( K I + (1-K1).*L).\"2 + end i f dont > 1 VI = Z l .* R Vv = Zv .* R d l = ( x l .* dv = ( ( y l . e l s e Vv = Zv .* VI = NaN; dv = ( ( y l . d l = NaN; end L p r i m e ( l , cnt) = L; p h i l 2 p r i m e ( 1 , c n t ) = p h i v 2 p r i m e ( 1 , c n t ) = K2prime(1,cnt) = K2 dvprime(1,cnt) = dv d l p r i m e ( 1 , c n t ) = d l Zvprime(1,cnt) = Zv Z l p r i m e ( 1 , c n t ) = Z l Vvprime(1,cnt) = Vv V l p r i m e ( 1 , c n t ) = VI Tprime(1,cnt) = T; y l p r i m e ( 1 , c n t ) = y l y2pr i m e ( 1 , c n t ) = y2 x l p r i m e ( 1 , c n t ) = x l x2prime(1,cnt) = x2 T ./ P; T . / P ; 18 + x2 .* 32) . / V I ; %must 18 + y2 .* 32)) ./ Vv; . * R . / P ; 18 + y2 .* 32)) ./ Vv; kg /T , phi12; p h i v 2 ; i f x2prime(1,cnt) > = T = T + 0.1; cn t = c n t + 1; P02 = y2prime .* P; 0.05, b r e a k , end Jromans s o l u b i l i t y model 224 kT = exp((0.046.*Tprime.*2 + 203.35.* Tprime .* l o g ( T p r i m e ./ 298.15) - (299.378 + 0.092 .* Tprime) .* (Tprime - 298.15)-(20591)) ./ (8.3144 .* T p r i m e ) ) ; gf = 1.016275 * 0.901725753; Co = p h i v 2 p r i m e .* kT .* P02 .* 0.9869233; I n e e d the p r e s s u r e i n a tmosphe res g i v e s moles p e r kg K20 % C a l c u l a t i n g t h e r a t i o o f c o r r o s i o n r a t e a t T t o c o r r o s i o n r a t e a t 2 98 Qt = 14227 + 46610000000000000000 .* Tprime (-6.8132); % A c t i v a t i o n e n e r g y T f i n l ( 1 , c n t -1) = T; r2 5 = C o ( l ) .* d l p r i m e ( l ) .* e x p ( - Q t ( l ) / (8 . 3145*298 . 15) ) ; r a t i o l = Co .* d l p r i m e .* exp(-Qt ./ (8.3145.*Tprime)) \u00E2\u0080\u00A2/ r25; ra t i o R K S = (x2prime .* 55.556 ./ (1 - x2prime)) .* d l p r i m e .* exp(-Qt ./ (8.3145 .* T p r i m e ) ) . / r25; % sigma. .12 - 0 .5 . * (34 .33 t 2 6 . 4 1 ) ; %nm %omegaD = 5e 12 . * T p r i m e . \" 4 l e - 8 . * T p r i m e . A 3 t 2e-5 . * T p r i m e . A 2 - 0 .0084 Tpr ime + 2 . 8 9 5 5 ; %D12 = 1 .8824e-3 . * s q r t ( T p r i m e . A 3 . * { 1 / 1 8 + 1 / 32)) . / (P . * 100 . * s.i .gmal2 A 2 . * omegaD); %m A 2/sec T f i n v ( l , c nt -1) = T; sigmal2 = 0.5 .* (34.33 + 26.41); %nm omegaD = 5e-12 .* Tprime. A4 - l e - 8 .* Tp r i m e . A 3 + 2e-5 . * Tprime. A2 - 0.0084 . * Tprime + 2.8955; D12 = 1.8824e-3 .* s q r t ( T p r i m e . A 3 .* (1 / 18 + 1 / 32)) ./ (P .* 100 .* sigmal2 A2 .* omegaD); % m A 2 / s e c p l o t r a t i o v = D12 .* lei ./ (r25 .* Vvprime); end F2 = y2prime . * P .* phiv2prime; % FUGACITY\" H21 = F2 ./ x2prime .* l e - 4 ; %HENRY\"S LAW CONSTANT H21 = l o g ( H 2 1 ) ; T f = 1000*(1./Tprime); FranckH = [-2.06 -1.59 -0.53 -0.35 -0.18 -0.03 0.15 0.30 0.62 0.82]; FranckT = [1.54 1.55 1.61 1.63 1.66 1.70 1.73 1.75 1.82 1.93]; %plot ( T p r i m e , x2pr i rae) %Coprime x 2 p r i m e . * t o t a l . / ( t o t a l . . * (1 - x2pr,i,:rne) . / molesH20) ; Coprime = x2prime ./ x l p r i m e .* 55.56; p l o t ( x 2 p r i m e , Tprime, y2prime, Tprime) 225 "@en . "Thesis/Dissertation"@en . "10.14288/1.0078566"@en . "eng"@en . "Materials Engineering"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "High temperature and high pressure corrosion of Ni-based alloys and stainless steels in ammoniacal sulphate solution"@en . "Text"@en . "http://hdl.handle.net/2429/30709"@en .