Environmental Degradation of Metals: Corrosion Technology Series 14

This Page Intentionally Left Blank

ISBN: 0-8247-9920-8 This book is printed on acid-free paper. Headquarters Marcel Dekker, Inc. 270 Madison Avenue, New York, NY 10016 tel: 212-696-9000; fax: 212-685-4540 Eastern Hemisphere Distribution Marcel Dekker AG Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland tel: 41-61-261-8482; fax: 44-61-261-8896 World Wide Web http://www.dekker.com The publisher offers discounts on this book when ordered in bulk quantities. For more information, write to Special Sales/Professional Marketing at the headquarters address above. Copyright  2001 by Marcel Dekker, Inc. All Rights Reserved. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, and recording, or by any information storage and retrieval system, without permission in writing from the publisher. Current printing (last digit): 10 9 8 7 6 5 4 3 2 1 PRINTED IN THE UNITED STATES OF AMERICA

Dedicated to the memory of our mentor and colleague Professor S. C. Sircar

This Page Intentionally Left Blank

Preface

All metallic components perform in one environment or another. At times the aggressiveness of the environment—or of species contained in the environment—brings about deterioration in physical and mechanical properties of the metal or alloy. The extent of damage may vary from a simple loss of appearance to the cracking of the component, leading to catastrophic failure. Any such degradation is of concern to the users of metals. Design and maintenance engineers should be well acquainted with the various types of damage and the conditions causing such damage in order to avoid or minimize them and to ensure reliability in service. Broadly speaking, the environment-assisted degradation of metals covers corrosion (both at ambient and higher temperatures), hydrogen damage, liquid metal attack, and radiation damage. The incidence of attack spreads from domestic to sophisticated nuclear and space applications. Excellent books are available dealing separately with these topics. However, a comprehensive book offering readers the basic knowledge of common types of environment-assisted damage and the measures to prevent such damage is both convenient and economical in terms of money and time, and the present book serves this purpose. The interested reader can make use of the references contained in the book for further assistance. This book has been designed to serve as a textbook or reference book in appropriate courses on materials for undergraduate and graduate students of metallurgical engineering, material science, chemical engineering, mechanical engineering, v

vi

Preface

and aerospace engineering. It can also be used as a reference for practicing engineers dealing with materials. As a prerequisite, a course on elements of physical metallurgy or material science is suggested. Thanks are due to Eric Stannard and Rita Lazazzaro of Marcel Dekker, Inc., without whose constant encouragement and perennial perseverance this book would not have seen the light of day. Our wives, Karuna, Aparna, and Supriya, deserve special thanks for sustaining us during the trying times of writing this book. U. K. Chatterjee S. K. Bose S. K. Roy

Contents

Preface

v

1. Introduction

1

2. Aqueous Corrosion: Fundamentals 2.1 Electrochemical Nature of Aqueous Corrosion 2.2 Thermodynamics of Aqueous Corrosion 2.3 Kinetics of Aqueous Corrosion 2.4 Passivity 2.5 Corrosion Rate Measurement References 3. Aqueous Corrosion: Forms 3.1 Classification of Aqueous Corrosion 3.2 General Corrosion 3.3 Galvanic Corrosion 3.4 Pitting 3.5 Crevice Corrosion 3.6 Intergranular Corrosion 3.7 Selective Leaching 3.8 Erosion Corrosion 3.9 Corrosion Cracking 3.10 Biologically Influenced Corrosion References

5 5 13 23 32 41 45 47 47 48 49 56 63 68 78 82 90 124 128 vii

viii

Contents

4.

Aqueous Corrosion: Prevention 4.1 Material Selection 4.2 Control of Environment 4.3 Protective Coatings 4.4 Cathodic Protection 4.5 Anodic Protection 4.6 Design Improvement References

131 132 147 155 162 170 172 177

5.

Tarnishing and Scaling Processes 5.1 Introduction 5.2 Thermodynamic Aspects of Metal-Single Oxidant Systems 5.3 Kinetic Aspects and Rate Equations 5.4 Defect Chemistry of Oxides and Other Inorganic Compounds 5.5 Mechanisms of Tarnishing and Scaling Processes 5.6 Scale Growth by Lattice and Grain Boundary Diffusion 5.7 Formation of Voids, Porosities, and Other Macrodefects in Oxide Scale and in the Substrate 5.8 Development of Stresses and Strains in the Growing Scales 5.9 Dissolution and Diffusion of Oxidant in Metals 5.10 Effects of Metal Surface Preparation and Pretreatment References

179 179 181 187 194 210 232 234 239 259 271 279

6.

Alloy Oxidation 6.1 Introduction 6.2 Doping Effect 6.3 Internal Oxidation and Catastrophic Oxidation 6.4 Sequences in Alloy Oxidation 6.5 Scaling of Binary and Ternary Alloys 6.6 Reactive Element Effects on the Oxidation Behavior of Alloys 6.7 Hot Corrosion 6.8 Protective Coatings for High-Temperature Applications 6.9 Conclusion References

283 283 288 295 304 312 330 346 374 408 409

7.

Liquid Metal Attack 7.1 Introduction 7.2 Liquid Metal Embrittlement 7.3 Corrosion by Liquid Metals References

413 413 413 431 434

8.

Hydrogen Damage 8.1 Introduction 8.2 Sources of Hydrogen

437 437 438

Contents 8.3 Types of Hydrogen Damage 8.4 Theories of Hydrogen Damage 8.5 Practical Examples 8.6 Preventive Methods References

ix 440 456 460 462 465

9. Radiation Damage 9.1 Introduction 9.2 Radiation-Induced Defect Production 9.3 Irradiation Growth 9.4 Void Swelling 9.5 Radiation-Enhanced Creep 9.6 Irradiation Strengthening and Embrittlement References Suggested Reading

467 467 469 470 472 475 479 484 485

Index

487

This Page Intentionally Left Blank

1 Introduction

Metals and alloys still constitute the most important group among engineering materials, and the demand for metallic materials with higher strength and special properties is on the increase with the advancement of technology. However, a serious drawback of metallic materials (and of other materials too!) is the deterioration in properties originating from their interaction with the environments in which they are to perform. Often this leads to a premature failure of metallic components with the allied hazards of plant shutdown and loss of economy, environmental pollution, and risk to human lives. The annual direct loss of natural resources, i.e., metals, due to environmental degradation is also substantial. The environment in question may vary from a simple atmospheric exposure at ambient temperatures to reactive gases at high temperatures, from soil to water, from weak chemicals to strong chemicals, from liquid metals to nuclear radiation. The nature of environmental degradation varies from environment to environment, although some features (e.g., loss of ductility, cracking) may be common to several of them. The various types of environmental degradation of metals are described by different terms such as corrosion, high-temperature corrosion, liquid metal attack, hydrogen damage, and radiation damage, which again have their narrower classifications in each group. The classification has been presented in Fig. 1.1. It is important to understand the nature of all types of environmental degradation of metals and alloys as vividly as possible so that preventive measures against metal loss and failures can be devised, and economy, safety, and reliability in the use of metallic components can be ensured. 1

2

Chapter 1

Figure 1.1 Environmental degradation of metals.

Corrosion of metals has received maximum attention of engineers and researchers in the field of environmental degradation of metals. Corrosion has been defined as the deterioration of a metal due to its chemical or electrochemical interaction with the environment. Corrosion has two broad distinct classifications: corrosion at ambient temperatures, which mostly involves aqueous phase or mois-

Introduction

3

ture, and high-temperature corrosion, which mostly involves reactive gases. The former is known as aqueous corrosion, which claims the major share in loss of metallic materials due to environmental degradation. In aqueous corrosion, localized types of attacks, rather than the general attack, are a matter of greater concern because of their unpredictability and lack of immediate visibility. Pitting, intergranular corrosion, and corrosion-assisted cracking among them account for many a premature failure of components. High-temperature corrosion involving the attack by reactive gases is also popularly known as oxidation of metals, although the electrochemical term ‘‘oxidation’’ applies to aqueous corrosion as well. The formation of scales and their instability account for the degradation due to oxidation by gases, and this behavior varies widely in single-component systems (pure metals) and in multicomponent systems (alloys). Another important type of high-temperature corrosion is hot corrosion, which is the attack of metals by molten salts in conjunction with gases. High-temperature corrosion is of particular concern in power-generating equipment that undergoes prolonged exposure to high-temperature contamination of fuel as aggravating the situation in some cases. Hydrogen damage refers to the results of the action of hydrogen in reducing the physical and mechanical properties of metals to a degree that renders them unreliable or useless. Hydrogen entry in the metal is possible from many sources, starting from its melting stage, during welding, or while undergoing corrosion in acid environments. The manifestation of damage is most generally a loss in ductility or the formation of cracks, external or internal. There are three main forms of hydrogen damage. One form results in internal pores, cracks, or fissures or external blister formation arising from hydrogen entrapment and subsequent pressure buildup in the flaws inside the metal. The second form of damage is embrittlement due to hydride formation. The third form of damage results despite the absence of a known chemical reaction or hydride formation; nevertheless, the hydrogen causes crack formation and growth, particularly in the presence of sustained loads. This form of damage has been described as ‘‘hydrogen stress cracking’’ or ‘‘hydrogen-assisted cracking,’’ and high-strength steels have been found to be particularly susceptible to this type of attack. Metallic components may come in contact with liquid metal during operations like brazing, soldering, or galvanizing and in some applications like the use of molten sodium as coolant in fast-breeder nuclear reactors. Liquid metal may corrode the solid metallic component or there may be diffusion-controlled intergranular penetration of liquid metal in the solid metal. However, the most drastic form of liquid metal attack is the instantaneous fracture of the solid metal in the presence of stress, a phenomenon described as ‘‘liquid metal embrittlement.’’ The flow behavior of the solid metal is not significantly affected, but a significant reduction in fracture stress or strain is encountered. With the advent of nuclear power generation, the concern about the radiation

4

Chapter 1

damage of structural components has increased considerably. The damage may manifest in four major forms: irradiation growth in anisotropic materials, irradiation-induced creep, void swelling, and degradation in impact properties. The affected component eventually suffers from dimensional instability as well as embrittlement. Keeping in view the importance of corrosion, both aqueous and high-temperature, the major portion of this book has been devoted to these forms of environmental degradation. Aqueous corrosion has been discussed in three chapters covering fundamentals, forms, and preventive measures. High-temperature corrosion is covered in two chapters: one dealing with pure metals and the other dealing with alloy systems in multicomponent oxidants. Hydrogen attack, liquid metal attack, and radiation damage have been discussed in separate chapters.

2 Aqueous Corrosion: Fundamentals

2.1 ELECTROCHEMICAL NATURE OF AQUEOUS CORROSION Corrosion in aqueous solutions is by far the most common of all corrosion processes. Aqueous medium is provided by water, seawater, and various process streams in industry. Moisture in atmosphere and water in soil account for the aqueous corrosion in these media. In all of these cases, water is hardly present in pure form. Rather, various salts and gases remain dissolved in it and their dissociation renders the water somewhat conducting. For all practical purposes, it acts as an electrolyte. The chemical nature of this electrolyte may be acidic, alkaline, or neutral.

2.1.1

Electrochemical Reactions

Aqueous corrosion can best be exemplified by the attack of zinc by hydrochloric acid. When a piece of zinc is placed in hydrochloric acid, vigorous reaction follows with the formation of zinc chloride and liberation of hydrogen gas. The reaction is chemically represented by Zn ⫹ HCl → ZnCl 2 ⫹ H 2 (g) ↑

(2.1)

5

6

Chapter 2

The attack on the surface is almost immediately visible. As the reaction proceeds, the piece of zinc thins down and loses its weight. The process continues until the entire piece disappears in the solution as zinc chloride. Equation 2.1 does not, however, reveal the entire picture of the reaction. The basis of this reaction is charge transfer, which is represented by two partial reactions: Zn → Zn2⫹ ⫹ 2e⫺

(2.2)

2H⫹ ⫹ 2e⫺ → H 2

(2.3)

The former represents the ionization of zinc in the acid solution with the liberation of valence electrons and the latter represents the discharge of the available dissociated hydrogen ions in the acid solution by the liberated electrons, which takes place on the metal surface itself. In terms of electrochemistry, the electron liberation reaction or the deelectronation reaction, as represented by Eq. 2.2, is called oxidation or anodic reaction, and the electron consumption or electronation reaction, as represented by Eq. 2.3, is called reduction or cathodic reaction. Figure 2.1 shows a schematic representation of these two reactions taking place at spatially separated sites on the metal surface. The site for oxidation or anodic reaction is called the anode and the site for reduction or cathodic reaction is called the cathode. The corrosion of zinc in hydrochloric acid is therefore an electrochemical reaction involving anodic and cathodic reactions of charge transfer.

Figure 2.1 Electrochemical reactions occurring during corrosion of zinc in air-free hydrochloric acid.

Aqueous Corrosion: Fundamentals

7

It is to be noted that although the actual dissolution process of metal is taking place through anodic reaction, cathodic reaction is equally important in the whole operation. The electrons liberated by anodic reaction are consumed in the cathodic process. A corroding metal does not accumulate any charge. It therefore follows automatically that these two partial reactions of oxidation and reduction must proceed simultaneously and at the same rate to maintain this electroneutrality. Some basic concepts of corrosion control also evolve from this simple electrochemical picture: Metal dissolution can be retarded by retarding the cathodic process; metal dissolution can also be retarded or stopped altogether by the supply of electrons to the corroding metal from any external source. The latter forms the basis of cathodic protection (Section 4.4). The picture as presented for the corrosion of zinc in hydrochloric acid is equally applicable for the corrosion of any other metal, say, iron or aluminum in hydrochloric acid or any other acidic medium in the absence of oxygen. In all cases, the metal will enter into the solution as ions through anodic process, i.e., Fe → Fe2⫹ ⫹ 2e⫺

(2.4)

Al → Al 3⫹ ⫹ 3e⫺

(2.5)

with the liberation of hydrogen gas through cathodic process. The anodic reaction for any corroding metal can thus be written in the general form: M → M n⫹ ⫹ ne⫺

(2.6)

where n is the valency of the metal involved. Cathodic reaction varies from medium to medium, but here also some generalization is possible. In acid media in the absence of oxygen in the examples discussed, the cathodic reaction is one of hydrogen evolution: 2H⫹ ⫹ 2e⫺ → H 2

(2.3)

However, if oxygen is present in acid solutions, the dominant cathodic reaction is that of oxygen reduction: O 2 ⫹ 4H⫹ ⫹ 4e⫺ → 2H 2O

(2.7)

Contamination of heavy salts such as FeCl 3 and CuCl 2 is common in some acids used industrially. Dissociation of these salts produces higher valence ions such as Fe3⫹ and Cu2⫹, which are also electron acceptors and get reduced to lower valency states, i.e., Fe 3⫹ ⫹ e⫺ → Fe 2⫹

(2.8)

Cu 2⫹ ⫹ e⫺ → Cu⫹

(2.9)

8

Chapter 2

Complete reduction of a contaminated metallic ion giving metal deposition is also possible: M⫹ ⫹ e⫺ → M

(2.10)

These additional cathodic reactions, if present, invariably accelerate the corrosion process. A very common example is the acceleration of corrosion of zinc in commercial grade of hydrochloric acid (called muriatic acid) because of the presence of FeCl 3 in the acid. All corrosive media are by no means acidic. Corrosion is widely experienced in neutral and alkaline aqueous media as well. Oxygen reduction provides the cathodic reaction in these media: O 2 ⫹ 2H 2O ⫹ 4e⫺ → 4OH⫺

(2.11)

Oxygen plays a big role in aqueous corrosion. Removal or lowering of oxygen content of the corroding medium has been an age-old practice for corrosion control. Since water gets partially dissociated into H⫹ and OH⫺ ions, simple reduction of water (which is actually the reduction of H⫹ ions) may provide the cathodic reaction under certain circumstances, even if oxygen is not available: 2H 2O ⫹ 2e⫺ → H 2 ⫹ 2OH⫺

(2.12)

Identification of cathodic reaction helps immensely in the understanding of corrosion in a particular system and its subsequent control. The familiar ‘‘rusting’’ of iron and ferrous alloys in the atmospheric exposure can thus be understood as: Anodic reaction: Fe → Fe 2⫹ ⫹ 2e⫺

(2.4)

Cathodic reaction: O 2 ⫹ 2H 2O ⫹ 4e → 4OH ⫺

⫺

(2.11)

A combination of the reaction products gives ferrous hydroxide, which in the presence of oxygen is further oxidized to ferric hydroxide, or rust. It may be mentioned that the term ‘‘rusting’’ is exclusive for ferrous materials only; other metals corrode but do not rust.

2.1.2

Electrochemical Cell Analogy

Referring back to Fig. 2.1 for the corrosion of zinc in hydrochloric acid, it becomes clear that the corroding system has four constituents: 1. 2.

An anode, where metal dissolution takes place according to Zn → Zn 2⫹ ⫹ 2e⫺ An electrolyte, the corrodent, into which the metallic ion passes

Aqueous Corrosion: Fundamentals

9

3. An electronic conductor, the metal itself, through which the electron passes from anodic site to cathodic site 4. A cathode, where electron is consumed according to 2H⫹ ⫹ 2e⫺ → H 2 The picture is analogous to a short-circuited current-producing electrochemical cell, such as a Daniell cell (Fig. 2.2). In a Daniell cell, zinc electrode constitutes the anode from which zinc passes into the ZnSO 4 electrolyte as zinc ions. Electrons released in the dissolution process pass through an electronic conductor, a metallic wire, and a high-resistance voltmeter, on their way to cathode, the copper electrode, where they get consumed by the available cupric ions that plate back as copper on the copper electrode according to Cu 2⫹ ⫹ 2e⫺ → Cu. The flow of positive current is assumed to be from zinc electrode to the copper electrode through the electrolyte, i.e., in the opposite direction of electron flow. If the resistance is withdrawn and the zinc and copper electrodes are connected by a metallic wire only, i.e., shortcircuited, zinc will dissolve at a higher rate, electrons will flow through the shortcircuited path, and copper will be deposited at the copper electrode. If now the short-circuited zinc and copper electrodes are taken from their respective compartments and placed in hydrochloric acid, zinc will still go into solution as zinc ion, electrons will flow through the short-circuited path, and in place of copper deposition a hydrogen evolution reaction will take place at the copper electrode. The corrosion of zinc in hydrochloric acid (Fig. 2.1) is analogous to this latest situation, except for the fact that the dissolution and reduction reactions are taking place on the same piece of metal at spatially separated sites. However, these sites are not eternally fixed. As corrosion proceeds, interchangeability of sites leads to a uniform attack on the surface and the piece thins down

Figure 2.2 Schematic representation of a Daniell cell.

10

Chapter 2

uniformly with the passage of time. This model of corrosion being analogous to a short-circuited electrochemical cell is often termed the local cell theory of corrosion. It will be useful to remember that in the corrosion cell the electrode from which current enters the electrolyte is the anode and this undergoes corrosion.

2.1.3

Types of Local Cell Formation

Three main types of local cell formation leading to corrosion are encountered in practice: 1. 2.

3.

Dissimilar electrode cells Concentration cells a. Salt concentration cell b. Differential aeration cell Differential temperature cells

Dissimilar Electrode Cells Dissimilar electrode cells may be formed when two dissimilar metals are in contact or due to the heterogeneity of the same metal surface. The Daniell cell is an example of the former. In practice, a copper pipe connected to a steel pipe or a bronze propeller in contact with the steel hull of a ship provides example for this type of corrosion cell. This is often referred to as galvanic coupling, in which the less noble metal becomes anode (discussed later in Section 3.3). A coldworked metal in contact with the same metal annealed leads to a similar situation, i.e., the cold-worked metal remaining anodic. On the same metal surface such type of cell formation may result from dissimilar phases and impurities, grain boundaries, differentially strained areas, and scratches or abrasions. In a single crystal, the different crystal faces differ in their electrochemical characteristics because of the difference in atomic orientation and, as a result, one crystal face tends to become anodic compared to others. Concentration Cells These are formed when the electrodes are identical but are in contact with solutions of differing composition. A salt concentration cell forms when one electrode is in contact with a concentrated solution and the other electrode with a dilute solution. On short circuiting, the electrode in contact with the dilute solution will be anodic (Fig. 2.3). The local variation of composition of the process stream inside the pipeline in a chemical plant may lead to such a situation in practice. A differential aeration cell forms when the identical electrodes are exposed to solutions of identical chemical composition that differ in oxygen content, which is

Aqueous Corrosion: Fundamentals

11

Figure 2.3 Salt concentration cell.

illustrated in Fig. 2.4. The electrode in contact with the less aerated or oxygenated solution will act as anode. Differential aeration cell formation is quite common in practice and is very important from the viewpoint of practical corrosion damages. A metallic bucket half-filled with water tends to get corroded just below the water line because of lower oxygen concentration compared to the area just above it near the water line. Corrosion damage invariably becomes pronounced underneath a corrosion product or at crevices where oxygen availability is low. Formation of concentration cells of both kinds account for the initiation of pits (discussed later in Section 3.4) in stainless steels or in some other metals and alloys exposed to seawater. Differential Temperature Cells These cells are formed when electrodes of the same metal, each of which is at a different temperature, are immersed in an electrolyte of the same initial composition. Such a situation may arise in practice in components of heat exchangers, boilers, and similar equipment. Polarity developed on an electrode varies from system to system. For a copper electrode in copper sulfate solution the electrode at higher temperature is cathode, but for lead the situation is just the reverse. For iron immersed in dilute aerated sodium chloride solutions, the hot electrode is initially anodic to the colder metal, but the polarity may reverse with the progress of corrosion.

12

Chapter 2

Figure 2.4 Differential aeration cell.

2.1.4

Homogeneous Theory of Corrosion

Corrosion essentially involves charge transfer reactions at the metal–electrolyte interface and these have been termed as anodic and cathodic reactions depending on whether the reaction releases electrons or consumes electrons. Both reactions proceed simultaneously and at the same rate, so that there is no charge accumulation in the corroding metal. In the local cell theory of corrosion, anodic and cathodic reactions have been visualized to occur at distinctively different sites, called anode and cathode. As discussed earlier, heterogeneities of one type or another tend to fix up the anodic and cathodic sites on a metal surface. Nevertheless, corrosion will proceed even if no such heterogeneity is present, as in the case of ultrapure metals. This is explained as follows: The necessary and sufficient condition for corrosion is the metal dissolution reaction and some electronation (reduction) reactions proceed simultaneously at the metal–environment interface. For these two processes to take place simultaneously, it is necessary and sufficient that the potential difference across the interface be more positive than the equilibrium potential of the reaction M ⫽ M n⫹ ⫹ ne⫺

(2.6)

and more negative than the equilibrium potential of the electronation reaction, say,

Aqueous Corrosion: Fundamentals 2H⫹ ⫹ 2e⫺ ⫽ H 2

13 (2.3)

involving electron acceptors contained in the electrolyte. Under these circumstances the metal dissolution and electronation reactions will occur randomly over the surface with regard to both space and time. This model of corrosion is termed the homogeneous theory of corrosion, the fundamental aspects of which will be discussed in detail in the Section 2.3.

2.2 THERMODYNAMICS OF AQUEOUS CORROSION The study of energy changes associated with chemical reactions comes within the scope of thermodynamics. These energy changes provide the driving force and control the spontaneous direction for a chemical reaction. Corrosion reactions being electrochemical in nature, by calculating the amount of energy associated in a given chemical reaction it is possible to indicate whether or not natural corrosion can take place in a given set of environmental conditions. It is, however, to be remembered that thermodynamics cannot predict the rate of reaction and despite an indicated large driving force, the rate of corrosion can be low because of the kinetic factors involved.

2.2.1

Free Energy Change

In any chemical or electrochemical reaction the magnitude of the change in free energy, ∆G, is a measure of the tendency of the reaction. A negative value of ∆G indicates the tendency for a reaction to proceed, whereas a positive value of ∆G is indicative of no tendency for reaction. A large negative value of ∆G indicates a pronounced tendency for reaction in the forward direction. The free energy change for a reaction at 25°C and with unit activity of reactants and products is represented by ∆G°, termed standard free energy change, is calculated from thermodynamic data. For example: Mg ⫹ H 2O(l) ⫹

1 O 2 (g) ⫽ Mg(OH) 2 (s), ∆G° ⫽ ⫺420, 600 cal 2

(2.13)

Cu ⫹ H 2O(l) ⫹

1 O 2 (g) ⫽ Cu(OH) 2 (s), ∆G° ⫽ ⫺28, 600 cal 2

(2.14)

Au ⫹

3 3 H 2O(l) ⫹ O 2 (g) ⫽ Au(OH) 3 (s), ∆G° ⫽ ⫹15, 700 cal 2 2

(2.15)

Thus, the tendency for magnesium to react with water in the presence of oxygen at 25°C is much larger compared to that of copper in the same environment. The positive value for gold indicates no tendency to react or corrode. If ∆G ⫽ 0, the system is in equilibrium, i.e., there is no tendency for backward or forward reac-

14

Chapter 2

tion. The change in free energy is a state function and is independent of the reaction path, but the reaction rate is dependent on the path followed. This is exemplified in Fig. 2.5. Position 1 is at a higher free energy state than position 2, and the difference in free energy (∆G) has a negative value when the transformation takes place from position 1 to position 2. The negative value is indicative of a spontaneous direction of transformation and it is the same whether the path followed is A, B, or C. The path B or C is visibly longer and the transformation or reaction rate along these paths has to be slower than along the path A. The path C has a hump. The reaction will not proceed from the trough position across the hump to position 2C unless some additional energy, called activation energy, is provided. Thus a negative free energy change is neither a guarantee for the reaction nor an indication for the rate at which it may proceed. On the other hand, a positive value of free energy change, like that for the transformation or reaction from position 2 to position 1, indicates that it is to be achieved only with the supply of additional energy and is not a spontaneous direction for reaction. The free energy change accompanying an electrochemical reaction, like a corrosion reaction, can be calculated as follows: ∆G ⫽ ⫺nFE where ∆G is the free energy change, in joules, n is the number of electrons involved in the reaction, F is the Faraday constant, in coulombs, E is the cell potential, in volts.

Figure 2.5 Effect of reaction path on reaction rate.

(2.16)

Aqueous Corrosion: Fundamentals

15

For all of the reactants participating in standard state, ∆G° ⫽ ⫺nFE°

(2.17)

It follows from the negative sign in Eq. 2.16 that a positive value for the cell potential E will ensure a negative free energy change and thus a spontaneous reaction. The concept of electrode potential, cell potential, and means for their determination are discussed below.

2.2.2

Electrode Potential and Cell Potential

A metal electrode immersed in an electrolyte develops a charged interface, a simplified picture of which is shown in Fig. 2.6. The interfacial structure of separated charge is commonly referred to as the electrical double layer, and it behaves much like a charged capacitor. Arising from this situation, a potential difference is developed at the electrode–electrolyte interface that is termed electrode potential. This corresponds to the establishment of an equilibrium of metal ionization reaction and the reaction of its recombination with electrons as represented by M i Mn ⫹ ne⫺ or

(2.18) M ⫽ Mn ⫹ ne⫺

Since an electrochemical cell has two electrodes, each of the electrodes has its own characteristic potential developed at the electrode–electrolyte interface, which is referred to as single-electrode potential or half-cell potential and the algebraic sum of the two single-electrode potentials constitutes the cell potential. This can be expressed as E ⫽ E1 ⫹ E2

(2.19)

where E is cell potential and E 1 and E 2 are the single-electrode potentials of the constituent electrodes. Referring back to the electrochemical cell comprising a copper electrode in equilibrium with cupric ions at unit activity and a zinc electrode in equilibrium with zinc ions at unit activity (Fig. 2.2), the cell potential is 1.1 V, which is easily measured with the help of a high-resistance voltmeter. This cell potential is an algebraic sum of the single-electrode potentials of the copper and zinc electrodes corresponding to the equilibria: Zn i Zn 2⫹ ⫹ 2e⫺

(2.20)

Cu 2⫹ ⫹ 2e⫺ i Cu

(2.21)

16

Chapter 2

Figure 2.6 Charge separation at electrode–electrolyte interface.

2.2.3

Standard Hydrogen Electrode Scale

There is no valid practical method to determine the absolute value of the potential difference existing at the metal–electrolyte interface, i.e., of the single-electrode potential. Any method of measurement of the potential difference needs a measuring instrument dipped in the electrolyte, which means the introduction of a second electrode–electrolyte interface. Thus, the measured value is always the potential difference between the two electrode–electrolyte interfaces. The difficulty in measuring the value of single-electrode potentials is over-

Aqueous Corrosion: Fundamentals

17

come by using a reference electrode whose potential has arbitrarily been assigned a zero value. A standard hydrogen electrode is such a reference electrode. When this is coupled with another electrode, a cell is formed whose potential eventually corresponds to the single-electrode potential of the second electrode because the reference electrode has a zero potential. The standard hydrogen electrode comprises a platinized platinum electrode immersed in a solution having unit activity of hydrogen ions through which hydrogen gas is bubbled under one atmospheric pressure (Fig. 2.7). The entire setup is maintained at 25°C. The potential of this electrode is expressed by E0H⫹/H2 ⫽ 0 and this corresponds to the equilibrium of the reaction, 2H⫹ ⫹ 2e⫺ i H 2

Figure 2.7 Schematic representation of standard hydrogen electrode (SHE).

(2.22)

18

Chapter 2

Figure 2.8 Cell containing reversible zinc and standard hydrogen electrode.

It must, however, be emphasized that the actual absolute value for E0H⫹/H2 is not zero. A standard hydrogen electrode is coupled with another electrode kept in its standard state, i.e., a metal in equilibrium with its ions at unit activity at 25°C, and the measured potential is termed the standard single-electrode potential or standard half-cell potential for the reaction occurring at that electrode. For example, as in Fig. 2.8, the single-electrode potential of the zinc electrode, E0Zn2⫹/Zn, corresponding to the equilibrium Zn 2⫹ ⫹ 2e⫺ i Zn

(2.20)

is ⫺0.76 V. This value is often written as E0Zn2⫹/Zn ⫽ ⫺0.76 V(SHE) where SHE refers to standard hydrogen electrode. Similarly, the standard singleelectrode potential for the copper electrode corresponding to the equilibrium Cu 2⫹ ⫹ 2e⫺ i Cu

(2.21)

is ⫹0.34 V (SHE).

2.2.4

EMF series and Corrosion Prediction

A listing of the standard single-electrode potentials constitutes the electromotive force series, or the emf series (Table 2.1). The potentials are referred to as redox potentials as well, meaning that this potential is the equilibrium potential for reduction and oxidation reactions. It is to be pointed out that the reactions shown in the table include reactions of O 2 /H 2O and O 2 /OH⫺ equilibria that are of immense importance in corrosion processes. It is also important to note that some

Aqueous Corrosion: Fundamentals

19

Table 2.1 Standard oxidation– reduction (redox) potentials a Au ⫽ Au 3⫹ ⫹ 3e O 2 ⫹ 4H⫹ ⫹ 4e ⫽ 2H 2 O Pt ⫽ Pt 2⫹ ⫹ 2e Pd ⫽ Pd 2⫹ ⫹ 2e Ag ⫽ Ag⫹ ⫹ e 2Hg ⫽ Hg 22⫹ ⫹ 2e Fe 3⫹ ⫹ e ⫽ Fe 2⫹ O 2 ⫹ 2H 2 O ⫹ 4e ⫽ 4OH⫺ Cu ⫽ Cu 2⫹ ⫹ 2e Sn4⫹ ⫹ 2e ⫽ Sn 2⫹ 2H⫹ ⫹ 2e ⫽ H 2 Pb ⫽ Pb 2⫹ ⫹ 2e Sn ⫽ Sn 2⫹ ⫹ 2e Ni ⫽ Ni 2⫹ ⫹ 2e Co ⫽ Co 2⫹ ⫹ 2e Cd ⫽ Cd 2⫹ ⫹ 2e Fe ⫽ Fe 2⫹ ⫹ 2e Cr ⫽ Cr 3⫹ ⫹ 3e Zn ⫽ Zn 2⫹ ⫹ 2e Al ⫽ Al 3⫹ ⫹ 3e Mg ⫽ Mg 2⫹ ⫹ 2e Na ⫽ Na⫹ ⫹ e K ⫽ K⫹ ⫹ e

⫹1.498 ⫹1.229 ⫹1.2 ⫹0.987 ⫹0.799 ⫹0.788 ⫹0.771 ⫹0.401 ⫹0.337 ⫹0.15 0.000 ⫺0.126 ⫺0.136 ⫺0.250 ⫺0.277 ⫺0.403 ⫺0.440 ⫺0.744 ⫺0.763 ⫺1.662 ⫺2.363 ⫺2.714 ⫺2.925

a

25°C, volts vs. normal hydrogen electrode. Electrode potential values are given and are invariant (e.g., Zn ⫽ Zn 2⫹ ⫹ 2e, and Zn 2⫹ ⫹ 2e ⫽ Zn, are identical and represent zinc in equilibrium with its ions with a potential of ⫺0.763 V vs. normal hydrogen electrode). Source: A. J. de Bethune and N. A. S. Loud, Standard Aqueous Electrode Potentials and Temperature Co-efficients at 25°C, Clifford A. Hampel, Skokie, IL, 1964.

of the reactions in the table have been represented as reduction reactions, following the international convention. Accordingly, the redox potentials above hydrogen have a positive sign and those below hydrogen have a negative sign. The sign will change if any reaction is considered for a reverse reaction, i.e., an oxidation reaction. Referring again to the copper-zinc cell in Fig. 2.2, the standard single-electrode potential for the copper electrode where reduction takes place is ⫹0.34 V (SHE), whereas for the zinc electrode, which undergoes an oxidation

20

Chapter 2

reaction, the standard single-electrode potential will be ⫹0.76 V (SHE). Thus the cell potential as an algebraic sum of these two values will yield a value of 1.1 V. This has a positive sign and, therefore, ∆G° will yield a negative value (Eq. 2.17) and the reactions as considered, i.e., the oxidation of zinc metal and reduction of Cu 2⫹ ions are thermodynamically favored. Thus in any electrochemical reaction, the most negative or active half-cell tends to be oxidized and the most positive or noble half-cell tends to be reduced. The larger positive value of redox potential is synonymous with high oxidizing ability of the reaction. It is immediately apparent from the emf series that the metals below hydrogen will corrode (at least, they will have the tendency to corrode) in oxygen-free acid solutions with the evolution of hydrogen. If oxygen is available, the other reaction of oxygen reduction forming water will also take place. The metals above hydrogen will have no tendency to react or corrode with the evolution of hydrogen in oxygen-free acid solutions, e.g., Cu in H 2 SO 4 solution. However, if oxygen is available, copper will readily corrode in acid solutions owing to the higher position of E0O2/H2O with respect to E0Cu⫹/Cu in the series. Copper will also have a tendency to corrode in neutral solutions with dissolved oxygen because E0O2/OH⫺ ⬎ E0Cu⫹/Cu. Gold and platinum are noble metals because they exhibit large positive redox potentials and have no tendency to corrode except in the presence of extremely powerful oxidizing agents. All of these considerations, however, are applicable to the systems at unit activity. With the change of concentration of reactants and products, single-electrode potentials change, and the tendency for reaction also changes accordingly. The change in electrode potential with change in concentration and temperature is calculated from the Nernst equation: E ⫽ E0 ⫹ 2.3

a RT log oxid nF a red

(2.23)

where E is the single-electrode potential E0 is the standard single-electrode potential R is the gas constant T is absolute temperature n is the number of electrons transferred in the reaction F is the Faraday constant a oxid and a red are the activities of oxidized and reduced species, respectively. Note that with an increase in the activity (concentration) of the oxidized species, the potential becomes more positive. Thus, potential can be considered to be a measure of the oxidizing power of the solution.

2.2.5

Potential-pH Diagrams

Potential-pH diagrams, also known as Pourbaix diagrams, are the graphical representation of the stability of a metal and its corrosion products as a function of

Aqueous Corrosion: Fundamentals

21

the potential and pH (acidity or alkalinity) of the aqueous solution. The potential is shown on the vertical axis and the pH on the horizontal axis. Such diagrams are constructed from calculations based on the Nernst equation and the solubility data for various metal compounds. The potential-pH diagram for an Fe-H 2 O system is shown in Fig. 2.9. In the diagram, the horizontal lines represent pure electron transfer reactions, dependent solely on potential, but independent of pH: (1) Fe ⫽ Fe 2⫹ ⫹ 2e⫺

(2.24)

(2) Fe 2⫹ ⫽ Fe 3⫹ ⫹ e⫺

(2.25)

These lines extend across the diagram until the pH is sufficiently high to facilitate the formation of hydroxides, represented by vertical lines, thereby reducing the concentration of Fe 2⫹ and Fe 3⫹ ions. The boundary is often set arbitrarily at the concentration of these ions at 10⫺6 g-ion/liter, which is indicative of a negligible dissolution or corrosion of the metal in the medium. The vertical lines correspond to the reactions: (3) Fe 2⫹ ⫹ 2H 2 O ⫽ Fe(OH) 2 ⫹ 2H⫹

(2.26)

(4) Fe 3⫹ ⫹ 3H 2 O ⫽ Fe(OH) 3 ⫹ 3H⫹

(2.27)

Figure 2.9 Potential-pH (Pourbaix) diagram for Fe-H 2 O system.

22

Chapter 2

There is no electron transfer involved and these reactions are solely dependent on pH. The sloping lines represent equilibria involving both electron transfer and pH, e.g., (5) Fe 2⫹ ⫹ 3H 2 O ⫽ Fe(OH) 3 ⫹ 3H⫹ ⫹ e⫺

(2.28)

(6) Fe ⫹ 2H 2 O ⫽ HFeO 2⫺ ⫹ 3H⫹ ⫹ 2e⫺

(2.29)

The hydrogen and oxygen are also shown in the diagram by dotted lines. The hydrogen line represents the equilibria: 2H⫹ ⫹ 2e⫺ ⫽ H 2

in acid solutions

(2.30)

or 2H 2 O ⫹ 2e⫺ ⫽ H 2 ⫹ 2OH⫺

in neutral or alkaline solutions

(2.31)

These two reactions are equivalent and their pH dependence of single-electrode potential is represented by: EH⫹/H2 ⫽ E0H⫹/H2 ⫺ 0.059 pH

(2.32)

At pH ⫽ 0, i.e., for [H⫹] ⫽ 1, E0H⫹/H2 ⫽ 0 and the slope is ⫺0.059 V. Similarly, for oxygen equilibrium with water the corresponding reactions at lower and higher pH are: O 2 ⫹ 4H⫹ ⫹ 4e⫺ ⫽ 2H 2 O

(2.33)

and O 2 ⫹ 2H 2 O ⫹ 4e⫺ ⫽ 4OH⫺

(2.34)

The pH dependence of single-electrode potential is represented by E O2 /H2O ⫽ E0O2 /H2O ⫺ 0.059 pH

(2.35)

At pH ⫽ 0, E0O2/H2O ⫽ 1.226 V and at pH ⫽ 1, i.e., for [OH⫺] ⫽ 1, E0O2/H2O ⫽ 0.401 V. Here again the slope of the line is ⫺0.059 V. Water is stable in the area delineated by these two lines. Below the hydrogen line it is reduced to hydrogen gas and above the oxygen line it is oxidized to oxygen. The potential-pH diagram shows three clear-cut zones: 1. 2. 3.

Immunity zone. Under these conditions of potential and pH, iron remains in the metallic form. Corrosion zone. Under these conditions of potential and pH, iron corrodes forming Fe 2⫹ or Fe 3⫹, or HFeO 2⫺. Passive zone. Under these conditions of potential and pH, protective layers

Aqueous Corrosion: Fundamentals

23

of Fe(OH) 2 or Fe(OH) 3 form on iron and further corrosion of iron does not take place. Such diagrams can be used for 1. Predicting the spontaneous direction of reactions 2. Estimating the stability and composition of corrosion products, and 3. Predicting environmental changes that will prevent or reduce corrosion. With reference to Fig. 2.9, corrosion prevention can be achieved by lowering the electrode potential down to the zone of immunity, raising the electrode potential up to the region of passivity, or raising the pH or alkalinity of the solution so that a passive film is formed. There are, however, a number of limitations in using such diagrams. The most important of these is that they represent equilibrium conditions and hence cannot be used for predicting the rate of a reaction. The tacit assumption that corrosion products, e.g., oxides, hydroxides, and so forth, lead to passivity may not always be true because they may not precipitate on the metal surface. The possibility of precipitation of other ions such as chloride, sulfate, and phosphate has been ignored. Finally, the pH at the metal surface may vary drastically because of side reactions and a prediction of corrosion based on the bulk pH of the solution may be misleading.

2.3 KINETICS OF AQUEOUS CORROSION Corrosion reactions are analogous to what happens in a short-circuited cell (Section 2.1.2). The system is no longer in equilibrium and the reaction proceeds either in the forward or backward direction at the electrodes generating a finite current flow in the circuit. The magnitude of current is a direct measure of the extent of corrosion and rate of current flow is a measure of the rate of corrosion. It is important to know the rate of corrosion in practice to determine the applicability of a metallic component in a given environment. Even if the tendency for a reaction is high, as may be evident from thermodynamic considerations, a negligibly low rate of corrosion can ensure the reliable use of a metal even in an apparently corrosive medium. On the other hand, a metal in a given environment may corrode at a rapid rate in a medium having a relatively low tendency for reaction (potential difference). The study of the rate of reaction comes under the purview of kinetics. Electrochemical dissolution reactions obey the Faraday laws of electrolysis. When a current of magnitude I ampere is flown through the circuit, the amount of mass of metal dissolved or deposited in given by:

24

Chapter 2

W⫽

Ita nF

(2.36)

where F is the Faraday constant (96,500 coulombs/equivalent) n is the number of equivalents exchanged a is the atomic weight of the metal, and t is the time in seconds The rate of metal dissolution, or corrosion, r, is obtained by dividing Eq. 2.36 by the surface area A and time t. Thus, r⫽

ia W ⫽ tA nF

(2.37)

where i, defined as current density, equals I/A. The total current I or the current density i, which is equivalent to corrosion rate, will appear in all discussion related to the kinetics of corrosion.

2.3.1

Exchange Current Density

For any electrode reaction at equilibrium current does not flow through the circuit, but there is always a finite exchange of ions and atoms at the interface. For example, for the reaction: Zn i Zn 2⫹ ⫹ 2e⫺

(2.20)

some moles of zinc atoms are leaving the surface and entering the electrolyte as zinc ions; at the same time, an equal number of zinc ions from the electrolyte are getting reduced on the electrode surface. Since electron transfer is involved, the rate of exchange can be expressed in terms of current density using Faraday’s law: r oxid ⫽ r red ⫽

i0 nF

(2.38)

where r oxid and r red are the equilibrium oxidation and reduction rates, and i 0 is the exchange current density. Thus, exchange current density can be defined as the rate of oxidation or reduction at an equilibrium electrode expressed in terms of current density. The magnitude of exchange current density is a measure of how easily a reaction attains equilibrium. It varies for different reactions and, for a particular reaction, the magnitude of exchange current density is a function of the electrode composition, surface roughness, surface impurities, and temperature. The equilibrium of hydrogen evolution reaction on different metal surfaces presents interesting data in terms of exchange current density, which is shown

Aqueous Corrosion: Fundamentals

25

Figure 2.10 Hydrogen–hydrogen ion exchange current densities on different metals.

graphically in Fig. 2.10. On the mercury surface, the equilibrium is attained with much sluggishness, whereas on a platinum surface it is readily established giving a large value of exchange current density. A platinized platinum surface, because of its large projected surface area, gives a still higher value of exchange current density. Presence of trace impurities such as arsenic, sulfur, and antimony compounds reduces the exchange current density for hydrogen equilibrium drastically, whereas an increase in temperature raises exchange current density.

2.3.2

Polarization

As the electrodes of a cell, such as in Fig. 2.2, are short-circuited, current starts flowing through the circuit, indicating that net oxidation and reduction reactions are taking place at the electrodes. The potentials of these electrodes start deviating from their equilibrium potential values. This deviation from equilibrium potential is called polarization and the extent of deviation is termed overvoltage which is expressed by the Greek letter, η. Let us consider the hydrogen evolution reaction at equilibrium: 2H⫹ ⫹ 2e⫺ i H 2 , E0H⫹/H2 ⫽ 0 volt

(2.22)

Actual hydrogen liberation will not take place unless more electrons are supplied to the electrode, i.e., the electrode is made more negative. Similarly, the oxidation reaction will take place only at the potentials more positive than 0 V (SHE). Metal dissolution reaction is of basic importance in corrosion. Let us consider the zinc dissolution equilibrium: Zn s Zn2⫹ ⫹ 2e⫺ ⫽ E0Zn/Zn2⫹ ⫽ ⫺0.76 V(SHE)

(2.20)

26

Chapter 2

Dissolution of zinc will proceed only at potentials more positive than ⫺0.76 V (SHE). There are two principal types of polarization, i.e., activation polarization and concentration polarization. These are discussed below. Activation Polarization Activation polarization arises out of a slow step in the electrode reaction for which an activation energy in the form of an increment in potential is required for the reaction to proceed. This will be best exemplified by the hydrogen evolution reaction. The hydrogen evolution reaction consists of several steps as shown in Fig. 2.11. Either the electron transfer step (no. 2) or the formation of hydrogen molecules (no. 3) is deemed to be the slowest step in the reaction sequence and the rate of overall reaction will depend on how fast or how slow it is proceeding. Therefore, in order to have a higher rate of reaction, expressed in terms of increased current density, an increase in potential is to be effected. The relationship between reaction rate and change in potential (overvoltage) is expressed by the Tafel equation: η a ⫽ ⫾β log

i i0

(2.39)

where η a is overvoltage activation polarization, in volts β is a constant, called Tafel constant (also expressed in volts), and is usually of the order of 0.1 V A graphical representation of the above equation, as applied to hydrogen evolution reaction, with a β slope of 0.1 V is shown in Fig. 2.12. It can be noted from the graph that 0.1 V change in overvoltage can effect a 10-fold increase or decrease in the reaction rate.

Figure 2.11 Steps involved in hydrogen reduction reaction.

Aqueous Corrosion: Fundamentals

27

Figure 2.12 Activation-polarization curve of a hydrogen electrode.

Dissolution reactions (anodic) in corrosion are usually controlled by activation polarization where the solvation of ion is the probable rate-controlling step. Hydrogen evolution reactions (cathodic reactions) are controlled by activation polarization where the concentration of hydrogen ions is high. Concentration Polarization A buildup or depletion of ions at the electrode surface as a result of reaction will change the value of the electrode potential according to the Nernst equation. For example, for a corroding zinc electrode, concentration of zinc will increase with dissolution in the vicinity of the electrode. The value of a oxid in the equation increases, causing the electrode potential to shift in a positive direction. For the hydrogen evolution reaction, the higher rate of discharge of hydrogen ions at the electrode surface brings down the value of a oxid and the electrode potential according to the Nernst equation, shifts in a negative direction. However, the rate of discharge of hydrogen ions at the electrode surface is dependent on the diffusion of hydrogen ions from the bulk of the solution to the surface; a maximum or limiting value of this reduction reaction is given by iL ⫽

DnFC x

(2.40)

where i L is called limiting diffusion current density, amp/cm 2 D is the diffusion coefficient for H⫹ ion n is the number of electrons transferred F is the Faraday number C is the bulk concentration of H⫹ ions in the solution and x is the thickness of diffusion layer adjacent to the electrode surface through which the concen-

28

Chapter 2

Figure 2.13 (a) Concentration polarization curve for reduction process and (b) effect of environmental variations on concentration polarization curve.

tration of the reacting species (H⫹ ions) changes from C in the bulk to zero at the electrode surface A mathematical expression for concentration polarization involves i L and is given by ηc ⫽

冢

2.3RT i log 1 ⫺ F iL

冣

(2.41)

where η c is overvoltage due to concentration polarization, in volts. A graphical representation of the equation is shown in Fig. 2.13. It can be seen from the graph as well as from Eq. 2.41 that as i approaches i L , η c tends to infinity. As evident from Eq. 2.40, factors like increasing velocity (smaller x), increasing temperature (higher D), and increasing concentrations increase the value of i L , i.e., a shift of the vertical portion of the curve in Fig. 2.13 more to the right. There is no question of concentration polarization where the supply of reacting species is abundant. Hence in metal dissolution reaction its effect is negligible, as the supply of metal atoms for dissolution is unlimited. On the other hand, for a hydrogen evolution reaction, concentration polarization becomes significant in the solutions of low H⫹ ion concentration. More often the reduction process is controlled by a combined polarization, i.e., activation polarization at lower reaction rates and concentration polarization at higher reaction rates, as i approaches i L . A graphical representation of such combined polarization is shown in Fig. 2.14.

2.3.3

Mixed Potential Theory

In the case of zinc corroding in hydrochloric acid, as has been discussed in Section 2.1.1, the constituent anodic and cathodic reactions are:

Aqueous Corrosion: Fundamentals

29

Figure 2.14 Combined curve of activation and concentration polarizations for reduction process.

Zn → Zn 2⫹ ⫹ 2e⫺

(2.2)

2H⫹ ⫹ 2e⫺ → H 2

(2.3)

These reactions again are the anodic and cathodic partials, respectively, of the equilibrium reactions: Zn i Zn 2⫹ ⫹ 2e⫺

(2.20)

2H⫹ ⫹ 2e⫺ i H 2

(2.22)

These equilibria correspond to the potentials ⫺0.76 V (SHE) and 0 V, respectively, with their characteristic exchange current density values which can be conveniently shown in a potential versus current density (usually log of current density) plot. As the current flows, polarization sets in. Assuming the polarization to be activation polarization only, straight line variation corresponding to the Tafel equation will be obtained and the polarization plots for the relevant partial equations will tend to intersect as shown in Fig. 2.15. At the point of intersection the rate of anodic reaction is equal to the rate of cathodic reaction, which corresponds to the corrosion current density, i corr : i a ⫽ i c ⫽ i corr as no accumulation of charge is observed in a piece of corroding metal. The potential corresponding to this point of intersection is the potential of the corroding metal, which is represented by E corr and is called corrosion potential. It can be readily noticed, therefore, that the corrosion potential is not the equilibrium potential of either of the constituent partial reactions, but rather some intermediate potential determined by simultaneous occurrence of these partial

30

Chapter 2

Figure 2.15 Polarization of constituent electrode reactions representing corrosion of zinc in air-free acid solution giving mixed potential, E corr , and corrosion current, i corr .

oxidation and reduction reactions at the same rate. This theory, propounded by Wagner and Traud in 1938 [1], is known as mixed-potential theory of corrosion. Mixed-potential theory paves the way for understanding the homogeneous theory of corrosion (Section 2.1.4). For corrosion to occur, spatially separated anodic and cathodic sites are not required if the system provides a potential where there is simultaneous occurrence of oxidation and reduction reactions. This potential obviously will be more positive than the equilibrium potential of the anodic reaction and more negative than the equilibrium potential of the cathodic reaction.

2.3.4

Importance of Kinetic Considerations

The electrode kinetic parameters i o , i L , β a , and β c determine the rate of a corrosion reaction. The rate of corrosion in terms of current density for a single-electron transfer reaction is expressed by the Bulter-Volmer equation [2]: i ⫽ i o (e(1⫺β)Fη/RT ⫺ e⫺βFη/RT )

(2.42)

The difference between the redox potentials of the cathodic and anodic reactions is the driving force for a corrosion reaction. However, as Eq. 2.42 shows,

Aqueous Corrosion: Fundamentals

31

the rate of corrosion depends on exchange current density and Tafel slope values. An example will clarify this. In the emf series, the redox potential for zinc is more negative (⫺0.76 V) than that for iron (⫺0.44 V). In an acid solution, however, the corrosion rate of iron is greater than that of zinc. The higher values of exchange current densities for dissolution of iron lead to this behavior. Graphically, the situation is represented in Fig. 2.16. The effect of the addition of an oxidizing agent on the corrosion rate can also be appreciated clearly through graphical representation of kinetic behavior. For example, the corrosion rate of zinc in hydrochloric acid increases in the presence of ferric chloride. Two reduction reactions, those of ferric ions and hydrogen ions, are active in this case and the total reduction rate equals the corrosion rate according to: i corr ⫽ i (Fe 3⫹ →Fe 2⫹) ⫹ i (H⫹→H2)

(2.43)

In the graphical representation shown in Fig. 2.17, the resultant cathodic polarization is shown by dotted lines. It is to be noted that i corr ⬎ i′corr and E corr has also shifted slightly. Rate of hydrogen evolution shows a decrease from a value of i′corr to i H⫹→H2 although the overall increase in corrosion rate has resulted from the additional cathodic reduction reaction.

Figure 2.16 Comparison of kinetic behavior of iron and zinc in acid solution.

32

Chapter 2

Figure 2.17 Effect of oxidizer addition on the corrosion behavior of a metal in acid solution.

In the situations where the reduction reaction is diffusion-controlled, the anodic polarization curve may intersect the cathodic polarization curve in the i L range (Fig. 2.18), the value of which increases with increasing velocity. The corrosion rate increases as the velocity or i L increases, but it assumes a steady value where i L ⬎ i a and the reaction becomes activation-controlled. The kinetic considerations and graphical representations will be taken up further in the discussion on passivity (Section 2.4) and galvanic corrosion (Section 3.3).

2.4 PASSIVITY Passivity refers to the phenomenon of loss of chemical reactivity of a metal or an alloy in an environment where thermodynamically the reaction ought to have occurred. It results from the formation of a thin, oxidized, protective film on the surface of a metal. Many metals active in the emf series, including important structural metals like aluminium, iron, nickel, chromium, titanium, and their alloys, can be passivated simply by exposure to strong oxidizing media or by anodic polarization or both. Other metals that show passivity include silicon, tantalum, niobium, molybdenum, and zirconium. Usual corrosion conditions are not sufficiently oxidizing to induce passivity in iron, but they do passivate aluminium and titanium. Iron can be rendered pas-

Aqueous Corrosion: Fundamentals

33

Figure 2.18 (a) Electrochemical behavior of a normal metal corroding with a diffusion-controlled cathodic process showing the effect of velocity. (b) Corrosion rate as a function of velocity.

sive by an initial exposure to fuming nitric acid. Its subsequent exposure to a corrosive media, such as dilute sulfuric acid, will bring down the corrosion rate drastically, by an order of 10 4 to 10 6. However, this induced passivity is not stable and is destroyed by vibration or scratching of the surface. Addition of highly oxidizing chemicals like chromates or nitrites, called passivators, to the corrosive media helps iron to remain in the passive state. With the addition of chromium in iron, passivity is achieved in relatively low oxidizing conditions. Stainless steels, with chromium content above 12%, constitute a class of passive alloys used as corrosion-resistant materials under oxidizing conditions; they lose their passivity under reducing conditions.

34

Chapter 2

2.4.1

Electrochemical Behavior of Active-Passive Metals

If a metal exhibiting passivity, e.g., iron in 1 N H 2 SO 4 , is anodically polarized by the glavanostatic method (i.e., in which the current is maintained constant and the potential is allowed to change), the polarization curve so obtained has the shape as shown in Fig. 2.19. The initial portion of the curve shows an increase in potential in the positive direction with increasing applied current density in conformity with the typical Tafel behavior. Beyond a certain value of current density the potential jumps abruptly to a higher value accompanied by oxygen evolution. At this high potential range, the anodic discharge of oxygen takes place according to reverse of the reaction (2.7), i.e., 2H 2 O → O 2 ⫹ 4H⫹ ⫹ 4e⫺

(Reverse of Eq. 2.7)

However, if the metal is polarized potentiostastically (i.e., whereby the increment in potential is given by potentiostat and the current is allowed to adjust itself), the polarization curve takes the shape as shown in Fig. 2.20. The dissolution shows Tafel behavior initially, the current density increasing with increasing applied potential. This is the active region. At E pp , which is called the primary passive potential, the current density shows a maximum value, i cr , the critical current density for passivity. Polarization beyond E pp lowers the dissolution drastically, as characterized by the low value of current density, which remains essentially independent of potential over a considerable potential range. This is termed the passive region. At still higher potentials, the current density

Figure 2.19 Galvanostatic anodic polarization curve for an active-passive metal.

Aqueous Corrosion: Fundamentals

35

Figure 2.20 Potentiostatic anodic polarization curve for an active-passive metal.

again shows an increase, and we have what is termed transpassive region. The increase in current density may be indicative of oxygen evolution, as for iron in 1 N H 2 SO 4 , or of increased anodic dissolution as for chromium or 18-8 stainless steel in the same solution according to the reaction; 2Cr ⫹ 7H 2 O → Cr 2 O 7 2⫺ ⫹ 14H⫹ ⫹ 12e⫺

(2.44)

All active-passive metals exhibit such typical S-shaped anodic polarization curves with the exception of titanium, which does not possess a transpassive region. The corrosion behavior of an active-passive metal can be understood from consideration of the mixed electrodes involving the cathodic reduction process. This has been illustrated in Fig. 2.21. The cathodic polarization curve, depending on its exchange current density and Tafel slope, will intersect the anodic polarization curve in any one of the three ways shown as 1, 2, and 3. For case 1, i corr corresponds to A, which is in the active region and is obviously high. For case 2, i corr corresponds to B, C, and D, but C being electrically unstable, the corrosion current assumes either the low value at D or the high value at B depending on whether the metal is in the passive or active state. The unstable passivity achieved by exposing iron to fuming nitric acid gives rise to such a situation. For case 3, where the cathodic curve clears the nose (i cr ), i corr corresponds to i passive and this is the most desirable situation from the viewpoint of corrosion prevention. The attainment of i cr is an important criterion for achieving passivity. An in-

36

Chapter 2

Figure 2.21 Corrosion behavior of an active-passive metal under different reduction reactions.

crease in the rate of cathodic reaction induces a higher rate of dissolution, taking the value to i cr . This is achieved either through the availability of more of oxygen at the metal surface or through the use of oxidizers or by increasing the exchange current density for cathodic reaction. A low value of i cr ensures passivity under relatively low oxidizing conditions and this is achieved by suitable alloying additions. Passivation can also be effected by maintaining the potential in the passive range by using a potentiostat, as described above; this forms the basis of anodic protection of metals (Section 4.5). These factors have been discussed in subsequent sections.

2.4.2

Effect of Environmental Factors

Various environmental factors, such as temperature, pH, and velocity of the corrosive medium, addition of oxidizers or halides to the corrosion medium, and galvanic coupling, influence the passivity vis-a`-vis the corrosion behavior of an active-passive metal. The effects are spectacularly different from those on a metal that does not show active-passive transition. 1.

2.

Temperature and pH. An increase in temperature and increasing acidity of the medium (higher H⫹ concentration or lower pH) increases the value of i cr and decreases the passive potential range as shown in Fig. 2.22. The primary passive potential, E pp , is affected only slightly. Velocity. An increasing velocity of the corrosive medium increases the limiting diffusion current density because of the increasing availability of oxygen. The effect of increasing velocity on the corrosion rate of a normal metal has been illustrated in Fig. 2.18. The corrosion rate increases up to the

Aqueous Corrosion: Fundamentals

37

Figure 2.22 Effect of temperature and hydrogen ion concentration on anodic polarization of an active-passive metal.

situation marked by D, beyond which the rate assumes a steady value. In contrast, for an active-passive metal, an increase in i L up to i cr increases the corrosion rate and beyond i cr the corrosion rate falls drastically, as the intersection of the cathodic polarization curve with the anodic polarization curve takes place only in the passive region (Fig. 2.23). 3. Addition of oxidizers. The effect of addition of oxidizing agents such as ferric, cupric, or chromate ions in the electrolyte on the corrosion behavior of active-passive metals has been illustrated in Fig. 2.24. Additions of increasing amounts of oxidizer shifts the reversible electrode potential in more positive direction in accordance with the Nernst equation: E ⫽ E0 ⫹ 2.3

a RT log oxid nF a red

(2.23)

For the sake of simplicity, it has been assumed that the exchange current density of the redox system and the Tafel slope remain unchanged. The cathodic polarization curves shift progressively upward and in the case of a normal metal an increase in corrosion rate is encountered. In the case of an active-passive metal, however, the corrosion rate increases initially, decreases drastically as the cathodic polarization curves intersect the anodic polarization curve in the passive region, and increases again if the intersection takes place in the transpassive region.

38

Chapter 2

Figure 2.23 (a) Electrochemical behavior of an active-passive metal corroding with a diffusion-controlled cathodic process showing the effect of velocity. (b) Corrosion rate as a function of velocity.

4.

5.

Addition of halides. The presence of halides, especially chloride, in the electrolyte has the similar effect of increasing temperature or increasing acidity in that i cr increases and the passive region is shortened. Halides have a tendency to destroy the passive film and hence in its presence the passivity is achieved with difficulty, i.e., requiring higher i cr . The passivity shows a local breakdown at some potential below the normal transpassive region. This potential is often termed breakdown potential and is associated with the onset of pitting (Section 3.4). Galvanic coupling. Galvanic coupling with a more noble metal increases the corrosion rate of a normal corroding metal (Section 3.3). In the case of metals exhibiting passivity, the situation turns out to be different. Coupling of Pt

Aqueous Corrosion: Fundamentals

39

Figure 2.24 Effect of oxidizer concentration on the corrosion behavior of an activepassive metal.

or Pd with corroding titanium or chromium brings down the corrosion rate drastically, whereas such a coupling with corroding iron enhances corrosion. A similar effect is obtained by noble metal alloying of titanium or chromium where after an initial corrosion the surface is left with islands of noble metals on the base metal matrix causing, in effect, a galvanic coupling. The situation here is effecting an increase in the cathodic exchange current density and its subsequent effect on the polarization behavior, as depicted in Fig. 2.25. The exchange current density for hydrogen evolution on platinum or palladium is much higher than that on titanium or chromium, and the polarization is also sluggish. That is why the cathodic polarization curve initially intersecting the anodic curve in the active region intersects it in the passive region after galvanic coupling and the corrosion rate decreases. In the case of iron, however, the passive region is beyond the redox potential of hydrogen evolution reaction. Therefore, the increased exchange current density and a flatter cathodic polarization curve on galvanic coupling enhances the corrosion rate because of an intersection at a higher current density value in the active region (Fig. 2.26).

40

Chapter 2

Figure 2.25 Passivation of titanium by galvanic coupling with platinum air-free acid solution.

Figure 2.26 Effect of galvanic coupling with platinum on corrosion behavior of an iron in air-free acid solution.

2.4.3

Theories of Passivity

There are two distinct theories of passivity: oxide film theory and adsorption theory. The oxide film theory [3] holds that the passive film is a diffusion barrier layer of reaction products, often an oxide, which separates the metal from the environment. Many metals resist further corrosion with the buildup of an initial corrosion product film, e.g., lead sulfate film on lead immersed in sulfuric acid.

Aqueous Corrosion: Fundamentals

41

However, the films formed on active-passive metals when exposed to highly oxidizing solutions are often very thin and invisible. Though investigators [4] have become successful in identifying γ-Fe 2 O 3 in thin films isolated from the surface of passivated iron, controversy exists regarding the formation of a stoichiometric oxide film during passivation. The adsorption theory [5] holds that the passivity arises out of the chemisorption of oxygen on the metal surface. The adsorption layer may only be monoatomic, but it is effective in providing a kinetic limitation reducing the exchange current density for the dissolution reaction. Chemisorption of oxygen is favored by the presence of uncoupled α electrons in transition metals; since the activepassive metals like Fe, Ni, Cr, and Ti are transition metals, the adsorption theory tends to get support. However, it has also been argued that both theories are supplementary in that the adsorbed film in the process of thickening gradually develops into an oxide film. Studies on the amount of charge required to produce passivation have given information on the thickness of passivating films produced on initially film-free surfaces [6]. Platinum, gold, cobalt, nickel, and iron in alkaline solutions require only a monolayer of oxide for passivation, which has been indicated by a rise of potential at constant current density. Copper and silver require approximately four molecular layers of oxide in alkaline solutions. In acid solutions, silver and lead require films of visible thickness. All of these indicate that the nucleation and growth characteristics of the film are important in deciding how thick it must become before being compact enough to produce passivation.

2.5 CORROSION RATE MEASUREMENT Measurement of corrosion rate is essential for the purpose of material selection. The compatibility of a metal to its environment is a prime requirement for its reliable performance. Corrosion rate measurement may become necessary for the evaluation and selection of materials for a specific environment or a given definite application or for the evaluation of new or old metals or alloys to determine the environments in which they are suitable. Often the corrosive environment is treated to make it less aggressive, and corrosion rate measurement of a specific material in the untreated and treated environments will reflect the efficacy of the treatment. These apart, corrosion rate measurement is also essential in the study of the mechanisms of corrosion.

2.5.1

Corrosion Rate Expressions

Corrosion involves dissolution of metal as a result of which the metallic part loses its mass (or weight) and becomes thinner. Corrosion rate expressions are,

42

Chapter 2

therefore, based on either of these two manifestations, i.e., weight loss and penetration into the metal. The most widely used rate expression based on weight loss is mg/dm 2 /day (mdd) and the rate expressions based on penetration are inch penetration/year (ipy) and mils penetration/year (mpy). One mil is one-thousandth of an inch. The last expression is very convenient as it does not involve a decimal point or zeros. Thus 0.002 ipy is simply expressed as 2 mpy. The expression is readily calculated from the weight loss of the metal specimen during the corrosion test by the empirical formula: mpy ⫽

534W DAt

(2.45)

where W is weight loss, mg D is density of metal, g/cm 3 A is area of specimen, in. 2 t is exposure time, h mdd and mpy are convertible through the following multiplying factors: mpy ⫽

mdd ⫻ 1.44 density

mdd ⫽ mpy ⫻ 0.696 ⫻ density

(2.46) (2.47)

Corrosion rates are often measured electrochemically (discussed later) in terms of current density. However, the corrosion data are rarely presented in terms of current density; they are converted to mdd or mpy. 1 mdd ⫽

2.5.2

1.117n ⫻ 10⫺5 A/cm 2 at. wt.

(2.48)

Exposure Tests

The simplest of corrosion rate measuring tests is the exposure test. A specimen of known initial weight is exposed to the corrosive environment for a specified length of time at the end of which it is taken from the solution, weighed to ascertain the weight loss, and the corrosion rate is then calculated by dividing with the area of specimen and the time of exposure. However, specimens of any size or form cannot be used for the sake of standardization and comparison of data from different sources. Surface preparation also should follow standardized practices, usually a finish with no. 120 abrasive cloth or paper. The details of specimen preparation are provided in ASTM Standard No. G1-88. The exposure of the specimens varies from simple hanging in the test solution

Aqueous Corrosion: Fundamentals

43

in laboratory experiments to elaborate arrangements of encasing them in spools or racks for exposure in process streams. The angle of exposure is also important where velocity is involved. Care should always be taken in terms of electrically insulating the specimens from their metallic holders in order to avoid galvanic effect (Section 3.3). The duration of exposure may vary from hours to years depending on the severity of corrosion. General guidance for the minimum test time is given by the formula: 2000 ⫽ hours (duration of test) mils per year Under many circumstances the corrosion rate varies with time. The initial rate is usually high but it dwindles with the accumulation of corrosion products on the specimen surface. The measured average corrosion rate over a specified length of time therefore fails to indicate the changes of corrosion rate at different stages of exposure. Some planned internal tests have been devised to overcome these drawbacks [7].

2.5.3

Electrochemical Methods

Aqueous corrosion is electrochemical in nature. It is therefore possible to measure corrosion rate by employing electrochemical techniques. Two methods based on electrochemical polarization are available: the Tafel extrapolation and linear polarization. Electrochemical methods permit rapid and precise corrosion rate measurement and may be used to measure the corrosion rate in systems that cannot be visually inspected or subjected to weight loss tests. Tafel Extrapolation Method Mixed-potential theory forms the basis of the Tafel extrapolation method, which is illustrated in Fig. 2.27. The dotted lines represent the anodic and cathodic components of the mixed electrodes involved in the corrosion process, the intersecting point of which corresponds to i corr and E corr . When a corroding specimen is polarized by the applied current, usually cathodic, the experimental polarization curve originates at E corr and at high current densities becomes linear on a semilogarithmic plot. This linear portion coincides with the extended reduction curve as shown by the bold line in the figure. It is clear, therefore, that an extrapolation of the linear portion of the experimental curve will intersect the E corr horizontal at a point that corresponds to i corr . The method is rapid. However, the linear portion should extend over a considerable length, not less than over one order of magnitude, to ensure accuracy in extrapolation. When more than one reduction process is prevailing, the linearity is

44

Chapter 2

Figure 2.27 Tafel extrapolation method of corrosion rate measurement through cathodic polarization.

also affected. These disadvantages are largely overcome in the linear polarization method. Linear Polarization Method It is observed that within 10 mV more noble or more active than the corrosion potential, the applied current density is a linear function of the electrode potential. This is shown in Fig. 2.28. The slope of this linear polarization curve is given by [8]: βaβc ∆E ⫽ ∆i app 2.3(i corr )(β a ⫹ β c )

(2.49)

where β a and β c are Tafel slopes for anodic and cathodic reactions, respectively. The slope has its unit in ohms and is referred to as polarization resistance, R p . Hence the method is also known as the polarization resistance method. Although the linearity of the curve deviates at higher overvoltages, the slope of the curve at the origin is independent of the degree of linearity. The slope of the linear curve is thus seen to be inversely proportional to the corrosion, current, i corr . Assuming β a ⫽ β c ⫽ 0.12 V, Eq. 2.48 reduces to: 0.026 ∆E ⫽ ∆i app (i corr )

(2.50)

One can calculate the corrosion rate from this equation without knowledge of the kinetic parameters. This principle has been utilized in devising commercial instruments for corrosion rate measurement. Such instruments are based on galva-

Aqueous Corrosion: Fundamentals

45

Figure 2.28 Applied-current linear polarization curve for corrosion rate measurement.

nostatic circuitry and have two-electrode or three-electrode configurations. These are now being widely used in chemical process and cooling water streams for corrosion monitoring purposes.

REFERENCES 1. C. Wagner and W. Traud, Z. Electrochem., Vol. 44, p. 391, 1938. 2. J. A. V. Butler, Trans. Faraday. Soc., Vol. 19, p. 729, 1924; T. Erdey-Gruz and M. Volmer, Z. Physik Chem. (Leipzig), Vol. 150, p. 203, 1990. 3. U. R. Evans, J. Chem. Soc., p. 1024, 1927. 4. J. Mayne and M. Pryor, J. Chem. Soc., p. 1833, 1949. 5. H. H. Uhlig, Z. Electrochem, Vol. 62, p. 626, 1958. 6. T. P. Hoar, Corros. Sci, Vol. 7, p. 341, 1967. 7. A. Wachter and R. S. Treseder, Chem. Engg. Progr., Vol. 43, p. 315, 1947. 8. M. Stern and A. Geary, J. Electrochem. Soc., Vol. 104, p. 56, 1957.

This Page Intentionally Left Blank

3 Aqueous Corrosion: Forms

3.1 CLASSIFICATION OF AQUEOUS CORROSION Aqueous corrosion of metals and alloys takes place in a variety of environments containing water. A classification based on the environments has often been described in different textbooks. Thus, the terms atmospheric corrosion, marine corrosion, and soil corrosion are in wide use. While these terms serve to describe certain typical characteristics of corrosion pertaining to an environment, corrosion attacks having common features are not unlikely in different environments. Such a classification is useful in assessing the extent and type of attack with the changing conditions of the environment. For example, atmospheric corrosion of steels decreases drastically at a relative humidity of below 60%, the increase in salinity above a certain level increases localized attack, and so on. Some definite forms or types of attack are evident irrespective of environments and it is convenient to classify corrosion on the basis of these forms. Broadly, corrosion can be classified as general and localized. The latter takes different forms and can be described as (1) galvanic or two-metal corrosion, (2) pitting, (3) crevice corrosion, (4) intergranular corrosion, (5) dealloying or selective leaching, (6) erosion corrosion, and (7) corrosion cracking. Localized attack is an outcome of metallurgical heterogeneities (as in galvanic corrosion, intergranular corrosion, and selective leaching), local aggravation of corrosive conditions (as in pitting and crevice corrosion), and involvement of mechanical factors (as in erosion corrosion and corrosion cracking). Because of its localized nature, attack is often intense leading to the premature failure of metallic components. These different forms of corro47

48

Chapter 3

sion are described in subsequent sections. Biologically influenced corrosion has been dealt with in a separate section.

3.2 GENERAL CORROSION Corrosive attack in many cases is distributed uniformly or, rather, without any preference for sites throughout the surface of the exposed metal. A piece of zinc dissolving in dilute hydrochloric acid (Section 2.1.1), a steel tank or steel roof rusting in atmosphere are examples of such corrosion. The metal thins down uniformly and eventually fails if no protective measure is taken. Atmospheric corrosion is a kind of general corrosion. It takes place under damp and wet conditions. Relative humidity is an important factor in atmospheric corrosion. For iron and steels, rust begins to form above a relative humidity of 60%. Even an invisible thin film of moisture formed on the metal surface can serve as an electrolyte. The aggressiveness of the electrolyte increases with dissolved salts in marine atmospheres or dissolved sulfur dioxide (SO2) in industrial atmosphere. Dust particles can be very detrimental as they absorb water and retain it for a longer time on the metal surface. Chloride-contaminated dust particles may be instrumental in breaking down the protective surface films on metals like aluminum and stainless steels and thus initiate corrosion. Temperature is another important factor. The rate of corrosion increases with the increase of surface temperature, but at a temperature where the electrolyte starts evaporating the corrosion rate falls sharply. The general corrosion in some metals tends to slow down or stop as some protective film develops on the metal surface after the initial attack. Aluminum forms an oxide film in normal atmospheric exposures or in any oxidizing medium. Copper forms a protective surfate ‘‘patina’’ when exposed to industrial atmospheres containing SO2 or a basic copper chloride film in seacoast environments. Lead also forms a protective sulfate film when exposed to SO2-containing atmospheres or waters. The general corrosion may increase with the passage of stray currents from the external sources through the structures (stray current corrosion) or in contact with a nobler metal (galvanic corrosion). These are described in Sections 4.4.3 and 3.3, respectively.

3.2.1

Remedial Measures

The protective measure against general corrosion is often the provision of corrosion allowance in wall thickness in the designing of a tank or pipe. All corrosion rate measurements and corrosion rate expressions are based on the tacit assumption of uniform attack of the metal surface and uniform thinning of the component. So a pipe showing a corrosion rate of 20 mils penetration/year (mpy) in a

Aqueous Corrosion: Forms

49

given environment and with an expected life of 10 years should have 200 mils. i.e., 0.2 in. of additional wall thickness over and above the thickness required for meeting the mechanical requirements of pressure, weight, and stress concentrations. However, since in practice no attack is completely uniform and there is a demand for material conservation, other measures for prevention of corrosion are used as well. For the prevention of uniform corrosion, measures such as the application of protective coatings, use of inhibitors, and use of cathodic or anodic protection are widely adopted. The selection of a suitable material is very important in minimizing general corrosion. Special corrosion-resistant alloys have also been developed of which stainless steels are most important. So far as the atmospheric corrosion of steel is concerned, the performance of special weathering steels containing small amounts of copper (0.3–0.4%) or of microalloyed steels containing copper, phosphorus, nickel, and chromium, particularly in temperate climates, has been commendable. The rust formed on the steels is adherent and protective. The corrosion preventive measures have been discussed in Chapter 4.

3.3 GALVANIC CORROSION When two dissimilar metals are in contact or electrically connected in the same electrolyte, one of them is preferentially corroded whereas the other remains free from corrosion. This localized corrosion of one of the constituent members of a couple is referred to as galvanic corrosion or two-metal corrosion. When placed singularly in the corrosive, both members may corrode, but at different rates. The metal showing a higher corrosion rate becomes the corroding member in the couple, analogous to the anode in a galvanic cell (e.g., in a Daniell cell, Section 2.1.2). The other member becomes cathode. This is essentially a corrosion due to dissimilar electrode cell formation (Section 2.1.3). While galvanic corrosion occurs on the anodic member of the couple, the cathode may suffer from hydrogen damage (Chapter 8). Sometimes nonmetallic conductors may act as cathodes in galvanic couples. Both carbon brick in vessels made of common structural metals and impervious graphite in heat exchanger applications are examples. Conductive films such as mill scale (Fe2O3) or iron sulfide on steel or of lead sulfate on lead are cathodic to the base metal or to some metallic components in their contact.

3.3.1

Prediction of Galvanic Corrosion

The cell potential, which is the algebraic sum of the single-electrode potentials, is the driving force for corrosion reaction. The standard single-electrode potentials of metals have been presented in the emf series (Table 2.1). In a galvanic coupling, the metal having a more noble (i.e., more positive) redox potential acts

50

Chapter 3

as cathode. Coupling of metals wide apart in the series should be avoided as this would produce a large driving force. However, the magnitude of potential difference does not always give the correct prediction of the magnitude of the increase in corrosion rate. For example, zinc coupled to platinum will corrode at a higher rate in an acid medium than when it is coupled to gold, though the position of gold is higher up in the emf series. (The explanation of such a behavior lies in kinetic factors. The exchange current density for hydrogen evolution reaction is greater on platinum than on gold and the resulting icorr is therefore higher.) The situation is described by graphical representation in Fig. 3.1. The importance of area effect in galvanic corrosion should also be well understood. Since the same corrosion current flows through the anodic and cathodic areas in the couple, the current density assumes a higher value on a smaller area. A higher anodic current density essentially means a higher rate of dissolution and so a small anode–large cathode combination should be avoided by all means. Overlooking of this simple rule has led to many hazardous consequences in practice. From the mixed-potential point of view, the area effect can be described by plotting a graph of log (total current) versus potential as shown in Fig. 3.2. A 1-cm2 piece of zinc corroding in an acid solution at a rate of iA corrodes at a higher rate of iB when coupled with a 1-cm2 piece of platinum. In these cases current and current densities are equal because electrodes of unit area have been used. When a 10-cm2 piece of platinum is coupled with a 1-cm2 piece of zinc, the cathode sustains a reduction rate of hydrogen ions 10 times that of the previous case. In other words, the exchange current on this larger cathode will be 10 times the exchange current on the 1-cm2 piece. The intersection of the cathodic polarization curve with the anodic polarization curve in this case yields a corrosion current of iC , which is icorr for the 1-cm2 piece of anode.

Figure 3.1 Corrosion behavior of zinc-gold and zinc-platinum galvanic couples.

Aqueous Corrosion: Forms

51

Figure 3.2 Effect of cathode–anode area ratio on galvanic corrosion of zinc-platinum couples.

In the foregoing discussion the effect of coupling a corroding metal with an inert metal has been discussed. When both metals are corroding in a particular electrolyte, the effect of coupling on their corrosion behavior is described in Fig. 3.3. The metal M is nobler than the metal N. M corroding singularly in the corrosive medium shows a corrosion rate corresponding to icorr(M), whereas the corrosion rate for the metal N is icorr(N). As M and N are coupled, Ecouple shifts to the nobler

Figure 3.3 Corrosion behavior of two corroding metals in a galvanic couple.

52

Chapter 3

direction with respect to the corrosion potential of the metal N and the corrosion rate increases to icorr(M-N). Galvanic coupling may not always increase the corrosion rate of the anodic member. Coupling of titanium or chromium with platinum or palladium drastically reduces the corrosion rate of the active metals through active-passive transition (Section 2.4.2). Polarization behavior of metals also influences galvanic corrosion. For example, titanium shows excellent resistance in seawater, yet the galvanic corrosion of a less resistant metal coupled to titanium does not increase appreciably because titanium cathodically polarizes readily in seawater. The selective attack shows a marked environmental dependence at times. In a copper matrix–tantalum filament composite, the copper matrix has been reported [1] to be selectively attacked after a brief exposure to 70% nitric acid, whereas in 45% hydrochloric acid the tantalum filament has been attacked. The passivation of tantalum in nitric acid and the lack of it in hydrochloric acid explains the difference in behavior.

3.3.2

Galvanic Series

The limitation of the emf series in the prediction of galvanic corrosion because of kinetic factors involved has been discussed in the previous section. There are other limitations as well. The emf series lists the half-cell potentials of metals measured under standard conditions of the electrolytes, i.e., solutions containing unit activity of ions of the respective metal and at a constant temperature. In practical situations, galvanic coupling rarely occurs between metals in equilibrium with their ions. Thus, the actual driving force for galvanic corrosion is quite different from the potential difference calculated from the emf series data. Moreover, the emf series lists only pure metals as it is not possible to establish a reversible potential for alloys containing two or more reactive components. In practice, alloys are widely used and the prediction of galvanic corrosion from the emf series becomes difficult in couples involving an alloy or where both members are alloys. Passivity is another factor that influences the galvanic behavior of a couple as explained in the previous sections. In the passive state, the metal behaves like a nobler metal. The emf series does not provide for the passive state of metals. The galvanic series has been formulated to overcome these difficulties. The galvanic series is an arrangement of metals and alloys according to their actual measured corrosion potentials in a given environment. Since the series is meant for obtaining qualitative information about the tendencies for galvanic corrosion, the measured potentials are not always indicated. Table 3.1 illustrates the galvanic series of commercial metals and alloys in seawater. It should be noted that alloys have been included in the series, as has a nonmetal of electrochemical interest, i.e., graphite. Active-passive metals and alloys occupy different positions in the series. The positions of platinum and gold have

Aqueous Corrosion: Forms

53

Table 3.1 Galvanic series of metals and alloys in seawater Metal Active

Noble

Magnesium Zinc Alchid 3S Aluminum 3S Aluminum 61S Aluminum 63S Aluminum 52 Low-carbon steel Alloy carbon steel Cast iron Type 410 (active) Type 430 (active) Type 304 (active) Type 316 (active) Ni-Resist (corrosion-resisting, nickel cast iron) Muntz metal Yellow brass Admiralty brass Aluminum brass Red brass Copper Aluminum bronze Composition G bronze 90 : 10 Copper-nickel 70 : 30 Copper-nickel—low iron 70 : 30 Copper-nickel—high iron Nickel Inconel, nickel-chromium alloy 600 Silver Type 410 (passive) Type 430 (passive) Type 304 (passive) Type 316 (passive) Monel, nickel-copper alloy 400 Hastelloy, alloy C Titanium Graphite Gold Platinum

54

Chapter 3

been interchanged with respect to their positions in the emf series. To minimize galvanic corrosion, the rule of thumb is to avoid the coupling of metals and alloys far apart in the series. The galvanic series in seawater is widely used to predict the galvanic behavior in other environments as well. However, this should be done with caution. Changes in electrolyte composition and temperature can significantly change the electrode potential (see Nernst equation) and hence cause a change in position in the galvanic series. Tin has been shown as nobler than iron in Table 3.1. In ‘‘tin cans,’’ however, certain food constituents combine chemically with Sn2⫹ ions to form soluble tin complexes. This lowers the activity of Sn2⫹ ions, shifting the potential of tin to more active direction so that iron may become cathodic. Ideally, there should be a galvanic series for each environment, but this is not practicable because it would require an infinite number of tests. Even then it may not help in a quantitative prediction of galvanic corrosion.

3.3.3

Sacrificial Anode

The fact that the cathodic member in the galvanic couple remains free from corrosion is utilized to protect a structure or component by making it cathode. This is accomplished by coupling the structure or coating it with a less noble metal. The anode protects the structure by sacrificing its life through preferential dissolution, hence the name ‘‘sacrificial anode.’’ A ship hull made of steel is protected by insertion of magnesium blocks at places. Such a protection is referred to as cathodic protection, which is discussed in detail in Section 4.4. Galvanizing is a common protective measure for steel parts. A layer of zinc is provided on steel by hot dipping in molten zinc. The layer is never perfect. Also, local breakdown or cracking of the layer occurs due to mechanical damage. At these discontinuities the corrosive comes in contact with both steel and zinc. Zinc being the anode dissolves, leaving the steel intact. Although the anodic area is large, the rate of dissolution is slow and this ensures a reasonable coating life as well. Coating steel with a more noble metal, say tin, would have provided a reverse situation of small anode (steel)–large cathode (tin) at the discontinuities, inviting a rapid attack on steel (Fig. 3.4). However, it may be noted that in some

Figure 3.4 Galvanic corrosion at discontinuities in tin- and zinc-coated steel.

Aqueous Corrosion: Forms

55

systems polarity of a couple can reverse. For example, iron-zinc couple in certain domestic waters reverses polarity at temperatures over 80°C. Cathodic product formed on zinc apparently causes this reversal.

3.3.4

Remedial Measures

1. Avoid coupling of metals far apart in the galvanic series. 2. Small anode–large cathode combinations should be avoided. Fasteners should be cathodic to the parts being fastened. 3. Whenever coating is applied on a galvanic couple, the cathodic member is to be coated, and not the anodic member. This is because any discontinuity in the coating on anodic member will only provide an unfavorable anode/ cathode ratio. 4. Whenever possible, insulation should be provided between dissimilar metals. 5. Anodic parts may be made thicker to ensure a longer life, e.g., water heaters with copper tubes and heavy steel tube sheets. 6. A third metal anodic to both metals of the couple may be installed.

3.3.5

Practical Examples

Practical examples of galvanic corrosion are varied and widespread. Rapid corrosion of steel rivets and bolts is commonplace when these are used to fasten copper or Monel plates. Contact with graphite packing has led to the failure of steel pump shafts or valve stems. Aluminum wire used as replacement of copper wire for connecting the auxiliary anode (high-silicon cast iron) in an impressed current cathodic protection system suffered rapid failure. The incidence of galvanic corrosion at times is triggered by unexpected sources. Dissolved copper from upstream may plate on downstream steel components forming in situ galvanic couples. Chance contact of a vernier fuel line made of aluminum alloy with helium pressurization line made of 301 stainless steel in a liquid-fueled missile led to a leak formation in the aluminum pipe in a 16month exposure to a marine environment [2]. The failure of an automatic water sprinkler system was due to incompatible metal combination and unfavorable area ratio [2]. The system had a copper plate made of a cast copper alloy that was held in position by a clapper latch made of malleable iron (Fig. 3.5). The malleable iron lip at the contact point had been corroded to such an extent in only 21 months of service that it failed by plastic deformation. Replacement of the latch material by silicon bronze prevented galvanic corrosion and the latches were reported to be in satisfactory condition after more than 14 years of service.

56

Chapter 3

Figure 3.5 Schematic illustration of the sprinkler system in which the malleable iron clapper latch failed from galvanic attack.

3.4 PITTING This form of corrosion is associated with the formation of pits, i.e., small holes or cavities with surface diameter equal to or less than the depth. The depth eventually increases, leading to a thorough perforation or a massive undercut in the thickness of the metallic part. The width of the pit may also increase with time, but not to the extent to which the depth grows. Most often the pit opening remains covered with the corrosion product, making it difficult to detect during inspection. This, along with a negligible loss in weight or absence of apparent reduction in the overall wall thickness, gives little idea about the extent of damage. Pitting may thus lead to an unpredictable leakage in a pipeline or storage tank or the rupture of a pressure pipe due to the reduction in wall thickness as a result of extensive undercutting. Pits may also assist in brittle fracture, fatigue failure, environment-assisted cracking like stress corrosion cracking, and corrosion fatigue by providing sites of stress concentration.

3.4.1 1. 2. 3. 4.

Characteristic Features

The attack is spread over small discrete areas. Pits are sometimes isolated and sometimes close together, giving the area of attack a rough appearance. Pits usually initiate on the upper surface of the horizontally placed parts and grow in the direction of gravity. Pitting usually requires an extended initiation period before visible pits appear. Pitting is autocatalytic in nature. The conditions prevailing inside the pit makes it self-propagating without any external stimulus. Thus, once initiated, the pit grows in an ever-increasing rate.

Aqueous Corrosion: Forms

57

5. Stagnant solution conditions lead to pitting, whereas even a very susceptible material remains free from pitting in an aggressive medium in flowing condition. 6. Stainless steels and aluminum and its alloys are particularly susceptible to pitting. Carbon steels are more resistant to pitting than stainless steels. Most failure for stainless steels occurs in neutral-to-acid chloride solutions. Aluminum and steels pit in alkaline chloride solutions. 7. Most pitting is associated with halide ions, with chlorides, bromides, and hypochlorites being particularly aggressive. Cupric, ferric, and mercuric halides are extremely aggressive because their cations are cathodically reduced and sustain the attack.

3.4.2

Evaluation of Pitting Damage

Weight loss methods are inappropriate for pitting damage evaluation because the negligible loss in weight is not indicative of the seriousness of damage. Extent of pitting damage can be characterized by pit density, surface size, and depth with the help of standard rating chart provided in Standard Practice G46-76 of American Society for Testing and Materials (ASTM) (Fig. 3.6). For example, a pitted specimen having average spacing (surface density) of 1 ⫻ 104 /m2, surface diameter of 8 mm2, and pit depth of 3.2 mm will be characterized as A-2, B-3, C-4. However, for the sake of component life prediction and reliability pit depth, rather the maximum pit depth, is of significance. Maximum pit depth increases with time and with sample size. The probability of finding the pit of a certain depth is higher as the specimen size becomes larger. Several methods are adopted to measure the pit depth: metallographic, machining, micrometer depth gauge, and microscopic. The first two are destructive. In the metallographic method, the sample is sectioned and polished through a selected pit followed by microscopic measurements. The method is tedious and uncertainty prevails in respect to selection of the deepest pit. Machining out to a depth where no evidence of pit remains obviously will require samples of regular shape. In the method using a micrometer, readings between surface and pit bottoms are compared with a needle probe; in the microscopic method, calibrated fine focus is used to determine depth difference between surface and pit bottoms. In both cases, pits must have a large opening for the insertion of the needle probe or for the light to reach the bottom of the pit. Both fail in proper assessment if the pit is directionally oriented or undercut. A special ultrasonic thickness testing has been reported [3] that is capable of determining the pit depth. In this technique the pit is used as an acoustic lens to ensure strong front surface and back surface reflection (Fig. 3.7). The ultrasonic transducer is centered over the pit center by maximizing the front surface signal arrival time. This is accomplished by using a microcomputer that is also capable

58

Chapter 3

Figure 3.6 ASTM standard rating chart for evaluation of pitting corrosion. of acquiring ultrasonic waveforms through the pit center, and analyzing waveforms for thickness determination. The ratio of the maximum metal penetration to average metal penetration as determined by weight loss method has been termed pitting factor, which is sometimes used to quantify pitting attack. A pitting factor of unity represents uniform attack.

Aqueous Corrosion: Forms

59

Figure 3.7 Focused transducer for pit depth measurement. The curvature of the pit collimates the diverging beam in the steel plate resulting in a strong back surface reflection.

3.4.3

Mechanism

Any mechanism proposed for pitting should explain its discrete initiation and autocatalytic propagation. Again, since pitting is most prevalent in the presence of chloride ions and occurs mostly in active-passive metals, the mechanism described should involve these systems. It has been observed that pitting initiates at a critical pitting potential, Epit, which is illustrated in Fig. 3.8. The anodic polarization curve of an active-passive alloy, say 304 stainless steel, shifts to higher current density values with chloride in the medium and the passivity breakdown with accompanying large increase in current density is observed at some intermediate potential in the passive region. This potential is referred to as Epit. In a potentiodynamic procedure, if the direction of polarization is reversed after some degree of anodic polarization above Epit, the return polarization curve follows an active path and a hysteresis is observed. Its intersection with the passive region is referred to as protection potential, Eprot, below which existing pits cannot grow. At intermediate potentials between Epit and Eprot, a new pit cannot initiate, but a pit initiated above Epit can grow. A more positive value of Epit implies more resistance to pitting. Chloride ions are held responsible for breakdown of passivity and initiation of pitting. Chloride ions adsorb on the outer side of the passive film [4] and the adsorption intensifies at potentials near Epit, apparently because of a stronger electrostatic attraction between the negatively charged chloride ions and an in-

60

Chapter 3

Figure 3.8 Schematic representation of cyclic polarization to determine pitting potential, Epit , and protection potential, Eprot .

creasing positively charged substrate. In the absence of chlorides, the passive film dissolves slowly, according to: FeOOH ⫹ H2O → Fe3⫹ ⫹ 3OH⫺

(3.1)

where FeOOH represents the passive film. Chloride ion catalyzes the liberation of Fe3⫹ according to the reactions [4]: FeOOH ⫹ Cl⫺ → FeOCl ⫹ OH⫺ FeOCl ⫹ H2O → Fe3⫹ ⫹ Cl⫺ ⫹ 2OH⫺

(3.2) (3.3)

Accumulation of relatively thick chloride salt ‘‘islands’’ has been observed on the surface of iron at the critical pitting potential [5] and FeOCl approximates the composition of the salt islands. These reactions proceed to remove the passive film at a preferred site until a pit is initiated through direct anodic dissolution to Fe2⫹. The preferred sites may arise from a surface scratch, from an emerging dislocation or other defects, or from random variations in solution

Aqueous Corrosion: Forms

61

Figure 3.9 (a) Slow dissolution of passive film forming Fe3⫹. (b) Accelerated dissolution at a soluble salt island forming Fe3⫹. (c) Direct anodic dissolution at a pit initiation site forming Fe2⫹.

composition. The sequence of events is described in Fig. 3.9. All initiated pits are not stable and many become inactive after some growth. However, some get stabilized. A stabilized pit starts growing according to the following mechanism as illustrated in Fig. 3.10. Since oxygen in the stagnated solution inside the pit is used up in the cathodic reduction process within a short period, anodic dissolution concentrates there with accompanying cathodic reduction of oxygen just outside the pit. A situation of small anode and large cathode prevails. Additionally, as metal cations accumulate inside the pit and the chloride ions rush to the pit promoting the hydrolysis reaction: Fe2⫹ ⫹ 2H2O ⫹ 2Cl⫺ → Fe(OH)2 ⫹ 2HCl

(3.4)

the hydrochloric acid produced aggravates the dissolution further. The propagation thus becomes self-stimulating and along a narrow front. In a way, the sides of the pits are cathodically protected. The requirement of a concentrated solution for the initiation and propagation of a pit is met preferably by a stagnant condition of the solution as well as its accumulation on the upper side of the component. That is why pits tend to grow

62

Chapter 3

Figure 3.10 Growth of a corrosion pit.

more in the direction of gravity. In 304 stainless steel, chromium and nickel provide additional hydrolysis reactions similar to Eq. 3.4 and more of acidic condition is produced. Thus the 304 stainless steel is more susceptible to pitting damage than the carbon steels. However, stainless steels containing higher amounts of nickel and chromium or with molybdenum addition (type 316) show high resistance to pitting, apparently because of a more stable passive surface produced by these alloying additions.

3.4.4 1. 2. 3.

4.

5.

Remedial Measures

Solution aggressiveness can be reduced by eliminating chlorides or by decreasing its concentration. Stagnation of the solution should be avoided. Addition of passivating inhibitors helps in reducing or eliminating pitting attack. However, they should be used in sufficient amount to ensure complete passivation. Otherwise, the attack will aggravate (‘‘dangerous inhibitor,’’ Section 4.2.3.1). Cathodic protection, through sacrificial anode or by impressed current, has been found to prevent pitting in marine applications (i.e., 304 stainless steel propellers). However, this may not work out in more aggressive solutions. Using a metal or alloy of higher resistance to pitting is often a practical solution, but with increase in investment cost which may prove to be economical in the long run. A stainless steel with 2–3% molybdenum (type 316) is

Aqueous Corrosion: Forms

63

Figure 3.11 Pitting attack on a type 304L stainless steel pump component.

far superior to type 304 in pitting resistance in seawater. Nickel alloys like Hastelloy or Chlorimet are even better. Titanium and its alloys show excellent pitting resistance in the most aggressive media, like those containing FeCl3 or CuCl2, and are often recommended for such applications.

3.4.5

Practical Examples

Figure 3.11 illustrates pitting attack on a pump component made of type 304L stainless steel experienced in a medium comprising hydrochloric and nitic acid mixtures [6]. Temperature and concentration are unknown. The damage marks on the surface inflicted during material handling apparently provided the nucleation sites for pitting. The lateral broadening of pits with time is also indicated. Figure 3.12 shows the subsurface enlargement of one of the pits in a type 321 stainless steel aircraft freshwater storage tank that failed in service because of leak formation as a result of pitting. The tank was sterilized with sodium hypochlorite solution, but the rinsing was delayed for 68 h after the draining out of the solution. Apparently, a small amount of the solution remained in the bottom of the tank causing the damage. The tank eventually failed after only 10 h of service.

3.5 CREVICE CORROSION A crevice is a small gap created by contact of a material with another material. The crevice area of a metal or alloy in contact with another material, metal or

64

Chapter 3

Figure 3.12 Subsurface enlargement of pit in a type 321 stainless steel aircraft freshwater storage tank (95⫻).

nonmetal, tends to get corroded preferentially in a corrosive environment compared to the area outside the crevice. This type of localized attack is known as crevice corrosion. Examples of crevices are lap joint, areas under bolts and rivets, screw threads penetrating a metal or wood, area under a rubber gasket, and areas under dirt or corrosion debris. The attachment of barnacles or other biofouling organisms in marine applications provides crevice areas. The terms gasket corrosion and deposit corrosion are also used to describe crevice corrosion in the latter cases. Crevice corrosion may take place on any metal and in any corrosive environment. However, metals like aluminum and stainless steels, which depend on their surface oxide film for corrosion resistance, are particularly prone to crevice corrosion, especially in environments containing chloride ions, such as seawater. The gap defining a crevice is usually large enough for the entrapment of a liquid but too small to permit flow of the liquid. The width is of the order of a few thousandths of an inch, not exceeding 1/8 in. (3.18 mm).

3.5.1

Mechanism

To start with, the areas inside the crevice and outside it undergo corrosion in the same manner in a corrosive environment. In a neutral chloride solution the anodic dissolution is supported by the cathodic reduction of oxygen: Anodic

M → Mn⫹ ⫹ ne⫺

(2.6)

Aqueous Corrosion: Forms Cathodic

O2 ⫹ 2H2O ⫹ 4e⫺ → 4OH⫺

65 (2.11)

As the reactions proceed, the dissolved oxygen in the small volume of stagnated solution inside the crevice is used up. However, this does not prevent the dissolution reaction inside the crevice because the electrons reach outside the crevice through the metal where plenty of oxygen is available for reduction. A sort of concentration cell (differential aeration) is set up between the crevice area and the area outside the crevice (Section 2.1.3). The situation is further aggravated in the presence of chloride ions. The accumulated cations inside the crevice attract the negatively charged chloride anions from the bulk solution. Hydroxide anions also migrate, but they are less mobile than chloride ions. The metal chloride formed hydrolyzes to produce metal hydroxide and hydrochloride acid: MCl ⫹ H2O → MOH ⫹ HCl

(3.5)

The nascent hydrochloric acid destroys the passive film and accelerates the rate of dissolution of the metal inside the crevice. The cathodic reduction remains restricted to the areas outside the crevice that remain cathodically protected. Figure 3.13 depicts the happenings during the initial and later stages of crevice corrosion.

Figure 3.13 Crevice corrosion. (a) Initial stage. (b) Later stage.

66

Chapter 3

3.5.2 1.

2.

3. 4. 5. 6.

Remedial Measures

Crevice corrosion can be controlled by proper design to avoid crevices. Welded butt joints should be preferred to riveted or bolted joints. Crevices in existing lap joints may be closed by welding or caulking. Tanks and vessels should be designed to ensure complete drainage, thus preventing solid deposits from forming on the bottom of the vessel. If the tank is placed on a masonry platform, closing of the gap along the periphery with tar or bitumen is advocated to avoid seepage of rainwater. Regular inspection and removal of deposits should be emphasized. Impervious gasket materials are preferable to porous ones. Also, the removal of wet packing materials during long shutdown periods is necessary. Solution aggressiveness may be reduced, where possible, by decreasing the chloride content, acidity, and temperature. Alloys resistant to pitting are usually also resistant to crevice corrosion. Where practicable, a vulnerable material may be substituted by a more resistant material. Increased chromium, nickel, molybdenum, and nitrogen increase the resistance to crevice corrosion of stainless steels. Type 316 stainless steel containing 2–3% Mo is fairly resistant. Nickel alloys are more resistant than stainless steels. Titanium alloys, however, are prone to crevice corrosion in halide solutions above 70°C.

3.5.3

Filiform Corrosion

Filiform corrosion is a special type of crevice corrosion sometimes encountered under protective coatings on metals. The attack manifests itself in the form of thread-like filaments spreading in a zigzag manner. The filaments are 0.1 to 0.5 mm wide, grow steadily, but, interestingly, do not cross each other. Filaments show an active head and an inactive tail. If an advancing head meets another filament, it gets reflected from there and starts growing in another direction. The reflected filaments at times enter into a ‘‘death trap’’ for not being able to cross the inactive tail and with the decrease of the available space for growth. An example of filiform corrosion is shown in Fig. 3.14. Filiform corrosion has been observed on steel, aluminum, zinc, and magnesium, usually under organic coatings like paints and lacquers. Attack under tin, enamel, and phosphate coatings is also experienced. The attack does not damage the metal to any great extent, but the coated surface loses its appearance. On steel, the head of the filament is usually blue and the tail is red-brown, indicating the presence of Fe2⫹ ions in the head and Fe2O3 or Fe2O3⋅nH2O at the tail as corrosion product. The growth mechanism of filiform corrosion is explained by the formation of a differential aeration cell and is illustrated in Fig. 3.15. The head absorbs water from the atmosphere because of the presence of a relatively

Aqueous Corrosion: Forms

67

Figure 3.14 Filiform corrosion under clear varnish on steel (10⫻).

concentrated solution of ferrous salts and hydrolysis creates an acidic environment (i.e., pH 1–4). Oxygen that diffuses through the film tends to accumulate more at the periphery of the solution, i.e., at the interface between the head and the tail. Lateral diffusion of oxygen serves to keep the main portion of the filament cathodic to the head. Filiform corrosion can be prevented by reducing the relative humidity of the environment to below 65%. Protection can also be achieved by the use of films of very low water permeability.

Figure 3.15 Schematic cross-section of filiform corrosion illustrating the growth mechanism.

68

Chapter 3

Figure 3.16 Crevice corrosion of stainless steel in seawater due to barnacles attachment.

3.5.4

Practical Examples

The practical examples of crevice corrosion or deposit corrosion are quite commonplace. Deposits of dirt and leaves on tin roofs often lead to leak formation. Similarly, a welding debris inadvertently left on the inside wall of a stainless steel digester filled with demineralized water produced leakage during a shutdown period. Rapid attack with red rust developed under a stainless steel bolt that had fallen to the bottom of a 18-8 stainless steel tank containing a hot saline solution in a dyeing plant has been reported [7]. Flange faces underneath gaskets, particularly fibrous gaskets having a wick action, are prone to attack by crevice corrosion. Figure 3.16 illustrates crevice corrosion of stainless steel in seawater as a result of barnacles attachment to the surface [8].

3.6 INTERGRANULAR CORROSION Intergranular corrosion is a preferential attack on the grain boundary phases or the zones immediately adjacent to them. Little or no attack is observed on the main body of the grain. The continuous attack along the grain boundary can readily be detected under microscope in a metallographically polished and etched specimen (Fig. 3.17). The surface of the affected stainless steel part shows visible rusting. The alloy loses its mechanical strength like a brick wall having its mortar degenerated. The alloy tends to disintegrate and in the extreme case the grains fall out. Since the grain boundary region is an area of crystallographic mismatch between the orderly structures within the adjacent grains, it is chemically slight-

Aqueous Corrosion: Forms

69

Figure 3.17 Photomicrograph illustrating intergranular corrosion in 316 stainless steel.

ly more active than the grain interior. This is reflected in the controlled etching of the metallographic specimens whereby the grain boundaries are revealed. An ‘‘overetching’’ equalizes the attack. However, under certain conditions the grain boundaries remain very reactive and under corrosive conditions the attack along the grain boundary is sustained resulting in intergranular corrosion (IGC). The factors that contribute to the increased reactivity of the grain boundary area are as follows: 1. Segregation of specific elements or compounds at the grain boundary, as in aluminum alloys or nickel-chromium alloys 2. Enrichment of one of the alloying elements at the grain boundary, as in brass, and 3. Depletion of the corrosion-resisting constituent at the grain boundary, as in stainless steels All of these factors contributing to intergranular corrosion have their origin in the thermal processing of the materials, such as welding, stress relief, and other heat treatments.

3.6.1

Austenitic Stainless Steels

Intergranular corrosion of austenitic stainless steels poses a big industrial problem. Many failures have occurred in welded components exposed to environ-

70

Chapter 3

ments normally unaggressive toward these stainless steels. This deterioration of corrosion resistance of stainless steels on welding has been termed ‘‘weld decay.’’ However, welding is not the only causative factor for intergranular corrosion. Austenitic stainless steels become susceptible to intergranular corrosion when heated in the temperature range of about 500°C to 800°C. The material is then said to be sensitized. The extent of sensitization effect is a function of both time and temperature. Exposure to temperatures near the middle of this range for a few minutes is equivalent to several hours near the upper and lower limits. However, the limits are not fixed and they are influenced by the composition, particularly the carbon content, of the steel. Figure 3.18 shows the sensitization diagram for type 304 stainless steel for various carbon compositions. The steel does not sensitize if the carbon content is below 0.02%. The higher the nickel content of the alloy, the shorter is the time for sensitization to occur at a given temperature. Molybdenum has the opposite effect. Mechanism Depletion of chromium in the grain boundary areas is usually the cause of intergranular corrosion in austenitic stainless steels. In the sensitizing range, carbon that is almost completely dispersed throughout the alloy rapidly diffuses to the grain boundaries where it combines preferentially with chromium to form chromium carbide, Cr23C6. The adjusting areas thus get depleted of chromium. As chromium diffuses much more slowly at these temperatures, the loss is not made up and the chromium content at the grain boundaries drops to a value much lower than 12%, which is the minimum required to maintain passivity. In these areas, the stainless steel behaves no better than ordinary steel.

Figure 3.18 Sensitization diagram for 18Cr-8Ni stainless steels of varying carbon content.

Aqueous Corrosion: Forms

71

Active-passive cells of appreciable potential difference are set up between the chromium-depleted alloy at the grain boundary and the chromium-rich grain body. An unfavorable large cathode/small anode ratio leads to the rapid attack of the grain boundary in corrosive environments. The chromium carbide precipitated at the grain boundary, which has a leaf-like structure, is not attacked; rather, as a cathodic constituent it aggravates the attack in the adjacent areas. The affected volume of the alloy extends over a small distance on either side of the grain boundary and the boundary thus appears to be broadened when viewed under the microscope. In alloys of low carbon content (below 0.02%), sensitization is absent as a sufficient amount of carbon is not available to produce a continuous chromiumdepleted grain boundary. The precipitated carbides are sparse and remain dispersed without affecting the corrosion resistance of the alloy. In the normal grades of austenitic stainless steels having 0.06–0.08% C, sensitization can be avoided if the alloy is rapidly cooled through the sensitizing range as carbon does not get time to reach the grain boundaries or react there with chromium. Austenitic stainless steels, quenched from 1050°C, undergo slow intergranular attack in strongly oxidizing media like boiling 5 N HNO3 with added oxidizing ions such as Cr6⫹ or Mn7⫹. The attack is attributed to the segregation of specific impurities to the grain boundary during quenching. Weld Decay and Knife-Line Attack The sensitization of austenitic stainless steels during welding is known as weld decay. The affected zone is usually a band in the parent plate somewhat removed from the weld bead. When exposed to corrosive environments, intergranular corrosion takes place in this zone and the attack gives a granular appearance (Fig. 3.19). During welding, this area remains in the sensitizing temperature range for a sufficient length of time resulting in chromium carbide precipitation, whereas at weld pool and its adjacent areas the temperature is high and time insufficient for the precipitation to take place. Areas further remote from the weld decay zone do not attain the sensitizing temperature and remain unsensitized. The exact location of the affected zone depends on the metallurgical history, plate thickness, and rate of heat input and cooling. Since temperature and time combinedly contribute to sensitization, a prolonged welding operation or slow cooling after welding induces susceptibility. Arc welding is therefore preferred to gas welding for stainless steels. Spot welding is still preferable as the heating is intense and rapid, followed by a rapid cooling. Knife-line attack (KLA) is yet another type of intergranular corrosion encountered in stabilized stainless steels (see section ‘‘Remedial measures’’) on welding. A narrow band adjacent to the weld is the zone of attack. The thermal history of the material for KLA is different from that for weld decay. Titanium carbide and niobium carbide precipitate at temperatures higher than

72

Chapter 3

Figure 3.19 Weld decay of type 304 stainless steel in 25% HNO3 solution (2⫻).

that for chromium carbide precipitation. For example, niobium carbide precipitates in the temperature range of 815°C to 1230°C where chromium carbide dissolves. During stabilization (i.e., cooling down from melt), niobium carbide forms in this temperature range leaving no carbon to form chromium carbide at temperatures below 815°C. However, above 1230°C, niobium carbide dissolves. So, when a stabilized steel is heated above this temperature all carbides get dissolved. A rapid cooling to room temperature prevents the carbides from reprecipitating. This is exactly the situation faced by the area just adjacent to the weld.

Aqueous Corrosion: Forms

73

If now the weldment is heated to 500–800°C for stress-relieving purposes or in service, chromium carbide forms and sensitization follows. Since niobium carbide needs a higher temperature to form, the presence of niobium in the alloy cannot prevent chromium carbide formation. The same situation prevails for the titanium-stabilized variety. Remedial Measures Three effective means are available for avoiding intergranular corrosion or weld decay in austenitic stainless steels. They are as follows: 1. Solution annealing. The sensitized components are heated to a temperature above 815°C (usually 1000°C) followed by quenching in water. The hightemperature treatment dissolves the carbides and quenching prevents their reformation. While this treatment can be given to small components, it is not practicable for large components or structures. 2. Reduction of carbon content. Austenitic stainless steels with low carbon content (⬍0.03%), e.g., 304L, 316L, etc., can be welded or heated in the critical range without sensitization problems. However, grease or oil should be cleaned from the surface of such steels before welding to prevent carbon pickup during welding. 3. Stabilization. Austenitic stainless steels are available in ‘‘stabilized’’ grades having alloying additions of Ti (type 321) or Nb⫹Ta (type 347). These elements have a higher affinity for carbon than chromium and when added in requisite amounts they react with carbon to form various carbides at temperatures higher than 815°C. Chromium is prevented from carbide formation at the sensitizing temperatures as no more carbon is available.

3.6.2

Ferritic Stainless Steels

Like austenitic stainless steels, ferritic stainless steels also exhibit sensitization and intergranular corrosion by a chromium depletion mechanism. However, there are several differences between the two. First, the solubility of nitrogen being low in austenitic stainless steels, precipitation of chromium nitride is not the prime factor for sensitization. In ferritic stainless steels, this contributes significantly. The second difference is the temperature range of sensitization. In ferritic stainless steels this lies above 925°C where the solubility of carbon and nitrogen becomes significant in ferrite. Because of this difference in the sensitizing temperature range, the zone of intergranular corrosion also differs. In ferritic steels the attack takes place at areas adjacent to the weld or in the weld itself. The immunity to intergranular corrosion is restored by heating the sensitized steel between 650–815°C for a short time. As the diffusion of chromium in the body centered cubic (bcc) lattice of ferritic steel is significantly higher than in austenitic steel, the uniform chromium composition is reestablished. Reduction of interstitials, both carbon and nitrogen, provides an effective step

74

Chapter 3

towards the reduction of susceptibility, but in order to achieve complete immunity this is to be lowered to very low value. For example, for 18Cr-2Mo alloys to be immune to intergranular corrosion, the maximum level of carbon plus nitrogen is 60–80 ppm. Ferritic stainless steels stabilized with titanium or niobium have also been developed.

3.6.3

Tests for Sensitization

Special tests have been designed to detect the susceptibility of stainless steels to intergranular corrosion. These serve as acceptance tests for the material for construction or repair and also indicate the actual metallurgical condition of the material in the evaluation of a plant process. Both chemical and electrochemical tests are available. The sulfuric acid–copper sulfate test, or Strauss test (ASTM A708-86), involves exposure of the specimen in a boiling 16% H2SO4 ⫹ 6% CuSO4 solution for 72 h and evaluation by bending of the specimen over a mandrel and examining macroscropically for fissures or peeling caused by intergranular corrosion. Since the rate of attack in this test is low, an accelerated test has been devised by coupling the stainless steel specimen to a copper sheet (ASTM A262-86, Practice E). The Huey test (ASTM A262-86, Practice C) consists of exposure of specimens to boiling 65% nitric acid for five 48-h periods and corrosion rates are calculated. Accepted rates for different materials have been specified. The test is time consuming. The Streicher test (ASTM 262-86, Practice B), or oxalic acid etch test, is a rapid test. It consists of etching the 3-zero emery paper finished specimens in 10% oxalic acid for 1.5 min under an applied current density of 1 amp/cm2 and then examining the surface at a magnification of 250–500⫻. The specimen acts as an anode and a stainless steel beaker as cathode. A ‘‘step’’ structure indicates a nonsensitized condition, whereas a ‘‘ditch’’ structure indicates susceptibility to intergranular corrosion. Figure 3.20 illustrates these features. Electrochemical tests involve anodic polarization of the specimen. Potentiostatic polarization shifts the passive region of the polarization curve progressively toward higher current density value with increasing sensitization. A difference of one to two orders of magnitude in the current density value with respect to nonsensitized specimen depending on the degree of sensitization is indicated. The electrochemical potentiokinetic repassivation (EPR) technique is based on a reverse or reactivation potential scan from a potential back to the corrosion potential. Either hot concentrated sulfuric acid or a solution of 0.5 M H2SO4 ⫹ 0.01 M KSCN at 30°C is employed. A polished stainless steel sample of controlled surface area is passivated at ⫹2 V (SCE) for 2 min. Potential is then scanned in the active direction at a rate of 6 V/h, down to the corrosion potential, where no current is indicated. Sensitized material shows a current peak that in-

Aqueous Corrosion: Forms

75

Figure 3.20 (a) Step structure from electrolytic-oxalic acid etch test in solution-annealed type 304 stainless steel (400⫻). (b) Ditch structure in sensitized steel (400⫻).

creases with degree of sensitization. No such peak is observed for unsensitized material. The difference in behavior is shown schematically in Fig. 3.21.

3.6.4

Other Alloys

Precipitation-hardenable nickel alloys like Inconel X-750 are susceptible to intergranular corrosion in hot caustic solutions, in boiling 75% nitric acid, and in high-temperature water containing low concentrations of other salts. Solid solution–type alloys like Inconel 600 are also susceptible in the same media when thermally treated in the range of 540–760°C, which gives rise to the grain boundary precipitation of carbides. In Alloy C, grain boundary precipitation of molyb-

76

Chapter 3

Figure 3.21 Schematic representation of procedure for electrochemical potentiokinetic reactivation (EPR) study to determine sensitization.

denum carbide causes molybdenum depletion and susceptibility to intergranular attack in hot reducing acids as well as in HNO3. Reduction of carbon and silicon has imparted improved resistance. Hastelloy B and C become susceptible when heated in the range of 500–705°C in which immunity can be restored by suitable heat treatment at higher temperatures (1150–1240°C). Copper alloy 260 (70–30 brass) corrodes intergranularly in dilute aqueous solutions of H2SO4, Fe2(SO4), BiCl3, and other electrolytes. Enrichment of grain boundary region in zinc through segregation is considered to be the cause of attack. Segregated iron at the grain boundary makes the high-purity aluminum susceptible to intergranular corrosion. The precipitated phases in the high-strength alloys are the cause for intergranular attack. For example, CuAl2 precipitation in Duralumin-type alloys causes copper depletion and a substantial potential difference is set up between the depleted area and the adjacent material. In 5000 and 7000 series alloys, precipitated phases like FeAl3, Mg5Al8, MgZn2, etc., at the grain boundary are anodic with respect to the grain matrix. Solution heat treatment eliminates the susceptibility to intergranular corrosion at the cost of loss in strength. Attack along the boundaries of grains elongated in the rolling direc-

Aqueous Corrosion: Forms

77

tion in high-strength aluminum alloys in marine and industrial environments causes some blocks of grains to lift up from the metal surface because of the pressure of the voluminous corrosion product. Such attack is traditionally referred to as exfoliation corrosion. Exfoliation may be minimized by the use of extended aging cycles for Al-Cu alloys, use of organic and sprayed metal coatings, by avoiding graphite-bearing lubricants that act as cathode, and by promoting an equiaxed grain structure at the surface or throughout the alloy. Titanium and some of its alloys show intergranular corrosion in red fuming nitric acid at room temperature. Addition of 1% NaBr inhibits attack. Commercially pure titanium is intergranularly attacked in methanol-halide solutions. A small addition of water acts as inhibitor.

3.6.5

Practical Examples

The brake pedal support plate made of Al-8Si-3.5Cu alloy from an aircraft refueling tanker truck fractured in service [9]. The contaminating liquid was found to be alkaline, containing borax. Metallographic examination revealed an intergranular attack along the needle-like copper-rich phase (Fig. 3.22). Figure 3.23 shows two impellers made of 53Ni-16Cr-16Mo alloy put in service in 25% hydrochloric acid with 100 ppm chlorine at ambient temperature [6]. The impeller on the left was solution-annealed and the one on the right was sensitized before being placed in service. The sensitized one shows severe intergranular corrosion with grain dropping occurring at the tip of the blades.

Figure 3.22 Intergranular corrosion along the needle-like copper-rich phase of an AlSi-Cu alloy casting.

78

Chapter 3

Figure 3.23 Intergranular corrosion and its absence respectively, in a pump impeller in sensitized (right) and solution-annealed (left) condition.

3.7 SELECTIVE LEACHING Selective leaching is the removal of one of the components of the alloy by corrosion. It is usually the less noble component, e.g., zinc in brasses. The selective removal of zinc from brass is known as dezincification, which is the most common example of selective leaching. Preferential dissolution of iron from gray cast iron has been named graphitic corrosion. The phenomenon also takes place in some other alloy systems. Selective leaching is the general term to describe these processes. Table 3.2 lists the combinations of alloys and environments where selective leaching is encountered, some of which belong to high-temperature corrosion.

3.7.1

Dezincification

Dezincification occurs in brasses containing more than 15% zinc. It may take place either uniformly all over the surface, called layer type, or in localized area on the metal surface, called plug type (Fig. 3.24). Both are easily recognized by the naked eye because of the coppery appearance of the attacked areas in contrast to the normal yellow color of brasses. Sections of the affected areas, viewed under microscope, reveal porous or spongy structure. The strength of the material is impaired and the layer-type attack may lead to rupture of pipe with an increase in water pressure; the plug type may lead to hole formation. The layer-type attack is more prevalent in alloys of high zinc content; lowzinc-containing alloys favor plug-type attack, although exceptions have been reported. The chemistry of the environment seems to have a greater effect in determining the type of attack. Uniform attack is produced by slightly acidic water,

Aqueous Corrosion: Forms

79

Table 3.2 Combinations of alloys and environments for selective leaching Alloy Aluminum Bronzes Brasses Cupronickels Gray iron Gold alloys High-nickel alloys

Environment

Element removed

Hydrofluoric acid, acid chloride solutions

Aluminum

Many waters High heat flux and low water velocity (in refinery condenser tubes) Soils, many waters Nitric, chromic and sulfuric acids, human saliva Molten salts

Zinc Nickel

Iron-chromium alloys Medium- and high-carbon steels Monel

High-temperature oxidizing atmospheres

Nickelmolybdenum alloys Silicon bronzes Tin bronzes

Iron Copper or silver Chromium, iron, molybdenum, tungsten Chromium

Oxidizing atmospheres, hydrogen at high temperatures

Carbon

Hydrogen and other acids

Oxygen at high temperatures

Copper in some acids, nickel in others Molybdenum

High-temperature steam, acidic solution Hot brine, steam

Silicon Tin

low in salt content and at room temperature. Plug-type attack is favored in neutral and alkaline water, high in salt content and above room temperature. Dezincification gets accelerated under the conditions of high temperatures, stagnant solutions, and porous inorganic scale formation. Crevice conditions under a deposit or scale tend to aggravate the situation. Mechanism Two different mechanisms have been suggested for dezincification: (1) simultaneous dissolution of the components and redeposition of copper as a porous layer, and (2) selective dissolution of zinc from the lattice leaving behind a porous copper-rich structure. The selective dissolution mechanism envisages the following steps: 1. Anodic dissolution of zinc, which may proceed in pure water even in the absence of oxygen with the cathodic reduction of water to hydrogen and

80

Chapter 3

Figure 3.24 (a) Layer-type (uniform) and (b) plug-type dezincification of brass pipe.

2. 3.

hydroxyl ions (Eq. 2.11). Vacancies, particularly divacancies, are created by the anodic dissolution. Diffusion of vacancies into the interior of the alloy, thereby aiding in diffusion of zinc. Diffusion of zinc in the opposite direction.

Objections to this mechanism have been raised as the diffusion of zinc seems to be too slow to account for the relatively rapid rate of penetration observed. The redeposition mechanism seems to derive support from the beneficial effect of alloying additions of Sn, As, Sb, and P against dezincification in α brass. Apparently, these elements are redeposited on the alloy as a film and thereby

Aqueous Corrosion: Forms

81

hinder deposition of copper. However, since the dezincification of β brass is not prevented by such alloying, the question remains regarding the mechanism. More recent studies indicate that both processes occur in separate but overlapping potential regimes. Remedial Measures 1. Use of a more resistant alloy is the most practical approach to minimize the occurrence of dezincification as the control of environment is not often feasible or economical. Red brass with less than 15% zinc is almost immune. Improvement in the performance of yellow brass (70–30) has been achieved with the addition of 1% tin (admiralty brass) and small amounts of arsenic, antimony, or phosphorus. Cupronickels provide better substitute in severely corrosive environments. 2. Removal of stagnation of corrosive, particularly if acidic. 3. Periodic removal of scales and deposits from the inside surface of pipelines. 4. Use of cathodic protection.

3.7.2

Dealloying of Other Copper Alloys

Selective leaching of aluminum takes places in copper-aluminum alloys (aluminum bronze) when exposed to hydrofluoric acid or acids containing chloride ions. The alloys containing more than 8% Al are particularly susceptible; these are two-phase alloys containing a copper-rich α phase and an eutectoid of α phase and an aluminum-rich γ phase. The γ phase is anodic to α and is preferentially attacked. Dealloying of nickel in copper-nickel alloys has been observed in refinery condenser tubes under the condition of high heat flux and low water velocity at temperatures above 100°C. Selective leaching of tin in tin bronzes in hot brine or steam and desiliconization of silicon bronzes in high-temperature steam containing acidic species have been reported.

3.7.3

Graphitic Corrosion

Selective leaching of iron in gray iron is termed graphitic corrosion. This is observed in gray iron pipes buried in soil or manhole covers exposed to mildly corrosive waters and in similar appliances. Iron leaches out selectively, leaving the cathodic interconnected graphite flakes as a porous mass on the metal surface (Fig. 3.25). There is little change in metal thickness, but the entire structure is rendered weak and tends to give way to fluid pressure from within or to impact load from outside. Graphitic corrosion does not occur in ductile iron or malleable iron because in these varieties graphite does not form a network to hold the metal residue. The term ‘‘graphitization’’ is sometimes used to describe this phenomenon,

82

Chapter 3

Figure 3.25 Graphitic corrosion of gray cast iron showing the residual skeleton of graphite flakes and phosphide eutectic.

but this should be avoided because the term refers to graphite formation through breakdown of Fe3C in steels on prolonged heating.

3.7.4

Practical Examples

The micrograph of dezincified area experienced in an alloy 260 (70–30 brass) pipe for domestic water supply is shown in Fig. 3.26 [2]. Area A shows plug-type attack on the nickel-chromium-plated exterior surface where untreated municipal water supply leaking from the faucet packing gland ran down the pipe and attacked the brass through a break in the plating caused by mechanical damage. The attack on the interior surface of the pipe was a fairly uniform layer-type dezincification. It may be noted that in the area shown only about one-third of the original wall has remained as sound metal. Figure 3.27 shows the fragmented parts of an aluminum bronze seawater pump impeller that were retrieved from the casing of the pump [9]. The impeller was missing; apparently, it had disintegrated in service. Sectioning of the fragments revealed the porous structure caused by dealuminification from both sides. Weakening of the impeller through dealloying was established as the cause of failure. Solution heat treatment minimizes dealuminification, but this particular impeller was in as-cast condition.

3.8 EROSION CORROSION The term erosion applies to deterioration due to mechanical factors. When the factors contributing to erosion accelerate the rate of corrosion of a metal, the attack is called erosion corrosion. A corrosive fluid, aqueous or gaseous, flowing

Aqueous Corrosion: Forms

83

Figure 3.26 Micrograph showing difference in dezincification of inside and outside surfaces of a plated 70-30 brass pipe for domestic water supply. (A) Plug-type attack of the outside surface. (B) Layer-type attack of the inside surface.

over the metal surface or impinging on it provides the usual condition for erosion corrosion. However, the mechanical deterioration may be aggravated by sheer presence of a corrodent as in the case of fretting corrosion or corrosive wear. Sometimes movement of environment decreases the rate of attack, e.g., pitting of stainless steels in flowing seawater as compared to pitting in stagnant seawater, but this is not erosion corrosion because the attack is not increased. The attack takes the form of grooves, i.e., scooped-out, rounded areas, horseshoe-shaped depressions, gullies, waves, all of which often showing directionality. Sometimes the attack appears as an assembly of pits. Ultimate perforation due to thinning or progress of pits and rupture due to the failure of the thinneddown wall to sustain the internal fluid pressure are also common. All sorts of equipment exposed to flowing fluid are subject to erosion corrosion among which piping systems and heat exchangers are most common.

3.8.1

Environmental Factors

The environmental factors that affect erosion corrosion are: velocity, turbulence, impingement, presence of suspended solids, temperature, and prevailing cavitation conditions. The acceleration of attack is due to destruction or removal of the protective surface film by mechanical forces exposing fresh metal surfaces that are anodic to the uneroded neighboring film. The nature of the surface film is important. A hard, dense, adherent, and continuous film such as on stainless

84

Chapter 3

Figure 3.27 (a) Fragments of a seawater pump impeller that failed by dealuminification. (b) Section through a fragment showing the depth of dealuminification.

Aqueous Corrosion: Forms

85

steels is more resistant than a soft, brittle film as on lead. The nature of the protective film depends largely on the corrosive itself. In most metals and alloys corrosion rate increases with increase in velocity, but a marked increase is experienced only when a critical velocity is reached. For example, an admiralty brass (70Cu-29Zn-1Sn) showed a corrosion rate of 1–2 mdd in seawater in the solution velocity range of 1–4 ft/s, and above 300 mdd at the solution velocity of 27 ft/s. However, stainless steels, nickel alloys such as Monel and Hastelloy, and titanium show a remarkably low corrosion rate even at this high velocity. Their critical velocities are of still higher value. Turbulence is caused when the liquid flows from a larger area to a smalldiameter pipe as in the case of the inlet ends of tubings in heat exchangers. Internal deposits in the pipes or any obstruction to the flow inside a pipe by a foreign body, such as a carried-in pebble, can also cause turbulence. The liquid hits the wall of the pipe more vigorously and a rapid corrosion ensures. Impingement, i.e., direct hitting of the corroding fluid on the metal surface, occurs at bends, elbows, and toes in a piping system and causes intense attack. Impingement is also encountered on surfaces of impellers and turbines, in areas in front of inlet pipes in tanks, and in many other situations. The attack appears as horseshoe-shaped pits with deep undercut and the end pointing in the direction of flow (Fig. 3.28). The corrosive effect of high velocity, turbulence, and impingement is further aggravated when the solution contains solids in suspension and at higher temperatures. Rapid attack on type 316 stainless steel was reported to have been experienced in a sulfuric acid–ferrous sulfate slurry moving at a velocity of 39 ft/s. The rate of corrosion increased to about 4500 mpy, whereas the laboratory tests in stagnant solution at room temperature indicated a nil corrosion rate. Lead resists attack by static dilute sulfuric acid at all temperatures up to the boiling point of the solution, but at high velocity of the solution the corrosion rate increases steadily with a rise in temperature (Fig. 3.29). Steam carrying water condensate droplets provides an aggressive medium for erosion corrosion of steel and cast iron pipes. The impingement of water droplets at the return bends destroys the protective oxide film and accelerates the attack on the substrate.

Figure 3.28 Longitudinal section through condenser tube showing impingement attack by seawater in the direction of flow (7⫻).

86

Chapter 3

Figure 3.29 Erosion corrosion of lead by 10% sulfuric acid flowing at 39 ft/s. Crossmarks (*) on abssica indicate the results of tests in static medium.

3.8.2

Metallurgical Factors

Soft and low-strength metals like copper, aluminum, and lead are especially susceptible to erosion corrosion. So are the metals and alloys that are inherently less corrosion-resistant, like carbon steels. Stainless steels of all grades are, in general, resistant to erosion corrosion and their performance improves with increasing nickel, chromium, and addition of molybdenum. In iron-chromium alloys of lower grades, the resistance progressively increases with increase in chromium content. Stainless steels and chromium steels are resistant because of their tenacious protective surface films. In other alloy systems also, development of a more stable protective film improves their resistance to erosion corrosion, e.g., addition of iron to cupronickels or aluminum to brasses. As a rule, solid solution alloys provide better resistance than alloys hardened by heat treatment because the latter are heterogeneous in structure. Cast irons usually perform better than steel. Alloy cast irons containing nickel and chromium show better performance. Durinon, containing 14.5% Si, gives excellent performance under severe erosion corrosion conditions.

3.8.3

Remedial Measures

Improvement in Design Erosion corrosion can be minimized or eliminated altogether by change of design, i.e., the shape and geometry of the components, and also by increasing the wall

Aqueous Corrosion: Forms

87

thickness at the vulnerable areas. Impingement attack can be avoided by smoothing the bends in the piping systems. Increasing the diameter of the pipe will ensure a laminar flow and less turbulence. Provision of baffles or barriers will restrict the impact of fluid. Design should also provide for the easy replacement of the affected parts. These are only a few examples. A well-regarded modification of the design has in many instances proven to be the most economical solution to erosion corrosion problems. Selection of Materials As discussed in the previous section, inherently resistant materials may be used to replace the frequently affected parts. However, galvanic corrosion must also be considered if such replacement forms part of an integrated system involving components made of more than one metal or alloy. Alteration of Environment Lowering of solution corrosivity through control of pH, inhibitor addition, and lowering of temperature can have beneficial effects. Cathodic Protection Provision of sacrificial anodes as a component or additional attachment may be effective in some cases.

3.8.4

Cavitation Damage

When the conditions of velocity are such that repetitive low- and high-pressure areas are developed, bubbles form and collapse at the metal–liquid interface. This phenomenon is called cavitation and the damage to the metal caused by it is known as cavitation erosion or cavitation damage. A pressure drop develops when a high-velocity liquid flows across a curved surface. A pressure drop below the vapor pressure of the liquid causes local boiling and bubble formation. The bubbles collapse as the bulk of liquid falls on them. The collapse of vapor bubbles produces shock waves with high pressures sufficient to dislodge metal particles or destroy the protective film. If the liquid is corrosive, rapid attack takes place on the exposed metal until the protective film is restored. The sequence is shown schematically in Fig. 3.30. The repetition of the process leads to a cluster of pit formation resembling the appearance of a honeycomb. Eventually, perforation may take place. The trailing faces of ship propellers and pump impellers are particularly vulnerable to cavitation damage and many failure cases have been reported. Cavitation damage is also encountered on rotors of pumps, water turbine blades, and on the water-cooled side of diesel engine cylinders. Cavitation damage can be minimized by the proper choice of materials. Titanium alloys, austenitic and precitation-hardening stainless steels, nickel-chro-

88

Chapter 3

Figure 3.30 Schematic representation of sequence in cavitation damage.

mium alloys, and nickel-chromium-molybdenum alloys are useful under extremely severe conditions. Nickel-copper, copper-nickel, and other copper alloys in general perform better than carbon and low-alloy steels. Hard facing of the surface with a resistant alloy is beneficial. In an opposite approach, resilient coatings of rubber and other elastomers have been successfully used to combat cavitation damage. They essentially reflect the shock waves and reduce the damage on metals. Aeration of water or hydrogen produced on cathodic protection serves to cushion the shock waves. As smooth surfaces do not provide sites for bubble nucleation, smooth finishes on pump impeller or ship propellers are advocated. Damage can be reduced by operating rotary pumps at the highest possible head of pressure to avoid bubble nucleation. In closed systems like the cooling water in diesel engine cylinder liners, use of appropriate inhibitors like the addition of 2000 ppm sodium chromate has proved effective.

3.8.5

Corrosive Wear and Fretting Corrosion

Wear is a surface phenomenon that occurs by displacement and detachment of materials. Corrosive wear refers to the aggravation by corrosion of the wear produced by the hard projection of a mating surface or with hard particles moving relative to the wearing surface. The chemical reaction may take place first followed by the removal of corrosion products by mechanical abrasion. Mechanical action may also precede chemical action in which the small particles dislodged by abrasion subsequently react with the environment. In both cases, the wear rate enhances.

Aqueous Corrosion: Forms

89

Fretting is also a wear phenomenon that occurs between two mating surfaces under loading and having a relative slip of extremely small amplitude. The slip is usually oscillatory as, for example, that caused by vibration. Under such conditions, asperities, i.e., the minute protrusions of one surface ‘‘plough’’ through the mating surface dislodging metallic particles or breaking the protective film (Fig. 3.31). Fretting corrosion is the aggravation of this phenomenon in the presence of a corrosive liquid. Here also, like corrosive wear, either a mechanism of oxidation–wear or wear–oxidation operates. In fretting corrosion the damage is characterized by discoloration of the metal surface and formation of pits. Fatigue cracks may eventually nucleate at the pits. In ferrous materials, fretting corrosion creates debris of reddish brown ferric oxide particles. The consequence of fretting corrosion is loss of dimensional accuracy, loosening of the parts, sometimes seizure of the parts because of the accumulation of corrosion products, and, at times, fatigue failure. The magnitude of damage increases with increasing load, but decreases with increasing temperature and increasing moisture, indicating that the mechanism is not fully electrochemical. Fretting and fretting corrosion are encountered in joints, connecting rods, shrink fits, oscillating bearings, splines and couplings, and in many parts of vibrating machinery. Fretting often occurs in bearings and other contact points during shipping of components and equipment. Any measure that reduces wearing action would minimize fretting corrosion. Lubrication of contacting surfaces is beneficial. That is why increased moisture during rainy season provides a lubricating effect to reduce the extent of damage. Reducing the load is effective, but at the same time, increasing the load and thus reducing relative slip is also helpful. Separation of surfaces by insertion of a material of high elastic strain limit, such as rubber or Teflon, prevents fretting. In press-fitted assemblies, induction of residual stresses by shot peening or other

Figure 3.31 Fretting action at a metallic surface with oxide layer.

90

Chapter 3

treatments is helpful to prevent fatigue crack propagation initiated by fretting. Elimination or reduction of vibration would be ideal to reduce relative slip, but often this measure is not practicable.

3.8.6

Practical Examples

Leakage was detected in a malleable iron elbow after only 3 months in service, whereas the life expectancy was 12–24 months (Fig. 3.32) [2]. The component was exposed to alternate steam and cooling water supply for at least 16 h a day. Investigations revealed a ferritic malleable iron structure, which was less resistant to erosion corrosion than pearlitic malleable iron. A proper heat treatment improved the performance. Figure 3.33 illustrates a cavitation damage of a pump impeller made of 55Ni21Cr-3Mo-3.5Cu-1W alloy working in a water contain 550 ppm chloride at 44°C [6]. Closely spaced pits are clearly visible. It may be noted that the material used belongs to the group of most erosion corrosion–resistant materials. Change in operating conditions of the pump was recommended as a remedial measure.

3.9 CORROSION CRACKING Many metals and alloys develop cracking under the combined action of stress and corrosion. Corrosion-induced cracking processes are categorized as: 1. 2. 3.

Stress corrosion cracking, Corrosion fatigue, and Hydrogen-induced cracking.

Figure 3.32 Section through the malleable iron elbow showing the area of erosion corrosion attack.

Aqueous Corrosion: Forms

91

Figure 3.33 Cavitation degree of a pump impeller showing a honeycomb type of attack.

Stress corrosion cracking (SCC) involves static tensile stress, whereas corrosion fatigue takes place under the conditions of cyclic or alternating stresses with a tensile component. Hydrogen-induced cracking (HIC) also involves tensile stresses, but the cracking process is distinctly related to the entry of atomic hydrogen into the metal. Cathodic reduction of hydrogen ions in a corrosion process is one of the sources of atomic hydrogen. Since hydrogen-induced cracking falls under the broad category of hydrogen damage, this will be discussed in a subsequent chapter (Chapter 8).

3.9.1

Stress Corrosion Cracking

Stress corrosion cracking (SCC) is defined as the delayed failure of alloys by cracking when exposed to certain environments in the presence of static tensile stress. The importance of a conjoint action of corrosion and stress is reflected in the definition; an alternate application of stress and corrosive environment will not produce SCC. The stress level at which the failure occurs is well below the stress required for a mechanical failure in the absence of corrosion. The minimum stress below which SCC is not encountered is called threshold stress (Fig. 3.34), but this may be as low as 10% of the yield strength in some systems. Again, corrosion alone in the absence of stress does not cause SCC. The earliest report on SCC is probably the occurrence of ‘‘season cracking’’ in brass cartridge cases in ammonia-bearing environments in the beginning of the present century. Caustic embrittlement of riveted steel boiler plates is another classical example of SCC encountered in the early steam-driven locomotives.

92

Chapter 3

Figure 3.34 Stress–time to fracture relationship for stress corrosion cracking of 0.8% C steel in sodium acetate and acetic acid buffer solution showing the threshold stress.

The cause of SCC in this case has been the residual stress developed due to riveting operations and concentrated sodium hydroxide present as corrosive in these areas. Many alloy–environment combinations leading to SCC have been discovered with the passage of time. It is a matter of concern that almost all alloys of engineering interest are susceptible to this type of catastrophic degradation in some environment or other, the only silver lining being that all alloys are not susceptible to SCC in all environments. There is some specificity in this regard. Earlier, the pure metals were considered to be immune to SCC. Though pure metals have been induced to crack under severe test conditions in the laboratory, practical instances are absent. On the other hand, their alloys show ready and rapid cracking. A list of alloy–environment systems exhibiting SCC is presented in Table 3.3. It should, however, be remembered that the table is not exhaustive; newer combinations and newer environments causing SCC in a particular alloy are being regularly discovered and are being added to this list. It is, therefore,

Aqueous Corrosion: Forms

93

Table 3.3 Alloy–environment combinations exhibiting stress corrosion cracking Alloy Aluminum alloys

Carbon steels

Copper alloys

Nickel alloys

Stainless steels: Austenitic

Austenitic (sensitized )

Ferritic Martensitic Titanium alloys

Environment Air with water vapor Potable waters Seawater NaCl solutions NaCl-H2O2 solutions Caustic NaOH solutions Calcium, ammonium, and sodium nitrate solutions HCN solutions Acidified H2S solutions Anhydrous liquid ammonia Carbonate/bicarbonate CO/CO2 solutions Seawater Ammoniacal solutions Amines Nitrites Caustic alkaline solutions High-temperature chloride solutions High-purity steam Hydrofluoric acid Acidic fluoride solutions Hot acid chloride solutions NaCl-H2O2 solutions NaOH-H2S solutions Seawater Concentrated caustic solutions Neutral halides: Br⫺, I⫺, F⫺ Polythionic acids (H2SnO6) Sulfurous acid Pressurized hot water containig 2 ppm dissolved oxygen H2S, NH4Cl, NH4NO3 Hypochlorite solutions Caustic NaOH solutions Red-fuming nitric acid Hot salts, molten salts N2O4 Methanol/halide

94

Chapter 3

advisable to test the compatibility of a given alloy to a new environment where it may be expected to perform. Sources of Stress The existence of a tensile stress, either applied or present as a residual stress, is essential for the cracking process. In order for SCC to occur a threshold stress or stress intensity (see section ‘‘Testing Methods’’) must be exceeded. Residual stresses of high magnitude, 70% of the yield strength of the material or even higher, may be produced by cold-forming operations like bending, rolling, deep drawing, etc., or by thermal processing like welding, solidification of casting with large differences in section thickness, severe quenching processes, etc. Often design stress in service is sufficiently high to precipitate the occurrence of SCC. The presence of stress raisers may aggravate the situation. These stress raisers may be the geometrical stress raisers or notches related to design, inclusions, rough machining marks, welding strikes, and so on, and also may result from localized corrosion attack such as pitting, selective leaching, or intergranular corrosion. Exposure of metal parts to high and low temperatures results in nonuniform heating rates and sharp thermal gradients which, in turn, may give rise to thermal stresses of very high magnitude. This is quite common in heat exchangers. If the environment on the cold side of the heat exchanger is conducive to stress corrosion, cracking is produced by these thermal stresses. Corrosion products accumulating in fissures or tightly jointed parts may produce wedging action providing sufficiently high stress for SCC. General Features of Stress Corrosion Cracks Stress corrosion environments being usually mildly corrosive from the viewpoint of general attack, the surface of the stress-corroded part exhibits only faint signs of corrosion while fine cracks penetrate deeply into the part. The mode of cracking can be intergranular or transgranular (Figs. 3.35 and 3.36). The cracks proceed in a direction perpendicular to the stresses and often show branching. One type of cracking usually occurs more readily in a given alloy. For example, carbon steels exhibit intergranular mode of cracking in nitrate solutions, whereas austenitic stainless steels crack transgranularly in boiling MgCl2 solutions. However, mixed mode of cracking may also be observed; depending on environment, alloy composition, cold work, or metallurgical conditions, there can be a transition from one type of cracking to the other. For example, α brass cracks intergranularly in ammoniacal solutions, but β brass (⬎40% Zn) cracks transgranularly in the same environment. Again, α brass shows transgranular cracking if heavily cold-worked or in highly alkaline ammoniacal solutions. Austenitic stainless steels in sensitized condition crack intergranularly. Stress corrosion cracks have the appearance of a brittle mechanical fracture.

Aqueous Corrosion: Forms

95

Figure 3.35 Intergranular stress corrosion cracking in 70Cu-30Zn brass.

The brittle nature of fracture, whether intergranular or transgranular, is quite evident in scanning electron microscope (SEM) fractography (Fig. 3.37). Testing Methods The SCC testing methods can be categorized as 1. 2. 3. 4.

Constant strain or constant deformation tests Tests on statically loaded smooth samples Tests on statically loaded precracked samples Slow strain rate testing (SSRT)

Figure 3.36 Transgranular stress corrosion cracking in a type 304 stainless steel in boiling MgCl2 solution.

96

Chapter 3

Figure 3.37 SEM fractographs of (a) transgranular SCC of austenitic stainless steel in hot chloride solution and (b) intergranular SCC of sensitized type 316 stainless steel in polythionic acid.

Constant strain tests make use of U-bend, C ring, bent beam specimens or tensile specimens in a small loading frame with nuts on the threaded ends of the specimen (Fig. 3.38). The specimens in the stressed condition are exposed to the stress-corroding medium and checked periodically for the appearance of cracks. Although the stress distribution within the specimen cannot be precisely known, such tests are most simple for the qualitative evaluation of resistance of various alloys to SCC to different solution conditions. Time to failure, tf , is the parameter used to measure the resistance to SCC. Tests on statically loaded smooth samples are conducted at various fixed stress levels, and the time to failure is measured. A dead-weight load is hung from one end of the specimen, directly or by means of a lever arrangement, whereas the specimen is simultaneously exposed to the test solution contained in a glass cell. A threshold stress below which the time to failure approaches infinity is encountered in some systems, as shown in Fig. 3.34, whereas in some other systems it is not observed. Time to failure, tf , is essentially made up of two components: the time for crack initiation (ti ), and the time for crack propagation, tp, so that ti ⫹ tp ⫽ tf

Aqueous Corrosion: Forms

97

Figure 3.38 Constant strain testing for stress corrosion cracking: (a) U bend; (b) C ring; (c) bent beam; (d) tensile.

Different metallurgical and electrochemical variables affect tf , but it is difficult to ascertain these effects separately on ti and tp. In smooth specimens usually ti ⬎⬎ tp, whereas in practical situations a pit or a surface roughness feature acts as an already initiated crack and the question of its propagation under different variables assumes more importance. The use of precracked specimens, notched or fatigue-precracked, and the application of linear elastic fracture mechanics (LEFM) technique in stress corrosion crack propagation have evolved as a consequence. Tests on statically loaded precracked samples are usually conducted with a constant applied load and the velocity of crack propagation as a function of stress intensity factor, k, is measured. The value of k is calculated from k ⫽ σC 1/2, where σ is the applied stress and C is the crack length. Figure 3.39 gives the schematic representation of a typical da/dt versus k plot. Three regions in the plot are identified as stage I, II, and III. No crack propagation is observed below some threshold stress-intensity level kISCC . In stage 1, the crack propagation rate increases rapidly with the stress-intensity factor. In stage 2, the crack propagation rate approaches some constant velocity, referred to as the plateau velocity, which is characteristic of the alloy–environment combination and is a result of ratelimiting diffusion of the reactant species to the crack tip. The effect of variations in composition, changes in heat treatment, electrochemical variables, and changes to the environment gets reflected in the plateau velocity and such data become helpful in alloy selection. Alloys that have a low-plateau velocity can be chosen

98

Chapter 3

Figure 3.39 Schematic relationship between stress intensity factor and crack growth rate for stress corrosion cracking.

in preference to the alloys with a high-plateau velocity under the similar conditions of the environment. It has been shown [10] that high-strength titanium alloys that were thought to be immune to SCC in dilute aqueous chloride environments on the basis of smooth specimen stress corrosion tests are in fact highly susceptible when evaluated using fatigue-precracked specimens. Actually, these alloys are resistant to pitting in these environments and stress corrosion crack initiation through pitting does not take place. However, preexisting flaws and defects make these alloys susceptible to SCC. In slow-strain rate testing, a smooth or a precracked specimen exposed to the corrosive environment is pulled at a low cross-head speed (10⫺5 to 10⫺9 m/s) to failure. The elongation to failure (or any other tensile property such as reduction in area, ultimate tensile strength, or fracture energy) is plotted against strain rate, as shown in Fig. 3.40. The plot shows that a narrow range of strain rate exists where the ductility is minimum, which is indicative of SCC. At higher strain rates film formation, which is important in the initiation of a stress corrosion crack, cannot keep pace with the mechanical plastic strain. At very low strain rates, the ruptured film is healed before a stress corrosion crack can be initiated through an intense corrosive attack at the rupture sites. In both cases, the pulled sample fractures in a ductile manner. The slow strain rate technique has the advantage over the constant strain or

Aqueous Corrosion: Forms

99

Figure 3.40 Schematic presentation of the effect of strain rate on SCC and hydrogeninduced cracking.

constant load test methods in that the experiment can be completed in a short time (2 days maximum) to ascertain the susceptibility to SCC in an alloy in a given environment. However, the results may not be always indicative of SCC in a similar environment in service where a forced, continuous strain is absent. Metallurgical Aspects of SCC Virtually all alloys are susceptible to SCC in the appropriate environments. Susceptibility to SCC is affected by the chemical composition of the alloy, size and preferential orientation of grains, composition and distribution of precipitates, dislocation interactions, and the progress of phase transformation. Bulk alloy composition can affect passive film stability and phase distribution (e.g., chromium in stainless steel), minor alloying elements can cause local changes in passive film–forming elements (e.g., carbon in stainless steel causing sensitization), impurity elements can segregate to grain boundaries and cause local difference in the corrosion rate (e.g., phosphorus in nickel or nickel-base alloys), and inclusions can cause changes in the local crack–tip chemistry as the cracks intersect them (e.g., manganese sulfide in steel). The effect of bulk alloy composition on SCC susceptibility is best exemplified in stainless steels and copper alloys. Figure 3.41 shows the effect of nickel content on SCC in stainless steels exposed to boiling 42% magnesium chloride solution, a common SCC medium used in the laboratory to simulate the cracking in hot

100

Chapter 3

Figure 3.41 Effect of nickel content on SCC of iron-chromium-nickel alloys in boiling 42% MgCl2 solution.

chloride environments. The beneficial effect of high nickel content is clearly visible. However, a ferritic stainless steel devoid of nickel, which is generally resistant to SCC in most common service environments that attack austenitic stainless steels, becomes susceptible with the addition of 1.5% to 2% nickel. Duplex stainless steels with high chromium and low nickel content are more resistant to SCC than austenitic stainless steels. Pure copper, which is almost immune to SCC in ammoniacal environments, becomes readily susceptible with the alloying of 1% silicon or 0.2% phosphorus. Brasses with a zinc content below 15% are resistant to SCC. Addition of a third element like Si, Al, to 70Cu-30Zn brass changes the mode of cracking from intergranular to transgranular. Grain boundary precipitation and grain boundary segregation play a major role in intergranular SCC. Austenitic stainless steels, heated in the temperature range of 500–850°C, develop a chromium-depleted zone along the grain boundary through precipitation of chromium carbide, a phenomenon known as sensitization (Section 3.6.1). Susceptibility to intergranular SCC and the crack growth rate are related to the degree of sensitization. Intergranular SCC of sensitized austenitic stainless steels is a recurring problem in cooling-water piping of boiling-water nuclear power plants. The Ni-Cr-Fe alloy 600 also gets sensitized through chromium depletion at the grain boundary and has been reported to fail by intergranular SCC in high-temperature pressurized water. Grain boundary pre-

Aqueous Corrosion: Forms

101

cipitation has been identified as a contributing factor in the intergranular SCC of aluminum alloys. The 7000 series alloys (Al-Zn-Mg-Cu) are most susceptible in the peak-aged condition; the susceptibility is reduced with overaging. Grain boundary segregation of alloying elements or impurities has been identified as the causative factor of intergranular SCC in many alloys. Grain boundary enrichment of magnesium in Al-Mg alloys accounts for the increased anodic activity or possible formation of magnesium hydride along the grain boundaries. Grain boundary enrichment of impurities such as phosphorus, sulfur, carbon, and silicon contributes to the intergranular SCC of iron-base alloys, austenitic stainless steels, and nickel-base alloys. The enrichment of impurities in the grain boundaries can be as high as 50% within a region 1–2 nm thick, facilitating the propagation of a stress corrosion crack along the grain boundary. The effect of variation of carbon on the SCC of very low carbon steels is shown in Fig. 3.42. Carbon segregation at the grain boundary is considered to provide suitable imperfection sites for adsorption of nitrates to promote SCC, and at a very low level of bulk carbon content such segregation is not attainable. It has been shown experimentally [11] that about 0.01% carbon was required to cause SCC in nitrate or caustic environments, which is above the room temperature solubility limit. Phosphorus segregation has been shown by several authors to promote intergranular SCC of low-alloy steels (Cr-Mo or Ni-Cr-Mo-V) in caustic or water environments at relatively oxidizing potentials [12–14]. Phosphorus segregation has also been identified as a contributory factor in the intergranular SCC of austenitic stainless steels containing less than 0.002% carbon and also in nickel alloy 600 [15]. Grain boundary segregation of impurities is deemed responsible for the intergranular SCC of pure iron [16]. In transgranular SCC, alloying effect on slip planarity is a major factor. A number of crack growth models have been proposed based on the planar sliplocalized corrosion processes. Planar slip occurs in alloys with low stacking-fault energy, alloys containing ordered phases, or alloys exhibiting short- or long-range ordering. The additions of nickel to stainless steels [17] and manganese to copper [18] have been reported to develop planar array of dislocations due to the lowering of stacking-fault energy and consequently a susceptibility to transgranular SCC. Alloys, on the other hand, are prone to dealloying. It has been suggested [19] that the dealloyed layer acts as a cleavage crack initiator in brass, copper, gold, and stainless steels. Electrochemical Aspects of SCC A number of environmental parameters influence SCC susceptibility, in terms of both crack initiation and crack growth, and the cracking mode in an alloy. Important among these parameters are temperature, nature of solute species and its concentration, pH, and the electrochemical potential. Although the ‘‘specificity’’ of an environment is not strictly valid, still only a few environments can induce

102

Chapter 3

Figure 3.42 Effect of carbon content in very low carbon steels quenched from 920°C on SCC in a calcium nitrate–ammonium nitrate solution.

SCC in any particular alloy system and the susceptibility varies significantly with potential and pH. The electrochemical mechanisms of SCC (see next section) visualize the initiation and growth of crack as an intense anodic dissolution along a narrow front that is possible only if the metal surface and the crack wall remain inactive. To provide such a condition, the stress-corroding solution should either be able to form a protective surface film or locally destroy any such existing film. Therefore, the alloys with high inherent corrosion resistance due to the presence of protective or passive films such as aluminum and titanium alloys and stainless steels require aggressive halide species to promote SCC. On the other hand, inherently reactive metals like carbon steels or magnesium-base alloys require the presence of an

Aqueous Corrosion: Forms

103

environment that is itself partially passivating. Carbon steels stress-corrode in the solutions containing nitrates, hydroxides, or carbonates, which are anodic inhibitors for steel. SCC of magnesium-base alloys is encountered in the presence of both chromate (CrO42⫺) and chloride ions, the former providing passivity and the latter destroying it locally. The specificity of the environment is explained by these considerations. The dependence of cracking time of alpha brass on the solution pH is illustrated in Figure 3.43. The most susceptible pH range is also the stability range for Cu2O in the potential-pH diagram for Cu-H2O-NH3 system [20]. For SCC of carbon steels in various environments, potential-pH conditions for severe cracking susceptibility have been identified which also coincide with the regions of thermodynamically stable protective films [21]. The susceptible ranges for iron-base alloys with respect to the schematic anodic polarization curve are shown in Fig. 3.44. The upper zone corresponds to the breakdown potential range for stainless steels initiating pitting in chloridecontaining solutions. At more negative (active) potentials, pitting is absent and so is the SCC originating from pits. The lower zone represents the borderline passivity condition for carbon steels in nitrates, hydroxides, or carbonates. The addition of small amounts of nitrates to concentrated NaOH solution shifts the corrosion potential significantly in the positive direction beyond the cracking potential range in alkali and the SCC stops. On the other hand, the free corrosion

Figure 3.43 Time to cracking as a function of pH for brass in ammoniacal copper sulfate solution.

104

Chapter 3

Figure 3.44 Schematic anodic polarization curve of iron-base alloys showing zones of susceptibility to SCC.

potential of some mild steels lies at the boundary of cracking range in hydroxide, and SCC occurs with the addition of a small amount of lead oxide that causes a potential shift into the cracking range. The SCC of brass has also been shown to be potential-dependent (Fig. 3.45) with cracking mode changing over from intergranular to transgranular beyond a certain value of anodic potential [22]. The figure also indicates the effect of bulk solution composition (copper concentration) on cracking time as well as on the susceptible potential range. The potential and pH conditions inside the crack are not the same as those in the bulk of the solution or at the surface of the alloy, and so a correlation may not always be possible with the potential-pH diagram. However, it is reasonable to believe that the conditions prevailing inside the crack essentially meet the electrochemical requirements as presented above for the SCC to occur. If electrochemical dissolution at the crack tip accounts for the propagation of

Aqueous Corrosion: Forms

105

Figure 3.45 Effect of applied potential on time of fracture of α brass in ammoniacal copper sulfate solutions with different copper contents.

stress corrosion cracks, then the crack velocities should be related to the current density at the crack tip through Faraday’s law according to: v⫽

iaM Z Fρ

(3.6)

where ia is the current density, M the molecular mass of the metal, Z the valency of the sovated ions, F is Faraday’s constant and ρ is the density of the metal. A straight-line relationship between crack velocity and current density, conforming to Eq. 3.6, has actually been observed in many systems supporting the postulate (Fig. 3.46). The deviation can be attributed to dissolution in the crack initiation process alone and a mechanical fracture propagation step. Mechanisms of SCC Any mechanism of SCC should explain the characteristic features of the process, i.e., the susceptibility of an alloy in a limited number of environments, the role of stress, and the nature of cracking apart from the influence of various metallurgical and environmental factors in the cracking process. Since SCC essentially involves a corrosive environment, mechanisms based on localized dissolution of metal have been postulated. On the other hand, the brittle fracture feature of SCC has led to the postulation of mechanisms based on cleavage. There are several

106

Chapter 3

Figure 3.46 Observed crack velocities versus current densities in different stress corroding systems. The straight line is calculated from Eq. (3.6).

variations of these two basic mechanisms. According to some models, e.g., the so-called continuous electrochemical mechanisms, both the initiation and propagation of cracks are dissolution-controlled. According to the periodic electrochemical-mechanical mechanism, the initiation of cracks or the embrittlement of the region ahead of the crack tip has been dissolution-controlled, whereas propagation is a mechanical brittle fracture step. There is a third group wherein the role of dissolution is deemphasized in the embrittlement process. Dissolution Mechanisms. According to this mechanism, a crack advances by preferential dissolution at its tip. The restriction of the dissolution process at the crack tip has been attributed to either a preexisting active path or a strain-generated active path. The preexisting active path mechanism is the oldest mechanism proposed [23]. Grain boundary segregation or precipitation (as discussed in the section ‘‘Metal-

Aqueous Corrosion: Forms

107

lurgical Aspects of SCC’’) brings about substantial change in the microchemistry of the grain boundary with respect to the bulk alloy. These segregated solutes or precipitated phases may act as anode in the local cell or, by acting as an efficient cathode, may cause the dissolution to be localized on the immediately adjacent matrix. The role of stress here is to keep the crack open for the accessibility of the corrosive medium to the crack tip. The stress concentration at the crack tip producing yielding has also been considered to accelerate dissolution due to the so-called mechanochemical effect [24]. Intergranular stress corrosion cracking (IGSCC) of sensitized stainless steel in various environments, including highpurity water, of carbon steels exposed to nitrate, hydroxide, or carbonate-bicarbonate solutions and of aluminum alloys exposed to chloride solutions have been explained in these terms. However, a propensity for intergranular corrosion is no guarantee for IGSCC. It has been shown that although the nickel-base superalloy IN718 and Ticolloy show intergranular corrosion in NaCl solution, in deaerated solutions they may crack transgranularly [25]. Strain-generated active path mechanism. The SCC in alloys that do not have a preexisting path has been explained by dissolution of strain-generated active path. There are two distinct models: film rupture [26] and slip-step dissolution [27]. According to the film rupture model, the localized plastic deformation at the crack tip ruptures the passivating film, exposing the bare metal which dissolves rapidly, resulting in crack extension. The film heals and the cycle is repeated. It is also assumed by some that once the propagation starts, the crack tip remains bare as the rate of film rupture exceeds the rate of repassivation. According to the slip-step dissolution model, the local plastic deformation produces a slip step and the bare slip step sustains dissolution until the next repassivation. Figure 3.47 gives the schematic representation of these two models. Since many of the stress corroding systems are associated with film formation and the SCC occurs under some sort of borderline passivity condition, the models involving film rupture have received considerable support. However, controversy persists with regard to a crack propagation by continued dissolution. Fracture surface features such as crystallographic cleavage and crack arrest marks for transgranular SCC, and well-defined grain boundary for intergranular SCC, often matching with those on the opposing fracture surface, are indicative of a brittle mechanical cleavage with very little dissolution. A corrosion tunnel model [28] has emerged where dissolution and mechanical fracture have been combined. It assumes that a fine array of small corrosion tunnels forms at the emerging slip steps. These tunnels grow in diameter and length until the stress in the remaining ligaments causes ductile deformation and fracture (Fig. 3.48a). However, such a model should produce a grooved fracture surface with evidence of microvoid coalescence at the broken ligaments, which have not been observed experimentally. So this model has been modified subsequently [29] by the suggestion that the application of a tensile stress results in

108

Chapter 3

Figure 3.47 Strain-generated active path mechanisms. (a) Film rupture model. (b) Slip-step dissolution model.

a change in the morphology of corrosion damage from tunnels to thin, flat slots, as shown in Fig. 3.48b. The width of the corrosion slots has been shown to approach atomic dimensions and a close correspondence of matching surfaces is expected. Cleavage Mechanisms. Several mechanisms have been postulated to explain the cleavage-type cracking encountered in SCC. These are hydrogen-assisted cracking, turnish rupture, film-induced cleavage, adsorption-induced cleavage, and atomic surface mobility mechanisms. Hydrogen-assisted cracking mechanism is very often described by the surface energy lowering model [30] in which atomically dissolved hydrogen acts to weaken the interatomic bonds in the plain-strain region of the crack tip (Fig. 3.49) by lowering the surface energy γs in the Griffith equation: σc ⫽ (2Eγs /πc) 1/2

(3.7)

where σc is the fracture stress necessary to cause the propagation of an elliptical crack of length 2c and E is Young’s modulus. Blocked glide planes have been considered to provide the initial Griffith cracks in steels. The decohesion may also be caused by hydrogen influx to the dilated lattice [31]. It has also been suggested that hydrogen decreases the stacking-fault energy to induce coplanar deformation at the tip. Embrittlement would then result from Lomer–Cottrell supersessile dislocations on the intersecting slip planes. The embrittlement has also been ascribed to the stress-assisted formation of brittle hydride phase ahead of the crack tip which facilitates crack growth by cleavage, with cracks arresting at the boundary where the relatively tough matrix is encountered. Another particle

Aqueous Corrosion: Forms

109

Figure 3.48 Corrosion tunnel models. (a) Initiation of a crack by the formation of corrosion tunnels at slip steps and ductile deformation and fracture of ligaments. (b) Dissolution of tunnels to flat-slot formation.

of hydride then forms in the region of crack tip and the process is repeated, resulting in discontinuous crack growth. Hydrogen-assisted crack growth in the manner described above has been suggested as the SCC mechanism for ferritic steels, nickel-base alloys, austenitic stainless steels, aluminum alloys, and intermetallic compounds. The most probable source for atomic hydrogen to enter the metal is the cathodic reduction of hydrogen ions accompanying anodic dissolution in aqueous environments. Materials with impurities like sulfur, phosphorus, antimony, and tin segregated at grain boundaries have been found to be more susceptible to hydrogen-induced cracking because these impurities act as hydrogen evolution poisons. On the other hand, the growing stress corrosion cracks have been effectively stopped in low-strength ductile alloys by cathodic polarization, refuting the validity of hydrogen-assisted

110

Chapter 3

Figure 3.49 Hydrogen-assisted cracking mechanisms. (a) Crack tip adsorption and bond rupture. (b) Embrittlement due to anchoring of dislocations by adsorbed hydrogen. (c) Decohesion by hydrogen influx to dilated lattice. (d) Crack extension due to brittle hydride phase formation.

SCC in these alloys. Again, all of the environments and conditions in which SCC is encountered do not produce hydrogen, and some of them produce surface films that constitute an effective barrier to hydrogen entry. These restrict the universal applicability of hydrogen-induced cracking as an SCC mechanism. The adsorption-induced cleavage mechanism, or the stress-sorption cracking mechanism [32], is based on the hypothesis that adsorption of environmental species lowers the interatomic bond strength and the stress required for cleavage fracture—an idea similar to the decohesion model of hydrogen-induced cracking. Adsorption is assumed to be potential-dependent, which accounts for the stoppage of SCC by cathodic polarization below a critical potential. The specificity of the species for inducing SCC in a particular alloy can also be conveniently explained in terms of preferential adsorption. However, this model does not explain how the crack maintains an atomically sharp tip in a normally ductile material, and it also fails to explain the discontinuous nature of crack propagation. The tarnish rupture mechanism was originally proposed to explain discontinu-

Aqueous Corrosion: Forms

111

ous transgranular crack growth of α-brass single crystals in ammoniacal solutions [33]. According to this model, a brittle surface film or tarnish forms on the metal that fractures under the applied stress. The exposed bare metal reacts rapidly with the environment, the film grows again, and the cycle of growth and fracture of the tarnish is repeated. The model was subsequently modified [34] for intergranular SCC proposing that the tarnish penetrates along the grain boundary ahead of the crack tip. The models are depicted in Fig. 3.50. However, the validity of grain boundary penetration in all systems has been doubted. Also, this model predicts a discontinuous intergranular crack growth, but the acoustic emission signals do not confirm this. The film-induced cleavage mechanism developed [35,36] as a consequence to the tarnish rupture model. In this model it is assumed that dissolution leads to the formation of a surface film and a crack growing in the film propagates further into the underlying metal. The successive events occurring during the propagation of transgranular cracks by this mechanism have been illustrated in Fig. 3.51. From an initial arrest position, the crack advances by cleavage for a limited distance, after which the crack becomes progressively blunted by plastic deformation until the propagation stage is repeated. The film may be a dealloyed layer or oxide, in which nanoporosity is considered to be the key feature leading to the injection of a sharp crack into the substrate. Experimental results showing the correlation of SCC with dealloying have been given for Cu-Zn and Cu-Al alloys [19] and also for stainless steel [37]. Certain objections have been put forward against this mechanism. Dealloyed layers are unlikely to have sufficient adhesion to the substrate or inherent brittleness to sustain cracking. Also, the crack advance distance as measured from the arrest marks is typically of the order of a few micrometers, and doubts have been raised [38] as to whether a cleavage crack could penetrate that far from an initiating film that is only of the order of 30 nm thick. The atomic surface mobility mechanism [39] suggests that many forms of environmentally induced cracking, including hydrogen embrittlement, SCC, and liquid metal embrittlement, grow by the surface diffusion of atoms away from the highly stressed crack tip to a new site at the less stressed crack sides. A schematic representation of the mechanism is shown in Fig. 3.52. The coefficient of surface self-diffusion will dominate the crack growth rate and the role of the environment is to change that diffusivity which is enhanced with the presence of low-melting surface compounds. However, it has been pointed out [40] that although carbon steel cracks in nitrates that form low-melting compounds with iron, it also cracks in the presence of high-melting-point Fe3O4. An opposite viewpoint has also been suggested [41], i.e., that the flux of atoms in the region of a stressed crack tip should be the reverse of what the mechanism requires, by analogy with diffusional creep. It has been criticized [42] that many of the correlations postulated in this mechanism would work equally well if the criterion was surface energy

Aqueous Corrosion: Forms

113

Figure 3.51 Film-induced cleavage mechanism. A schematic illustration of events during the propagation of transgranular stress corrosion cracks by cleavage. Figures a–c represent a section at the crack tip, whereas Figs. d–f represent a plan view of a semicircular crack radiating from the initiation site. ∆x indicates the crack advance distance per event.

reduction, rather than increase in diffusivity, and thus it becomes indistinguishable from earlier surface energy models. The proposed mechanisms of SCC thus differ immensely from one another with some suggesting an exclusive role of dissolution and others giving dissolution a marginal role or no role at all in the cracking process. While the proponents of a particular mechanism have produced enough evidence in support of the mechanism, observations not conforming to the mechanism have also been cited in plenty. Though there have been attempts from time to time to find a unified theory explaining all features of SCC in all of the alloy–environment systems, it appears more reasonable to believe that the mechanism varies from system to system. It is also probable that no particular mechanism operates exclusively in one system but that more than one mechanism is at play. Parkins [43] has proposed that the different mechanisms of stress corrosion should be considered as occurring within a continuous spectrum, with a gradual transition from one to the others as the dominance of corrosive processes is replaced by stress or strain leading to a brittle fracture, which is shown in Table 3.4.

Figure 3.50 Tarnish rupture models. (a) Alternate tarnish formation and rupture leading to transgranular cracking. (b) Penetration of tarnish along the grain boundary and its rupture causing intergranular cracking.

114

Chapter 3

Figure 3.52 Atomic surface mobility mechanism. Atoms migrate out of the crack tip from A to B, then subsequently to C and D. The vacancy created at A amounts to the advancement of crack by one atomic distance.

Remedial Measures Control of Stress. Lowering of applied stress below threshold stress or stress intensity by suitable changes in design and lowering or removal of residual stress by annealing treatment are effective means to reduce the incidence of SCC in service. The season cracking of brass cartridge cases, mentioned earlier was prevented by proper stress relief annealing treatment. It is a recommended practice to give such a treatment to brass tubes after drawing or to steel tubes after welding for use in sour oil wells. However, annealing may be impractical for some stainless steels that sensitize and become susceptible to intergranular attack. Since a tensile component of stress is required in SCC, it can be prevented by putting the surface of a component into compression, e.g., by shot peening. The treatment needs to be applied uniformly. It will not be effective if pitting occurs on the compressive layer. Control of Corrosion. The elimination or reduction of the damaging species is desirable, but often it is difficult to achieve in practice. The initial nondamaging concentration of the species may become high in the crevices or under hightemperature conditions bringing about failures. However, the prevention of intergranular SCC of sensitized stainless steel, a recurring and expensive problem in cooling-water piping of boiling water nuclear power plants, has been achieved by minimizing both dissolved oxygen and chloride [44]. Use of inhibitors reduces or even eliminates SCC, possibly by moving the

Stress corrosion spectrum Corrosion-dominated (solution requirements highly specific)

INTERGRANULAR CORROSION

Carbon steels in NO3⫺ solns.

Al-ZnMg alloys in Cl⫺ solns.

Cu-Zn alloys in NH3 solns.

Intergranular fracture along preexisting paths

←→ Fe-CrNi steels in Cl⫺ solns.

Mg-Al alloys in CrO42⫺ ⫹ Cl⫺ solns.

Stress-dominated (solution requirements less specific) Cu-Zn alloys in NH3 solns.

Transgranular fracture along strain-generated paths

Ti alloys in methanol

Aqueous Corrosion: Forms

Table 3.4

Highstrength steels BRITTLE in Cl⫺ FRACTURE solns.

Mixed crack paths by adsorption, decohesion, or fracture of brittle phase

115

116

Chapter 3

corrosion potential outside the range of cracking. The addition of small amounts of nitrates to concentrated NaOH prevents SCC of steel, as discussed in the section ‘‘Electrochemical Aspects of SCC.’’ Substances like H3PO2, Na2O4, and CO(NH2)2, which may be expected to form insoluble products with iron, retard or prevent cracking in nitrates. Addition of traces of NO to N2O4 or 1–2% H2O to red fuming nitric acid or CH3OH/HCl mixtures prevent SCC of titanium alloys in these media [45]. The addition of water causes anodic inhibition by shifting the potential to the safe passive potential range. However, there are some practical limitations for the use of inhibitors. Many failures occur in steam or under condensation conditions, and in both cases the transport of inhibitors to sites of crack initiation is not feasible. Increasing the corrosion rate to reduce SCC might appear to be a ridiculous proposition. But because SCC is a form of highly localized corrosion, extending corrosion over the whole of the surface will usually lessen the probability of such failures. It has been employed in making up mixtures containing HCl to clean austenitic stainless steel parts in chemical plants; the corrosion rate is maintained ⬎10 mpy [46]. Electrochemical Protection. Cathodic protection will control SCC in alloys that crack by anodic dissolution mechanism, but is likely to accelerate hydrogeninduced cracking, particularly in high-strength alloys. SCC failure of Kraft continuous digesters in the pulp and paper industry in an NaOH-Na2S environment at 140°C has been reported [47] to have been mitigated by the application of anodic protection. Material Selection. Choosing a different alloy resistant to the particular environment is a popular option to prevent SCC. An alloy having the lowest plateau velocity, as discussed in the section ‘‘Testing Methods,’’ can be chosen from among a number of susceptible alloys. Mitigation of SCC by alloy development has been a rare endeavor. However, it is reported [42] that a relatively inexpensive stainless steel has been developed that resists SCC up to 140°C in crevice tests with 20% NaCl in which 304 stainless steel will fail at 60°C. The addition of copper raises the lower critical potential for SCC. Minimization of phosphorus in austenitic stainless steels is also a key approach to making them resistant to chloride-induced SCC. The cladding of high-strength aluminum alloys with pure aluminum has been successfully employed to mitigate SCC in aircraft components. The relatively low susceptibility of pure metals to SCC has been utilized in this preventive measure. Practical Examples Practical examples of SCC in engineering alloys are abundant and varied. Figure 3.53 shows intensive SCC in a high-pressure autoclave made of 18-8 stainless

Aqueous Corrosion: Forms

117

Figure 3.53 Stress corrosion cracking of type 304 stainless steel autoclave due to chloride contamination.

steel [7]. The failure in service took place in a matter of hours and was traced to the buildup of chloride concentration on the outside surface because of evaporation of the water used for cooling, which originally contained only a negligible amount of chloride. Yet another interesting example of SCC is depicted in Fig. 3.54, where the development of longitudinal cracks in a cold-drawn brass tube had its origin at the bird droppings that provided the ammonia necessary for corrosion [9]. The occurrence of SCC has been reported in high-strength aluminum alloy components in low-flying British military aircrafts arising from chloride-contaminated moisture, which led to the ban on the use of the material until a solution was found through cladding of the material with pure aluminum. Carbonate-bicarbonate environments have been identified as the probable cause of cracking in natural gas transmission steel pipelines [48].

Figure 3.54 Stress corrosion cracking of cold-drawn 70–30 brass tube originating at bird droppings.

118

Chapter 3

Numerous cases of SCC in chemical process plants and power generation plants have been reported in the literature. Oxygenated pure or impure hightemperature water has caused failures in steam generator shells made of carbon or low-allow steel [21]. Extensive failures have taken place in austenitic 18Mn4Cr steel rotor end-retaining rings in contamination with oxygenated high-temperature water [49]. SCC of austenitic stainless steel pipes caused by the same environment in boiling water reactors (BWRs) has cost the world’s nuclear power industries as much as $10 billion [50]. SCC of nickel-base alloy 600 in reducing high-temperature water and supercritical steam in pressurized water reactors (PWRs) has been reported [51]. Room temperature cracking of sensitized austenitic stainless steels produced by polythionic acid was first experienced in catalytic reformers used in the petroleum industry [52].

3.9.2

Corrosion Fatigue

Corrosion fatigue is the cracking of a metal or alloy under the combined action of a corrosive environment and repeated or fluctuating stress. As in SCC, successive or alternative exposure to stress and corrosion does not lead to corrosion fatigue. Metals and alloys fail by cracking when subject to cyclic or repetitive stress even in the absence of a corrosive medium, a phenomenon known as fatigue failure. The greater the applied stress, the less is the number of cycles required and the shorter is the time to failure. In steels and other ferrous materials, no failure occurs for an infinite number of cycles at or below a stress level that is called the endurance limit or the fatigue limit. In a corrosive medium, the fatigue limit is lowered or no longer observed, i.e., the failure occurs at any applied stress if the number of stress cycles is sufficiently large. Corrosion fatigue may thus also be defined as the reduction in the fatigue life of a metal in a corrosive environment. These behaviors are illustrated in Fig. 3.55. In nonferrous metals and alloys, fatigue limit is not indicated, but nevertheless a corrosive environment brings down the fatigue life. It may be noted here that unlike SCC, corrosion fatigue is equally prevalent in pure metals and their alloys. Corrosion fatigue cracks, like mechanical fatigue cracks, are usually transgranular and are rarely branched (Fig. 3.56). The cracks propagate perpendicular to the principal tensile stress. Whereas in mechanical fatigue a single crack leads to failure, corrosion fatigue cracks in an affected component are often numerous. The fracture surface may show macroscopic beach marks typical of fatigue failure and corrosion products. Electron microscopy reveals striation marks, each one indicating the advancement of fatigue crack for each cycle of stress (Fig. 3.57). Stress Factors Mean stress, frequency of cyclic stress, and stress amplitude affect corrosion fatigue. If the mean stress has a tensile value, the growing fatigue crack is held

Aqueous Corrosion: Forms

119

Figure 3.55 Fatigue curves for ferrous materials in air and in corrosive environments.

Figure 3.56 Corrosion fatigue cracks in carbon steel.

120

Chapter 3

Figure 3.57 Fatigue striations in aluminum alloy 2024-T851 (2000⫻).

open for the entire cycle and the attack is aggravated in contrast to the situation when the mean stress is zero, as in the case of a sinusoidally varying stress of equal amplitude. The situation is further aggravated if a static tensile stress is superimposed on the cyclic stress. The mean stress in such a case is raised to a higher tensile value. Cyclic stresses of low frequency and high amplitude lead to greater crack propagation per cycle. This arises from the fact that such conditions allow longer interaction between material and environment. However, if the frequency is high, the crack tip may not get the chance to be exposed to the environment, and the corrosion fatigue behavior becomes one like the mechanical fatigue behavior in an inert environment. The presence of stress raisers, such as notches or surface roughness, increases susceptibility to corrosion fatigue. The fatigue limit decreases sharply (Fig. 3.58). Corrosion pits also act as stress raisers and cracks have often been observed to initiate from corrosion pits. Environmental Factors Unlike SCC, corrosion fatigue is not restricted to specific environments. Any environment causing general attack in a metal or alloy beyond a certain minimum corrosion rate is capable of causing corrosion fatigue. For steels, the minimum corrosion rate required is about 1 mpy [53]. Thus corrosion fatigue of steel is encountered in varied environments like freshwater, seawater, combustion product condensates, and many chemicals. The fatigue strength or fatigue life of a material decreases sharply when ex-

Aqueous Corrosion: Forms

121

Figure 3.58 Fatigue curves showing reduced endurance limit in the presence of (a) corrosive environment and (b) added notches.

posed to a corrosive environment. For example, the fatigue strength at 10 million cycles in salt water could be reduced to as little as 10% of that in moist air. The oxygen content of the environment greatly influences corrosion fatigue. It may be mentioned here that fatigue tests conducted in vacuum yield much higher values of endurance limit than that encountered in normal tests in air containing oxygen and moisture. Fatigue in their presence therefore may be considered as a case of corrosion fatigue. Complete elimination of oxygen from a neutral solution has been found to eliminate corrosion fatigue of low-carbon steel. Corrosion fatigue increases almost proportionately with the increase of general aggressiveness of the corrodent. Thus an increase in temperature, a lowering of pH, or an increase in the concentration of corrosive species leads to an aggravation of corrosion fatigue.

122

Chapter 3

Mechanism In mechanical fatigue, the crack initiation is attributed to the formation of extrusions and intrusions at the metal surface because of localized slip within the grain caused by the stress (Fig. 3.59). The slip is localized because the stress is not large enough to produce massive slip. Once the crack initiates the tip of the crack undergoes localized plastic deformation in the same manner and the crack propagates. Stresses below the endurance limit cannot overcome the work hardening caused by plastic deformation and further slip is impeded. Since corrosion fatigue involves the lowering of endurance limit or the change of the fatigue behavior of ferrous metals to that of nonferrous metals exhibiting no endurance limit, it has been envisaged that the corrosive environment helps in the process of localized plastic deformation at the crack tip. This may be achieved either by the removal of the barriers to plastic deformation, such as dislocation piled up at the metal surface at slip steps, or by favoring plastic deformation by the reduction of surface energy. The process of formation of extrusions and intrusions thus continues at lower stress levels. Although the exact mechanism has not been quantitatively presented, it nevertheless explains the combined role of stress and corrosion. The more aggressive the solution, the more of this effect is achieved. Also, the shortening of fatigue life with the application of cyclic stresses of lower frequency is explained by the

Figure 3.59 Extrusions and intrusions in copper after 6 ⫻ 105 cycles in air silvercoated after test.

Aqueous Corrosion: Forms

123

fact that the corrosive environment gets more time to interact with the metal. A pit is expected to provide aggressive corrosive conditions congenial to the process, and very often the corrosion fatigue crack initiates at the base of a pit. Remedial Measures Lowering of Stress. Lowering of operating stresses by lowering the mean stress or the amplitude of the cyclic stress leads to an increase of corrosion fatigue life. This can be accomplished by changes in design. Points of stress concentration, like sharp fillets, may be suitably modified. Control of Environment. Corrosion fatigue is reduced if the corrosive factors are controlled. Deaeration of saline solution has been reported [53] to restore the normal fatigue limit in air for steels. Addition of inhibitors is also effective. Use of Coatings. Organic coatings, like paints, act as a physical barrier between the metal and the environment and lower the incidence of corrosion fatigue. Inhibitors can be incorporated in such coatings. Coatings of zinc or cadmium on steel provide cathodic protection to the base metal. Noble metal coatings on steel act as barriers, but they should be sufficiently dense and thick. Breakage and discontinuity not only increases corrosion at those points but provides a readymade site for corrosion crack initiation. Polarization. Cathodic protection by impressed current reduces corrosion and, subsequently, corrosion fatigue. For steels, cathodic polarization to ⫺0.49 V (SHE) in salt water provides protection. In high-strength steels, however, cathodic polarization may lead to hydrogen-induced cracking. In stainless steels, protection can be achieved by producing passivity by anodic polarization. Shot Peening. Shot peening introduces compressive stresses to the surface and is effective in combating fatigue in air. It is marginally effective for corrosion fatigue, as severely aggressive environments tend to remove the compressive layer by dissolution. Practical Examples Vibrating parts or metal structures that have been designed to operate safely in air below the fatigue limit often fail due to corrosion fatigue. Examples include wire ropes, springs, marine propellers, oil-well sucker rods, and heat exchangers. Corrosion fatigue failure has been reported [54] in a hollow, spindled alloy steel aircraft shaft that in operation was subject to static radial, cyclic torsional, and cyclic bending stresses. The inner surface of the hollow shaft was continuously exposed to hydraulic oil at temperatures of 0–80°C. The failure investigation revealed that fatigue cracks had originated from corrosion pits. The hydraulic oil was originally noncorrosive, but water contamination made it corrosive.

124

Chapter 3

Figure 3.60 Radiograph of boiler tube showing many fine transverse corrosion fatigue cracks.

Figure 3.60 is the radiograph of a marine boiler tube that failed by corrosion fatigue [9]. The presence of a large number of cracks are immediately noticeable. The material is mild steel. The corrosion has been attributed to overheating. On surfaces with an excessive rate of steam generation the net effect was that of concentrating the boiler water, which then became corrosive to steel. Cyclic stresses of the order of several 10,000 psi were considered to be produced by the expansion and contraction of the tube under restrain.

3.10 BIOLOGICALLY INFLUENCED CORROSION Biologically influenced corrosion refers to the degradation of metals caused by the activity of living organisms. Both micro- and macroorganisms have been found to contribute to corrosion in a variety of environments including soil, groundwater, seawater, domestic and industrial freshwasters, natural petroleum products, and oil-emulsion cutting fluids. It does not represent any special form of corrosion. The aggravation of corrosion under environmental conditions whereby the corrosion rates are expected to be low is often traced back to the presence of bioorganisms. The biological activities may directly influence the anodic or cathodic reactions, produce end products that are corrosive, or influence protective surface films. The corrosion may also aggravate under the deposits formed by the organisms or their bioactivity products.

Aqueous Corrosion: Forms

125

3.10.1 Microbiological Corrosion A number of bacteria have been identified that aggravate corrosion of metals such as carbon steels, cast irons, stainless steels, aluminum alloys, and copper alloys. These are classified as aerobic and anaerobic according to their ability to grow in the presence or absence of air. They survive under a wide range of temperature (0–50°C) and pH (3–10). The important microorganisms contributing to corrosion are listed in Table 3.5. The affected areas are characteristically associated with excessive deposits, tubercles, or biological slimes. Buried grayiron structures suffer from graphitic corrosion (Section 3.7.3) where bacteria help in the process of leaching out of iron. Anaerobic bacteria grow in waterlogged clay-rich soils, in deaerated waters, in stagnant cutting fluids, underneath the deposits caused by aerobic bacteria, and under macroorganisms like mollusks and barnacles. Sulfate-reducing bacteria, e.g., D. desulfuricans, are the most important anaerobic bacteria that influence the corrosion behavior of steels. These bacteria reduce inorganic sulfates to sulfides in the presence of hydrogen according to the schematic equation: SO42⫺ ⫹ 8H → S2⫺ ⫹ 4H2O

(3.8)

Hydrogen is provided by the cathodic reaction or by the organic matters present in soil or in the general environment. For each equivalent of hydrogen atom consumed by the bacteria, one equivalent of Fe2⫹ enters the solution to form rust and FeS. The presence of sulfide in rust is often an indication of the corrosion having been influenced by sulfate-reducing bacteria. Qualitative detection is made possible by the treatment of the deposit with dilute hydrochloric acid, which produces H2S with its characteristic smell. Aerobic bacteria contribute to corrosion by oxidizing or sulfide and iron. Sulfur-oxidizing bacteria, like Thiobacillus thiooxidans, oxidize elemental sulfur or sulfides to produce sulfuric acid according to the equation: 2S ⫹ 3O2 ⫹ 2H2O → 2H2SO4

(3.9)

Highly corrosive conditions are thereby produced. Also the slime, which is a polymeric excreta of these bacteria, provides a localized anaerobic environment for sustaining the growth of sulfate-reducing bacteria. Iron-oxidizing bacteria, such as Thiobacillus ferrooxidans, oxidize soluble ferrous ions to less soluble ferric ions and thereby accelerate the anodic reaction for dissolution of iron. Hydrated ferric oxides along with biological slimes tend to form tubercles often impairing fluid flow inside the pipe, which has been illustrated in Fig. 3.61. The deposits also produce areas shielded from oxygen, thus providing localized anodic areas leading to crevice corrosion. In soils, aerobic and anaerobic conditions may alternately prevail depending on seasonal variation, with the former prevailing during dry seasons and the latter

Table 3.5

Bacteria known to cause microbiologically influenced corrosion

126

Genus or species

pH range

Temp. range (°C)

Oxygen requirement

Desulfovibrio Best known: D. desulfuricans

4–8

10–40

Anaerobic

Iron and steel, stainless steels, aluminum, zinc, copper alloys

Utilizes hydrogen in reducing SO42⫺ to S2⫺ and H2S; promotes formation of sulfide films

6–8

10–40 (some 45–75) 10–40

Anaerobic

Iron and steel, stainless steels

Reduces SO42⫺ to S2⫺ and H2S (spore formers)

Anaerobic

Iron and steel

0.5–8

10–40

Aerobic

Iron and steel, copper alloys, concrete

Thiobacillus ferrooxidans

1–7

10–40

Aerobic

Iron and steel

Gallionella

7–10

20–40

Aerobic

Iron and steel, stainless steels

Sphaerotilus

7–10

20–40

Aerobic

Iron and steel, stainless steels

Reduces SO42⫺ to S2⫺ and H2S Oxidizes sulfur and sulfides to form H2SO4; damages protective coatings Oxidizes ferrous (Fe2⫹) to ferric (Fe3⫹ ) Oxidizes ferrous (and manganous) to ferric (and manganic); promotes tubercule formation Oxidizes ferrous (and manganous) to ferric (and manganic); promotes tubercule formation

S. natans Pseudomonas

— 4–9

— 20–40

— Aerobic

P. aeruginosa

4–8

20–40

Aerobic

Aluminum alloys Iron and steel, stainless steels Aluminum alloys

Desulfotomaculum Best known: D. nigrificans (also known as Clostridium) Desulfomonas Thiobacillus thiooxidans

—

Metals affected

Some strains can reduce Fe3⫹ to Fe2⫹

Chapter 3

From S. C. Dexter, Metals Handbook, Vol. 13, Corrosion, 9th ed., ASM International, p. 114, 1987.

Action

Aqueous Corrosion: Forms

127

Figure 3.61 Tubercules formed due to microbial activities in carbon steel pipe.

during rainy seasons. Thus, sulfur-oxidizing bacteria and sulfate-reducing bacteria may grow in cycles causing continuous damage to the underground structures.

3.10.2 Macrofouling Macrobiological organisms, such as shells, mollusks, barnacles, etc., cause both fouling and corrosion of metallic structures, vessels, and pipelines exposed to seawater as well as freshwater. Fouling is often more important than the attendant corrosion. These organisms remain attached to the metal surface, and their accumulation at the bottom of a ship’s hull increases the drag and power requirement. In heat exchangers, such accumulations may clog the pipeline and impair fluid flow as well as heat transfer. Stagnant and low rate of fluid flow are conducive to biofouling. Biofouling contributes to corrosion in several ways. The underlying metal remains sheltered from dissolved oxygen, and a crevice condition is created. The metabolic byproducts of these organisms are often acidic and hence corrosive. Moreover, the anaerobic conditions prevailing underneath the macroorganisms can favor the growth of anaerobic bacteria which, in turn, accelerate corrosion of the metal.

3.10.3 Remedial Measures For Microbiological Corrosion 1. Aeration and chlorination. It is important to identify the type of bacteria involved in corrosion through culturing. Aeration of water in a closed or recirculating system reduces the activity of anaerobic bacteria. Chlorination and treatment with biocides help control populations of some bacteria,

128

2. 3. 4.

5.

Chapter 3 though they are not effective in all cases. Also, the bacteriocides fail to reach the areas underneath deposits where the bacteria thrive. Coatings. Coating the buried structure with tar, enamel, plastic, or the like is often an effective means to preclude the bacteria from the metal surface. Cathodic protection. Cathodic protection in combination with coatings can be used to prevent or arrest microbiological corrosion. Maintenance. During storage or after hydrotesting, water should not be allowed to stand for a long period. Complete drainage and drying up are advocated. Inhibitors may be used in stagnating water and cutting-oil fluids. Periodic cleaning of pipelines is also essential. Substitute materials. In worst affected soils, steel pipes may be replaced by asbestos or plastic pipes to avoid microbiological corrosion.

For Macrobiofouling 1. Paints. The application of antifouling paints is probably the most effective and most widely used means to prevent biofouling in sea water. Ships and piers are coated with specially formulated paints containing compounds toxic to the organisms. Copper compounds are often used, as the released copper ions poison the growth of barnacles and other marine organisms. 2. Cleaning. Periodic mechanical cleaning of surfaces of structures and inside of pipelines help reduce the growth of bioorganisms and the creation of crevice sites. 3. Use of biocides. In closed systems, fouling can be mitigated by chlorination and periodic injection of suitable biocides, including copper compounds.

3.10.4

Practical Examples

Figure 3.16 illustrates a crevice corrosion attack on stainless steel in seawater as a result of barnacle attachment. Microbiological corrosion has been reported [8] to be encountered in stainless steel tanks, pipelines, and flanged joints, particularly in welded areas after hydrotesting with natural river or well waters in chemical processing industries as well as in nuclear power generation. Copper-nickel, brass, and aluminum-bronze pipes and tubes are also affected. Sulfate-reducing bacteria have been observed to cause damage to buried steel pipelines, particularly in oil fields. It has been reported [55] that a well water caused failure of 2in. -diameter galvanized water pipe within 2 years by the action of sulfate-reducing bacteria, whereas municipal water employing similar wells, but chlorinated beforehand, was much less corrosive.

REFERENCES 1. V. J. Colangelo and F. A. Heiser, Analysis of Metallurgical Failures, John Wiley and Sons, New York, p. 182, 1974.

Aqueous Corrosion: Forms

129

2. ASM Metals Handbook, 8th ed., Vol. 10, p. 168, American Society for Metals, Metals Park, Ohio, 1975. 3. A. Singh, R. McClintock, T. W. Rudwich, and R. L. Brackett, Mat. Eval. Vol. 41, p. 568, 1983. 4. D. A. Jones, in Corrosion Processes, R. N. Parkins (ed.), Applied Science, Englewood, NJ, p. 180, 1982. 5. M. Janik-Czakor, A. Szummen, and Z. Szklarska-Smialowska, Corros. Sci., Vol. 15, p. 775, 1975. 6. Satish D. Patel, Mechanics of Corrosion, LaBour Pump Company, Elkhart, IN, 1990. 7. M. G. Fontana and N. D. Greene, Corrosion Engineering, McGraw-Hill, New York, p. 44, 1967. 8. ASM Metals Handbook, 9th ed., Vol. 13, American Society for Metals, Metals Park, Ohio, 1989. 9. R. D. Barer and B. F. Peters, Why Metals Fail, Gordon and Breach, New York, 1970. 10. B. F. Brown and C. D. Beachem, Corros. Sci., Vol. 5, p. 745, 1965. 11. R. N. Parkins, in Fundamental Aspects of Stress Corrosion Cracking, R. W. Staehle, A. J. Forty, and D. Van Rooyen (eds.), NACE, Houston, p. 361, 1969. 12. R. P. Harrison, D. D. G. Jones, and F. G. Newman, in Stress Corrosion Cracking and Hydrogen Embrittlement in Iron-Base Alloys, R. W. Staehle, J. Hochman, R. D. McCright, and J. E. Slater (eds.), NACE, Houston, p. 659, 1977. 13. G. T. Burnstein and J. Woodward, Metall. Sci., Vol. 17, p. 111, 1983. 14. N. Bandyopadhyay and C. L. Briant, Metall. Trans., Vol. 14A, p. 2005, 1983. 15. R. H. Jones, in Proc. Int. Symp. Environmental Degradation of Materials in Nuclear Power System–Water Reactors, American Nuclear Society, p. 173, 1985. 16. R. N. Parkins, in Environment-Induced Fracture of Metals, R. P. Gangloff and M. B. Ives (eds.), NACE, Houston, p. 1, 1990. 17. H. W. Pickering and P. R. Swann, Corrosion, Vol. 19, p. 45, 1963. 18. U. K. Chatterjee, S. C. Sircar, and T. Banerjee, Corrosion, Vol. 26, p. 141, 1970. 19. K. Sieradzki and R. C. Newman, J. Phys. Chem. Solids, Vol. 40, p. 1101, 1987. 20. E. Mattsson, Electrochim. Acta, Vol. 3, p. 279, 1961. 21. J. Congleton, T. Shoji, and R. N. Parkins, Corros. Sci, Vol. 25, p. 633, 1985. 22. S. C. Sircar, U. K. Chatterjee, M. Zamin, and H. J. Vijayendra, Corros. Sci., Vol. 12, p. 217, 1972. 23. E. H. Dix, Trans. Am. Inst. Metall. Engg., Vol. 137, p. 11, 1940. 24. T. P. Hoar and J. M. West, Proc. Roy. Society, Vol. 268A, p. 304, 1962. 25. P. C. Wang et al., in Parkins Symposium on Fundamental Aspects of Stress Corrosion Cracking. S. M. Bruemmer et al. (eds.), TMS, Warrendale, p. 399, 1992. 26. H. L. Logan, J. Res. Natl. Bur. Stand., Vol. 48, p. 99, 1952. 27. D. A. Vermilyea, Ref. 11, p. 15. 28. P. R. Swann and H. W. Pickering, Corrosion, Vol. 19, p. 102t, 1963. 29. J. M. Silcock and P. R. Swann, in Environment-Sensitive Fracture of Engineering Metals, Z. A. Forulis (ed.), The Metallurgical Society, London, p. 561, 1991. 30. N. J. Petch and P. Stables, Nature, Vol. 169, p. 842, 1952. 31. R. A. Oriani, Berichte Bunsen-Gesellschaft & Physik. Chem., 76, p. 848, 1972.

130

Chapter 3

32. H. H. Uhlig, in Physical Metallurgy of Stress Corrosion Fracture, T. N. Rhodin (ed.), Interscience, New York, p. 1., 1959. 33. C. Edeleanu and A. J. Forty, Phil. Mag., Vol. 5, p. 1029, 1960. 34. A. J. McEvily and P. A. Bond, J. Electrochem. Soc., Vol. 112, p. 141, 1965. 35. E. N. Pugh, Corrosion, Vol. 41, p. 517, 1985. 36. K. Sieradzki and R. C. Newman, Phil. Mag., Vol. 51, p. 95, 1985. 37. R. C. Newman, R. R. Corderman, and K. Sieradzki, Br. Corros. J., Vol. 24, p. 145, 1989. 38. W. W. Gerberich and S. Chen., Ref. 16, p. 167. 39. R. Galvele, Corros. Sci., Vol. 27, p. 1., 1987. 40. R. N. Parkins, Ref. 16, p. 1. 41. R. A. Oriani, Ref. 16, p. 263. 42. R. C. Newman and R. P. M. Procter, Br. Corros. J., Vol. 25, p. 259, 1990. 43. R. N. Parkins, in Corrosion, LL Shreier (ed.), Newman Butterworths, London, p. 24, 1976. 44. W. L. Williams, Corrosion, Vol. 13, p. 539t, 1959. 45. M. J. Blackburn, W. H. Smynl, and J. A. Feeny, in Stress Corrosion Cracking in High Strength Steels and Titanium and Aluminum Alloys, B. F. Brown (ed.), Naval Research Lab., Washington, D.C., 1972. 46. F. L. Laque and H. R. Copson, in The Corrosion Resistance of Metals and Alloys, Reinhold, New York, p. 392, 1963. 47. D. Singbell and A. Garner, Mat. Perform., Vol. 26, p. 31, 1987. 48. R. N. Parkins, Br. Corros. J., Vol. 14, p. 5, 1980. 49. R. Viswanathan, in Retaining Ring Failures: EPRI Workshop on Retaining Rings, EPRI, Palo Alto, CA, 1982, p. 24. 50. R. W. Staehle, Ref. 16, p. 561. 51. H. Coriou, L. Grall, P. Olivier, and H. Willermoz, Ref. 6, p. 352. 52. A. Dravnicks and C. H. Samans, Proc. Am. Petroleum Inst., Vol. 37, p. 100, 1957. 53. D. J. Duquette and H. H. Uhlig, Trans. Am. Soc. Materas, Vol. 61, p. 449, 1968. 54. ASM Metals Handbook, 9th ed. Vol. 11, p. 261, American Society for Metals, Metals Park, Ohio, 1989. 55. H. H. Uhlig, Corrosion and Control, John Wiley and Sons, New York, p. 97, 1971.

4 Aqueous Corrosion: Prevention

The preventive measures pertaining to each form of corrosion have been presented in the previous chapter. It is possible to classify the preventive measures under six general headings: 1. 2. 3. 4. 5. 6.

Materials selection Control of environment Protective coatings Cathodic protection Anodic protection Design improvement

Since corrosion involves interaction between material and environment, the first approach to corrosion prevention is obviously the use of the most compatible material in a given environment and the next approach is to reduce the aggressiveness of the environment toward the material in use. Provision of a coating to separate the environment from the material in order to prevent their interaction is also a natural consideration, and many such coatings have been developed and are being used. Cathodic and anodic protection are preventive measures based on electrochemical principles of corrosion. Lastly, an improvement in design of the metallic component or the system to which it belongs can reduce corrosion in many cases.

131

132

Chapter 4

4.1 MATERIAL SELECTION Economics dictates the selection of materials for corrosion applications. The most corrosion-resistant metal or alloy may not always be the choice because of its low abundance, high cost, fabrication difficulty, or unsuitability to meet the engineering requirements. Gold and platinum thus fail to be considered as viable engineering materials though they find exotic applications because of their excellent corrosion resistance. Cast irons and steels as plain carbon or alloyed varieties, on the other hand, still constitute the major bulk of engineering components and also find use in domestic applications. Iron or steel appliances may need elaborate corrosion prevention measures at times, but this will still be more economical in many cases than their replacement by costlier materials having higher corrosion resistance. Also, long-life equipment made with a higher corrosion resistance may not always be preferred by the industry as the process may become technically obsolete in a relatively short time. Conversely, a costlier metal may be the choice for a component to prevent its frequent replacement due to corrosion failures that hamper the production process and consequently the economics as a whole. Some metal–environment combinations have been found to be the most compatible from a corrosion performance and economic point of view. For example, carbon steels are the best material to handle concentrated sulfuric acid (above 65%), whereas lead is recommended for dilute sulfuric acid. Nitric acid is most economically handled by stainless steels, and caustic solutions by nickel and its alloys. Titanium, though expensive, still proves to be the most economical choice to handle hot, strong oxidizing solutions containing FeCl3 or CaCl2 in which all other metals tend to corrode or pit severely. In applications such as human body implants where a little bit of corrosion is too much, safety considerations override economics and the metal of ultimate corrosion resistance, like tantalum or palladium-coated titanium, is the choice. Corrosion performance data of metals and alloys in different environments are available from various reference sources like handbooks, corrosion charts, and corrosion data books. The behavior of individual metals and alloys in corrosive media of industrial importance is often presented as isocorrosion charts. The charts constitute the curves representing the same corrosion rate as a function of temperature and concentration. The isocorrosion chart for carbon steels in concentrated sulfuric acid (above 60%) is shown in Fig. 4.1. The curves represent corrosion rates of 5, 20, 50, and 200 mpy. In the regions contained by the curves, intermediate corrosion rates are expected to be encountered. The dotted lines indicate the lack of reliable data. Combined isocorrosion charts for a group of metals and alloys in a particular medium are also presented. Figure 4.2 shows such a chart for six important metals and alloys, i.e., chemical lead, iron, Durimet 20, high Ni-Mo alloy, high Ni-Cr-

Aqueous Corrosion: Prevention

Figure 4.1 Isocorrosion chart for steel in sulfuric acid.

Figure 4.2 Combined isocorrosion chart for six alloys in sulfuric acid.

133

134

Chapter 4

Mo alloy, and Duriron normally used for handling sulfuric acid. This particular chart has been constructed by superimposing the individual isocorrosion charts and selecting the 20 mpy curve for each material. For applications where a corrosion rate of up to 20 mpy can be tolerated, the suitability of the materials at different concentrations and temperatures is indicated in the chart. Yet another type of summary chart is available showing demarcated areas of identical corrosion rate for a wide variety of materials as a function of concentration and temperature. The summary chart of corrosion rate of less than 20 mpy for various materials is shown in Fig. 4.3 and the materials falling in each zone are detailed in the table that follows. It should be remembered that the corrosion rates indicated in these charts correspond to the rate of general corrosion under the static condition of the medium. The propensity for localized attack is not indicated. The flowing conditions of the medium, particularly high velocities, tend to bring about a drastic increase in corrosion rate. Also, the erosion corrosion behavior of steels (and other materials as well) varies widely depending on their composition and heat treatment. For example, a hot water sprinkler used in a rubber curing plant was reported to have failed prematurely because of the widening of the orifices due to erosion corrosion as the material used was a 0.1% carbon steel against the prescribed material of 0.2% carbon steel, which proved too soft for the operation concerned. The corrosion behavior of some of the important metals and alloys of construction are discussed briefly in the following sections. It should be borne in mind that nonmetallic materials like ceramics, polymers, and concrete perform better than metals in some corrosion applications and are economical as well.

4.1.1

Cast Irons

Cast irons can be grouped into three categories on the basis of their alloy content: unalloyed irons, low- and moderately alloyed irons, and the high-alloy irons. The unalloyed irons of gray varieties inherently contain silicon from 1.5% to 3%. The corrosion behavior of cast irons differs from that of steel because of their high silicon content and the presence of graphite flakes. Due to its cathodic nature, graphite remains at the casting surface on selective dissolution of the iron matrix (Section 3.7.3). The attack on the underlying metal retards subsequently if the corrosion products are retained in the network of graphite flakes. The presence of silicon leads to the formation of a dense and adhering iron oxide–iron silicate subscale that retards further attack. This behavior of gray iron contrasted with mild steel is schematically represented in Fig. 4.4. Under conditions where corrosion rate continues to be high, iron castings are still preferred to steel because these can be produced with thicker sections providing a satisfactory length

Aqueous Corrosion: Prevention

135

of service. Unalloyed white cast irons because of their inherent high hardness are resistant to erosion corrosion. Low and moderately alloyed irons contain a few tenths of copper and moderate additions of nickel and chromium, which impart better resistance to atmospheric corrosion and erosion corrosion. Ni-Hard is an example of this group, which contains 4% Ni and 2% Cr. The high-alloy irons are classified into three types: 1. The austenitic alloys. These irons are alloyed with nickel and chromium, with or without copper, for improved corrosion and abrasion resistance. Mechanical properties and heat resistance are also improved. These are known as Ni-Resist. Seven varieties of Ni-Resist contain 14–32% Ni and 1.75– 5.5% Cr. 2. The high-chromium alloys. These contain 15–30% Cr and are white irons. They are particularly resistant to high-temperature oxidation and erosion corrosion. 3. The high-silicon irons. The addition of more than 12% Si to irons makes them extremely hard and corrosion-resistant because of the formation of a passive SiO2 surface layer. Duriron is an example of this group, which contains 14.5% Si and 0.95% carbon. Unalloyed and low-alloy cast irons perform better than steels in many atmospheric exposures because the corrosion products form a protective film. Sulfur dioxide in industrial atmospheres and chlorides in marine atmospheres tend to increase corrosion. For domestic water supply cast iron pipes have been successfully used over centuries. Cast iron water pipe installed at Versailles, France in 1664 is still in service. Gray cast irons, however, suffer from severe graphitization and pitting when exposed to seawater. Ductile iron gives a better performance and is being increasingly used for marine appliances such as machinery foundations, valve bodies, tanks, cargo pipings, etc. Unalloyed gray iron is commonly used for stills in the manufacture of acetic and fatty acids, but it corrodes rapidly in dilute solutions of organic acids such as acetic, oxalic, citric, or lactic. Low-alloy cast irons perform better. Unalloyed and low-alloy cast irons are also not resistant to corrosion in dilute to medium concentrations of sulfuric acid and nitric acid and in hydrochloric acid of all concentrations. However, Ni-Resist alloys are fairly resistant. High-chromium cast irons have excellent resistance to dilute nitric acid. Duriron shows excellent resistance to sulfuric acid (Fig. 4.2) and nitric acid of all concentrations, and the modified grades (Durichlor) containing 2–3% Mo or Mo and Cr possess improved resistance to hydrochloric acid, chlorides, and pitting. These are widely used for drain lines, pumps, valves, and as anodes for impressed current cathodic protection.

136

Chapter 4

Figure 4.3 Corrosion resistance of materials to sulfuric acid; codes shown below. Materials in shaded zones having reported corrosion rate less than 20 mpy 10% aluminum bronze (air-free) Illium G Glass Hastelloy B and D Durimet 20 Worthite Lead Copper (air-free) Monel (air-free) Haveg 43 Rubber (up to 170°F)

Zone 1 Impervious graphite Tantalum Gold Platinum Silver Zirconium Nionel Tungsten Molybdenum Type 316 stainless (up to 10% aerated)

Figure 4.3 (Continued) Glass Silicon iron Hastelloy B and D Durimet 20 (up to 150°F) Worthlie (up to 150°F) Lead Copper (air-free) Monel (air-free) Haveg 43 Rubber (up to 170°F) 10% aluminum bronze (air-free) Glass Silicon iron Hastelloy B and D Durimet 20 (up to 150°F) Worthite (up to 150°F) Lead Monel (air-free) Steel Glass Silicon iron Hastelloy B and D Lead (up to 96% H2SO4) Durimet 20 Worthite Glass Silicon iron Hastelloy B and D Durimet 20 (up to 150°F) Worthite (up to 150°F) Glass Silicon iron Hastelloy B and D (20–50 mpy) Glass Silicon iron Tantalum Glass Steel 18Cr-8Ni Durimet 20 Glass 18Cr-8Ni Durimet 20 Glass Gold

Zone 2 Ni-Resist (up to 20% at 75°F) Impervious graphite Tantalum Gold Platinum Silver Zirconium Nionel Tungsten Molybdenum Type 316 stainless (up to 25% at 75°F) aerated Zone 3 Impervious graphite Tantalum Gold Platinum Zirconium Molybdenum Zone 4 Ni-Resist Type 316 stainless (above 80%) Impervious graphite (up to 96% H2SO4) Tantalum Gold Platinum Zirconium Zone 5 Lead (up to 175°F and 96% H2SO4) Impervious graphite (up to 175°F and 96% H2SO4) Tantalum Gold Platinum Zone 6 Tantalum Gold Platinum Zone 7 Gold Platinum Zone 8 Worthite Hastelloy C Gold Platinum Zone 9 Worthite Gold Platinum Zone 10 Platinum

137

138

Chapter 4

Figure 4.4 Relative performance of steel and cast iron as a function of time.

Numerous salts and salt solutions can be handled by cast iron without excessive corrosion. Cast iron is also used extensively in the exhausts of flue gas systems where the formation of sulfurous acid on surfaces below the dew point of the flue gases at the cold end is less detrimental for cast iron than steel because thicker sections can be used.

4.1.2

Carbon and Low-Alloy Steels

Carbon steels have poor corrosion resistance. They exhibit high rates of corrosion in nearly all aqueous environments including atmospheric exposures. The hydrated oxide corrosion product produced on atmospheric exposures is called rust. This being flaky and nonadherent, corrosion continues unabated. Carbon steels, however, are the most economical choice to handle sulfuric acid above concentrations of 65% because of the formation of the passive ferrous sulfate protective film. Carbon steels are also passive in alkaline solutions of high pH and are used to handle such solutions at ambient temperatures. Carbon steels rebars are used to reinforce concrete and are often used bare as the leaching of concrete provides an alkaline medium. However, coating or cathodic protection of rebars is necessitated in structures exposed to marine conditions or to aggressive salts, as well as to prevent stress corrosion cracking of the highly stressed rebars by the accumulated hydroxyl ions. Carbon steels are nevertheless the most widely used structural material because of their wide range of mechanical properties developed through compositional variation and heat treatment. Tanks, pipelines, railroad wagons, automobiles, ship hulls, and marine structures are a few applications of carbon steels.

Aqueous Corrosion: Prevention

139

All that is needed to save them from corrosion is to provide suitable corrosion protection measures. These measures have been discussed in the sections to follow, mostly with reference to steels. Low-alloy steels have principally been developed for improved mechanical properties. They normally contain alloying elements like Ni, Cr, Mo, V, Nb, and Ti with a total content of not more than 2%. The corrosion behavior of such steels in aqueous media is not very different from that of carbon steels. However, the addition of Cu, Si, Cr, Ni, and P improves the resistance to atmospheric corrosion because the rust produced becomes nonporous and adherent. Copper plays a definite role in altering the morphology of the rust and is used to the extent of 0.1–0.4% in all weathering steels.

4.1.3

Stainless Steels

Stainless steels are basically iron-chromium alloys containing a minimum of 11% Cr, the amount necessary to keep the alloy in the passive state in normal oxidizing exposures. Chromium apart, nickel (or manganese) is another principal alloying element in several grades of stainless steels. Molybdenum, titanium, niobium, silicon, and copper are also added to several grades for specific effects. Stainless steels are available both as wrought and cast products and are of four principal types: ferritic, martensitic, austenitic, and precipitation hardening. Duplex stainless steels containing both ferrite and austenite constitute the fifth group. The wrought varieties are known generally by their numerical AISI (American Iron and Steel Institute) designations (Table 4.1). The cast varieties contain 1–2.5% silicon additionally for better castability. The ferritic stainless steels are characterized by their ferrite microstructure. For low-carbon iron-chromium alloys the high-temperature austenite phase exists only up to 12% Cr; beyond this composition the alloys are ferritic at all temperatures up to the melting point. They can be hardened moderately by cold working but not by heat treatment. Ferritic stainless steels with straight chromium were the first stainless steels to be developed. The low solubility of carbon and nitrogen in the ferrite phase tends to render them easily sensitized, hence embrittled. Welding has also been a problem with the conventional ferritic steels. These difficulties have been overcome in the recently developed high-purity ferritic stainless steels containing the total interstitials below 250 ppm. These grades contain high chromium (26–29%) with 1–4% molybdenum and in some varieties up to 2% nickel. Typical applications of ferritic stainless steels are architectural and automobile trims, tableware, ammonia oxidation plants for making nitric acid, tank cars and tanks for storage of nitric acid, thin-wall tubings for heat exchangers, and other high-temperature applications. The martensitic stainless steels contain a higher percentage of carbon than ferritic steels, so that the austenite phase is available at higher temperatures and

140

Table 4.1

Chemical compositions of stainless steels UNS number

%C

%Cr

410

S41000

0.15 max

11.5–13.5

416 420 431 440A

S41600 S42000 S43100 S44002

0.15 max 0.35–0.45 0.2 max 0.60–0.75

405 430 442 446

S40500 S43000 S44200 S44600

0.08 0.12 0.25 0.20

max max max max

201 202 301 302 302B

S20100 S20200 S30100 S30200 S30215

0.15 0.15 0.15 0.15 0.15

max max max max max

304 304L

S30400 S30403

0.08 max 0.03 max

%Ni

% Other elements

12–14 Se, Mo, or Zr 12–14 15–17 1.25–2.5 16–18 Group II Ferritic nonhardenable steels 11.5–14.5 0.5 max 0.1–0.3 Al 14–18 0.5 max 18–23 0.5 max 23–27 0.5 max 0.25N max Group III Austenitic chromium-nickel steels 16–18 3.5–5.5 5.0–7.5 Mn, 0.25N max 17–19 4–6 7.5–10 Mn, 0.25N max 16–18 6–8 2 Mn max 17–19 8–10 2 Mn max 17–19 8–10 2–3 Si 18–20 18–20

8–12 8–12

1 Si max 1 Si max

Remarks Turbine blades, valve trim ‘‘Free’’ machining Cutlery Improved ductility Very hard: cutters Al prevents hardening Auto trim, tableware Resists O and S at high temperatures Mn substitute for Ni Mn substitute for Ni Strain hardens Architectural uses Si for high-temp oxidation Continuous 18–85 Very low carbon

Chapter 4

AISI type

S30800 S30900 S30908 S31000 S31008 S31400

0.08 max 0.2 max 0.08 max 0.25 max 0.08 max 0.25 max

316 316L 317 321 347 Alloy 20a

S31600 S31603 S31700 S32100 S34700 J95150

0.10 0.03 0.08 0.08 0.08 0.07

322 17-7PHb 17-4PHb 14-8MoPHb AM350b CD4MCuc

S17700 S17400 S13800 S35000

0.07 0.07 0.05 0.05 max 0.10 0.03

max max max max max max

19–21 22–24 22–24 24–26 24–26 23–26

10–12 12–15 12–15 19–22 19–22 19–22

1 Si max 1 Si max 1 Si max 1.5 Si max 1.5 Si max 1.5–3.0

16–18 10–14 2 3 Mo 16–18 10–14 2–3 Mo 18–20 11–14 3–4 Mo 17–19 8–11 Ti 4 ⫻ C(min) 17–19 9–13 Cb ⫹ Ta 10 ⫻ C(min) 29 20 3.25 Cu, 2.25 Mo Group IV Age-hardenable steelsa 17 7 0.07 Ti, 0.2 Al 17 7 1.0 Al 16.5 4.25 4.0 Cu 14 8.5 2.5 Mo, 1% Al 16.5 4.3 2.75 Mo 25 5 3.0 Cu, 2.0 Mo

‘‘High’’ 18–8 25–12, heat resistance Lower carbon 25–20, heat resistance Lower carbon Si for high-temp oxidation 18–8S Mo Very low carbon Higher Mo Ti stabilized Cb stabilized Best corrosion resistance

Aqueous Corrosion: Prevention

308 309 309S 310 310S 314

a

Typical compositions. Commercial designations. c Cast form only. Source: M. G. Fontana, Corrosion Engineering, 3rd ed., McGraw-Hill, New York, 1987. b

141

142

Chapter 4

quenching from these temperatures produces martensitic microstructure. These steels are used where high strength or hardness along with moderate corrosion resistance is required. Typical applications are ball bearings, surgical instruments, cutlery, cutters, valve parts, turbine blades, nuts and bolts, and the like. The austenitic stainless steels retain austenite at room temperature because of the addition of nickel and/or manganese and nitrogen. These are readily deformed and can be hardened by cold working. Austenitic stainless steels are in general more corrosion-resistant than the ferritic and martensitic grades. This coupled with their good weldability has made austenitic stainless steels the principal choice for structural units in process industries. Other typical applications include kitchenware, automobile parts, pump shafts, fasteners, architectural trims in industrial and marine atmospheres, and cryogenic applications. The varieties containing molybdenum are better suited in marine applications for their resistance to pitting. The age-hardening or precipitation-hardening stainless steels are alloys of higher mechanical strength because of the precipitation of intermetallic phases on aging after quenching from high temperature. Copper and aluminum are added for this purpose. These alloys usually have poorer corrosion resistance than the other types and are used in relatively mild corrosive environments. Typical applications are aircraft and missile components. The duplex stainless steels contain both ferrite formers (Cr, Mo) and austenite stabilizers (Ni, Mn) in such amounts as to have a favorable combination of both phases. These alloys are particularly resistant to stress corrosion cracking (SCC). Although the stainless steels have been developed for corrosion resistance, they nevertheless are not the answer to all corrosion problems. Rather, they are more susceptible to localized attack like pitting, crevice corrosion, and SCC than carbon steels, particularly in chloride-containing media. Moreover, there is a misnomer that nonmagnetic stainless steels have better performance as corrosionresistant materials. Some alloys belonging to the nonmagnetic austenitic group become magnetic on cold working because of the transformation of the metastable austenite to ferrite, but this does not hamper their corrosion resistance. Stainless steels are, in general, resistant to nitric acid over a wide range of concentrations and temperatures, aerated dilute sulfuric acid at room temperature, sulfuric acid of somewhat higher concentration (e.g., 10%) and at boiling temperature if Fe3⫹, Cu2⫹, or HNO3 is added as inhibitor, or at lower temperature if small amounts of Cu, Pt, or Pd are alloyed. They can handle sulfuric acid of all concentrations up to the boiling point, if anodically protected. They are resistant to many organic acids. Alkalis do not attack stainless steels, but some varieties are susceptible to SCC in hot concentrated caustic solutions. Stainless steels do not rust in atmospheric exposures. Stainless steels are not resistant to reducing acids HCl, Br, or HF because the passive film is damaged in these media. Stainless steels are not suitable for sea-

Aqueous Corrosion: Prevention

143

water application if not protected otherwise because they develop pitting within months. Austenitic grades are better than other grades in this respect, particularly the molybdenum-containing varieties (e.g., AISI 316), but they too develop pitting over a period of 1 to 2.5 years. An exception is 29Cr-4Mo grade, which is comparable to titanium in its performance in seawater. Pitting in solutions containing FeCl3, CuCl2, or HgCl2 develops within hours at room temperature, but this can be inhibited with the addition of NO3⫺. Stainless steels, both austenitic and ferritic, are susceptible to sensitization and subsequent intergranular corrosion, though their thermal history is different (Section 3.6). Austenitic stainless steels are particularly susceptible to SCC in the presence of Cl⫺ and OH⫺. Ferritic and duplex stainless steels are resistant to SCC and are used where it is a problem.

4.1.4

Aluminum and Its Alloys

Aluminum is a reactive metal, but it develops a thin protective oxide layer when exposed to air or aerated environments that resists further corrosion. The oxide is stable in the neutral or near-neutral ranges of aqueous solutions but is attacked by strong acids and alkalis. Aluminum is a soft, light metal that can be rolled down to very thin sheet or foil. Since aluminum does not rust or lose its luster on atmospheric exposure and is resistant to mild acidic conditions, the sheets and foils are used for architectural trims, vessels in food processing and pharmaceutical units, cans for storage and transportation of food and beaverages, and as a packaging material. The strength of aluminum is remarkably increased by alloying additions with some sacrifice of corrosion resistance and these alloys constitute a wide range of structural materials. The wrought alloys are designated by four-digit numbers, i.e., 1000 to 7000 series depending on the alloying elements present (Table 4.2). The 2000 series age-hardenable alloys containing about 4% copper possess the highest strength and the lowest corrosion resistance, particularly to pitting and SCC. The SCC is encountered in sodium chloride solution, seawater, and even in air and water vapor. The susceptibility SCC can be reduced by overaging. Alclading is another measure adopted to prevent SCC in these alloys. Aluminum alloys have extensive applications in aircraft and aerospace vehicles, automobiles, window and door frames, roofing, cold and hot storage vessels and pipings, food processing equipment, and household cooking utensils.

4.1.5

Copper and Its Alloys

Copper is the noblest among the common structural metals. It exhibits good resistance to atmospheric corrosion and aqueous environments. In reducing acids copper does not corrode in the absence of oxygen or oxidizing agents because the

144

Chapter 4

Table 4.2 Alloy class

Wrought aluminum alloys and their corrosion behavior Corrosion Resistanceb

Typical temper a

Alloying elements

General

Pitting

Exfoliation

SCC

E

E

E

I

F F E F E E F F

P P E G G G F F

P F E G G E F-P G

VS R I G IR 1 S-VS R

1xxx

All

2xxx

T3, T4, T8

Natural impurities in refinery Al Cu

3xxx 4xxx 5xxx 6xxx 7xxx

All All Most All T6, T73

Mn, Mn ⫹ Mg Si Mn, Mg, Cr Mg, Si Zn, Mg, Mn, Cu

a

T3, T4, T6: age hardened; T8, T73: overaged. E, excellent; G, good; F, fair; P, poor; I, immune; R, resistant; S, susceptible; VS, very susceptible. Source: Ref. 3. b

equilibrium potential of the copper dissolution reaction is far more noble than that of the cathodic hydrogen evolution reaction. However, the following cathodic reactions are possible in the presence of oxygen and oxidizing agents, and anodic dissolution of copper is facilitated (see Table 2.1): O2 ⫹ 4H⫹ ⫹ 4e ⫽ 2H2O Fe ⫹ e ⫽ Fe O2 ⫹ 2H2O ⫹ 4e ⫽ 2OH⫺ 3⫹

2⫹

(2.7) (2.8) (2.11)

Copper and its alloys are, in general, resistant to seawater, neutral solutions, and alkalis with the exception of ammonia which increases general corrosion through complex formation, as well as SCC in alloys. Commercially pure copper has traditionally been used for the handling of potable water and for roofing of churches and monuments. The green copper sulfate formed on weathering on the roofs is termed ‘‘patina,’’ which provides its characteristic aesthetic appearance. Pure copper, however, is soft and is very much susceptible to impingement attack. Copper alloyed with zinc, tin, nickel, and aluminum are stronger and are widely used; the nominal compositions of some of these are given in Table 4.3. Brass are copper-zinc alloys with zinc content ranging from 10% to 40%. High brasses (with higher zinc content) are prone to dezincification and SCC in ammonia. Addition of 1% Sn and a small amount of P, As, or Sb improves the resistance of brass to dezincification. Bronzes are copper-tin alloys containing

Aqueous Corrosion: Prevention

145

Table 4.3 Some common copper alloys and their nominal chemical composition Material

Composition (%)

Beryllium copper Red brass Cartridge brass (yellow brass) Admiralty brass Aluminum brass Muntz metal Phosphor bronze Aluminum bronze Cupronickels Nickel silver

98 Cu, 1.9 Be, 0.2 Ni or Co 85 Cu, 15 Zn 70 Cu, 30 Zn 71 Cu, 28 Zn, 1 Sn, 0.04 As 77 Cu, 21 Zn, 2 Al, 0.04 As 60 Cu, 40 Zn 90–95 Cu, 5–10 Zn, 0.25 P 92 Cu, 8 Al 69–88 Cu, 10–30 Ni, 0.5–1 Fe, 0.5 Mn 65 Cu, 25 Zn, 10 Ni

Source: Adapted from: M. G. Fontana, Corrosion Engineering, 3rd ed., McGraw-Hill, New York, 1983.

5–10% tin, often with addition of a small amount of phosphorus. They are more resistant to SCC and impingement attack than brasses. Aluminum bronzes with 5–8% aluminum and cupronickels containing 10–30% nickel with small iron additions have superior resistance to erosion corrosion. Copper alloys are used extensively as heat exchanger tubes because of their high thermal conductivity, and resistance to erosion corrosion is an important consideration in such applications. Copper alloys are also used for piping, valves, pumps, gears, bushings, shafts, tanks, and other vessels.

4.1.6

Nickel and Its Alloys

Nickel is inherently resistant to caustic solutions at all concentrations and temperatures, seawater, and slightly acidic solutions. Addition of nickel to cast irons, steels, and stainless steels improves the corrosion performance of these alloys. However, nickel is not resistant to strongly oxidizing solutions, e.g., nitric acid and ammonia solutions. At high temperatures, nickel is attacked by sulfur-bearing gases and is embrittled. Nickel alloys constitute a major group of corrosion-resistant alloys. Some important nickel alloys are listed in Table 4.4. Monel, containing about 30% copper, is considered to be a natural for handling hydrofluoric acid. A host of alloys have been developed with chromium and molybdenum as principal alloying elements. Chromium addition induces passivity and improves corrosion resistance in oxidizing solutions, including nitric acid. Nickel-chromium-molybdenum alloys (Hastelloy) are best suited to handle hydrochloric acid at all concentrations.

146

Chapter 4

Table 4.4 Some common nickel alloys and their nominal chemical composition Material

Composition (%)

Monel Inconel 600 Chlorimet 2 Chlorimet 3 Hastelloy B Hastelloy C Hastelloy D Incoloy 800 Duranickel 301 Carpenter

63–66Ni, 29–31Cu, 1.4–2Fe 76Ni, 16Cr, 8 Fe, 0.2Cu 62Ni, 32Mo, 3 Fe, 1Si 60Ni, 18Cr, 18Mo, 3Fe, 0.6Si 62Ni, 28Mo, 5Fe, 1Si 56Ni, 15Cr, 17Mo, 5Fe, 1Si, 4W 85Ni, 3Cu, 10Si 32Ni, 21Cr, 46Fe 94Ni, 0.05Cu, 0.15Fe, 4.5Al, 0.5Ti 35Ni, 20Cr, 25Mo, 3Cu

Source: Adapted from M. G. Fontana, Corrosion Engineering, 3rd ed., McGraw-Hill, New York, 1983.

Nickel alloys are also resistant to sulfuric acid at intermediate and higher concentrations. The 9–11 silicons are resistant to hot concentrated sulfuric acid. Nickelchromium-iron alloys with or without molybdenum addition are more resistant to pitting, crevice corrosion, and SCC in chlorides than austenitic stainless steels. Inconel (76Ni-16Cr-8Fe) has excellent resistance to SCC. Nickel alloys are extensively used in chemical process plants, oil refineries, acid-producing plants, and in seawater applications in the form of pipings, pumps, tanks, vessels, and so forth.

4.1.7

Titanium and Its Alloys

Titanium, like aluminum, is a highly reactive metal that owes its corrosion resistance to the adherent and highly protective oxide film developed on exposure to oxygen and moisture. Titanium shows excellent corrosion resistance to many aggressive solutions, i.e., nitric acid including fuming nitric acid; seawater and chloride solutions including those containing heavy oxidizing chlorides such as FeCl3 and CuCl2; wet chlorine; oxidizing acids like chromic and perchloric acids; weak reducing acids like sulfurous, boric, and carbonic acids; and titanium is resistant to alkalis. In strong reducing acids, like sulfuric and hydrochloric acids, titanium performs well in the presence of oxidizing inhibitors. However, titanium has a pyrophoric tendency in red fuming nitric acid and is attacked by hydrofluoric acid and fluoride solutions. Titanium has the interesting properties of low density, high melting point, and outstanding corrosion resistance. Its density is intermediate of aluminum and steel

Aqueous Corrosion: Prevention

147

and possesses a higher strength-to-weight ratio compared to aluminum and steel. This provides a distinct advantage in automotive and aerospace industries. However, in high-temperature applications care should be taken to avoid adsorption of gases, which makes it brittle. Commercially pure titanium is well suited to many corrosion applications. Titanium is alloyed for strength with Al, V, Mn, Cr, and Sn and the alloys have comparable corrosion resistance. Ti-6Al-4V is a widely used alloy. Titanium alloys, however, suffer from SCC in many environments, i.e., in halides at temperatures of 287–427°C and at room temperature in NaCl solution, methanol/ HCl solution, N2O4, and red fuming nitric acid. Ti-0.2Pd and Ti-5Ta alloys have been developed as special corrosion-resistant materials that are extremely successful in handling reducing acids like sulfuric and hydrochloric acid. An alloy with 30% Mo is also resistant to hydrochloric acid. The major applications of titanium and titanium alloys in chemical industry are as follows: 1. 2. 3. 4. 5. 6.

Titanium anodes in the chloralkali industry Pressure vessels and columns in chemical, urea, and fertilizer plants Desalination plants Heat exchangers in coastal power stations Synthetic fiber industry Electroplating industry for various tanks used in nitric acid industry, pickling, nitriding baths, etc. 7. In electrometallurgy plants as anodes for plating of metals Besides these major uses, titanium is used in automotive parts, in ship building, as auxiliary anodes in cathodic, protection, and in medical implants and equipment. As titanium gives vivid interference colors when subjected to controlled oxidation, it is used in fancy applications such as spectacle frames, tennis rackets, and jewelery.

4.2 CONTROL OF ENVIRONMENT In many cases corrosion can be effectively reduced through the change or modification of the corrosive environment. This can be achieved by: 1. A change of operating variables 2. Removal of corrosive constituent, and 3. Use of inhibitors

4.2.1

Change of Operating Variables

Temperature, velocity, and pH are the operating variables that can sometimes be suitably modified to reduce corrosion. Since the rate of corrosion increases at

148

Chapter 4

higher temperatures, lowering of temperature reduces the corrosion rate. The severity of erosion corrosion can be reduced by decreasing the velocity of fluid flow. On the other hand, pitting can be reduced in passivating metals and alloys by avoiding stagnant conditions and ensuring some fluid flow. Potential pH diagrams indicate immunity or passivity of metals at certain pH ranges, and an effective means of corrosion control is to alter the pH of the electrolyte to such pH values. For example, iron attains passivity and does not corrode at pH above 8.5. The addition of alkali is a widely used practice in the treatment of boiler feedwater, industrial and municipal waters. Many natural and municipal waters contain calcium carbonate, which reacting with carbon dioxide is converted to soluble calcium bicarbonate: CaCO3 ⫹ CO2 ⫹ H2O i Ca(HCO3) 2

(4.1)

Addition of lime causes insoluble calcium carbonate to precipitate according to the reaction: Ca(HCO3)2 ⫹ Ca(OH)2 → 2CaCO3 ⫹ 2H 2O

(4.2)

The deposit of CaCO3 forms a hard and smooth protective film at a pH where the precipitation just begins to occur. At higher pH values, the deposit is slimy and porous. The difference between the pH of the water and the pH required to precipitate CaCO3 is referred to as Langlier index, which often is used to express the condition of a water with respect to its tendency to deposit CaCO3. The boiler tubes in oxygen-free hot water forms a protective film of magnetite according to the reaction: 3Fe ⫹ 4H2O → Fe3O4 ⫹ 4H2

(4.3)

The formation of the film is ensured by maintaining the pH of the feedwater at a level of 9.5–11 with the addition of NaOH. In some high-pressure boilers, NH3 is added instead of NaOH and the pH is maintained at 8.5–9.0. However, excess alkali is damaging as the corrosion rate increases rapidly at pH values above 13 (Fig. 4.5). Since steel is resistant to concentrated sulfuric and phosphoric acids at moderate temperatures, increasing the concentration of acid, wherever permissible, constitutes an effective means to reduce corrosion.

4.2.2

Removal of Corrosive Constituents

The corrosive tendency of the environment sometimes increases greatly in the presence of certain corrosive constituents or chance contaminants. Chloride ions present in water coolant are responsible for corrosion of nuclear reactors and their elimination reduces corrosion. When such oxidizers as FeCl3 or CuCl2 are present in hydrochloric acid, the corrosion of steel increases because of the additional cathodic reaction provided by the reduction of these oxidizers. The contam-

Aqueous Corrosion: Prevention

149

Figure 4.5 Corrosion of iron by water at various pH values.

inated acid is also highly corrosive toward nickel-molybdenum alloys, which possess excellent corrosion resistance in pure hydrochloric acid. Removal of such oxidizers naturally reduces corrosion. Moisture is an aggressive corrosive constituent in the atmosphere. Lowering of relative humidity of air by increasing the temperature by 6–7°C above ambient in storage areas brings down the rate of corrosion. Removal of moisture by use of silica gel in small closed spaces reduces corrosion. Oxygen is a strong cathodic depolarizer and the presence of dissolved oxygen is the cause for corrosion of steel in waters above pH 6.0. Neutral water contains about 8 ppm of dissolved oxygen at 20°C, whereas only 0.1 ppm of oxygen is required to increase corrosion rates in a dynamic system, e.g., in boilers. The removal of oxygen from feedwater is therefore a necessary step in boiler operations. In high-pressure boilers, the maximum allowable oxygen concentration in feedwater is 0.005 ppm. Removal of dissolved oxygen from water is accomplished either by deaeration or by deactivation. Deaeration is the process of distilling off of the oxygen in suitable equipment and deactivation refers to the removal of oxygen by chemical reaction. Deaeration is carried out by spraying water countercurrent to steam, by inert

150

Chapter 4

gas purging, or by vacuum treatment. The older practice of deactivation was to pass the water over hot steel plates or scraps, the oxygen dissolved in water getting consumed by reacting with iron in the process. The modern practice is to use oxygen scavengers or ion exchange resins. Sodium sulfite and hydrazine are the common oxygen scavengers. They react with oxygen according the following reactions and remove oxygen in the process: 2Na2SO3 ⫹ O2 → 2Na2SO4 N2H4 ⫹ O2 → N2 ⫹ 2H2O

(4.4) (4.5)

One ppm of oxygen requires 7.3 ppm of sodium sulfite. The reaction rate of sulfide with oxygen at low temperature is increased with the addition of catalysts like cobalt, manganese, or copper salts. Similarly, hydrazine also requires catalysts like activated charcoal, metal oxides, and alkaline solutions of Cu2⫹ and Mn2⫹ for speeding up of the reaction at low temperatures. Hydrazine is preferred in high-pressure boilers because it reacts faster than sodium sulfide at elevated temperatures and because it does not increase the salt content of the boiler as its reaction products are all volatile. Ion exchange resins are available with metal sulfites, ferrous hydroxide, or manganous hydroxide, which react rapidly with oxygen. Reduction of the oxygen level to below 0.002 ppm is achieved through the use of such resins. The used resins can be regenerated by chemical treatment.

4.2.3

Inhibitors

An inhibitor is a chemical substance that when added in small concentrations to an environment decreases the corrosion rate. The amount of inhibitor required for an effective decrease in corrosion rate may be as low as a few ppm; a few hundreths are often satisfactory, and rarely exceeding 1%. Hundreds of inhibitors, both inorganic and organic compounds, are available and are being employed for corrosion prevention in various water systems, in acid pickling solutions, and in oil production and refining. Some common inhibitors used for such purposes are summarized in Table 4.5. A special class of inhibitors, called vapor phase inhibitors, are used in closed boxes and packages for protection against atmospheric corrosion during storage and transit. Inhibitor for a particular metal in a particular environment may not be effective for the same metal in another environment or for another metal in the same environment, and at times it increases the corrsion rate of another metal. A combination of two or more inhibitors sometimes is more effective than a single inhibitor (synergistic effect). The overall effectiveness of an inhibitor depends on corrosivity, pH, and temperature of the solution. Use of inhibitors is preferred in closed and recirculating systems. However, in once-through systems like municipal wa-

Aqueous Corrosion: Prevention

151

Table 4.5 Some systems and inhibitors used for protection of steel components Systems Waters: Potable water Seawater Brines Recirculating cooling water Automotive coolants Vapor condensates Acids: Sulfuric acid Hydrochloric acid

Phosphoric acid Oil production: Oil field brines Primary and secondary recoveries Refining

Inhibitors Ca(OH)2, polyphosphates, silicates Na-nitrite, na-silicate, NaH2PO4 ⫹ NaNO2 Na-chromate, Na-benzoate, na-nitrite Chromate, nitrate, polyphosphates, silicates, merpholine Nitrite, benzoate, borax, benzotriazole, mercaptobenzothiazole Morpholine, cyclohexylamine, ammonia, ethylenediamine, benzylamine Phenylacridine, phenylthiourea, mercaptans, sulfides, aromatic amines, arsenic Ethylaniline, quinoline, various amines, phenylthiourea, pyridine ⫹ phenylhydrazine, mercaptobenzothiazole NaI, thiourea, sulfonated castor oil, As2O3 Na-silicate, Na2SO3, amine acetates, formaldehyde Imidazoline, various amines Imidazoline and derivatives

ter supply and industrial pipelines use of inhibitor is also common, where the choice is obviously the inexpensive chemicals. Some inhibitors like chromates and arsenic compounds are toxic, and these cannot be used in potable water supply or where the effluent might cause pollution problems. Based on their mechanism of inhibition, the inhibitors may be classified as 1. 2. 3. 4.

Passivators Cathodic inhibitors Organic inhibitors, and Vapor phase inhibitors

However, the demarcation is not always very sharp and more than one mechanism may be operative in the action of an inhibitor. Passivators Passivators or passivating inhibitors or oxidizers, as they are sometimes called, constitute the most important group of inhibitors for the protection of steel. There are two types of passivators: oxidizing and nonoxidizing. Chromates, nitrites,

152

Chapter 4

and nitrates belong to the first group. They are effective even in the absence of oxygen, as they themselves are readily reduced. Tungstates, molybdates, phosphates, silicates, borates, and benzoates are examples of nonoxidizing passivators. They require dissolved oxygen in order to inhibit corrosion. Passivators are essentially anodic inhibitors. The oxidizing passivators get adsorbed and the nonoxidizing passivators facilitate the adsorption of dissolved oxygen on the metal surface. The redox potential of the reduction system shifts progressively in the nobler direction with the increase in passivator concentration. At the same time, the rapid reduction of passivators or oxygen accelerates the anodic dissolution of steel (an active-passive metal) to a value above the icritical, taking the anodic polarization to the passive region. The intersection of the cathodic polarization curve with the anodic polarization curve in this region gives a corrosion rate equal to ipassive, which is of insignificantly small value (Section 2.4). To ensure this situation, the passivator should be used in sufficient concentration; otherwise, the intersection of the polarization curves will be in the active region giving a higher corrosion rate. Since the passivators accelerate corrosion when added in insufficient amount, they are also known as ‘‘dangerous inhibitors.’’ The situations are shown in Fig. 4.6. A shift from 1 to 2 increases the corrosion rate from A to B, whereas for situation 3, the cathodic polarization curve clears the nose of the anodic polarization curve and passivity is achieved. The critical concentration of passivators is usually of the order of a few ppm

Figure 4.6 Effect of passivator concentration on corrosion of iron.

Aqueous Corrosion: Prevention

153

to 500 ppm. At higher temperatures and higher hydrogen ion activity, icritical increases and the critical concentration of passivators also rises. Chloride and sulfate ions compete with the passivating ions for adsorption, thus interfering with the process of passivation. The local breakdown of passivity in their presence leads to pitting and higher amounts of passivators are required to prevent this. Passivators are used extensively for the protection of steel in all types of water. However, the choice of passivators varies from system to system. Chromates are almost the inevitable choice for recirculating cooling waters, but they are not to be used in antifreeze cooling waters because they tend to react with alcohols or ethylene glycol. On the other hand, nitrites are not used in cooling tower waters because they are gradually decomposed by bacteria. Chromates are toxic, which restricts their application. For potable waters, nontoxic silicates or polyphosphates are used. Cathodic Inhibitors Cathodic inhibitors interfere with the cathodic processes and the rate of corrosion is thereby decreased. They fall into three categories: cathode precipitates, oxygen scavengers, and hydrogen evolution poisons. Cathode precipitates and oxygen scavengers have been discussed in Sections 4.2.1 and 4.2.2, respectively. Calcium and magnesium carbonates, which are often present in natural waters, can be precipitated to form protective cathodic deposits with the adjustment of pH. The addition of zinc sulfate also inhibits corrosion by precipitating insoluble Zn(OH)2 at increased alkalinity on the cathodic areas according to the reaction: ZnSO4 ⫹ 2NaOH → Zn(OH) 2 ⫹ Na2SO4

(4.6)

Hydrogen evolution poisons interfere with the formation of hydrogen gas (2Hads → H2) to retard the overall rate of the cathodic reaction of hydrogen evolution. The corrosion rate is consequently decreased. Sulfides, selenides, and compounds (usually oxides) of arsenic, antimony, and bismuth act as hydrogen evolution poisons. They are effective inhibitors in strong acids where the hydrogen evolution reaction is rate controlling in the corrosion process. One difficulty is presented by these inhibitors: They cause blistering and hydrogen embrittlement in certain grades of steel because of the entry of the atomic hydrogen into the metal. Also arsenic being toxic is restricted in use. Organic Inhibitors A good number of organic compounds are used as inhibitors. Common among them are amines, imines, thiourea, mercaptans, guanidine, and aldehydes. The compounds are chemisorbed on the metal surface forming a monolayer that interferes with both the anodic and cathodic processes, although in many cases the effect is unequal. The chemisorption is effected through the presence of a polar

154

Chapter 4

group in the molecular structure by which the molecules can attach themselves to the metal surface. These include sulfur, nitrogen, amine, phosphorus, and hydroxyl groups. Cationic inhibitors, such as amines, or anionic inhibitors, such as sulfonates, will be adsorbed preferentially depending on whether the metal is charged negatively or positively, respectively, with respect to the solution. Amines show better performance as inhibitors for steel in phosphoric acid in the presence of iodides. The explanation for this synergism is that the adsorption of iodide ions shifts the surface charge of steel to more negative values where the adsorption of positively charged amines is favored. The fact that a certain organic compound acts as a good inhibitor for some metal but not another is explained from this specific electronic interaction of the polar groups with the metal surface. Molecular structure and size of the organic compound influence their inhibitive action. The structures with benzene rings are particularly effective inhibitors. Primary amines become more efficient as the chain length is increased. This is presumably because of the steric effect, i.e., diffusion barrier provided by long chains. However, for mercaptans and aldehydes, the efficiency is decreased with the increase in chain length. The sharp decrease in corrosion rate with organic inhibitors is shown in Fig. 4.7. Organic inhibitors find wide application as pickling inhibitors. Acid pickling of hot-rolled steel is necessary to remove mill scale. Pickling inhibitors resist corrosion of the substrate metal. These inhibitors are also used for the acid cleaning of the pipes clogged with rust or to remove limestone crust from inside the boiler tubes. Typical examples of pickling inhibitors are quinolin ethiodide, o- and p-tolythiourea, hexamethylene tetramine, formaldehyde, and p-thiocresol. They are added in the concentration of 0.01–0.1%. Organic inhibitors are also added to oils, greases, and waxes used as slushing compounds to temporarily protect steel surfaces from rusting during shipment or storage. Vapor Phase Inhibitors Vapor phase inhibitors (VPIs) are compounds with low vapor pressure (0.0002– 0.4 mm Hg). In a closed system, they volatilize and the vapor condenses on the metal surface to provide protection. In boilers, volatile basic compounds such as morpholine and ethylenediamine are transported with steam to the condenser tubes that prevent corrosion of the tubes by neutralizing carbonic acid and making the environment alkaline. In closed containers and packages, volatile solids such as the nitrite, carbonate, and benzoate salts of dicyclohexylamine, cyclohexylamine, and hexamethylene imine are used for temporary protection of critical machine parts, ball bearings, cold-rolled steel coils, etc., during storage or transportation. Dicyclohexylamine nitrite is a widely used vapor phase inhibitor that is often impregnated in the waxed paper or cardboard used for wrapping and packaging.

Aqueous Corrosion: Prevention

155

Figure 4.7 Effect of concentration of organic inhibitors on corrosion rate.

One gram of this inhibitor saturates about 550 M3 of air and the protection to steel is provided over years. Cyclohexylamine carbonate has higher vapor pressure and is used in packages that must be opened and closed repeatedly. The mechanism of inhibition is not the same for all vapor phase inhibitors. Nitrite ions and benzoate ions in association with the oxygen present passivate the steel surface. Carbonate provides alkalinity to the environment and the organic amine portion of the inhibitor effectively provides protection through adsorption. While the vapor phase inhibitors, in general, are effective in prevention of corrosion in steel, they accelerate the corrosion of some nonferrous metals. The vapor phase inhibitors based on nitrobenzoate organic compounds have been reported to protect ferrous, copper, and other alloy systems.

4.3 PROTECTIVE COATINGS The application of a coating, metallic or nonmetallic, on steel and other metals is an age old and widely used practice for protection against corrosion. The primary objective of a coating is to provide a physical barrier between the metal and the

156

Chapter 4

environment so that they do not come in contact with each other. This is achieved fully with some types of coatings and partially with others. Apart from this, the coatings may release substances that inhibit the corrosive attack of the environment on the substrate or the coatings may act as a sacrificial anode (Section 3.3.3) to protect the substrate cathodically. However, it is to be remembered that all coatings have their own life, i.e., they will not resist the attack permanently. Nonmetallic coatings are divided into two groups: inorganic and organic. Inorganic coatings comprise vitreous enamel, cement, and chemical conversion coatings. Organic coatings include paints, varnishes, and lacquers. Thicker coatings of enamel, cement, tar, or rubber are sometimes applied to negotiate aggressive chemical conditions; such coatings are referred to as linings. Most of the coatings have been developed to protect steel in varied aqueous environments and the discussion that follows will be with reference to steel unless otherwise stated.

4.3.1

Metallic Coatings

Classification From a corrosion point of view, metal coatings can be divided into two classes: noble and sacrificial. Silver, copper, nickel, chromium, tin, and lead coatings on steel constitute the first group, whereas the coatings of zinc, aluminum, and cadmium belong to the second group. As discussed in Section 3.3.3, any damage or discontinuity in the noble metal coating creates a small anode–large cathode condition leading to rapid localized attack on the substrate at the damaged areas. On the other hand, such damages in the sacrificial coating will not pose a problem as the exposed substrate will be cathodic with respect to the coating metal and will be protected at the cost of the corrosion of the coating metal. Naturally, the noble metal coatings should be free from pores and this is usually achieved through an increase in the coating thickness. Reversal of polarity between zinc and steel occurs in many aerated waters above 60°C, which means that the zinc coating behaves as a noble metal coating on steel. Under the circumstances the base steel becomes vulnerable to attack at coating discontinuities. Tin is cathodic to iron, but tin coating inside food cans becomes anodic to steel because stannous (Sn2⫹) ions are complexed with the food product, thereby greatly reducing the stannous ion activity (see Nernst equation). However, the galvanic protection of steel by tin is lost in the presence of dissolved oxygen and food should not be retained inside the tin cans after opening to avoid contamination by corrosion products. Methods of Application Metallic coatings are applied by several methods, the principal among these being hot dipping, electroplating, flame spraying, cementation or diffusion-coating cladding, and vapor deposition.

Aqueous Corrosion: Prevention

157

Hot dipping is the application of coating by immersion of the metal piece in the molten bath of a low-melting-point metal like zinc, tin, aluminum, and lead. Galvanizing of steel by hot dipping in molten zinc is a widely used practice for wires, roofing sheets, and a host of other components. Galval coating, accomplished with the incorporation of aluminum in molten zinc, provides improved performance in marine conditions. Electroplating is the deposition of a metal from its aqueous salt solution by passing direct current in a cell in which the metal to be plated constitutes the cathode. Electroplating is normally suited for small components. Zinc, nickel, tin, and cadmium are plated on steel in large tonnage. Nickel plating is particularly popular for anticorrosive and decorative applications. A base of copper coating on steel is usually employed for better adhesion of nickel coating. It is often overlaid with a thin chromium coating for appearance and also for resisting the attack of sulfur-bearing industrial atmospheres. Such coating is known as ‘‘chrome plate.’’ Flame spraying is a suitable method for providing metallic coatings on large structures like bridges, ship hulls, tank cars, and vessels of all kinds. Metal wire or powder is fed to the steel surface through a moving flame and the coating is achieved. Oxygen or oxyacetylene is commonly used for the melting flame. Plasma jet is used for coatings of high-melting metals. The coatings achieved by flame spraying are usually porous, but they provide a good base for a subsequent paint coating. Flame spraying is industrially known as ‘‘metallizing.’’ Diffusion coating involves alloy formation on the metal surface by high-temperature diffusion of the coating metal. The latter may be supplied in the powder form, dissolved in molten calcium, or obtained through a gaseous environment containing the metal. Coatings of zinc, chromium, and aluminum are delivered by this method; the corresponding processes are called sherardizing, chromizing, and calorizing. Cladding refers to a surface layer of a corrosion-resistant sheet metal, usually accomplished by rolling, on a cheaper and stronger base. The cladding can also be applied by explosive bonding or spot welding. Nickel, aluminum, copper, titanium, and stainless steels are often used as cladding for steel. A lot of saving on the costlier and more corrosion-resistant materials is achieved. Cladding of high—strength aluminum alloys with commercially pure aluminum saves them from SCC. The vapor deposition processes fall into two major categories: physical vapor deposition (PVD) and chemical vapor deposition (CVD). In PVD process the coating metal is vaporized in a high-vacuum chamber and the vapor deposits on the parts to be coated. The coating can also be achieved by sputtering, which involves the transport of the coating material from a source (target) to the substrate by means of the bombardment of the target by gas ions that have been accelerated by a high voltage. There is also a third method, ion plating, which

158

Chapter 4

is a combination of evaporation and glow discharge (plasma formation). In CVD the coating metal is produced and alloyed at the surface of the workpiece through gas phase reaction with a volatilized compound. The films produced by vapor deposition are dense and pore-free; therefore, they are resistant to penetration by moisture and gases. However, the cost and complexity of the processes have restricted their industrial application for corrosion protection purposes. Nevertheless, the vapor-deposited aluminum coating is finding wide applications, particularly in the aerospace industry. The brittle intermetallic layer of iron-aluminum alloy produced in hot-dipped steel, which hampers the formability of steel, is absent in steels with vapor-deposited aluminum coating. Cadmium electroplating produces hydrogen embrittlement in some varieties of steel; replacement by vapor-deposited aluminum coating eliminates this problem. Sputtered chromium and stainless steel are also making inroads in corrosion applications. Surface Modification Surface composition or structure of a metallic part can be modified drastically with respect to the bulk material by use of directed energy or particle beams. The modification is principally metallurgical in that it involves alloy formation or the formation of an amorphous layer, in contrast to chemical conversion coatings (see section ‘‘Chemical Conversion Coatings’’) which are inorganic in nature. However, the structural changes achieved differ significantly from those obtained by conventional processes and this becomes a contributing factor to the improved performance of surface-modified metals in corrosion applications. The principal surface modification methods are ion implantation and laser processing. Ion implantation involves the deposition of ions of metals or nonmetals on the metal substrate and several techniques are employed. Both conventional and unconventional alloying have been carried out, the latter being more attractive. For example, tantalum, which cannot be alloyed easily with iron, can be easily implanted on steel. Amorphous layers produced by phosphorus or boron implantation in iron-chromium alloys increase corrosion resistance and improve passivation characteristics of the alloys. Even the implantation of chromium or molybdenum on iron to achieve the properties of stainless steel means a lot of saving on these expensive alloying elements. Similarly, the noble metal bulk alloying can be avoided by simple ion implantation, e.g., palladium on titanium. Laser processing is carried out by high-energy laser beams ranging from 0.5 to 10 kW. The surface of the metal may simply be modified to an amorphous layer by melting and rapid quenching. Alloying elements added to the beam produces surface-alloying plain-carbon steels. Laser surface alloyed with chromium and molybdenum has been reported to exhibit the same passivation behavior of the bulk alloys of corresponding composition. Laser-modified type 304 stainless steel

Aqueous Corrosion: Prevention

159

exhibits decreased critical current density for passivation in sulfuric acid and higher pitting potential compared to unprocessed material. Industrial applications of surface modification are still limited, principally because of the cost of installation. Irregular contours of the components sometimes pose a problem. However, successful applications of ion implantation have been reported in orthopedic implants and precision bearings for aircraft.

4.3.2

Inorganic Coatings

Inorganic coatings include numerous classes of materials which include cement coatings, glass or vitreous enamel coatings, ceramic coatings, silicones and silicates, and chemical conversion coatings. Cement Coatings Cements are commonly used to coat the inside and outside of steel and cast iron pipes that are burried or submerged, e.g., sewer lines. The coatings are applied by centrifugal casting and are usually thick. The coating is cured for several days before being placed in service. Addition of organic and inorganic setting agents reduces the curing time. A cement coating on steel is protective to steel because the cement contains sodium silicate or potassium silicate as a constituent, which provides an alkaline reaction. The cement-coated steel pipes in soils and water have excellent records of performance. The potassium silicate–based cements are resistant to attack in the acidic range of pH from 1 to 7. Glass Coatings Glass coatings are known by various names, such as, vitreous enamels, porcelain enamels, or—simply—glass linings. Glass powders or frits are applied to a metal surface and on heating in a furnace a glassy coating is obtained. The composition of the glass determines the corrosion resistance of the coating, but in general glass linings are highly resistant to acids (except hydrofluoric acid) and to mildly alkaline corrosives. High-silica glasses with additions of TiO2 and ZrO2 are more resistant to acids, whereas higher B2O3 content with BaO addition provides better resistance to alkalis. Since the coatings are virtually impermeable to water and oxygen, the coatings remain protective over years or decades in water and atmospheric exposures. Glass coatings are mainly applied on steel but are also used on cast iron, copper, brass and aluminum. Enameled steels have found wide applications in advertising signs, gasoline pump casings, decorative building panels and so forth because of their ability to retain color and gloss for decades in weather and sunlight. Glass-lined pipes, valves, pumps, and vessels of all kinds are widely used in chemical, pharmaceutical, and food industries, particularly where contamination by corrosion products must be avoided. Enameled kitchenwares and glass-lined hot water tanks are

160

Chapter 4

examples of domestic applications. Glass coatings are susceptible to mechanical damage and cracking by thermal shocks and as such care should be taken against such eventualities. Chemical Conversion Coatings Chemical conversion coatings are different from other types of coatings in that these are formed in situ by chemical reaction with the metal surface. Coatings comprise inorganic compounds rather than surface alloying (cf. chemical vapor deposition). Sometimes such coatings form naturally when exposed to corrosives and further corrosion is prevented. The formation of lead sulfate coating on exposure to sulfuric acid is an example. In other cases, such coatings are artificially developed for protection against subsequent corrosive exposures. Anodizing of aluminum is a prominent example of chemical conversion coating. Aluminum is inherently associated with a surface oxide film of a few micrometers thickness. The film is thickened (1–7 mils) by anodic oxidation of aluminum in a suitable electrolyte, usually dilute sulfuric acid or chromic acid, with the passage of current. Anodized coatings are hard and tightly adherent to the base. The hardness approaches that of corundum. The corrosion resistance does not improve much on anodizing, as the oxides are tubular and porous. However, the pores can be ‘‘sealed’’ by exposure to boiling water or steam for several minutes and the corrosion resistance of the scaled anodized layer will improve in a wide variety of corrosives. The anodized layer also provides a good base for application of paints, which is otherwise difficult on aluminum. Magnesium can also be anodized. Phosphatizing of steel is another important example of chemical conversion coating. The steel surface is treated with a cold or hot solution of dilute zinc or magnesium phosphate or orthophosphate with phosphoric acid. A network of porous metal phosphate is produced on the metal surface. The phosphate coating does not provide much corrosion protection, but it also provides a good base for subsequent paint application. Phosphatizing also ensures protection against corrosion during the interval between fabrication of the part and application of paint. Automobile bodies are the best known examples of phosphatizing treatment. Oxide coatings are produced on steel by controlled high-temperature oxidation in air or by treatment in hot alkali solutions containing some oxidizing additives such as nitrates, chlorates, or persulfates. Black, brown, or blue coatings are developed, depending on film thickness. The oxide coatings are not protective but are made so by rubbing with inhibitor-containing oils or waxes. Gun barrels provide example for oxide-coated utility. Chromate coatings are produced on zinc, cadmium, and on coatings of these metals on steel. An immersion in sodium dichromate solution acidified with sulfu-

Aqueous Corrosion: Prevention

161

ric acid produces chromate on the surface. Zinc chromate imparts a light yellow color and increases the corrosion resistance in atmospheric exposures.

4.3.3

Organic Coatings

Paints, lacquers, and varnishes constitute organic coatings of which paints of varied formulations are in use. Coal tar or asphalt is another member of this group and probably the earliest one applied on steel to prevent rusting. It is still in use for protection against corrosion. The cathodically protected buried pipelines in soils provide an example. Paint is a mixture of particles of pigment suspended in a continuous organic or aqueous vehicle. The most common natural vehicle is a drying oil such as linseed or tung, each of which when exposed to air oxidizes and polymerizes to solids. Modern paint formulations use synthetic resins with a solvent—water or alcohol—as vehicles. Some catalysts are sometimes added to hasten the process of polymerization. Lacquers are resins dissolved in a volatile thinner and varnishes are a mixture of drying oil, dissolved resins, and a volatile thinner. Pigments are usually particles of inert inorganic compounds. Examples are oxides such as TiO 2, Pb2O3, Fe 2O3, ZnO 2, etc., and other compounds such as ZnCr4 , BaSO4 , PbCO3, clays, etc. Metallic zinc powder or flake is also used as a pigment. As the vehicles dry up and polymerize, they form a solid coating along with the bonded pigment particles on the metal substrate. Pigments confer the bulk, barrier, opacity, and color to the coating. Some pigments are inhibitors and provide additional protection to the metal surface. Metallic zinc provides cathodic protection to the steel substrate. Copper and arsenic, and some other toxic compounds as well, are used as pigment in marine paints and their release from the paint limits the growth of marine organisms. Distinction should be made between zinc-pigmented paints and zinc-rich paints. In the former, pigments constitute about 80% of the paint, of which 20% is ZnO2. Zinc-rich coatings have loadings of zinc dust, often over 90% by weight. Zinc-rich coatings provide excellent galvanic protection in aggressive environments. They are supplied as inorganic or organic zinc, the difference being the vehicle in which the zinc fillers are carried. Silicates are common vehicles for inorganics which, when reacted, produce an electrically conducting matrix in which the particles of zinc are embedded. Chlorinated rubber, catalyzed epoxy, polystyrene, and polyurethane are the recommended organic vehicles for zinc. Zinc-rich coatings have a long life and are more economical than some other high-quality three-coat systems. These are used to protect ship hull superstructures, marine structures, highway bridges, chemical process plant equipment, and other installations exposed to high humidity and salt. Paints are broadly classified as primers and top coats. Primers are applied

162

Chapter 4

directly on the metal surface. They contain inhibitive pigments or zinc, and perform the principal job of corrosion protection. Top coats are used on the primer coating, mainly for the sake of appearance. Nevertheless, they also provide a diffusion barrier and close the pores in the primary coat. Pores, which are called ‘‘holidays’’ in paint terminology, are the starting points of paint failures and they are minimized with the application of multiple top coats. In industrial and marine applications, three to five top coats is often recommended. Paints, whether primers or top coats, are classified according to the type of resin used as vehicle in the paint formulation. Among the most commonly used synthetics are alkyds, phenolics, chlorinated rubber, vinyls, and epoxies. Alkyd resins have found wide application in the protection of home appliances and machinery due to their fast drying property and durability in atmospheric exposures. Phenolic finishes have excellent resistance to acids, chemicals, moisture, and cold alkalis. They are used frequently as linings for drums used to ship corrosive chemicals. Vinyls and chlorinated rubber have the widest range of resistance to corrosives from strong acids to strong alkalis and have good resistance to penetration by water. Epoxies are also resistant to alkalis and many other chemical media. Table 4.6 shows the resistance of various coatings exposed to several environments. Surface preparation constitutes a very important factor in the application of all types of coatings, particularly paints. The surface should be free from mill scales, dust, rust, grease, welding flux, and impurities. A clean and rough surface is required for good adhesion of paint; otherwise the paint film fails prematurely. Different methods of surface preparation include chemical cleaning, scraping and wire brushing, abrasive cleaning, water blasting, flame cleaning, and cathodic cleaning. Abrasive cleaning by sand blasting is considered to be the best. Often phosphatizing is resorted to for prolonging the life of the paint coating. It is a false economy to save on surface preparation because the cost of labor for repainting far exceeds the cost involved in a good surface preparation carried out in the first place.

4.4 CATHODIC PROTECTION Cathodic protection implies that the component to be protected from corrosion is made cathode. This is accomplished by the flow of current from an external source to the component. The rate of corrosion is brought to zero so long as the sufficient amount of current is passed. It is a widely used corrosion prevention practice employed to protect steel structures, which include pipelines buried in soil or immersed in water, water tanks, ship hulls, chemical equipment, reinforcing rods in concrete, and others. Steel protected by galvanizing (zinc coating) is also an example of cathodic protection. Cathodic protection can also be applied for the protection of other metals from

Performance of organic coatings in different environmentsa

Oil base Alkyd Chlorinated rubber Coal-tar epoxy Catalyzed epoxy Silicone aluminum Vinyl Urethane Zinc (inorganic)

Acids

Alkalies

Salts

Solvents

Water

Weather

Oxidation

Abrasion

1 6 10 8 9 4 10 9 1

1 6 10 8 10 3 10 10 1

6 8 10 10 10 6 10 10 5

2 4 4 7 9 2 5 9 10

7 8 10 10 10 8 10 10 5

10 10 8 4 8 9 10 8 10

1 3 6 5 6 4 10 9 10

4 6 6 4 6 4 7 10 10

Aqueous Corrosion: Prevention

Table 4.6

a A value of 10 represents the best protection. Source: Adapted from M. G. Fontana, Corrosion Engineering, 3rd ed., McGraw-Hill, New York, 1983.

163

164

Chapter 4

general corrosion as well as from localized corrosion, such as pitting, dezincification, intergranular corrosion, and SCC. However, the necessary condition is that the surface of the protected structure be in intimate contact with the corrosive medium. That is why, it cannot be applied to the portions above the water line in an immersed structure. Neither can it protect structures that are electrically shielded, such as the inner members in a bundle of pipes or rods. Cathodic protection was introduced as early as in 1924 by Sir Humphrey Davy to protect the copper seathing of the ship hulls of that time, only to be discontinued as the stoppage of corrosion prevented the release of toxic copper ions and the growth of marine foulings adversely affected the speed of ships under sail. It took nearly 100 years to be reintroduced to the concept of the protection of buried oil pipelines in the United States. Application of cathodic protection to steel ship hulls started in 1950.

4.4.1

Principle

Cathodic protection is based on an understanding of the simple electrochemical principle that in the process of corrosion the cathode member in the cell is saved from corrosion. The corrosion reaction in an acid medium is represented by: M → Mn⫹ ⫹ ne (anodic)

(2.6)

2H ⫹ 2e → H2 (cathodic)

(2.3)

⫹

Supply of electron to the metal structure to be protected makes the first equation (2.6) proceed from left to right and the protection is achieved. Since the electrons flow through the metallic path, the corresponding positive current can be imagined to enter the structure from the electrolyte. Cathodic protection, therefore, is achieved though the flow of current from an external source through the electrolyte to the structure concerned. This will be made clearer from the consideration of corrosion in a pipeline buried in soil, as discussed below. Figure 4.8 represents corrosion of a buried pipeline in soil by the action of local cells, A being the anodic area and C the cathodic area. Electrons flow from A to B through the pipe. So, the positive current (or, simply, the current) flows from A to B through the electrolyte (soil). From the point of view of the direction of current flow, therefore, the area where the current is entering is protected. If now an external current of a bigger magnitude is allowed to enter the pipeline at all points countering the action of local cells, the entire pipeline will be protected, and this forms the basis of cathodic protection. With regard to the magnitude of the current required for protection, the effect of external cathodic current on the Evans diagram of the corroding system may be considered (Fig. 4.9). Impressing a cathodic current is essentially the polarization of the cathodic line (dc) further toward the active (anodic) direction, as represented by the dotted line. When the polarization line reaches the level of EA, the

Aqueous Corrosion: Prevention

165

Figure 4.8 Corrosion of a buried pipeline. Arrows show the direction of current flow.

open-circuit potential of anodic reaction, the corrosion stops completely as the driving force (Ec ⫺ EA) becomes zero. Iapplied is the current requirement for complete protection. However, a current of lower magnitude will provide some degree of protection. For example, and applied current corresponding to the point e will bring down the corrosion current from Icorros to that corresponding to the point b. If polarization is continued beyond the point f, i.e., if the current applied is

Figure 4.9 Polarization diagram illustrating principle of cathodic protection.

166

Chapter 4

more than required for complete protection, the situation is described as overprotection. An overprotected steel structure will still be protected, but this will mean an unnecessary wastage of current. The risk of stray current corrosion (Section 4.4.3) of the neighboring metallic structure also enhances with the excess current. For amphoteric metals like zinc and aluminum, overprotection increases corrosion because of the excess alkalis generated at the metal surface. Such a situation is sometimes referred to as cathodic corrosion.

4.4.2

Methods of Application

There are two methods for the application of cathodic protection: 1. 2.

Use of a sacrificial anode, and Use of an impressed current

Sacrificial Anode System The action of sacrificial anodes has been discussed in Section 3.3.3. A less noble metal in galvanic contact with another metal higher up in the emf series behaves as an anode member in the couple and sacrifices its life to protect the nobler metal, which acts as a cathode. All that is required is to provide galvanic coupling between the two metals. This may be achieved by direct contact, as in a coating, e.g., galvanizing, or by electrical contact through a conducting wire. Magnesium, zinc, and aluminum are used to protect steel. The schematic diagram for the protection of a buried pipeline by use of sacrificial anode is shown in Fig. 4.10. For protection of the inner wall of a pipe or tank the anode is to be suitably held inside and connected electrically with the wall. To act as a sacrificial anode a metal should fulfill the following requirements: 1. 2.

The potential difference between the anode and the corroding structure must be large enough to overcome the local cells on the corroding metal. The anode material must have sufficient electrical energy content to last for a reasonably long period before replacement. The electrical energy content is expressed by the term ‘‘ampere-hour per pound’’ (or ‘‘kg’’), meaning

Figure 4.10 Schematic representation of cathodic protection by sacrificial anode.

Aqueous Corrosion: Prevention

167

thereby that a pound of material will last for so many hours if 1 ampere of current is continuously discharged by it. 3. The self-corrosion of the anode material should be minial so that it is used up effectively for the current output only. Among the three common sacrificial anode materials, magnesium has the largest potential difference with steel, which is increased further by alloying with 1% Mn. Magnesium is often alloyed with aluminum and zinc. Although the efficiency of magnesium is low (about half of the theoretical efficiency of 1000 amp-h/1b), it is preferred as an anode material because of the advantage of higher potential. Zinc is usually used in its commercially pure form and has an efficiency of 95% of the theoretical. The potential difference between steel and pure aluminum is not large and under several conditions polarization of aluminum tends to render it cathodic with respect to steel. To avoid such a situation and also to increase efficiency, aluminum is alloyed with tin, zinc, mercury, and iron. Newer alloys of aluminum have also been developed for use as sacrificial anodes, which are often proprietary. In soils, the anodes are usually surrounded by a backfill comprising coke breeze, gypsum, and bentonite. The purpose of the backfill is to provide conducting surroundings that help in the uniform discharge of current. When aluminum alloy is used as anode, some amount of NaCl is added to the backfill, which counters passivation of the alloy. Cathodic protection should invariably be employed with an insulating coating on the surface to be protected. A bare surface needs a large amount of current for protection, which means a shorter life for the sacrificial anode. Coated surfaces need to be protected only at the pores or leaks (holidays) and the current requirement decreases drastically. A single magnesium anode can protect a length of 8 km of a coated pipeline, whereas only a few meters can be protected by it when the pipeline is bare. The coatings applied to the buried pipelines are usually tar- or asphalt-based backed up by layers of adhesive tapes. Thick and rugged coatings are necessary to resist damage during transport and installation. Impressed Current System In an impressed current cathodic protection system the protective current is supplied by an external dc supply, which is usually a rectifier but may be a generator or a battery as well. Its negative terminal is connected to the structure to be protected and the positive terminal to an auxiliary anode that discharges the current. In soils, the anode or, more often, a ‘‘ground bed’’ comprising a number of anodes is surrounded by backfill for better conductivity. Figure 4.11 shows the arrangement of an impressed current system protecting an underground tank. The common materials used for the auxiliary anode are steel scrap, graphite, aluminum, lead, and high-silicon iron (Duriron). Theoretically, the anodes should

168

Chapter 4

Figure 4.11 Schematic representation of an impressed current cathodic protection system.

not be consumed, but in practice they are consumed slowly and need replacement. Platinized titanium has indefinite life as an auxiliary anode and is being increasingly used in the protection of marine structures. It is important to remember that all buried parts of the ground bed assembly connected to the positive terminal of the power source may discharge current if they come in contact with the soil and corrode. The cable from the rectifier to the anodes or the cables interconnecting the anodes in the ground bed may eventually snap in that process, disrupting the electrical contact of the installation. Proper insulation of these parts is, therefore, of utmost importance. By contrast, all buried wires connecting to the sacrificial anodes collect current from the environment and are protected. Protecting currents are usually determined empirically. The current requirement is low in static environments but increases under flowing conditions of the environment. It becomes enormously high in aggressive media such as for steel in hot sulfuric acid, and cathodic protection is not economically viable in such cases. For buried steel structures in soils, a potential difference of ⫺0.85 V with respect to a copper–copper sulfate reference electrode is held as the criterion of protection. Current is adjusted to maintain this potential difference. For the protection of lead-seathed cables, this value is ⫺0.70 V, and for aluminum between the limits of ⫺1.00 and ⫺1.20 Volt. The upper limit is specified for aluminum to avoid alkalinity buildup at the cathode (‘‘cathodic corrosion’’). As in the case of a sacrificial anode system, insulating coatings must be used on the structures to be protected in order to reduce the total current requirement. The effectiveness of protection is monitored through potential measurement. Test coupons made of the same metal as that in the protected structures are often exposed in electrical contact with the protected structure. Any loss in weight will indicate the inadequacy of protection. Cotton soaked in potassium ferricyanide is sometimes placed in contact with the protected structure. It turns blue with

Aqueous Corrosion: Prevention

169

the formation of ferrous ferricyanide by reacting with ferrous ions available from corrosion of steel, if the protection is inadequate.

4.4.3

Stray Current Corrosion

Stray currents are those that travel away from the intended circuit. A part of the current discharged by the ground bed in an impressed-current cathodic protection system may enter a metallic structure in the vicinity of the protected structure. To complete the circuit it leaves the structure, enters the electrolyte (soil), and then goes to the cathode (i.e., the protected structure). Figure 4.12 illustrates such a situation. The steel pipe is the neighboring structure to the cathodically protected steel tank. It is clear from Section 4.4.1 that the portion of the steel pipe where the current is entering from the soil gets the benefit of cathodic protection, whereas the portion where the current leaves the pipeline to enter the soil acts as anode and undergoes corrosion. This type of corrosion is called stray current corrosion. Eventually, a leak may appear in the pipeline. Application of a paint coating to the pipeline simply aggravates the situation as the current now gets discharged from a few defect sites in the paint coat where anodic current density increases enormously. The structure can be saved from stray current corrosion by simply short-circuiting it to the protected tank. In that case the current will follow the metallic path to complete the circuit and the pipeline will also be a part of the protected structure. Such a situation will not be uncommon in congested industrial areas. A common cathodic protection system should be organized by the owners of the installations for their mutual benefit. Although stray current corrosion has been discussed in the context of cathodic protection, the stray current may emerge from other sources as well. In a metropolis stray currents from tram car lines often bring hazards to the metallic structures

Figure 4.12 Strong current corrosion resulting from an impressed current cathodic protection system.

170

Chapter 4

Figure 4.13 Stray currents produced by an on-shore welding generator damaging a ship hull.

underneath; particularly the municipal water supply lines and sewage lines are affected. The hull of a ship becomes prone to stray current corrosion if a welding operation is carried out on it with the help of a welding motor-generator located on shore. This is because current leaking from the hull enters water that returns to the source. The situation is shown in Fig. 4.13. The incidence of corrosion can be avoided by shifting the dc generator to the ship.

4.5 ANODIC PROTECTION Anodic protection is based on the principle of passivation of metals. It has been discussed in Section 2.4 that active-passive metals such as iron, chromium, nickel, titanium, and their alloys attain passivity on anodic polarization and in this state their rate of corrosion decreases many fold, sometimes by the order of 105 –106. In anodic protection, the potential of the system is maintained in the passive region with the help of a potentiostat. The structure to be protected is connected to one terminal of the potentiostat, the second terminal is connected to an auxiliary cathode, and a constant potential is maintained with respect to a reference electrode convected to the third terminal. The anodic protection arrangement for a steel tank containing sulfuric acid is shown in Fig. 4.14. The principle of anodic protection is further illustrated in Fig. 4.15. An activepassive metal corroding in an acid medium has been depicted. For the attainment of passivity the anodic current must be increased to icritical , which corresponds to Epp . The current automatically adjusts itself as the potentiostat sets the potential to this value. The magnitude of applied current has been shown as iapp(1). However, the potential has to be raised further to EProt to maintain the metal in the passive region and the corresponding applied current is iapp(2). The large difference in magnitude of these two applied current values may be noted. The current required

Aqueous Corrosion: Prevention

171

Figure 4.14 Anodic protection of a steel storage tank containing sulfuric acid.

to maintain passivity corresponds nearly to ipassivity, which again indicates the rate of corrosion under protection. Anodic protection has been very successfully employed for the protection of carbon and stainless steel storage and transportation vessels for sulfuric acid as well as for the stainless steel heat exchangers in sulfuric acid plants. The corrosion rate of the unprotected structures in these media is very high, and it would be highly uneconomical to provide cathodic protection because the current requirement will also be enormously high, as represented by iapp(c) in Fig. 4.15. On the other hand, the current requirement for maintenance of anodic protection is usu-

Figure 4.15 Polarization diagram illustrating current requirement in anodic protection.

172

Chapter 4

ally very low. For example, type 316 stainless steel requires 0.6 mA/cm2 for passivation in 67% H2SO4 at 24°C and 0.1 µA/cm2 for the maintenance of passivity. The requirement of current in cathodic protection of the same system will be several thousand times higher. The decrease in the rate of corrosion is significant. For example, the corrosion rate of type 304 stainless steel in 10NH2SO4 drops from 1930 mpy to 0.016 mpy when protected anodically. Corrosion rates of type 316 stainless steel heat exchangers used for cooling of sulfuric acid have been reported to reduce from 200 mpy to 1 mpy. Anodic protection has also found application in the protection of stainless steels in phosphoric acid and caustic soda solutions, in carbon steels in oxalic acid and in NH4NO3 fertilizer mixtures. However, anodic protection is not possible in hydrochloric acid or in solutions of large chloride content as the passivity is destroyed by chloride ions and the current requirement becomes high (high icritical and ipassivity). When compared with cathodic protection, it can be concluded that both systems are often complementary to each other. Cathodic protection is effective in low to moderately high corrosive environments, whereas anodic protection is effective in low to highly aggressive media. The cost of installation for anodic protection is higher than that of cathodic protection, but this is compensated by low maintenance cost.

4.6 DESIGN IMPROVEMENT Many corrosion problems can be eliminated or reduced if proper considerations are made at the design stage. It is obviously wiser to reduce the possibilities of corrosion in the beginning than to take elaborate arrangements for the prevention of corrosion that might be avoidable. Broadly, the design considerations should include appropriate site selection for the plant, plant layout, selection of material for various components, design of components, and selection of corrosion prevention measure. A design engineer is not basically a corrosion engineer, but it is necessary to convey the basic knowledge of corrosion to the designer so that the obvious factors aggravating corrosion are not overlooked. For the location of the plant and its layout due considerations are to be given to the direction of winds. Airborne particulates and saline droplets from the nearby sea can lead to adverse deterioration of structures. The stacks should be suitably located so that the effluent emitted by one does not directly hit the other structures in the plant or another stack located nearby. The basic requirements of a component from a corrosion point of view need to be established before the selection of material for the component and its design proper. The considerations include the environment to which the component is likely to be exposed, the expected life of the component, accessibility for mainte-

Aqueous Corrosion: Prevention

173

nance, the expected type of corrosion, and the risk of product contamination by corrosion products. When the design involves contact of two dissimilar metals, their compatibility must be checked to avoid the incidence of galvanic corrosion. The geometrical shape of the components is of the utmost importance. A faulty geometrical design may be the cause of failure of a component by corrosion even if the design satisfies all of the mechanical requirements. Some important rules relating to geometrical design to minimize corrosion are discussed below.

4.6.1

Some Design Rules

1. Provide corrosion allowance. The simplest way to ensure the expected full life of a component like pipe, tank, or a civil engineering structure is to provide corrosion allowance to the wall thickness. This is simply an added thickness over the mechanical design that will be corroded out over the expected lifetime and is calculated from the rate of corrosion multiplied by the number of years of life. However, with reliable corrosion protection and monitoring, this additional material cost may be avoided. 2. Avoid galvanic contact. Whenever possible, the same or similar material should be used throughout the entire structure. Otherwise, compatibility must be checked when two dissimilar metals in contact are used. Providing an insulation between two dissimilar metals is advisable, provided the insulation does not produce crevice conditions. Lastly, if galvanic corrosion is

Figure 4.16 Crevices arising from design and assembly.

174

Chapter 4

inevitable, the corroding part should be designed in such a way that it can be easily replaced. The relative surface areas of anodic and cathodic surfaces should not be underestimated. Fastener material should be more noble. When paint is applied, it is important to paint either the whole assembly or the noble member of the couple. Application of paint only on the active metal will aggravate corrosion at the pores leading to rapid pitting. In a flowing system, nobler metal should be avoided as upstream component, since ions carried downstream deposit on the less noble metal and induces galvanic corrosion. 3. Avoid crevices. Crevice sites may arise due to the contact of two metallic components (Fig. 4.16a), incorrect trimming of gaskets (Fig. 4.16b), improper provision of base for a vessel (Fig. 4.16c), riveting (Fig. 4.16d), and many other situations that should be avoided. Welding rather than riveting is prescribed to do away with crevices. Use of proper filler material or sealing is recommended at the crevice sites.

Figure 4.17 Improper and proper drainage in tanks and pipes.

Aqueous Corrosion: Prevention

175

Figure 4.18 Design and arrangement of channels and angles to avoid water accumulation.

4. Provide for easy drainage. This applies particularly to tanks. Any accumulation of corrosive liquid due to improper drainage is harmful. Steel tank storing concentrated sulfuric acid is an example. The residual acid absorbs moisture, gets diluted, and aggravates corrosion of steel. Even the residual moisture will create differential aeration conditions for corrosion to take place. Improper and proper drainage provisions are exemplified in Fig. 4.17. Channels should be provided with drain holes (Fig. 4.18a) and V angles should be used inverted (Fig. 4.18b) to avoid accumulation of moisture. Provision of ventilation holes in closed structures is a must to avoid accumulation of residual moisture. 5. Avoid sharp corners. Sharp corners should be replaced by rounded corners as the applied paint coats tend to be thinner at the sharp corners. Paint failures often start from such points. 6. Avoid sharp bends and protrutions in piping systems. Sharp bends cause impingement attack in high-velocity fluids (Fig. 4.19a), which can be avoided by smooth rounding off (Fig. 4.19b). Protrutions and troughs cause turbulence and result in erosion corrosion (Fig. 4.20). It is also proper to avoid T joints where the flowing fluid directly impinges on the facing wall (Fig. 4.21a). The design can be improved and baffle plates provided in vulnerable places (Fig. 4.21b). 7. Avoid mechanical stresses. Excess mechanical stresses and stress concentration should be avoided in the design of structures to minimize the occurrence of SCC. 8. Avoid vapor pockets. Frequently, interfaces between air and liquids are the

176

Chapter 4

Figure 4.19 (a) Undesirable and (b) desirable blends in pipes carrying fluids.

Figure 4.20 Protrusion in pipe causing turbulance and erosion corrosion.

Figure 4.21 (a) Direct impingement of fluid in a T-jointed pipe system. (b) Avoided by better design and provision of a baffle plate.

Aqueous Corrosion: Prevention

177

Figure 4.22 Avoiding vapor pockets in piping system by use of a valve.

sites of corrosion attack. The vapor pockets may be avoided by suitable use of vent lines. Figure 4.22 illustrates this. 9. Avoid hot spots. In heat exchangers and heat transfer devices, design should ensure uniform temperature gradients. Nonuniform temperature distribution produces local heating and such hot spots may lead to rupture. SCC may also be encountered due to the development of local thermal stresses along with local increase in the concentration of damaging species. 10. Exclude air. Since air provides oxygen, the effective cathodic depolarizer, air is to be excluded whenever possible. This is particularly recommended for chemical process equipment.

REFERENCES 1. Mars G. Fontana, Corrosion Engineering, 3rd ed., McGraw-Hill, New York, 1987. 2. Herbert H. Uhlig, Corrosion and Corrosion Control, 2nd ed., John Wiley and Sons, New York, 1971. 3. Denny A. Jones, Principles and Prevention of Corrosion, Macmillan, New York, 1992. 4. John O’M Bockris and Amulya K. N. Reddy, Modern Electrochemistry, Vols. 1 and 2, Plenum Press, New York, 1973. 5. L. L. Shrier (ed.), Corrosion, Vols. 1 and 2, Newness-Butterworths, Sevenoaks, Kent, England, 1976. 6. B. E. Wilde, Passivity and Its Breakdown in Iron Base Alloys, NACE, Houston, 1976. 7. NACE Basic Corrosion Course, NACE, Houston, 1971. 8. N. D. Tomashov, Theory of Corrosion and Protection of Metals, Macmillan, New York, 1966. 9. J. H. Morgan, Cathodic Protection, Leonard Hills [Books] Ltd., London, 1959. 10. P. Markus and J. Oudon, Corrosion Mechanics in Theory and Practice, Marcel Dekker, New York, 1995.

This Page Intentionally Left Blank

5 Tarnishing and Scaling Processes

5.1 INTRODUCTION A metal in a gaseous environment constitutes a very complex chemical system. In most situations an interpretation of the reaction behavior can be achieved only with great difficulty. The terms metal oxidation, tarnishing, and scaling are used in literature whenever oxidizing gases such as oxygen, sulfur vapor, the halogens, water vapor, CO 2 , etc., attack a metal or an alloy at low, intermediate, or high temperatures. In general, the oxidation processes at high temperatures leading to the formation of thick oxide layers are referred to as the scaling process, whereas at lower temperatures the phenomena of thin oxide film formation are termed the tarnishing process. Oxidative attack on metals can take place under widely varied conditions from the ‘‘mild’’ oxidizing one that prevails in air at room temperature to the extremely ‘‘severe’’ conditions of mixed environments at elevated temperatures encountered by high-temperature components such as superheater and reheater tubes, gas turbine blades and vanes, alloy components in fuel conversion and power-generating units, reformer tubes, heat exchanger tubes, and the like, which are often subjected to thermal fluctuations and thermal cyclings under actual service conditions. The severity of the operating condition can be inflated due to thermal fluctuations and sudden high-temperature excursion. It is a matter of common experience that most of the metals with the exception of gold and platinum, when exposed to different atmospheres like oxygen, halogen, sulfur vapor, etc., take up the electronegative constituents and form the com179

180

Chapter 5

pounds even at room temperature due to fulfillment of favorable thermodynamic conditions. Such reactions can be represented as follows: aM(s) ⫹

b X 2 (g) ⫽ M a X b (s) 2

(5.1)

For example: 1 O 2 (g) ⫽ NiO(s) 2 1 2Cu(s) ⫹ S 2 (g) ⫽ Cu 2 S(s) 2 1 Ag(s) ⫹ Cl 2 (g) ⫽ AgCl(s) 2

Ni(s) ⫹

Moreover, this type of reaction can be either reversible or nonreversible depending on the temperature under consideration and the partial pressure of the oxidant. Normally, the reversible type of uptake of the electronegative component by the metal occurs at sufficiently low temperatures and low pressures of oxidant when the film thickness is restricted to a few monolayers only. But with gradual increase in temperature, this adsorption process leads to the formation of distinct metallic compounds. From a thermodynamic point of view, the formation of metal oxide, metal sulfide, metal halide on a metal surface is very feasible at elevated temperatures and appears to be one of the simplest reactions, but in reality this is not the case. The formations of such thermodynamically stable compounds frequently appear as compact and adherent layers on the metal substrate. As a consequence, the two reacting species, metal and oxidant, are spatially separated from each other and further reaction is only possible if at least one of the reactants can diffuse through the scaling layer to the reaction partner. In such cases the course of reaction or the thickening of the product layer is no longer determined by the equations of the type 5.1 but by diffusional transport of reacting species and phases boundary reactions, for which the mechanisms can be quite complex. It is a well-established fact that the rates of thickening of various oxide scales or oxidation kinetics are governed by the reaction mechanisms, which in turn are controlled by the relevant structural, morphological, thermodynamic, and transport properties of the solid reaction product. Day by day, new experimental data and novel ideas are pouring into this field of research. Some of these are in support to the already proposed theories, whereas others do not fit to the existing multiplicity of kinetic laws and oxidation mechanisms. Therefore, the confrontation between the theory and experimental results for both low- and high-temperature regimes is not uncommon in recent literature.

Tarnishing and Scaling Processes

181

5.2 THERMODYNAMIC ASPECTS OF METAL–SINGLE OXIDANT SYSTEMS The primary approach to tackle the degradation of a metallic component exposed to a certain environment in an industrial installation should be to examine the extent of affinity of the metal to oxidizing gases, which leads to its degradation manifested by the formation of some compounds with the oxidant. For judging the relative stability of different metallic compounds, one has to derive information from equilibrium thermodynamics. Therefore, for any reaction of the following type involving a metal in an environment of single oxidant: M(solid metal) ⫹ O 2 (g) ⫽ MO 2 (solid compound)

(5.2)

one has to know the value of Gibbs’ free energy change (∆G) for the total reaction. If ∆G is negative, the reaction will be spontaneous in the forward direction, leading to the formation of metal oxide (MO 2 ), whereas if ∆G is positive, MO 2 will not be stable at the temperature and pressure of the oxidant, thereby leading to spontaneous dissociation of the oxide. But thermodynamics deals with the equilibrium attainment for any metal–compound–oxidant system that needs a scrutiny of the standard free energy change (∆G 0 ) of the above reaction. For ready reference, to judge the relative stability of various metal–oxidant systems, one has to have a look at the graphical representation of ∆G 0 vs. T plots [1] for various systems (Ellingham-Richardson diagrams). Such types of graphical representation are shown in Fig. 5.1(a)–(d) for metal–oxide–oxygen, metal–sulfide–sulfur vapor, metal–halide–halogen, and metal–carbide–carbon systems, respectively. The preferential choice of alloying elements like Cr, Al, Si, etc., in the development of high-temperature alloys is best judged from such diagrams. The compounds of these elements are comparatively more stable than the base metal compounds, as evidenced by more negative ∆G 0 values in the above-mentioned figure parts. It would be appropriate to mention that when a metallic component is exposed to an environment containing more than one oxidant, say SO 2 and O 2 , and the ∆G 0 values are found to be negative for the formation of both oxide and sulfide; however, preferential growth of either of the compounds will no longer be guided by thermodynamic considerations alone. The equilibrium oxidant pressure for the formation of a metal oxide could be evaluated from the standard free energy change of the reaction by using ∆G 0 ⫽ ⫺RT ln K

(5.3)

where K is the equilibrium constant that is related to the activities of species involved in the reaction. For example, K for the reaction (5.2) is given by: K(T) ⫽

a MO2 a M ⋅ a O2

(5.4)

182

Chapter 5

Figure 5.1(a) Free energies of formation of some oxides. (Courtesy of Mr. Olette and Mrs. Ancey-Moret.) [Ref. 1].

Tarnishing and Scaling Processes

183

Figure 5.1(b) Free energy of formation of sulfides. (Courtesy of Mr. Olette and Mrs. Ancey-Moret.) [Ref. 1].

184

Figure 5.1(c) Free energy of formation of some chlorides [Ref. 1].

Chapter 5

Tarnishing and Scaling Processes

185

Figure 5.1(d) Free energy of formation of carbides. (Courtesy of Mr. Olette and Mrs. Ancey-Moret.) [Ref. 1].

186

Chapter 5

In pure state, a MO2 ⫽ 1, a M ⫽ 1, and a O2 ⫽ (PO2 /P 0O2 ) ⫽ PO2 (since the standard state of O 2 is unity, i.e., P 0O2 ⫽ 1 atm). Therefore ∆G 0T ⫽ RT ln PO2 ⫽ ∆H 0 ⫺ T ∆S 0

(5.5)

From the thermodynamic data available in literature, at any temperature T, equilibrium PO2 values at M/MO 2 and MO 2 /O 2 can be estimated. If a metal is exposed at some elevated temperature in dry air, there may be formation of single oxide or multilayered oxides. For some common metal–air (dry) systems, assuming compact oxide scale formation, the equilibrium PO2 values developed at various interfaces at 1173 K are presented in Fig. 5.2a–i. Another relevant graphical representation of metal–oxidant equilibria, which

Figure 5.2 Schematic presentation of single-layered and multilayered oxide scale formation in dry air at 1173 K for different metal–oxygen systems showing equilibrium values of PO 2 at different reaction interfaces (assuming compactness of the scales).

Tarnishing and Scaling Processes

187

are referred to as phase diagrams, provide valuable information in understanding the mechanism of the sequential oxide layer formation. If the metal can possibly form two or more oxides, the most oxygen-deficient one will be formed adjacent to the metal substrate and the most oxygen-rich oxide will be in contact with the gas phase. As illustrations, the Fe-O and the Ti-O phase diagrams are presented in Fig. 5.3a and b, respectively. The Fe–O 2 system depicts formation of three oxide phases, namely (1) wu¨stite (Fe 1⫺δ O), (2) magnetite (Fe 3 O 4 ), and (3) Haematite, (Fe 2 O 3 ). Wu¨stite phase is known to be highly metal-deficient but unstable below 843 K where it disproportionates to Fe and Fe 3 O 4 . Therefore, during oxidation of Fe above 843 K in air, the total scale will consist of three layers having the most Fe-deficient oxide (wu¨stite) adjacent to the metal and the most oxygenrich oxide (haematite) in contact with the gas phase with Fe 3 O 4 sandwiched between them [2]. Experience shows that the wu¨stite layer will be the thickest and the haematite will be the thinnest as shown in Fig. 5.4. Reasons for such observation will be discussed subsequently. If one has to deal with binary or ternary alloys in a single-oxidant system, the thermodynamic description of the condensed phase equilibria will be more complicated due to the presence of a second or third element as a variable. In such a situation, it will be more convenient to consider a stability diagram of the alloy–oxidant system at a constant temperature [3]. Such an isothermal phase diagram for the Fe-Cr-O system at 1573 K is presented in Fig. 5.5. Here, mole fraction of Cr in binary Fe-Cr alloy has been plotted against logarithm of oxygen partial pressure. This figure depicts the stability regions of different solid solutions and mixed oxides with variation in Cr content and oxygen potential.

5.3 KINETIC ASPECTS AND RATE EQUATIONS In general, whenever one examines the reaction behavior of any metal with a single oxidant, one must consider a large number of phenomena and partial processes of which one would be the slowest and hence the rate-determining step. The various partial or sequential processes could be the following: 1. Phase boundary reactions (chemisorption of the nonmetal molecule and simultaneous electron exchange with the substrate, thereafter splitting of the molecules at the oxide–gas interface and transfer of the metal from the metallic phase in the form of ions and electrons to the oxide film at the metal– oxide interface with further reaction of the individual reactants and formation of a distinct reaction product), nucleation, and growth of oxide crystals. 2. Diffusional transport of cations, anions, electrons, or positive holes through the scale, which gets complicated by a special migration mechanism because of the development of electrochemical potential gradients set up across the growing layer.

188 Chapter 5

Figure 5.3 (a) The iron–oxygen phase diagram. (b) The titanium–oxygen phase diagram [Ref. 4].

Tarnishing and Scaling Processes

189

Figure 5.4 Microsection of scale formed on iron in air at 625°C after 24 h [Ref. 2].

3. Predominant transport processes in space charge boundary layers in the case of thin tarnishing films, especially at low temperatures. Besides these, there are two more significant aspects that need due consideration: 1. Thermodynamic stability of reaction product, and 2. Crystal structure and morphology of the oxide scale as well as that of the underlying metal or alloy. Some of the main aspects of partial processes during metal oxidation are presented schematically in Fig. 5.6 [4]. If one starts with a theoretically clean metal surface exposed to an oxidant, the initial step in the interaction is adsorption of the oxidant on the metal surface. On continued exposure to the reacting environment, the initial nuclei of the reaction product are formed at some preferential sites. These often grow laterally to produce a continuous film that ultimately covers the entire surface of the metallic substrate. In certain situations the oxidant may get dissolved in the metal substrate to an extent determined by the solubility and diffusivity of the oxidant in the metal at the temperature under consideration. Once the continuous adherent film is formed, it separates the reactants by creating a physical barrier for which further

190

Figure 5.5 Stability diagram for the Fe-Cr-O system at 1300°C [Ref. 3].

Chapter 5

Tarnishing and Scaling Processes

191

Figure 5.6 Schematic illustrations of some of the main aspects of metal–oxygen reactions [Ref. 4].

reaction will be governed by the rate of migration of either of the reactants or both in a countercurrent direction through the film. The solid state diffusional transport takes place by various processes such as (1) diffusion through oxide lattice and (2) diffusion along oxide grain boundaries and other easy diffusion paths, e.g., dislocation networks, microchannels, etc. Furthermore, during scale growth, many additional phenomena and secondary processes may occur. Depending on the growth mechanism, cavities and closed pores may develop in the scale as well as in the metal. These features will, in turn, affect transport processes through the scale. It is also a common experience that large stresses may build up in the scale and in the underlying metal. Such stresses may cause plastic deformation or cracking of the scale. If cracking takes place in segments or in repeated manner,

192

Chapter 5

the scale loses its protective ability. In such cases, reaction may be governed by diffusion through (1) a thin reaction product layer of approximately constant thickness adjacent to the metal or (2) by phase boundary reactions. The linear oxidation rate commonly observed in alkali metals is due to this type of reaction behavior. Depending on the system under investigation and the reaction conditions, other processes may also take place. This may involve the formation of reaction products that are liquid or that may continuously evaporate (MoO 3 , CrO 3 , etc.). In general, the oxide layer that forms on the metal during oxidation are found to be compact, adherent, and protective if the criterion formulated by Pilling and Bedworth (PB) [5] in 1923 is satisfied. The PB ratio is expressed as: φ⫽

Voxide (molar volume of oxide) (5.6) Vmetal (volume of metal consumed for formation of 1 mol of oxide)

In general, when this ratio is greater than unity, protective film or scale growth is expected to occur on the corresponding substrate. However, if it is too much greater or too much lower than unity, then the compactness of the scale is lost and there will always be free access of gaseous reactant to the fresh metal surface. The PB ratios of some common metals are listed in Table 5.1. Although the rule has many exceptions, it is historically important and is still used even today as a rough guideline in predicting the protective nature of the compound layer formed. Schottky first pointed out the extent to which this volume quotient idea will play its role in maintaining adherence and compactness of the scale, so that solid state diffusional transport remains the rate-limiting step. Furthermore, it would be decided by the level of stresses developed in the oxide and the underlying metal as well as by the extent of simultaneous stress relief through plastic flow of the oxide and the substrate. Although attempts have been made to rationalize the situation by considering all of these mechanical aspects, it is really difficult to develop a unified theoretically sound background that will be fully reliable to explain the behavior and abnormalities of various systems. In this respect, knowledge of ratio of linear thermal expansion coefficients for

Table 5.1

Pilling–Bedworth (PB) ratios of some typical oxide–metal systems

K2 O

Li 2 O

MgO

CdO

Al 2 O 3

ZnO

Cu 2 O

NiO

FeO

0.45

0.58

0.81

1.21

1.28

1.55

1.64

1.65

1.68

TiO 2

CoO

Cr 2 O 3

SiO 2

Ta 2 O 5

Nb 2 O 5

U3 O8

MoO 3

WO 3

1.76

1.86

2.07

2.15

2.50

2.68

2.77

3.30

3.35

Tarnishing and Scaling Processes

193

Table 5.2

Ratios of linear coefficient of expansion of some metal–oxide systems (in general valid up to ⬃1173 K)

Fe/FeO 1.25 Ti/TiO 2 1.30

Ni/NiO

Co/CoO

Cr/Cr 2 O 3

Cu/Cu 2 O

1.03

0.93

1.30

4.32

Al/Al 2 O 3

Si/SiO 2

Zr/ZrO 2

Hf/HfO 2

3.22

0.475

0.91

1.03

various metal–metal oxide systems, which are presented in Table 5.2, will be very useful. It is important to point out that during the 1920s Tammann [6] for the first time provided experimental evidences for obeyance of a definite mathematical relation between the extent of reaction and exposure time. Such a relation was found to be parabolic in nature. His experimental results inspired a large number of investigators to derive a variety of mathematical relations to be obeyed during the progress of oxidation, depending on the individuality of the systems and prevailing experimental conditions. Broadly speaking, two types of oxide growth rate have been observed, i.e., thick-layer growth and thin-film growth where the demarcation between the two ˚ (1000 nm), as suggested by Cabrera and Mott [7]. is a thickness of ⬃10,000 A In both cases, the oxide film formed increases in its thickness with time but the rate laws obeyed in the two cases are different. For the thin-film growth, four kinetic laws have been proposed and experimentally verified for a number of metal–oxidant systems. These are presented below in their integral forms: 1. 2. 3. 4.

Inverse logarithmic law: 1/ζ ⫽ k 1 ⫺ k 2 ln t Normal logarithmic law: ζ ⫽ k′1 ⫺ k′2 ln t Cubic law: ζ 3 ⫽ k c t, and Parabolic law: ζ 2 ⫽ k′p t

where ζ ⫽ instantaneous thickness of the oxide film at any time (t), and k 1 , k 2 , k′1 , k′2 , k c , and k′p are constants in the corresponding kinetic equations. For thick layer growth or scaling processes, two possible rate laws are observed: (1) parabolic law (ζ 2 ⫽ k p t) and linear law (ζ ⫽ k 1 t), where k p and k 1 are parabolic and linear rate constants, respectively. The above rate expressions could also be presented in their differential forms. Instead of thickness, the extent of reaction can be expressed in terms of mass gain per unit area as a function of time [8]. The various oxidation–time relationships are presented in Fig. 5.7. One must realize that in the study of ambient or near-ambient temperature

194

Chapter 5

Figure 5.7 Oxidation–time relationships [Ref. 8].

aqueous corrosion, which is frequently referred to as ‘‘wet corrosion,’’ emphasis is mainly placed on the electrochemical aspects, i.e., electrode kinetic studies are more important. In contrast, in the study of elevated temperature gaseous reaction of metals or alloys, referred to as ‘‘dry corrosion,’’ emphasis is placed on thermodynamics or phase stabilities of the various reaction products, the defect structures and transport properties of the scale, and the morphology of the product scale, besides the usual electrochemical aspects.

5.4 DEFECT CHEMISTRY OF OXIDES AND OTHER INORGANIC COMPOUNDS Before one can understand the mechanisms of oxide film or scale growth processes in the thin- and thick-film ranges for protective oxidation of metals, one is required to have knowledge of the mechanisms of solid state diffusion. It is well established that diffusion in solids takes place because of the occurrence of imperfections or point defects within them. The presence of such point defects in the compounds, their concentrations, and the nature of the defect structure would play the most important role in controlling the progress of oxidation reactions. Continued degradation of the metal becomes possible only when the reacting species, i.e., either cations or anions, migrate through the oxide layer so

Tarnishing and Scaling Processes

195

that they can meet at one of the interfaces, i.e., metal/oxide or oxide/oxygen, and react with one another. The migration of reacting species through this barrier layer is only possible if the oxide lattice is defective in nature, i.e., defective in atomistic scale. Imperfections or defects in solids may be divided into two groups: (1) point or lattice defects and (2) line and surface defects. The point defects include vacancies either on anion or cation sublattices or both, interstitials of either kind of ions, misplaced atoms, as well as foreign cations or anions. The line and surface defects include dislocations, grain boundaries, the inner and outer surfaces of the product layer, etc. However, discussion here will be mainly focused on the types of point imperfections and their concentrations, and the corresponding defect equilibria, because they can help us to understand the oxidation behavior of the various metal–oxidant systems. In order to describe the point defects and to express their formation in terms of reaction equations, it is absolutely necessary to follow a system of notation. Several systems of notation exist in the literature of defect chemistry; however, Kro¨ger and Vink’s method has found more or less universal acceptance. Accordingly, this type of notation has been used in the present discussion. In a compound MO, the point defects are represented by: V O —anion vacancies V M —cation vacancies O i —anion interstitials M i —cation interstitials Atoms on normal lattice positions are written as O O , M M , i.e., oxygen atoms and metal atoms are at their normal lattice sites. Correspondingly, an unoccupied or vacant interstitial site is written as V i . Real crystals always contain some imperfections. When a foreign (F) cation occupies a regular M site, the foreign atom is represented by FM . If it occupies an interstitial site, it is written as Fi , and when it occupies an oxygen site, it is written as FO . The creation of a point defect in a perfect crystal increases both the internal energy (hence enthalpy) and the entropy of the system. The equilibrium concentration of the defects will be reached only when the free energy of the system is at a minimum. Thermodynamically, point defects will always be present in a crystal above zero Kelvin. Furthermore, all types of defects will, in principle, be formed. However, the free energies of formation of the different types of defects will usually have widely different values; correspondingly, it is found that one type of defect structure commonly predominates in a particular solid. The relative concentrations of the different types of defect will be a function of temperature and other variables such as state and composition of the compound. Thus, defect equilibria with a large positive enthalpy of formation, for instance, which are not

196

Chapter 5

thermodynamically favored at low temperature, may become important at high temperatures.

5.4.1

Neutral and Charged Defects and Electroneutrality Conditions

In a crystal of ionic compounds (oxides, halides, sulfides, etc.) the atoms are charged, i.e., the compound crystal is made up of cations and anions. Point defects may be neutral or charged. Neutral defects are designated by a cross-sign as superscript, e.g., V xO , V xM , M xi , O xi , etc., which signifies zero charge. It is generally more convenient to write defect formation reaction in a crystal by considering the charge on the defect relative to the perfect crystal. The relative charge is termed as the ‘‘effective charge’’ on the defect. As examples, singly and doubly ionized oxygen vacancies are represented by V O• and V ••O , respectively. A doubly charged interstitial cation is written as M ••i . Similarly, singly and triply ionized metal vacancies are represented by V′M , and V⵮M , respectively. Correspondingly, an interstitial oxygen ion will have effective negative charge, e.g., O″i . Also, substitutionally dissolved foreign ions, which have valence different from that of the host lattice ions (i.e., aliovalent foreign ion), will also show an effective charge represented by F M• or F′M or F O• or F′O . In the study of band structure for ionic solids or semiconductor compounds, the energy levels of V O , M i , and FM (F having higher valence than M) are nearer to the conduction band of the host lattice and hence these centers act as donors, i.e., they donate electrons to the conduction band and themselves become positively charged. Similarly, the energy levels of V M , O i , and FM (F having lower valence than M) will be nearer to the valence band of the host crystal. These centers act as acceptors taking up electrons from the valence band, thereby becoming negatively charged and injecting positive holes in the valence band. Quasi-free electrons (excess electrons) in the conduction band and positive holes in the valence band will exhibit effective negative and positive charges, respectively. They are represented in defect reactions as e′ and h•, respectively. It should also be kept in mind that a crystal as a whole should be electrically neutral, i.e., the sum of all effective positive charges must equal the sum of all effective negative charges. This principle of electroneutrality condition serves as the basic concept in formulating and treating defect equilibria and for evaluating defect concentration in crystals.

5.4.2

Stoichiometry and Nonstoichiometry

Stoichiometric compounds also referred to as daltonides, follow the law of constant proportion, i.e., in the case of oxides, a fixed proportion between metal and oxygen is maintained. Schottky and Wagner [9] for the first time established

Tarnishing and Scaling Processes

197

through detailed thermodynamic treatment of defects that the law of constant proportion is a special case. In principle, it is an exception rather than the rule. Thus, it is more correct to speak of ‘‘nonstoichiometry’’ in compounds, i.e., all compounds, referred to as bertholides, exhibit deviation from stoichiometry. However, the extent of deviation is very small (undetectable) in the so-called stoichiometric compounds. Therefore, any compound should always be represented by M 1⫺δ X (e.g., FeO, NiO, CoO, Cu 2 O, etc.) or MX 2⫹δ (e.g., UO 2 ) or MX 2⫺δ or M 2 X 5⫺δ (e.g., SiO 2 , TiO 2 , ZrO 2 , Ta 2 O 5 , Nb 2 O 5 , etc.) or M 1⫹δ X (e.g., ZnO, CdO, BeO, Al 2 O 3 , etc.), where X represents the electronegative component. At a specific temperature and Px 2 (s), the compound would be stoichiometric. The deviation from stoichiometry, δ, is related to Px 2 (g) and n i , the intrinsic defect concentration. The variations of δ as a function of Px 2 (g)/Px 2 (s) for two different compounds having different n i values are presented in Fig. 5.8 [10]. It can be seen that for low intrinsic defect concentrations, it is necessary to have X 2 (g) pressures different from the pressure above the stoichiometric compound by several orders of magnitude before one can obtain measurable deviations from stoichiometry. In contrast, if n i is high, relatively small changes in the pressure of X 2 only will cause significant deviation from the stoichiometric composition. For example, if n i is considered to be 10⫺4, increasing the X 2 pressure 10,000 times the pressure that is in equilibrium with the stoichiometric compound will cause a change of 0.01 in δ. In contrast, at n i ⫽ 5 ⫻ 10⫺3, the deviation is 0.015 when Px 2 (g) is only 10 times Px 2 (s). As examples, one can consider the compounds like KCl, AgBr, TiO, and FeO. At 25°C, n i for KCl is known to be 10⫺17. Thus, one would have to change the pressure of chlorine by many orders of magnitude before any change in stoichiometry could be detected. Even at 973 K, the intrinsic defect concentration is only 4 ⫻ 10⫺5 for which the compound is considered to be stoichiometric. In the cases of FeO and TiO, the deviation from stoichiometry is readily detectable. In AgBr, the deviation is small at and around room temperature (n i ⫽ 8 ⫻ 10⫺6 at 293 K), but heating the crystal to 573 K would make it susceptible to measurable deviations from stoichiometry (n i ⫽ 4 ⫻ 10⫺3 ).

5.4.3

Defect Structure in Stoichiometric Compounds

Kro¨ger and Vink [11] have provided a list of six basic types of defect structure (internal atomic disorder, thermal disorder, and intrinsic defects) that are possible in a stoichiometric compound of type MO. These include: 1. Cation and anion vacancies (V M ⫹ V O ), 2. Vacancies and interstitial ions of the same component (V O ⫹ O i ) or (V M ⫹ M i ), 3. Misplaced atoms (M O ⫹ O M , antistructure),

198

Chapter 5

Figure 5.8 Deviation from stoichiometry (δ) as a function of the more volatile component [Ref. 10].

4. 5. 6.

Vacancies and misplaced atoms for the same type of atom (V M ⫹ M O ), Interstitial and misplaced atoms (O i ⫹ M O ), and Interstitial atoms (M i ⫹ O i , anti-Schottky).

Of these, the types 1 and 2 are known as Schottky and Frenkel disorders, respectively. As yet they are the only ones found to be important in inorganic compounds for which they are illustrated below. Schottky Disorder A stoichiometric crystal with Schottky disorder contains equivalent concentration of cation and anion vacancies. Such type of disorder in a compound (MX) is

Tarnishing and Scaling Processes

199

Figure 5.9 (a) Schematic illustration of Schottky disorder with doubly charged vacancies, V″M and V ••X in a compound, MX. (b) Schematic illustration of Frenkel disorder involving doubly charged cation vacancies and interstitial ions in a compound, MX.

illustrated schematically in Fig. 5.9a. Systems showing predominantly this type of disorder are KCl, NaCl, PbI 2 , PbCl 2 etc. AgCl near its melting point shows almost 5-10% of Schottky defects. Frenkel Disorder A crystal with Frenkel disorder contains the same concentrations of vacancies and interstitial ions for the same kind of atom. A schematic illustration of this

200

Chapter 5

type of disorder involving cations in a compound (MX) is shown in Fig. 5.9b. The most common examples of a system exhibiting such a disorder on metal sublattice include AgBr and AgCl. As a rough rule, Schottky disorder is energetically favored in stoichiometric compounds where the cations and anions are of comparable sizes, whereas Frenkel disorder predominates in systems where the sizes of the cations and anions are appreciably different. To achieve nonstoichiometry and maintain charge balance in an ionic crystal, secondary defects also must exist. Whether the defects will remain free and randomly distributed or in association will depend on the concentration of the defects and temperature. Usually at low concentration and high temperature, the defects will be in free form or unassociated. At low temperatures with high concentrations of point imperfections, the defects are expected to be in associated form leading to formation of defect complexes or clusters.

5.4.4

Defects in Nonstoichiometric Compounds

Nonstoichiometric compounds are of two main types: (1) oxidant-deficient or metal excess with respect to stoichiometric composition and (2) metal-deficient or oxidant excess with respect to stoichiometric composition. The electrical neutrality of nonstoichiometric compounds is conserved through the formation of point defects and complementary electronic defects. The subsequent discussion relates to compounds formed in metal–oxygen systems. Oxygen-Deficient Oxides (n-Type Semiconductors) This could be due to presence of vacant sites in the oxygen sublattices as shown in Fig. 5.10a. Typical oxides in this group ZrO 2 , SiO 2 , TiO 2 , Ta 2O 5 , Nb 2 O 5 , U 3 O 8 , V 2 O 5 , etc. It could also be due to metal excess at interstitial sites as shown in Fig. 5.10b. Examples of this group are ZnO, CdO, BeO, Al 2 O 3 , etc. It may be noted that the formation of both oxygen vacancies and interstitial cations leads to the formation of complementary quasi-free electrons. In oxygendeficient oxides, therefore, the electronic conductivity is due to the transport of electrons in the conduction band. Such oxides are termed n-type semiconductors (negative charge carrier). Metal-Deficient Oxides (p-Type Semiconductors) When a nonstoichiometric oxide is metal deficient, the predominant defects may be either metal vacancies or interstitial oxygen atoms. Figure 5.11 shows a defect structure involving predominantly metal vacancies (M 1⫺δ O). The formation of charged metal ion vacancies leads to the formation of complementary electronic defects (positive holes in the valence band). In such

Tarnishing and Scaling Processes

201

Figure 5.10 (a) Schematic illustration of an oxidant-deficient compound MX 2⫺δ in which doubly charged oxygen vacancies predominate. The electrons are assumed to be localized at M atoms on regular lattice sites, M′M . (b) Schematic illustration of an oxidantdeficient compound (excess metal), M 1⫹δ X 2 , in which doubly charged interstitial cations predominate. The electrons are assumed to be localized at M′M .

oxides the electronic conductivity involves transport of positive holes for which they are called p-type semiconductors (positive charge carrier). In metal-deficient oxides where the disorder is limited to the cation sublattice, particularly in those cases when the concentration of defects is high, the point defects may be more complex than that presented in Figs. 5.10 and 5.11. Neutron diffraction studies by Roth [12] on wu¨stite (Fe 1⫺δ O) have indicated that the vacancy concentration is not in accord with the ideal defect model concept. The resultant complex defect for this system may thus be considered as an interstitial ion adjoined by two vacancies, (V″Fe Fe i••• V″Fe ), as illustrated in Fig. 5.12. They are referred to as Roth clusters. Such complexes can further increase in size with association of additional vacancies, leading to clusters containing 4 trivalent inter-

202

Chapter 5

Figure 5.11 Schematic illustration of a metal-deficient compound, M 1⫺δ X, predominantly containing single-charged metal vacancies. The electron holes are assumed to be localized at M atoms on regular lattice sites, M •M .

stitial cations and 13 cation vacancies (4Fe i••• ⋅ 13V″Fe ), better known as CochCohen clusters [13]. Examples of systems belonging to p-type semiconductors are MnO, FeO, CoO, NiO, and Cu 2 O. Examples of metal-deficient oxides due to interstitial oxygen ion are less. A typical example is UO 2⫹δ . However, the actual situation is much more complex as has been reported through neutron diffraction studies.

Figure 5.12 Schematic illustration of a metal-deficient compound, M 1⫺δ X, with hypothetical complex metal vacancy defect (Roth cluster).

Tarnishing and Scaling Processes

5.4.5

203

Rules for Writing Defect Reaction

In formulating any type of defect formation reaction, it is essential to have mass balance and charge balance, i.e., the electroneutrality condition must be satisfied. For evaluating the defect concentrations in a crystal, the defect equilibria are to be treated in terms of the law of mass action. Use of this law implies the applicability of Boltzmann statistics, i.e., random distribution of defects are considered to be valid. However, at very large concentration of defects this approach breaks down. Moreover, at very low temperatures there will be every likelihood of defect association and less probability for the presence of free defects. Schottky Disorder Formation This type of defect creation occurs as per the following reaction: M M ⫹ O O ⫽ V xM ⫹ V xO ⫹ M M ⫹ O O (at the surface of the oxide lattice) The net reaction is represented by Null ⫽ V xM ⫹ V xO

(5.7)

where null designates a perfect crystal of unit activity. Under the situation of δ ⬍⬍ 1, Henry’s law is applicable. However, it is convenient to use concentration units in terms of number of defect per cubic meter of the crystal represented by [V xM ] and [V xO ] (atom fraction can also be used). Then, one can write from Eq. 5.7: [V xM ] [V xO ] ⫽ K S ⫽ K 1 exp

冢 冣 ∆G 01 kT

(5.8)

It is further known that ∆G 01 ⫽ ⫺RT ln K S ⫽ ∆H 01 ⫺ T∆S 01 Therefore, K 1 exp

冢 冣 冢

冣

∆H 0 ∆S 01 exp ⫺ 1 ⫽ K S k kT

(5.9)

Frenkel Disorder Formation This can be presented by either MO ⫽ M 1⫺δ O ⫹ δV M ⫹ δM i or, simply, M M ⫽ M xi ⫹ V xM

(5.10)

204

Chapter 5

where O sublattice remains undisturbed. Now, according to the law of mass action one can write the following expression:

冢

冣

冢 冣 冢

∆G 0 ∆S 0 ∆H 0 [V xM ] [M xi ] ⫽ K F ⫽ K 2 exp ⫺ 2 ⫽ K 2 exp ⫹ 2 exp ⫺ 2 kT k kT

冣

(5.11)

Some of the various point defects may also be ionized according to the following reaction: V xM ⫽ V′M ⫹ h•

(5.12)

That is, metal vacancies are acceptors that take up electrons from the valence band and inject positive holes to the valence band. Applying law of mass action in Eq. 5.12, one can write the following expression: K′2 ⫽

[V′M ] ⋅ p [V xM ]

(5.13)

where p represents the concentration of positive holes in the valence band. Similarly, M xi ⫽ M i• ⫹ e′

(5.14)

That is, interstitial metal atoms are donors that donate electrons to the conduction band. Hence one can write: K′3 ⫽

[M i• ] ⋅ n [M xi ]

(5.15)

where n represents the concentration of excess electrons in the conduction band. For a metal oxide (MO) exhibiting Frenkel disorder, the defect equilibrium can also be written as: MO ⫽ M xi ⫹

1 O 2 (g) 2

(5.16)

for which K R ⫽ [M xi ]P O1/22

(5.17)

Moreover, the concentrations of excess electrons in the conduction band and positive holes in the valence band are interrelated through intrinsic ionization in the manner: Null ⫽ e′ ⫹ h•

(5.18)

Tarnishing and Scaling Processes

205

Furthermore, the law of mass action allows one to write:

冢 冣

E np ⫽ K i ⫽ A i exp ⫺ i kT

(5.19)

where E i represents the band gap energy. Since the crystal as a whole should be electrically neutral, the following condition should also be fulfilled: n ⫹ [V′M ] ⫽ p ⫹ [M i• ]

(5.20)

In a crystal of MO exhibiting Frenkel disorder, there are six unknown concentration terms like n, p, [V xM ], [V′M ], [M xi ], [M i• ] and six equations (5.11, 5.13, 5.15, 5.17, 5.19, and 5.20). However, solution of these equations frequently becomes quite tedious, particularly in the presence of other types of disorders and foreign ions. Kro¨ger and Vink’s [11] approach has been followed for solving these equations. If one takes the logarithm on both sides of the above-mentioned equations, it is readily seen that all except the last (charge neutrality condition) lead to linear relations between the logarithms of the unknown concentration terms and the known values of constants. For example: log[V xM ] ⫹ log[M xi ] ⫽ log K F

(5.21)

log[V′M ] ⫹ log p ⫺ log[V xM ] ⫽ log K′2

(5.22)

log[M i• ] ⫹ log n ⫺ log[M xi ] ⫽ log K′3

(5.23)

log[M xi ] ⫽ log K R ⫺

1 log PO2 2

log n ⫹ log p ⫽ logK i

(5.24) (5.25)

Equation 5.20, i.e., the one dealing with charge neutrality condition, offers difficulties in taking logarithm because it contains a sum instead of the product of concentrations. However, fortunately any two terms on either side of this equation are usually greater than others depending on the PO2 or PM (g). As a consequence, the neutrality condition simplifies to the type X ⫽ Y, so that one gets a linear relation like log X ⫽ log Y. The pair of the dominant terms in the neutrality condition will be decided by the values of the constants K F , K′2 , K′3 , K R , K i , and PO2 . At extremely high values of PO2 (oxidizing condition), the electroneutrality condition simplifies to p ⫽ [V′M ]

or

log p ⫽ log[V′M ]

(5.26)

206

Chapter 5

On the other hand, at extremely low oxygen pressures, n ⫽ [M i• ]

or

log n ⫽ log[M i• ]

(5.27)

At intermediate PO2 , the electroneutrality condition may be n⫽p

(5.28)

where K i ⬎⬎ K′F , or [V′M ] ⫽ [M i• ]

(5.29)

when K′F ⬎⬎ K i (K′F ⫽ [V′M ][M i• ]). All five sets of above-mentioned equations, together with either of the equations representing simplified charge neutrality condition, will result two types of graphical presentation of concentrations of defect as a function of PO2 , which are known as Kro¨ger-Vink diagrams [11] as depicted in Fig. 5.13a [for n ⫽ p] and 5.13b (for [V′M ] ⫽ [M i• ]), respectively. These are the most simplified diagrams and they obviously become more complicated if there are mixtures of SchottkyWagner and Frenkel disorders along with some foreign aliovalent ions. Figure 5.13a and b clearly exhibits that a compound, MX, would behave either as an n-type or p-type conductor at log PX2 less or greater than stoichiometric compositions.

5.4.6

Defect Formation Reaction in Nonstoichiometric Compounds

Oxygen-Deficient Oxides An oxygen vacancy formation is favored by the transfer of an oxygen atom from a normal lattice site to the gas phase, or by transfer of a metal atom from the gaseous state to the normal cation site, or by moving a metal atom from the normal lattice site to an interstitial site with corresponding release of electronegative component to the gas phase. Such situations can be presented in the following manner: O xO ⫽

1 O 2 (g) ⫹ V xO ; K r ⫽ [V xO ] ⋅ P O1/22 2

(5.30)

or M(g) ⫽ M xM ⫹ V xO ; K r ⫽

[V xO ] PM (g)

(5.31)

or M xM ⫹ O xO ⫽ M xi ⫹

1 O 2 (g); K r ⫽ [M xi ] ⋅ P O1/22 2

(5.32)

Tarnishing and Scaling Processes

207

Figure 5.13 Kro¨ger-Vink plots of log (defect) vs. log PX 2 at a fixed temperature with charge neutrality conditions in the intermediate PX 2 range. (a) n ⫽ p. (b) [M •i ] ⫽ [V′M ].

208

Chapter 5

Examples: ZnO ⫽ Zn xi ⫹

1 O 2 (g); [Zn xi ] ⬀ P ⫺1/2 O2 2

(5.33)

or Zn(g) ⫽ Zn xi ; [Zn xi ] ⬀ PZn

(5.34)

ZrO 2 ⫽ O 2 (g) ⫹ 2V xO ⫹ Zr Zr ; [V xO ] ⬀ P ⫺1/2 O2

(5.35)

1/2 Zr(g) ⫽ Zr Zr ⫹ 2V xO ; [V xO ] ⬀ P Zr(g)

(5.36)

or

Metal-Deficient Oxides 1 O 2 (g) ⫽ O xO ⫹ V xM ; [V xM ] ⬀ P O1/22 2

(5.37)

1 O 2 (g) ⫽ O xi ; [O xi ] ⬀ P O1/22 2

(5.38)

M xM ⫽ M(g) ⫹ V xM ; [V xM ] ⬀ P M⫺1

(5.39)

or

or

Examples: 1 O 2 ⫽ Cu 2 O ⫹ 2[V xCu ]; [V xCu ] ⬀ P O1/42 2

(5.40)

⫺1 Cu 2 O ⫽ 2Cu(g) ⫹ O xO ⫹ 2V xCu ; [V xCu ] ⬀ P Cu(g)

(5.41)

or

Similarly, 1 O 2 ⫽ NiO ⫹ V xNi ; [V xNi ] ⬀ P O1/22 2

(5.42)

⫺1 NiO ⫽ Ni(g) ⫹ O xO ⫹ V xNi ; [V xNi ] ⬀ P Ni(g)

(5.43)

or

Regarding dependence of nonstoichiometry on oxygen pressures or metal pressures, one may conclude that nonstoichiometry increases with increase of oxygen

Tarnishing and Scaling Processes

209

pressure in the gas phase for p-type oxides, whereas defect concentration or nonstoichiometry increases with decrease of oxygen potential in the gas phase for n-type semiconductors. On the basis of defect equilibria, the dependence of theoretical electrical conductivity on oxygen pressure can be predicted. For a large number of systems, such estimated values match fairly well with the experimental data, lending support to the defect models proposed by Wagner and Schottky [14], Hauffe [15], Verway [16], Kro¨ger [11,17], and others. Examples: 1 O 2 (g) ⫽ Cu 2 O ⫹ 2V′Cu ⫹ 2h• 2

(5.44)

for which K(T) ⫽

[V′Cu ] 2 ⋅ p 2 P O1/22

(5.45)

According to the following electroneutrality condition: [V′Cu ] ⫽ p

(5.46)

therefore, one obtains, p ⬀ σ h• ⬀ P O1/82

(5.47)

The experimentally found relation is σ h• ⬀P O1/72 , which supports the defect model proposed by Wagner and Gru¨newald [18]. Similarly, 1 O 2 (g) ⫽ NiO ⫹ V″Ni ⫹ 2h• 2

(5.48)

for which K(T) ⫽

[V″Ni ] ⋅ p 2 P O1/22

(5.49)

When the electroneutrality condition becomes 2[V″Ni ] ⫽ p

(5.50)

one would be able to write the following from Equation 5.49: p ⬀ σ h• ⬀ P O1/62

(5.51)

210

Chapter 5

This type of pressure dependence matches with experimental results in a certain temperature range. On the other hand, in the case of a Zn/ZnO/O 2 (g) system, one can visualize the following defect formation reaction as ZnO ⫽ Zn i• ⫹ e′ ⫹

1 O 2 (g) 2

(5.52)

for which K(T) ⫽ [Zn i• ] nP O1/22

(5.53)

When n ⫽ [Zn i• ], one finds the following relation: n ⬀ σ e′ ⬀ P ⫺1/4 O2

(5.54)

When interstitial Zn ion is doubly ionized, the charge neutrality condition becomes n ⫽ 2[Zn ••i ]. Therefore, one can visualize the following defect formation reaction: 1 ZnO ⫽⫽ Zn ••i ⫹ 2e′ ⫹ O 2 (g) 2

(5.55)

The equilibrium constant for Eq. 5.55 is K(T) ⫽ [Zn ••i ]n 2 P O1/22

(5.56)

Therefore, n ⬀ σ e′ ⬀ P ⫺1/6 , O2

(5.57)

Measured conductivity dependence is reported to be between P ⫺1/4 and P ⫺1/5 O2 O2 , suggesting the presence of some singly charged as well as doubly charged Zn interstitial ions. In conclusion, the electrical conductivity of a p-type compound increases with increase of the oxidant pressure whereas that for an n-type conductor decreases with increased oxidant pressure in the environment.

5.5 MECHANISMS OF TARNISHING AND SCALING PROCESSES In 1920, Tammann [6] for the first time proposed a mathematical relation between the extent of reaction of a metal with the environment and the corresponding exposure time of reaction at elevated temperatures for a number of metal–oxidant systems. This relation was experimentally found to be parabolic in nature. Subsequently, in 1923, the important concept of the Pilling-Bedworth [5] ratio had emerged, which still serves as a rough guideline for the prediction of protective

Tarnishing and Scaling Processes

211

or nonprotective oxide scale growth on the metal substrate. However, it was not until 1933 and subsequently in 1936 that the pioneering theoretical relation for the parabolic film/scale growth process was proposed by C. Wagner [19] for a number of metal–oxidant systems at high temperatures. He proposed that for thick scale formation, the reactions at metal–oxide and oxide–oxygen interfaces are fast enough and migration of ionic or electronic species through the reaction product layer should be the rate-limiting process for subsequent thickening of the film/scale. This mathematical formulation marked the beginning of theoretical developments in the field of oxidation and tarnishing reactions of metals and alloys. It was rightly pointed out by Wagner that the migration mechanism must be closely related to the types and concentrations of point defects in the growing product lattice and is controlled by the difference in electrochemical potentials between the metal/scale and scale/gas phase boundaries. It is also known from the chemical physics of semiconductors and ionic compounds that the presence of a favorable electrical field across the growing layer tends to increase the mobility of ions and electrons via lattice imperfection and complementary electronic defect in the crystalline solids of the reaction product. Thus, measurements of electrical conductivity, Hall coefficient, and thermoelectric power of the solid reaction product with identification of the relative position of an inert marker within the compact scale would provide enough information as to the kind and extent of lattice defects present in product compounds. These would help in drawing conclusions about the type and extent of migration process for ions and electrons in the scaling layer. Through an elaborate mathematical treatment, Wagner could correlate the parabolic rate constant with some of the physical parameters of the scale that are highly dependent on the types and concentrations of point defects in the growing film/scale. In deriving the parabolic law for thick film or scale growth process, he considered the diffusing species to be essentially induced to migrate through the growing film by the action of two kinds of driving forces. The first one is a chemical potential gradient that is set up across the layer as a result of attainment of equilibrium conditions at the two reaction interfaces, i.e., metal/scale and scale/gas, respectively. The second is an electrical potential gradient that arises due to the formation of negatively charged ionized species of the oxidant at the outer interface of the scale at the expense of electrons available from the metal through the conduction band of the product layer. The occurrence of such chemical potential gradients or concentration gradients is schematically illustrated in Fig. 5.14a and 5.14b for oxide scales having a predominance of metal vacancies and oxidant vacancies, respectively. The partial pressures of oxygen at the metal– oxide interface as depicted in Fig. 5.14a and 5.14b refer to the equilibrium dissociation pressure of the oxide in contact with the metal substrate [P (d) O 2 ], whereas the corresponding ones at the oxide–oxygen interface refer to the oxygen pres-

Figure 5.14 (a) Schematic representation of concentration gradient of metal ion vacancies and the transport processes occurring in an oxide scale containing mostly metal ion vacancies. (b) Schematic representation of concentration gradient of oxygen ion vacancies and the transport processes occurring in an oxide scale containing mostly oxygen ion vacancies [Ref. 57].

Tarnishing and Scaling Processes

213

sures prevailing in the gas phase [P (g) O 2 ]. For an oxide scale with a predominance of metal vacancies, the metal ions diffuse outward from the M/MO to MO/O 2 (g) interface. The metal vacancies created at the outer surface migrate in the opposite direction and their equilibrium concentrations at the interfaces are given by the relevant defect equilibrium as presented in Fig. 5.14a. Since normally P (g) ⬎ P (d) O2 ⬎ O 2 , the metal vacancies are continuously produced at the MO–O 2 (g) interface and are consumed or annihilated at the M–MO interface. On the other hand, oxygen ion vacancies migrate in opposite direction to the metal ion vacancies and their equilibrium concentrations at the two reaction interfaces of the oxide are presented in Fig. 5.14b. In such a system, oxygen vacancies are continuously created at the M/MO phase boundary and are simultaneously ⬍ P (g) consumed at the MO–O 2 (g) interface because P (d) O2 ⬍ O 2. For thick film or scale formation at relatively high temperatures, of the abovementioned two driving forces, the contribution from electrical field to the film growth process becomes negligible. Therefore, under such a situation, thermally aided diffusion of ionic or electronic species due to the existence of a chemical potential gradient provides the major driving force. If the temperature is too low, diffusion of ions is not easily possible due to lack of thermal activation. Under such circumstances, Wagner’s parabolic law [19] is not satisfied. This indirectly brings a thought in one’s mind regarding two additional factors that need consideration. The first is the type of migrating species (electrons, positive holes, ions, vacancies, interstitials, etc.) and the second is the necessary thermal energy, which is again highly dependent on the nature of the product layer, i.e., on some of its physical and chemical properties. As was pointed out by Wagner, the migrating or diffusing species in a tarnishing or scaling layer are usually the cations, anions, electrons, or positive holes. The slowest moving species automatically controls the movement of other species, as otherwise the charge neutrality condition will not be maintained in the bulk film. The absence of associated species or defects has also been assumed.

5.5.1

Thick Scale Formation Mechanism; Wagner’s Parabolic Law and Its Applications

The basic assumptions of Wagner’s parabolic law [19] are as follows: 1. Scale should be adherent to the metal substrate, compact and nonporous in nature such that solid state diffusional transport through the product lattice remains the main process. 2. Phase boundary reactions being quite fast, equilibrium attainment at the metal–scale and scale–gas interfaces is assumed to be true. 3. Since thick scale formation is being considered, the chemical potential gradient across it acts as the major driving force, with a minimal contribution from electrical field gradient.

214 4. 5.

6. 7.

8.

Chapter 5 Transport of only charged species is considered. Fulfillment of the charge neutrality condition in the total scale is assumed, i.e., any section of the growing scale perpendicular to the direction of migrating species is essentially electrically neutral. Interfacial space charge effect is neglected. Lattice diffusion (or volume diffusion) in the product scale is only considered; no short circuit or easy diffusion paths such as grain boundaries, dislocation pipes, and so on are taken into account. Absence of associated species or imperfections is assumed, i.e., dilute solution model has been assumed to be applicable, which means nonexistence of interactions among ionized imperfection centers for formation of defect clusters or complexes. The migrating species are free to move.

Keeping in mind the above-mentioned basic assumptions, one can write the following generalized flux equation for the transport of migrating species in a compound layer like MX (X stands for oxygen, halogen, sulfur vapor, etc.): Ji ⫽

冢

冣

冢

δφ number 1 δµ i ⫹ Zi e ⫽ ⫺C i B i 2 m s N δξ δξ

冣

(5.58)

Such equations can be formulated for the three different migrating species, such as cations, anions, and electrons or positive holes. where Ci Bi N δµ i /δξ δφ/δξ e Zi

⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽

concentration of ith species in number per meter cube, mobility of ith species under unit force in m 2 J⫺1 s⫺1, Avogadro’s number, 6.023 ⫻ 1023, chemical potential gradient in J ⋅ m⫺1, electrical potential gradient in V ⋅ m⫺1, charge on the species i in coulombs, valence of ith species.

Along with it one has to consider the following charge neutrality relation: | J 1 | ⫺ |J 2 | ⫺ |J 3 | ⫽ 0 where | J 1 | ⫽ magnitude of cationic flux, | J 2 | ⫽ magnitude of anionic flux, and | J 3 | ⫽ magnitude of electronic flux.

(5.59)

Tarnishing and Scaling Processes

215

Furthermore, one should consider the following thermodynamic relations µ1 ⫹ | Z1 | µ3 ⫽ µM and

(5.60)

µx ⫹ | Z2 | µ3 ⫽ µ2 where µ 1 , µ 2 , µ 3 , µ M , µ x ⫽ chemical potentials of cations, anions, electrons, metal atoms, and oxidant atoms, respectively; Z 1 and Z 2 are valences of metal and nonmetal atoms. Thereafter one may utilize the Gibbs-Duhem equation as given by n M dµ M ⫹ n X dµ X ⫽ 0

(5.61)

along with the stoichiometric relation: n M |Z 2 | ⫽ nX |Z 1 |

(5.62)

where n M and n X represent number of kg atoms of metal and oxidant, respectively. Finally, the parabolic rate equation as derived by Wagner takes the following form in which the roles played by both ionic and electronic defects and their corresponding mobilities become apparent:

冢

1 1 dn ⫽ 2 A dt F | Z2 |

冮

µ (o) X

(i)

µX

冣

(t 1 ⫹ t 2 ) t 3 σdµ X

1 kr ⫽ ξ ξ

(5.63)

where dn/dt ⫽ the rate expressed in kg equivalent s⫺1, t 1 , t 2 , t 3 ⫽ transport numbers of cations, anions, and electrons (or positive holes), respectively, A ⫽ area of the oxide perpendicular to the direction of the diffusing species (or area of the sample, m 2), F ⫽ Faraday number ⫽ 96,500 ⫻ 103 coulomb (kg equiv.)⫺1, dµ X ⫽ difference in chemical potentials of oxidant, expressed in Joules (kg atom)⫺1, (o) (i) µ X and µ X ⫽ chemical potentials of nonmetal (oxidant) at the oxide–oxygen and metal–oxide interfaces, respectively, σ ⫽ total electrical conductivity of the scale or oxidation product, in S ⋅ m ⫺1, ξ ⫽ instantaneous thickness of the product scale, and the term in parentheses is the rational rate constant (k r ) expressed in kg equivalent m⫺1 s⫺1, i.e.,

216

Chapter 5

冢

1 kr ⫽ 2 F | Z2 |

冮

µ (o) X

(i) µX

冣

(t 1 ⫹ t 2 )t 3 σdµ X

(5.64)

Regarding temperature of exposure, Wagner’s equation (5.63) provides no clue as to what should be the minimum temperature for its validity. Experimental results on the oxidation of copper show that activation energy for the oxide growth rate is identical to that for diffusion rate of Cu⫹ ions in Cu 2 O. Consideration of experimental data on Cu–Cu 2 O–O 2 (g) and a few other systems roughly indicates that the minimum temperature for validity of Wagner’s parabolic rate law will be around 0.4Tm , where Tm is the melting point of scale compound. Further investigations relating to halogenation kinetics of silver and copper (halide growth process) have shown the obeyance to Wagner’s rate law mechanism even at a temperature as low as 303 K. Of course, this is true beyond a certain halide thickness range depending on the individuality of the system. The electrochemical steps in diffusional growth of oxide scales involving two kinds of mass and charge transport are schematically presented in Fig. 5.15. These clearly depict that the diffusional growth of a corrosion product formed on a metal substrate at elevated temperature is essentially an electrochemical process in which the oxidation and reduction reactions are taking place at the two reaction interfaces, i.e., metal–oxide and oxide–oxygen, respectively. The scales are both ionic and electronic conductors and the galvanic cell formed remains active by internal short circuiting. One can now examine the application and verification of Wagner’s theoretical expression for different metal–oxidant systems as reported by a large number of investigators. As illustrations, three metal–oxidant systems are being chosen in which oxidation product layers are (1) predominantly ionic, (2) predominantly p-type oxide (metal-deficient or oxygen excess), and (3) predominantly n-type oxide (metal excess or oxygen-deficient). (1) If one considers an iodide film growth on silver, it is to be remembered that an isolated crystal of AgI is an ionic conductor exhibiting predominance of Frenkel type of disorder. That is, the concentration of vacancies on silver sublattice and an equal number of silver ions at interstitial sites is of very high order compared to intrinsic electronic defects like positive holes in the valence band and excess electrons in the conduction band of AgI film. However, when AgI film initially grown on Ag is further exposed to iodine vapor, it exhibits electronic conductivity due to injection of positive holes in the valence band. Subsequent iodine incorporation into the growing AgI lattice can be represented by the following defect formation reaction: 1 I 2 (v) ⫽ I XI ⫹ V′Ag ⫹ h• 2

(5.65)

Tarnishing and Scaling Processes 217

Figure 5.15 Electrochemical steps involved in diffusional growth of scales.

218

Chapter 5

where h• is a positive hole and V′Ag is a negatively charged vacancy on Ag sublattice. This means that iodine on the lattice surface of AgI will create extra acceptor centers like vacancies on silver (V XAg ), which becomes negatively charged (V′Ag ) by accepting electrons from the valence band, thus injecting positive holes in the valence band. Assuming the validity of simple mass action law, one can write the following expression for equilibrium at the outer interface of the growing film: K e (T) ⫽

[V′Ag ] ⋅ p P I1/2 2

(5.66)

where K e (T) is the equilibrium constant at any temperature T, [V′Ag ] is the concentration of singly negatively charged metal vacancies in number per m 3, p is the concentration of positive holes in number per m3, and P I 2 is the pressure of iodine vapor in the environment. Since inherent concentration of cation vacancies are already high (predominantly ionic lattice) with fulfillment of the existence of virtual ionic current equilibrium across the iodide film, incorporation of extra iodine into the growing halide lattice will not bring in much change in the concentration of cation vacancies. So the concentration of positive holes ( p) will be decided by the pressure of the oxidant (iodine). Accordingly, Eq. 5.66 is modified to the following form assuming [V′Ag ] to remain virtually constant: K e (T) p ⫽ K′e (T) ⫽ 1/2 [V′Ag ] P I2

(5.67)

As pointed out earlier, virtual ionic current equilibrium does exist across the iodide film for such a system, so the rate of iodide layer thickening will be controlled by the transport of positive holes, i.e., by the conductivity of positive holes (σ h• ). Then one can write from Eq. 5.67: σ h• ⬀ p ⬀ P I1/2 2

(5.68)

The above relation has been experimentally verified for the growth of AgCl, AgBr, and AgI films on silver [20] for film thicknesses exceeding certain values. The electrical field strength (E ⫽ V/ξ ⫽ volt/m) in such cases is very weak and can be neglected. The expression (5.68) is obtained from the relation σ ⫽ q i c i v i , where q i ⫽ charge (in coulomb) on the rate limiting species, v i ⫽ charge mobility in m 2 V⫺1 s⫺1 which remains virutally constant at any temperature (T), and C i ⫽ concentration of the rate-limiting species expressed in number per m 3. One can reexamine the Wagner’s theoretical relation (5.63) for Ag–AgI–I 2 (v) system. In this case, t 1 ⫽ tAg⫹, ⯝1, t 2 ⫽ tI⫺ ⯝0 (iodide sublattice remains almost (at constant stationary), and t 3 ⫽ t h•. Therefore, one may write, t h• σ ⫽ σ h• ⫽ σ°h• P I1/2 2 temperature), where, σ°h• ⫽ hole conductivity at PI2 ⫽ 1 atm (1.01325 ⫻ 10 5 N. m ⫺2 ) and Z 2 ⫽ Z I⫺1 ⫽ unity.

Tarnishing and Scaling Processes

219

It is also known from thermodynamics that dµ I ⫽

1 RT dµ I 2 ⫽ d ln PI 2 ⫽ dµ x 2 2

(5.69)

Now utilizing this expression (5.69) in the rational rate constant (k r ) given by Eq. 5.64, one obtains kr ⫽

⫽

冮

1 RT ⋅ F2 2 RTσ°h• 2F 2

P (o) I2

(i) PI 2

冮

σ°h• (PI 2 ⫽ 1 atm) ⋅ P I1/2 d ln PI 2 2

P (o) I2

(i) PI 2

P I1/2 2

dPI 2 PI 2

(5.70)

(5.71)

Therefore, kr ⫽

RT (o) (i) ⋅ σ°h• [P I1/2 ⫺ P 1/2 I2 ] 2 F2

(5.72)

Since P (o) I 2 at the AgI/I 2 (v) interface at any temperature is much greater than the equilibrium P I(i)2 at the Ag/AgI, i.e., P (o) ⬎ P I(i)2 , the expression for rational rate I2 ⬎ constant (k r ) simplifies to the form: kr ⫽

RT • 1/2 σ°h P I 2 (kg equivalent m⫺1 s⫺1 ) F2

(5.73)

Parabolic rate constant values for the growth of thick films of AgCl, AgBr, and AgI in corresponding halogen atmospheres conform to the theoretically predicted pressure dependence [20] and thus confirm the defect model as has been proposed by Wagner [19] for such types of ionic halide lattice. (2) When oxidation product layers are predominantly p-type conductors, e.g., FeO, NiO, CoO, MnO, Cu 2 O, Cu 2 S, FeS, CuI, UO 2 , etc., the compounds are nonstoichiometric in nature, exhibiting either metal deficiency or oxidant excess. Oxidation of copper at high temperatures can be cited as an example. During oxidation of copper, incorporation of oxygen into Cu 2 O lattice leads to the following defect formation equilibrium: 1 O 2 (g) ⫽ O xO ⫹ 2V′Cu ⫹ 2h• 2

(5.74)

In p-type oxides, virtual electronic current equilibrium condition is fulfilled across the oxide layer. Since concentration of defects are small compared to the total number of lattice sites, application of ideal mass action law to Eq. 5.74

220

Chapter 5

will result in the following expression of equilibrium constant [K e (T)] at any temperature (T): K e (T) ⫽

[V′Cu ] 2 ⋅ p 2 P O1/22

(5.75)

Furthermore, the validity of the following charge neutrality condition at the outer interface of the oxide layer suggests [V′Cu ] ⫽ p

(5.76)

So Eq. 5.75 simplifies to [V′Cu ] 4 ⬀ K e (T) ⋅ P O1/22 or [V′Cu ] ⬀ P O1/82

(5.77)

Since virtual electronic current equilibrium prevails in the oxide layer, migration of Cu⫹ ions through vacant cation sites will become rate limiting for subsequent thickening of Cu 2 O scale on copper substrate in oxygen atmosphere. In the expression of Wagner’s equation (5.64) for rate constant, since t 1 ⫽ t Cu⫹ ⬍⬍ t 3 ⫽ t h• ⯝ 1 and t 2 ⫽ t O2⫺ ⯝ 0 (almost stationary oxygen sublattice), | Z O⫺2 | ⫽ 2; therefore, one may write σt Cu⫹ ⫽ σ Cu⫹ ⫽ σ 0Cu⫹ (PO2 ⫽ 1 atm) ⋅ P O1/82 (theoretical). However, experimentally obtained parabolic rate constants exhibit the following oxygen pressure dependence relation: k r ⬀ P O1/72

(Wagner and Gru¨newald [18] at 1273 K).

Subsequent studies at 1123 K by Ananth et al. [21–23] have confirmed the validity of this relation. In either investigation, with plots of k r vs. P O1/72 when extrapolated to k r ⫽ 0, the corresponding values of PO2 agree quite well with the equilibrium PO2 at the Cu–Cu 2 O interface calculated from thermodyamic data, lending support to the defect model as proposed by Wagner and coworkers [18] for Cu–Cu 2 O–O 2 (g) system. Slight deviation from theoretically expected pressure dependence relation (k r ⬀ P O1/82 ) has been attributed to the presence of some neutral vacancies (V XCu ) along with singly charged vacancies whose concentration predominates over the concentration of neutral vacancies. However, observation of k r ⬀ P O1/42 by Mrowec [24,25] has been attributed to a predominance of neutral vacancies in the Cu 2 O lattice that is not in accord with reports by Wagner et al. [18], Ananth et al. [21– 23], and Bose et al. [26,27]. Reasons for the discrepancy have been elaborated in recent publications [21–23,26,27]. Similar oxygen pressure dependence of rate constants can also be assumed as

Tarnishing and Scaling Processes

221

valid for ionic conductivity, i.e., σ Cu⫹ ⫽ σ 0Cu⫹ P O1/72 (as per Wagner [28]). Furthermore, one can write the following relation from a thermodynamic standpoint: dµ O ⫽

1 RT dµ O2 ⫽ d ln PO2 2 2

(5.78)

Now, utilizing the above expressions, the rational rate constant for copper oxidation simplifies to the form: kr ⫽

RT 0 ⫹ σ Cu 4F 2

冮

P (o)

O2

P O1/72 d ln PO2

(5.79)

⫽

7 RT 0 ⫹ 1/7 (o) (i) σ Cu [P O2 ⫺ P 1/7 O2 ] 4 F2

(5.80)

⫽

7 RT 0 ⫹ 1/7 (o) σ Cu P O2 4 F2

(5.81)

(i) PO 2

(i) [since P O1/72 (o) ⬎⬎ P 1/7 O2 ] The close agreement of theoretically estimated rate constant values to those experimentally obtained further confirms the defect model of Cu 2 O as proposed by Wagner [18]. (3) When the oxidation products are predominantly n-type conductors, e.g., ZnO, CdO, BeO, SiO 2 , ZrO 2 , TiO 2 , etc., the compounds formed on the corresponding metals are nonstoichiometric, exhibiting either excess metal ions at interstitial sites or vacancies at oxidant ion sites. Oxidation behavior of zinc in the temperature range of 573–673 K [18] can be considered as an example to illustrate such systems. The following defect formation equilibrium occurs when oxidation of zinc takes place in an oxygen atmosphere at elevated temperatures:

ZnO ⫽ Zn ••i ⫹ 2e′ ⫹

1 O 2 (g) 2

(5.82)

Here conductivity is mainly due to the transport of excess electrons in the conduction band of ZnO lattice, i.e., virtual electronic current equilibrium exists across the growing ZnO on the Zn substrate. Therefore, the thickening process of ZnO film will be controlled by the transport of interstitial Zn ions through the ZnO lattice. One can write the following expression for equilibrium constant of the above defect equilibrium (5.82) at any temperature (T), assuming the validity of the mass action law: K e (T) ⫽ [Zn ••i ] ⋅ n 2 P O1/22

(5.83)

222

Chapter 5

where n represents number of excess electron in the conduction band per m 3 of ZnO and [Zn ••i ] represents concentration of interstitial Zn ions per m 3 of ZnO. Utilizing validity of the following charge neutrality condition, n ⫽ 2[Zn ••i ] one obtains from Eq. 5.83 [Zn ••i ] ⬀ P ⫺1/6 O2

(5.84)

Since ZnO is mainly an electronic conductor, t 3 ⯝ 1, t Zni ⬍⬍ 1, and tO⫺2 ⫽ 0 (almost stationary oxygen sublattice). Therefore, one can write: σ Zn••i ⫽ σ

t Zn ••i ⫽ σ 0Zn••i (PO2 ⫽ 1 atm) ⋅ P ⫺1/6 O2

(5.85)

The expression for rational rate constant can thus be presented as: kr ⫽

冤

3 RT 0 •• 1 1 σ Zn i ⫺ (i) 1/6 (o) 1/6 2 F2 P O2 P O2

冥

(5.86)

The lattice defect structure of ZnO exhibits the concentration of interstitially positioned Zn ••i ions at the Zn–ZnO phase boundary to be considerably greater than that at the ZnO–O 2 (g) interface. Furthermore, it follows from the standard free energy of formation of ZnO ⬍ P (o) that P (i) O2 ⬍ O 2 and thus 1 1 ⬎⬎ (o) (i) P O2 P O2 and this is more true for the sixth root. So the second term in parentheses of Eq. 5.86 can be neglected. Therefore, the expression of rate constant simplifies to: kr ⫽

冢 冣

3 RT 0 •• 1 σ Zn i 2 F2 P O(i)2

1/6

(5.87)

Since P O(i)2 is constant at a constant temperature, k r should be independent of variation of oxygen pressure at the outer surface of ZnO film. Actually, Wagner and Grunewald’s [18] experiment on Zn oxidation at 663 K (Fig. 5.16) yielded the scaling rate constant values lying between 0.72 ⫻ 10⫺10 g 2 cm⫺4 S⫺1 (at PO2 ⫽ 0.022 atm) and 0.75 ⫻ 10⫺10 g 2 cm⫺4 s⫺1 (at PO2 ⫽ 1 atm). Therefore, it is concluded that for the growth of a compact, adherent, n-type conducting layer on a metal substrate, the parabolic rate constant becomes almost independent of variation of oxygen pressure in the environment at a particular temperature. One can now well appreciate that the oxidation rate of any metal will be largely dependent on the type and concentration of point defects in the product film/ scale as well as on its transport properties, i.e., electrical conductivity, diffusivity,

Tarnishing and Scaling Processes

223

Figure 5.16 Concentration of zinc ions in interstitial lattice positions through the ZnO layer, according to Wagner. Lines 1 and 2 show the concentration decrease of the zinc ions in the interstitial lattice positions at low and high oxygen pressures. As can be seen, (dc ZnO• /dξ) 1 ⬇ (dc ZnO• /dξ) 2 [Ref. 15].

etc. Al 2 O 3 , Cr 2 O 3 , or SiO 2 layer growth rates on corresponding substrate are much slower in comparison to the majority of base metal oxides. This is because these compounds are nearly stoichiometric in nature. In other words, their inherent defect concentrations are quite small. On the other hand, FeO, NiO, or Cu 2 O growth rates on the corresponding metal are much faster, and this is attributable to the presence of appreciable concentrations of point defects in such oxides. It has been reported (29) that under similar experimental conditions, the concentration of defects in FeO is ⬃10 at. %, that in NiO ⬍0.1 at. %, and that in Cr 2 O 3 negligibly small (⬃0.001 at. %). Figure 5.17 represents the comparative oxidation rates of Fe, Ni, and Cr at 973 K in one atmosphere oxygen pressure. This figure shows that the chromium oxidation rate is only 1/1000th the value of that for iron. Moreover, it is known that the rate constants are related to electrical conductivity of the compounds being formed (2). Some of the conductivity values for various oxides are presented in Table 5.3. Actually, the rate constant values are controlled by the transport coefficient factor, which is nothing but the product of diffusivity (D i ) and concentration (C i ) of the species i, the value of which will decide the rate-limiting step for thickening of the oxide layer. Subsequently, Wagner [30] derived another expression of rational rate constant similar to the original one (5.64). In the latter expression, the rate constant

224

Chapter 5

Figure 5.17 Parabolic oxidation rate plots for iron, chromium, and nickel at 700°C in oxygen atmosphere [Ref. 29].

Table 5.3 Electrical conductivity of some oxides at 1273 K [Ref. 2] Oxide BeO Al 2 O 3 CaO SiO 2 MgO NiO Cr 2 O 3 CoO Cu 2 O FeO σ(ohm⫺1 cm⫺1 ) 10⫺9 10⫺7 10⫺7 10⫺6 10⫺5 10⫺2 10⫺1 10⫹1 10⫹1 10⫹2

Tarnishing and Scaling Processes

225

has been related to the self-diffusion coefficients, D 1 and D 2 of metal and nonmetal ions, respectively, for an oxide of t e ⯝ 1. Using Einstein’s relation, D i ⫽ B i kT, where B i ⫽ drift velocity of species i per unit force; Wagner finally obtained the following expression for k r : k r ⫽ C equiv

冮

a (o)

冮

a (i)

O

(i) aO

冤|Z| Z || D ⫹ D 冥 d ln a 1

1

2

O

(5.88)

M

(5.89)

2

or k r ⫽ C equiv

M

(o) aO

冤D ⫹ |Z| Z || D 冥 d ln a 2

1

2

1

where C equiv ⫽ (concentration in kg equivalent of compound) ⋅ m⫺3 D 1 and D 2 ⫽ self-diffusion coefficient of cations and anions, respectively, in the compound layer, expressed in m2 s⫺1 Z 1 and Z 2 ⫽ valence of metal and nonmetal atoms, respectively, (i) a (o) O and a O ⫽ activity of oxidant at the outer oxide–oxygen and the inner metal–oxide interfaces, respectively (i) a (o) M and a M ⫽ activity of metal atoms at the outer oxide–oxygen and the inner metal–oxide interfaces, respectively Wagner’s theoretical relation (5.88) for parabolic rate constant has been successfully tested and found to be valid for a large number of metal–oxidant systems. However, his expression considered only lattice or volume diffusion in the growing product layer but did not take into account the diffusion through grain boundary and other easy diffusion paths. In many a situation, porosities and microchannels are developed within the scale where this model is not fully satisfied. Moreover, no specific mention has been made regarding the temperature of applicability of this theoretical relation. Subsequent experimentations have established that the temperature of validity of this mechanism will vary from one system to another.

5.5.2

Determination of Self-Diffusion Coefficient from Parabolic Rate Constant

Fueki and Wagner (31) and subsequently Pettit (32) evaluated self-diffusion coefficient from the knowledge of oxygen pressure dependence of rational rate constant (k r ) by following an inverse procedure, i.e., by differentiating Eq. 5.88 as proposed by Wagner (30), which correlates k r and diffusivity values of ionic species. One would obtain the following expression by differentiating Eq. 5.88:

226

Chapter 5

dk r ⫽ C equiv

冤|Z| Z || ⋅ D ⫹ D 冥 d ln a 1

1

2

O

2

or

冤|Z| Z || ⋅ D ⫹ D 冥 ⫽ C1

dk r equiv d ln a O

(5.90)

2 dk r | Z1 | D1 ⫹ D2 ⫽ (since a O ⫽ P O1/22 ) |Z 2 | C equiv 2.303 d log PO2

(5.91)

1

1

2

2

or

In the case of nickel oxidation to NiO, where D 2 ⫽ D O2⫺ which is negligibly small compared to D 1 ⫽ D Ni2⫹ and | Z 1 |/| Z 2 | ⫽ 1; Eq. 5.91 simplifies to: D Ni 2⫹ ⫽

2 ⋅ M oxide dk r 2 ⋅ ρ oxide ⋅ 2.303 d log PO2

(5.92)

where M oxide /2 ⫽ molecular mass of oxide in kg/equivalent number and ρoxide is expressed in kg ⋅ m⫺3. Furthermore, the relation between k r (kg equiv. m⫺1 s⫺1 ) and k″p (practical rate constant, kg 2 m⫺4 s⫺1 ) is given by: kr ⫽

M oxide ⋅ k″p ρ oxide ⋅ M 2O

(5.93)

where M O ⫽ atomic mass of oxygen in kg. Finally one obtains the following expression for diffusivity of Ni 2⫹ in NiO as: D Ni

2⫹

冢

M NiO ⫽ ρ NiO ⋅ M O

冣

2

⋅

dk″p 1 2.303 d log PO2

(5.94)

The self-diffusion coefficient values of Ni 2⫹ ions in NiO (31), Fe 2⫹ ions in FeO (33), and Cu⫹ ions in Cu 2 O [21,22] thus estimated match fairly well with those obtained by other techniques as reported in the literature. Subsequently, Mrowec employed a much simpler method [24,25] for the estimation of self-diffusivity through oxidation kinetics study. The procedure he adopted is as follows: It is known that k r ⫽ (constant) P O1/n2 at any constant temperature. Moreover, D 1 ⫽ D o1 (PO 2 ⫽ 1 atm) ⋅ P O1/n2 . Therefore, the expression of k r is given by: kr ⫽

冮

(o) PO

(i) PO 2

2

|Z | C equiv ⋅ 2.303 1 ⋅ D 01 P O1/n2 d log PO2 2 |Z 2 |

2.303 ⋅ C equiv |Z 1 | 1/n nD 01 P (o) ⫽ O2 , 2 |Z 2 |

[since P

(5.95) (o) O2

⬎⬎ P ] (i) O2

Tarnishing and Scaling Processes

227

Taking logarithms on both sides, one obtains: log k r ⫽ log

⋅C 冤2.303 2

equiv

冥

1 |Z 1 | ⋅ nD 01 ⫹ log P (o) O2 |Z 2 | n

(5.96)

Therefore, a plot of log k r vs. log P (o) O 2 will give the value of slope (1/n). Now, 0 from the intercept at log P (o) ⫽ 0, i.e., P (o) O2 O 2 ⫽ 1 atm, one gets D 1 and hence D 1 at any PO2. Such a procedure has been utilized for estimation of D Cu⫹ in Cu 2 O [21,22,27] and the values match fairly well to those calculated following others’ methods.

5.5.3

Thin Film Growth Mechanisms

It is a common experience that many metals, perhaps all, that oxidize readily exhibit similar behavior when exposed to oxygen/dry air at sufficiently low temperatures. In such cases, oxidation rate is initially extremely rapid, but after a few minutes or hours of exposure to oxidizing atmosphere, the rate drops to a ˚ being formed. The very low or negligible value, with a stable film of 20–100 A kinetics of thin oxide film growth at room temperature for some of the metals are presented in Fig. 5.18. As an example, at relatively low temperatures below 323–333 K, copper oxidizes rapidly in the beginning and then virtually ceases ˚ thickin a logarithmic fashion, forming a stable and protective film of 40–50 A ness. An explanation to such a behavior was first given by Cabrera and Mott [7], which is based on the idea of sustenance of a strong electrical field across the oxide film. Consistent with the observed phenomena, they proposed a theory for

Figure 5.18 Thickness of oxide films formed on various metals at room temperature [Ref. 2].

228

Chapter 5

formation of very thin oxide films and their derivation of rate equation is based on the following steps: 1.

2. 3.

4.

Adsorption of oxygen on the oxide surface (a few monolayers only) followed by its ionization due to electrons captured from the metal at the metal–oxide interface by electron tunneling, which is a quantum mechanical phenomenon. Formation of cation vacancies at the oxide–oxygen interface. Subsequent migration of these vacancies through the oxide or, in other words, the migration of metal ions through these vacant cationic sites under the action of an electrical field produced by step (1). Destruction or annihilation of the cationic vacancies at the metal–oxide interface.

˚ ), the acting electrical Since thickness of the oxide film is of low order (⬍100 A field strength (E ⫽ V/ξ) will be as high as 10 8 V/m. Such an extremely high field strength will facilitate the movement of migrating species by lowering the energy necessary for migration. They also made three more simplifying assumptions: 1. 2.

3.

In the thin and very thin films, the space charge may be considered uniform and thus their effects may be neglected. The electrical potential difference between the metal–oxide and the oxide– oxygen interfaces is independent of the oxide thickness. Subsequently, this concept received support from Fromhold’s [34,35] detailed theoretical analysis. A uniform defect concentration is maintained in the oxide layer.

Considering the above assumptions, after an elaborate mathematical treatment, Cabrera and Mott arrived at an equation of the form: ξ dξ ⫽ K exp 1 dt ξ where K ⫽ NΩν exp(⫺W/kT ) ξ1 ⫽

1 qa ⋅V 2 kT

where ξ dξ/dt N Ω ν

⫽ ⫽ ⫽ ⫽ ⫽

thickness of film in meters at any time (t), in seconds, rate of oxidation or rate of film growth at ξ, in m s⫺1 number of mobile species available per m 2 volume of oxide per metal ion consumed, in m 3 frequency factor ⬃ 10 12 s⫺1

(5.97)

Tarnishing and Scaling Processes

229

W ⫽ sum of the energy of solution of a metal ion in the oxide and the activation energy for the ions to transit from one position to the next site, expressed in joules k ⫽ Boltzmann constant ⫽ 1.38 ⫻ 10⫺23 J ⋅ K⫺1 q ⫽ electronic charge in 1.602 ⫻ 10⫺19 coulombs a ⫽ lattice constant (cation–cation distance ⬃ 4 ⫻ 10⫺10 m in the case of Cu 2 O lattice) V ⫽ potential developed across the film in volts V/ξ ⫽ field strength in V ⋅ m⫺1 A closer look at the above equation (5.97) suggests that as ξ increases the value of exp(ξ 1 /ξ) goes on decreasing, and after a particular value of ξ (i.e., when ξ approaches ξ 1 ), dξ/dt value becomes so small that the rate may be considered to be negligible and the oxidation process almost comes to a stop. This particular thickness has been defined as the ‘‘limiting thickness.’’ However, this equation is valid only in the very thin film range over which electron tunneling is possible requiring almost no activation energy (i.e., thermal activation). It is apparent from the above equation that the number of migrating species at the metal–oxide interface will determine the rate. Beyond the limiting thickness, the field-creating electrons will no longer be available at the outer surface by tunneling mechanism. At such low temperature, the electron availability by thermoionic emission is also not possible. With an increase in temperature to the intermediate range (e.g., for copper, 330–473 K) and the thickness of the film formed being limited to the so-called thin range, oxidation kinetics becomes different from that followed at low temperatures, conforming in many cases, to a normal logarithmic rate law. Theoretical explanations of this process have mainly been provided by Cabrera and Mott [7], William and Hayfield [36], and Uhlig [37]. William and Hayfield’s equation is virtually a simplified version of Uhlig and others for which the same is briefly discussed here. Emission of electrons from the metal was considered as the rate-limiting step in the oxidation of copper at least at temperatures below 393 K. This led them to visualize a mechanism of logarithmic oxidation in which the rate is mainly controlled by the arrival of electrons at the oxide–oxygen interface. Since the oxidation rate is controlled by the electron flow from the metal, space charge effect plays a prominent role in controlling the rate of electron flow and hence the oxidation process itself. The essential step is then the provision of electrons at the oxide–oxygen interface in order that the oxygen atoms can be ionized and incorporated into the oxide lattice according to the following defect equation: O 2 (g) ⫹ 4e′ → 2O xO ⫹ 4V′Cu

(5.98)

230

Chapter 5

Considering electron supply to be due to thermoionic emission from the metal through the conduction band of the oxide, the rate law is expressed in the following form: dξ ⫽ α exp(⫺βξ) dt

(5.99)

where dξ/dt ⫽ rate of oxide growth in m⋅s⫺1. α ⫽ N O Ω exp

冤⫺e(φ ⫹εkT4πan e)冥 s

and β⫽

4πan v e 2 εkT

The terms contained in α and β are as follows: N 0 ⫽ number of Fermi electrons moving toward the barrier per unit area per unit time, Ω ⫽ volume of oxide per metal ion consumed in m3, e ⫽ electronic charge in coulombs, φ ⫽ work function of the metal in volts, a ⫽ cation–cation spacing in Cu 2 O (⬃ 4 ⫻ 10⫺10 m), n s ⫽ number of surface sites per unit area, n v ⫽ number of electron trapping centers per m3, ε ⫽ dielectric constant of the oxide, k ⫽ Boltzmann constant, 1.38 ⫻ 10⫺23 J ⋅ K⫺1. The validity of this kinetic law and mechanism thereof has been confirmed by Roy et al. [38] for Cu-O 2 (g) system in dry air over a temperature range of 348–374 K. It is also not uncommon for some metal–oxidant systems to show the obeyance of a parabolic law at fairly low temperatures or at intermediate temperatures being governed by a different mechanism other than that of Wagner [19]. With an increase in temperature, the availability of electrons at the outer surface of the oxide film becomes plenty and as such it is likely that the availability of electrons will no longer be the rate-limiting step. Considering such situations, Cabrera and Mott theoretically predicted that the ion flux due to an electrical field can be directly dependent on the field strength. This concept led them to derive a parabolic rate law for the formation of relatively thin tarnishing layers of metal-deficient compounds. The rate law expressions for such electrical field– induced migration of defects like vacancies and interstitials are the following:

Tarnishing and Scaling Processes

231

冤

冥

(5.100)

冤

冥

(5.101)

eV 1 dξ ⫽ C v ΩD v (for metal vacancies) dt kT ξ and dξ eV 1 ⫽ C i ΩD i (for interstitial ionic species) dt kT ξ where Cv Ci Dv Ω E Di k ξ

⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽

concentration of cation vacancies, in (number) ⋅ m⫺3 concentration of interstitial ions in (number) ⋅ m⫺3 diffusion coefficient of vacancies in m2 s⫺1 volume of oxide per metal ion consumed in m 3 V/ξ ⫽ field strength in V ⋅ m⫺1 diffusion coefficient of ionic species in m2 s⫺1 Boltzmann constant, in 1.38 ⫻ 10⫺23 J ⋅ K⫺1 instantaneous oxide thickness in meters.

Roy et al. [38] tested and established the validity of this rate law for Cu-O 2(g) system in the temperature range of 348–374 K under the action of an externally impressed direct current during film growth by connecting the substrate metal to the negative terminal of a direct current source. Similar mechanism is also reported to be operative during oxidation of chromium-doped copper. Subsequent study by Bose et al. [27] has further confirmed that even at a temperature of 523 K the contribution to the cuprous oxide growth on copper by Wagner’s mechanism is hardly 3%. Changeover in the mechanism of film growth process from Hayfield’s logarithmic to Cabrera-Mott’s parabolic one has been attributed to the huge supply of electrons at the oxide–oxygen interface, thereby altering the rate-controlling step from electron availability at the outer surface to the electrical field–induced ion migration. Cabrera and Mott [7] also derived a cubic rate equation for a situation when the concentration of charged defects at the outer interface of the film, which is responsible for the electrical field creation, becomes dependent on the thickness of the layer. Assuming electrons to be available at the outer film–gas interface from the metal substrate by thermionic emission, the rate of arrival of these electrons will go on decreasing with increasing film thickness and hence the defect concentration at the outer interface will change accordingly. Taking this into consideration, the following cubic rate equation has been proposed for the growth of a metal-deficient oxide layer on a metal:

冤

冢 冣冥

dξ Q 3᏷D v ΩV 2 ⫽ exp ⫺ dt 4πakT RT

1 ξ2

(5.102)

232

Chapter 5

where ᏷ Dv a V Q

⫽ ⫽ ⫽ ⫽ ⫽

permittivity of the layer, in coulomb V⫺1 m⫺1 diffusion coefficient of vacancies, in m2 s⫺1 cation–cation distance, in meters potential across the layer, in volts summation of activation energy values for the formation of defects and their diffusion through the film, in joules.

Bose and Sircar observed the obeyance of a similar cubic law for Ag–Br 2 [39] and Ag–Cl 2 [40] systems during halide film growth in the thin film range at and around room temperature. They used the above Eq. 5.102 in the following modified form considering positive holes instead of ions or cation vacancies:

冤

冢 冣冥

Q dξ 3᏷D h˙ ΩV 2 ⫽ exp ⫺ dt 4πakT RT

1 ξ2

(5.103)

where D h˙ ⫽ diffusion coefficient of positive holes in m2 s⫺1. This type of cubic law is generally to be observed only at a lower thickness level and at such a temperature range so that the field-creating electrons are available. However, it is difficult to define the exact temperature necessary for electron availability because the energetics of thermionic emission for metal in contact with the film will vary from system to system. In conclusion, the mechanism of cubic rate law is still a controversial issue.

5.6 SCALE GROWTH BY LATTICE AND GRAIN BOUNDARY DIFFUSION In Wagner’s theory of oxidation for thick, compact, nonporous, adherent oxide scale or other reaction product layer (sulfide, halide, etc.) formation, it is assumed that the transport of the reactants through the scale takes place by lattice diffusion or volume diffusion due to the presence of point defects in the growing scale. This theory has been very successful in explaining the reaction mechanism for a number of metal–oxidant systems at high temperatures. As a general rule, lattice diffusion tends to predominate at high temperatures for product scales with a relatively large concentration of point imperfections. But one has to appreciate that solid state diffusional transport also takes place along grain boundaries and through dislocations, especially at low and intermediate temperatures. Thus in the overall treatment of oxidation process, the effects and contributions of nonlattice transport in the growing product scale must also be considered. The principal difference between the diffusional properties of mono- and polycrystalline substances results from the fact that polycrystalline substances contain additional linear and planer defects along which diffusion of different migrating

Tarnishing and Scaling Processes

233

species can take place at a much faster rate than through point defects in the bulk crystal lattice. For this reason, defects of this type are called high-diffusivity paths and the most important of them are the grain boundaries. Studies on polycrystalline materials have shown that diffusion along grain boundaries and other paths of low diffusion resistance may under many circumstances, and particularly at lower temperatures, be an important if not the predominating mode of transport. Furthermore, theoretical considerations indicate that the atomic packing density along grain boundaries is of lower order than that in the grain body. This suggests that the activation energy for diffusion along grain boundaries should be much lower than that necessary for lattice diffusion. Shewmon [41] has suggested that for metals E gb ⫽ 1/2 E 1 , where E gb denotes activation energy for grain boundary diffusion and E 1 is the activation energy for lattice diffusion. It is also an established fact that at higher temperatures scale growth on different metals takes place mainly by lattice diffusion, whereas at low and intermediate temperatures it is dominated by transport of matter and charged species through grain boundaries and other short-circuit paths (dislocations). Important work of Graham et al. [42] has demonstrated that cold-worked nickel oxidizes faster than the annealed one where cold working produced finer oxide grain size during oxidation process. Accordingly, it is concluded that the oxide grain boundaries act as easy diffusion paths for transport of nickel through NiO scale. It is also reported in the literature that the instantaneous parabolic rate constant values decrease with prolongation of exposure. This is a direct consequence of the fact that with prolonged exposure, grain coarsening takes place reducing the grain boundary areas, thus ultimately attaining a value that becomes more or less steady, i.e., when further oxide grain coarsening or scale densification does not occur. At this stage the rate constant corresponds more or less to that of lattice diffusion. Scale growth under such situation tends to conform to Wagner’s mechanism provided the scale is sufficiently thick, compact, and adherent. Various models have been suggested in the literature [4] where the overall transport process constitutes both lattice diffusion and transport along grain boundaries. Under such situations the overall effective diffusion coefficient is expressed as D eff ⫽ D l (1 ⫺ f) ⫹ D gb f

(5.104)

where f denotes the fraction of the total number of diffusion sites at the grain boundaries. For example, during oxidation of titanium at 773 K, it is estimated that the grain boundary areas account for nearly 5–10% of the total oxide area. Furthermore, this model has been utilized to estimate the relative contribution of lattice and grain boundary diffusion during oxidation of annealed and coldworked nickel, where the ratio of k p(gb) /k p(l) has been reported to be of the order 5 ⫻ 10 5 –10 7 at temperatures of 1073–773 K.

234

Chapter 5

The high-temperature corrosion resistance of numerous practical alloys are provided mainly by scales consisting of either Cr 2 O 3 , Al 2 O 3 , SiO 2 , or more complex oxides of these compounds. Since the concentration of point defects in Cr 2 O 3 and Al 2 O 3 are of negligible order compared to those in base metal oxides like NiO, Cu 2 O, FeO, CoO, etc., the growth of protective Cr 2 O 3 and Al 2 O 3 scales are expected to take place by countercurrent grain boundary diffusion of metallic and nonmetallic species whereby the new oxide formation takes place at the grain boundaries of the growing scales. Therefore, at relatively low and intermediate temperatures, for the interpretation of oxidation kinetics, one should consider the contribution of grain boundary diffusion along with that of lattice diffusion. The mathematical models suggested for such estimation have been discussed in detail by Kofstad [4].

5.7 FORMATION OF VOIDS, POROSITIES, AND OTHER MACRODEFECTS IN OXIDE SCALE AND IN THE SUBSTRATE Metals like Cu, Ni, Co, Fe, etc., which oxidize predominantly by outward cation migration through the film/scale, are found to generate vacancies at the compound layer–gas interface. These vacancies diffuse inward through the growing oxide, e.g., Cu 2 O, NiO, CoO, FeO, etc., and accumulate at the metal–oxide interface, thereby nucleating and subsequently coalescing in the forms of voids and porosities near the inner interface and sometimes within the metal itself. It has been demonstrated during oxidation of Fe-19%Cr alloy and many other metal/ alloy-oxide systems [4] that these interfacial voids lead to loss of adhesion and subsequent spallation of the scale. Moreover, supersaturation of cationic vacancies at the metal–oxide interface is accompanied by development of a vacancy concentration gradient within the metallic substrate. In particular, vacancies have been found to migrate into nickel by Hancock and Fletcher [3], into copper by Jaenicke et al. [44] and Appleby and Tylecote [45], and into iron by Cagnet and Moreau [46]. These vacancies have generally been found to precipitate preferentially along grain boundaries of the metals. When the growth of oxide on a metal solely involves outward migration of metal ions (oxygen ions are considered to be essentially immobile in the oxygen sublattice), fresh oxide is formed at the oxide–oxygen interface. If the situation is such that the oxide formed is a completely compact envelope around the metal core, voids and porosities should develop beneath the scales, and the total volume of the void should be equal to the volume of the metal that is being consumed and converted to oxide. A classical example of such circumstances has been demonstrated by Mackenzie and Birchenall [47] who found that an iron wire rod when fully oxidized formed a central cavity of dimensions almost identical to

Tarnishing and Scaling Processes

235

those of the original metal core. Similar observation was also made by Bose and Sircar [48] who found that iodination of a copper wire ultimately transformed it into a pipe of CuI with a central hole throughout the length of the wire. Such observation hints to outward migration of Cu ions and ingress of cation vacancies through the CuI layer, leading to vacancy coalescence and formation of central cavity. However, depending on the exposure conditions (temperature and oxidant pressure in the gas phase), only a fraction of this volume might develop as voids and pores. Kofstad [4] has illustrated the following two major reasons for decohesion of scale from the substrate: 1. Extent of plastic deformation of the oxide scale 2. Vacancy injection and void formation within the metal Due to vacancy injection, voids in the metal may also be formed by hot deformation or creep of the metal substrate induced by the oxidation process. The voids can then be considered to nucleate at the grain boundaries due to grain boundary sliding accompanying the hot deformation process. When the oxide scale grows by outward metal ion migration and is less nonstoichiometric in nature, being too thick to deform or to maintain contact with the receding metal substrate, vacancies condense out and form porosities near the metal–oxide phase boundary. For this reason, void formation has often been termed ‘‘vacancy condensation.’’ However, voids may also form at the inner metal–oxide interface when metal ions diffuse outward through the scale as interstitials. It is established that the tendency of void formation in the retreating metal as well as in the scale is a function of geometry of the specimen and its surfaceto-volume ratio (radius of curvature). For a metal having an infinite plane surface and oxidizing uniformly over the entire surface, the oxide being formed will continuously collapse on the retreating metal surface without any deformation of the scale. If, instead of uniform attack, nonuniform oxidation takes place for a specimen with finite dimensions, constraints are imposed on the system, particularly at the edges and corners which become highly prone to formation of voids and cavities. Study on oxidation of high-purity nickel rods of various diameters in 1 atm O 2 at 1373 K has demonstrated that the density of voids formed increased with increasing oxidation time and with decreasing diameter of the rods. The effect of surface-to-volume ratio has been attributed to the fact that the oxide envelopes (with equal thickness) have less ability to deform, the smaller the diameter of the rod. The presence of carbon in metal has a dramatic effect on the extent of cavity formation and detachment of scale from the metal. Kuiry et al. [49] have shown excessive cavitation, wrinkling, and bulging of the scale in mild steel due to the formation of CO and CO 2 beneath the scale. It is further demonstrated that during oxidation of plain carbon steel with superficially applied coatings containing mixtures of ferromanganese, ferrosilicon, and bentonite or CaSi 2 and bentonite pow-

236

Chapter 5

Figure 5.19 Appearances of the surfaces for 0.16% C steel oxidized nonisothermally up to 1273 K and held at this temperature for 2 h. (a) Coated with bentonite and CaSi 2 ; (b) uncoated.

der from a slurry bath, formation of CO ⫹ CO 2 gas bubbles could be prevented by the achievement of very reduced oxygen activity at the alloy–scale interface. Such coatings also minimized scale spallation even under thermal cycling, as demonstrated in Fig. 5.19.

5.7.1

Consequences of Void Formation

Void formation in the scale and in the metal is expected to affect the rate and mechanism of oxidation as well as the mechanical properties of metal/scale combination. If vacancy injection in the metal predominates and if these are annihilated at sites of vacancy sinks such as dislocations, interfaces, and grain boundaries, the metal surface is likely to recede. On the other hand, when the injected metal vacancies partially condense as voids at the grain boundaries of the metal, scale/metal contact may easily be maintained. However, accumulation of voids in the region of grain boundaries can adversely affect the mechanical integrity of the metal substrate. When pores are formed at the metal–oxide interface or within the scale itself, the path for outward solid state transport of the metal is partially blocked. This may reduce oxidation rate but such effect may be temporary. It has been demonstrated through a large number of investigations on nickel oxidation that after

Tarnishing and Scaling Processes

237

prolonged exposure the nature of the NiO scale becomes double-layered with an outer, fairly compact, columnar crystal layer followed by an inner porous layer of more fine-grained oxide. The following two mechanisms have been proposed to explain such continued oxidation: 1. Dissociative transport across the voids 2. Development of microchannels in the outer layer that allow the oxidant to penetrate into the inner layer. Dissociative Gas Phase Transport Across Voids When a void coalesces at the inner metal–oxide interface, metal ions will continue to move outward from the void surface to the outer oxide–gas interface. Due to depletion of metal ions at the region of dissociation, there will be a corresponding increase in the partial pressure by the liberated oxygen in the void. The oxidant is thus transported in the gas phase across the void and new oxide is formed on the interior surface of the void. Such a model is presented in Fig. 5.20. At the initial stage of void formation, partial pressure of the oxidant is equal to the equilibrium dissociation pressure at the metal–oxide interface at the temperature under consideration. For example, the partial pressure of oxygen at the Ni–NiO interface at 1273 K is approximately 1 ⫻ 10⫺10 atm. Therefore, instantaneous pressure of oxidant is very small. But with continued outward nickel transport from the void surface to the NiO/O 2 (g) interface, the pressure

Figure 5.20 Model for the growth of scales by dissociative transport across voids in the scale. Metal ions migrate outward through the scale by solid state diffusion, while oxygen is transported inward across the voids as gaseous species [Ref. 4].

238

Chapter 5

of oxygen in the void gradually builds up and eventually becomes adequately high to sustain reaction for new oxide formation at the interior surface. With continued oxide dissociation and new oxide formation, the void is expected to move gradually outward in the scale. This will ultimately lead to opening of the void at the oxide–gas interface. Development of Microchannels A model for the development of microchannels in the scale has been proposed by Mrowec [50]. In this scheme an initial void is assumed to be formed at the inner interface just below an oxide grain boundary as shown in Fig. 5.21. The

Figure 5.21 Model for the development of microchannels above cavities at the metal– oxide interface or in the oxide scale. Solid state diffusion in the oxide takes place by outward metal diffusion in the lattice and along grain boundaries. When grain boundary diffusion is much faster than lattice diffusion, the grain boundary opens into a microchannel. This, in turn, permits inward transport of molecular oxygen [Ref. 4].

Tarnishing and Scaling Processes

239

oxide grain boundary is further assumed to extend from the location of a void to the outer oxide–gas interface. When the outward transport of metal ions is faster along the grain boundary than through the oxide lattice (this is always true except at very high temperatures), the grain boundary area has a strong tendency to ‘‘open up’’ next to the void. Continued transport of metal ions in such a fashion ultimately leads to an opening up of the grain boundary as a channel that penetrates the scale from the gas interface to the void. The overall oxidation process under such conditions will involve the inward transport of oxygen molecules along the microchannels associated with the outward transport of metal ions in the scale by solid state diffusion. However, for several reasons, this apparently simple scheme may not explain many other observations of scale growth with development of porosity and microchannels. A major doubt arises regarding the maintenance of ‘‘all-through’’ open channels. When oxygen molecules can easily diffuse inward through the channel, the oxygen potential in the channel is expected to be quite high, leading to faster oxidation of metal at the surface of the channel. It is expected that eventually the channel will close up due to lateral growth of new oxide grains. Some experimental observations strongly suggest that gaseous species infiltrate even through relatively thick-growing scale in which lattice diffusion predominates, e.g., in wu¨stite (highly defective lattice). Furthermore, the above model implicitly assumes that the scale must grow by outward diffusion of metal ions. Therefore, this model will be inappropriate for the cases where scale growth takes place predominantly by ingress of oxygen. An alternative explanation has been advanced where it is proposed that development of porosity and channel is closely linked to plastic deformation and creep of the scale as a consequence of induced growth stresses. It is believed that cavitation in the underlying metal substrate can occur by the condensation of injected vacancies, stresses accompanying the growth of oxide, etc., which cause plastic deformation and subsequent creep cavitation of the substrate. In cases of mild steel oxidation, internal oxidation of carbon at metal grain boundaries generates gas pressures (CO ⫹ CO 2 ) high enough to form cavities. Carbon in the substrate is likely to facilitate the nucleation of deformation cavities.

5.8 DEVELOPMENT OF STRESSES AND STRAINS IN THE GROWING SCALES It is an accepted view that when metallic components are exposed to high temperatures in aggresive environments, their useful life will be governed by the presence of a protective surface film that acts as a barrier to the reactants, i.e., metal and oxidant. Therefore, the mechanical maintenance of the protective film is of

240

Chapter 5

utmost importance in prolonging the service life of many engineering components because the failure or rupture of the surface film creates easy access of the oxidant to the fresh metal surface, facilitating further degradation. The net result is a significant loss of metal each time the overlying film/scale failure occurs. Therefore, properties of the growing film/scale must be modified in such a fashion so that it can withstand both internal and superimposed stresses (e.g., thermal and growth stresses) yet remain protective to the underlying metallic substrates. The most exhaustive and systematic recent review of stress effects on hightemperature oxidation behavior has been put forward by Evans [51] who discussed different theoretical models of stress development and its measurement. The main factors that have been identified to influence the development of stresses in metal/scale combinations during isothermal oxidation are as follows: 1. 2. 3. 4. 5.

Molar volume ratio of the oxide to the metal transforming into oxide (PBR), Epitaxial relationships between the structures of metal and its oxide, Compositional changes that occur in both the metal and the surface oxide, Vacancies generated during oxidation, and Component geometry.

5.8.1

Stresses and Strains Due to Volume Changes

The concept of the Pilling-Bedworth ratio (PBR) [5] concerning protective or nonprotective scale growth on metals has already been discussed in an earlier section of this chapter. According to this, if the PBR (φ) is greater than unity, one can expect protective, compact, and adherent scale formation on a metal surface, whereas if φ is less than unity, porous nonprotective scale growth will take place. It is important to note that these authors had confined their studies entirely to cylindrical geometries because their aim was to determine the qualitative oxidation rates of wire specimens of various diameter, and the effect outlined a geometrical one that would strongly depend on the diameter of the wire in question. It is readily realizable that the geometrical constraint indeed approaches zero as the radius of curvature of the metal surface approaches infinity as that for planar metal specimens oxidizing uniformly. Even in the case of planar metal surface, if nonuniform oxidation takes place, it could give rise to some effects from the variation of mechanisms because the geometry would then deviate from the planar metal–oxide interface. Such situations are illustrated schematically in Fig. 5.22 [52]. Though the consideration of PBR nearly approaching unity acted as a handy guideline for prediction of protective scale formation, it was felt necessary to modify the concept in order to explain nonprotective oxide formation for those metals having PBRs greatly exceeding unity. Kubaschewski and Hopkins [2] cite evidence where metals having PBRs either much less or much more than unity

Tarnishing and Scaling Processes

241

Figure 5.22 Role of the Pilling-Bedworth ratio in producing compressive or tensile stresses during nonuniform oxidation of a planar metal sample [Ref. 52].

exhibit a strong tendency toward linear nonprotective oxidation. The linear oxidation of metals having particularly high PBRs (e.g., the ratio is 2.68 for Nb 2 O 5 and 2.5 for Ta 2 O 5 ) follows an initial stage during which the scale remains fairly coherent. The linear rate for such systems is always associated with the formation of cracked or porous scales, probably due to development of stresses in the scale as well as in the metal substrate. Cracking of scales is a common observation in the cases where diffusion is primarily anionic. An ‘‘extension’’ to the PBR concept has subsequently emerged for oxides having considerably large volume ratios and for which oxide growth occurs primarily at the metal–oxide interface. When oxygen ions diffuse from the surface to the metal–oxide interface where they form new oxide which, owing to its high volume ratio, expands against the resistance of the existing oxide layer, severe biaxial stresses develop that eventually lead to rupturing of the film [52]. The consideration of volume ratio turns out to be of minor importance when

242

Chapter 5

the oxidation proceeds through outward migration of metal ions. In such cases, fresh oxide is formed at the outer free surface in an unconstrained manner and thus initial protective oxidation is expected under such conditions. It has been pointed out by Borie et al. [53] that point imperfections in a growing oxide lattice may usually lead to a change in the lattice constants. It is to be expected on the basis of elementary considerations of volume filling that the average expansion would be positive for defects of the interstitial type but negative (i.e., contraction) for accommodating lattice vacancies as illustrated in Fig. 5.23. In the absence of other effects, the expansion or contraction would be isotropic. However, this is a somewhat simplified view where each unit cell undergoes expansion or contraction in an unconstrained manner. As additional oxide forms, the defect density must gradually adjust throughout the already formed oxide to a new steady-state value, and this adjustment will take place under severe conditions of constraints. For example, a positive expansion in the directions parallel to the parent metal interface may be prohibited because the oxide already exists in a more or less uniform layer in these directions. This inhibition may cause an expansion to occur primarily in a direction perpendicular to the parent– metal interface. One approach that might prove fruitful is to compute the equivalent local stress F(ξ) that would be required to produce the average volume

Figure 5.23 Changes in the average unit cell size due to point defects. (The parameter a 0 denotes the unstrained lattice parameter in one direction.) [Ref. 52].

Tarnishing and Scaling Processes

243

change δV(ξ) associated with the point defect density C(ξ), and then solve the relevant equations of elasticity subject to the constraints in question. Therefore, it is natural that an anisotropic strain can be produced by the locally isotropic stress distribution. Other related problems arise when alloys of differing oxidation resistance are forced into close contact by means of external constraint. A particularly subtle example was the problem of bolt failures experienced in UK gas-cooled nuclear reactors [51] where rimming steel washers (low silicon) were used in conjunction with fully killed steel (high silicon) bolts. The low-silicon steel oxidized considerably faster than the bolt material in CO 2 environment at and around 673 K and the differential increase in volume produced was accompanied by straining and fracture of the oxidation-resistant bolts. Early evidence for the deformation produced by oxidation has been provided by Noden et al. [54] on austenitic stainless steel oxidized at 1173 K in CO 2 and/ or air. Some of their results clearly exhibit that creep strain of about 2% were developed in the thinnest samples (0.1 mm) in relatively short time. Even for the thickest samples (0.38 mm), creep strain of about 0.5% was recorded. Subsequently, similar effects have been observed in other metals and alloys. It has been concluded that the ‘‘origins’’ of such growth stresses are volume changes induced by the oxidation of process, occurring in a confined space. Attention was subsequently focused on the oxide–metal interface and the role it may have on stress development when the outward cation transport occurs by vacancy interchange. In such cases, there is a flux of vacancies counter to the direction of cation flux. It becomes important to recognize that for the metal to be consumed, such vacant sites, within the oxide must be annihilated at suitable locations (metal–oxide interface) generating stresses.

5.8.2

Epitaxial Relationships Between Structures of Metal and Its Corresponding Oxide

Atoms in the planar surface of a parent metal crystal are arranged in a periodic fashion. The periodicity is determined by the crystal structure of the metal and the relative crystal orientation of the surface. The surface atoms exert short-range molecular forces on any newly deposited or chemically formed atomic and molecular layer at the surface. A newly formed molecular layer of oxide is thus subjected to influences other than those associated with bulk thermodynamic properties of the oxide. Accordingly, the crystal structure and lattice constant in the monolayer may be quite different from those found in bulk quantities of the oxide. Such influence of the substrate in determining the structure, orientation, and lattice constant of deposited or chemically formed layers is referred to as epitaxy. When formation of the first oxide monolayer is complete, the periodic arrangement of atoms and molecules in this first layer exerts similar short-range

244

Chapter 5

molecular forces on new oxide molecules, which forms the second molecular layer. The second molecular layer thus formed, while subjected to influences other than those associated with bulk thermodynamic properties of the oxide in question, the crystal structure and the lattice constant in the second molecular layer, may again be quite different from those found in bulk quantities of the oxide. Each successive molecular layer can thus be viewed as a new substrate for formation of yet another molecular layer. So the initial influence of the metal surface can be propagated over distances within the newly formed oxide layers, which are large relative to those expected from the usual range of molecular forces. At each stage of new oxide layer formation, however, the driving force due to bulk thermodynamic properties always exists in competition with the influence of the metal substrate. It is thus intuitive that as the oxide film becomes thicker and thicker, the influence of the original metal substrate becomes weaker and weaker, so that eventually the crystal structure and lattice constant of the oxide must approach those characteristic of the bulk oxide. A simple mathematical model has been advanced by Fromhold [52] to explain the epitaxial influences of metal–oxide combination. In this model it has been assumed that there exists a possibility of some relaxation of constrained lattice toward the bulk lattice constant with each new epitaxial layer being formed. It is further assumed that the amount of this relaxation with formation of each new layer is proportional to the existing deviation from the bulk lattice constant, then one can write the following proportionality relation: d [a(ξ) ⫺ a o ] ⬀ [a(ξ) ⫺ a o ] dξ

(5.105)

where a o ⫽ the bulk lattice constant, and a(ξ) ⫽ the oxide lattice constant at distance ξ within the oxide measured in a direction perpendicular to the metal substrate. Further more, on denoting the proportionality constant determining the rate of such relaxation by ⫺1/δ, Fromhold finally arrived at the following expression:

冢 冣

ξ a(ξ) ⫺ a o ⫽ [a(o) ⫺ a o ] exp ⫺ δ

(5.106)

where a(o) ⫽ lattice constant of the oxide at the metal–oxide interface. Equation 5.106 can be further simplified to

冤

冢 冣冥

a(ξ) ⫽ a o 1 ⫹ γ exp ⫺ where γ ⫽ [a(o) ⫺ a o ]/a o .

ξ δ

(5.107)

Tarnishing and Scaling Processes

245

The minus sign of the proportionality constant (⫺1/δ) implies that relaxation occurs toward (and not away from) the bulk lattice constant with increasing ξ, and δ is a positive quantity having the dimensions of length that determines how slowly or quickly the relaxation takes place with increasing distance from the metal substrate. It is thus expected that δ will be a characteristic of the oxide medium and in particular will depend on the elastic constants of the medium. The predictions of Eq. 5.107 have been illustrated by Fromhold [52] in a qualitative way as shown in Fig. 5.24. In these schematic diagrams, sketch (a) indicates the difference in the metal substrate and oxide periodicities that might be ordinarily expected if each substrate had its normal bulk lattice constant. Sketch (b) illustrates a situation in which the metal substrate influences the first monolayer to take on a ‘‘reduced’’ value for the lattice constant in order to make the periodicities the same for the two substances, in which case γ ⬍ 0. Sketch (c) illustrates the alternative situation in which the metal substrate influences the first oxide monolayer to take an ‘‘expanded’’ value for the lattice constant in order to equalize the periodicities of the two substances, in which case γ ⬎ 0. Thus the stresses developed in scales may be related to epitaxial relationships and oxide–metal mismatch. Such stresses are the largest at the metal–oxide interface. Studies of epitaxial relationship have, for example, been reported for oxide film formation on copper [55]. It has been shown by means of x-ray diffraction studies that there existed a strain of 2–2.5% at the copper–oxide interface while there was no strain at the oxide–oxygen interface. Epitaxial relationships are a function of the orientation of metal grains. In polycrystalline materials stresses may also develop at grain boundaries owing to the difference in oxidation rates between neighbouring grains and a preferential diffusion and oxide formation along grain boundaries.

5.8.3

Compositional Changes in the Metal and Surface Oxide

Compositional changes in the metal as well as in oxide scale are likely to alter the mechanical properties of both phases. Differential doping of the growing scale by second or third alloying element or by the presence of impurity elements in the scale has been experimentally observed during oxidation of dilute alloys for which a theoretical model has also been proposed. Thus nonstoichiometry will vary from the inner to outer interface of the oxide layer. Accordingly, it is expected that different segments of the total oxide will have different creep properties or stress relaxations. The dopant may enrich either in the outer layer or at the metal–oxide interface, as, for example, segregation of sulfur at the metal– oxide interface often leads to decohesion of scale. Similarly, difference in interdiffusivity values of elements in the metallic phase or internal precipitation of oxide, carbide, nitride, etc., will also greatly influence the mechanical properties

246

Chapter 5

Figure 5.24 Epitaxially produced changes in the oxide lattice parameter adjacent to the parent metal interface due to molecular forces exerted by the parent metal substrate [Ref. 52].

Tarnishing and Scaling Processes

247

of the substrate. Besides, precipitation within metallic matrix will create local differential stress distribution and strain rate. With increased point imperfections, the scale is expected to have a better tendency to relaxation through elastic and plastic deformation. Moreover, at elevated temperatures, individual oxide grains may undergo some sort of sintering process facilitating a better barrier to mass transport rendering protectiveness to the underlying substrate. However, it is to be realized that stress and relaxation would be decided by a complex interplay of mechanical and chemical properties of any metal–scale combination.

5.8.4

Effect of Vacancies Generated During Oxidation

For p-type oxide growth like NiO on Ni, Cu 2 O on Cu, CoO on Co, etc., due to oxygen pickup at the scale–gas interface, cation vacancies are generated that migrate through the oxide layer toward the metal–oxide interface, where they are annihilated or consumed at suitable sinks like dislocation defects, grain boundaries, etc. These metal vacancies may accumulate at the metal–oxide interface followed sometimes by their coalescence, producing voids that result in scale–metal decohesion at some locations. Alternatively, they may diffuse through grain boundaries into the substrate leading to formation of discontinuous or continuous pores, or be annihilated at dislocation sites. Sometimes voids may also form within the scale. Compactness of the scale and its integrity to substrate depend on mechanical properties of both the growing scale and the substrate metal. In such cases, stresses are of course generated but of lower magnitude because new oxide formation takes place at the free oxide–gas interface. However, in cases of n-type product layer formation such as ZrO 2 , Ta 2 O 5 , Nb 2 O 5 , TiO 2 , etc., on the corresponding metal substrates, oxygen vacancies are created at the metal–oxide interface and are annihilated at the outer surface of the scale. In such cases, oxidant migrate inward and new oxide formation mainly takes place at the metal–oxide interface, generating more stresses and strains in the layer. Evans [51] has pointed out that at the oxide–metal and oxide–gas interfaces, the free energy of formation of point defects (vacancies) will be changed by an amount σ H ∆Ω, where σ H is the hydrostatic component of the local stress and ∆Ω is the local change in volume. This factor will change the vacancy concentrations at the boundaries of the oxide layer. In addition, the presence of a stress gradient across the oxide layer will impose a bias to the random migration of defects. A treatment of this problem has been advanced [51] for the growth of a protective zirconia film on zircalloy-2 (Zr-1.5Sn) where the stress at the oxide–gas interface is considered to be zero, so that the anion vacancy concentration (C II ) there remained in equilibrium with the environment. Furthermore, a biaxial compressive stress is assumed to exist at the oxide–metal phase boundary. Since the

248

Chapter 5

formation of new oxide at this interface involves a volume increase, work of the order σ H ∆Ω is needed to be supplied by the chemical reaction to insert fresh oxide at this location. The resultant effect is to make it less easy for the reaction to proceed. The stress dependence of vacancy concentration at this interface is then given by: (C I ) σ ⫽ C I exp

冢

σ H ∆Ω kT

冣

(5.108)

where C I ⫽ anion vacancy concentration at the metal–oxide interface. Considering the flux of vacancies through the stressed oxide, the following stress-dependent oxidation rate expression can be obtained:

冢

dξ D diss σ ∆Ω ⫽ O exp H dt ξ kT

冣

(5.109)

where D diss O is the diffusion coefficient of oxygen within ZrO 2 equilibrated at its dissociation oxygen pressure. This expression reduces to the conventional parabolic form at zero stress but predicts a lower rate of reaction when σ H ∆Ω is negative and a higher rate when this stress term is positive. Figure 5.25 depicts the main volume changes that need to be considered for the growth of an anion-deficient oxide. Interchange of an oxygen gas atom with an oxide vacancy produces a volume change as given by: ∆Ω B ⫽ Ω O ⫺ Ω V

(5.110)

where Ω O and Ω V are, respectively, the ionic and vacancy volumes in the oxide. It is this change that also needs to be considered when discussing the influence of stress on the bias term for diffusion through the oxide layer. A second important volume change arises at the oxide–metal interface when oxygen and metal ions react to form a molecule (MO) of fresh oxide. ∆Ω A ⫽ Ω MO ⫺ Ω M

(5.111)

where Ω MO ⫽ volume of oxide Ω M ⫽ volume of metal ion However, since a vacancy is also expected to be involved with the reaction at this interface, the total volume change will be given by: ∆Ω ⫽ ∆Ω A ⫺ ∆Ω B

(5.112)

The values of ∆Ω B are not well established, whereas those of ∆Ω A are much better known through the concept of Pilling-Bedworth ratio (φ ⫽ Ω MO /Ω M ). Utilizing this, one can obtain: ∆Ω A ⫽ Ω M (φ ⫺ 1)

(5.113)

Tarnishing and Scaling Processes

249

Figure 5.25 Volume changes associated with oxide growth by anion diffusion [Ref. 51].

For the particular case of Zr/ZrO 2 considered above, φ ⫽ 1.56 so that ∆Ω A ⫽ 0.56Ω M . In such calculations for oxidation of zircalloy-2, Evans [51] estimated ∆Ω to be positive but about two-thirds the value of φ. The rate of reaction was thus expected to be less in the stressed oxide, i.e., σ H ∆Ω was negative in Eq. 5.109. The negative stress in the oxide is known to increase with oxide thickness which means that retardation in the rate will also increase with thickness. As a consequence, the kinetics of reaction will be subparabolic and thus in accord, at least qualitatively, with common observations on this alloy. It has been further pointed out that there exists the possibility for the stress, either originating from the growth process or deriving from external constraints,

250

Chapter 5

to prevent or reduce the oxidation process. The most straightforward approach is first to establish the level of stress required to prevent the chemical process of oxidation. This is simply obtained by equating the energy obtained from stress through the volume change ∆Ω A , with the free energy released ∆Go, on oxidation: σ ∆Ω A ⱖ ∆Go

(5.114)

to suppress further reaction. Estimates of the stresses needed to satisfy the above inequality often exceed typical material flow or fracture stresses by over two orders of magnitude. This suggests that under most experimental or engineering conditions, the direct effect of stress on the chemical process of oxidation is negligible. In actuality, more subtle effects may become important through their action on the rate of diffusional supply of reacting material. The flux of vacancies (J v ) through stressed oxide is given by: JV ⬀

冤

冢

冣

冥

DV σ ∆Ω C I exp H ⫺ C II ξ kT

(5.115)

where D v signifies the diffusion coefficient of vacancies. Considering the same oxygen-deficient oxide in contact with metal (e.g., ZrO 2 /Zr), from Eq. 5.115, it is clearly revealed that the flux of anion vacancies, and hence the flux of oxygen ions, becomes zero when the local stress at the metal–oxide interface is obtained through fulfillment of the following condition: C I exp

冢

冣

σ H ∆Ω ⫺ C II ⫽ 0 kT

(5.116)

Therefore, σH ⫽

kT C II ln ∆Ω C I

(5.117)

where σ H is a compressive stress. Equation 5.117 describes the minimum condition for suppressing oxidation, at least for the particular example considered. If this equation is rewritten in terms of PBR and the volume change (∆Ω) identified as simply that due to this ratio, utilizing Eq. 5.113 one obtains: σH ⫽

冢 冣

kT C ln II Ω M (φ ⫺ 1) CI

(5.118)

which is the minimum condition to suppress oxidation. For ZrO 2 film where the principal defects seem to be anion vacancies, its

Tarnishing and Scaling Processes

251

highest concentration exists in the oxide in equilibrium with the metal and is insensitive to the partial pressure of oxygen in the external environment. Then either of the concentration terms, i.e., C I or C II, may be given by: C II ⬀ P ⫺1/6 ext

C I ⬀ P ⫺1/6 * where P ext ⫽ pressure of the oxidant at the scale–gas interface, P* ⫽ dissociation pressure of the oxide in contact with Zr. Therefore, by expressing vacancy concentration in terms of local oxygen partial pressure, Eq. 5.118 can be transformed into: σH ⫽

and

P kT ln * 6Ω M (φ ⫺ 1) P ext

(5.119)

This equation is plotted in the form of compressive stress vs. log oxygen pressure and depicted in Fig. 5.26 for the case of zirconia (the line marked ‘‘100’’). The calculations have been performed using values of φ ⫽ 1.56, Ω M ⫽ 10⫺29 m 3, and a ZrO 2 dissociation pressure of P* ⫽ 8 ⫻ 10⫺68 atm at 773 K. This figure also

Figure 5.26 Stress-induced reductions in oxidation rate for an oxide (zirconia) growing by anion diffusion at 773K. Numbers on curves give percentage drop in oxidation rate under stress [Ref. 51].

252

Chapter 5

shows an estimate of the fracture stress (σ F ) for zirconia of 2.1 GN m⫺2 as deduced from an observed fracture strain of about 0.01 [51] assuming elastic behavior. It is evident that the ‘‘kinetic limit’’ as represented by Eq. 5.119 exceeds this fracture stress for most values of oxygen partial pressure. However, as the dissociation pressure is approached, the critical stress to achieve the kinetic limit falls below the oxide fracture stress. Thus, in this regime it should be possible to suppress the oxidation reaction completely without producing fracture of the oxide layer. Lower stresses are required to effect a change in the oxidation kinetics through their effect on the driving force for diffusion through the zirconia layer. Estimates of the stresses required to reduce the oxidation rate by 10% and 50% are also depicted in the above figure. It may be noted that the rate of supply of anions can be enhanced in the presence of a tensile stress at the oxide–metal interface. This of course applies for an increase in volume on formation of the oxide. In principle, similar considerations apply to cation-deficient oxides where fresh oxide is formed at the oxide–gas interface. However, such a situation is complicated by the potential for layer growth in the direction normal to the unstressed free surface. It is then difficult to define the work done by any in-surface stress system if relaxation occurs partially. Accommodation of epitaxial strains can also arise by the presence of amorphous phase at the oxide–metal interface. A classical example is the formation of a thin (ⱕ0.1 µm) interlayer of amorphous silica between the outer chromia layer and metal substrate of austenitic steels [51]. Even though no measurements of strain gradients have been made in its vicinity, a reduction in these may contribute to the superior spallation resistance of the chromia layer when the silica phase is present as an interlayer. Interfacial regions of amorphous material have also been reported in zirconia layers grown on various zirconium alloys. A related effect is the formation of denser allotropic forms that are favored by the high local compressive stresses, e.g., tetragonal rather than monoclinic zirconia. Borie et al. [53] have also observed the distortion of Cu 2 O films from a cubic to an orthorhombic structure in the vicinity of copper substrate.

5.8.5

Effect of Component Geometry on Stresses Developed in Oxides

Specimen geometry exhibits significant influence on the stress patterns in the scale and, hence, the overall oxidation rate. It has been emphasized that when thick scales are formed there is a continuous change in the dimensions of test specimens. In interpretations of oxidation kinetics and mechanisms this effect has to be taken into account. It is agreed on that large flat disks are the most desirable (no corners present) shape of specimens for kinetic studies, whereas cylindrical specimens are the least suitable. It has already been discussed that outward transport of metal ions through the

Tarnishing and Scaling Processes

253

growing scale is a function of the shape of the specimens and their surface-tovolume ratio. For a metal with an infinite plane surface oxidizing uniformly over the surface, the overlying scale in principle collapses on the retreating metal core without any deformation of the scale. But if nonuniform oxidative attack occurs (more common), stress amd straims are produced in the scales. For specimens with finite dimensions, constraints are imposed on the system, and this produces voids and cavities. For metal rods or cylindrical specimens, large deformations are necessary if the oxide is to remain in contact with the metal and large voids are likely to form. A classical example of this is given in Fig. 5.27, which exhibits the cross-section of an oxdized iron wire [56] where decohesion and void formation has appreciably affected the oxidation rate. Stresses and strains, void formation, scale detachment, and deformation and fracturing of scales will be dependent on whether the oxidizing surface is concave or convex. In these cases also there will be major differences with respect to predominance of inward oxygen transport or outward metal transport. The various limiting cases have been well addressed in detail by Hancock and Hurst [55], as well as by Evans [51].

5.8.6

Compressive Failure of Scales and Spallation

Oxide scale spallation frequently occurs, particularly under conditions of thermal cycling and poses a severe threat both to the maintenance of protective layer and, ultimately, to the endurance of the component. It is only in the past decade or so that significant progress has been made in identifying crack models for the spallation process. In early studies, spallation was frequently reported during cooling cycles because it is then that the oxide is usually subjected to compression (for flat specimens), as a result of the differences in thermal expansion coefficient between oxide and metal. For spallation to take place, it is necessary to generate cracks through the oxide layer to the interface at which spallation will take place and also to produce decohesion along that interface. This interface may be one between layers of oxide of different composition, or, indeed, a plane of weakness within a single oxide layer. Here in order to establish the principles, it is simply assumed that the interface is the one existing between the oxide and metal. Two distinct processes of spallation of a compressively stressed oxide can be identified. If the oxide–metal interface has a high adhesive strength relative to the cohesive strength of the oxide, cracking of the oxide will be the consequence before decohesion. On the other hand, poor interfacial adhesion will lead to decohesion before the occurrence of cracking of the oxide. These two distinct situations of stronger and weaker interfacial adhesion provide ideas of two entirely different routes leading to spallation of oxide as schematically shown in Fig. 5.28. When adhesion between the metallic substrate and the oxide is quite strong

254

Chapter 5

Figure 5.27 (a) Loss of oxide adherence after oxidation of an iron wire in oxygen for 266 min at 750°C (96⫻); (b) Effects of loss of oxide adherence of oxide scale on the kinetics of oxidation on iron wire (d⫽0.48 mm) at 715 °C [Ref. 4].

Tarnishing and Scaling Processes

255

Figure 5.28 Cracking and spallation caused by compressive oxide stresses. Route I (a) shear cracks develop; (b) wedge crack grows between shear cracks; (c) local athermal stress relaxation; (d) thermal stress relaxation. Route II: (e, f) localized decohesion can lead to buckling; (f) buckles may spread laterally and coalesce; (g, h) tensile cracks develop in regions of tensile stress and lead to spallation [Ref. 51].

(route I), initial failure of the oxide occurs by compressive shear cracking. Subsequent cooling of the sample results in differential contraction strains that drive wedges of the adjacent oxide layer under the segment bounded by the shear cracks and thus produce gradual decohesion at the interface. Alternatively, if interfacial adhesion is poor, e.g., due to coalescence of voids or segregation of elements such as sulfur, then compressive straining will initiate wrinkling and buckling of the oxide (route II) over these areas of weaker adhesion. It may also happen for individual wrinkles or buckles to spread laterally by propagation of crack along the interface. Spallation results when through-thickness cracks form preferentially in regions of high tensile stress in the oxide. The above-mentioned two different routes to spallation are denoted as ‘‘wedging’’ and ‘‘buckling,’’ respectively [51]. The two modes of failure have been well demonstrated in the oxidation of 20Cr-25Ni austenitic steel where chromia is the protective oxide layer. The buckling configuration arises on the steel containing no silicon and the wedging configuration (characterized by an inclined oxide failure plane) on a steel containing 0.6% Si. Presence of this small amount of silicon in the second steel

256

Chapter 5

results in formation of a thin amorphous layer of silica between the chromia and the steel. It has been speculated that this amorphous interlayer might have improved adherence by its ability to accommodate epitaxial misfit strains. However, for a condition of good oxide–metal bonding, it is still possible to demonstrate prevalence of both wedging and buckling modes while characterizing spallation at different oxide thicknesses even within the same oxide–alloy system. Researchers normally conduct oxidation experiments for a number of thermal cycles in order to have a better assessment on the performance of high-temperature alloys. It will be of interest to determine the value of ∆Tc (critical temperature drop to initiate spallation) for spallation in the first cycle when the protective surface oxide remains undamaged and its properties can be well characterized. A number of methods such as use of thermobalances, acoustic emission, or a combination of both have been developed to determine the ∆Tc values. However, the most direct approach is the use of thermobalances where spallation is readily detected as a loss of mass. Use of this simple technique on the oxidation behavior of 20Cr-25Ni-Nb-stabilized austentic stainless steel has been extensively employed and reviewed. The detailed attention paid to this material derives from its use as thin-walled (0.38-mm) cladding for UO 2 fuel in advanced gas cooled (nuclear) reactors. On oxidation at temperatures around 1123 K the alloy forms a protective oxide layer of chromia with an underlying thin interlayer of silica (amorphous). The variation of the critical temperature drop, ∆Tc , to initiate spallation with the estimated thickness of the chromia layer is shown in Fig. 5.29. It depicts a distinct trend of decreasing ∆Tc values with increasing oxide thickness. This is the dependence expected from the wedging process, whether this is described in the phenomenological manner of a critical strain energy or by the mechanistic finite element calculations. It was already stated that the initiation of oxide spallation in a given thermal cycle may occur either by a wedging or buckling mode as illustrated in Fig. 5.28. Both of these mechanisms have been theoretically examined and expressions derived [51] that adequately predict the critical temperature drop, ∆Tc to effect spallation. These expressions differ in their parametric dependence on oxide thickness such that the preferred mode of mechanical failure for any given oxide– metal system will change as the oxide layer thickens. These different relations can be demonstrated graphically by using ‘‘spallation maps.’’ An example of such maps is presented in Fig. 5.30 for chromia scale formed on 20Cr-25Ni austenitic stainless steel and being cooled from 1173 K. The line curving downward and the line curving upward with increasing oxide thickness correspond to the wedging and the buckling criteria, respectively. Basically, this diagram (Fig. 5.30) consists of four regions within which four distinctly different modes of mechanical response to the cooling cycle occur. To the left-hand side of the map there is a region in which buckling failure dominates

Tarnishing and Scaling Processes

257

Figure 5.29 Variation of critical temperature drop to initiate chromia spallation from austenitic steel as function of gross weight gain or oxide thickness [Ref. 51].

and spallation occurs by the formation of cracks in tensile regions around the buckle. This buckle configuration is stable, i.e., it will not extend laterally by tensile crack growth along the oxide–metal interface. Such lateral growth will occur in a small region as shown at the top of the map bounded by the buckling and wedging lines after their interaction. The dominant failure mode for the example cited is the wedging as indicated by a large area to the right-hand side of the map. Of much interest is the region to the bottom of the map within which the oxide layer suffers no significant mechanical damage. Alloys having high spallation resistance will have this area in an enlarged form. A significant feature of Fig. 5.30 is the rapid increase in the value of ∆Tc for buckling with only modest increases in oxide thickness, e.g., over the range 0.5– 1.0 µm. A consequence is that buckling failure is likely to be important only for thin layers or for those that already exhibit large-scale decohesion. This appears unlikely in alloys that have been designed for high-temperature oxidation resistance where extensive interfacial void formation is rare. A recent study for the interfacial voids during high-temperature oxidation of alumina forming FeCr alloy (Fe-22Cr-5A1) and Haynes 214 (Ni-16Cr-4.5A1) alloys exhibited nothing

258

Chapter 5

Figure 5.30 Spallation map for 20Cr-25Ni steel cooled from 1173 K [Ref. 51].

larger than the detection limit of 0.1 µm. Such alloys have generally been found to form a protective oxide thickness of around 10 µm under typical service conditions, leading to the inevitable conclusion that spallation by buckling is unlikely to be relevant in industrial usage. The wedging line as shown in Fig. 5.30 is calculated using an effective fracture energy (γ F ) of 6 J m⫺2. This value is in reasonable agreement with spallation data for a chromia forming austenitic steel but it also includes a contribution from creep deformation of the alloy. Identification of the appropriate value of γ F is rather difficult because many factors, notably creep deformation, are involved. Indeed the energy of adhesion at the oxide–metal interface and the surface energies of the oxide and metal phases may only be minor components of the actual total energy required to produce decohesion and subsequent spallation. The benefits of a weak creeping substrate and a small defect size at the oxide– metal interface can best be illustrated by a modified spallation map as shown in

Tarnishing and Scaling Processes

259

Figure 5.31 Notional spallation map for alumina on FeCr alloy steel cooled from 1373 K. Note expansion of region of no damage [Ref. 51].

Fig. 5.31. This is thought to be representative of an alumina layer on FeCr alloy steel being cooled from an oxidation temperature of 1373 K. The movement of the wedging line to the right (a weak substrate) and the buckling line to the left (a small defect size) has widened the central area of the map so that no serious spallation damage will be expected for most experimental conditions. It seems likely that this combined influence of low creep strength of the alloy and absence of large interfacial defects is the principal reason for the excellent spallation resistance of this type of alloy. In conclusion, it can be stated that there exists a direct influence of oxide growth stresses on defect diffusion through the layer and hence on the hightemperature oxidation behavior of metals. In particular, the stress factor becomes significant when constrained volume changes accompany growth of the layer.

5.9 DISSOLUTION AND DIFFUSION OF OXIDANT IN METALS All metals exhibit a tendency to absorb oxygen and other oxidants (like nitrogen, sulfur, carbon, etc.) to a lesser or greater extent, especially at elevated tempera-

260

Chapter 5

tures. The gases may be present in metals as interstitially dissolved atoms; as molecular gas in voids, microchannels, and microcracks; or as separate phases such as oxides, nitrides, sulfides, etc. Dissolution and diffusion of oxidant in metals may become an important factor for assessing the high-temperature oxidation behavior of metals, particularly for metals belonging to groups IVA (Ti, Zr, Hf) and VA (V, Nb, Ta) of the periodic table. These metals are very much prone to readily dissolve relatively large amounts of oxygen. Dissolved gases greatly affect the mechanical properties of metals used for various structural applications at high temperatures and may become a crucial factor determining the life span and usefulness of such metals and their alloys. The solubility of gases in metals may qualitatively be correlated with the position of metals in the periodic table. As mentioned earlier, oxygen is readily soluble in the transition metals belonging to groups IVA and VA. However, the solubility in group VI metals (e.g., Cr, Mo, W, Se, Te) is extremely small and that in noble metals (e.g., Au, Ag, Pt) is quite high. As a measure of the solubility it may be referred that the oxygen solubility just below 1173 K, i.e., in the α phase of Ti, Zr, and Hf amounts to about 30 at. %, 28.5 at. %, and 20 at. %, respectively. The solid solubility of oxygen in the corresponding β phases is comparitively small; however, with an increase of temperature, solubility increases in both phases. In contrast, group VA metals show relatively smaller solubility which also increases with temperature and for Nb and Ta it amounts to hardly 5 at. % at 1773 K [57].

5.9.1

Diffusion of Oxygen in Metals

In the lattice structures of metals, interstitially dissolved atoms may occupy two types of interstitial sites: octahedral and tetrahedral. Octahedral sites are larger and can accommodate larger atoms than the tetrahedral sites. Thus, carbon, nitrogen, and oxygen atoms are expected to occupy octahedral interstitial sites. Diffusion of interstitial solute atoms in metals takes place by the interstitial mechanism in which solute atoms successively jump from one interstitial site to another. Empirically the temperature dependence of the diffusion coefficient of any species in a matrix phase can be given by the relation:

冢 冣

D ⫽ D 0 exp ⫺

Q RT

(5.120)

where D is the diffusion coefficient and D 0 is a preexponential function. Theoretical relations for the diffusion coefficient have been developed by Wert and Zener [57] on the assumption of random motion of the solute atoms and thus negligible interactions between interstitial atoms and substitutionally dissolved impurity atoms or parent lattice atoms. It has been observed that these relations are in

Tarnishing and Scaling Processes

261

fairly good agreement with experimental results for a number of parent metals. For diffusion of carbon, nitrogen and oxygen in bcc metals, values of D 0 often vary between 0.001 and 0.01. Experimentally determined values of activation energy for oxygen diffusion in bcc metals generally vary between 105 and 150 kJ/mol. The corresponding values in hcp metals are considerably higher and in the range of 180–220 kJ/mol. Similarly, activation energy values for carbon diffusion in Fe-Ni-austenite (1173–1373 K) of various compositions are reported to be about 120–160 kJ/mol [58]. Surface penetration of oxidant into metal matrix takes place by diffusion of oxidant solute into the solvent metal. The driving force for such diffusion is the chemical potential gradient from the surface (maximum oxygen activity) to the interior (minimum oxygen activity). Except for the few noble metals, all metals exhibit some reactivity with the oxidant that can be predicted by thermodynamic calculations. During dissolution and diffusion, at some stage, oxygen activity may reach a critical value finally leading to the formation of a continuous layer of some stable oxide that may be adherent or porous depending on the physical and chemical properties of the overlying oxide and underlying metal as discussed in earlier sections. In the case of adherent oxide layer formation, there will not be direct access of oxygen to the metal surface. Oxygen must diffuse through this layer. At the metal–oxide interface, since oxygen activity is brought down to a low level, further oxygen ingress into the metal matrix may become restricted. On the other hand, if oxide layer is nonadherent, there can be deeper surface penetration into the metal substrates. Extent of these phenomena, i.e., penetration of oxygen into metal matrix and simultaneous formation of a barrier oxide layer, are decided very much by the thermodynamics and kinetics of individual metal–oxidant reactions.

5.9.2

Grain Boundary Diffusion

Along with lattice diffusion of substitutional atoms it may be expected that an enhanced diffusion of interstitial solutes takes place along grain boundaries, which are supposed to be easy diffusion paths. It has generally been observed that the activation energy for the diffusion of interstitial solutes is of smaller values at relatively lower temperatures, and this hints that grain boundary diffusion predominates at low temperatures. Such effects have been observed for oxygen diffusion in zirconium, as studied by the oxygen penetration method, in the temperature range of 773–973 K [57].

5.9.3

Thermal Diffusion

Diffusion may also occur under the influence of a temperature gradient (LudwigSoret effect) in addition to migration under a concentration or electrical potential gradient. Thermal diffusion may be of considerable importance when a metal is

262

Chapter 5

being oxidized under a temperature gradient, e.g., zircalloy cladding in nuclear reactors. When starting with a homogeneous solid solution in a temperature gradient, the solute atoms migrate and build up a concentration gradient, and a steady state is reached when the opposing constraints of thermal and concentration diffusion just balance. The steady-state condition in a temperature gradient can be described by the expression: ln N ⫽

Q* ⫹C RT

(5.121)

where N is the atomic fraction of the solute at position ζ along the temperature gradient, T is the temperature in absolute scale at ζ, Q* is the heat of transport, and C is a constant. For a positive heat of transport, the solute atoms migrate toward the colder region as has been observed for migration of hydrogen, oxygen, and nitrogen in zirconium.

5.9.4

High-Temperature Oxidation Behavior of Group IV and V Metals

Many common features are observed during the oxidation of Ti, Zr, and Hf. High-temperature exposure of these metals to oxidizing environment results in simultaneous oxygen dissolution in the metallic phase and oxide scale formation. During the initial stage of oxidation all the three metals exhibit protective scale formation, whereas after extended exposures at high temperatures nonprotective oxide scale formation takes place. In such a situation it is essential to assess the oxidation behavior of these metals in terms of the relative importance of oxygen dissolution and oxide scale formation under different experimental conditions. In such studies a critical issue draws special attention to whether the oxygen gradient in the metal beneath the oxide scale and the dissolution process would bring about a behavioral change from protective to nonprotective. Studies on such reactive metals have recently gained special attention considering their wide applications in industry, especially in alloy forms.

Titanium Whenever titanium is exposed to an oxidizing environment, thermodynamically one should always expect the oxide scale to consist of a sequence of layers of different stoichiometries (TiO, Ti 2 O 3 , Ti 3 O 5 , TiO 2 ). However, at temperatures below 1273 K, only the rutile modification of TiO 2 has been detected in significant quantities. Oxidation of Ti follows different rate equations depending on temperature and

Tarnishing and Scaling Processes

263

Figure 5.32 Schematic diagram of rate equations observed in the oxidation of titanium [Ref. 57].

time of exposure. Figure 5.32 is a schematic diagram of the different rate equations observed in different temperature ranges at approximately 1 atm oxygen pressure. Figure 5.33 shows a double logarithmic plot of weight gain vs. time illustrating the oxidation behavior in the temperature range 673–1273 K. However, it is to be kept in mind that the observed oxidation behavior will be strongly influenced by the purity of the metal, pretreatment, surface preparations, etc. In general, below about 673 K, the oxidation follows a logarithmic rate equation, whereas between 673 and 873 K, a transition from logarithmic to parabolic kinetics or approximately a cubic rate equation is obeyed. Above 873 K the oxidation is of parabolic type and, after extended exposure, transforms into an approximately linear rate. The transition from parabolic to linear kinetics involves the formation of a continuous crack between the oxide scale and the metal when the stresses in the oxide reach a critical value. In such a situation the scale that is not at all adherent to the substrate forms a porous border of constant thickness. This border acts as a diffusion barrier resulting in linear kinetics. During the linear period, a set of layers of constant thickness are successively formed by the same mechanism and are separated from each other by a continuous crack. Above 1273 K, the linear oxidation behavior is followed by a decrerasing rate, again conforming to parabolic kinetics. While considering protective oxidation of Ti the question arises as to whether oxygen ions or titanium ions are the faster moving ionic species in TiO 2 . Rutile falls in the group of n-type semiconductors, but there is disagreement as to whether interstitial metal ions or oxygen vacancies are the predominant defects in rutile. It has been proposed that the defects may be more complex in nature than single unassociated point defects. It is not unreasonable to assume that diffusion of both titanium and oxygen ions contributes to the growth of rutile scale during oxidation of Ti. However, the activation energies of diffusion of the two types of ions being different, the relative importance of the two diffusion species will be governed by temperature. From inert marker studies, most of the research-

264

Chapter 5

Figure 5.33 Double logarithmic plot of weight gain vs. time in the temperature range 400–1000°C during oxidation of annealed van Arkel titanium [Ref. 57].

Tarnishing and Scaling Processes

265

ers have concluded that oxygen diffusion predominates below 1173 K, whereas diffusion of titanium ions becomes more important at higher temperatures. Parabolic oxidation of Ti above 873–973 K comprises of two simultaneous processes—oxygen dissolution and oxide scale formation. At such high temperatures, the oxide grains are larger and better developed with faceting than those at lower temperatures and correspondingly volume diffusion becomes the dominant diffusion process. During the parabolic oxidation period of the metal, attempts have been made to estimate the solubility of oxygen by using conventional metallography, microhardness, and x-ray diffraction techniques. Measurement of lattice parameter at the metal–oxide interface as a function of time suggests that, at 923 K and 973 K, oxygen is readily dissolved in Ti up to concentrations of 14–15 at. %, which corresponds to a composition Ti 6 O, at which order-disorder transformation may occur. Similar studies by Kofstad et al. [59] indicate that the oxygen concentration in an outer layer of the metal tends to a limiting value of TiO 0.35 when titanium is oxidized at 1173 K. It is further reported that 80% of the reacting oxygen gets dissolved in the metal during the initial parabolic oxidation at 1173 K, whereas Stringer [60] has reported a corresponding value of only 45% at 1223 K and admitted that scatter in the estimated values increases with temperature reaching a maximum at 1223 K. The activation energy for oxidation of titanium in the temperature range of 873–1273 K has been reported to lie between 209–230 kJ/mol, whereas at lower temperature, a smaller value has been suggested which is attributed to enhanced grain boundary migration. It is also important to note that Ti undergoes a phase transformation from α-Ti to β-Ti at 1155 K. It is argued that oxygen probably diffuses faster in the more open bcc structure, but at the same time it is to be remembered that oxygen is an α stabilizer. So during oxidation above the transformation temperature, there will always be an outer layer of α-Ti. On this basis it has been suggested that the phase change may be of minor importance with regard to the oxidation mechanism except during the very initial stages of the reaction. Once the parabolic oxidation changes to linear kinetics, the protective behavior is lost. Such transition from parabolic to linear oxidation takes place at shorter times, the higher the temperature of exposure. The linear oxidation rate is reflected through an increased rate of oxide formation whereas dissolution of oxygen continues at a rate governed by the prevailing oxygen gradient and the diffusion coefficient of oxygen in the metal. It is reported [57] that during linear oxidation the oxide has a lamellar structure parallel to the metal surface and the color of the thick scale formed at 1173 K for 17 h in 1 atm O 2 ranges from yellow to white. Pt-wire marker study has shown its position at the top of the scale, i.e., at the oxide–oxygen interface. The more or less uniformly light color of the entire scale suggests that there is no change in nonstoichiometry of the oxide across the scale. This leads to the logical conclusion that this portion of the scale is nonprotective and porous to oxygen which might

266

Chapter 5

have penetrated to the oxide–metal interface. Thus, the linear rate is attributed to phase boundary–controlled processes like nucleation and growth of oxide. It is further postulated that the lamellar structure is possibly due to repetitive exfoliation of a very thin outer surface layer of the oxygen-saturated metal core, which is readily converted to rutile. During the linear oxidation the metal–oxide interface tends to move inward at a linear rate, which will eventually become faster than, and will begin to overtake, the parabolic displacement of the oxygen gradient into the metal. However, when this happens, the oxygen gradient into the metal becomes steeper and causes a faster oxygen diffusion into the metal. As a result, a steady-state oxygen gradient should eventually be approached during the linear oxidation. After an extended period of oxidation beyond 1173 K, the rate begins to decrease with time and at 1473 K the linear stage is not detected at all. Some researchers have suggested that the decreasing rate is due to rate-limiting oxygen diffusion through the porous scale for which a model has also been proposed in which the pores are identified with screw dislocations. In contrast, Kofstad et al. [57] attributed this decreasing rate to a sintering process promoting grain growth of the outer surface scale. Once a compact oxide layer is formed, it can act as a solid state diffusion barrier causing decrease in reaction rate. The sintering and grain growth phenomena further suggest that the mobility of Ti in rutile becomes important at high temperature, which has been supported by Pt-wire marker study conducted at 1478 K in 1 atm 0 2 where the marker is positioned within the scale close to the metal–oxide interface.

Zirconium The reaction between zirconium and oxygen always leads to dissolution of oxygen in the metal and formation of ZrO 2 films or scales. Lower oxides of zirconium are not known. Similar to titanium, the oxidation of zirconium is logarithmic below 573–673 K. At higher temperatures, the reaction behavior changes and the general rate equation can be described by W n ⫽ kt, where the value of n has been found to vary from 3 to 2 depending on temperature, i.e., the oxidation kinetics vary from cubic to parabolic. Some investigators have proposed that such change in kinetics may be attributed to the surface preparation of samples because oxidation of mechanically abraded specimens could best be fitted to a cubic law, whereas chemically polished specimens conform to a parabolic relationship. From a number of studies on the oxidation behavior of zirconium it has been concluded that high-temperature protective oxidation deviates from ideal diffusion-controlled behavior, particularly during early stages of reaction. But after extended exposure, the oxidation tends to become approximately parabolic. After protective behavior, the oxidation rate suddenly increases (breakway oxidation)

Tarnishing and Scaling Processes

267

and eventually becomes linear. The onset of breakaway oxidation appears to be markedly dependent on impurities in the metal, and for high-purity zirconium the breakaway oxidation is significantly delayed. The loss in protective properties of ZrO 2 scale is of particular concern as regards the use of zircalloys in nuclear reactors. The ZrO 2 films formed on zirconium have been found to consist of three modifications depending on experimental conditions. Most x-ray diffraction studies on oxide films formed below 1273 K have revealed only the presence of monoclinic ZrO 2 , whereas electron diffraction studies on thin films have also indicated the formation of a cubic (or tetragonal) modification. The monoclinic form is thermodynamically stable below about 1473 K, whereas the tetragonal modification becomes stable at higher temperatures. It is generally accepted that ZrO 2 may be oxygen-deficient, but the extent of nonstoichiometry of ZrO 2 in equilibrium with oxygen-saturated metal is not exactly known. From marker studies on oxidation of zirconium it is concluded that inward migration of oxygen ions through oxygen vacancies of the ZrO 2 film/scale is the predominant factor in scale growth process. However, exact nature of the defect structure of ZrO 2 , the transport properties and possibility of any complex defect formation are still controversial issues. For temperatures above 873 K, a number of researchers [57] have experimentally determined the amount of oxygen going into solution as a function of time during oxidation of zirconium. In all of these investigations it has been shown that oxygen dissolution process follows a parabolic rate, which leads to the conclusion that the dissolution is governed by volume diffusion of oxygen in zirconium. The oxygen solubility has been reported to be about 60% of the reacting oxygen at 1123 K and this dissolution becomes increasingly important with a rise in temperature. The temperature dependence of the total parabolic rate constant and that for oxygen volume diffusion in α-Zr as reported by different investigators are summarized in Fig. 5.34. It is observed that the activation energy of the dissolution process is larger (212.5 kJ/mol) than that of the total oxidation process (151 kJ/ mol). Consequently, the dissolution process becomes increasingly important at higher temperatures of oxidation, whereas at temperatures below 773–873 K the dissolution process is of minor importance. It is appropriate to point out that the importance of oxygen dissolution will in reality be smaller than that indicated in Fig. 5.34 because of the simultaneous oxide formation, which has been neglected in the theoretical estimation of the contribution due to oxygen dissolution. For temperatures below 873 K Hussey and Smeltzer [61] have determined the fraction of the reacting oxygen that goes into solution by means of mass gain measurements and from estimates of oxide film thickness. Such results clearly reveal that large experimental errors are involved in estimating the thickness of thin films. If it is assumed that the estimated values are more or less correct,

268

Chapter 5

Figure 5.34 The parabolic rate constant for oxidation of zirconium as a function of 1/T. Data of Cubiciotti [66], Gulbransen and Andrew [67], Mackay [68], Wallwork et al. [69] and Hussey and Smeltzer [61]. The broken line shows the calculated maximum possible reaction rate due to volume diffusion of oxygen in α-Zr [Ref. 57].

then the contribution due to oxygen dissolution at 673–773 K is appreciably higher than would be expected from the maximum estimated mass gain due to volume diffusion of oxygen in zirconium (Fig. 5.34). The results as such indicate a contribution from grain boundary or short-circuit diffusion. Since the oxygen dissolution makes a parabolic contribution to the total oxidation process, the deviation from parabolic behavior must be due to oxide scale formation. At relatively lower temperatures (⬍873 K), the nonparabolic oxidation is attributed to enhanced diffusion along dislocation pipes and grain boundaries in the initially formed oxide. Similar to Ti and Hf, a simple model has been proposed assuming the density of oxygen sites in the short-circuit diffusion paths to decrease by a first-order rate mechanism. This implies that the effective diffu-

Tarnishing and Scaling Processes

269

sion coefficient is time-dependent and its value as a limit becomes equal to that of lattice diffusion. Under such circumstances, the rate of reaction would become faster than parabolic during initial stages, whereas after an extended period it will tend to approximate parabolic behavior. During oxidation of Zr at temperatures above 1248 K, voids have been found within the compact oxide layer. Such voids should act as barrier to the diffusion process resulting slower kinetics than parabolic. It has been proposed that void formation takes place through condensation of oxygen vacancies. But if oxygen diffusion takes place solely through a vacancy mechanism, it becomes difficult to account for condensation of vacancies in the middle of the oxide layer. Therefore, it appears that another migration mechanism would also have to operate in the oxide scale for condensation to take place. If one assumes a defect structure of Frenkel type for near-stoichiometric ZrO 2 , where both oxygen vacancies and interstitial oxygen are involved, it may be assumed that vacancy condensation within the scale might be a possibility. It has further been suggested through emf measurement studies across growing ZrO 2 film at 973 K that electron transport is the rate-limiting factor in the film growth process. After the initial protective oxidation, Zr and Zr-base alloys start oxidizing at an accelerated rate (breakaway oxidation). During protective oxidation, a compact oxide film is supposed to be formed that adheres tightly to the metal substrate even after cooling down to room temperature. The onset of breakaway oxidation is found to be accompanied by formation of white surface oxide, particularly at edges and corners of the specimens, and this color suggests that the oxide at the surface is close to being stoichiometric. On continued oxidation in the breakaway region, the surface is completely covered with white oxide, which is probably porous and poorly protective. It has been suggested that such change in kinetics may be due to a change in the modification of ZrO 2 from cubic and tetragonal to monoclinic, the latter having poor protective properties. It is significant to note that such breakaway oxidation is delayed during oxidation of high-purity zirconium and the presence of impurities or some specific alloy additions is considered to be an important cause of breakaway oxidation. Such observations have lended support to the fact that inhomogeneously distributed impurities, alloy additions, or intermetallic particles might play an important role. Analogous to the proposed mechanism by Smeltzer et al. [61], diffusion through the oxide is assumed to take place along some preferred paths. These may be provided by the misorientation in the oxide at grain boundaries and inhomogeneous region due to foreign elements originally present in the metal. A scheme of oxide film growth as suggested by Cox [62] is depicted in Fig. 5.35. This suggests that transport mechanism may lead to small-scale local variations in the growth rate, which, in turn, result in stresses nucleating failure in the oxide. Such nucleation sites are expected to increase with time. The cracks are also

270

Chapter 5

Figure 5.35 Proposed mechanism for film growth and failure in zirconium and zirconium alloys [Ref. 62]. expected to occur at sites of maximum stress in the oxide, preferentially at edges and corners. However, it is hoped that further studies on Zr-O 2 system will present a better picture of the different stages and mechanism of its oxidation.

Hafnium The oxidation behavior of hafnium is very similar to at of Ti and Zr. It is reported that the oxidation of Hf in the temperature range 623–1473 K could be well described by a logarithmic followed by parabolic and ultimately by a linear rate equation at higher temperatures. The logarithmic law is valid at low temperatures, and the duration of this stage decreases with a rise in temperature. Similar to Ti and Zr, the oxidation of hafnium also leads to simultaneous oxygen dissolution in the metal and formation of monoclinic HfO 2 film. Marker studies suggest that oxygen migrates inward through oxygen vacancies in the scale. The parabolic rate constant is reported to be independent of oxygen pressure variation at 1073 K, hinting at the n-type nature of HfO 2 . The temperature dependency of the parabolic rate constant has resulted in an estimated activation energy of 150 kJ/mol, similar to that for ZrO 2 scale growth. It is also proposed that the transition from protective to nonprotective behavior may be associated with cracking of the embrittled oxygen-enriched zone of the metal. However, any conclusive evidence for such a mechanism is still awaited.

Tarnishing and Scaling Processes

271

Tantalum and Niobium Oxidation of Nb and Ta has been studied over wide temperature and oxygen pressure ranges. Their poor oxidation resistance presents a major problem restricting their application in oxidizing environments at elevated temperatures. These metals exhibit a complex oxidation behavior, and various stages of the process, such as oxygen adsorption, oxygen dissolution, oxide nucleation and growth, diffusion through compact and cracked scales, and oxide evaporation, have been identified to be the rate-limiting factors depending on temperature, oxygen pressure, exposure time, etc. Many problems have been encountered in the interpretation of oxidation behavior of both of these metals [57,63] for which more contemplated and careful experiments are essential to unravel the true mechanisms of film/scale growth on Nb and Ta.

5.10 EFFECTS OF METAL SURFACE PREPARATION AND PRETREATMENT It is now accepted that the oxidation rate of metals is strongly influenced by the oxide grain size because grain boundary diffusion often governs the cation or anion transport rate, particularly at low and moderate temperatures. Studies carried out by different groups have further established that fine-grained polycrystalline oxide thickens most rapidly due to the presence of many grain boundaries that act as preferential paths for rapid diffusion. In contrast, monocrystalline oxide, which forms on some preferred orientations of metals, thickens very slowly showing the highest activation energy. Oxide grain size and morphology are found be highly dependent on factors like metal orientation, chemical and other surface pretreatments, cold work and oxidation procedures. The following section is an attempt to review the correlation of oxidation rates with different surface pretreatments on the oxidation behavior of pure metals like Ni, Cr, and Fe, which are usually used as base metals in the design of high-temperature alloys. The common observations of Graham et al. [64], who did extensive studies in these directions, strongly suggest that the oxidation rates vary considerably with surface pretreatments and employed oxidation procedures, which govern the initial oxide grain size and structure of the oxide films.

5.10.1 Oxidation Procedures and Different Methods of Specimen Surface Preparation The specimens normally used for oxidation experiments could be polycrystalline sheets or single crystals. A proper surface for oxidation studies should have a low roughness factor and be representative of bulk composition. The methods

272

Chapter 5

of specimen surface preparation often include either chemical polishing followed by electropolishing (EP) or electropolishing followed by light etching (ETCHED). Abrasion, the most widely used technique, produces a cold-worked (CW) distorted surface. Etching of such a surface to get rid of the worked layer may leave a rough surface that may be enriched with a more noble metal of the alloy constituents by preferential dissolution or redeposition. In a similar way, the procedure for oxidation tests could also vary. The test runs are generally carried out either in an ultrahigh-vacuum manometric apparatus, which is a closed system, or with the help of a continuously recorded thermobalance in a gas flowing system. ‘‘Hot-bare’’ (HB) run may be conducted on metal surface free from prior oxide after electropolishing followed by cathodic reduction with hydrogen. In such experiments, oxygen is allowed to come in contact with oxide-free specimen at the predetermined experimental temperature. In the ‘‘furnace-raised’’ (FR) procedure, EP or ETCHED specimens are heated in the presence of oxygen to the oxidation temperature. Cold insertion (CI) experiments are conducted by quickly lowering the test samples into the hot zone of a vertical tube reactor in flowing oxygen at a particular pressure.

5.10.2

Oxidation of Nickel

Results reported by Graham et al. [64] on oxidation of Ni at 973 K in 0.5 torr O 2 pressure are presented in Fig. 5.36. This figure clearly depicts that HB specimens oxidize at a higher rate than the FR ones of either EP or ETCHED surfaces. Moreover, the oxide formed on the HB run is reported to be fine-grained and uniformly thick over the entire surface, whereas the oxides formed on the ETCHED specimens are of large variation in oxide thickness with differences in substrate grain orientation. NiO is a p-type semiconductor with a predominance of point defects such as vacancies on Ni sites. Thus, NiO on Ni grows by outward transport of Ni ions through cation vacancies (lattice diffusion) and preferentially via oxide grain boundaries. With a finer oxide grain size there are more oxide grain boundaries to act as easy diffusion paths and therefore the oxidation rate is faster. Under HB oxidation conditions a large number of randomly oriented oxide nuclei initially form on all metal orientations, yielding the maximum oxidation rate. At temperatures of 973 K and below, the kinetic data obey a parabolic rate law with an associated activation energy of 155 kJ/mol, an approximate value for the growth of NiO via easy diffusion paths. On the other hand, under FR conditions the epitaxial relationships between NiO and the different Ni orientations produce overgrowths with a relatively restricted range of leakage path populations. In Fig. 5.36, the FR oxidation rate of ETCHED Ni is found to be slowler than the EP Ni, probably due to formation of a few NiO grains which contribute to highleakage paths during oxidation.

Tarnishing and Scaling Processes

273

Figure 5.36 Effect of surface pretreatment on the kinetics of oxidation of polycrystalline Ni at 700°C and 0.5 torr O 2 pressure. Scanning electron micrographs of the outer surface of the oxides illustrate a uniform fine-grained oxide on H. B. (‘‘hot-bare’’) Ni compared with a marked variation in oxide thickness with substrate orientation on ‘‘furnace-raised’’ ETCHED Ni [Ref. 64].

Figure 5.37 shows the early stages of Ni oxidation at 873 K in PO2 ⫽ 0.5 torr, demonstrating the effect of oxidation procedures and surface pretreatments. It clearly depicts how substrate orientation affects oxide grain size and oxidation rate. Similar to observations at 973 K, the HB specimen exhibits the fastest oxidation; however, EP Ni is found to oxidize at a slower rate than ETCHED Ni, contrary to the results at 973 K. After about 40 min exposure, however, an enhanced rate of oxidation is observed for EP Ni. At greater oxide thickness, ETCHED Ni undergoes oxidation at a relatively slower rate consistent with the results as shown in Fig. 5.36. On polycrystalline Ni, the kinetic data represent an average oxidation rate of different substrate orientations. Experimentation on a single crystal hemisphere shows that orientations close to (112) after EP oxidizes at a much reduced rate, where only ⬃10 nm of oxide film is formed in 1 h. Reflection electron diffraction has established the oxide to be single crystal [(111) antiparallel NiO on (111) steps of the (112) macrosurfaces], and its low growth rate is a consequence of the

274

Chapter 5

Figure 5.37 Effect of surface pretreatment on the early stages of oxidation of Ni at 600°C. H.B. (‘‘hot-bare’’) oxidation of polycrystalline Ni in 0.5 torr O 2 , F.R. (‘‘furnaceraised’’) oxidation of ETCHED and E.P. polycrystalline Ni in 0.1 torr O 2 , F.R. oxidation of E.P (112) Ni in 5 ⫻ 10⫺3 torr O 2 producing a singly orientated overgrowth [Ref. 64].

formation of oxide via lattice diffusion only. Experiments on other EP surfaces of Ni single crystals have exhibited decreasing oxidation rate in the following order: (100), (111), and (112). This difference in rate with orientation as well as changes in oxidation rate with surface pretreatments can be explained in terms of modification in the structure of the growing oxide. When a single crystal oxide, as on (112) Ni, persists during growth the oxidation rate is very low; when the oxide contains twin boundaries [as on (111) Ni] or becomes polycrystalline [as on (100) Ni], the oxidation rate is enhanced because of the availability of additional easy diffusion paths. The role of easy diffusion paths in governing oxidation anisotropy has been addressed by many research groups. The effect of cold work (CW) on oxide grain size and oxidation rate of Ni is depicted in Fig. 5.38. This figure clearly exhibits that at each temperature CW Ni oxidizes at a faster rate than the corresponding annealed ones. Such observations suggest that CW Ni oxidizes faster because of more oxide nucleation sites, leading to growth of finer grained oxide. Observation of surface scale morphology at

Tarnishing and Scaling Processes

275

Figure 5.38 Oxidation curves for Ni from 700 to 1270°C in 1 atm of O 2 , using the ‘‘cold insertion’’ (CI) oxidation procedure. At each temperature oxidation is greater for cold-worked (C.W.) Ni (broken lines) than annealed Ni (solid lines) [Ref. 64].

276

Chapter 5

different durations of exposure has established that the oxide grain size gradually increases with time (scale densification). Accordingly, the parabolic oxidation rate constant {k p ⫽ 2[d(∆W)/dt]} decreases progressively, consistent with decreasing population of easy diffusion paths due to scale densification. In comparison, the oxide grain size formed on annealed Ni is larger in the beginning and changes little with time; accordingly, k p remains fairly constant. The k p for CW Ni is made up of two components, i.e., the normal lattice diffusion, plus a contribution from leakage paths that is the greatest at initial stages of oxidation at lower temperatures where the oxide grain size is small. By correcting oxidation rates of CW Ni for lattice diffusion (using theoretical relation with contributions from both processes of migration), and by taking into account the changes in leakage path density during growth, it is possible to estimate the parabolic rate constants and an activation energy for growth via leakage paths only. The estimated activation energy is reported to be 155 ⫾ 8.4 kJ/mol, which is in agreement with the value for HB oxidation studies in the temperature range 573–973 K where growth of oxide takes place exclusively by cation transport across the film via highdiffusivity paths (grain boundaries and dislocation pipes).

5.10.3

Oxidation of Chromium

Similar to the oxidation behavior of Ni the electropolished Cr surface is also reported to result a fine-grained oxide layer and, consequently, a very rapid rate of oxidation as depicted in Fig. 5.39. Experiments conducted at 1363 K in 1 atm O 2 on EP and ETCHED Cr show that EP Cr oxidizes at a much faster rate than the ETCHED one. This is due to formation of fine-grained prior oxide film on EP Cr during electropolishing. Such a situation leads to the development of a fine-grained Cr 2 O 3 layer during a high-temperature oxidation test. In contrary, a well-epitaxed coarse-grained oxide forms on ETCHED Cr. Some orientations of ETCHED Cr are thought to develop monocrystalline scale for which the parabolic growth constant and the associated activation energy are minimum. The evidence of multilayered and highly blistered scale on EP Cr is a consequence of the compressive stresses that develop in the growing oxide layers. In contrast to NiO, Fe 3 O 4 , and single-crystal Cr 2 O 3 , which are considered to grow by outward cation diffusion, polycrystalline Cr 2 O 3 , possibly forms by outward diffusion of cation as well as ingress of oxygen along the oxide grain boundaries. As a result of such two-way traffic of both Cr and oxygen, new oxide formation within the growing scale leads to development of high compressive stresses. When the oxide contains a large population of easy diffusion paths, the high flux of cation vacancies is expected to promote scale separation at the metal–oxide interface. However, because of the high volatility of Cr, oxidation is sustained by Cr vapor transfer across the gap; generation of compressive stress from the two-way diffusion being continued leads to buckling and wrinkling of scale. The progressive stress

Tarnishing and Scaling Processes

277

Figure 5.39 Oxidation curves for E.P. and ETCHED Cr at 1090°C in 1 atm O 2 using the ‘‘cold insertion’’ (C.I.) oxidation procedure. Metallographic cross-sections show the oxide on E.P. Cr after 21 h to be a six-layered polycrystalline, wrinkled scale, whereas thin single-crystal oxide forms on some metal grains of ETCHED Cr [Ref. 64].

development and wrinkling leads to eventual failure of the scale by cracking and subsequently another fine-grained oxide layer is formed on the metal. This sequential process of scale wrinkling, failure, and new oxide formation may progress in segmental or repeated manner producing multilayered separated scales. In situ oxidation studies on high-purity Cr under scanning electron microscopy with different pretreatments have further demonstrated [65] that there exists a close correlation between the morphology of the thin films formed and the oxidation procedures.

5.10.4 Oxidation of Iron The oxidation rate of pure Fe is also reported to increase with a decrease in the initial oxide grain size. The behavior of CW Fe as depicted in Fig. 5.40 clearly shows a similar trend of decreasing k p with exposure time as discussed above for CW Ni. This obviously signifies subsequent oxide grain growth with exposure time. The very high value of k p is primarily due to rapid grain boundary diffusion

278

Chapter 5

Figure 5.40 Oxidation curves for Fe at 500°C in 1 atm O 2 using the ‘‘cold insertion’’ (C.I.) oxidation procedure showing the increased oxidation rate of cold-worked (C.W.) metal, together with sections through the oxide formed in 20h [Ref. 64].

through the fine-grained Fe 3 O 4 that initially forms on CW Fe. In addition, cold working suppresses the formation of voids at the Fe–Fe 3 O 4 interface by providing sinks to the inward-moving cation vacancies, resulting in formation of a uniformly thick, fairly adherent scale. In contrast, the film formed on annealed Fe is relatively thin and poorly adherent. Such poor adherance and separation of the

Tarnishing and Scaling Processes

279

oxide layer from the substrate at some locations can be attributed to void coalescence as a result of the arrival of vacancies. Moreover, these gaps cause a slower rate of reaction by impeding metal transport (at such moderate temperatures, metal vapor transport through the gaps is negligible due to very low volatility). The oxidation kinetics of pure Ni, Cr, and Fe have provided ample evidences to the fact that prior oxide grain size and hence surface pretreatment or adopted oxidation procedure strongly affect the subsequent scale growth process. So long as fine-grained oxide is produced and maintained, the oxidation rate is rapid because of enhanced cation transport via oxide grain boundaries. At times the effect of initially formed oxide grain size on subsequent oxidation rate can be large. For example, the parabolic rate constant of Ni oxidation at 873 K is ⬃10 4 times smaller for the growth of monocrystalline NiO than that for fine-grained polycrystalline nickel oxide. These features of oxidation tests clearly imply some practical utility considering the fact that proper choice of surface preparation, pretreatment, or specific oxidation procedure can be conducive to protective and adherent oxide layer growth.

REFERENCES 1. L. Coudurier, D. W. Hopkins, and I. Wilkormirsky, Fundamentals of Metallurgical Processes, International Series of Materials Science and Technology, Pergamon Press, Oxford and New York (1978), p. 79, 140, 141, 144. 2. O. Kubaschewski and B. E. Hopkins, Oxidation of Metals and Alloys. 2nd Edn., Butterworth, London (1967). 3. N. Briks and G. H. Meier, Introduction to High Temperature Oxidation of Metals, Edward Arnold, London (1983). 4. P. Kofstad, High Temperature Corrosion, Elsevier London and New York (1988). 5. N. B. Pilling and R. E. Bedworth, J. Inst. Met., 29:529 (1923). 6. G. Tammann, Z. Anorg. Chem., 111:78 (1920). 7. N. Cabrera and N. F. Mott., Rept. Progr. Phys., 12:163 (1948–49). 8. O. Kubaschewski and B. E. Hopkins, Oxidation of Metals and Alloys, 2nd edition, Butterworth, London (1967), p. 36. 9. P. Kofstad, Nonstoichiometry, Diffusion and Electrical Conductivity in Binary Metal Oxides, Wiley, New York (1972). 10. G. G. Libowitz, Energetics in Metallurgical Phenomena, (ed. W. W. Mueller), Gordon and Breach, IV:71 (1968). 11. F. A. Kro¨ger and H. J. Vink, Solid State Phys., 3:307 (1956). 12. W. L. Roth, Acta Cryst., 13:140 (1960). 13. F. Kock and J. B. Cohen, Acta Cryst., 25:275 (1969). 14. C. Wagner and W. Schottky, Z. Physik. Chem., B11:163 (1930); C. Wagner, Z. Physik (Bodenstein-Festhand) 177: (1931); B22:181 (1933). 15. K. Hauffe, Oxidation of Metals, Plenum Press, New York (1965). 16. E. J. W. Verwey, P. W. Haayman and F. C. Romeyn, Chem. Weekblad, 44:705 (1948).

280

Chapter 5

17. F. A. Kro¨ger, The Chemistry of Imperfect Crystals, 2nd edition, North Holland Publishing Company, Amsterdam, (1974). 18. C. Wagner and K. Gru¨newald, Z. Physik Chem, B40:455 (1938). 19. C. Wagner, Z. Physik Chem, B21:25 (1933); B32:447 (1936). 20. S. K. Bose and S. C. Sircar, Met. Trans, 5:2015 (1974); Trans. Indian Inst. Metals, 33:37 (1980); 33:45 (1980). 21. V. Ananth, S. C. Sircar, and S. K. Bose, Proc. Int. Conf. Corros. Sci. Tech. (ICMS85), (S. K. Bose and U. K. Chatterjee, eds.), (1985), p. 320. 22. S. K. Bose, V. Ananth and S. C. Sircar, Proc. 10th Congr. Metallic Corrosion, Madras, IV: Oxford and IBH, New Delhi, (1987), p. 3615. 23. S. K. Roy, V. Ananth and S. K. Bose, Oxid. Met., 43:185 (1995). 24. S. Mrowec and A. Stoklosa, Oxid. Met., 3:291 (1971). 25. S. Mrowec, A. Stoklosa and K. Godlewski, Crys. Lat. Def., 5:239 (1971). 26. S. K. Roy, S. K. Bose and S. C. Sircar, Oxid. Met., 35:1 (1991). 27. S. K. Bose, S. K. Mitra and S. K. Roy, Oxid Met., 46:73 (1996). 28. J. Gundermann, K. Hauffe and C. Wagner, Z. Physik Chem.; B37:148 (1937). 29. W. W. Smeltzer, Canadian Min. Met. Bull., November:1 (1962). 30. C. Wagner, Atom. Movements, ASM, Cleveland, Ohio:151 (1951). 31. K. Fueki and J. B. Wagner, J. Electro. Chem. Soc., 112:384 (1965). 32. F. Pettit, J. Electro. Chem. Soc., 113:1250 (1966). 33. V. Ananth, S. C. Sircar and S. K. Bose, Trans. Jap. Inst. Met, 26:123 (1985). 34. A. T. Fromhold, J. Phy. Chem. Solids, 33:95 (1972). 35. A. T. Fromhold, Theory of Metal Oxidation, Vol. 1: Fundamentals, North Holland Publishing Co., Amsterdam, (1976), p. 135. 36. E. C. William and P. C. S. Hayfield, Vacancies and Other Point Defects in Metals and Alloys, Institute of Metals (London), Monograph No. 23, (1957), p. 131. 37. H. H. Uhlig, Acta Met., 4:541 (1956). 38. S. K. Roy, P. K. Krishnamoorthy, and S. C. Sircar, Acta Met, 18:519 (1970). 39. S. K. Bose and S. C. Sircar, Brit. Corros. J., 8:177 (1973). 40. S. K. Bose and S. C. Sircar, Brit. Corros. J., 8:279 (1973). 41. P. G. Shewmon, Diffusion in Solids; McGraw Hill Book Company Inc., New York, (1963). 42. M. J. Graham, D. Caplan, and R. J. Hussey, Can. Met. Q, 18:283 (1979). 43. P. Hancock and R. Fletcher, Metallurgie, 6:1 (1966). 44. W. Jaenicke, S. Leistikow, and A. Stadler, J. Electro. Chem. Soc, 111:1031 (1964). 45. W. K. Appleby and R. F. Tylecote, Corr. Sci., 10:325 (1970). 46. M. Cagnet and J. Moreau, Compt. Rend, 244:2924 (1957). 47. J. D. Mackenzie and C. E. Birchenall, Corrosion, 13:783 (1957). 48. S. K. Bose and S. C. Sircar, unpublished work. 49. S. C. Kuiry, S. K. Roy and S. K. Bose, Oxid. Met., 41:65 (1994). 50. S. Mrowec, Corro. Sci., 7:563 (1967). 51. H. E. Evans, Int. Mat. Rev, 40(1):1 (1995). 52. A. T. Fromhold, in Stress Effects and the Oxidation of Metals, (J. V. Cathcart, ed.) American Institute of Mining Metallurgical and Petroleum Engineers, New York, (1975).

Tarnishing and Scaling Processes

281

53. B. Borie, C. T. Sparks, and J. V. Cathcart, Acta Met., 10:691 (1962). 54. J. D. Noden, C. J. Knights, and M. W. Thomas, Brit. Corr. J., 3:47 (1968). 55. P. Hancock and R. C. Hurst, in Advances in Corrosion Science and Technology, Vol. 4, (M. G. Fontana and R. W. Staehle, ed.) Plenum Press, New York (1974). 56. H. J. Engell and F. Wever, Acta Metall, 5:695 (1957). 57. P. Kofstad, High Temperature Oxidation of Metals, John Wiley & Sons Inc., New York (1966). 58. S. K. Bose and W. J. Grabke, Z. Metal Kunde, 69:8 (1978). 59. P. Kofstad, P. B. Anderson, and O. J. Krudtaa, J. Less Common Met., 3:89 (1961). 60. J. Stringer, J. Less Common. Met., 6:207 (1964). 61. R. J. Hussey and W. W. Smeltzer, J. Electrochem. Soc., 111:564 (1964). 62. B. Cox, J. Electrochem. Soc., 108:24 (1961). 63. J. Stringer, Reviews on High Temp. Mat., 1(3):256 (1973). 64. M. J. Graham, D. Caplan, and R. J. Hussey, Can. Met. Quarterly, 18:283 (1979). 65. S. K. Bose and R. A. Rapp, unpublished work. 66. D. Cubiciotti, J. Am. Chem. Soc., 72:4138 (1950). 67. E. A. Gulbransen and K. F. Andrew, Trans. AIME, 209:39A (1957). 68. T. L. Mackay, Trans. AIME, 227:1184 (1963). 69. G. R. Wallwork, W. W. Smeltzer, and C. J. Rosa, Acta Met, 12:409 (1964).

This Page Intentionally Left Blank

6 Alloy Oxidation

6.1 INTRODUCTION A large number of metallic components in a variety of industrial installations are exposed to oxidizing environments at elevated temperatures. Examples of such components are gas turbine blades and vanes, superheater and reheater tubes of boilers, reformer tubes, heat exchanger tubes, parts for fuel conversion and power-generating units, and parts for petrochemical industrial applications. Sometimes these components are exposed to mixed environments which simultaneously exhibit oxidizing, carburizing, and sulfidizing capacities. The extreme conditions imposed on metals and alloys can be exacerbated by thermal cycling and sudden high-temperature excursion. This may eventually reduce the service life of metallic components, leading to sudden failure and rupture. Furthermore, a type of accelerated corrosion, known as ‘‘hot corrosion,’’ is often encountered in gas turbine components in the presence of deposits of molten salts, such as sulfate, nitrate, and vanadate. The recent trend is to run the gas turbines at increasingly higher temperatures for superior efficiency. Superalloys have been designed to retain the high-temperature strength along with improved high-temperature corrosion resistance. Suitable coatings and claddings, such as MCrAlY, Ni-Al, and so forth, on the metallic components have been developed to combat the environmental degradation of such materials. Research in this direction is exhaustive for further improvement of high-temperature performance of components made up of special alloys and superalloys. It has also been reported that rare earth additions in the elemental 283

284

Chapter 6

form or their oxide dispersoids in alloys are quite beneficial with regard to retardation in degradation rate and improvements in oxide scale adherence. Development of high-temperature alloys has taken place through the efforts of two different research groups. The activities of the first group were mainly devoted to improvement of high-temperature strength, while the second group focused attention on improving high-temperature corrosion resistance. Therefore, the present demand for high-temperature alloys requires that the above-mentioned concurrent approaches be used so that strength property and corrosion resistance are simultaneously achieved in the alloy component under consideration. Many of the factors discussed earlier on the oxidation behavior of pure metals also apply to alloys. However, alloy oxidation is generally much more complex as a result of some or all of the following reasons: 1. 2. 3. 4. 5. 6.

The individual elements in the alloy will have different affinities for oxygen governed by the different free energies of formation of the respective oxides, Ternary or higher oxides may be formed, e.g., the occurrence of various spinel-type compound formation is a possibility, A degree of solid solubilities between the oxides may exist; if aliovalent, a doping effect may be seen. The various metal ions will have different mobilities in the oxide phase, The different metals will have different diffusivities in the alloy phase. Dissolution of different oxidants into the alloy phase may result in subsurface precipitation of oxides, carbides, nitrides, etc., of one or more alloying elements (internal oxidation).

Therefore, the primary requirements of high temperature alloys are as follows: 1. 2. 3.

Acceptable high-temperature mechanical properties, i.e., best combination of strength and toughness along with reasonable ductility, Ease of fabrication in forming and assembling the system components, Oxidation and corrosion resistance to the gaseous environment and perhaps to fused salt condensate (e.g., in gas turbine parts).

Until now, in the development of high-temperature alloys, most efforts had been directed to the development of materials for gas turbine engines. Over the past five decades, in the evolution of turbine materials based on Ni and Co alloyed with Cr, there has been a gradual increase in secondary alloying elements for strengthening purposes, with a corresponding reduction in Cr%. As pointed out earlier, situation demands that both oxidation resistance and strength property of an alternative alloy progress concurrently; however, position is such that strength development is more satisfactory than oxidation resistance. For high-temperature use it has become necessary to consider alternatives to presently used alloys. This has been done in the past by selecting metals on the basis of availability, melting point, physical properties, and chemical stability, in that order.

Alloy Oxidation

285

As an empirical measure of properties, half the melting temperature, i.e., 0.5 Tm, has been used as a criterion (where Tm is in °C, a temperature at which most metals still retain sufficient strength for engineering purposes). Thus, high-temperature materials can be defined as those having melting points above 1300°C where the performance temperature would be 0.5 Tm, i.e., 923 K [1]. On this basis, all of the metals of the first three groups of the periodic table, as well as the rare earth metals, are unsuitable. Some of the rare earth metals have acceptable melting points, but when alloyed, especially with other transition metals, they form eutectics with very low melting points. Sn, Pb, and Bi, due to their low melting points, may also be disregarded, and the actinides are unsuitable due to their strong radioactivity as well as scarcity. Scarcity also eliminates the transition metals of the Pd and Pt groups. This leaves the possible metals as arranged in Table 6.1, which are divided into two groups according to a system described by Kofstad [2]. The usefulness of the metals in group A at high temperatures is doubtful because of their high affinity for interstitial elements like O, N, C, and H. These elements are easily dissolved in such metals, forming ordered and sometimes metastable martensitic phases that not only affect the physical properties of the metals but also appear to have adverse effects on their oxidation resistance. The group B metals have been studied extensively, and the basic mechanisms of their strengthening are fairly well understood, particularly for Ni-based alloys. The basic mechanisms involved include solid solution hardening; dispersion hardening; precipitation of borides, nitrides, and carbides; and intermetallic precipitation and fiber reinforcement. Therefore, in high-temperature alloy development it is necessary to establish the compatibility of the alloy additions made to promote these mechanisms with those promoting oxidation resistance. The optimum conditions for oxidation resistance occur when a compact, adherent, protective oxide layer develops on the metal surface. The rate of further degradation is then governed by the growth rate of the scale, which in turn is controlled by the solid-state diffusion of cations or anions through it. Develop-

Table 6.1 Division of Possible High-Temperature Metals into Two Groups [2] Group A Ti Zr V Hf Nb Ta

m.p. (°C)

Group B

m.p. (°C)

1668 1852 1900 2222 2468 2996

Ni Co Fe Cr Mo W

1453 1495 1535 1875 2610 3410

286

Chapter 6

ment of an oxidation-resistant alloy therefore concerns the properties of the different oxidation products formed of the alloys. Most of the oxides of the base metals considered as possible high-temperature alloys are unsatisfactory for such reasons as volatility (WO 3, MoO 3, CrO 3, SiO), fast diffusion of anions (ZrO 2, TiO 2), fast diffusion of cations (FeO, CoO, NiO), low melting temperature, and low free energy of formation relative to other oxidation products in the alloy (Table 6.2). Therefore, alloying elements must not only provide the desirable mechanical properties to the alloys but must also promote the formation of an oxide with a high melting point, a relatively high negative free energy of formation, and low diffusion coefficients for both cations and anions. Accordingly, the protective oxides should be as close to stoichiometric composition as possible. Their defect structure and diffusion properties must not be adversely affected by the impurities. CaO and MgO can be excluded from Table 6.2 because they are highly prone to water and CO 2 contamination. MnO, FeO, CoO, and NiO can also be eliminated due to their large deviations from stoichiometry, relatively fast diffusion of cations, and relatively small free energies of formation. The group of rare earth oxides are impractical because such metals generally form extremely low-temperature eutectics with base metals such as Ni, Co, Fe. Therefore, by elimination, one is left only with the oxides, such as SiO 2, Al 2O 3, and Cr 2O 3. The first two are well known as the passive elements of coatings, whereas Cr 2O 3 currently provides the oxidation resistance in most uncoated superalloys and heat-resistant steels. SiO 2 may be used as a protective oxide, but

Table 6.2 Free Energies of Formation of Oxides of Possible Use as Protective Scales Oxide

∆G° at 1000°C/ mol O 2 (kcal)

CaO MgO Al 2O 3 CeO 2 TiO 2 SiO 2 MnO Cr 2O 3 FeO CoO NiO

⫺242 ⫺228 ⫺201 ⫺189 ⫺165 ⫺153 ⫺140 ⫺125 ⫺87 ⫺67 ⫺56

Alloy Oxidation

287

at elevated temperatures its reduction to the volatile SiO may occur or it may react with other metallic ions to form complex molten oxides. It is also known that the protective properties of Cr 2O 3 are reduced at temperatures exceeding 1273 K and at high flow rates of oxidizing environment. Under such conditions, Cr 2O 3 may be converted to the volatile CrO 3 or, in the presence of water vapor, to volatile Cr(OH) 3. At higher temperatures, an Al 2O 3 scale appears to provide greater protection. The diffusion rate of the cations through the Al 2O 3 layer is also much smaller than through the Cr 2O 3 scale. Furthermore, evaporation of a volatile oxide does not pose any serious problem in the case of an Al-O system in contradistinction to a Cr-O system. It therefore emerges that the optimum protective simple oxide scale should be alumina. However, thermodynamic considerations predict that internal oxidation of Al will always occur unless the concentration of Al in the alloy exceeds a critical value. Mixed oxides of the general composition AB 2O 4 (A and B represent two metallic components) have often been identified as oxidation products of Fe-, Ni-, and Co-based alloys. The most important properties of these compounds with respect to oxidation are the diffusion coefficients of the cations and anions, which are usually much smaller than in their parent oxides. The diffusion coefficient very much depends on the stoichiometry of the particular spinel. However, the ability of a spinel to provide oxidation resistance depends not only on diffusion coefficients but on the morphology of the spinelcontaining scale. Generally, two different modes of oxide layer growth are expected to occur. First, it can be the only oxidation product allowed to grow as a continuous scale. Second, it can form as a byproduct of the solid-state reaction of its parent oxides and eventually be interdispersed in these oxides. In the first case, the diffusion coefficients of the transported species determine the oxidation rate. However, in the second case, the effect of the spinel is only marginal because the matrix of the parent oxides with their higher diffusion coefficients may serve as easy transport channels, thus outflanking the beneficial effect of spinel phase. A mechanism of this type has been identified for the oxidation of some alloys in the Co-Cr system by Kofstad et al. [3], where the spinel CoCr 2O 4 was found to be dispersed in the CoO scale. Even in situations where the spinel forms the bulk of the scale, the presence of very small CoO channels in the spinel has been found to undermine the protective properties of the spinel. It is further emphasized that oxidation in such a situation proceeds at a much slower rate than through bulk CoO scale, though at a faster rate than through Al 2O 3 or Cr 2O 3 scales. It should be kept in mind that only under very special conditions of temperature, oxygen pressure, and alloy composition do compact spinel layers develop, as has been reported for the case of Udimet 700 at 1311 K [4]. However, the ability of a spinel phase to improve oxidation resistance of an alloy depends largely on the type of spinel being formed.

288

Chapter 6

From the above discussion, it can be concluded that high-temperature strength or creep resistance is best offered by FCC (austenitic) alloys based on Fe, Ni, or Co. Additions of Cr, Al, and perhaps Si to these alloys will impart oxidation and hot corrosion resistance. Other alloying elements chosen for strengthening purposes include C, Mo, W, V, Nb, Ta, Ti, and Zr, together with austenite-promoting elements such as Ni and Mn to permit heat treatment. Under severe and adverse environmental conditions, surface protection by means of coatings or claddings is frequently used for improving resistance to environmental attack with simultaneous retention of strength. The upper limit of temperature at which alloys can be used in most engineering applications appears to be about 1473 K. Above this temperature, they start losing their mechanical and chemical stabilities. The presently available superalloys lose their strength above 1523 K and melt at around 1723 K.

6.2 DOPING EFFECT The phenomenon of doping is relevant to an understanding of the defect models of the oxide film or scale material as well as the mechanism of film or scale growth processes on metals. It is therefore important in the study of physical chemistry and electrochemistry of ionic compounds and semiconductors. It is known that electrically active aliovalent impurities dissolved in the corrosion product scale bring about changes in the concentrations of ionic and electronic defects. Since the transport of these defects through the film/scale decides the rate of diffusion-controlled film/scale growth process, the presence of such impurities in the film/scale is expected to bring about a change in the rate of film/scale growth. This is widely known as Wagner–Hauffe rule or the valence approach to alloy oxidation, also referred to as the doping effect [5]. According to this rule, the effect of doping can be observed only when 1. 2. 3.

The impurity atom or intentionally added atom has a valence different from that of the parent atom, i.e., host lattice atom, In practice, the doping effect is applicable to alloys with minor alloying additions only, and The impurity oxide (solute oxide) must be soluble to some extent in the host oxide lattice.

The effects of a higher or lower valency foreign ion addition on the oxidation rates of dilute alloys are discussed below where the host oxide lattice exhibits predominance of p-type, n-type, or ionic conductivity. In such discussion, NiO as a typical p-type oxide, ZnO as a typical n-type oxide, and AgBr as a typical ionic conductor are considered.

Alloy Oxidation

6.2.1

289

n-Type Oxide (ZnO)

The native defect structure of such oxides involves excess cations at interstitial positions of the oxide lattice with excess electrons in its conduction band. As discussed earlier, this may be represented by the following defect formation equilibrium: ZnO s Zn ••i ⫹ 2e′ ⫹

1 O 2(g) 2

(6.1)

for which one can write the following equilibrium constant at a constant temperature, T: K(T) ⫽ [Zn ••i] n 2P O1/22

(6.2)

where [Zn ••i] and n represent the concentrations of interstitial zinc ions and excess electrons per m3 of ZnO lattice. To represent the addition of higher valency cations into the ZnO lattice, one may choose Cr 2O 3 as the impurity oxide. The solid solution of Cr 2O 3 in ZnO will bring about changes in defect concentration of the host lattice in the following ways: 1. Two Cr3⫹ ions may occupy the two normal lattice sites of Zn2⫹ ion. Because only two corresponding anion lattice sites are available for the three oxygen ions, one of them must be discharged, releasing oxygen to the gas phase and putting two electrons in the conduction band as per the following defect equilibrium: • ⫹ 2e′ ⫹ 2O xO ⫹ Cr 2O 3 s 2Cr Zn

1 O 2(g) 2

(6.3)

2. The introduction of extra electrons to the conduction band will upset the equilibrium between them and interstitial Zn2⫹ ions as per Eq. (6.2) at a constant temperature. Thus, Cr 2O 3 may also dissolve in a manner whereby some interstitial Zn2⫹ ions are eliminated according to the following equilibrium: • Cr 2O 3 ⫹ Zn ••i s 2CrZn ⫹ Zn xZn ⫹ 3O Ox

(6.4)

Therefore, the net result of doping ZnO with Cr 2O 3 is to reduce the concentration of interstitial Zn ions as well as to enhance the concentration of excess electrons in the conduction band, thus reducing the cationic conductivity (σ i) and increasing the electronic conductivity (σ e′). Since ZnO film growth is governed by its cationic conductivity, effect of doping with Cr 2O 3 would decrease the oxidation rate of an alloy of Zn containing a small amount of Cr.

290

Chapter 6

To represent the addition of a lower valency cation to the ZnO lattice, one may consider the solution of Li 2O in ZnO. This will also affect the defect concentration of the parent ZnO lattice in two ways: 1.

Two Li⫹ ions can be accommodated in two normal Zn2⫹ sites when only one anionic site would be occupied. The second anionic site could be filled by the incorporation of 1/2O 2(g) from the gas phase with simultaneous capture of electrons from the conduction band in order to ionize the gas as per the following defect equilibrium: Li 2O ⫹ 2e′ ⫹

2.

1 O 2(g) s 2Li′Zn ⫹ 2O Ox 2

(6.5)

Since removal of electrons from the conduction band upsets the equilibrium condition of Eq. (6.2), an alternative scheme for Li 2O incorporation would be to displace one Zn2⫹ ion from the normal Zn site to interstitial site as per the following defect equilibrium: Li 2O ⫹ ZnO s 2Li′Zn ⫹ Zn ••i ⫹ 2O xO

(6.6)

Therefore, the net result of doping ZnO with Li 2O would be to increase the concentration of interstial Zn ions and reduce the concentration of electrons in the conduction band. As a consequence, there will be an increase in the cationic conductivity (σ i) and a reduction in electronic conductivity (σ e′). Such doping would therefore lead to an increased oxidation rate of Zn containing a small amount of Li. The lattice defect models of pure ZnO along with its modified versions due to doping by Cr 2O 3 and Li 2O are presented in Fig. 6.1.

6.2.2

p-Type Oxide (NiO)

The intrinsic defect structure of such compounds involves cation vacancies and complementary electron holes in the valence band. In the case of an NiO lattice, the incorporation of O 2 from the gas phase into the oxide lattice can be represented as: 1 O 2(g) s O Ox ⫹ V″Ni ⫹ 2h• 2

(6.7)

for which one can write the following expression of equilibrium constant at a temperature, T: K(T) ⫽

[V″Ni]p2 P O1/22

(6.8)

where [V″Ni] and p are the concentrations of cation vacancies and positive holes, respectively, per m3 of NiO lattice.

Alloy Oxidation

291

Figure 6.1 Idealized lattice structure of ZnO, an n-type semiconductor. (a) pure ZnO; (b) effect of Cr⫹⫹⫹ additions; and (c) effect of Li⫹ additions.

The consequences of dissolving cations of higher and lower valences than the host ion may be considered in a similar way as dealt for n-type oxides. In order to satisfy Eq. (6.8), the effects may be reflected in the following ways. 1. Two Cr3⫹ ions may occupy two normal Ni2⫹ sites with evolution of the extra oxygen ion from Cr 2O 3 to the gas phase, contributing two electrons to neutralize the two positive holes in the valence band [since K i(T) ⫽ n.p], i.e., two positive holes will be consumed as per the following defect equilibrium: • Cr 2O 3 ⫹ 2h• s 2Cr Ni ⫹ 2O xO ⫹

1 O 2(g) 2

(6.9)

292 2.

Chapter 6 Two Cr3⫹ ions may occupy two normal Ni2⫹ ion sites and three oxygen ions may occupy three normal O2⫺ sites; thus, an Ni-ion vacancy will be created as per the following defect equilibrium: • Cr 2O 3 s 2Cr Ni ⫹ V N″i ⫹ 3O Ox

(6.10)

So the net effect of dissolution of higher valent cation into cation-deficient ptype oxide (NiO) leads to the creation of more cation vacancies and reduction in the concentration of electron holes. As a consequence, there will be an increase in cationic conductivity and a decrease in electronic conductivity. The net result would be to increase the oxidation rate as a result of easier diffusion of Ni2⫹ ions through the increased number of cation vacant sites. This means that Cr-doped Ni would oxidize at a faster rate than pure Ni at the same temperature and partial pressure of oxygen. The effect of dissolving a cation of lower valency such as Li⫹ ion into NiO may lead to the following situations with regard to defect creation or defect annihilation: 1.

When two Li⫹ ions occupy two normal Ni⫹2 ion sites, an excess oxygen atom would be ionized from the gas phase to fill the second anionic site for which two electrons are to be taken up from the conduction band. This means [since K i(T) ⫽ n.p] that the two positive holes will be injected into the valence band as per the defect equilibrium: Li 2O ⫹

2.

1 O 2(g) s 2Li′Ni ⫹ 2h• ⫹ 2O Ox 2

(6.11)

One of the two Li⫹ ions may also occupy an Ni-vacant site, i.e., it blocks the vacant site as per the following defect equilibrium: Li 2O ⫹ V N″i ⫹ O Ox s 2Li′Ni ⫹ 2O Ox

(6.12)

As a consequence, Ni-vacancies will be consumed and electron holes will be produced to maintain the equilibrium in Eq. (6.8). This means that the cationic conductivity (σ i) would decrease and hole conductivity (σ h•) would increase. Therefore, it can be inferred that Li-doped Ni would be expected to oxidize at a slower rate than that of undoped Ni. Figure 6.2 shows the schematic lattice structures of pure NiO, Cr 2O 3-doped NiO, and Li 2O-doped NiO, respectively. The above models have been verified for oxidation of copper [6] showing enhancement and reduction in the oxidation rates of Cr-doped copper and Lidoped copper compared with corresponding undoped copper.

6.2.3

Ionic Conductor (AgBr)

For pure AgBr, incorporation of Br 2(v) into a growing AgBr film on Ag can be represented by the following defect equilibrium:

Alloy Oxidation

293

Figure 6.2 Idealized lattice structure of NiO, a p-type semiconductor. (a) Pure NiO; (b) effect of Cr⫹⫹⫹ additions; and (c) effect of Li⫹ additions.

1 Br 2(v) s Br xBr ⫹ V A′ g ⫹ h• 2

(6.13)

for which one can write the following expression of equilibrium constant at any temperature, T. K(T) ⫽

[V A′ g]p 1/2 P Br 2

where [V A′ g] represents the concentration of cation vacancies.

(6.14)

294

Chapter 6

Furthermore, due to a predominance of Frenkel-type disorder, pure AgBr is always in equilibrium with some cation vacancies and cation interstitials. For example: AgBr s Ag i• ⫹ V A′ g ⫹ Br xBr

(6.15)

and also [V′Ag][Ag i•] ⫽ K′F

(6.16)

where K′F is the equilibrium constant for ionized Frenkel defect formation at a constant temperature, and [V′Ag] and [Ag i•] represent the concentration of ionized cation vacancies and cation interstitials per m3 of AgBr. When AgBr is doped with CdBr 2, the following defect formation reactions are expected: • x (a) CdBr 2 s Cd Ag ⫹ V A′ g ⫹ 2Br Br

(6.17)

• CdBr 2 ⫹ Ag i• s Cd Ag ⫹ Ag xAg ⫹ 2Br xBr

(6.18)

or

and • x (b) CdBr 2 ⫹ h• s Cd Ag ⫹ Br Br ⫹

1 Br 2(v) 2

(6.19)

According to Eq. (6.14), an increase in [V′Ag] must decrease p, i.e., with increase of σ i, σ h• should decrease. As AgBr is predominantly an ionic lattice, positive hole transport through AgBr is the rate-limiting step for subsequent growth of the halide film. Thus, the rate of bromination decreases for Cd-doped Ag compared with that for undoped Ag. This model has been verified for Ag– halogen systems [7,8], and the kinetic data as reported by Hauffe [5] are presented in Fig. 6.3. As regards doping effect, one has to remember that the effect will be more pronounced only when the intrinsic defect concentration of the reaction product is small. If the inherent defect concentration of a compound is already quite high, very little influence is expected, e.g., doping effect of the FeO growth rate on Fe will invariably be less pronounced than those for NiO or Cu 2O on the corresponding metals. This is because of the fact that the inherent defect concentration in FeO is more than 10 at.% while those in NiO or Cu 2O are less than 0.1 at.% at 973 K while exposed to the same oxygen pressure.

Alloy Oxidation

295

Figure 6.3 Parabolic course of the weight increase during the bromination of silver and silver-cadmium alloys at 330°C and 170 mm Hg bromine partial pressure, according to Hauffe and Gensch (∆m/q in g/cm2; the numbers on the straight lines denote at.% Cd).

6.3 INTERNAL OXIDATION AND CATASTROPHIC OXIDATION 6.3.1

Internal Oxidation

Internal oxidation is a process whereby oxygen diffuses into an alloy and causes subsurface precipitation of the oxides of one or more alloying elements. This subject has been well reviewed by Rapp [9], Swisher [10], and Meijering [11]. It is already known that at high temperatures there exists a possibility for dissolution of nonmetallic species, such as O, C, and S, into metals and alloys. The dissolution and diffusion of oxidants into the alloys introduces some embrittlement and causes subsurface precipitation of oxides, carbides, or sulfides of the more reactive components in the alloys. Dilute solid solution alloys comprising such base metals as Fe, Ni, Co, Cu, and Ag, with less noble alloying elements

296

Chapter 6

such as Cr, Mn, Al, Si, Zr, Be, or In, when exposed to either single oxidant or mixed environments such as a mixture of CO, CO 2, CH 4, H 2O, H 2S (i.e., in simultaneous presence of oxygen, carbon, and sulfur potentials) exhibit considerable solubility and diffusivity of the oxidants in the alloys at high temperatures, leading to internal oxidation. For simplicity and convenience, oxygen is considered here as the single oxidant. Then the process occurs by dissolution of the oxidant in the base metal (either at the external surface or at the alloy–scale interface in the presence of an external scale), which diffuses inward through the base metal matrix containing previously precipitated internal oxide particles. At an advancing reaction front (parallel to the alloy external surface), the critical solubility product, a BaνO(B ⫹ νO ⫽ BO ν) for the nucleation of precipitates is established by the inward diffusion of oxygen and the outward diffusion of the reactive solute atoms, B. Subsequently, nucleation and growth of the oxide precipitate occur until the reaction front moves forward and depletes the supply of solute atoms arriving at the precipitate. Further growth of the precipitate takes place only by capillary-driven coarsening (Ostwald ripening). The necessary criteria for the occurrence of internal oxidation in a binary alloy, A-B (B is a more reactive solute and A is a noble solvent) during its isothermal oxidation at a constant oxygen pressure are as follows: 1.

2.

The standard free-energy change of formation (per mole O 2) for the solute metal oxide, BO ν (ν is the number of oxygen ions per B ion in the oxide), must be more negative than the standard free-energy change of formation (per mole O 2) for the lowest oxide of the base metal. The free-energy change for the reaction B ⫹ νO ⫽ BO ν

3. 4.

must be negative. This implies that the base metal must have a significant solubility and diffusivity for atomic oxygen at the oxidation temperature to establish the required activity of dissolved oxygen (O) at the reaction front. The solute concentration of the bulk alloy must be lower than that required for transition from internal to external oxidation. The surface layer formed during chemical or mechanical preparation of the alloy surface must not prevent dissolution of oxygen into the alloy at the start of oxidation.

In practice, internal oxidation of an alloy that fulfills conditions 1 and 2 may be prevented by intentional elimination of conditions 3 and 4. From a practical standpoint, internal oxidation was initially considered to be undesirable since by such process unwanted inclusions can be introduced into an otherwise clean alloy. However, internally oxidized copper alloys are now commercially available, and

Alloy Oxidation

297

the development of dispersion-strengthened alloys gained momentum from the understanding of this phenomenon. The earliest comprehensive literature on internal oxidation, published by Rhines et al. in 1942 [12], provides not only experimental data on a number of copper alloys but also a formal treatment of the kinetics of the process. Subsequently, Darken [13] presented a more generalized treatment of the process combining occurrences of diffusion and precipitation. However, the most lucid treatment on kinetics of internal oxidation was provided by Wagner [14] in 1959, and the same approach is presented below in estimating the depth of internal oxidation zone (IOZ) as a function of exposure time. Internal oxidation requires that the rate of diffusion of oxygen in the alloy is appreciably faster than that of the solute element. In such a situation, an oxygen gradient is established in the alloy, and the dissolved oxygen reacts to form an oxide of the solute element in a zone beneath the alloy surface. For simplicity, one may consider the planar specimen geometry and the quasisteady-state approximation. A binary alloy, A-B is chosen in which B is a dilute solute that forms a very stable oxide. It is further assumed that the ambient partial pressure of oxygen is too low to oxidize A but high enough to oxidize B. In such a situation, the concentration profiles of dissolved oxygen and the alloying element during internal oxidation of A-B alloy free from any initially formed surface scale is schematically presented in Fig. 6.4. The quasi-steady-state approximation implies that the dissolved oxygen concentration varies linearly across the IOZ. Therefore, the oxygen flux through the IOZ will be given by Fick’s first law as

Figure 6.4 Schematic concentration profiles of dissolved oxygen and solute element for the internal oxidation of A-B alloy.

298

Chapter 6

J⫽

dm N (s) ⫽ D O O (mol cm⫺2 s⫺1) dt ξV m

(6.20)

where N (s) O is the oxygen solubility in A (atom fraction), V m is the molar volume of the solvent metal or alloy (cm3 mol⫺1), D O is the diffusivity of oxygen in A (cm2 s⫺1), and ξ is the instantaneous thickness of IOZ. If counterdiffusion of solute B is assumed to be negligible, the amount of oxygen accumulated in the IOZ per unit area of reaction front is given by m⫽

N B(O)νξ (mol cm⫺2) Vm

(6.21)

where N B(O) is the initial solute concentration in the A-B bulk alloy. Differentiating Eq. (6.21) with respect to time, one obtains an alternative expression for the flux as dm N B(O)ν dξ ⫽ dt V m dt

(6.22)

Equating (6.20) and (6.22), one can write DO

N (O)ν dξ N (s) O ⫽ B ξV m V m dt

(6.23)

On rearrangement of Eq. (6.23), one obtains ξ dξ ⫽

N (s) O DO dt νN B(O)

(6.24)

Integration of Eq. (6.24), assuming ξ ⫽ 0 at t ⫽ 0, yields 1 2 N (s) D ξ ⫽ O (O)O t 2 νN B

(6.25)

or ξ⫽

冤

冥

2N (s) O DO t νN B(O)

1/2

(6.26)

Equation (6.26) gives the penetration depth of the IOZ as a function of oxidation time. The following points emerge from the consideration of Eq. (6.26): 1. 2.

The penetration depth has a parabolic time dependence: ξ ⬀ t 1/2. The penetration depth for a fixed time is inversely proportional to the square root of the atom fraction of solute in the bulk alloy.

Alloy Oxidation

299

3. Careful measurements of the front penetration as a function of time for an alloy of known solute concentration can yield a value of the solubility–diffusivity product, N (s) O D O (permeability) for oxygen in the metal matrix. The results of similar derivation for a cylindrical specimen [10] yield

冢冣

1 2N (s) (r 1)2 r O DO ⫺ (r 2)2 ln 1 ⫹ ⫽ t 2 r2 2 νN B(O)

(6.27)

and that for a spherical specimen [10] is 2 (r 2)3 2N (s) (r 1)2 O DO ⫺ (r 2)2 ⫹ ⫽ t 3 3 (r 1) νN B(O)

(6.28)

where r 1 is the specimen radius and r 2 the radius of the unoxidized alloy core. Many of the effects of internal oxidation, both on the overall corrosion process for an alloy and on the mechanical, electrical, and magnetic properties of the alloy, are intimately related to the morphology of the oxide precipitates. Formation of oxide particles by internal oxidation is a nucleation and growth process. Accordingly, for precipitation of very stable oxides, the driving force for the process is large, which facilitates nucleation of finer particles. On the other hand, for less stable oxides, the driving force being less, nucleation occurs with much difficulty rather favoring growth of existing particles. Transition from Internal Oxidation to External-Scale Formation Upon increasing the concentration of reactive solute in an internally oxidizable alloy system (e.g., Cu-Be, Cu-Si, Cu-Al, Ni-Cr, Ni-Al, etc.), a critical solute concentration will be reached above which a protective external scale is formed and the alloy is no longer oxidized internally. Assuming the formation of a compact, pore-free oxide scale, Wagner [14] has given a theoretical analysis of the transition from internal oxidation to external scale formation. This model is based on the fact that the cross-sectional area available for oxygen diffusion into the alloy is reduced due to the presence of internally formed oxide particles. So above a critical volume percent of oxide particles formed in the metal matrix, further oxygen diffusion into the metal gets retarded, creating a situation for formation of surface oxide layer only. This transition to external scale formation is the basis for the design of Fe-, Ni-, and Co-based high-temperature alloys, which contain sufficiently high concentration of a solute, such as Cr, Al or Si, to produce an external layer of a slow-growing stable oxide (i.e., Cr 2O 3, Al 2O 3 or SiO 2), which prevents further degradation of the alloy. The condition that favors external scale formation is N BD B ⬎⬎ N OD O, i.e., (1) high concentration of solute B associated with its rapid diffusion to form a continuous blocking layer of BOν and stop internal oxidation and (2) low solubility

300

Chapter 6

associated with low diffusivity of oxygen into the alloy. Under such conditions, the outward flux of the solute element exceeds the inward flux of the oxygen atoms. The decrease in the inward flux of oxygen can be achieved by lowering N(s) O , i.e., by maintaining lower pO2 in the atmosphere; the increased outward flux of solute B can be achieved by cold-working the alloy (increased contribution of short circuit diffusion paths), and by doing so, one can reach transition to external scale formation at lower solute concentrations in the alloy systems. According to Wagner, at the critical alloy composition, the volume fraction of precipitated oxide is just sufficient in developing a physical barrier to the ingress of oxygen and the existing oxide particles grow laterally to form a continuous protective oxide film. Therefore, the conditions under which transition from internal to external oxidation will take place are represented by: N B(O) ⱖ

冤

冥

πg (s) D OV NO 2ν D BV ′

1/2

(6.29)

where g DO DB V, V ′

⫽ ⫽ ⫽ ⫽

volume fraction of internal oxide zone occupied by BOν phase diffusion coefficient of oxygen in A diffusion coefficient of B in alloy molar volumes of alloy and BOν oxide, respectively.

With larger amount of precipitation due to the presence of increased reactive element concentration, the oxygen partial pressure at the advancing precipitating front is reduced to so low a level such that oxygen ingress to the nobler metal matrix decreases. Since reactive solute supply from the interior to the precipitation front is maintained, it favors more oxidation of the reactive metal, their lateral growth and subsequent coalescence to provide a continuous protective scale, thereby completely stopping internal oxidation. The critical solute concentration for an alloy will, however, be decided by the partial pressure of oxygen at any given temperature. The lower the oxygen partial pressure in the environment, the lower will be the critical amount of reactive solute requirement for exclusive external scale formation as demonstrated by Rapp and illustrated in Fig. 6.5 [15] for Ag-In alloy system at 823 K. At pO2 ⫽ 1 atm (105 N/m2), the transition was observed at N In(O) ⫽ 0.15, while by decreasing the oxygen partial pressure to 10⫺7 atm (10⫺2 N/m2), the transition occurred at N In(O) ⫽ 0.04. Such a decrease in the concentration of solute (In) is due to reduction in the flux of oxygen into the alloy. In the same investigation [15], the transition from internal to external oxidation at a particular oxygen pressure was demonstrated to depend on the initial surface preparation of the alloy. For mechanically polished specimens exposed to air at 823 K, the transition took place at N In(O) ⫽ 0.10, while for specimens etched in hot concentrated HNO 3 and electroetched in 1-N HNO 3, the

Alloy Oxidation

301

Figure 6.5 Transition from internal to external oxidation in Ag-In alloys at 883 K [15].

transition occurred at N In(O) ⫽ 0.12 and N In(O) ⫽ 0.15, respectively. The retardation of internal oxidation in mechanically polished specimens has been attributed to increased outward solute diffusivity along short circuit paths in the deformed layer or decreased oxygen solubility due to excessive deformation of the surface layer. The internal oxidation behavior of various alloy systems, such as Cu-Be, NiCr, Fe-Cr, and Fe-Si, Nb-Zr, has been discussed at length by Birks and Meier [16].

6.3.2

Catastrophic Oxidation

Until now, the formation of only solid oxide scale on metals and alloys has been considered. In some cases, however, liquid oxide phases may also form as a part of the reaction product. This may arise under the following two situations: 1. If metals or alloys are exposed to the vapor of a low-melting oxide during oxidation, or 2. During oxidation of some alloys with an alloying element which itself forms a low-melting oxide or which may form a low-melting eutectic of oxide mixtures.

302

Chapter 6

It is natural that liquid oxide phase cannot provide protection to the underlying substrate alloy and may lead to excessively fast oxidation and eventual disintegration of the alloy. This phenomenon is commonly termed catastrophic oxidation, as postulated in 1948 by W. C. Leslie. Important examples of catastrophic oxidation are found during oxidation of metals and alloys in the presence MoO 3 (m.p. 795°C), V 2O 5 (mp. 674°C), Bi 2O 3 (m.p. 820°C), and PbO (m.p. 888°C). Leslie and Fontana [17] investigated the special features of this type of oxidation for such systems as Fe, Ni, Cr, stainless steel, Inconel, Hastelloy, and other commercial high-temperature alloys in the presence of MoO 3 and V 2O 5. In all cases, increased oxidation rates were observed compared to those in dry air. Since rapid oxidation took place, the oxide scales were invariably porous, spongy, or nonadherent to the substrate alloy. Rathenau and Meijering [18] also studied the oxidation behavior of a few pure metals, binary and ternary alloys when buried in MoO 3 powder. They too observed accelerated attack above certain critical temperatures depending on the metal or alloy system. However, they [18] were the first to point out the importance of a liquid oxide phase in such aggressive attack. They reported that accelerated attack occurred at the eutectic temperatures of the binary or ternary oxides involved. For Cr-containing alloys, the onset of catastrophic oxidation occurred near the temperature at which liquid MoO 3 dissolves Cr 2O 3. Unfortunately, no conclusive mechanism for this type of degradation has been elucidated. In the presence of oxide vapors, the liquid phase is probably initially formed on the scale surface. But when the metal forming the low-melting oxide is present as an alloying element, rapid attack may be initiated at the scale–alloy interface. Accordingly, it may be hypothesized that the liquid phase penetrates the scale along the initial oxide grain boundaries or microchannels or microcracks to the alloy surface where rapid degradation takes place. Presence of a liquid phase along the grain boundaries may also act as easy diffusion paths for both cations and anions, leading to accelerated attack. An elaborate study by Brenner [19] on ternary alloys like Fe-Ni-Mo and FeCr-Mo has shown that when Ni and Cr are added to binary Fe-Mo alloys, catastrophic oxidation occurs in certain concentration regions, as illustrated in Fig. 6.6, where low-melting ternary or quaternary compounds might form. From these results, the mechanism of such attack has been suggested for Fe-Mo-Cr alloys as depicted in Fig. 6.7. During initial oxidation, Fe, Ni, and Cr are preferentially oxidized, and the oxidation is protective in nature. With progress of oxidation, Mo becomes enriched at the alloy interface, leading to formation of an inner layer of MoO 2. Catastrophic oxidation is initiated by the formation of a crack in the scale due to some sort of stress development; MoO 2 becomes oxidized to molten MoO 3, which penetrates along the alloy–scale interface. As Mo is less noble than the other alloying elements, MoO 3 will be reduced to a lower oxide of Mo or even to Mo. Simultaneously, molten MoO 3 may exert dissolving action

Alloy Oxidation

303

Figure 6.6 Boundaries of the composition range showing catastrophic oxidation for Fe-Mo-Cr and Fe-Mo-Ni alloys after 2 h in air at 1273 K [19].

on other oxides, such as Fe 2O 3 and Cr 2O 3, and this fluxing may be further accelerated by the enthalpy of formation of Fe 2O 3 and Cr 2O 3, which tends to increase the temperature at the alloy–scale interface. In such a situation, Fe 2O 3 and Cr 2O 3 are transported through the molten MoO 3 layer and precipitated at the MoO 3 (liquid)–solid oxide interface, which is comparatively cooler. In the above-described mechanism, it has been assumed that a crack in the solid oxide initiates catastrophic oxidation. However, it may also be hypothesized that the crack is the result rather than the initiating cause of such attack. During

Figure 6.7 Mechanism of catastrophic oxidation for Fe-Mo-Cr alloys [19].

304

Chapter 6

the process, Mo gets accumulated at the alloy interface; accordingly, a liquid MoO 2-MoO 3 mixture (eutectic melting point is 778°C) or a more complex liquid mixture may be formed, which may penetrate the scale along grain boundaries to the external oxide surface, where MoO 3 gets evaporated producing a porous nonprotective oxide. It is further reported [19] that increased content of nickel decreases the tendency for catastrophic oxidation. Thus, an Fe-Ni-Cr-Mo alloy with 20% Mo and 40% Ni did not demonstrate catastrophic oxidation. However, no suitable explanation has been provided for such observation. Similar types of attack are also encountered in alloy components when exposed to combustion products or ash from fuel oils containing vanadium. The combustion products may contain V 2O 5, SO 2-SO 3, etc., and, with the simultaneous presence of NaCl, oxygen, and water vapor, low-melting mixtures of Na 2SO 4 and (Na 2O)xV 2O 5 may be formed, which get deposited on the alloy surfaces. Such a situation leads to accelerated attack and degradation of the constructional materials. This type of corrosion damage is termed ‘‘hot corrosion,’’ to be discussed in Sec. 6.7.

6.4 SEQUENCES IN ALLOY OXIDATION Morphologies of the scales formed on alloys are time-dependent. The three generally observed stages in alloy oxidation are presented schematically in Fig. 6.8, where stages I, II, and III represent transient, steady-state, and break-away oxidation, respectively. Upon initial exposure of an alloy to an oxidizing environment at a fixed temperature and oxygen pressure, the oxides of essentially every reactive element (for which ∆G values of oxide formation are negative at the temperature and

Figure 6.8 Stages in alloy oxidation; I, transient stage. II, steady state. III, break-away stage.

Alloy Oxidation

305

partial pressure of oxygen under consideration) are formed in a proportion given by the composition of the bulk alloy. Oxide nuclei with high intrinsic growth rates due to high level of point defect concentrations, such as FeO, CoO, NiO, Cu 2O, etc., overgrow the nuclei of the slower growing oxides and spinels (e.g., Cr 2O 3, Al 2O 3, SiO 2, FeCr 2O 4, NiCr 2O 4, etc.). The rapid kinetics of overgrowth formation contribute to the high initial rate of oxidation. While the overgrowth is continuing, the underlying nuclei of the slower growing and usually more stable oxides grow laterally. Eventually they impinge one another and form a continuous layer, or may remain as isolated particles or precipitates in the faster growing oxide matrix. Which one of the two materializes is decided by a number of factors involved in determining the steady-state configuration of the scale, e.g., percentage of reactive solute in the alloy, chemical potential of the oxidant in the gas phase, interdiffusion coefficients of the elements in the alloy, surface finish and pretreatment of the alloy, oxidation procedure, etc. Eventually, the transient oxidation stage gives way to a steady-state scale formation stage, which essentially means that the morphology and composition of the scale remain independent of exposure time. Generally, the overall oxidation rate is governed by the transport of one or more ionic species through a particular layer in the total scale. As a consequence, the kinetics approximately conforms to a parabolic rate law. This further implies that the interface concentrations and indeed the concentration profiles through the scale and alloy, when expressed as a function of ξ/√t, are independent of exposure time. From a practical point of view, the duration of this steady-state period is the most critical factor for oxidation-resistant alloys. Any oxidation-resistant alloy depends on this period for continued protection. Accordingly, the steady-state period should be as long as possible. However, it cannot last indefinitely, since selective oxidation of one of the alloying elements of the alloy is taking place, and as a consequence, the effective service life will end when the supply of alloying element in the alloy is exhausted. Severity of operating conditions, such as higher temperature, rapid thermal fluctuation or cycling, higher gas flow rate, and higher level of mechanical or thermal stresses, will shorten this period. The steady-state scaling behavior of some commercially important binary and ternary alloy systems is discussed in Sec. 6.5. The ultimate (equilibrium) state of an alloy under oxidizing conditions is an oxide scale containing the alloy components in the same ratio as in the original alloy. Usually the end of the steady-state period occurs before all of the selectively oxidizable elements have been consumed by some type of mechanical failure of the oxide scale. This last stage is often related to mechanical influences, such as adherence, spallation, thermal or mechanical stresses, and void formation. Since after mechanical rupture of the scale, the alloy is of different composition than originally, it can no longer regain its original steady-state oxide composition and morphology, thereby leading to a more devastating situation for the depleted

306

Chapter 6

Figure 6.9 Oxidation–time curve for periodic cracking of scale.

alloy. The break-away or periodic cracking of the scale is presented schematically in Fig. 6.9. Transport of either metal ions or oxidizing species along preferential paths through scales such as grain boundaries, cracks, or pores must obviously lead to deviations in morphologies based on uniform one-dimensional transport process. Depending on the species preferentially transported, there may be possibilities of nonuniform precipitation of a second phase within the scale and scale growth at preferred sites, which may cause stress buildup, followed by scale fracture. As oxidation continues, the alloy gets consumed, and if alloy–scale adhesion is poor, a gap may be formed between the alloy and scale without fracturing of the scale. On the other hand, when fracture of the scale occurs, the oxidizing species in the atmosphere can easily have access to the underlying alloy, which has probably been denuded of the less noble metal. If the less noble element concentration becomes insufficient to ensure reestablishment of the original scale composition, a completely new oxidation reaction starts at each fracture site, with incorporation of larger fractions of the more noble elements in the newly formed scale than the original one.

Transient Stage High-temperature alloys in general are often expected to perform their major period of service life under steady-state conditions; however, a considerable time elapses, and a significant thickness of scale develops before attainment of this steady-state situation. This initial unsteady or transient oxidation process is governed not only by composition of the alloy and the properties of its corresponding oxides but by a number of practical factors, such as the method of heating up, the oxidation conditions, the degree of cold work, geometrical shape, and surface finish of the alloy. These factors are particularly important when operation is

Alloy Oxidation

307

close to the borderline between ‘‘protective’’ and ‘‘nonprotective’’ scaling, a situation common in practice where alloys are being exploited as close as is practicable to their useful limits. In industrial situations the pre-steady-state oxidation data are very important in deciding the conditions for commissioning new equipment. One has to recognize that the nature of the scales (protective or nonprotective) formed during the transition period, i.e., the spatial distribution, amount, composition, and structure, will determine to a large extent the nature of the steady-state scale. Even though the relative free energies of formation of the oxides predict the formation of thermodynamically favored oxide, they do not guarantee the composition of the initial oxide or of the steady state. The composition of the bulk alloy plays a major role in determining whether the thermodynamically favored oxide will appear as an external, often protective, scale or as an internal oxide. The free energies of formation of the oxides also determine the possibility of additional solid-state reactions between oxides of the scale and between the oxide and the alloy. A large alloy interdiffusion coefficient ensures rapid replenishment of the appropriate alloying element at the surface and thus help in establishment of the healing layer (i.e., eventual development of a complete layer of a protective oxide). Higher oxygen solubility and diffusivity in the alloy substrate tend to promote internal oxidation. The relative growth rates of elemental oxides and complex oxides will eventually determine the relative development and overgrowth rates of the initial nuclei and the scale encroachment rate on the alloy. The main scale may tend to incorporate the internal oxide particles, thus preventing them from producing a healing layer. The interplay of all of these factors under the prevailing oxidizing condition will determine the oxidation behavior of the alloy. Chattopadhyay and Wood [20] made detailed investigations on the pre-steady-state or transient oxidation behavior of a number of alloys comprised of three different sets of binary alloy systems: (1) Fe-Cr, NiCr, and Co-Cr alloys (in which the reactive solute metal is the same but the solvent noble metals are different and the oxides are partially miscible or they react); (2) Ni-Al, Ni-Cr, Ni-Si, Ni-Mn, and Ni-Co alloys (in which the solvent noble metal is the same but the less noble solutes are different having different affinities for oxygen and the oxide phases produced include solid solutions, largely immiscible simple oxides, and complex oxides); and (3) Cu-Ni, Cu-Zn, and Cu-Al alloys (in which the solvent noble metal is the same but the less noble solutes are different and their resultant oxides are immiscible). These studies have advanced our understanding of the mode of development of transient oxides on binary alloys which could be readily explained at least in a qualitative manner. To illustrate, the coalescence of initially formed oxide nuclei to give rise to the transient scale and its subsequent development into the steady-state scale is presented schematically in Fig. 6.10 for two different alloys of the Ni-Cr system [21]. At high temperatures (1273–1473 K), alloys in this system are of single phase up to 40% Cr, but the respective oxides, such as NiO and Cr 2O 3, are only

308

Chapter 6

Figure 6.10 Schematic representation of the transient oxidation of Ni-Cr alloys [21]. It should be noted that some NiCr 2O 4 is probably formed directly immediately and by reaction between NiO and Cr 2O 3, but is omitted from the diagrams during the early stages for simplicity’s sake.

slightly miscible; rather they tend to react in forming the compound NiCr 2O 4. Because Cr is more reactive, it is preferentially oxidized and produces the thermodynamically favored Cr 2O 3 doped with Ni⫹2 ions on the alloy surface. Nevertheless, some NiO doped with Cr⫹3 ions and NiCr 2O 4 may also form directly during the transient stage, but the compound NiCr 2O 4 can also form by the reaction between NiO and Cr 2O 3. On exposure of the alloys to an oxidizing atmosphere at high temperature, small NiO, Cr 2O 3, and NiCr 2O 4 nuclei develop rapidly on and in the alloy surface from the amorphous skin or prior air-formed skin and grow quickly until they impinge on one another. The nuclei of NiCr 2O 4 are omitted in the figure during

Alloy Oxidation

309

the early stages for simplicity even though they may play an important role, but this phase is incorporated at a later stage to illustrate the reaction between NiO and Cr 2O 3. For the alloy Ni-20–40%Cr, the NiO overgrows the more slowly developing Cr 2O 3. As the grains coarsen and the scale becomes more uniform, lattice transport tends to become rate determining in the growth of the nuclei and alloy interdiffusion starts playing a greater role. The Cr 2O 3 starts growing laterally as Cr at the alloy–oxide interface enters these nuclei directly accompanied by possible reduction of NiO in contact with the alloy by a displacement reaction. Densely distributed internal oxide particles of Cr 2O 3 behind the NiO nuclei eventually form a complete layer of Cr 2O 3 at the scale base by coalescing and linking up with the initially developing Cr 2O 3 nuclei. Once the Cr 2O 3 layer is complete, it continues to thicken, there being little further Ni incorporation into the scale, since the oxygen potential at the alloy–oxide interface is lower than the dissociation pressure of NiO. Ultimately, the protective doped Cr 2O 3 becomes the ratedetermining layer. However, Cr 2O 3 can react with the outer NiO to form NiCr 2O 4 also, and if this layer turns out to be sufficiently thick and compact, it can become rate determining. In comparision, the behavior of Ni-5%Cr alloy is different because its low Cr content never favors development of a complete Cr 2O 3 layer at the scale base, even though the initial stages of oxide formation are similar. Accordingly, NiO in the vicinity of the alloy–oxide interface partly dissociates, supplying the alloy with atomic oxygen, which diffuses inward, producing a front of internal Cr 2O 3 particles. During scale thickening, these are incorporated into the inner layer as NiCr 2O 4 particles. The inner layer may be a porous one due to vacancy coalescence and void formation, and oxygen gas transport may become operative, supplied by the dissociation of oxide. Thus, a steady-state scale for Ni-5%Cr alloy consists of a two-phase surface oxide with NiCr 2O 4 in a matrix of NiO and internal oxide of Cr 2O 3. The NiO and Cr 2O 3 may be doped as in the case of higher Cr(20–40%)–containing alloys and such doping will influence the scaling rate. However, it should be noted that a complete layer of equilibrium-doped Cr 2O 3 is not formed on low Cr–containing alloys because of the spatial and diffusional factors. The alloy interdiffusion coefficient undoubtedly plays a vital role in the early stages of oxidation, determining how readily and preferentially the less noble oxidizable element can be supplied to the alloy–oxide interface. Giggins and Pettit [22] have demonstrated that the development of Cr 2O 3 on Ni-Cr alloy grain boundaries occurs more readily and rapidly than over the bulk areas of the grains. The higher diffusion rate of Cr along the grain boundaries permits an easy transition from internal oxidation to external scale formation with respect to Cr 2O 3 formation at these locations. Thus, finer grain alloys have been found to approach the steady-state configuration at a faster rate. Further, smaller alloy grain size

310

Chapter 6

may promote an irregular alloy–oxide interface, which is desirable for better scale adherence to the substrate alloy. Such corrogated interface provides better keying action of the scale to the underlying alloy substrate. Similarly, cold working of an alloy can promote the establishment and maintenance of a protective oxide layer, not always produced immediately on annealed alloy, by increasing the alloy interdiffusion coefficient. But it may also cause spalling of this layer due to the remaining residual stresses in the alloy. Reduction of oxygen potential is conducive to early development of a protective scale and minimization of internal oxidation. This is because of the reduced oxygen flux into the alloy without affecting the flux of the less noble constituent of the alloy to reach the alloy surface. Thus, there is a tendency to precipitate a compact, more prohibitive layer on the alloy surface rather than form an array of internal oxide particles. Addition of a second alloying element to binary alloys that can form an oxide intermediate stability between those of the eventual protective oxide and the noble metal sometimes plays an overriding role in gettering oxygen and helping the main reactive alloying element to develop a protective layer. The oxidation behavior of Cu-Zn-Al is one such classical example [23]. A Cu-30wt%Zn solid solution alloy on oxidation forms a relatively slow-growing ZnO scale. A binary Cu-5wt%Al alloy suffers from internal oxidation of Al and rapid scaling. However, addition of 2–4wt%Al to Cu-Zn alloy can substantially cut down its oxidation rate as depicted in Fig. 6.11, where protection is provided by the healing layer of Al 2O 3 formed immediately beneath the thin initially nucleated scale. When the surface of the bare ternary alloy is exposed to an oxidizing environment, Cu 2O, ZnO, and Al 2O 3 are nucleated at the beginning of oxidation. In view of depletion of Al in the alloy next to the surface, Al atoms migrate to the surface without being converted to Al 2O 3 on their way to the surface. Hence, there is sufficient supply of Al atoms with which oxygen can react preferentially in comparision to Cu and Zn in view of higher negative standard free energy of formation of Al 2O 3. Moreover, Al atoms diffusing toward the surface convert the initially formed nuclei of Cu 2O and ZnO to Al 2O 3 by virtue of displacement reactions, such as 2Al ⫹ 3Cu 2O ⫽ 6Cu ⫹ Al 2O 3

(6.30a)

and 2Al ⫹ 3ZnO ⫽ 3Zn ⫹ Al 2O 3

(6.30b)

Thus, the scale is supposed to consist exclusively of Al 2O 3 as depicted in Fig. 6.11c. Although Zn atoms do not enter the steady-state scale but here Zn acts as a gettering element, reducing oxygen flux into the alloy interior and allowing Al to diffuse up to the surface in forming the protective layer. Basically the lowering

Alloy Oxidation

311

Figure 6.11 Oxidation behavior of Cu-Zn-Al alloys [23]. (a) Oxygen uptake of Cu30%Zn alloy with 0–2 wt% Al after 3 h of oxidation at 1123 K. (b) ZnO surface film formation to prevent internal oxidation of Al during transient stage. (c) Al 2O 3 protective layer formation during the steady-state oxidation.

of oxygen solubility due to the presence of Zn has the same effect as the lowering of the oxygen solubility in the investigation by Rapp [15] on Ag-In alloys by reducing the outer oxygen partial pressure from 1 atm to 10⫺7 atm. Similar examples are the development of Al 2O 3 scale on Fe-Cr-Al, Ni-Cr-Al, and Co-Cr-Al alloy systems whose oxidation behavior is discussed in Sec. 6.5. As much as 20– 30 wt% Al is needed to form a protective layer exclusively of Al 2O 3 on the corresponding binary Fe-Al, Ni-Al, and Co-Al alloy systems, but the alloys tend to become brittle exhibiting extremely poor deformability due to such high content of Al. However, on addition of 15 wt % Cr to these alloys, the amount of

312

Chapter 6

Al required to develop the protective layer of Al 2O 3, reduces to only 5–6 wt %. Here Cr acts as an oxygen getter, inhibiting internal oxidation.

6.5 SCALING OF BINARY AND TERNARY ALLOYS Oxidation of alloys involves the same general phenomena as described for pure metals in the preceding chapter. However, alloy oxidation is generally much more complex as a result of the following: 1.

2.

3. 4. 5. 6.

7.

Alloys in general contain two or more oxidizable constituents having different affinities for oxygen reflected by different free energies of formation of the respective oxide. The reacting metal atoms do not diffuse at the same rates either in the oxide or in the alloy phases. As a result, the oxide scales on the alloys will not contain the same relative amounts of the alloy constituents as does the alloy phase. A degree of solid solubility between the oxides may exist. Ternary and higher oxides are likely to be formed. Dissolution of oxygen into the alloy may result in subsurface precipitation of oxides of one (or more) alloying element (internal oxidation). The composition and structure of the oxide scales on alloys often change as oxidation proceeds. The oxidation kinetics, in turn, often markedly deviates from the ideal simple rate equations. The scales can also crack, spall, develop voids, sinter, and produce multiple irregular layers.

Accordingly, there can be no unified theory of alloy oxidation, only a set of special cases can be treated to varying extents in a fundamental and quantitative way. Pure metals have poor engineering properties and are rarely used as construction materials. Therefore, oxidation of alloys, particularly the methods to achieve improved oxidation resistance, are in a practical sense the most important aspects of high-temperature oxidation of metals. Because of technological importance, numerous experimental studies on the oxidation behavior of different alloys are reported in literature. Since the field of alloy oxidation is vast, no attempt is made here to give a complete survey of the extensive literature on the topic; rather, examples to illustrate the important fundamentals are presented.

6.5.1

Preliminary Classification

The oxidation of alloys under normal conditions leads to formation of an external scale. However, for many alloys, depending on reaction conditions, simultaneous occurrence of internal oxidation also takes place. The composition and micro-

Alloy Oxidation

313

structure of the resultant scale are primarily governed by the alloy composition and the reaction mechanism under the prevailing conditions. Since the alloy oxidation behavior can be extremely complex, it is necessary to break down the field into various limiting cases, applicable specifically to binary alloys but in a more general way to commercial alloys, which can be treated quantitatively or at least semiquantitatively. Let us consider the oxidation of a single-phase alloy, A-B, in a single oxidant system, where B is the less noble metal. For simplicity it is assumed that the two metals form only one oxide of each, i.e., AO and BO. So for steady-state scale growth on alloy A-B, the following cases may arise: 1. Compositions near pure A, where AO is produced almost exclusively, at least in the external scale 2. At sufficiently high concentrations of B, where BO is formed exclusively 3. At intermediate composition range, where both AO and BO are formed In the intermediate composition range (3), depending on the properties of AO and BO, the two oxides may be completely miscible, producing an oxide solid solution; or they may be completely immiscible, producing multiphase scales; or they may combine to form another complex oxide compound, e.g., ABO 2. In addition, internal oxidation may also take place. For ternary and multicomponent alloys, the degree of complexity increases further. The schematic classification of the scale morphologies according to distribution of phases in the scale, obtained by oxidizing or sulfidizing alloys, was presented for the first time by Moreau and Be´nard [24]. Subsequently, this was adopted by Wood [25] and Wallwork [26]. Dalvi et al. [27] attempted to rationalize the complex processes that occur during the oxidation of binary alloys utilizing ternary diffusion theory, in particular the concept of a diffusion path on a ternary phase diagram, like A-B-O or A-B-S. Such analysis provides scope not only for theoretical predictions of the alloying elements likely to be preferentially oxidized but also the expected steady-state scale. They can also supplement information on subsequent reactions, such as the changes occurring when the internal oxide is incorporated in the main scale. Such theoretical analysis is based on the assumptions that a local thermodynamic equilibrium prevails at the alloy–scale and scale–gas interfaces and that the oxide scales are free of cracks and pores, and adherent to the alloy substrate. More recently, Bastow et al. [28], by considering the elemental distributions through the scales, presented a more comprehensive classification of the various scale morphologies for the oxidation and sulfidation of binary alloys, complementary to that of the diffusion path approach of Dalvi et al. [27]. The application of thermodynamic concepts, though extremely useful, is used sparingly in the present discussion; more emphasis is placed on factors affecting the spatial distribution of components in the scale and alloy,

314

Chapter 6

and their influence on the overall scaling behavior of single-phase binary and ternary alloys. Before proceeding to a discussion of the steady-state scaling behavior of different alloy systems, it is reasonable to enumerate the following pertinent parameters [21] determining the establishment of a steady-state scale: 1.

2.

3.

4.

5.

6. 7.

The standard free energies of formation of the different simple, doped, and complex oxides predict the occurrence of thermodynamically favored oxide. This factor is not always predominant initially, nor does it provide a guarantee for the formation of a complete layer of the favored oxide. The bulk alloy composition is of prime importance in deciding whether the favored oxide will be formed in sufficient quantity to produce a complete, often protective layer or merely appears as nonprotective internal oxide and finally as a precipitate incorporated in the main scale. The alloy interdiffusion coefficient determines how quickly the preferentially oxidizing alloy constituent can be replenished at the growing surface layer. The rapid development of a protective oxide layer is facilitated by a high alloy interdiffusion coefficient. The solubility and diffusivity of atomic oxygen in the alloy together with the alloy interdiffusion coefficient determine the internal oxidation characteristics. High values of solubility and diffusivity of atomic oxygen promote internal oxidation, and high alloy interdiffusion coefficient may permit easy and early transition from internal to external scale formation in achieving protective scaling behavior. The growth rates of base metal oxides decide the relative development, overgrowth, and undercutting rates of the initial nuclei and the subsequent scale. Rapid growth of the base metal oxide can absorb the internal oxide particles within it before the oxide particles can coalesce to form a complete protective layer. The microstructure of the alloy may also exert influences on factors such as epitaxial relationship, protective layer development, and so forth. The oxidizing conditions, including temperature and partial pressure of oxygen, as well as the procedure adopted for elevating the alloy to the steady temperature, can influence the scaling behavior.

The subject of alloy oxidation is vast because there are numerous alloys of varying compositions and their corresponding oxidation behaviors are also different. Accordingly, it is reasonable to break the field down into various limiting cases. The following simplified classification, based on the work of Wood [25], is presented schematically in Fig 6.12. In such an exercise, the entire scene of alloy oxidation is broadly subdivided into two classes, with class 1 representing the special case of the oxidation of one alloy component and class 2 representing the more general case where both components of the binary alloy oxidize.

Alloy Oxidation

315

Class 1 Only one of two elements constituting the binary alloy oxidizes under the prevailing conditions, producing BO. (A) The reactive element, B (in smaller concentration), oxidizes under the given conditions: 1. Internally oxidized BO particles are produced in a matrix of A. For example, oxidation of dilute Ag-Si alloys in an oxygen atmosphere less than the equilibrium dissociation pressure of Ag 2O, produces SiO 2 particles in a Ag matrix (Fig 6.12a). 2. Exclusively a single layer of BO is externally formed over the alloy matrix depleted in B. For example, oxidation of Ag-Si alloys richer in Si at an oxygen pressure less than the dissociation pressure of Ag 2O produces an external scale of SiO 2 (Fig 6.12b). However, when both the components of the binary alloy have possibilities to oxidize but the kinetics and geometrical conditions permit B to be selectively oxidized, as in the case of chromiumrich Fe-Cr, Ni-Cr, and Co-Cr alloys, essentially an external scale of Cr 2O 3 is developed, particularly at low oxygen pressure. In reality, a small amount of Fe, Ni, or Co does enter the Cr 2O 3 layer, producing solid solutions or doping effects. (B) The reactive element, B (in major proportion), oxidizes exclusively: 1. The nonoxidizable metal A gets dispersed in BO, as found in Cu-Au alloys rich in Cu (Fig. 6.12c). 2. An exclusive external scale of BO is formed leaving underneath the nonoxidizable metal A in an enriched zone of the unreacted alloy as found in NiPt alloys (Fig. 6.12d). Class 2 Both the elements (A and B) oxidize simultaneously at an oxygen pressure greater than the dissociation pressures of both the oxides to produce AO and BO. (A) AO and BO either are miscible (in the whole range or partially) to each other or react chemically to produce a complex compound. 1. AO and BO are completely miscible to produce a single solid solution of (A,B)O in the external scale, as in the case of Ni-Co alloys (Fig. 6.12e). However, in reality some internal oxide of (A,B)O richer in B than the surface scale may also be present. 2. The two individual oxides combine chemically, often to form a spinel (ABO 2). The particles are incorporated in a matrix of AO if the reaction is incomplete, as in the case of some Ni-Cr alloys (Fig. 6.12f).

316

Chapter 6

Alloy Oxidation

317

(B) AO and BO are virtually insoluble in each other. 1. Internal oxide particles of the reactive element B (solute) lie beneath a twophase outer layer of AO and BO (Fig. 6.12g), as found in the case of CuNi, Cu-Zn, Cu-Al, Cu-Be, etc., alloys. In such alloy systems, conditions may develop such that the internal oxide particles may link up to form a complete healing layer of BO at the scale base. 2. When the reactive element is the solvent in the alloy system, no internal oxidation can take place (Fig. 6.12h). In practice, the second phase, as depicted in Fig. 6.12g, may not be present in the outer regions of the scale if AO grows rapidly to produce an external scale exclusively of it. However, the outer regions of such a scale may be oxidized to higher oxides. For example, a thin outer layer of CuO is observed over Cu 2O during oxidation of Cu-Si alloys. It is therefore apparent that there cannot be a single comprehensive theory of alloy oxidation. It is to be recognized that in certain alloy systems several classifications of behavior are possible, depending on the alloy composition, oxidizing atmosphere, temperature, time of exposure, and so on. Changes from one type of behavior to the other may occur even for a single alloy.

6.5.2

Simultaneous Formation of Oxides of Both Components in the External Scale

At concentrations lower than that at which B becomes selectively oxidized, the external scale consists of oxides of both A and B, e.g., AO and BO. The various limiting cases of scale growth under such situations are considered below.

Figure 6.12 Schematic presentation of the different modes of oxidation of alloy AB with variable composition, where B is the less noble metal. (a) Solute B only oxidizes, producing internal oxide BO in the matrix of A. (b) Solute B only oxidizes, producing external scale of BO above the alloy depleted in B. (c) Solvent B only oxidizes, producing a matrix of BO in which A is distributed. (d) Solvent B only oxidizes, producing an external scale of BO above the alloy depleted in B. (e) A and B oxidize simultaneously, producing an external scale of a solid solution or compound of variable composition. (f ) A and B (solute) oxidize simultaneously, producing a compound ABO 2 dispersed in a matrix of AO as the external scale with internal oxide particles of BO at the alloy–oxide interface. (g) A and B (solute) oxidize simultaneously to give insoluble oxides with BO in a matrix of AO as the external scale and internal oxide particles of BO at the alloy– oxide interface. (h) A and B (solvent) oxidize simultaneously to give insoluble oxides, with AO in a matrix of BO as the external scale [25].

318

Chapter 6

Formation of Mutually Insoluble Oxides For some alloy systems, the oxides, AO and BO, may for all practical purposes be considered immiscible. However, under equilibrium conditions, the oxides will be mutually doped with the other cation. Classic examples of such binary alloy systems are Cu-Zn (forming Cu 2O/CuO and ZnO), Cu-Ni (forming Cu 2O/ CuO and NiO), Cu-Be (forming Cu 2O/CuO and BeO), etc. As an illustration, the behavior of Cu-Be alloys is discussed. The scale morphologies developed for different concentrations of Be and the oxidation mechanism are presented schematically in Fig. 6.13. Maak [29] studied the oxidation behavior of a number of Cu-Be alloys in air with 0.08–12.6 at. % Be at 1123 K. Alloys containing up to 6.7 at. % Be were found to oxidize at the same rate as pure copper. The oxide

(a)

(b)

Figure 6.13 (a) Schematic presentation of the scale morphologies developed in different Cu-Be alloys oxidized in air at 1123 K. (b) Schematic illustration of oxidation mechanism in Cu-Be alloys [2].

Alloy Oxidation

319

scale consisted of a thin outer layer of CuO over Cu 2O and an inner two-phase layer of Cu 2O ⫹ BeO. Internally oxidized BeO crystallites in the copper matrix were also detected beneath the scale–metal interface (Fig. 6.13a). The mutual solubilities of the two oxides (Cu 2O and BeO) are known to be almost nil, and the oxides do not react to form a ternary oxide as well. At a concentration of 12.6 at. % Be, the oxidation resulted in an external scale formation exclusively of BeO, where the rate was approximately 104 times slower than that for Cu-Be alloys with less than 7 at. % Be. The critical concentration of Be required for the transition from internal oxidation to external scale formation is theoretically calculated to be only 1.8 at. % against 12.6 at. % as experimentally observed. The discrepancy between the experimental value and the predicted value was attributed to additional factors such as (1) internal oxidation of Be in the alloy phase and (2) development of porosity in the Cu 2O layer—which were not considered in the ideal model. Pores are expected to develop when cations migrate outwardly through the scale. Even though the pores act as solid-state diffusion barriers to copper ion migration, the oxidation is sustained by dissociative transport of oxygen across the pores, as schematically illustrated in the reaction mechanism (Fig. 6.13b). Formation of Oxide Solid Solution Having Complete Solubility of Oxides When both components of the binary alloy (A-B) are oxidized and the resulting oxides are completely miscible in each other, both appear in the external scale in a proportion determined by the oxidation potentials of A and B as well as by their diffusion rates in the alloy and scale. Binary alloys of Co, Fe, Mn, and Ni oxidized under most conditions of temperature and oxygen pressure fall into this category. Their oxides, e.g., CoO, FeO, MnO, and NiO, all have a simple cubic, NaCl-like structure and form solid solutions over their entire composition range. Since the two cations involved in the reaction do not diffuse at the same rate through the scale, different concentration gradients for the ions develop in the scale. The situation is best illustrated by considering the oxidation of Ni-10.9%Co alloy in 1 atm O 2 at 1273 K [30]. The alloy was found to oxidize at a relatively uniform rate following parabolic growth kinetics, yielding a scale similar to that on pure Ni or Co except for the occurrence of internal oxidation in a limited way. Ni and Co produce a continuous range of solid solutions and their resultant oxides, NiO and CoO are also mutually soluble, producing (Ni,Co)O solid solution of variable composition. The concentration profiles of Ni and Co as determined through microprobe analysis in the unoxidized alloy together with the resultant scale are presented in Fig. 6.14. It is revealed that Co is enriched at the oxide–oxygen interface. Concurrently, the alloy beneath the scale is depleted in Co but enriched in Ni. This is possibly because Co has a higher affinity for oxygen than Ni and so is enriched in the steady-state scale. Enrichment in Co

320

Chapter 6

Figure 6.14 Concentration profiles of nickel and cobalt through alloy and scale for Ni-10.9%Co alloy oxidized in 1 atm O 2 at 1273 K for 24 h [30].

could also be due to that fact that it diffuses more quickly than Ni in the oxide solid solution. However, Co and Ni contents in the scale at the two reaction interfaces had remained essentially constant with increasing oxidation time. Partial Miscibility of Component Oxides and Formation of Compound Oxides Studies relating to oxidation behavior of Fe-, Ni-, and Co-based alloys have received much attention in relation to the development of heat-resistant alloys, where chromium is chosen as an important alloying element to provide hightemperature corrosion resistance by the formation of chromia scales. Accordingly, Fe-Cr, Ni-Cr, and Co-Cr alloy systems serve as the base alloys for many commercial high-temperature alloys. It is pertinent that in such binary alloy systems there may be composition ranges in which noble metal oxide and reactive metal oxide predominate but their oxides are partially miscible to form singlephase solid solution scales, and the respective oxides also react to form a new, distinct oxide phase [spinels of the type NiCr 2O 4, CoCr 2O 4, or FeFe 2⫺xCrxO 4 (0 ⱕ x ⱕ 2)]. Thus, the monoxides of Fe, Ni, and Co (MO) react with Cr 2O 3 to form spinels of composition MCr 2O 4. The main features of oxidation of these three groups of alloys are similar, and as an illustration Co-Cr-O [31] is discussed. At high temperatures, the thermodynamically stable phases that can exist in a Co-Cr-O system are CoO, Co 3O 4 (stable at ⬍ 1243 K in 1 atm O 2), Cr 2O 3, and CoCr 2O 4, of which Cr 2O 3 is the most stable one formed by selective oxidation. However, if Cr 2O 3 is to be formed externally as a protective layer, there has to

Alloy Oxidation

321

be a continuous supply of Cr from the inner part of the alloy to the alloy–oxide phase boundary, which means that both the concentration and diffusivity of Cr in the alloy are sufficiently high that it is in ample supply at the alloy–oxide reaction interface. If this condition is not fulfilled, the other component (more noble) of the alloy will be oxidized and the reaction mechanism becomes a function of several factors, including composition of the alloy and rates of transport in the alloy phase as well as in the oxide scale. As a clean Co-Cr alloy surface is exposed to an oxidizing atmosphere at high temperatures, both Co and Cr oxidize to form CoO (and Co 3O 4 at less than 1243 K in 1 atm O 2) and Cr 2O 3 as a thin film, which thickens by solid-state diffusion. Thereafter several competing processes take place at or near the interface between CoO and the alloy during the transient oxidation stage (Fig 6.15a): (1) Co continues to get dissolved in CoO and diffuses outwardly through CoO to react with oxygen at the oxide–oxygen interface; (2) Cr in the alloy also diffuses to the alloy–oxide interface and reduces CoO to metallic Co while it itself gets oxidized to Cr 2O 3; (3) the released O 2 from CoO dissolves and diffuses into the alloy where it reacts with Cr to form internal oxide particles of Cr 2O 3 beneath the scale–alloy interface. The establishment of a protective continuous Cr 2O 3 layer depends on factors like (1) the concentration and diffusivity of Cr in the alloy, (2) the diffusivity of oxygen in the alloy and (3) the growth rate of CoO layer. This suggests that only under certain specific conditions a protective continuous Cr 2O 3 scale can be grown, at and above a critical concentration of Cr in the alloy. This concentration in turn is a function of exposure temperature and partial pressure of oxygen. Under steady-state conditions, at lower concentration of Cr (less than 30%), a continuous layer of Cr 2O 3 fails to develop. The external scale consists of an outer CoO layer (growth rate of CoO is faster than that of Cr 2O 3) with an inner layer of CoO ⫹ CoCr 2O 4 (formed by solid-state reaction of CoO and Cr 2O 3), and internally oxidized Cr 2O 3 particles are dispersed in the alloy beneath the external scale (Fig. 6.15b). Since the overall oxidation process is governed by solid-state outward diffusion of Co2⫹ through CoO of inner and outer layers, the inner layer in particular develops voids and closed porosity. Inward dissociative transport of oxygen taking place across the voids allows the inner external layer to grow into the alloy, thereby oxidizing metallic Co to CoO in the internally oxidized zone. Since CoO grows inwardly, the internally oxidized Cr 2O 3 particles get embedded in CoO and finally react to form CoCr 2O 4 spinel. So one has to realize that for dilute Co-Cr alloys (Cr less than 10%), the increase in oxidation rate with increasing Cr concentration is partly due to the doping effect contributed by dissolution of Cr in CoO, which increases the vacancy concentration of the oxide and in turn enhances the diffusion of Co2⫹ in CoO (the native defect concentration in CoO is 1% in 1 atm O 2 at 1273–1473 K, and the solubility of Cr in CoO is 2.6% at 1473 K [30]). However, the effect is primarily believed to be

322

Chapter 6

Figure 6.15 (a) Schematic illustration of processes taking place during transient oxidation stage of Co-Cr alloy at 1273–1473 K in 10 torr to 1 atm O 2. (b) Steady-state scaling of Co-Cr alloys with insufficient Cr (⬍30%) showing a two-layered external scale and internal oxidation of Cr. (c) Development of a continuous layer of Cr 2O 3 on Co-Cr alloys with more than 30% Cr [32].

due to formation of voids and porosities in the inner part of the external scale, allowing rapid gaseous transport across the pores that overrides the effect of diffusional blockage by CoCr 2O 4-spinel particles in the inner layer. On increasing the concentration of Cr in the Co-Cr alloys (Cr greater than 30%), an increasingly larger fraction of the inner layer consists of CoCr 2O 4 spinel. Diffusion through the spinel being slow in comparison with CoO phase, the available area in the scale for easy diffusion continues to decrease. As a consequence, the rate of oxidation gradually decreases with increasing Cr concentration in the alloys until a critical concentration is reached where a continuous blocking layer of Cr 2O 3 is formed (Fig. 6.15c). The dependence of parabolic rate constants on Cr concentration for oxidation of Co-Cr alloys at 1373 K is illustrated in Fig. 6.16.

6.5.3

Comparison of the Behavior of Fe-Cr, Ni-Cr, and Co-Cr Alloys

The possible oxide phases that may be formed during oxidation of binary Fe-Cr, Ni-Cr, and Co-Cr alloys are: 1. 2.

Fe-Cr alloys—FeO, FeFe 2-xCrxO 4 (0 ⱕ x ⱕ 2), Fe 2O 3, and Fe 3O 4, of which the last two form a complete range of solid solutions Ni-Cr alloys—NiO, NiCr 2O 4, and Cr 2O 3

Alloy Oxidation

323

Figure 6.16 The parabolic rate constant for oxidation of pure Cr and Co-Cr alloys at 1373 K as a function of Cr content [32].

3. Co-Cr alloys—CoO, Co 3O 4 (at T ⬍ 1243 K in 1 atm O 2), CoCr 2O 4, and Cr 2O 3 It is possible for FeO, NiO, and CoO to dissolve a low percentage of Cr (1.5– 2% in FeO at 1273 K, 1.45% in NiO at 1273 K, and 2.6% in CoO at 1473 K [30]) to form solid solutions and similarly for Cr 2O 3 to dissolve some Fe, Ni, or Co (doped oxide). The native cation vacancy levels in pure NiO, CoO, and FeO are approximately 0.1, 1, and 10%, respectively in 1 atm O 2 at 1273–1473 K [30]. It is expected that doping at the concentration levels possible would be more effective in the more nearly stoichiometric oxides; its importance in increasing the cation vacancy level should decrease in the order NiO ⬎ CoO ⬎ FeO. The likelihood of a cation vacancy concentration gradient also lies in this order. When Fe-Cr, Ni-Cr, and Co-Cr alloys are oxidized, the different oxides as mentioned above are produced on the respective alloy in various proportions and locations and at different exposure timings, depending on conditions such as alloy composition, impurities, atmosphere, temperature, alloy surface geometry, surface finish, etc. Among the three binary alloy systems, Fe-Cr alloys are ferritic (BCC), whereas Ni-Cr and Co-Cr alloys are austenitic (FCC). Accordingly, it is

324

Chapter 6

expected that Ni-Cr alloys would behave in a similar way to that as discussed for Co-Cr alloys (in Sec. 6.5.2). However, solubility, diffusivity, and alloy interdiffusion coefficient are also important factors; consequently, the scale growth pattern and oxidation mechanism, even between the two types of FCC alloys, may not be exact. The major differences in the oxidation behavior between these three types of binary alloy system [21,25,30,31,32,33] are as follows: 1.

2.

3.

a.

b.

4.

Fe-Cr and Ni-Cr alloys display a minimum in their oxidation rates at about the 20% Cr level, whereas Co-Cr alloys must contain more than 30% Cr to exhibit a similar effect. For dilute alloys (Cr-5–10%), addition of Cr increases the oxidation rates in all three alloy systems. However, the rate for Fe-Cr alloys is several times faster than that for Co-Cr and Ni-Cr alloys when no protective healing layer of Cr 2O 3 can form. A continuous layer of doped Cr 2O 3 is more readily formed on Fe-Cr alloys (Cr ⬎ 15–30%) than on the corresponding Ni-Cr alloys, in 1 atm O 2 at 1073– 1473 K. This is basically due to the following [21]: The alloy interdiffusion coefficient for FCC. Ni-Cr (for Ni-20% Cr, 1.1– 2.8 ⫻ 10⫺11 cm2 /s at 1273 K) is an order of magnitude lower than that for BCC Fe-Cr alloys (for Fe–20% Cr, 6 ⫻ 10⫺10 cm2 /s at 1273 K [20]), whereas the solubility and diffusivity of atomic oxygen is higher in Ni-Cr alloys. Therefore, Cr 2O 3 tends to precipitate as densely populated internal oxide particles in Ni-Cr alloys, whereas in Fe-Cr alloys they tend to coalesce producing a continuous layer near the alloy surface. The wide composition range of FeFe 2-xCrxO 4 (0 ⱕ x ⱕ 2), the spinel phase in Fe-Cr alloys, permits it to form as a relatively extensive phase. This can prevent the rapidly growing FeO (with very high native defects) from overgrowth and undercutting the Cr 2O 3 nuclei. Such favorable influence is less in Ni-Cr alloys where NiCr 2O 4 is of relatively fixed composition rich in Cr, less extensive and mostly in particulate form. Fortunately, NiO, unlike FeO, does grow at a relatively slow rate for which overgrowth and undercutting by this oxide are also slow, and eventually a continuous layer of Cr 2O 3 is formed. In Cr-rich alloys of all the three systems, the main protective oxide is doped Cr 2O 3 and not the spinels. Doped Cr 2O 3 is less readily established on Ni-Cr alloys as initial scale, but it is more readily retained because of subscale keying, which is never the case with Fe-Cr alloys. For Fe-Cr alloys, chromia is the sole oxide except for a little dissolved iron as Fe 2O 3. However, for Ni-Cr alloys, there is a considerable outer layer of NiO containing dissolved Cr. In contrast, Co-Cr alloys form an external continuous Cr 2O 3 scale with dissolved Co or with CoCr 2O 4.

Alloy Oxidation

325

5. Protective scale failure that occurs by lifting and cracking is less prevalent in Ni-Cr alloys than in Fe-Cr alloys. This is partly due to keying action on the scale by internally oxidized Cr 2O 3. Subscale and associated internal oxide particle formation within the alloy are never observed behind the protective Cr 2O 3 scale on Fe-Cr and Co-Cr alloys. 6. A basic difference between the doped Cr 2O 3 scales growing on Fe-Cr and Ni-Cr alloys is their greater adhesion on the latter. This is related to the more convoluted alloy–scale interface and intergranular penetration of fingers of Cr 2O 3 into the Ni-Cr alloy base. Such convolutions are produced by the interplay of stress development in the oxide and substrate and by its relief through plastic deformation. An oxide scale of uniform thickness is stable only if diffusion of the less noble metal in the alloy phase is relatively fast compared to its diffusion in the oxide phase. Consequently, the greater the ratio of the oxidation rate constant to the alloy interdiffusion coefficient, the more convoluted and irregular the alloy–oxide interface is likely to be. This ratio is higher for Ni-Cr than for Fe-Cr alloys in the appropriate composition range. It is therefore significant to note that although an increase in alloy interdiffusion coefficient usually promotes the rapid development of a protective surface scale, if it is too large it may lead to generation of a flat alloy–oxide interface, which is detrimental to scale adhesion, at least on cooling. 7. Break-away oxidation, in which there is a dramatic increase in the oxidation rate after an induction period, follows from the lifting and cracking open of the scale on alloys that are eventually somewhat more dilute in Cr and following the exposure of the Cr-depleted substrates to the hot gas. Such breakaway occurrences are less evident on Ni-Cr alloys than on Fe-Cr alloys, at least during isothermal exposures. Potentially, break-away is more likely with Ni-Cr alloys because the Cr depletion at the alloy–oxide interface is greater. However, the overgrowth and undercutting rates of NiO are so low in comparison to FeO that oxidation of Ni-Cr alloys is never catastrophic and a protective Cr 2O 3 layer is readily reestablished. This occurs by coalescence of densely precipitated internally oxidized Cr 2O 3 particles yielding a healing layer at the scale base. 8. It is difficult to form a protective layer of Cr 2O 3 on Co-Cr alloys with Cr content less than 30%, other than at low oxygen partial pressures or after preferential oxidation treatments. This is due to the interplay of the different governing parameters, which is least favorable for this alloy system. Although CoCr 2O 4, spinel phase has a modest composition range but is generally insufficient to produce a complete layer and prevent the rapidly growing CoO from overgrowing and undercutting the Cr 2O 3 nuclei. The alloy interdiffusion coefficient for FCC Co-Cr alloys (Co–40% Cr, 6 ⫻ 10⫺12 cm2 /s at 1273 K [20]) is an order of magnitude lower than for Ni-Cr alloys and two orders of magnitude lower than that of ferritic Fe-Cr alloys. Co-Cr alloys

326

Chapter 6 probably behave partly like Fe-Cr and partly like Ni-Cr alloys in situations where doped Cr 2O 3 can be developed on them.

6.5.4

Oxidation Behavior of Fe-Cr-Al, Ni-Cr-Al, and Co-Cr-Al Alloys

The main principles already described for oxidation of binary alloys also apply to the ternary alloy systems, at least in simple atmospheres. It is to be envisaged that the reduction in oxidation rate and improvement in scale adhesion, with suitable alloying element additions, and within the restrictions of desired mechanical properties and cost, can provide one of the most promising ways for the development of high-temperature materials without resorting to the use of coatings. Most alloys/alloy coatings for high-temperature applications are based on iron, nickel, and/or cobalt and rely on the establishment of chromia (Cr 2O 3), alumina (Al 2O 3), or silica (SiO 2) healing layers for protection against oxidation. These oxides are thermodynamically very stable, have high melting points, and the transport processes through such scales are generally slow. For many hightemperatures applications, Cr 2O 3 scales can be effective on binary M-Cr alloys (where M is Fe, Ni, or Co). However, at temperatures above 1173 K, CrO 3 may further react with oxygen to form its higher oxide, such as CrO 3 (a volatile species), limiting the long-term application of such alloys to temperatures below about 1173 K. Accordingly, for applications in highly oxidizing environments at high temperatures up to 1473 or 1573 K, alloy/alloy coatings often are designed to develop surface layers of Al 2O 3 or SiO 2 that do not form volatile oxides (formation of volatile SiO limits applications of the latter in environments of low oxygen partial pressure). Unfortunately, the concentrations of Al or Si needed to establish such scales on the respective binary alloys often result in unacceptable mechanical properties. Fortunately, in practice, addition of Cr to alumina or silica-forming binary alloys reduces the concentration of aluminum or silicon required to establish the respective Al 2O 3 or SiO 2 scale to tolerable levels, e.g., 3–4% Al in an Fe-14% Cr or an Fe-27% Cr alloy or 1–2% Si in an Fe-14% Cr alloy [34]. Si is also expected to compete with Cr to form a surface layer of SiO 2 rather than that of Cr 2O 3, but it is reported [30] to be unsuccessful with Fe-26% Cr-1% Si and Fe-29% Cr-5% Si alloys at 1273–1473 K, even under preferential oxidation conditions. This is because SiO 2 grows so slowly that the initial nuclei are overgrown by Cr 2O 3, which continues to thicken, and SiO 2 fails in coalescing to provide a protective layer, slowing down the oxidation rate below that of Fe-CrAl alloys. α-Al 2O 3 can be more readily formed and maintained on Fe-Cr-Al alloys than on binary Fe-Al alloys. This is partly because Cr tends to stabilize α-Al 2O 3 rather than the less protective γ-Al 2O 3 (stable below 1173–1223 K). The more general

Alloy Oxidation

327

concept is that Cr acts as a getter, preventing oxygen from entering the alloy and thereby promoting a complete external layer of Al 2O 3 rather than its precipitation as internal oxide [25]. Moreover, during the transient oxidation state, Cr 2O 3 and α-Al 2O 3 are more predominantly formed and supply oxygen to the alloy more slowly than do iron oxides, with their high dissociation pressures. As a consequence, Al finds more opportunity to diffuse up to the alloy–oxide interface, permitting development of an α-Al 2O 3 layer beneath the thin transient scale. It has sometimes been claimed that the initially formed iron oxides and Cr 2O 3 nuclei can be reduced by Al to produce Al 2O 3 by displacement reaction [23]. However, in reality, Cr 2O 3 is sometimes found outside the Al 2O 3 layer, where it is relatively stable being out of contact with the alloy, although it tends to interdiffuse with the Al 2O 3 to form a solid solution. For low-aluminum-containing alloys, the aluminum acts as a getter, permitting Cr to diffuse to the alloy surface for development of a surface scale, solely of Cr 2O 3 with α-Al 2O 3 as internal oxide. A detailed analysis of the development of scale morphologies during isothermal oxidation of Fe-Cr-Al, Ni-Cr-Al, and Co-Cr-Al alloys containing 10–15% or 26– 30% Cr and 0.9–1.3% or 4.3–5.7% Al, in 1 atm O 2 at 1273 and 1473 K, was carried out by Stott et al. [35]. On immediate exposure of the alloy surfaces to oxygen at high temperature, the respective alloy surfaces are covered by nuclei of MO, M 2O 3 (M represents Fe, Ni, or Co), Al 2O 3, and spinels, bringing down the oxygen potential at the alloy–oxide interfaces to a considerably low value enabling occurrence of preferential oxidation; subsequent processes are governed by alloy composition. It should be noted that the ternary alloys of Ni and Co containing at least 4% Al are of two-phase structures. The oxidation kinetics behavior and the type of scale formed on some typical alloys comprised within the three systems [35] are shown in Fig. 6.1 and briefly discussed in the following: Co-Cr-Al alloys containing 10–15% Cr and 1% Al were found to oxidize very rapidly (scale type 8 followed by type 7) following a parabolic law at both 1273 and 1473 K. On the other hand, Co-30% Cr-1% Al, even though it initially showed somewhat protective behavior (type 3), switched to break-away (type 5) during the later stages at both temperatures. A more protective behavior, but with some slight break-away, was observed even for Co-15% Cr-4.5% Al at 1273 K (types 1 and 2) and for Co-28% Cr-4.5% Al at 1273 and 1473 K (types 1 and 2 followed by type 4). Fe-12.9% Cr-1.3% Al oxidized more rapidly at 1473 K (type 1 followed by types 6 and 4) but registered a more protective behavior at 1273 K. On the other hand, Fe-26.1% Cr-0.9% Al oxidized more slowly than the former alloy (with low Cr) at 1473 K (types 2 and 3), but at a similar rate at 1273 K (types 2 and 3). The addition of 5.7% Al to Fe-11.5% Cr alloy resulted in protective behavior at 1273 K (type 1), but at higher temperature (1473 K) break-away occurred (type 1 followed by type 6). Finally, Fe-27.8% Cr-4.9% Al exhibited continuous protective behavior at both temperatures (type 1).

328

Chapter 6

In contrast, Ni-Cr-Al alloys containing 15–30% Cr and 1% Al exhibited a total protective behavior (type 2) with no sign of break-away during its exposure period. On further increase of the Al content (4–4.5% Al) with the same Cr level, the alloy showed superior protective performance at both temperatures (type 1). The different morphologies of the scales developed on the three alloy systems are presented schematically in Fig. 6.17. Comparison of the scaling behavior of the three alloy systems [21,25,30,35] highlights the following interesting trends: 1.

2.

3.

The ease of formation of a protective α-Al 2O 3 layer on the alloys and the lowest mass gain necessary to achieve this protective layer are in the order Fe-Cr-Al ⬎ Ni-Cr-Al ⬎ Co-Cr-Al. Under rapid scaling conditions, as a result of break-away of protective αAl 2O 3, Cr 2O 3, or Cr 2O 3-α-Al 2O 3 oxide layers, the most reliable scale is produced on Ni-Cr-Al. Fe-Cr-Al gives the most disastrous local break-away by locally formed nodules that grow at a considerably faster rate than on CoCr-Al alloys, even though its surface area undergoing break-away at a catastrophic rate is less. The Co-Cr-Al system yields the most generally distributed break-away over the alloy surface for which the overall mass gain is found to be greater than the corresponding Fe-Cr-Al alloys. α-Al 2O 3 is the most desirable surface scale and doped Cr 2O 3 with an inner layer of doped α-Al 2O 3 is the next most protective scale. Doped Cr 2O 3 surface scale located above α-Al 2O 3-rich internal oxide is a less desirable mode of scaling but is definitely more protective than the various nodular and fully stratified scales containing the respective oxides of Fe, Ni, or Co, which may develop during the early stages of oxidation or after mechanical breakdown of the α-Al 2O 3 or Cr 2O 3 scales.

So it is clearly revealed that at certain critical concentrations of the alloy constituents the reaction during steady-state changes from one mechanism to the other. For a particular alloy system, with the help of available data, one can construct an ‘‘oxide map’’ in which the composition ranges of the alloys are delineated for the formation of different types of oxide scale and the reaction behavior. An example of such oxide map for Ni-Co-Al alloys at 1273 K [32] is illustrated in Fig. 6.18. In region 1, the scale consists of NiO, NiCr 2O 4, and NiAl 2O 4 spinels with internally oxidized particles in the alloy; region 2 represents Cr 2O 3 as the external scale, and in addition Al is internally oxidized; region 3 represents selective oxidation of Al producing α-Al 2O 3 scale. There exists a great challenge for future investigators to derive such maps for different alloy systems with the help of more basic data involving the interplay of thermodynamics, diffusion data, and interfacial reactions in predicting the expected scale morphology utilizing the diffusion path approach of Dalvi et al. [27].

Alloy Oxidation

329

Figure 6.17 Schematic presentation of the possible scale morphologies that can be produced on different Fe-Cr-Al, Ni-Cr-Al, and Co-Cr-Al alloys at 1273 and 1473 K in 1 atm O 2 [35].

330

Chapter 6

Figure 6.18 Oxide map for alloys in Ni-Cr-Al system at 1273 K delineating the composition ranges for the formation of different types of oxide scales [32].

6.6 REACTIVE ELEMENT EFFECTS ON THE OXIDATION BEHAVIOR OF ALLOYS The alloys designed to withstand high-temperature degradation by oxidation during their service must meet two essential requirements. Primarily they must form a surface compound (oxide, sulfide, halide, etc.) that thickens only at a slow rate, and secondly the respective compound layer must remain adherent to the alloy substrate under service conditions. It is also well known that all of the technologically important high-temperature alloys used in an oxidizing environment (oxygen as oxidant) receive protection from degradation by the formation and maintenance of a thermodynamically stable, coherent, crack-free, well-adherent, oxide scale of low diffusivity, such as chromia (Cr 2O 3) or alumina (Al 2O 3). Cr 2O 3forming alloys generally contain less than 2–3% Al and 15% Cr or more depending on the base alloy. On the contrary, Al 2O 3-forming alloys usually contain more than 5% Al with a substantial amount of chromium. On the basis of low volatility, relative chemical inertness, and slow-growth characteristics, Al 2O 3 is often the scale of choice. Although Cr 2O 3 and Al 2O 3 layers are effective barriers to the transport of both oxygen and metal ions, they are susceptible to failure by loss of adhesion, particularly spallation on thermal cyclings. However, continued resistance requires the maintenance of the protective barrier separating the environment from the alloy substrate. It has already been mentioned (Sec. 5.8) that the principal sources of stress during high-temperature exposure of a metal/alloy are twofold: oxide growth stresses, which develop during the isothermal formation of the scale, and thermal stresses, which develop due to the differences in thermal

Alloy Oxidation

331

expansion or contraction between the substrate and the scale. Factors influencing the development of stresses during isothermal exposure of the metal/alloy have already been discussed in detail (Sec. 5.8.1–5.8.6). Stresses generated during cooling arise from the difference in the thermal expansion coefficients of metal and oxide. The magnitude of such stress in the oxide may be expressed as [36]: σ ox ⫽

E ox∆T(α ox ⫺ α m)

冢 冣

(6.31)

E t 1 ⫹ 2 ox ox Em tm where E α t ∆T

⫽ ⫽ ⫽ ⫽

elastic modulus coefficient of thermal expansion thickness temperature difference

and the subscripts ‘‘ox’’ and ‘‘m’’ refer to the oxide and to the metal substrate, respectively. The linear coefficients ratios of thermal expansion of the common metals and their respective oxides have earlier been presented in the Table 5.2, which clearly demonstrates that, in general, the coefficient of thermal expansion of the metal is more than that of its oxide. This suggests that compressive stresses will be developed in the oxide during cooling and that the magnitude of this stress will be proportional to the difference in thermal expansion coefficients. Since the adhesive bond between metal and oxide is generally weaker than the cohesive bonds in metal or scale, development of large thermal stresses will result in spallation of the oxide from the metal substrate. Such a situation is more detrimental to the alloys, which undergo severe depletion of the protective scale forming elements during isothermal exposure because their subsequent exposure will involve less oxidation-resistant alloy surfaces. The oxide growth stresses and induced thermal stresses are relieved in the overall alloy–scale system by such processes as 1. 2. 3. 4. 5.

Plastic deformation of the alloy substrate Plastic deformation of the oxide scale Detachment at the scale–substrate interface Cracking of the scale, and Ultimate spallation of the scale

All of these mechanisms have been observed to operate in various metal/alloy– oxidant systems. The contribution of the various stress relaxation mechanisms are dependent on such factors as (1) plastic property of the substrate alloy, (2)

332

Chapter 6

plastic property of the scale, (3) cohesive forces between the scale grains, and (4) adhesive forces between the scale and the substrate. Three methods have been used extensively to improve the scale adherence on metals and alloys. They are: 1. 2. 3.

The small addition of oxygen active elements like Y, Ce, Hf, and the rare earths to the alloys Incorporation of an oxide dispersoid like ThO 2, into the alloy, and Addition of noble metals, such as platinum.

The first two methods are widely used and they are somewhat related because during exposure of the alloy to the environment the oxygen-active elements are internally oxidized due to their high reactivity, forming highly stable oxides below the scale–alloy interface. However, the third method finds limited applications due to high cost of the noble metal.

6.6.1

Active Element/Rare Earth Additions for Oxidation Resistance

In 1937, Griffiths and Pfeil [37] observed that rare earths added as melt deoxidant to Ni-20% Cr alloys had a beneficial effect on their lifetime when used as heating elements under thermal cycling conditions. Such empirical observations subsequently prompted Pfeil [38] to demonstrate that surface coatings of the same elements, their oxides, hydroxides, or other salts, which would get converted to stable oxides at high temperatures, could also provide an alternative means to alloy additions in improving oxidation resistance of the alloys without affecting their creep resistance property. However, this ‘‘rare earth effect/reactive element effect/active element effect’’ (REE/AEE) was not confined to rare earth element additions. In fact, Pfeil [38] indicated that elements from groups II, III, IV, and V of the periodic table could be used for such effects, although their effectiveness decreased on passing from group II to group V but increased with increasing atomic mass within a particular group. From the classification as suggested by Pfeil, subsequently only some elements as depicted in Table 6.3 [39] emerged to be of practical importance. The active elements (AEs), having a higher affinity for oxygen than the base metal constituents, as referred to in this table are those that when added in small quantity improve the oxidation resistance of the alloys not only by curtailing the degradation rates but by ensuring the integrity of the protective scales, developing or developed on high-temperature alloys under isothermal or thermal cycling exposure conditions. Accordingly, the choice of the active elements is governed primarily by their properties [39] related to 1.

Free energy or enthalpy of formation of AE compounds (oxides, sulfides, oxysulfides, etc.)

Alloy Oxidation

Table 6.3

Periodic Table Indicating Elements Found to Act as ‘‘Active Elements’’

IA

IIIA

IIA

IVA

VA

VIA

VIIA

IB

VIII

H Li Na K

Be Mg Ca

Sc

Ti

V

Cr

Mn

Fe

Co

Ni

Cu

Rb Cs

Sr Ba

Y* La*

Zr Hf

Nb Ta

Mo W

Tc Re

Ru Os

Rh Ir

Pd Pt

Ag Au

Fr

La

Ac*

* Ce

Pr

* Th

Pa U

Nd

Pm

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

Np

Pu

Am

Cm

Bk

Cf

Es

Fm

Md

No

Lw

IIB

IIIB

IVB

VB

VIB

VIIB

O

Zn

B Al Ga

C Si Ge

N P As

O S Se

F Cl Br

He Ne Ar Kr

Cd Hg

In Tl

Sn Pb

Sb Bi

Te Po

I At

Xe Rn

333

334 2. 3. 4. 5. 6. 7.

Chapter 6 Pilling–Bedworth ratio (PBR) of the compounds Crystallographic structure of their compounds Solubility of active elements in the main scale constituents and in the metallic substrate Diffusivity of the active elements in the scale and alloy Valency of the active element ion and its radius The thermal conductivity of oxides, which can play a significant role in the temperature distribution and consequently on the stress generation in the scale and at the scale–alloy interface.

On the consideration of the above-mentioned properties, it may be concluded that the three most relevant active elements are Y, Ce, and La, which are characterized by 1. 2. 3.

High stability of their oxides, sulfides, and oxysulfides Very low PBR values of their oxides Their fairly low melting points, which facilitate plasticity at high temperatures.

The active elements can be incorporated into or onto the metallic substrates by any of the following ways as depicted in Fig. 6.19: 1. 2. 3. 4.

As addition of pure active elemental metal or an alloy (‘‘mischmetal’’) during alloy preparation Added/formed as an active element oxide dispersion Active elements incorporated into the alloy surface by ion implantation Application of surface coatings of the active element or its oxide.

It is important to recognize that methods (1) and (2) may influence the mechanical properties of the alloys and method (3) is yet to be commercialized, now being restricted only to laboratory tests for limited surface area to receive protection. Method (4) is the most commonly used commercial technique, which consists of applying an MCrAlY-type protective overlay coating, where M designates the alloy substrate.

6.6.2

Beneficial Effects of Additions

Despite the numerous studies devoted to REE within the last six decades, their beneficial or detrimental effects are not yet well understood and are sometimes controversial [40–52]. However, available results indicate that the major beneficial effect is a dramatic improvement in the scale adherence [39], as depicted in Fig. 6.20 and 6.21. This is more marked when the exposure temperature is high or when the conditions are severe involving thermal cyclings. Detailed investigations on Cr 2O 3- and Al 2O 3-forming alloys, reported in an exhaustive review by Whittle and Stringer [40], indicate that the addition of reactive or REEs or a distribution of their oxides within the alloy or a surface coating

Alloy Oxidation

335

Figure 6.19 Different methods to incorporate reactive elements [39].

of the elements/oxides or ion implantation has the following beneficial effects on the oxidation behavior of the alloys: 1. The growth rate of Cr 2O 3 scales is undoubtedly reduced and that of the Al 2O 3 scales may be reduced (controversial). 2. Selective oxidation of the protective scale–forming element, i.e., Al or Cr, is enhanced, particularly during the early stages of oxidation. That is, continuous protective scales of Cr 2O 3 or Al 2O 3 are achieved at lower Cr or Al contents of the alloy compared with that in addition-free alloys. 3. The transport mechanism of Cr 2O 3 growth is changed from predominantly outward cation (Cr3⫹) transport to oxygen ingress. There is no evidence to suggest a similar change with Al 2O 3-forming alloys, for which the oxygen transport, apparently via grain boundary short-circuit paths, appears to be

336

Chapter 6

Figure 6.20 Isothermal oxydation of FeNiCr alloy with and without reactive element [39].

4. 5. 6. 7.

the dominant transport mechanism in the presence of a reactive metal oxide dispersion. The transient oxidation stage is shortened and less base metal oxidation occurs. Void formation/accumulation at the alloy–Al 2O 3 or alloy–Cr 2O 3 interface is reduced. The adherence of the scale to the alloy substrate is substantially increased in both the types of alloys. The integrity of the oxide scales is improved.

Figure 6.21 Oxidation of FeNiCrAl alloy with and without reactive element under thermal cycles [39].

Alloy Oxidation

337

A wide number of reactive elements or their oxide dispersion additions produce all or most of the above-mentioned effects but to differing degrees. The main critical factor is that a potential reactive element must have a higher affinity for oxygen than the element that is to form the protective scale. The effect of adding a reactive element to an alloy is more or less the same as that of adding a stable oxide dispersion since the reactive element oxidizes internally ahead of the scale–alloy interface.

6.6.3

Proposed Mechanisms

Scale Growth Mechanisms The prominent effects of RE additions on scale growth processes for Cr 2O 3-forming alloys are illustrated in Fig. 6.22, which compares the oxidation rates for TDNiCr (Ni-20%, Cr-2vol%, ThO 2) to Ni-30% Cr and Co-35% Cr alloys at 1200°C [50]. The platinum markers were found to be situated at the alloy–scale interface for the binary alloys without any active element/oxide additions, and at the scale–gas interface for the thoria dispersed alloy. This clearly suggests that outward chromium diffusion is the important process for the binary alloys, whereas inward oxygen diffusion is important for the thoria dispersed alloy. The

Figure 6.22 Comparison of the oxidation of Ni-30%Cr, Co-35%Cr, and TDNiCr (Ni20%Cr-2vol%ThO 2) at 1200°C [50].

338

Chapter 6

enhanced selective oxidation of chromium is interpreted as being due to formation and stabilization of a fine-grained substructure with high dislocation density by the thoria particles. This substructure favors rapid short-circuit diffusion of chromium, resulting in rapid development of a protective Cr 2O 3 layer on the alloy surface. The oxide particles of active element act as nucleation sites and promote selective oxidation of the protective scale–forming element, thereby curtailing the transient stage of oxidation as schematically illustrated in Fig 6.23 for NiCr alloy. Essentially three main mechanisms have been proposed [48] to explain the reduction in scale growth rate and possible modification to the transport processes brought about by the reactive metals or their oxide additions. These are (1) doping effects, (2) formation of a partial or complete blocking layer in the scale, and (3) short-circuit diffusion model.

Figure 6.23 Schematic illustration of reactive element on scale nucleation in Ni-Cr alloy with and without reactive element oxide dispersion.

Alloy Oxidation

339

Doping Effects. The defect structure of Cr 2O 3 is not fully ascertained but the fact that Cr 2O 3 is a p-type semiconductor suggests that the oxygen defects are oxygen interstitials. Cr 2O 3 scale growth kinetics further suggests that countercurrent diffusion of Cr ions and oxygen along grain boundaries results in oxide formation within the scale. It is also reported that Cr 2O 3 is an intrinsic semiconductor above 1200°C. So, Y being trivalent, no doping effect is expected if this element is dissolved in Cr 2O 3. Substitution of Ce or Th in the Cr 2O 3 lattice, which can have valencies of 4⫹, would lead to a reduction in the concentration of interstitial Cr3⫹ ions and so to a reduction in the outward transport of chromium, leading to n-type behavior. Thus, dispersions of the reactive elements, where the metal can assume trivalent state and occupy cation vacancies, can improve oxidation resistance. The main weakness of this model lies in the apparently equal effects of a wide range of both metallic and oxide dispersion additions. Moreover, addition of Y or Y 2O 3 has often been found to affect the rate, which is not expected from this model. Formation of a Partial or Complete Blocking Layer. In this model, it is postulated that a reduction in chromium flux toward the outer interface takes place owing to the precipitation of rare earth oxides in the alloy matrix acting as a ‘‘set of sieves in series,’’ thus blocking the supply of Cr. Further, in the oxidation study [40] of TDNiCr (Ni-20%, Cr-2vol%, ThO 2) it is proposed that the incorporation of the dispersoid particles in the scale reduces its growth rate by decreasing the cross-sectional area of Cr 2O 3 available for transport of chromium. However, the elegance of this model falls down when additions of even a small quantity of reactive oxide produce significant reduction in the oxidation rates, whereas a substantial amount of reactive element oxide is required to bring about a reduction in the cross-sectional area of the scale. Short-Circuit Diffusion Model. This model suggests that the dispersoid particles on the alloy act as nucleation sites for the first formed oxides, thus decreasing the internuclei spacing. So the time required for the lateral growth process to form a complete prohibitive layer of Cr 2O 3 and terminate the formation of base metal oxides is reduced. The dispersoid particles may only block the short-circuit diffusion paths (dislocations) for Cr. The presence of dispersoid in the alloy reduces the subsequent oxide grain size, and if it is reduced below a certain level corresponding to the inter-dislocation distance in a pile-up, a single dislocation, being unstable in the grain, will move to the grain boundary where it ceases to be a cation diffusion path. Thus, if dislocations are the short-circuit diffusion paths for transport of Cr ions in Cr 2O 3, reduction in the size of the oxide grain below a certain limit may reduce the density of such paths and hence the diffusion rate. On the other hand, the oxygen diffusion rate via grain boundaries will be increased because larger grain boundary area causes a reversal of the oxide growth process. Accordingly, the oxide-forming reaction will shift from the

340

Chapter 6

oxide–oxygen interface to the oxide–metal interface as often experimentally observed. Of the three mechanisms proposed above to explain the role of active elements, the doping effect and the formation of a diffusion barrier layer lack general applicability. The most probable mechanism is the third, where it is proposed that the dispersed active element oxide particles at the alloy surface act as heterogeneous nucleation sites for the first formed oxides, thereby decreasing the internuclei spacings. As a consequence, less time will be required for subsequent lateral growth processes to link the nuclei to form a protective Cr 2O 3 scale, as illustrated schematically in Fig. 6.24. Subsequent analytical electron microscopic (AEM), energy dispersive X-ray analysis (EDX), and microdiffraction studies [43], in establishing the composition and structure of YCrO 3 particles formed by Y 2O 3-Cr 2O 3 solid solution reaction, revealed that the Y 2O 3 and YCrO 3 particles acted as the sources of Y3⫹ ions that segregated along the oxide grain boundaries in the scale formed on Ni-20% Cr-0.6% Y 2O 3 alloy, Y-implanted Fe-20% Cr-25% Ni stainless steel and Y-implanted chromium. Such observations prompted researchers to conclude that inhibition of transport along the dominant grain boundary pathways by reactive element segregants limits the oxide growth by cation diffusion. The reactive element oxide particles act as sources for the segregants in the scale. As an extension of this model, Pint [51] suggested the ‘‘dynamic segregation theory,’’ which states that all of the reactive element effects in Cr 2O 3 and Al 2O 3 scale-forming alloy

Figure 6.24 Schematic illustration of the role of reactive element oxide particles in an alloy surface acting as heterogeneous sites for Cr 2O 3 nucleation and facilitating lateral growth processes to form a protective layer.

Alloy Oxidation

341

systems can be explained in terms of the interfacial segregation of reactive metal ions to metal/alloy–scale interface and the scale boundaries. They diffuse outwardly from the metal substrate to the gas interface, and the driving force for such diffusion is the oxygen potential gradient in the scale. By diffusing outward along the scale grain boundaries, the reactive element ions inhibit outward cation transport. Similar types of mechanisms have also been suggested for Al 2O 3-forming alloys [40–42]. Even though in the literature there is a general agreement that reactive elements influence the transport processes in Al 2O 3 scales, the situation is not as clear as for Cr 2O 3 scales. There exists considerable disagreement regarding the detailed mechanism. This is mainly due to the fact that the transport mechanism in undoped alumina has always remained a subject of controversy [50]. Many researchers have suggested that the preponderant diffusing species is the oxygen, while others hold the view that Al 2O 3 scales grow either by outward diffusion of aluminum or by diffusion of both oxygen and aluminum. So the beneficial effect of active elements is assumed to be due to a decrease of oxygen or aluminum diffusion. However, there is general agreement that, in contrast to the situation of chromia scales, active elements do not reverse the growth mechanism of alumina scales, although they influence the transport processes. There also exists controversy [52] about the beneficial or detrimental effects of certain active elements like Ti, Zr, Hf, etc., and this appears to be dependent on their amount, chemical state in the scale, alloy nature, oxidation conditions (temperature, gas composition), and so forth. Sulfur is always recognized as a detrimental element, whereas the effect of yttrium on the growth kinetics of Al 2O 3 scales is marginal compared to the growth kinetics of Cr 2O 3 scales. In some cases, Y decreases (marginally) the oxidation rate of alumina-forming alloys, but an increase has also been observed (for FeCrAl alloys at T ⬎ 1373 K) [52]. Therefore, at this stage it may be inferred that the predominant mechanism by which active elements improve the oxidation resistance of Al 2O 3-forming alloys probably is not directly related to their influence on transport properties but rather associated with some effects in the improvement of scale adhesion. Mechanisms for Improved Scale Adherence The most beneficial effect of active element additions on the oxidation behavior of Cr 2O 3- and Al 2O 3-forming alloys is the considerable improvement in scale adherence to the alloy substrate. This aspect has been studied in detail, and a number of hypotheses [40,42,48,49,50,52] have been advanced to account for this. It has to be realized that this effect is irrespective of whether the alloy contains RE in its elemental form or as a dispersion of its stable oxide. Only distribution of the oxide particles may be different in the two cases. The RE oxides formed during the high-temperature exposure often exist in the vicinity of the alloy grain boundaries at intersections with the scale–alloy interface. However, in the case of dispersion containing alloys, the oxides are more randomly distrib-

342

Chapter 6

uted in the alloy matrix. Even though several mechanistic models have been postulated to explain the improvement in scale adherence, these are not mutually exclusive and a consensus view is yet to emerge. The various suggested models are discussed below. Enhanced Scale Plasticity. Many researchers believe that the scales formed on metal/alloy-containing reactive element/oxide dispersoid have higher plasticity contributing to better adhesion than those formed on metal/alloy without RE. It is well known that improvement in scale plasticity can be obtained by a finer grain structure of the oxide scale. There is no direct evidence for improvement in scale plasticity of alumina scales by active elements, though there is extensive literature that shows that the reactive elements considerably improve the scale adhesion. In most cases, presence of the reactive element produces a columnar fine-grained oxide microstructure segregated at the grain boundaries. This will definitely affect the plastic deformation behavior. Diffusional creep is the most important mechanism for plastic deformation in fine-grained polycrystalline alumina. Besides, the creep strain rate is extremely sensitive to grain size and increases with decreasing grain size. Plastic deformation will occur more easily in doped scales than in undoped ones. The growth stresses will be relieved by plastic flow during oxidation. Vacancy Sink Model. There is much clear evidence in the literature to justify that the presence of a reactive metal or its oxide dispersion in the alloy minimizes the development of voids at the alloy–scale interface. In the usual alloys, the voids arise from condensation of vacancies at the alloy–scale interface. It has been proposed that the internal oxide particles of the reactive element, the reactive element atoms themselves, or the stable oxide dispersions provide alternative sites for vacancy condensation, thus eliminating interfacial porosity. This, in turn, helps to maintain better scale–alloy contact and minimize the possibility of scale spallation. Moreover, for Cr 2O 3-forming alloys with reactive element additions, the scale growth mechanism is reversed, with the oxide-forming reaction shifting from the scale–gas interface to the scale–alloy interface, whereas Al 2O 3 scales grow primarily by oxygen transport. So the possible source of vacancies is confined to the early transient stages of oxidation when faster-growing base metal oxides, such as NiO and CoO, are being formed. Once such vacancies are formed on reactive element–free alloys, voids continue to persist. But the presence of reactive element or its stable oxide dispersion in the alloy curtails this transition period, leaving only a few vacancies. It has been reported [40] that 0.05% addition of Y or Hf to Co-10% Cr-11% Al alloy could successfully eliminate all the voids. Graded Seal Mechanism. The graded seal mechanism is based on the assumption that the compound oxide layer developed between the surface scale and the alloy possesses a thermal expansion coefficient intermediate between the scale

Alloy Oxidation

343

and the substrate alloy. However, very few alloy systems develop such a continuous intermediate scale. Mechanical Keying/Oxide Pegging. This is perhaps the most important factor for improved scale adhesion during long-term exposures and under thermal cyclings. Many authors suggest that mechanical keying/oxide pegging is a result of internal oxidation of the reactive element additions, or of oxide dispersoid particles growing in size to form oxide stringers in the form of thin elongated intrusions extending into the alloy substrate and thereby anchoring the scale to the alloy as illustrated in Fig. 6.25. Improved Chemical Bonding. The adhesion of the scale to alloy substrate is primarily governed by the atomic bonds that develop across the oxide–alloy interface. It is reported that there are certain harmful impurities, such as S and P, which are often present as minor impurities in the alloy, segregate to the scale– metal interface, and embrittle the interface affecting scale–metal adhesion. This is better known as the ‘‘sulfur effect’’ [53]. It has been proposed that reactive element (like Y) additions reduce this segregation by interacting with the sulfur and thereby improved scale adhesion is achieved. Modification of the Scale Growth Process. The addition RE or oxide dispersion to chromia forming alloys in general reduces the oxide growth rate by a factor of 10 or so at high temperatures. Such a reduction in rate is not significant (if it happens at all) with alumina-forming alloys. Any reduction in rate is associated with a decrease in the growth stresses. It has already been mentioned that in the growth of chromia scales the important major effect of the reactive elements

Figure 6.25 Schematic diagram showing the mechanical keying of oxide scale developed on Fe-26wt%Cr-4wt%Al-0.82wt%Y alloy at 1200°C in O 2 at 1 atm [39].

344

Chapter 6

is to alter the diffusional transport process from predominantly outward cation migration to predominantly inward diffusion of oxygen. It is also interesting to note that the oxides of the reactive elements, e.g., Y 2O 3, La 2O 3, CeO 2, ZrO 2, and HfO 2, are all oxides in which oxygen defects predominate, making oxygen diffusion much faster than cation diffusion. When inward oxygen diffusion predominates, oxide formation takes place at the scale–alloy interface, reducing void or cavity formation, and thereby increased scale adhesion is achieved. The wrinkles and convolutions of the scale caused by countercurrent diffusion of oxygen and chromium are also reduced. Such modifications in the scale growth process in the presence of reactive elements are illustrated schematically in Fig. 6.26. Modification of Scale Morphology and Microstructure. For Al 2O 3-forming alloys, in most cases the scales developed on alloys without reactive element additions are non-uniform in thickness, convoluted, with protrusions at the scale– gas interface and/or intrusions at the scale–alloy interface. Literature suggests that Al 2O 3 primarily grows by oxygen transport; however, the wrinkling and convolutions of alumina scales hint that oxide formation takes place within the scale in a manner similar to that of chromia scales, and accordingly accompanied by outward diffusion of aluminum also, probably along line defects in the oxide. Hence, lateral growth of oxide produces rapid development of high compressive stresses resulting in localized detachment of the scale from the underlying alloy.

Figure 6.26 Schematic representation of the chromia scales growth in (a) absence of and (b) presence of reactive element [50].

Alloy Oxidation

345

The alumina scale generally consists of two zones comprising equiaxed and columnar grains, where the former are localized to the oxide–gas interface and the latter to the inner, alloy–scale interface. Even though the presence of reactive element in alumina-forming alloys does not exactly reverse the scale growth mechanism as observed in Cr 2O 3-forming alloys, it is often proposed that the outward transport of aluminum is suppressed. But the most spectacular effect that is brought about by the presence of a reactive element in alumina-forming alloys is a change in the microstructure and morphology of the oxide scale. Oxidation of yttrium-free β-NiAl produced a convoluted scale surface with large ridges and whiskers, whereas Y-implanted β-NiAl gave rise to a smooth scale surface without any whiskers. Studies on FeCrAl alloys showed that Y-doped samples developed a distinct columnar grain morphology (average width 0.5 µm) in contrast to equiaxed grains (1 µm) formed on Y-free samples [52]. So the presence of a reactive element in alumina-forming alloys decreases scale convolutions, induces oriented rather than randomly distributed oxide grains, and decreases the oxide grain size, favoring oxygen transport over aluminum diffusion. The reactive elements probably exert their influence by segregating to the grain boundaries in the oxide scale and thereby cause significant changes in the grain boundary transport. An enhanced oxygen transport and a reduced aluminum transport favor oxide growth at the alloy–scale interface and reduce the formation of voids and cavities at this interface. This, in turn, possibly reduces exfoliation and spallation of the alumina scales in RE-containing alloys. Buckling of the Cr 2O 3 scale on pure chromium is also thought to be due to the formation of new oxide within the existing layer, and the role of active element addition is to modify this. During high-temperature performance of either Cr 2O 3- or Al 2O 3-forming alloys, improvement in scale adherence is possibly the major beneficial effect contributed by the reactive element/oxide dispersion addition, particularly during thermal cyclings. Of the mechanisms suggested, elimination of interfacial voids and mechanical keying or peg formation seem to be of most significance. The formation of oxide pegs is perhaps the most decisive factor for oxide adhesion during the latter stages of growth, whereas vacancy sink processes are important in the early stages of oxidation. However, a fine, uniform distribution of tiny oxide pegs will produce the best effects. For Al 2O 3-forming alloys, there exists controversy about the beneficial or detrimental effects of certain active elements even though Y has shown beneficial effects in terms of both reduction of oxidation rate (marginally) and improved scale adhesion (most recognized). Reactive elements do modify the microstructure and morphology of the alumina scales not only by decreasing the grain size but by inducing intergranular segregation or second-phase precipitation, which influences both transport processes and the plasticity of the scales. The most recognized effect of the reactive element additions in Al 2O 3-forming alloys is the improvement in scale adhesion. It generally

346

Chapter 6

accounts for enhanced scale plasticity, even though there exists no direct evidence to support this. More experimentations are needed in this domain to ascertain the exact reason for enhanced scale plasticity. It is to be recognized that the REE phenomenon, which was realized nearly six decades ago by chance, still continues to allure the scientific community in search of a unified theory that can explain the effects so far reported in the literature. Often it is proposed that several mechanisms are operative simultaneously.

6.7 HOT CORROSION Various Ni-, Co-, and Fe-based alloys used in high-temperature structural components, such as superheater supports; superheater and reheater tubes of boilers, and first-stage nozzles, blades, and vanes of gas turbines in industrial, marine, and aircraft engines, are often subjected not only to complex thermal and mechanical stresses but also simultaneously to combustion gases generated through burning fossil fuels containing small amounts of impurities. The environment in a gas turbine is not that of a clean combustion gas. The source of the impurities can be either fuel or intake air. Sodium and vanadium are often present in the fuel as oil-soluble compounds. Sodium in the air can be present as an aerosol of sea salt. In seawater, the majority of the sodium is present as sodium chloride, but approximately 11% is present as sodium sulfate. Sulfur is present in the fuel, which may account for up to 0.4 wt % in aviation kerosenes even though the average is near 0.1%. Industrial turbines may burn light oils with similar sulfur contents; heavy distillates may contain up to 2 wt % of sulfur or more. On thermodynamic considerations, it has been established [54] that sodium chloride is unstable in the presence of even small concentrations of sulfur in an oxidizing environment and the following reaction goes virtually to completion: 2NaCl ⫹ SO 3 ⫹

1 O 2 ⫽ Na 2SO 4 ⫹ Cl 2 2

(6.32)

Therefore, when metals/alloys are exposed to combustion environments, deposition of salts rich in sodium sulfate and/or vanadium compounds on the alloy or oxide surface is a common experience. Experiences have proved beyond doubt that the nature of degradation of the metallic component by combustion gases in the presence of a salt deposit is definitely different from that occurring in the absence of deposit. The salt deposits are usually liquid or semiliquid at the operating temperature and cause enhanced metal deterioration. This type of severe attack, which can be catastrophic at times, is referred to as hot corrosion. Therefore, the process of hot corrosion can be defined as an accelerated attack/degradation in a high-temperature gaseous environment (containing oxygen, sulfur, alkali salts, vanadium, and a host of other contaminants) of a metallic material (metal/

Alloy Oxidation

347

alloy) whose surface is coated by a thin salt film. Turbine manufacturers and users became aware of such aggressive corrosion processes in the late 1960s. In this regard, Stringer [55] has presented an excellent review of the phenomenology and proposed mechanisms. In most early cases of severe hot corrosion, the salt deposits were analyzed and found to contain a substantial amount of sodium sulfate (Na 2SO 4) with different amounts of Ca, Mg, Pb, V, Zn, and chloride ion; however, the compositions also included a mixture of calcium, sodium, and magnesium sulfates. Vanadium, often present in fuel oils as metal–organic complexes (porphyrins) whose concentration may reach as high as 500 ppm in heavy distillates and residues, can also condense as molten vanadium oxide, typically on the vanadium-rich oxide of V 2O 5. In industrial applications there may be situations where the salt deposits consist of a mixture of sulfates and vanadates. Experience has shown that the combined vanadate-sulfate attacks can be more serious than individual attacks. Vanadates containing 10–20% Na 2O are reported to be most corrosive. Sodium vanadates are formed by the reaction of Na 2SO 4 and V 2O 5. But the exact composition of the vanadates may vary as (Na 2O)xV 2O 5, depending on the conditions. The molten vanadates are capable of fluxing most metal oxides, allowing rapid diffusion of oxygen through the melt to the metal surface. The combined attack of vanadate-sulfate deposits causing serious damage to the components is probably due to the fact that vanadate serves as an effective fluxing agent while the sulfate reacts according to the sulfate-sulfide mechanism. The presence of vanadium principally modifies the nature of sodium sulfate, rendering it more acidic so that both forms of corrosion may proceed simultaneously and interactively. This type of highly accelerated corrosion process is often termed vanadate-induced hot corrosion. Furthermore, accelerated corrosion induced by other salts, such as molten alkali carbonates and nitrates, has also been experienced in special applications such as molten carbonate fuel cells and solar heat exchangers at and above 873 K, and the involved processes have been explained in the light of general mechanism suggested for hot corrosion. Basically, the essential ingredients for hot corrosion processes are (1) elements in an alloy for oxidation, (2) species in the gas phase for reduction, and (3) a salt deposit on the surface of the alloy capable of influencing the oxidation– reduction processes. Even though this type of corrosion takes place in the presence of various types of salt deposits, in many practical situations it is caused by sulfates, e.g., Na 2SO 4, or mixtures of sulfates. Therefore, in the present context, it is neither the intention nor the desire to account for such corrosion processes in the presence of various salt deposit compositions, and accordingly the present discussion is restricted to Na 2SO 4-induced processes. Furthermore, with respect to materials, the gas turbine manufacturers normally choose Ni- or Co-based superalloys (for compositions, refer to Tables 6.4 and 6.5), at times with oxidation-resistant coatings, and

348

Table 6.4

Compositions of Nickel-Based Superalloys Composition in wt% (balance Ni)

Group I. Sheet alloys a. Solid solution strengthened

b. γ ′ or γ ″ precipitation strengthened

II. Bar or forgings a. Air-melted, γ′ precipitation strengthened

Alloy

Fe

Co

Mo

W

Ti

Al

Cb

C

Others

Hastelloy B Hastelloy X Nimonic 86 Inconel 600 Inconel 617 Nimonic 75 PK 33 C 263 Inconel 625

— 22 25 15 22 20 19 20 22

5 18 — 8 — — — — 5

— 1.5 — — 12 — 14 20 —

28 9 10 — 9 — 7 6 9

— 0.6 — — — — — — —

— — — — — 0.4 1.9 2.1 —

— — — — 1 0.2 1.9 0.5 —

— — — — — — — — 4.0

— 0.1 0.05 0.08 0.07 0.10 0.04 0.05 0.05

0.3 V — 0.03 Ce — — — — — —

Nimonic 80 A Nimonic 90 Nimonic 105 Inconel X750 Udimet 500 Waspaloy M 252

20 20 15 15 18 20 20

— — — 7 — — —

— 17 20 — 18 13 10

— — 5 — 4 4 10

— — — — — — —

2.5 2.5 1.2 2.5 2.9 3 2.6

1.5 1.5 4.5 0.9 2.9 1.3 1

— — — 1.0 — — —

0.05 0.08 0.15 0.04 0.08 0.08 0.15

B, Zr B, Zr B, Zr Zr B, Zr B B

Chapter 6

Cr

c. Vacuum melted, γ ″ precipitation strengthened d. Corrosion-resistant alloys e. Mechanically alloyed alloys III. Cast alloys

Nimonic 115 Nimonic 120 Astroloy Udimet 720 Inconel 718 Rene´ 95 Nimonic 91 Nimonic 101 MA 754 MA 6000 IN 713C IN 738 LC IN 100 IN 939 Rene´ 80 Mar M 006 Mar M 009 M 22 B 1900

15 12 15 18 19 13 28.5 24.5 20 15 12.5 16 10 22.5 14 9 9 5.7 8

— — — — 18 — — — — — — — — — — — — — —

15 10 17 15 — 8 20 20 20 — — 8.5 15 19 9.5 10 10 — 10

3.5 5.8 5.3 3 3 3.5 — 1.5 — 2 4.2 1.7 3 — 4 2.5 — 2 6

— — — 1.3 — 3.6 — — — 4 — 2.6 — 2 4 10 12 11 —

4 3.6 3.5 5 0.9 2.5 2.3 3.0 0.5 2.5 0.8 3.4 4.7 3.7 5 1.5 2 — 1

5 4.6 4 2.5 0.6 3.5 1.2 1.5 3 4.5 6.1 3.4 5.5 1.9 3 5.5 5 6.3 6

— — — — 5.2 3.5 0.7 1.0 — — 2.0 0.9 — 1.0 — — 1.0 — —

0.15 0.06 0.06 0.03 0.05 0.06 0.05 0.05 0.05 0.05 0.12 0.11 0.18 0.15 0.17 0.14 0.14 0.13 0.10

B, Zr B, Zr B, Zr B, Zr B B, Zr B, Zr B, Zr 0.6 Y 2O 3 2.5 Y 2O 3 B, Zr 1.7 Ta 1V 1.4 Ta B, Zr 1.5 Ta, 1.8 Hf 1.8 Hf 3 Ta 4 Ta

Alloy Oxidation

b. Vacuum-melted, γ ′ precipitation strengthened

Source: R. W. Cahn, P. Haasen, and E. J. Kramer, eds. Materials Science and Technology, Vol. 7. VCH, Weinheim, 1992, p. 672.

349

350

Chapter 6

Table 6.5 Compositions of Cobalt-Based Superalloys Composition in wt.% (balance Co) Cr I. Wrought alloys S 816 HA 188 L 605 Mar M 918 II. Cast alloys HS 21 X 40 FSX 414 Mar M 509 Mar M 302

Ni

Fe

Mo

W

Ti

Cb

C

Others

20 22 20 20

20 22 10 20

4 1.5 — —

4 — — —

4 14 15 —

— — — —

4 — — —

0.38 0.1 0.1 0.05

— 0.08 La — 7.5 Ta, 0.1 Zr

27 25.5 29.5 24 21.5

3 10.5 10.5 10 —

1 2 2 — —

5 — — — —

— 7.5 7 7 10

— — — 0.2 —

— — — — —

0.25 0.5 0.35 0.6 0.85

— 0.01 B 0.01 B 3.5 Ta, 0.5 Zr 9 Ta, B, Zr

Source: R. W. Cahn, P. Haasen, and E. J. Kramer, eds. Materials Science and Technology, Vol. 7. VCH, Weinheim, 1992, p. 685.

for which the present discussion is limited to either pure Ni or Co or their respective alloys. It is to be further recognized that once a deposit of Na 2SO 4 has been formed on the alloy surface, the extent to which the deposit affects the corrosion resistance of the alloy will be governed by whether or not the deposit melts at the operating temperature, how adherent it is, and the extent to which it wets the surface. Even though a liquid deposit is generally necessary for severe hot corrosion to occur, considerable degradation of alloys and oxidation-resistant coatings in marine gas turbines has also been experienced under dense, thick, solid deposits. Accordingly, hot corrosion processes have been subdivided into two categories: type I, high-temperature hot corrosion (HTHC) processes, which are experienced above the melting temperature of pure Na 2SO 4 (1157 K), and type II, low-temperature hot corrosion (LTHC) processes, which occur at temperatures below the melting point of Na 2SO 4 (i.e., in the range 873–973 K) where the deposited salt is expected to be solid [56]. The combustion gases arising from burning of coal and other types of fuel oil contain both SO 2 and SO 3; accordingly, to simulate the service environmental conditions, most of the hot corrosion tests employ either static air or flowing oxygen ⫹ SO 2 gas mixtures over a platinum catalyst, and/ or SO 3, since SO 2 ⫹ 1/2 O 2 s SO 3. SO 2 or a mixture of SO 2 ⫹ O 2 and/or SO 3 contains two oxidants, S and O. If one considers the thermodynamic equilibrium between a metal and a mixture of SO 2 ⫹ O 2, only one stable compound, e.g., an oxide or a sulfate, is the stable end product of the reaction. However, for kinetic

Alloy Oxidation

351

Figure 6.27 Phase stability diagram for the Na-S-O system at 1173 K [50].

reasons the corrosion process may lead to the formation of oxides, sulfides, and sulfates (if these are thermodynamically stable under reaction conditions).

6.7.1

Thermodynamic Considerations

In order to have an understanding of Na 2SO 4-induced hot corrosion, one requires knowledge not only of the properties and stability of Na 2SO 4 but also of the reactions of the metal under consideration with the two oxidants such as sulfur and oxygen. The stability of Na 2SO 4 is illustrated in terms of phase stability at a constant temperature, as depicted in Fig. 6.27, where the activities of O 2 and SO 3 are used as coordinates. In this figure, the activity of SO 3 can alternatively be expressed in terms of activity of Na 2O, as the equilibrium of the reaction: Na 2SO 4 s Na 2O ⫹ SO 3 can be expressed as: a Na2Oa SO3 ⫽ Constant ⋅ a Na2SO4 For pure Na 2SO 4, a Na2SO4 is unity. Thus, a Na2O ⫽ Constant ⋅ a SO⫺1 3

(6.33)

352

Chapter 6

In Fig. 6.27, the activity of Na 2O is considered to be unity at the Na 2O/Na 2SO 4 boundary. At sufficiently low oxygen and sulfur activities (not shown in the figure), Na (l) is the only stable phase in the Na-O-S system; otherwise Na 2O is the only stable phase at sufficiently low sulfur activities and Na 2S at sufficiently low oxygen activities. For interpretation of the reactions of a metal in a mixed environment consisting of two oxidants such as sulfur and oxygen, one needs to consider the phase stability diagram of the corresponding metal–oxygen–sulfur system also at the temperature under consideration. Such diagrams define the stability regions of various phases that may exist at different activities of the two oxidants. These are the high-temperature analogs of well-known Pourbaix diagrams used in aqueous corrosion processes. As an illustration, the phase stability diagram for the Ni-O-S system at 1173 K is presented in Fig. 6.28 in terms of activities of O 2 and SO 3, where activity of nickel has been considered equal to unity [50]. This diagram illustrates that at sufficiently low activities of both O 2 and SO 3, Ni (s) metal is the only stable phase; at low SO 3 activities and high O 2 activities NiO is the only stable phase. While at high activities of both O 2 and SO 3, NiSO 4 is the stable phase. The possible sulfide phases at this temperature are Ni-S liquid solution, NiS, and NiS 2. It is to be recognized that such diagrams do not provide any information about the mutual solubility of the phases, e.g., solubility of oxygen and sulfur in nickel metal, sulfur solubility in NiO, etc. To interpret the Na 2SO 4-induced hot corrosion process of nickel, it is necessary to know how the presence of Na 2SO 4 affects the Ni-O-S phase stability

Figure 6.28 Phase stability diagram for the Ni-S-O system at 1173 K [50].

Alloy Oxidation

353

Figure 6.29 Superimposed phase stability diagram of Ni-O-S on the stability region of Na 2SO 4 in Na-O-S diagram at 1173 K. The broken lines represent isoactivity lines for NiSO 4 and NaNiO 2 at activities of 10⫺2 and 10⫺4 [50].

diagram. The obvious choice would be to superimpose the Ni-O-S diagram (Fig. 6.28) on the stability region of Na 2SO 4 in Na-S-O diagram (Fig. 6.27). Such superimposition is illustrated in Fig. 6.29. However, while doing so, one should not forget to consider the formation of some other new phases (like Na 2NiO 2 and NaNiO 2) and the mutual solubilities of the different phases, e.g., NiO in Na 2SO 4 or NiSO 4 in Na 2SO 4. Solid NiSO 4 is stable only at high SO 3 activities but gets dissolved in molten Na 2SO 4 at activities lower than unity. In a similar way, NaNiO 2 also gets dissolved in Na 2SO 4, whereas Na 2NiO 2 is thermodynamically unstable in the temperature range of 1173–1273 K. The mutual solubilities of NiO and the nickel sulfides are insignificant. Gupta and Rapp [57] have measured the solubility of NiO in fused Na 2SO 4 at 1200 K and 1 atm O 2 as a function of a Na2O (or p SO3), and their results are presented in Fig. 6.30. This figure demonstrates that the solubility passes through a minimum at log aNa2O ⫽ ⫺10.3, which prompted the authors to conclude that at higher values of Na 2O activity, NiO dissolves as nickelate (NiO 2⫺), whereas at lower values it dissolves as Ni2⫹ ions. These two types of dissolution processes are termed as basic and acidic fluxing, respectively, which will be discussed in

354

Chapter 6

Figure 6.30 Solubility of NiO in fused Na 2SO 4 at 1200 K [57].

details in Sec. 6.7.4. These terminologies are the commonly accepted nomenclature in describing oxyanion salts in terms of acid–base equilibria. In such systems, the O2⫺ ions have the same role as that of the hydroxide ion in aqueous solutions: Base ⫽ acid ⫹ O2⫺

(6.34)

Sodium sulfate may be considered to consist of a basic component, Na 2O (or O⫺2 ions) and an acid component, SO 3. In liquid sodium sulfate deposit, the sulfate ion decomposes according to SO 42⫺ ⫽

1 O 2(g) ⫹ SO 2(g) ⫹ O2⫺ 2

(6.35)

So a Na 2SO 4 melt is composed of an oxygen ion or Na 2O, whose activity is defined by the oxygen and SO 2 potentials. When the oxygen ion concentration of the melt is low compared to the value required for maintaining equilibrium in the dissociation reaction of the metal oxide according to NiO s Ni2⫹ ⫹ O2⫺

(6.36)

Alloy Oxidation

355

the metal oxide reacts with SO 3 and gets dissolved in the molten salt as a cationic species. This is known as acid fluxing. If the oxygen ion activity of the sulfate melt is high compared to that required for complex anion formation according to the equilibrium MO ⫹ O2⫺ s MO 22⫺

(6.37)

then complex ion formation can take place and the metal oxide gets dissolved in the melt as a complex anion. This is referred to as basic fluxing. Figure 6.30 further demonstrates that from a minimum solubility of NiO corresponding to activity of Na 2O as 10⫺10.3, the solubility goes on increasing with the increase of a Na2O (i.e., decreasing p SO3) according to the reaction: 2NiO ⫹ O2⫺ ⫹

1 O 2 s 2NiO 2⫺ 2

(6.38)

and this corresponds to basic fluxing. It has been proposed [57] that NiO can dissolve in Na 2SO 4 as nickelate (NiO 2⫺), possibly due to instability of Na 2NiO 2. Similarly, for the right-hand side of the curve, solubility of NiO increases with decreasing a Na2O according to the reaction: NiO s Ni2⫹ ⫹ O2⫺

(6.39)

which corresponds to acid fluxing. Such solubility behavior in Na 2SO 4 is also reported for other oxides, such as Co 3O 4, Fe 2O 3, Al 2O 3, and Cr 2O 3, which are of principal interest in high-temperature alloys and coatings. The reported solubility curves for the different oxides maintain the same nature but with different minima values and are displaced to the left or the right of each other. The differences in magnitude of -log a Na2O values as observed for the solubility minima between the different types of oxides further emphasize that the local chemistry within a fused salt film is important. The terms acidic and basic are relative and refer to the reaction that occurs rather than the condition of the melt. To simplify the fluxing processes with reference to hot corrosion of alloys in the presence of a molten Na 2SO 4 deposit, it is useful to consider the appropriate stability diagrams as a means for predicting the feasible fluxing reactions. It is to be recognized that the initially formed oxides on the metal/alloy surface will virtually determine the nature of the salt to be either basic or acidic. This in turn will be governed by the oxide ion concentration at the salt–gas interface of the deposited salt and the affinity of the oxides as well as their metal ions for oxide ions. The affinity of different relevant metals for the oxide ions can be best described by using their stability diagrams superimposed on the Na-S-O diagram (for Na 2SO 4) at the temperature under consideration. As an illustration, such a superimposed stability diagram is presented in Fig. 6.31 for the phases of Ni, Al, and Cr that can exist in contact with a molten Na 2SO 4 layer on Ni-Cr-Al

356

Chapter 6

Figure 6.31 Stability diagram to illustrate the phases of nickel (– ––––), aluminum (⋅⋅⋅⋅⋅⋅⋅⋅), and chromium (xxxxx) that can exist in an Na 2SO 4 layer on a Ni-Cr-Al alloy at 1200 K [58].

alloy at 1200 K [58]. This diagram has been constructed by superimposing the Ni-O-S, Al-O-S, and Cr-O-S stability diagrams on the Na-O-S phase diagram as a function of log p O2 versus log pSO3. In this diagram, the Na 2SO 4 region is bounded by Na 2O and Na 2S and is indicated by solid straight lines. It has already been mentioned that basic fluxing is dependent on the ability of the salt to maintain a high concentration of oxide ions (O2⫺) for reaction with a protective layer. This diagram suggests that NiO will be dissolved in Na 2SO 4 in preference to Al 2O 3 and Cr 2O 3 at the composition of the melt designated by the symbol 䊟 due to generation of oxide ions, thus making the melt more basic. On the other hand, for the composition of the melt designated by the symbol 䊉, Al 2O 3 will be dissolved as sodium aluminate (Na 2Al 2O 4) in preference to Cr 2O 3. Thus, the melt is made more acidic due to depletion of oxide ion or release of SO 3 within the melt. These concepts are useful in understanding the mechanisms of hot corrosion processes.

Alloy Oxidation

6.7.2

357

Kinetics

Several authors have suggested that hot corrosion of all susceptible alloys and coatings is characterized by a two-stage process: an incubation period with a slow rate of reaction, possibly with the alloy being protected by a stable Cr 2O 3 or Al 2O 3 layer; and a propagation stage with rapid and often catastrophic material degradation as illustrated in Fig. 6.32. During the incubation period, elements in the alloy are oxidized, and electrons are considered to be transferred from metallic atoms to reducible substances in the salt deposit. When the reduced substances are the same as those that would have reacted with the alloy in the absence of the deposit, the reaction product barrier forms beneath the salt on the alloy surface. As the hot corrosion process is continued, certain features become apparent which indicate that the salt is affecting the corrosion process, and eventually the selective oxidation process is ineffective. The time, which may vary from a few hours to thousands of hours for which the most effective reaction product barrier is stable beneath the salt layer, is influenced by a number of factors as illustrated in Fig. 6.33. These factors include alloy composition and its microstructure, fabrication condition, pretreatment of alloys, environment composition, gas velocity, erosion, salt deposit condition and its composition, performance/test temperature,

Figure 6.32 Mass change versus time for IN-738 alloy coated with 1 mg/cm2 Na 2SO 4 in 1 atm O 2 at different temperatures [16].

358

Chapter 6

Figure 6.33 Schematic illustration of the conditions that develop during the initiation stage governing the time at which transition from the initiation to propagation stage occurs.

temperature cycles, test specimen geometry, and so forth. The effect of these factors influencing the initiation of hot corrosion attack has been well documented by Giggins and Pettit [58]. Hot corrosion process shows a strong temperature dependence. For most of the superalloys, the corrosion rate is maximal within the temperature range 1123– 1173 K and decreases markedly at temperatures up to 1273 K. However, the time required to initiate hot corrosion attack decreases as temperature is increased (Fig. 6.32). For a fixed amount of sulfur in the gas, the SO 3 pressure decreases as the temperature increases, resulting a lower hot corrosion rate. Moreover, for the same ingestion rate of salt, less is deposited on test specimens as the temperature is increased. Accordingly, less attack takes place with smaller amount of salt deposit at higher temperatures. The effect of alloy composition is still a matter of debate. It is generally agreed that the content of chromium in the alloy is the most important factor and for Ni-based alloys at least 15% Cr is required for better resistance to such corrosion processes. Much of the disagreement concerning the effects of other elements is possibly due to interactive effects within the alloy scale and salt. Certain alloying elements are found to have beneficial effects

Alloy Oxidation

359

Figure 6.34 Weight change versus time data for the cyclic hot corrosion of Ni-30% Cr and Ni-30% Cr-6% Al alloys exhibiting superior performance of the former at longer times [58].

over certain composition ranges but can be deleterious over others. This is clearly demonstrated in Fig. 6.34 for the corrosion of Ni-30% Cr-6% Al and Ni-30% Cr alloys. In the hot corrosion resistance of Ni-based superalloys apart from the beneficial effects shown by chromium; Co and Ta slightly improve the resistance; Ti seems to do little, whereas Mo and W are detrimental, especially at higher temperatures. It must be recognized that all of the factors mentioned above and illustrated in Fig. 6.33 have significance in preconditioning the alloy, which will ultimately determine the type of propagation mode to be followed. Such preconditioning may include (1) depletion of the element responsible for forming the initial oxide layer (Al or Cr), (2) formation of sulfides of the alloy constituents due to sulfur penetration through the scale, (3) dissolution of oxides into molten salt deposit, (4) development of growth stresses, and (5) alteration in the salt composition for more corrosive conditions. The end of the initiation stage is marked by the propagation of attack, and during this stage the characteristic hot corrosion morphology of the scale appears. Divergent results have been reported in literature by numerous investigators studying this phenomenon since their experimental conditions varied widely. Generalization of the test results indicates that the degradation of alloys in the

360

Chapter 6

Figure 6.35 Schematic diagram illustrating the three categories of protective scale breakdown to less protective reaction product during the process.

presence of salt deposits falls into three categories [58] as depicted in Fig. 6.35. In one of these categories, the salt is innocuous, and accordingly degradation during the propagation stage proceeds by the mechanisms determined by the alloy and environment composition. Such a situation is likely to occur with porous solid deposits through which the gas can penetrate. No definite example of such an effect is yet reported. In the second category, reactions are considered between the elements in the alloy and constituents of the gas phase in the presence of a liquid salt in the formation of nonprotective reaction products. The reactions that take place under such conditions are similar to those that occur in cleaning of metal surfaces by the use of salt baths for descaling. Accordingly, this mode of propagation is often referred to as salt fluxing. The third category involves propagation stages where a constituent of the salt deposit either is introduced to the alloy or reacts with the alloy or its corrosion products, so that nonprotective reaction product barriers are formed. Such propa-

Alloy Oxidation

361

gation stages are often termed salt component–induced hot corrosion processes. Some elements, such as chlorine, sulfur, and carbon, can produce such effects depending on the deposit composition. In the cases of Na 2SO 4-NaCl deposits, sulfur and chlorine are the significant elements. The importance of the element carbon is emphasized by the fact that in fluidized-bed reactors for combustion of coal, lime is added. The lime accepts sulfur from the coal and is converted to CaSO 4. Because the beds are characterized by low oxygen activity and because of simultaneous presence of CaO, CaSO 4, and CO ⫹ CO 2, relatively high SO 2 activities are produced according to CaSO 4 ⫹ CO ⫽ CaO ⫹ SO 2 ⫹ CO 2

(6.40)

which ultimately leads to rapid degradation of the metallic components involving oxidation and/or sulfidation due to the reaction with SO 2.

6.7.3

Scale Morphology

The morphology of the scales resulting from hot corrosion reactions should be different from that associated with simple oxidation of a metal/alloy in a single oxidant environment because oxidation and sulfidation commonly occur simultaneously. Investigations related to fundamentals of hot corrosion on pure metals, binary, and ternary alloys (mostly Ni- and Co-based) have revealed widely different scale morphologies due to varying experimental conditions [50]. Even case histories of commercially used Ni- and Co-based alloys that have suffered such damage under different operating conditions are reported to have revealed various types of scale morphologies [55]. In most service cases the general consensus is that a layer of sulfide particles forms beneath a region of porous oxide. The sulfides have often been identified as chromium-rich, even though nickel sulfides have also been reported in certain cases. Furthermore, the extent of internal sulfide layer can vary considerably depending on the service/test conditions. At times the sulfide layer appears to be virtually nonexistent, or it may appear as a very thin band of fine discrete sulfide particles or even as a large band of interconnected sulfide particles. In general terms, the corrosion product morphology of Ni- and Co-based superalloys can be described to consist of an outer porous oxide layer, an intermediate thick layer of oxide dispersed with metal-rich fragments, and an inner layer of fine sulfide particles in the metal matrix. The outer porous oxide layer consists primarily of simple oxide of the base metal, i.e., either NiO or CoO depending on the type of superalloy, and at times with some spinel. Beneath this zone, there exists an intermediate layer comprising Ni- or Co-rich fragments in a Cr 2O 3 matrix with some spinel. Finally, the inner zone, which is in the immediate vicinity

362

Chapter 6

of mixed oxide/metal layer, is composed of finer, light gray sulfide particles in the metal matrix. Since the sulfides are usually rich in chromium, the underlying metal consists of chromium-depleted discontinuous fragments of the base alloy. Under exceptionally severe service conditions, liquid Ni-rich or Co-rich sulfides may also develop in the inner layer, leading to very rapid degradation of the alloy [55].

6.7.4

Mechanisms

The mechanism of hot corrosion has been the subject of a number of investigations, which have been extensively reviewed by Stringer [55]. Some of the first proposals that protective scales could be removed from the surfaces of metals/ alloys by molten deposits were generated in studies related to fireside corrosion of boilers [58]. It was also mentioned earlier (Sec. 6.7.1) that at relatively elevated temperatures, comparatively high SO 3 pressures are required to form sulfates of such elements as Ni, Co, and Al. Furthermore, for the same amount of sulfur in the gas stream SO 3 pressures are lower at higher temperatures. Therefore, SO 3 can be considered to have a progressively less dominant role as the operating temperature of the metallic component is increased. Accordingly, investigators started developing mechanisms of hot corrosion attack without involving SO 3 in the gas phase at temperatures above about 1023 K. Two different models are reported in the literature to explain the mechanism for breakdown of the protective oxide scale formed on metal/alloy surface during the incubation period, leading to propagation stage of hot corrosion attack. These are (1) the acid–base fluxing model and (2) the electrochemical model. Perhaps the most popular is the acid–base fluxing model, which has been used extensively since it was first proposed. Bornstein and Decrescente [59] were the pioneers who proposed that hot corrosion of alloys involves a basic fluxing process, as opposed to acidic fluxing process involving SO 3. According to their model, the protective oxide scales are destroyed as a result of reactions with oxide ions in the salt where the oxide ions are produced by removal of sulfur from Na 2SO 4. Goebel et al. [60] subsequently extended the high-temperature fluxing reactions to acidic processes where the component to make the salt acidic was proposed to be oxides of the elements in the alloys (e.g., MoO 3, WO 3). The proposal embodies the fact that porous oxide scales are formed during basic or acidic fluxing by precipitation from the molten salts in which these oxide scales had initially dissolved. The dissolution and precipitation processes are eventually controlled by the oxide ion activity of the melts, which in turn is governed by the removal of sulfur from the salt (Na 2SO 4) or by addition of oxides of certain elements (e.g., MoO 3, WO 3) to the salt. An idealized model for fluxing a protective oxide scale in a molten layer of salt deposit [61] is shown schematically in Fig. 6.36. In this model, it is assumed that the metal is covered with a thin protective oxide scale that is continuously

Alloy Oxidation

363

Figure 6.36 Schematic representation of an idealized model for fluxing of a continuous, protective oxide scale in a layer of molten salt deposit [61].

dissolved in the molten salt. It further implies that if the reaction rate is a linear one, the protective scale is reformed at the same rate as it dissolves, i.e., the scale has a stationary thickness. The electrochemical model proposed by Rapp and Goto [62,63] states that an accelerated attack during hot corrosion is sustained through a dissolutionprecipitation process whenever there is a negative gradient in the solubility of the protective oxides across the salt film, i.e., at the oxide–salt interface, as illustrated in Fig. 6.36. Such gradients are developed by an electrochemical reduction reaction that accomplishes oxidation of the metallic elements. Accordingly, the dissolved protective oxide is transported away from the salt–oxide interface and precipitated at some distance from it in a region of lower solubility as a noncontinuous and nonprotective phase. This model is based on the solubility data of Cr 2O 3 and Al 2O 3 in fused Na 2SO 4 at 1200 K, and the slopes of the solubility lines substantiate the following acidic and basic dissolution reactions: Al 2O 3 s 2Al3⫹ ⫹ 3O2⫺ acid dissolution

(6.41)

Al 2O 3 ⫹ O2⫺ s 2AlO 22⫺ base dissolution

(6.42)

Cr 2O 3 s 2Cr

3⫹

⫹ 3O

2⫺

acid dissolution

Cr 2O 3 ⫹ 2O2⫺ s 2CrO 42⫺ base dissolution

(6.43) (6.44)

The oxide stability gradient is established by the nature and site of the electrochemical reduction step, which will generate basic ions. The interrelation of the gradient of the basic ions in the melt to the oxide solubility map determines the extent and continuance of hot corrosion reaction. The major advantage of this model over the foregoing fluxing model is that oxide ion production need not occur as a result of removal of sulfur from Na 2SO 4. Therefore, the electrochemi-

364

Chapter 6

cal model is more general and powerful in explaining the oxide scale dissolution and reprecipitation during hot corrosion of metals/alloys. Despite the elegance of the electrochemical model, investigations have most often utilized the acid–base fluxing model to explaining the observations. The mechanisms for basic and acid fluxing in the presence of Na 2SO 4 salt deposit are presented in the following sections. Basic Fluxing It is common in basic fluxing for the amount of attack to increase as the amount of deposited salt is increased. For a fixed quantity of salt under basic conditions, the melt is gradually consumed and becomes saturated with complex anion, whereupon the reaction eventually subsides. The microstructural features of the scale formed on a metal/alloy brought about by hot corrosion damage are largely governed by the compositions of the material and the gas phase for a particular composition of the salt deposit, apart from other factors. Accordingly, for illustrating the basic fluxing mechanism in hot corrosion, a simplified system, e.g., pure Ni under a thin Na 2SO 4 deposit exposed in pure oxygen at 1173 K, is considered here. The corresponding kinetic curves along with the superimposed micrographs indicating the reaction sequence are presented in Fig. 6.37. The corresponding reaction sequence is illustrated schematically in Fig. 6.38. Here it is worth mentioning that when nickel with Na 2SO 4 deposits undergoes a reaction in O 2 or air (containing insignificant SO 3), accelerated corrosion takes place only above the melting point of Na 2SO 4. At lower temperatures, Na 2SO 4 serves as a barrier to oxidation, and the reaction rate is reduced compared with of nickel containing no Na 2SO 4 deposit. The kinetic curves (Fig. 6.37) clearly demonstrate that Ni has undergone accelerated corrosion in the presence of a molten Na 2SO 4 film in comparison with the bare one but such accelerated damage lasts for a relatively brief period. Subsequently, the rate virtually equals that for Ni without the salt deposit. The very slow rate of reaction for Ni without salt is due to the formation of a highly tenacious, compact, and adherent film of NiO. Reaction Sequence. Upon initial exposure to the oxygen atmosphere at the designated temperature, NiO forms a layer covered by molten Na 2SO 4. The continued formation of NiO rapidly lowers the pO2 in the salt at the oxide–salt interface. Therefore, sulfur potential increases, leading to transport of sulfur through the thin oxide, and sulfide formation takes place at the scale–metal interface as illustrated in Fig. 6.38a. But there are varied opinions about the nature of sulfur transport. As sulfides are detected within a very short exposure time, such transport cannot occur by lattice diffusion of sulfur ions through NiO. A more feasible mechanism could be transport of SO 2 molecules penetrating through the microcracks of the scale. The source of SO 2 is dissociation of the sulfate ions according to Eq. (6.35).

Alloy Oxidation

365

Figure 6.37 Mass change versus time for pure Ni with and without Na 2SO 4 deposit and associated scale microstructures [16].

Equation (6.35) further suggests that with increasing consumption of SO 2 and O 2, the oxide ion (O2⫺) activity in the melt will increase to maintain equilibrium. As a result, the salt will turn more basic in nature. The increase in basicity will reach a maximum at the areas of sulfide formation, i.e., where SO 2 is consumed most rapidly. In these (Fig. 6.38b), the NiO scale will react to form soluble nickelate ions (NiO 2⫺) according to Eq. (6.38). These, in turn, will diffuse to the salt– gas interface where the oxide ion concentration is low and will allow the liquid salt to penetrate it and be spread along the scale–metal interface (Fig. 6.38c), lifting and cracking the scale. It is possible that such cracking is initiated by the formation of a liquid Ni-S phase at the scale–metal interface with a greater molar volume than that of nickel. The cracking of the scale allows oxygen penetration, resulting in oxidation of the sulfides and freeing of sulfur to further penetrate

366

Chapter 6

Figure 6.38 Schematic diagram illustrating the sequential development of scale in Na 2SO 4-induced hot corrosion of pure Ni in 1 atm O 2 [16].

into the metal. Repetition of this process eventually produces a porous, honeycomb-like NiO layer (Fig. 6.38d) and results in diffusion of sulfur and oxygen along the grain boundaries of the metal (Fig. 6.38e). Eventually as Na 2SO 4 is trapped in the porous scale, the rapid reaction ceases, leading to the formation of a dense protective NiO layer. The reaction path followed in molten Na 2SO 4 is illustrated in Fig. 6.39 by superimposing the stability diagram of the Ni-S-O system onto the stability region of the Na-S-O system. A similar basic fluxing mode of attack is also observed for protective oxides such as Cr 2O 3 and Al 2O 3 (usually formed on high-temperature alloys) as in the hot corrosion of Ni-8% Cr-6% Al alloy [58]. Initial formation of Cr 2O 3 and Al 2O 3 depletes the molten salt of oxygen and lowers the oxygen potential, creating an oxygen gradient across the melt. As a result of this gradient in oxygen pressure, the sulfur activity increases and nickel sulfide is formed at the surface of the alloy. The low sulfur and oxygen potentials of the salt due to formation of oxides and sulfides lead to an increase in the oxide ion or Na 2O activity in the melt, reaching values at which Al 2O 3 and Cr 2O 3 can dissolve. A process is therefore developed whereby sulfate ions diffuse toward the alloy, and as a region of lower oxygen pressure is approached, these ions sulfidize nickel whereupon oxide ions are produced. These in turn react with Al 2O 3 and Cr 2O 3 to form soluble products in oxide ion–enriched Na 2SO 4.

Alloy Oxidation

367

Figure 6.39 Superimposed stability diagrams for Ni-S-O and Na-S-O systems showing schematic reaction path during hot corrosion of pure Ni [16].

Once suitable conditions have been established, the basic fluxing of Cr 2O 3 and Al 2O 3 can occur by the following reactions: Cr 2O 3 ⫹ O2⫺ s 2CrO 22⫺

(6.45)

Al 2O 3 ⫹ O2⫺ s 2AlO 22⫺

(6.46)

forming chromate and aluminate ions in solution in the melt. These ions subsequently migrate through the melt to sites of higher oxygen potential close to the salt–gas interface where they are subsequently reprecipitated as Cr 2O 3 and Al 2O 3 by reverse reactions according to Eqs. (6.45) and (6.46), releasing oxide ions. The prevailing high oxygen potential at the salt–gas interface shifts Eq. (6.35) in the reverse direction such that the oxide ion or Na 2O activity becomes too low to support the existence of complex anions. Automatically they decompose by reverse reactions as per Eqs. (6.45) and (6.46). It is worth noting that sulfateinduced hot corrosion attacks of pure Ni as well as those of Ni alloys containing chromium and aluminum are not self-sustaining processes. As long as there is a steady supply of sulfate ions from Na 2SO 4, the hot corrosion process proceeds at approximately a linear rate. But as the source of sulfate ions get depleted, the attack also diminishes, and oxygen becomes plentiful in the corrosion product.

368

Chapter 6

The nickel sulfide particles in the scale become oxidized and the released sulfur moves deeper into the alloy where chromium sulfide is formed. Acid Fluxing Acid fluxing differs from base fluxing by the fact that the acid-induced attack is usually self-sustaining. Hence, a small amount of salt deposit produces much more attack in acid fluxing processes than in basic fluxing. Salt deposits can be made acidic by two different factors: alloy-induced acidity and gas-induced acidity. In alloy-induced acid fluxing, the acidic conditions in the salt melt are established by the dissolution of such oxides as MoO 3, WO 3, and V 2O 5 formed on the alloy surface, which have a greater affinity for Na 2O. As the oxides are dissolved in the melt, oxide ion concentration of the melt decreases, thus making the salt more acidic. The dissolution of oxides in the salt can occur by dissociation reaction according to Eq. (6.36). Giggins and Pettit [58] discussed in detail the sulfate-induced hot corrosion behavior of two important alloys, Co-25% Al-12% W and Ni-8% Cr-6% Al-8% Mo, which are widely known to suffer from alloyinduced acidic fluxing. During the hot corrosion process, oxides of such metals as tungsten and molybdenum become dissolved in Na 2SO 4, forming tungtates and molybdates, with some SO 3 being displaced from Na 2SO 4. The time required for such dissolution of the refractory metal oxides in Na 2SO 4 depends on the oxidation characteristic of the alloy. In certain alloys, the refractory elements are oxidized at the initial stage of oxidation, whereas in others selective oxidation of other elements causes delay before these are available to react with Na 2SO 4. The solubility of these oxides in Na 2SO 4 is known to be substantially high (e.g., WO 3 dissolves to the extent of more than 50 mol % at 1023 K). As dissolution continues, the Na 2SO 4 solution gradually becomes enriched with the oxides of refractory metals. Al 2O 3, Cr 2O 3, and CoO can also be dissolved in such refractory metal oxide–enriched melts by donating oxide ions to the melt according to 2Al ⫹ 3W ⫹ 6O 2 → Al 2O 3 ⫹ 3WO 3 ⫽ 2Al3⫹ ⫹ 3WO 42⫺

(6.47)

2Cr ⫹ 3W ⫹ 6O 2 → Cr 2O 3 ⫹ 3WO 3 ⫽ 2Cr3⫹ ⫹ 3WO 42⫺

(6.48)

Co ⫹ W ⫹ 2O 2 → CoO ⫹ WO 3 ⫽ Co⫹2 ⫹ WO 42⫺

(6.49)

Subsequently, these ions diffuse through the solution to the outer surface of the melt where the above-mentioned reactions proceed in reverse direction due to the prevalence of lower activity of the refractory metal oxides in that region. Such low activity is caused by evaporation loss of refractory metal oxides to the gas phase. Hence, Al 2O 3, Cr 2O 3, and CoO are dissolved at one side of the molten salt and reprecipitated as a nonprotective scale at the melt–porous oxide interface. The precipitation process results in some of the melt being incorporated into the outer, porous part of the scale.

Alloy Oxidation

369

The most important feature of alloy-induced acidic fluxing is that a zone of refractory metal oxide–rich liquid is formed in the immediate vicinity of the alloy surface and the oxides that are commonly relied on for protection against attack (e.g., Al 2O 3, Cr 2O 3) become nonprotective due to a solution-precipitation process. Therefore, it is the result of a local increase in salt acidity by removal of oxide ions in the salt through the formation of basic complexes (e.g., WO 42⫺, VO 3⫺, or MoO 42⫺) with other components in the alloy or salt. Gas-induced acid attack occurs when the atmosphere contains relatively high partial pressures of SO 3 or V 2O 5, which are introduced to the gas phase during combustion of fuel containing sulfur and vanadium. Under such situations, reaction (6.37) is forced to the left, resulting in a salt of low oxide ion or Na 2O activity. This form of attack is prevalent at low temperatures between 873 and 1073 K, where the salt deposit (Na 2SO 4) is expected to be solid. Although the applicability of the proposed fluxing theory is justified for the high-temperature hot corrosion (HTHC) processes of pure metals and alloys, its validity has been contested for low-temperature hot corrosion (LTHC), particularly in Co-based binary and ternary alloys. Turbines operating in marine environments suffer from severe degradation of Co-based CoCrAlY coatings at relatively low temperatures (873–1073 K), where Na 2SO 4 salt depositing on the surface of blades is expected to be solid. Damage is manifested irregularly in the form of pits rather than as a broad frontal attack. Luthra et al. [56,64] demonstrated that a minimum p SO3 of the order of 10⫺5 atm is required in the gas phase to stabilize an Na 2SO 4CoSO 4 liquid salt at temperatures below the melting point of Na 2SO 4 (less than 1157 K). Under such conditions, a liquid melt of Na 2SO 4-CoSO 4 can form on cobalt-containing alloys by the interaction of cobalt oxide and Na 2SO 4 with SO 3 in the gas phase according to CoO(s) ⫹ SO 3(g) ⫽ CoSO 4(s,l)

(6.50)

1 1 Co 3O 4(s) ⫹ SO 3(g) ⫽ CoSO 4(s,l) ⫹ O 2(g) 3 6

(6.51)

where the underline implies that sulfate is present in solution at an activity less than unity. Similar to HTHC of alloys, the degradation of Co-based alloys by LTHC is also a two-stage process: an initial stage during which an Na 2SO 4CoSO 4 liquid forms on the alloy surface, followed by a propagation stage during which the reaction occurs by transport of various reactants through the liquid melt. Degradation results from rapid dissolution of Co and/or CoO/Co 3O 4 on the alloy surface by Na 2SO 4. This prevents the formation of a continuous, impervious, protective film of Cr 2O 3 and/or Al 2O 3 on the alloy surface. Based on detailed considerations of the mechanisms of transport of various oxidants through the liquid melt, a model for LTHC has been proposed [56]. This model has been developed on the basis of dissolution of the more noble metal or metal oxide in

370

Chapter 6

liquid salts. In the proposed mechanism, it is considered that the predominant oxidant at the scale–salt interface is SO 3, which rapidly diffuses inward by an S 2O 72⫺ /SO 42⫺ exchange reaction [S 2O 72⫺ ⫹ SO 42⫺ ⫹ SO 42⫺ → SO 42⫺ ⫹ S 2O 72⫺ ⫹ SO 42⫺]. At the scale–salt interface, Cr and/or Al are preferentially oxidized, and Co and/or cobalt oxide dissolves in the melt, leaving behind a porous film of Cr 2O 3 and/or Al 2O 3 penetrated by an Na 2SO 4-CoSO 4 liquid. At longer times, when the liquid salt gets saturated with Co, a steady state is reached wherein Co dissolving at the scale–salt interface migrates outwardly through the liquid as Co2⫹ ions by the exchange reaction 3Co 22⫹ /2Co3⫹ [3Co2⫹ ⫹ S 2O 72⫺ ⫹ 1/2 O 2 (dissolved) s 2Co3⫹ ⫹ 2SO 42⫺]. The location of the interface, where it forms oxide and/or sulfate of cobalt, and the resultant product morphology depend on the relative transport rates of reactants. The major reactions involved in the dissolution of the oxide and precipitation process are identified as follows: A. Dissolution only (at the scale–salt interface until the salt is saturated with Co) CoO(s) ⫹ S 2O 72⫺ ⫽ CoSO 4(l) ⫹ SO 42⫺

(6.52)

B. Dissolution at the scale/salt interface and precipitation in liquid salt under steady-state condition: (a) Dissolution Co2⫹ ⫹ 2e⫺ ⫹ 2Co3⫹ ⫽ 3Co2⫹

(6.53)

(from the (inward inner scale) diffusion through the liquid) (b)

Precipitation (at intermediate levels of SO 3, i.e., of the order of 10⫺4 to 10⫺3 atm at 1023 K): 3Co2⫹ ⫹

2 1 O 2(g) ⫽ Co 3O 4(s) ⫹ 2Co3⫹ 3 3

(6.54)

The reaction sequence during LTHC of a Co-30% Cr alloy covered with an Na 2SO 4 deposit exposed to an O 2-SO 2-SO 3 environment at 1023 K is shown schematically in Fig. 6.40 [56]. Even though CoCr 2O 4 formation is also likely, in the figure only Cr 2O 3 is shown for simplicity’s sake. Initially as soon as the alloy is exposed to the environment (Fig. 6.40a) both cobalt oxide and Cr 2O 3 form below the surface of the salt layer. Subsequently, cobalt oxide interacts with SO 3 from the environment forming CoSO 4 which gets dissolved in Na 2SO 4. Consequently, formation of a liquid phase with cobalt oxide leaves behind a

Alloy Oxidation

371

Figure 6.40 Schematic representation of the reaction sequence during low-temperature hot corrosion of a Co-30% Cr alloy exposed to O 2-SO 2-SO 3 environment where both Na 2SO 4-CoSO 4 liquid and Co 3O 4 are stable. At higher concentration of SO 3 where Co 3O 4 is unstable at the gas–salt interface, the outward migrating cobalt will form CoSO 4(s) or Co 3O 4 and CoSO 4(s) [56].

porous Cr 2O 3 film (Fig. 6.40b). Further reactions occur by inward migration of SO 3 and outward transport of cobalt. Inwardly migrating SO 3 oxidizes Cr to form Cr 2O 3 and SO 2. The released SO 2 further reacts to form both sulfides and oxides of cobalt and chromium (Fig. 6.40c). With increasing exposure time, the earlier formed sulfide is oxidized, generating S 2 or SO 2 which can migrate further inward

372

Chapter 6

to form additional sulfide. As a result, both the earlier formed pervious Cr 2O 3 layer and the inner sulfide layer grow; consequently, the inner scale–salt interface and the scale–alloy interface migrate inward (Fig. 6.40d). At a longer time of exposure when the salt at the gas–salt interface attains the maximum activity of CoSO 4 (as dictated by the gas composition), the outwardly migrating cobalt forms CO 3O 4 (Fig. 6.40e). So, during LTHC of cobalt alloys, high degradation rates result from the continuous dissolution of cobalt and/or cobalt oxide at the scale– salt interface and precipitation of cobalt oxide occurs near the salt–gas interface. Such a process makes the initially formed Cr 2O 3 film discontinuous and pervious at the alloy–scale interface. Since the solubility of Cr 2O 3 or Al 2O 3 in the melt at the prevailing atmosphere is low, chromium or aluminum remains below the original alloy–scale interface as a pervious oxide layer containing Cr 2O 3 and/or Al 2O 3 and/or CoCr 2O 4. Thermodynamic considerations suggest that Ni-based alloys containing no cobalt are not supposed to undergo LTHC as the Co-based alloys do at intermediate concentrations of SO 3. This is justified by the fact that at temperatures above the melting point of Na 2SO 4-NiSO 4 eutectic (944 K), NiSO 4(s) is not expected to form under normal gas turbine operating conditions (pSO3 ⬍ 10⫺4 atm) even though Na 2SO 4-NiSO 4 liquid may form. However, at high SO 3 levels, where NiSO 4(s) is stable at the scale–gas interface, nickel dissolution and sulfate precipitation may occur, resulting an accelerated attack. In the above discussion, acid and basic fluxing mechanisms have been considered independent of each other. But in reality the situation may not be so simple, and interactive effects of both the processes leading to catastrophic attack are often encounted with multicomponent alloy systems under a vaguely defined atmosphere. The possible salt fluxing reactions with Na 2SO 4 deposit for basic and acidic processes are illustrated in Table 6.6 [58]. In this table, category A includes the basic processes that occur due to production of oxide ions in Na 2SO 4 as a result of removal of oxygen and sulfur from the melt by the alloy. As a result, the attack may occur either due to solution of the oxide in Na 2SO 4 or solution and reprecipitation. In both cases, the attack is not self-sustaining but rather is controlled by the amount of Na 2SO 4 deposit unless the gas phase contains SO 3. Category B illustrates the Rapp-Goto electrochemical model [62], whereby the attack is explained not in terms of sulfur removal but in terms of the existence of a negative solubility gradient of the corrosion product at the oxide–salt interface in the Na 2SO 4 film. Categories C, D, E, and F include the different acidic processes wherein C and D refer to gas-induced acidic fluxing processes involving dissolution and reprecipitation, with the acidic component supplied by the gas phase. Category E illustrates the Rapp-Goto concept for acidic melts. Category F demonstrates the alloy-induced acid fluxing processes whereby the acidic component comes from the alloy undergoing degradation.

Table 6.6 Possible Salt-Fluxing Reactions for Na 2SO 4 Deposits on Alloys [58] Basic Processes A. Dissolution of reaction product (i.e., AO) due to removal of sulfur and oxygen from Na 2SO 4 by the metal or alloy: SO 42⫺(sulfate deposit) →

1 3 S 2(for reaction with alloy) ⫹ O 2(for reaction with alloy) 2 2 ⫹ O2⫺(for reaction with AO)

Reaction between AO and oxide ions can follow two courses: (i) Continuous dissolution of AO A(alloy) ⫹

1 O 2 ⫹ O2⫺ → AO 22⫺ 2

Na 2SO 4 is converted to Na 2AO 2 and the attack is dependent on the amount of Na 2SO 4 initially present. (ii) Solution and reprecipitation A(alloy) ⫹

1 O 2 ⫹ O2⫺ → AO2⫺(solution) → AO(precipitate) ⫹ O2⫺ 2

A supply of SO 3 is required in order the attack can proceed indefinitely; otherwise the attack will stop when melt becomes sufficiently basic at the precipitation site. B. Solution and precipitation of AO as a result of a negative gradient in the solubility of AO in Na 2SO 4 Acidic Processes Gas phase–induced: C. Formation of ASO 4 in Na 2SO 4: A(alloy) ⫹ SO 3 ⫹

1 O 2 → A2⫺ ⫹ SO 42⫺ 2

Continuous solution of ASO 4 in Na 2SO 4 requires continuous supply of SO 3 and O 2 from the gas phase. D. Solution and precipitation of AO in Na 2SO 4 due to reduction of SO 3: A(alloy) ⫹ SO 3(from gas) → A2⫹ ⫹ SO 32⫺(in melt) A2⫹ ⫹ SO 32⫺ ⫹

1 O 2(from gas) → AO(precipitate) ⫹ SO 3 2

E. Solution and precipitation of AO as a result of a negative gradient in the solubility of AO in Na 2SO 4. Alloy phase–induced: F. Solution of AO in Na 2SO 4 modified by a second oxide from the alloy (i.e., BO 3) Modification of Na 2SO 4 by BO 3: B(alloy) ⫹

3 O 2 ⫹ SO 42⫺ → BO 42⫺ ⫹ SO 3 2

Solution reaction for AO; Na 2SO 4 becomes enriched in ABO 4: A(alloy) ⫹ B(alloy) ⫹ 2O 2 → A2⫹ ⫹ BO 42⫺ Solution and precipitation: A(alloy) ⫹ B(alloy) ⫹ 2O 2 → A2⫹ ⫹ BO 42⫺ → AO ⫹ BO 3 Precipitation of AO in Na 2SO 4 as a result of BO 3 loss from Na 2SO 4 permits substantial attack with small amounts of Na 2SO 4.

373

374

Chapter 6

In conclusion, it can be stated that degradation of multicomponent alloys by hot corrosion is extremely complex. It is agreed that such attack is a two-stage process comprising initiation or incubation and propagation. During the initiation stage, the alloy gets conditioned by the salt in the gaseous environment to degrade via a particular mode as discussed above. In the identification of the operative hot corrosion mechanisms it is important that the exposure conditions be well defined in terms of alloy composition, microstructure, salt composition, thickness of the salt, temperature, thermal cycling, gas composition, velocity, erosion, and so on.

6.8 PROTECTIVE COATINGS FOR HIGH-TEMPERATURE APPLICATIONS The continuous demand for high-temperature alloys to operate at higher temperatures in aggressive environments paved the way to the development of alloys with steadily increasing strength coupled with other favorable mechanical and corrosion resistance properties. To be oxidation-resistant at high temperatures, an alloy must meet two essential requirements: it must form a surface oxide that thickens slowly, and this oxide layer must remain adherent to the alloy substrate under all service conditions. The specific difficulties in the protection of metallic components are contributed not only by the extreme reactivity of the aggressive environment but by the very high mobility of the cations or oxygen, leading to rapid diffusion, coupled with instability of the protecting layer during the service life. Resistance to high-temperature corrosion of metals/alloys can be achieved by the establishment and maintenance of an impervious, stable, inert, adherent, protective layer on the substrate during the service period. The attainment of such a protective layer is possible either by direct application of, for example, a ceramic coating or, more usually, by interaction between the environment and the metallic surface. Accordingly, the role of coatings is to provide the most effective protective barrier layer separating the two reactants, i.e., the alloy and its environment of functioning, with long-term stability and resistance to cracking or spallation under the mechanical and thermal stresses induced during the operation of the component. Historically the motivation for high-temperature coating development has been linked to development of alloys for gas turbine components, where there is a steadily growing demand for the increased alloy strength at high temperatures coupled with increased corrosion resistance. So it is considered worthwhile to briefly trace the history of high-temperature alloy developments and their limitations in practical applications before the different issues related to coatings are discussed. The early metallic materials used at significantly higher temperatures were

Alloy Oxidation

375

cast or wrought ferrous alloys involved in furnaces, kilns, and ovens for a variety of industrial and domestic purposes. These suffered oxidation and hot corrosion, leading to heavy metal loss by scaling and at times catastrophic failure of components. Subsequently, improved resistance to oxidation at moderate levels of strength requirement was obtained by alloying cast irons with high silicon and later by austenitic alloys containing 18–25 wt % Cr and up to 25 wt % Ni together with small amounts of Si and Al. The high-temperature capability of iron-based alloys increases with increasing chromium content. The strength of these alloys also increases at higher temperatures with increasing chromium content. Therefore, the two essential requirements for structural ferrous materials—strength and corrosion resistance—are compatible, as illustrated in Fig. 6.41a and b, respectively. However, the temperature capability of these alloys is limited to less than 1273 K where a combination of strength and oxidation rate invariably becomes unacceptable. Electrical heating elements for domestic equipments and industrial furnaces, which require ductile wrought alloys, also undergo degradation. For the relatively small cross-section of the heating elements, the importance of scale adhesion to avoid generation of local hot spots has long been recognized. The incorporation of reactive elements (e.g., Ce, Y, La, Zr, or Hf) or their oxide dispersions to minimize scale flaking was first used in such alloys (refer to Sec. 6.6) and has more recently found applications in the coatings of high-strength superalloys. The widespread use of steam power both for electricity generation and for rail and marine transport led to the progressive increase in temperature in the search for higher thermal efficiency, but the involved temperature remained generally modest in comparison with those for the gas turbines. Failures of boiler and superheater tubings under the influence of internal pressure led to the recognition of creep as an important mechanism. The gas turbine has been known in principle for many years. Such engines are used in a wide variety of applications. The most demanding of these in terms of material durability and reliability requirements under severe service conditions are aircraft propulsion, marine propulsion, and electric power generation. But successful operation of the engines was limited by the difficulty of matching the efficiencies of the compressor and turbine to make the combination self-sustaining as well as by the limitation on the operating temperature imposed by available materials of construction. Gradually, stationary gas turbines producing shaft power were recognized to be thermally less efficient than steam turbines. The outbreak of World War II stimulated further development in use of the concept of turbine exhaust as a jet for aircraft propulsion. Soon the shortcomings of the high-temperature materials for turbine blades were recognized, and some researchers concentrated on air cooling of the blades using best available ironbased alloys, whereas others tried the use of high-alloy austenitic steels and

376

Chapter 6

Figure 6.41 Influence of chromium concentration on (a) oxidation resistance (scaling rate, 20 mg/cm2 in 1000 h) and (b) creep rupture strength (t ⫽ 104 h, T ⫽ 873 K) of ironbased alloys [65].

Alloy Oxidation

377

nickel-based alloys. In the process, Nimonic-80, the initial γ ′-hardened creepresistant Ni-Cr alloy, was invented and proved to be the forerunner of the extensive family of nickel-based superalloys which now provide the vast majority of materials used for the highest temperature gas turbine blades. The nickel- and cobalt-based superalloys are no doubt capable of developing higher strength at higher temperatures than the iron-based alloys, but such high strength is achieved only by reduction of their chromium content. The lower chromium content is required not only to maintain microstructural stability but to allow additions of alloying elements such as aluminum, titanium, and niobium for optimum creep properties. The influence of chromium content on the oxidation resistance and creep rupture strength of some of the currently used superalloys is shown in Fig. 6.42a and b, respectively [65]. Gradually, iron-, nickel-, and cobalt-based superalloys found applications in most of the high-temperature components of the aircraft jet engine, such as turbine blades, disks, combustion systems, tail pipes, and so on, and provided a major step forward in the development of shaft power gas turbines for stationary and transport purposes (particularly for naval ships). There is no doubt that the development of aircraft gas turbine engines has been spectacular over the past five decades; the thrust-toweight ratio of about 2:1 for the earliest engine has been increased to about 7: 1 in a modern military aircraft. Such gain has been achieved predominantly by enhanced thermodynamic efficiency as a consequence of the increased temperature of the inlet gas to the tubing. However, the major credit for such advances must go to the design and development engineers engaged in superior materials development, for whom the permissible temperature of operation for a given lifetime under specified mechanical stresses could be substantially increased. However, such significant developments in materials are reaching an asymptote because these alloys can be used at a maximum metal temperature of 1323 K as their melting points are of the order of 1523–1573 K. Another significant development is that despite the fact that use of the superalloys has allowed the gas turbine operating temperature to increase by 423 K since the 1960s, developments in blade cooling technology over the same period have allowed an additional increase in temperature of 573 K, as illustrated in Fig. 6.43 [65]. The successful performance of most high-temperature metallic materials in service conditions is often dependent on the formation and maintenance of an impervious, protective oxide scale. In the selection of high-temperature material for a specific application, it is generally a design requirement that the mechanical properties of the alloy substrate (e.g., strength, creep, fatigue) remain unaltered by the compositional and structural changes that may result from the degradation process caused by the environment during the service period of the component. But the alloy compositions and microstructures that provide optimum mechanical properties often do not provide the satisfactory high-temperature corrosion resistance property. It has already been pointed out that the development of alloy

378

Chapter 6

Figure 6.42 Influence of chromium content on (a) oxidation resistance (T ⫽ 1123 K) and (b) creep rupture strength (σ ⫽ 200 MN m⫺2, t ⫽ 104 h) of some superalloys [65].

strengthening mechanisms is such that higher strength can only be achieved at the expense of oxidation resistance. The situation is more adverse for refractory metals such as Mo, W, Nb, and Ta. These metals are characterized by very high melting points, but even so their alloys cannot be used in oxidizing environments without additional protection due to their poor resistance to oxidation. During oxidation of Mo and W, volatile oxides (MoO 3, WO 3) are formed; while Nb and Ta, apart from developing poor protective scales have high affinity for interstitial elements such as O, N, C, and H. These interstitial elements easily dissolve in the metals and form ordered and

Alloy Oxidation

379

Figure 6.43 Increase in operating temperature of Rolls Royce turbine, illustrating the contribution of blade cooling [65].

sometimes metastable martensitic phases that adversely affect the alloy physical and mechanical properties and oxidation processes. However, these metals can be successfully utilized at high temperatures with the use of suitable protective coatings. The scaling resistance of metallic materials is primarily dependent on the protective properties of the scales formed on their surfaces by reaction with the environment. At high temperatures, however, the scale layer frequently provides an inadequate protection to the underlying alloy substrate, as a result of which it undergoes degradation either partially or at times catastrophically within a short period of its expected service life. Accordingly, the service life of an alloy of satisfactory mechanical properties may be extended by coating the alloy with a special protective layer. Such a layer should have protective properties characterized by satisfactory adherence, compactness, low mobility of the reactants (i.e., constituents of the alloy and the aggressive environment as well as the coating constituents), etc. Therefore, the basic purpose of a high-temperature coating is to act as an effective solid-state barrier between the oxidants and the alloy, thereby decreasing the rate of degradation of the metallic component, i.e., to increase the service life of the underlying alloy. Apart from this, the coatings also help to minimize the consumption of the critical and scarce raw materials used in the manufacturing of alloys. So the best technical solution to combat the degradation of structural materials in oxidizing environments is to protect the same by means of surface layers consisting of high-melting, thermally stable, and chemically resistant oxides.

380

Chapter 6

In an oxidizing environment, a metallic material may be protected from degradation in two ways: by alloying with suitable elements or by coating. In either case, the objective is the same, i.e., to form or obtain a layer on the metallic surface that acts as a barrier separating the two reactants (the underlying metallic substrate and reacting gas), thereby minimizing the reaction between them. The high-temperature corrosion resistance of numerous alloys in practice is provided by scales consisting of chromia, alumina, and silica or more complex oxides of these, e.g., various spinels. Such scales are achieved by preferential or selective oxidation of chromium, aluminum, or silicon present as constituents of the alloys or coatings. The oxide usually preferred at less than 1273 K is Cr 2O 3 (because at such temperatures volatile CrO 3 formation becomes appreciable at or near atmospheric pressures), whereas Al 2O 3 and SiO 2 are chemically more stable at higher temperatures. Formation of such oxide layers can be considered in various ways [66], as depicted in Fig. 6.44. In the consideration of coatings, it is important to recognize the coating and the underlying substrate as an integral system under the operating conditions of the component. Prefabricated oxides can be developed with maximum chemical inertness as shown in Fig 6.44a, but they usually suffer from unfavorable mechanical properties, such as brittleness. Moreover, in most such cases these are physically incompatible with the metallic substrate. If chemical

Figure 6.44 Schematic illustration of the possible situations for protective oxide layer formation rendering oxidation resistance and mechanical properties of metallic materials at high temperatures [66].

Alloy Oxidation

381

inertness of the coating is put on the priority list along with improved mechanical properties of the alloy substrate, one may also consider the use of nonoxide ceramic coatings, keeping in mind that components of the coating material will diffuse into the substrate. Due to reactivity of such coating constituents with the substrate metallic elements during long-term application at high temperatures, a sequence of intermetallic phases may be formed. Such a situation is illustrated in Fig. 6.44b. Embrittlement of the substrate alloy by solid-state reactions with the coating material is often a possibility during a long exposure period and should be avoided since the composition and microstructure of the substrate alloy is always optimized to achieve the desired mechanical properties. The unique technical solution to the problem is the combination of a high-strength alloy with a highly alloyed coating of the preferentially oxidizable alloy constituents having the capability to form self-healing layers as illustrated in Fig 6.44c. Since it is recognized that the addition of elements like Cr, Al, and Si in sufficient quantity to the substrate alloys seriously affects the mechanical properties of the alloys, these elements are often used in limited quantity in the alloy manufacturing process, which provides limited resistance to the substrate alloys as shown in Fig. 6.44d. On the other hand, if mechanical properties of the alloy become the prime concern, and to eliminate the chance of embrittlement caused by aluminum and silicon, alloys having the best mechanical properties could be produced without the addition of these elements, thus allowing totally uncontrolled degradation of the alloys under service conditions as illustrated in Fig. 6.44e. Therefore, the main goal of researchers and developers during recent decades has been to achieve a coating–alloy combination of maximum or full chemical resistance on the exposed surfaces of the metallic material with unaltered mechanical properties of the substrate during the expected service period of the component.

6.8.1

Requirements of Coating–Substrate System

In oxidizing environments at high temperatures, a coating in general owes its oxidation resistance to the formation of a protective oxide layer. Therefore, in selecting coating materials, it is important that the coating–substrate system meet the following requirements [2,67,68]: 1. The coating should be chemically and thermally stable (forming an integral coating–metal/alloy system) during service life of the component. 2. It should have properties compatible with those of the metallic substrate. 3. The rate of interdiffusion of the elements in the integral system (i.e., between coating and substrate alloy) must be slow during the desired service life. 4. The protective layer and the metallic substrate should have matching thermal expansion coefficients to avoid cracking and exfoliation of the coating during thermal cycling.

382 5. 6. 7.

8.

Chapter 6 A protective coating should exhibit some mechanical ‘‘elasticity’’ under operating conditions to accommodate creep and plastic deformation. A coating material should resist damages from impact, erosion, and abrasion depending on the specific applications of the metallic components, It should exhibit a spontaneous ‘‘self-healing’’ property for self-repair in case failure occurs due to cracking or spallation of the layer. So the coating should act as a reservoir for the highly oxidizable metallic constituent/constituents for early development of a protective scale, It should be relatively easy to apply the coating on substrates, and the defects that may occur during handling of the component should be reparable without accompanying adverse effects on the sound neighboring areas.

Consequently, the development of a truly satisfactory coating that meets all of the above requirements is a difficult task. Accordingly, compromises are often made, depending on the specific application of the coated material in a particular environment. Moreover, because of coating–environment and coating–substrate reactions, the structures of the actual protective coating systems are complex. Multilayered coating systems are often found to be the most successful; in practice, even single-layer coatings often become multilayered during service due to coating–substrate and coating–environment reactions (e.g., silicide coatings on refractory metals such as Mo, W, etc.). So, in the selection of protective systems for components of high-temperature utility, three main factors deserve consideration: the service or application conditions of the component, the structural alloy, and the system of protection itself. The design of the component sets the service stresses, temperature, thermal cyclings, and so forth. The alloy properties are governed by composition, microstructure, and the processing steps, which control the high-temperature stability. Finally, the selection of a suitable protective system is decided by its resistance to environmental effects. The possible interactions of these three constituents as illustrated in Fig. 6.45 serve as a guide in the design, formulation, and evolution of the best possible coating–alloy system that guarantees the expected service life of the component [68]. Stability and Compatibility of Oxides Since the high-temperature oxidation resistance of metallic materials is contingent on the development and maintenance of a protective oxide layer, it is pertinent to have a first-hand knowledge of the stability, diffusion characteristics, compatibility with the substrate, evaporation, and so forth, of such oxides to serve as a basis for the assessment of potential coating materials. The protective layer should consist of an oxide or mixture of oxides with the maximal stability. On the basis of a thermodynamic property like free energy of formation, the oxides of primary interest include BeO, MgO, CaO, Al 2O 3, Y 2O 3, La 2O 3, SiO 2, TiO 2,

Alloy Oxidation

383

Figure 6.45 Interactions between coating, substrate, and environment [68].

ZrO 2, HfO 2, ThO 2, and Cr 2O 3 (as single oxide), their complex oxides, spinels, and so forth. Of these, CaO and La 2O 3 hydrate rapidly in air; BeO is markedly toxic; and pure ZrO 2 undergoes polymorphic transformations. However, this transition can be eliminated by stabilizing ZrO 2 in a cubic form with the addition of other oxides, such as CaO, MgO, and Y 2O 3. It is pertinent to point out that most refractory oxides undergo chemical reactions among themselves at temperatures well below their individual melting points, with the formation of low-melting eutectic liquids. Accordingly, the useful temperature range of their applications becomes limited. Cr 2O 3 is stable only at temperatures below 1273 K in atmospheric oxygen pressure. Diffusion in Oxides For a protective oxide film, its effectiveness in combating further degradation of the underlying metal/alloy is usually determined by the rate of solid-state diffusion through the film. The most effective diffusion barriers are provided by oxides having the slowest rates of diffusion of the reactants. Accordingly, it is essential to have knowledge about the diffusion rates of both cations and oxygen. A comparative plot of self-diffusion coefficients of cations and oxygen in some simple oxides of interest [2] is illustrated in Fig. 6.46. This figure clearly demonstrates that oxides like CaO, MgO, and Al 2O 3, which exhibit small deviations from stoichiometry, have the smallest diffusion coefficients. On the other hand, oxides like

384

Chapter 6

Figure 6.46 Comparative plot of cation diffusion coefficients in Al 2O 3, MgO, CaO, Cr 2O 3, NiO, CoO, and oxygen diffusion in stabilized zirconia and SiO 2 [2].

NiO and CoO, which are characterized to possess appreciable nonstoichiometry, exhibit high rates of diffusion due to their high concentrations of point defects. Similarly, stabilized zirconia with a large concentration of oxygen point defects provides a poor diffusion barrier. To the contrary, the transport of molecular oxygen through SiO 2 is much lower than Al ions through Al 2O 3 above 1773 K. Accordingly, in generalized form, it may be stated that simple oxides with large ionic character and small deviations from stoichiometry are expected to act as better protective layers. To the contrary, oxides like FeO, NiO, and CoO cannot act as effective diffusion barriers unless their point defect concentrations are decreased by doped element oxides. During high-temperature exposure of alloys in oxidizing environments, it is a common experience that various types of spinels are also formed along with

Alloy Oxidation

385

Figure 6.47 Comparative plot of diffusion coefficients in some spinels [2].

simple oxides. These spinels are contemplated to serve as effective diffusion barriers since they exhibit low diffusion rates [2] in comparison with simple oxides, as illustrated in Fig. 6.47. The above discussion clearly brings out the following characteristic properties of the three most commonly preferred alloying elements, Cr, Al, and Si, which can form stable, self-healing, oxide layers on alloy surfaces in providing protection under oxidizing environments: 1. Transport rates of aluminum cations are the slowest among the three simple oxides in the temperature range below 1673 K. 2. In the cases of Cr 2O 3 and Al 2O 3 layers, the diffusing species are the respective cations, whereas in the case of pure SiO 2, it is mainly nonionized oxygen. 3. At temperatures above 1573 K, the mobility of the diffusing species, i.e., oxygen in SiO 2, is the lowest among the three simple pure oxides.

386

Chapter 6

Such observations suggest that the addition of aluminum within the permissible limit that does not cause embrittlement to high-temperature alloys should provide sufficient protection by self-forming oxide layers at temperatures below 1673 K. Chromium is known to be the mildest of the three elements (i.e., Al, Cr, and Si) insofar as its influence on alloy embrittlement is concerned. But to guarantee self-forming oxide layers, there is a need to add more than the minimum highest permissible limit (20 wt %). Moreover, chromium as the single alloying element provides only limited protection above 1273 K, due to formation of its volatile oxide (CrO 3). Accordingly, it is desirable to use aluminum (in low concentrations) in combination with higher contents of chromium to guarantee protection above 1273 K. Another interesting observation is that at very high temperatures (above 1773 K), the transport rates of molecular oxygen through SiO 2 layer are low in comparison with aluminum ion diffusion through Al 2O 3. This provided impetus for the development of silicides and silicon ceramics as suitable coating materials in oxidizing environments at extremely high temperatures (1773–2073 K). However, it is to be recognized that the element silicon, apart from causing embrittlement to high-temperature alloys, similar to aluminum, has an additional negative effect on the coating–alloy performance. SiO 2 can form low-melting eutectics for which the use of silicon in iron-based alloys as a stable oxide forming addition is avoided at performance temperatures above 1273 K.

6.8.2

Coating Methods

In the present state of knowledge, the most successful coatings are the metallic coatings, particularly the intermetallics, wherein one constituent preferentially oxidizes to produce a protective scale. Under service conditions of metallic materials, protection by coatings from subsequent degradation is obtained in two different ways: 1.

2.

Constituents of the coating material may react with the corrosive constituents of the environment under service conditions of the metallic material, forming product layers of positive protective properties, and Layers of coating material can be applied over the metallic component that mechanically isolate the underlying substrate from the corrosive environment.

Materials used for protective coatings on metal/alloy surfaces are broadly classified into four groups: 1. 2.

Scaling-resistant metals or alloys, which on reaction with the environment develop protectiveness Intermetallic compounds of the type silicides, aluminides, borides, etc.

Alloy Oxidation

387

3. Ceramic coatings 4. Noble metals, which under operational conditions do not form compounds with the aggressive constituents of the environment. During service of the coating–alloy system at high temperatures, interdiffusion occurs between the alloy and the coating constituents. This leads to significant changes in the composition and microstructure of both alloy and the coating material in the vicinity of the coating–alloy interface. Such interdiffusion in many instances leads to an increase in adherence of the coating to the substrate and to an improvement in its mechanical properties. Accordingly, coated alloys are often subjected to a special thermal treatment known as ‘‘diffusion annealing.’’ During this process, a partial intermixing of the alloy constituents and the coating material takes place, owing to which a gradual compositional change occurs. This is particularly important in the cases of alloys used under thermal cycling conditions because it accommodates differences in thermal expansion of the coating and the substrate, thus ensuring improved stability. On the other hand, a rapid diffusion process may lead to reduction or complete exhaustion of the element providing protection during the short service period of the alloy. At the same time, mechanical properties of an alloy may also get adversely altered due to a change in its composition. A particularly disadvantageous situation may arise due to formation of chemical compounds having a negative effect on the mechanical properties of the coating–alloy system. A large number of methods are available for developing coatings on alloy substrates. These include electroplating, hot dipping in molten metals or fused salts, spraying (oxyfuel and plasma techniques), slurry spraying, cladding, enameling, vapor deposition or chemical transport reactions (pack cementation, fluidized-bed technique, pyrolitic deposition), vacuum evaporation, etc. Depending on the technique adopted for the production of protective coatings with enriched content of the stable oxide–forming elements, they are broadly classified into two groups: diffusion coatings and overlay coatings. Diffusion coatings are formed through diffusional interactions between the coating material and the substrate alloy. On the other hand, overlay coatings do not involve a direct alloying reaction with the substrate, though often a diffusion step is included to improve the bonding between the coating and the substrate. Some methods for producing the two types of coatings are discussed below. Diffusion Coatings Pack Chromizing. The pack chromizing process by pack cementation [69] as developed during World War II is still often used to increase the service life of stationary gas turbine blades and is considered as a representative of the diffusion technique. Coatings so developed are considered to be among the most effective

388

Chapter 6

protective systems to combat high-temperature degradation. In this process, the metal/alloy component to be coated is packed in a powdered mixture (cement) of the coating element (Cr), a small amount of easily decomposable activator (e.g., NH 4Cl) to produce the gas phase and an inert ballast material (usually Al 2O 3) to prevent sintering. Reaction is carried out under inert or hydrogen atmosphere at elevated temperatures (1073–1373 K) for certain hours, depending on the nature and thickness of the coating desired. The protective mechanism of such coatings is analogous to that of the Cr-rich superalloys and depends on their ability to develop a compact, dense, coherent oxide coating (Cr 2O 3) as a diffusion barrier against further oxidation or sulfidation. As a consequence of the interaction of hot gas and fuel ash deposit, such coatings undergo a continuous consumption that is further accelerated by thermal and mechanical stressing. During service, at times, the creep stress may develop above a critical limit, which may cause cracking of the protective scale at a faster rate than the regrowth rate of the oxide layer. Chromium’s progressive impoverishment through successive Cr 2O 3 layer formation and its diffusion into the substrate alloy ultimately leads to a situation of insufficient chromium content in the coating for the development of a subsequent protective barrier layer. Thus, the chromium transport mechanism plays the most important role in the life expectancy and ultimate failure of the protective system. It is pertinent that in the establishment of a diffusion coating layer, one of the prerequisites is to have sufficient solubility of the coating element in the substrate alloy and the resultant solid solution should have good physical compatibility with the substrate without affecting its mechanical properties to a large extent. Such a situation is ideal for chromizing iron-based alloys and adherent, nonbrittle coatings can easily be achieved with chromium contents of more than 30 wt% at the surface. Pack Aluminizing. Similar to pack chromizing, in this process [70,71,75] the component is embedded in a powder mixture (cement) containing Al or Al-rich metallic powders (e.g., Ti-Al, Ni-Al, Cr-Al alloy powders), inert filler, Al 2O 3 to prevent the sintering of the pack, and 1–2 wt% ammonium halide activators. Subsequently, the whole assembly is heated to a temperature of 1073–1373 K in H 2 or Ar atmosphere. In this temperature range, aluminum halides are formed that diffuse through the porous pack and react at the surface of the alloy component to deposit aluminum either by disproportionation of aluminum halides or by hydrogen reduction reaction. The deposited aluminum diffuses into the substrate to form NiAl coatings. The aluminum concentration profile during the progressive steps in the formation of the aluminide coatings by pack aluminizing is illustrated schematically in Fig. 6.48. The formation of aluminide coatings can be broadly categorized as low- or high-activity processes, depending on the preferential diffusion of nickel or aluminum that occurs in the different layers formed during the heat cycles associated with the techniques.

Alloy Oxidation

389

Figure 6.48 Schematic diagram of aluminum concentration profile during pack aluminizing.

In the low-activity pack process, direct formation of NiAl compound takes place in a single thermal cycle during the coating operation, involving preferential outward diffusion of nickel in the temperature range 1273–1373 K. The structural development of NiAl-type coating on a nickel-based superalloy in a low-activity process is illustrated schematically in Fig. 6.49. In the high-activity process,

Figure 6.49 Schematic illustration of an NiAl type of coating structure on a nickelbased superalloy in a low-activity process [70].

390

Chapter 6

Figure 6.50 Schematic illustration of formation of NiAl coating by diffusion annealing from Ni 2Al 3 layer on a nickel-based superalloy in a high-activity process [70].

which is used more often, Ni 2Al 3 phase is initially formed with preferential diffusion of aluminum during aluminizing treatment in the temperature range 973– 1123 K. Subsequently, during its diffusion annealing in argon atmosphere at temperatures of 1273–1373 K in the absence of aluminum source, Ni 2Al 3 transforms to NiAl by reacting with the substrate. The development of an NiAl coating during diffusion annealing, by interdiffusion between a layer of Ni 2Al 3 and the nickel-based superalloy substrate, is illustrated schematically in Fig. 6.50a–c. The final layer of NiAl on the alloy substrate, as depicted in Fig. 6.50c, consists of three distinct regions. The external zone is almost as thick as the initial Ni 2Al 3 layer, containing various precipitates that existed in the initial layer. The central region is devoid of precipitates. The internal zone consists of precipitates similar to those observed in the internal zone of coatings obtained by a ‘‘low-activity’’ aluminizing treatment (Fig. 6.49). In fact, the last two zones may be considered to be a low-activity coating, with the initially formed Ni 2Al 3 layer virtually playing the role of cement. So it is the external zone containing precipitates formed in the initial Ni 2Al 3 layer that distinguishes the high-activity NiAl coatings from the low-activity ones. The typical microstructures of the two types of aluminide coatings formed on U-700 alloy by high- and low-activity processes are illustrated in Fig. 6.51a and b, respectively. The structural features of the coatings can be explained in terms of the diffusion mechanism wherein outward diffusion of Ni through β-NiAl phase occurs during the entire coating process in a lowactivity mode, whereas inward Al diffusion followed by outward Ni diffusion during the subsequent heat treatment takes place in a high-activity mode. However, it is to be noted that the composition and microstructure of the coatings are dependent on the composition of the substrate alloy, making it necessary to optimize the process parameters for a particular alloy. This implies that the coat-

Alloy Oxidation

391

Figure 6.51 Typical aluminide coating microstructures formed on U-700 by (a) highactivity process and (b) low-activity process. Magnification 1000⫻, both parts reduced to 50% for reproduction [71].

ings are generally tailor-made for a given alloy composition. Significant improvements in the oxidation and hot corrosion resistance of aluminide coatings can be achieved by predepositing of a noble metal, specifically platinum and/or rhodium. The noble metal is generally electroplated as a thin (0.40 µm) layer before the aluminizing process. In the low-activity process for nickel-based superalloys, to avoid the deleterious effect of titanium, which is often present in the alloy, a suitable predeposit of titanium-free Ni-based alloy is often recommended to trap Ti in the form of Ti(C,N). Similarly, in a high-activity process, predeposit of a

392

Chapter 6

suitable Ni-Cr alloy followed by aluminizing produces a coating with a superficial region rich in chromium, which imparts superior oxidation resistance. Overlay Coatings Weld Overlays. The most important type of overlay coating is weld overlay, which is the most widely applied of all the methods for bulk coatings production in largest tonnage. It is often used to protect the inner surfaces of reaction vessels in chemical engineering processes. In this process [72], the deposit is laid down by melting the coating alloy on the work surface by gas or arc welding processes. The coating material is supplied in the form of powder, paste, rod, strip, or wire. Many of the standard welding processes, such as oxyacetylene, manual metal arc (MMA), metal inert gas (MIG), tungsten inert gas (TIG), submerged arc, etc., in manual, semiautomatic, and fully automatic modes, are often employed to develop such coatings on components ranging from small intricate shapes to large areas of flat or cylindrical shape. Roll Bonding and Coextrusion. Metallurgical methods of bonding protective layers to the outside surfaces of components without resulting fusion at the interface are widely used in industry. In such techniques, the coating is bonded to the substrate by solid-state methods that rely on a combination of surface cleanliness, temperature, and pressure to generate an intimate atomic contact and interdiffusion, which produce a true metallurgical bond, often usually stronger than the parent metal itself. High-rate techniques [72] comprise explosive cladding and electromagnetic impact bonding (EMIB), while a slower rate process utilizes hot isostatic pressure (HIP), to bond both powder and solid layers onto a component. Medium rate methods involve processing times of the order of 1 min and comprise such processes as roll cladding and coextrusion. The virtue of these methods is that the coating material can be carefully controlled and must itself have a reasonable ductility, so that it is unlikely to adversely affect the mechanical properties of the component. Explosive cladding is a solid-phase welding process in which bonding is produced by an oblique, high-velocity collision between the two pieces to be joined, the required force for deformation being supplied by chemical energy. HIP is a diffusion bonding process in which a metallurgical bond is formed by diffusion across the interface. In EMIB, the force is applied to the component by an intense magnetic field developed by a sudden surge of current through a coil. Coextrusion is most frequently used for products of rod and tubular form, particularly for creep-resisting steel bodies with internal or external cladding of stainless steels and nickel-based alloys. Ion Implantation. This technique is of very recent origin and is carried out in a fairly high vacuum chamber. The process involves bombardment of the metal/ alloy surface with a high-energy beam of ions of the chosen element to be implanted. The ions are accelerated to an energy of about 100 keV, but their penetra-

Alloy Oxidation

393

tion into the surface is limited to less than 0.5 µm with an approximately Gaussian distribution. Heating of the substrate by such interactions is minimal; once the ions are implanted, no diffusion occurs. The fraction of the implanted ions can be as high as 0.30 locally. In high-temperature applications, its principal use has been limited to the implantation of reactive elements such as Y, Ce, La, etc., into metal/alloy substrate. Since the implanted element is confined initially to a nearsurface layer, its subsequent redistribution during oxidation can provide unique information about atomic migration mechanisms. This technique has remained as a research tool and has yet to find industrial applications. Flame and Plasma Spraying. These methods are similar to the weld overlay techniques. They are generally based on processes [73,74] by which a metallic or nonmetallic powder or wire is injected into the flame or plasma, where it melts down to form small molten droplets. These droplets are then projected to the metal/alloy surface to be coated, splatting and freezing on impact. The integrity of the coating depends on atomization, melting point of the particles, degree of oxidation of the droplets, and velocity at impact. In plasma spraying (in atmospheric air), a direct-current electric arc is struck between the nozzle and the electrode, while a stream of mixed gases (commonly used are nitrogen, hydrogen, argon, and helium or their mixtures) is passed through the arc. This results in dissociation and ionization of the gases, thereby producing a high-temperature plasma (temperatures up to 16273 K) stream from the gun nozzle, although in practice most coatings are deposited with a flame temperature in the range of 6273–11273 K. The plasma torch acts as a high enthalpy heat source and accelerates the powders to velocities upto 300 m/s. In the short residence time of a few milliseconds, the powders transform to molten droplets, which hit and flatten on the component to be coated. By repeated movement of the gun, the coating is built up layer by layer. A cross-sectional view of the plasma spray gun head is shown in Fig. 6.52 [73]. However, the coatings formed by this technique are generally porous (porosity in the 2–10% range) and their bonding to the substrates is often not satisfactory. Spraying in the air causes oxidation of the powders due to turbulent mixing of the ambient air with the plasma gas. For improved bonding, one approach is to use a first layer of a material that undergoes an exothermic reaction with the substrate, thus developing a metallurgical bond. The intended coating material is then sprayed over the bond layer. Often a third layer is also applied to seal the top surface of the coating. A typical micrograph of plasma-sprayed ZrO 2 ⫹ CaO ⫹ Al 2O 3 coating showing lamellar structure is depicted in Fig. 6.53 [74]. The impetus to improve the adhesive strength and homogeneity of coatings by increased particle velocity and to minimize the disadvantage associated with spraying in air brought changes in plasma coating technology. The latest state of technology is the low-pressure plasma spraying (LPPS) process, where the

394

Chapter 6

Figure 6.52 Cross-sectional view of a plasma spray gun head [73].

Figure 6.53 Micrograph of a typical lamellar structure for a plasma sprayed coating: top coating: ZrO 2 ⫹ CaO ⫹ Al 2O 3; bond coating: NiCrAlY [74].

Alloy Oxidation

395

entire unit is maintained at a low pressure (50–70 mbar). This technique not only allows longer jet length and higher particle velocity accompanied by heating of the substrate to 1073–1173 K, but unwanted gas–metal reactions are also avoided, producing coatings with a high degree of density and good adhesion. The greater processing flexibility and closer compositional control of LPPS have permitted deposition of coatings with desired compositions and microstructures that cannot be achieved by electron beam physical vapor deposition (EPPVD). The most corrosion-resistant overlay coatings of current use for gas turbine foils that rely on the formation of protective Al 2O 3 scales are the MCrAlY (having a nominal composition in wt% of 18% Cr, 23% Co, 12% Al, 0.5% Y, and balance Ni) coatings with minor element contents and microstructural homogeneity, and are produced by LPPS. The current generation of LPPS coatings of the type Ni-Co-Cr-Al-Hf-Si-Y protect-single crystal airfoils in production of JT9D and PW2037 engines. Another technique of recent origin is laser-assisted spraying, whereby the coating powder is blown from the side into a high-power laser beam, which heats it to the melting point so that the added elements are either alloyed with the substrate or embedded as solid particles in the molten surface. Vapor Deposition and Related Techniques. Physical vapor deposition (PVD) consists of evaporating the elements required to form the coating, typically by directing an electron beam onto the substrate, in a high-vacuum chamber and allowing the elements to condense on it, which may be preheated to improve the adhesion and is usually rotated to improve the uniformity of the coating [75]. Accordingly, application of coatings to the interiors of holes or into hidden cavities is difficult, and since only thermal energy is involved and no vacuum glow discharge is used, the adhesion of the coating is also poor. Ion plating is a related approach where, by increasing the gas pressure (⬇ 1 MPa) in the deposition chamber and creating a glow discharge, the energy of the ionized gas atoms, which are usually argon, can be used to clean the component surface by sputtering for improved adhesion of the coating. Chemical vapor deposition (CVD) involves volatilization of a molecular species containing the element or elements required for the coating; the molecules subsequently decompose onto the component surface, depositing the element/ elements. Thus, the coating is obtained either by thermal decomposition (pyrolysis) or chemical reaction in the gaseous phase. A typical example of thermal decomposition of the gaseous phase is 1473 K TiI 4(g) → Ti (deposit) ⫹ 2I 2(g)

(6.56)

Similar decomposition of the gaseous phase by reduction reaction with hydrogen or a metallic vapor can be illustrated as

396

Chapter 6

WF 6(g) ⫹ 3H 2(g) → W (deposit) ⫹ 6HF (g)

(6.57)

and TiCl 4(g) ⫹ 2Mg (g) → Ti (deposit) ⫹ MgCl 2(g)

(6.58)

The gas phase may be composed of metal halides, metal carbonyls, metal hydrides, organometallic compounds, etc. Reactions generally take place in the temperature range of 423–2473 K and frequently between 773–1373 K. Both open-loop and closed-loop reactor systems are used to produce CVD coatings. Most of the CVD processes are carried out in open-loop systems, where the reactant gases are continually supplied from one end of the reactor and removed from the other. The major advantages of CVD processes are: 1. 2. 3. 4.

Possibility of forming various types of deposits (elemental metal/alloy, TiC, TiN, Al 2O 3 etc.) Good-quality deposits with widely varying structures (amorphous, crystalline, epitaxial, whiskers, etc.) High rates of deposition Possibility of coating complex-shaped bulk components.

Hybrid processes like plasma-assisted CVD (PACVD) and laser-assisted CVD (LACVD) are presently gaining popularity. PACVD is being used on an industrial scale in the fields of microelectronics, optics, solar energy devices, and so forth. In this technique, chemically reactive species, such as ions and free radicals, are created from the gas phase with the help of a plasma. In LACVD, the laser is used as a heat source, and under appropriate conditions only the surface of the substrate is heated to the CVD reaction temperature.

6.8.3

Thermal Barrier Coatings

In gas turbine operation, a considerable amount of energy derived from the fuel combustion process is dissipated through the engine structure and the cooling system. During combustion in a diesel engine, components like pistons, valves, liners, cylinder covers, etc., attain high surface temperatures. The exchange of heat occurs by convection and radiation or by direct contact with the flame. The absorbed heat must be transported away to retain the mechanical, thermal, and corrosion resistance properties of the component materials. The energy balance of a cylinder unit in a diesel engine demonstrates that nearly 50% of the energy produced in the combustion process is removed with cooling water/air and through the exhaust gas. To save energy, it is essential and advantageous to protect the hot parts of an engine by a thermally insulating layer. A ceramic layer would prevent heat transfer from the combustion zone to the coolant and surroundings. It would not only contribute to the reduction in temperature of the metallic components during service but provide protection against corrosion. Re-

Alloy Oxidation

397

Figure 6.54 Schematic representation of a TBC-bond coating system, indicating the temperature profile on an air-cooled engine component [77].

duced heat flow implies that a larger fraction of the released energy due to combustion process would be converted to mechanical power, involving increased efficiency of the engine. The incentive to develop such ceramic coatings is quite high since they can decrease the metal temperature by 313–323 K and thus can improve the creep and oxidation life of the components. As a result, thermal barrier coatings (TBCs) are increasingly being used [74,76] to insulate and protect critical air-cooled components of gas turbine engines. Most TBC systems consist of an insulating ceramic coating adhering to an underlying oxidation-resistant metallic bond coating. The insulating nature of a TBC is shown schematically in Fig. 6.54 [77]. Such a system allows the use of higher gas temperatures in a turbine engine for the same metallic component temperature and thereby increases the thermodynamic efficiency of the engine. In addition to being insulative (since it is a ceramic), TBC along with oxidationresistant bond coating also protects the metallic components from corrosive environments of the engine by lowering their temperatures. So the essential requirements for materials to be used as TBC are low thermal conductivity, resistance to corrosive and erosive environments, high thermal coefficients of expansion (for compatibility with metallic materials), and thermal shock resistance. Stabilized ZrO 2 (e.g., MgO ⋅ ZrO 2, Y 2O 3 ⋅ ZrO 2 etc.) seems to satisfy the above properties for which ZrO 2-based coatings are most commonly used as TBCs. A comparison of the thermal conductivity for different materials shows gray iron of about 20 W/mK, aluminum alloys of 117 W/mK, ceramic silicon nitride as high as 20 W/mK, whereas plasma-sprayed, stabilized ZrO 2 is measured to about 1.5–2.4 W/mK. ZrO 2 is known to possess not only low thermal conductivity but also relatively high coefficients of thermal expansion, which reduce the detrimental

398

Chapter 6

interfacial stresses between the TBC and the bond coating. Even though ZrO 2 is polymorphic in nature having crystal structures as monoclinic (room temperature to 1373 K), tetragonal (1373–2543 K), and cubic (2543–2953 K), the high-temperature cubic phase can be stabilized by the addition of Y 2O 3, MgO, CaO, CeO 2, etc. With the addition of sufficient stabilizer, the ZrO 2 structure can be fully stabilized and retained in cubic form even at ambient temperatures. However, by maintaining the addition of stabilizer at a sufficiently low value it is possible to obtain partial stabilization of ZrO 2 and all three phases (cubic, tetragonal, and monoclinic) can be retained on cooling to room temperature. Such partially stabilized ZrO 2 (PSZ) is a superior TBC material compared with a fully stabilized ZrO 2. The superiority of PSZ is due to its better thermal shock resistance and lower linear thermal expansion coefficient compared with that of fully stabilized ZrO 2. Accordingly, ZrO 2 partially stabilized by addition of 6–8 wt% Y 2O 3 is increasingly being used as a TBC due to its superior mechanical stability under thermal cycling conditions prevalent in gas turbine environments. Even though the porous nature of TBCs enhances the thermal shock resistance, which is very much desired, the thermal expansion mismatch between the TBC and the substrate results in the development of interfacial residual stresses. Moreover, the porous nature of TBC allows the corroding gases to penetrate, resulting in a high corrosion rate of the substrate alloy. Hence, to reduce these effects, a bond coating with high corrosion resistance is employed as an intermediate layer between the TBC and the substrate. Bond coating also minimizes the thermal expansion mismatch. In general, TBCs are deposited on superalloys by air plasma spraying on top of a vacuum plasma-sprayed M-Cr-Al-Y bond coating. The microstructure of such a coating is demonstrated in Fig. 6.53. Performance tests of various compositions of Ni-Cr-Al-Y bond coating suggest that the optimum chromium and yttrium contents should be 14–18 wt% and 0.3 wt%, respectively, for reducing the tendency for spallation at the TBC–bond coating interface, which is a common mode of TBC failure. So the major factors affecting thermal cycling survivability and high-temperature performance of plasma-sprayed TBC (ceramic) coatings include thermal expansion mismatch between the oxide coating and the alloy substrate, phase transformations of the coating material, interfacial interactions during thermal treatment, compositional effects, bond coating corrosion, and oxidation resistance. These factors will, of course, interact and determine the coating behavior.

6.8.4

Degradation of Coatings

The efficacy of a metal/alloy coating system in service environments is judged in terms of coating–environment and coating–substrate reactions and interactions along with accompanied vaporization processes (if any). The vaporization processes may include (1) vaporization of the protective external oxide film, (2)

Alloy Oxidation

399

volatile products formed by reaction with the environment, and (3) vaporization from an internal layer. Accordingly, degradation of coatings during performance may mainly occur by two processes: 1. Diffusional interaction between coating and the substrate 2. Degradation of the coating through reaction and interaction with the environment Degradation Through Diffusional Interaction Between Coating and Substrate It is worth emphasizing that during performance of a coated metallic component (coatings may be metallic or intermetallic) at high temperatures, chemical reactions take place between the reactive element or elements of the coating and the reactive species of the environment in the formation of a stable, impervious, protective barrier layer, which protects the underlying alloy from degradation. But it must also be realized that at the same time diffusional interaction between the coating and substrate continues. Such interactions may occur either through outward diffusion of the base metal/alloy constituents into the coating or through inward diffusion of the element/elements of the coating into the alloy substrate. It was mentioned earlier that during the aluminizing process of nickel-based superalloys a top layer of β-NiAl forms, beneath which a thin multiphase layer also occurs containing, in addition to β-NiAl, γ′-Ni 3Al and precipitates of the carbide-forming elements (Cr, Ti, Nb, W) composing the alloy. Beneath this, there exists an aluminum-saturated substrate alloy. In a similar way, during aluminizing of cobalt-based superalloys, the top layer consists of CoAl, beneath which there exists an α-Co layer enriched in aluminum along with other constituents of the alloy. Furthermore, the aluminized coatings are known to be formed on Ni- and Co-based alloys by outward diffusion of Ni and Co from the respective alloy substrate [75]. Protective properties of the aluminized coatings are determined by NiAl and CoAl phases which on oxidation form a stable barrier layer of Al 2O 3 [78]. During the high-temperature performance of the coated superalloys, deterioration of the coating continues as it becomes gradually diluted in the component that forms the protective oxide scale. The NiAl phase is gradually transformed to Ni 3Al phase having considerably poorer protective properties. Accordingly, on the one hand, aluminum is consumed in the formation of protective Al 2O 3 scale, which as a result of thermal shock and the erosive gas flux action is continuously destroyed; on the other hand, aluminum diffusion from the coating into the alloy substrate occurs simultaneously. In such a situation, Ni 3Al phase forms simultaneously at the NiAl–Al 2O 3 interface and at the substrate–aluminum-depleted NiAl boundary layer. Precipitation of Ni 3Al phase at the Al 2O 3 / coating boundary is more detrimental since along such precipitates an increased oxidation occurs. Such a situation may arise when continuous growth of Ni 3Al

400

Chapter 6

Figure 6.55 Schematic illustration of aluminide coating deterioration mechanism on nickel-based alloys. (a) Coating prior to exploitation; (b) changes during exploitation; and (c) breakdown of coating and spalling [78].

crystallites connects the inner surface of outer coating layer to the inner Ni 3Al layer adjoining the substrate, causing a complete loss of protective properties of the diffusion coating. Rapid oxidation along Ni 3Al crystallite boundaries together with penetration of the oxidant to the inner Ni 3Al layer causes exfoliation of the protective scale. The mechanism of such coating deterioration process is shown schematically in Fig. 6.55 [78]. In a similar way, deterioration of aluminide coatings on cobalt-based superalloys is mainly attributed to development of aluminum-depleted CoAl phase through simultaneous Al 2O 3 layer formation and outward diffusion of cobalt, which in the course of time transforms to α-cobalt phase. The problems associated with the use of molybdenum or tungsten or their alloys as high-temperature materials in oxidizing environments have already been discussed, wherein it is mentioned that the only practical solution to protect them from high-temperature degradation lies in the use of coatings. Attempts to develop protective scales by alloying the refractory metals with scale-forming ele-

Alloy Oxidation

401

ments did not meet with success. Self-healing SiO 2 layers formed on silicide coatings offer a much better oxidation-resistant barrier than Al 2O 3 layers developed on aluminide coatings. The protectiveness of silicide coatings is mainly due to (1) formation of glassy SiO 2 scales, (2) low diffusivity of molecular oxygen in these scales, (3) high flexibility of SiO 2 in being able to form modified glasses or silicates as a result of uptake of elements from the substrate or environment, (4) self-healing properties of SiO 2, (5) adjustment of thermal expansion by the use of additive elements, and (6) inertness of silicon against sulfidation [79]. Silicide coatings are normally produced by the diffusion of silicon into the refractory metal substrates having acceptable mechanical properties, thus forming MoSi 2 or WSi 2, which on oxidation at high temperature gives rise to the formation of a thin, protective, glassy layer of SiO 2. This layer exhibits protective properties to as high a temperature as 1973 K, which is close to the melting point (1998 K) of crystobalite-modified SiO 2. Such silicide coatings are not preferred for protecting nickel-based alloys, mainly due to brittleness of the coating and the rapid rate of Si diffusion into the alloy substrate during high-temperature exposures. However, attempts to modify the standard M-Cr-Al-Y coating systems with silicon additions up to about 2.7% on nickel-based superalloys have demonstrated encouraging results with respect to oxidation and hot corrosion resistance [79]. Moreover, the addition of boron markedly improves the spalling resistance of the SiO 2 scales due to formation of SiO 2-B 2O 3 glass scale with a higher thermal expansion coefficient and a lower softening temperature [79]. Even though hightemperature capabilities of silicide coatings are excellent, the useful upper temperature limit of their applications in oxidizing environments are determined by refractoriness of the coating components, their rates of conversion to oxide, the necessity that the oxide be silica, the rate and site of vapor phase material loss, and the rate of diffusional reactions between coating and substrate [80]. The refractoriness of the coating depends on the melting point of the initial material and the product of reactions between the substrate and environment. The silicide coatings formed on refractory metals are primarily composed of a layer of the most silicon-rich intermetallic in the binary system, which gets converted to lower silicides by solid-state diffusion and silica during their high-temperature exposure to oxidizing environments as depicted in Fig. 6.56a and b [80]. Nevertheless, MoSi 2 and WSi 2 are the desired intermetallics to be formed on Mo and W during the coating process, although during their service conditions they are converted to lower silicides by diffusion processes. In the case of Mo, in addition to MoSi 2, lower silicides, such as Mo 5Si 3 and Mo 3Si, can also form during performance of the coating by inward diffusion of silicon toward the substrate metal [2]. In Fig. 6.56a, Mo 3si phase is not observable for the limited exposure time of the coating. Similar silicide phases also occur in tungsten. The oxidation resistance of these silicides decreases in the order MoSi 2 ⬎ Mo 5Si 3 ⬎ Mo 3Si, i.e., with decreasing silicon content in the silicide [81]. The two lower silicides may eventually show

402

Chapter 6

Figure 6.56 Schematic illustration of the degradation of disilicide coatings to lower silicides by coating–substrate interaction during oxidation: (a) MoSi 2 on Mo after 30 h at 1948 K; (b) WSi 2 on W after 13 h at 1923 K [80].

protective behavior, but due to their low silicon content, a longer period of oxidation is needed. During diffusion-controlled interaction of the silicide coating with the substrate, the thickness of the different layers is governed by the relative rates of diffusion in the respective layers and the chemical potential gradient of the diffusing species. If diffusion through one layer is rapid, the corresponding layer will be thick. On the contrary, if the rate of diffusion through a layer is low compared with the neighboring layers, the thickness of the grown layer will automatically be thin, provided the temperature dependencies of growth of the layers are identical. Diffusion of the coating constituent into the substrate metal/alloy may be reduced by predepositing a high-melting-point metal prior to application of the coating [79]. Oxidation of molybdenum and tungsten silicides leads, on the one hand, to

Alloy Oxidation

403

the formation of protective SiO 2 formation and, on the other, to the formation of trioxides of the respective metals (e.g., MoO 3, WO 3), which are highly volatile in nature. Accordingly, the oxidation behavior of such silicides is determined by the degree to which the coating constituents are oxidized, the rate of evaporation of the trioxides, and the structure and composition of the reaction products. Silicide Pest. Oxidation of molybdenum disilicide (MoSi 2) in the temperature range of 723–873 K exhibits an interesting phenomenon, wherein accelerated corrosion occurs at an almost linear rate, but the rate decreases rapidly with the rise of temperature. This phenomenon is termed silicide pest, which involves not only surface oxidation of the disilicide but mainly intercrystalline oxidation of MoSi 2, in which each grain eventually becomes surrounded by reaction products. Ultimately, the silicide coating disintegrates to a voluminous heap of powder, its individual grain being surrounded by a thin layer of oxidation products. It is pertinent to note that the process occurs within such a temperature range where MoO 3 is stable. The pest phenomenon is not restricted to MoSi 2 only, but all silicides suffer from such limitations in their practical applications. Although the onset of pest can sometimes by delayed by modification of the coating, it cannot be prevented. The mechanism proposed for the occurrence of such phenomenon is preferential intergranular diffusion of the reacting gas coupled with a temperature-dependent hardening reaction [82]. Degradation Through Reaction with the Environment In general, the degradation of coatings during their performances at high temperatures follows the same principles and processes by which the corresponding bulk materials degrade. But the reaction behavior of the coating constituents with the environment may be different from the bulk homogeneous alloy for a number of factors. In the previous discussion it was illustrated that the composition and microstructure of the coatings may change during service due to coating–substrate interaction by diffusional processes and, as a consequence, the corrosion behavior of the coating is also expected to be altered. Depending on the technique used in the coating formation, a coating may contain porosities (e.g., in plasmasprayed coatings), which permit easy penetration of the reactive constituents of the gaseous reactant into the coating or even beyond it to the coating–substrate interface. This may not only result in accelerated corrosion of the coating itself but may even cause degradation to the underlying valuable alloy, which was otherwise thought to be well protected. Vaporization, as well as simple evaporation, may be promoted by chemical reactions between the coating and the environment. A previously formed protective oxide layer may dissociate when the environmental temperature is raised or the pressure is decreased or when the product of oxidation is volatile in nature. In both cases, changes in environmental pressure affect the vaporization rate. The

404

Chapter 6

rate of vaporization due to dissociation alone decreases steadily with increasing environmental pressure. However, vaporization due to volatile oxide formation increases initially with increasing oxygen pressure because the rate of oxide formation and subsequent vaporization depends on the rate of oxygen availability. Ultimately, the vaporization rate will decrease with increasing pressure due to formation of a blanket of the volatile oxide. If metals from the platinum group are used as coatings in oxidizing environments at high temperatures, they are eroded through volatilization of the oxides like PtO 2 [2]. Protective coatings are classified into three main groups depending on the type of protective scales they form during service: (1) chromia formers, (2) alumina formers, and (3) silica formers. Among these three, the chromia- and silica-forming coatings suffer from the following limitations contributed by reactions with the environment producing volatile oxides. (i). At high temperatures and high oxygen pressures, Cr 2O 3 is oxidatively evaporated to CrO 3 according to 1/2Cr 2O 3 ⫹ 3/4O 2 ⫽ CrO 3(gas). Therefore, chromia scale–forming coatings should not be used at temperatures above 1273 K in oxygen at or near atmospheric pressures, but may be used at higher temperatures in lower oxygen partial pressures. (ii). Silica scales on silicide coatings are not stable at reduced oxygen pressures and at high temperatures due to the formation of volatile SiO according to Si ⫹ SiO 2 ⫽ 2SiO (gas). High-temperature oxidation studies of elemental silicon have established that SiO (g) can form at the silica film–element interface by active oxidation process [50] under specific partial pressures of oxygen. The SiO (g) so formed tries to move upwardly and finds its escape route through pinholes and fissures (microchannels) in the SiO 2 film, thus rupturing the film locally. The SiO (g) formation at a particular temperature continues so long as the pressure of SiO (g) is greater than the ambient oxygen pressure, i.e., pSiO ⬎ pO2, thus destroying protective property of the silica film. In general, SiO 2-forming coatings should not be used at temperatures above 1873–1973 K because low viscosity of silica is maintained only below such a temperature range. It is the viscosity of silica, rather than its melting point, that deserves due consideration in applications, even though in practice, silicide coatings are sometimes used at temperatures above silica’s melting point. Silica is relatively fluid in the temperature range 2273–2773 K [80]. Moreover, the glassy films of silica are not prone to break-away, possibly because of their excellent ductility and thinness. Vapor phase loss of silica is possibly responsible for the oxide film thinness formed on silicides. Such losses can occur by two processes: simple evaporation and decomposition to volatile SiO, which bubbles out through the outer protective layer of silica, thus causing early coating failure [80].

Alloy Oxidation

405

In the oxidation studies of MoSi 2 and WSi 2 coating systems [78], the general kinetic behavior revealed that a net mass gain occurs at lower temperatures (⬍1273 K) followed by a net mass loss in the intermediate temperature range (1273–1473 K) and, again, a net mass gain at higher temperatures (⬎1473 K). The only behaviorial difference between the two disilicides being that for MoSi 2, the net mass loss in the intermediate temperature range gets displaced toward lower temperatures because of higher volatility of MoO 3. Since both the constituents of the coatings are oxidizable, elemental silicon forms an external scale of SiO 2, whereas the highest oxides of Mo and W (MoO 3 and WO 3), which are volatile at high temperatures, are formed simultaneously. In the low-temperature region (⬍1273 K), increased mass gain can be corroborated to solid reaction product formation with a negligible contribution from trioxide evaporation. The oxide layer consists of crystalline silica of the crystobalite form along with a complex Mo-Si oxide phase [50]. The net mass loss in the intermediate temperature range (1273–1473 K) suggests the predominance of volatile trioxides formation and their escape from the coating system over growth of the silica layer. There is no doubt about the formation and evaporation of some trioxides also at higher temperatures (⬎1473 K), but the net mass gain indicates that such evaporation process is of minor importance and Si is preferentially oxidized to form a glassy, protective layer, for which subsequent oxidation proceeds only at an exceedingly low rate. In describing the reaction mechanism, it is proposed [83] that initially a low-melting mixture of SiO 2 and MoO 2 is formed that flows freely to fill cracks on the surface before MoO 3 volatilizes and the oxide solidifies. The glassy outer scale is not to be considered as pure SiO 2 but rather as a ternary oxide containing less than 0.1% Mo. During isothermal oxidation of MoSi 2 coating on Mo substrate, the coating deteriorates due to the conjugal effect of selective silicon oxidation and inward silicon diffusion, resulting in transformation of the MoSi 2 layer to Mo 5Si 3 and subsequently to Mo 3Si by coating–substrate interaction, as discussed earlier. Since the coating deterioration is limited by diffusion, its effective life under isothermal conditions is a function of temperature and the coating thickness, and as such decreases exponentially with increasing temperature. At and above 1973 K, MoSi 2 does not provide satisfactory oxidation resistance. This is due to a combination of many factors, such as (1) rapid silicon diffusion into the substrate; (2) SiO 2 (crystobalite form) melts at 1998 K and, accordingly, rapid diffusion of oxygen through a liquid layer; (3) oxide evaporation (SiO 2, SiO, MoO 3); and (4) in case volatile oxides form at the scale–substrate interface, the scale may get ruptured. The effective life of a silicide coating is also markedly reduced by thermal cyclings, which lead to cracking of the protective layer as a result of differences in thermal expansion between the substrate and the scale. Crack formation during cooling of the component does not necessarily lead to failure of protective proper-

406

Chapter 6

ties of the coating because SiO 2 can again spontaneously grow during subsequent heating as a self-healing layer. As a result, during subsequent heating cracks are again healed partly by thermal expansion and sintering, and partly by formation of new oxide. It has been suggested [83] that the deteriorating effects of thermal cyclings can be minimized by the deposit of a layer of molybdenum boride between the disilicide and the metal. The positive effect of such a layer is due to better filling and sealing of the cracks resulting from fluxing action of boron oxides. It is also important to note that the structure of the SiO 2 layer changes with temperature, which has a strong bearing on its protective properties. At temperatures above 1473 K it is glassy, whereas at lower temperatures it becomes increasingly crystalline. At 1073 K, the scale on MoSi 2 consists of crystalline SiO 2 (crystobalite) associated with a complex MoO 2-SiO 2 phase. The glassy layer formed above 1473 K offers the best protective properties. During cooling in the intermediate temperature range (1273–1473 K), the SiO 2 layer undergoes transformation from glassy to crystalline structure. Thus, it appears advantageous to preoxidize MoSi 2 coatings at 1673–1773 K before using the coated component at lower temperatures [78]. Another factor that affects the protective properties of MoSi 2 adversely is the reduced oxygen partial pressure in the environment. At reduced pressures, the degradation process gets localized, resulting in numerous pinholes. Under such conditions, the oxidizing power of the gas phase is not sufficient to allow formation of a continuous protective film of SiO 2. Moreover, SiO (g) evaporation also becomes an important phenomenon at low oxygen pressures. Therefore, if silicide-coated molybdenum is to be employed in atmospheres having low oxygen activity, preoxidation of the coating in an environment having high oxygen activity ought to be carried out to develop a thin protective SiO 2 layer on the coating surface for its subsequent satisfactory performance. In all practical coating systems containing silicon as a major protective element, there is a pronounced gradient in the physical, mechanical, and chemical properties of the coating–metal/alloy system. Brittleness and thermal expansion mismatch give rise to cracking and spallation during thermal cyclings. The inherent brittleness of silicide intermetallics is one of the most important reasons for their early failures, although the ductile-brittle transition temperatures of silicide coatings are higher than those for other coating systems [79]. To overcome this disadvantage and maintain the silicon content in the surface layer of the coating unaffected from loss due to oxidation over longer periods of exposure, it is often suggested that tough matrices and dispersed silicide reservoir phases in the NiCr-Si and Ni-Cr-Si-Ta overlay coating systems be used. The problems associated with interdiffusion in silicide coating–substrate systems can be minimized by the addition of refractory metals, such as Ta, Ti, Nb, or Mo, to reduce the diffusion rate, Ta being the most effective.

Alloy Oxidation

407

It can be concluded that silica scales offer better protection than chromia scales, which are susceptible to vaporization loss via CrO 3(g) at temperatures above 1273 K. There is no doubt about the superiority of alumina scales over silica scales at gas turbine operating temperatures. However, above 1773 K, silica scales are more effective barriers in inhibiting degradation for which silicide coatings are used to protect refractory metals at superhigh temperatures (approx. 1973 K). Durability of TBCs. The use of ceramic thermal barrier coatings in the protection of components of gas turbine engines to significantly improve the service life of components and engine performance has already been discussed (Sec. 6.8.3). TBCs are attractive because they can resist corrosion and oxidation (due to inertness) of the high-temperature components and thereby extend their service life through lowering of functional temperature and thermal stresses (strains) induced during engine transients. They simultaneously provide an opportunity to increase the turbine inlet temperatures or reduce cooling airflow in turbine airfoils, contributing to higher engine efficiency and power output. The preferred material for TBC is partially stabilized zirconia (ZrO 2-22 wt% MgO or ZrO 2-6 to 8 wt% Y 2O 3), for its inertness, insulating property with low thermal conductivity, high resistance to corrosive and erosive environments, high coefficient of thermal expansion to enhance compatibility with metallic substrates, and thermal shock resistance properties. Although such material properties of the partially stabilized zirconia (PSZ) are of importance in performance, other properties, such as adhesion strength, residual stresses, porosity, etc., markedly influence the integrity of the coating system. The thermal expansion mismatch between the TBC and the alloy substrate results in interfacial residual stresses, leading potentially to coating delamination. The residual stresses depend on a variety of mismatch strains and on the extent to which these strains result in mismatch stresses. It is well known that thermal gradients and the transition from molten to solid state may generate stresses in plasma-sprayed ceramic coatings. The liquid–solid volume shrinkage, which may often be as high as 10% (for ZrO 2), results in large strains in all melt-coating processes. Such shrinkage is not only an important source of porosity but also of stress concentration [74]. While the porous nature of the TBC enhances its thermal shock resistance, it also allows easy penetration of corrodents through the coating, resulting in fast degradation of the substrate alloy. Hence, to reduce such effects, a material having high oxidation and corrosion resistance property is employed as an intermediate bond coat between the TBC and substrate, which also minimizes the thermal expansion mismatch. Ni-Cr-Al-Y coatings are commonly used for such bond coats. The FCC matrix of the Ni-based alloy is favored for high-temperature applications because of its nearly filled third electron shell, which results in phase stability even in the pres-

408

Chapter 6

ence of alloying elements. Chromium and aluminum increase the strength of the nickel matrix by solid solution strengthening and also form protective Cr 2O 3and Al 2O 3-rich films by reaction with the environment. Aluminum further helps in the formation of γ ′ phase, resulting in antiphase boundary strengthening. Spallation of the protective film is reduced by chromium in the superalloy–coating matrix. Minor amounts of Y are added for further improvement in scale adherence [84]. Most failures of the TBC–bond coating system often occur at the ceramic– bond coating interface because of high compressive stresses in the TBC. The extent of these stresses are determined by the amount of strain relieved, e.g., by plastic deformation or by microcracking. The transformation strains, which are inversely dependent on the degree of stabilization in ZrO 2, may or may not counteract some of the thermal expansion strain depending on the degree and nature of texturing. Furthermore, microcracking associated with phase transformation or porosity (as obtained in plasma-sprayed coating) is a potential mechanism of relieving mismatch strains. During performance of the TBC–bond coating system at high temperatures, the interfacial bonding oxide, which is predominantly Al 2O 3- or Cr 2O 3-rich, grows in thickness. Once the oxide layer reaches sufficient thickness, its own thermal shock resistance property comes into play. It has been suggested [85] that this oxidation is the single most time-dependent factor that limits the service life of TBCs. Moreover, it has been further stressed that the role of oxidationinduced strains that combine with cyclic strains to promote slow crack growth in the ceramic layer should not be ignored. Hence, the bond-coating material should be such that during service it forms an impervious, tenacious oxide bonding layer that doesn’t allow the oxide to grow in thickness with time at its operational temperature. It is well known that Al 2O 3 has a poor thermal shock resistance for which the compositions of currently preferred bond coatings [84] are adjusted to low aluminum with higher chromium in order to utilize the superior thermal shock resistance properties of Cr 2O 3.

6.9 CONCLUSION In general, the reaction behavior of protective coatings in environments of their use and their interactions with the substrates during high-temperature performance is not well understood. This is basically due to the limited research undertaken in this area in comparison with the investigations on bulk homogeneous metals/alloys. The demand for better performance of components has led to development of new coating compositions and appropriate methods of their formation. Such development has encompassed an interaction between the physical metallurgy of the coating and its processing. A fundamental understanding of the mechanism of formation of different coatings and their mode of degradation

Alloy Oxidation

409

is essential for design and development of improved coatings in the future. Since the performance of a practical coating system depends on many factors, it often becomes obligatory to strike a compromise among the different requirements; accordingly, a coating system is often tailored to a particular metal/alloy to suit a specific application. Hence, it is prudent to design a coating–substrate system as a single entity as opposed to the practice of trying to marry the existing ones to various alloys, with minor modifications here and there.

REFERENCES 1. G. R. Wallwork and A. Z. Hed, Oxid. Met., 3(2):171 (1971). 2. P. Kofstad, High Temperature Oxidation of Metals, John Wiley and Sons, New York (1966). 3. P. Kofstad and A. Z. Hed, Oxid. Met., 2:101 (1970). 4. S. T. Wlodek, Trans. AIME, 230:1078 (1964). 5. K. Hauffe, Oxidation of Metals, Plenum Press, New York (1965). 6. S. K. Roy and S. C. Sircar, Oxid. Met., 15:9 (1981). 7. S. K. Bose and S. C. Sircar, Met. Trans., 5:2015 (1974). 8. S. K. Bose and S. C. Sircar, Trans. IIM, 33(1):37,45 (1980). 9. R. A. Rapp, Corrosion, 21:382 (1965). 10. J. H. Swisher, Internal Oxidation, in Oxidation of Metals and Alloys (D. L. Douglass, ed.), ASM, Metals Park, Ohio, 1971), p. 235. 11. J. L. Meijering, Internal oxidation in alloys, in Advances in Materials Research (H. Herman, ed.), Wiley, New York, Vol. 5, 1971; p. 1. 12. F. N. Rhines, W. A. Johnson, and W. A. Anderson, Trans. TMS-AIME, 147:205 (1942). 13. L. S. Darken, Trans. TMS-AIME, 150:157 (1942). 14. C. Wagner, Z. Electrochem, 63:772 (1959). 15. R. A. Rapp, Acta. Met., 9:730 (1961). 16. N. Birks and G. H. Meier, Introduction to High-Temperature Oxidation of Metals, Edward Arnold, London, 1983. 17. W. C. Leslie and M. G. Fontana, Trans. Am. Soc. Met., 41:1213 (1949). 18. G. W. Rathenau and J. L. Meijering, Metallurgia, 42:167 (1950). 19. S. S. Brenner, J. Electrochem. Soc., 102:7, 16 (1955). 20. B. Chattopadhyay and G. C. Wood, Oxid. Met., 2(4):373 (1970). 21. G. C. Wood, Werkstoffe. Korro, 22(6):491 (1971). 22. C. S. Giggins and F. S. Pettit, Trans. TMS-AIME, 245:2495; 2509 (1969). 23. C. Wagner, Corr. Sci, 5:751 (1965). 24. J. Moreau and J. Be´nard, L’Oxydation des Metaux, Vol. 1, Gauthier/Villars, Paris, 1962. 25. G. C. Wood, Oxid. Met., 2:11 (1970). 26. G. R. Wallwork, Rep. Prog. Phys., 39:401 (1976). 27. A. D. Dalvi and D. E. Coates, Oxid. Met., 5(2):113 (1972). 28. B. D. Bastow, G. C. Wood and D. P. Whittle, Oxid. Met., 16(1/2):1 (1981).

410

Chapter 6

29. F. Maak, Z. Metallkd, 52:538; 545 (1961). 30. G. C. Wood, The structures of thick scales on alloys, in Oxidation of Metals and Alloys (D. L. Douglass ed.), ASM, Metals Park, Ohio, 1971, p. 204. 31. P. Kofstad and A. Z. Hed, J. Electrochem. Soc., 116:224, 229,1542 (1969). 32. P. Kofstad, Proc. Int. Symp. on the Reactivity of Solids (J. Wood, O. Lindqvist, C. Hegesson, and N. G. Vanneberg eds.), Plenum, Press, New York, 1972, p. 15. 33. G. C. Wood, T. Hodgkiess, and D. P. Whittle, Corr. Sci., 6:129 (1966). 34. F. A. Golightly, F. H. Stott, and G. C. Wood, J. Electrochem. Soc., 126:1035 (1979). 35. F. H. Stott, G. C. Wood, and M. G. Hobby, Oxid. Met., 3:103 (1971). 36. G. D. Oxx, Prod. Eng., 29:61 (1958). 37. W. T. Griffiths and L. B. Pfeil, U. K. Patent No. 459848 (1937). 38. L. B. Pfeil, U. K. Patent No. 574088 (1945). 39. G. Beranger, F. Armanet, and M. Lambertin, in The role of Active Elements in the Oxidation Behavior of High Temperature Metals and Alloys (E. Lang, ed.), Elsevier, London, 1989, p. 33. 40. D. P. Whittle and J. Stringer, Phil. Trans. R. Soc. London, A295:309 (1980). 41. F. H. Stott and G. C. Wood, Mat. Sci. Eng., 87:267 (1987). 42. A. M. Huntz, Mat. Sci. Eng., 87:251 (1987). 43. M. J. Bennet and D. P. Moon, in The Role of Active Elements in Oxidation Behavior of High-Temperature Metals and Alloys (E. Lang, ed.), Elsevier, London, 1989, p. 111. 44. D. G. Lees, Oxid. Met., 27:75 (1987). 45. D. G. Lees, Oxid. Met., 30:267 (1988). 46. K. L. Luthra and C. L. Briant, Oxid. Met., 30:257 (1988). 47. J. G. Smeggil, N. S. Bornstein, and M. A. Decrescente, Oxid. Met., 30:259 (1988). 48. D. P. Moon, Mater. Sci. Technol., 5:75 (1989). 49. T. N. Rhys-Jones, H. J. Grabke, and H. Kudielka, Corr. Sci., 27:49 (1987). 50. P. Kofstad, High Temperature Corrosion, Elsevier, London, 1988. 51. B. A. Pint, Oxid. Met., 45:1 (1996). 52. A. M. Huntz, in The Role of Active Elements in the Oxidation Behavior of High Temperature Metals and Alloys (E. Lang, ed.), Elsevier, London, 1989, p. 81. 53. J. G. Smeggil, A. W. Funkenbusch, and N. S. Bornstein, Met. Trans., 17A:923 (1986). 54. M. A. Descrescente and N. S. Bornstein, Corrosion, 24:127 (1968). 55. J. Stringer, Annu. Rev. Met. Sci., 7:477 (1977). 56. K. L. Luthra, Met. Trans., 13A:1647, 1843, 1853 (1982). 57. D. K. Gupta and R. A. Rapp, J. Electrochem. Soc., 127:2194, 2656 (1980). 58. C. S. Giggins and F. S. Pettit, Hot Corrosion Degradation of Metals and Alloys: A Unified Theory, Final Scientific Report to Air Force Office of Scientific Research on Contract No. F44620-76-C-0123, Pratt and Whitney Aircraft Group, June 1979. 59. N. S. Bornstein and M. A. Decrescente, Trans. TMS-AIME, 245:1947 (1969). 60. J. A. Goebel, F. S. Pettit and G. W. Goward, Met. Trans., 4:261 (1973). 61. D. A. Shores, Proceedings of the Conference on High Temperature Corrosion (R. A. Rapp, ed.), NACE, Houston, 1983, p. 493. 62. R. A. Rapp and K. S. Goto, Proc. 2nd Int. Sym. on Molten Salts (J. Braunstein and J. R. Selman, eds.), Electrochemical Society, Pennington, NJ, 1981, p. 81.

Alloy Oxidation 63. 64. 65. 66. 67. 68. 69.

70.

71. 72. 73. 74. 75. 76. 77.

78.

79. 80. 81. 82. 83. 84. 85.

411

R. A. Rapp, Corrosion, 42(10):568 (1986). K. L. Luthra and D. A. Shores, J. Electrochem. Soc., 127:2202 (1980). P. Hancock, Mater. Sci. Eng., 88:303 (1987). F. Fitzer and J. Schlichting, Proceedings of the Conference on High Temperature Corrosion (R. A. Rapp, ed.), NACE, Houston, 1983, p. 604. F. S. Pettit, in Coatings for High Temperature Applications (E. Lang, ed.), Applied Science Publishers, London, 1983, p. 341. A. R. Nicoll, in Coatings for High Temperature Applications (E. Lang, ed.), Applied Science Publishers, London, 1983, p. 269. R. Bauer, H. W. Grunling, and K. Schneider, in Materials and Coatings to Resist High Temperature Corrosion (D. R. Holmes and A. Rahmel, eds.), Applied Science Publishers, London, 1978, p. 369. R. Pichoir, in Materials and Coatings to Resist High Temperature Corrosion, (D. R. Holmes and A. Rahmel, eds), Applied Science Publishers, London, 1978, p. 271. P. J. Anderson, D. H. Boone, and G. F. Paskiet, Oxid. Met., 4(2):113 (1972). I. A. Bucklow, in Coatings for High Temperature Applications (E. Lang, ed.), Applied Science Publishers, London, 1983, p. 139. H. D. Steffens, in Coatings for High Temperature Applications (E. Lang, ed.), Applied Science Publishers, London, 1983, p. 121. I. Kvernes, in Coatings for High Temperature Applications (E. Lang, ed.), Applied Science Publishers, London, 1983, p. 361. C. Duret and R. Pichoir, in Coatings for High Temperature Applications (E. Lang, ed.), Applied Science Publishers, London, 1983, p. 33. J. W. Fairbanks and R. J. Hecht, Mater. Sci. Eng., 88:321 (1987). R. J. Bratton and S. K. Lau, Proc. Int. Conference on Science and Technology of Zirconia, in Advances in Ceramics, Vol. 3 (A. H. Hener and L. W. Hobbs, eds.), American Ceramics Society, Columbus, OH, 1981, p. 226. S. Mrowec and T. Werber, Gas Corrosion of Metals, translated from Polish by W. Bartoszewski, published by National Bureau of Standards and the National Science Foundation, Washington, DC 1978. H. W. Gru¨nling and R. Bauer, Thin Solid Films, 95:3 (1982). C. D. Dickinson, M. G. Nicholas, A. L. Pranatis, and C. I. Whitman, J. Met., 15: 787 (1963). R. A. Perkins, in The Science and Technology of Molybdenum, Tungsten, Niobium and Tantalum and Their Alloys (N. E. Promisel, ed.), Pergamon, London, 1964. J. H. Westbrook and D. L. Wood, J. Nucl. Mater., 12:208 (1964). G. Todd and E. Parry, Nature, 203:967 (1964). H. Herman and N. R. Shankar, Mater. Sci. Eng., 88:69 (1987). R. A. Miller, J. Am. Ceram. Soc., 67:517 (1984).

This Page Intentionally Left Blank

7 Liquid Metal Attack

7.1 INTRODUCTION The attack of solid metals by liquid metals may be manifested in the following forms: 1. Instantaneous failure of a solid metal under applied or residual stresses when in contact with a liquid metal 2. Delayed failure of a solid metal in contact with a liquid metal at a static stress level below the tensile strength of the metal 3. Stress-independent grain boundary penetration of a solid metal by liquid metal 4. High-temperature corrosion of a solid metal by a liquid metal The first two types are generally referred to as liquid metal embrittlement (LME) or liquid metal–induced embrittlement (LMIE). Liquid metal embrittlement and liquid metal corrosion have been discussed in the following sections.

7.2 LIQUID METAL EMBRITTLEMENT The catastrophic brittle failure of a solid metal under stress in contact with a liquid metal is known as liquid metal embrittlement (LME). The loss of ductility of a normally ductile metal or alloy is manifested as a reduction in fracture stress, strain, or both. A change in the fracture mode from ductile to brittle intergranular or brittle transgranular (cleavage) is also normally encountered, though embrittle413

414

Chapter 7

ment can also occur by a ductile dimpled rupture mode in certain steels, copper alloys, and aluminum alloys. The failure arising from LME may be instantaneous or may take place after a certain lapse of time after the exposure of the stressed metal in the liquid metal environment. The former is treated as the ‘‘classical’’ LME and the latter is often referred to as delayed failure or static fatigue. In either case, the presence of stress is a requirement, which may be tensile, shear, or tortional in nature, but not compressive. In this respect, LME is analogous to stress corrosion cracking (SCC), but the propagation of fracture is much faster in LME than in SCC. Intergranular penetration of liquid metal may render a solid metal brittle even in the absence of stress if sufficient time is allowed, e.g., aluminum by liquid gallium, but this is not considered as LME. On the other hand, stressed aluminum in contact with liquid gallium breaks immediately and provides an example for LME.

7.2.1

Characteristics of LME

Effect on Stress–Strain Curve The elongation and reduction in area of the metal or alloy are lowered as a result of LME. The fracture stress is also reduced and in cases of severe embrittlement may be less than the yield stress of the material. However, there is no change in the yield strength and strain-hardening behavior of the solid metal. The stress– strain curve remains unaltered up to the point of failure. The liquid metal acts only to limit the total ductility before fracture or the stress at fracture if the failure occurs below the normal yield point. This is illustrated in Fig. 7.1 for the embrittlement of copper by various molten Bi-Pb alloys and in Fig. 7.2 for the embrittlement of various Fe-Al alloys in liquid mercury. The failure of mild steel occurs at only 2–3% elongation in molten lithium, but the lower yield point, upper yield point, and yield point elongation remain unaffected. Fracture Morphology The fracture mode changes from ductile to brittle as a result of LME. In polycrystalline materials, the brittle intergranular mode is the most common, e.g., α-brass in mercury or aluminum in liquid gallium. However, mixed inter- and transgranular cracking has been encountered in some systems, e.g., aluminum alloy 2024T4 in Hg-Zn amalgam [1]. Single crystals essentially fail by cleavage fracture in liquid metal environments, e.g., Cd in liquid indium. As mentioned earlier, ductile intergranular dimpled fracture has been encountered in some steels, copper alloys, and aluminum alloys. The fracture surfaces show complete coverage by liquid metal, which is sometimes difficult to detect as well as to remove for metallographic examination. It may be pointed out here that apart from the contamination of the fracture surface, there is little or no intergranular penetration of the liquid metal into the solid metal.

Liquid Metal Attack

415

Figure 7.1 Stress–strain curves of cold-rolled copper with various molten Bi-Pb alloys at 350°C.

LME cracking originates at the solid–liquid interface, which may be internal in some cases, e.g., in lead- or tellurium-containing steels. These elements are added to steel for improvement of machinability, but being low melting, they are causative factors for LME of the steel component at elevated temperature processing or applications. In most instances, the initiation and propagation of cracks appear to occur instantaneously. The velocity of crack propagation in LME has been estimated to be 10–100 cm/s or higher. The fracture may be nucleation-controlled or propagation-controlled. In the former, once a crack is initiated it propagates to failure in the presence or absence of the liquid metal at the crack tip. In the propagationcontrolled failure, microcracks are formed at some low stress in a liquid metal environment, but the propagation of these microcracks becomes possible only at higher stresses. Therefore, in this type of failure, unpropagated microcracks may be found beside the main fracture. Specificity of Environment Like SCC, all liquid metals do not embrittle all solid metals and it appears that some specificity exists in this regard. For example, liquid mercury embrittles zinc

416

Chapter 7

Figure 7.2 Stress–strain curves of various iron-aluminum alloys tested in air and mercury-indium solutions [8].

but not cadmium; liquid gallium embrittles aluminum but not magnesium. Table 7.1 enlists the known embrittlement couples. Two generalizations have been made from the survey of the embrittlement couples: 1. 2.

In a majority of cases of embrittlement, the solubility of the liquid in the solid and of the solid in the liquid is very small. There is an absence of intermetallic compound formation in the two metals involved.

There are exceptions to this generalization, e.g., Al-Zn and Fe-Zn show considerable mutual solubilities. Also, systems having small terminal solubilities, e.g., Al-Cd and Al-Pb, do not show LME. In regard of intermetallic compound formation, Fe-Zn and Mg-Zn are exceptions. However, embrittlement of Fe or

Summary of Embrittlement Couples |

Solid Sn Bi Cd Zn Mg Al Ge Ag Cu

Ni

Fe

Pd Ti

P P P P LA CA P CA P P LA CP LA CA P LA CA P LA CA P LA CA

|

Hg P

A

X X ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ X X ⋅⋅⋅ X ⋅⋅⋅ X X X X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X X ⋅⋅⋅ X X ⋅⋅⋅ X ⋅⋅⋅ X ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X

|

Cs P

|

P

A

⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

X ⋅⋅⋅ X X ⋅⋅⋅ ⋅⋅⋅ X X X X ⋅⋅⋅ ⋅⋅⋅ X X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ X X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

Ga

|

Na P

|

P

A

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ X X X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

In

Li P

Liquid Sn | | P A

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X X X ⋅⋅⋅ X X X ⋅⋅⋅ ⋅⋅⋅ X X X X ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ X X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

|

Bi

|

P

A

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X X X ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

Ti P

Cd P

|

P

A

|

Zn P

Te P

Sb P

Cu P

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ X

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ X X X ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ X X ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

Pb

|

417

P, element (nominally pure); A, alloy; C, commercial; L, laboratory.

⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ X ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ (?) ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅ ⋅⋅⋅

|

Liquid Metal Attack

Table 7.1

418

Chapter 7

Cu by molten lithium and the nonoccurrence of embrittlement of Al by molten lithium support the postulate, as in the latter system several intermetallic compounds form and none forms in the former two systems. Figure 7.3 shows the phase relationships most commonly associated with embrittlement couples. Embrittlement may not be observed in a possible embrittlement couple because the solid metal is quite ductile at the liquid metal temperature such that a brittle crack cannot be initiated or propagated, or it is excessively soluble in the liquid metal which leads to the blunting of the crack. Cadmium (m.p. 329°C)– liquid indium (m.p. 165°C) is an example. In such cases, the embrittling liquid metal may be dissolved in an ‘‘inert carrier’’ liquid metal of lower melting point. Mercury does not embrittle cadmium; it dissolves indium up to 70 at. % at 25°C and has been utilized as inert carrier for indium to effect LME in solid cadmium. The variation in ductility of cadmium with varying indium content in mercury carrier is shown in Fig. 7.4.

Figure 7.3 Phase relationships most commonly associated with embrittlement couples.

Liquid Metal Attack

419

Figure 7.4 Variation of ductility of polycrystalline cadmium as a function of indium content of mercury-indium surface coatings at 25°C [8].

Embrittlement has been observed even below the melting point of the embrittling metal, i.e., even when it is in the solid state and is in intimate contact with the ‘‘solid’’ metal of the couple. Embrittlement by solid cadmium of titanium and steel [2] has been reported. Presumably, the embrittlement is caused by the vapor phase of the solid metal with the lower melting point. Although this type of embrittlement is called solid metal–induced embrittlement (SMIE), this apparently follows the same mechanism of LME. Severity of Embrittlement The severity of embrittlement is related to the chemical nature of the embrittling species and depends on such factors as strength, alloying elements, and grain size, which determine the properties of the solid metal. It does not depend on the time of exposure to the liquid metal before testing or whether the liquid is presaturated with the solid. However, the severity as well as the occurrence or nonoccurrence of LME depend on the presence or absence of stress concentration such as preexisting cracks or flaws in the solid metal. High-stress concentrations lead to a greater severity of embrittlement. With regard of the dependence of severity on the chemical nature of the embrittling species, there appears to be a correlation with the electronegativities of the metals involved. Zinc (electronegativity 1.6) is more severely embrittled by liquid gallium (1.6) than by mercury (1.9). Cadmium (1.7) is most severely embrittled by liquid indium (1.7), less by liquid gallium (1.6), and not at all by thallium (1.8) or mercury (1.9). The embrittlement of pure aluminum by various

420

Chapter 7

mercury solutions is shown in Fig. 7.5. From the observations made in various systems it can be concluded that (1) the maximum embrittlement occurs when the solid metal and the embrittling metal have the similar electronegativity, and (2) the degree of embrittlement induced in the solid metal decreases as the difference in electronegativity between the metals of the couple increases. The quantitative definition of electronegativity is the power of an atom in a molecule to attract electrons to itself. The difference in electronegativity is a measure of the tendency for two elements to form ionically bonded compounds. For the metals with the similar electronegativity such affinity is the least, which means that there will be little mutual solubility. Apart from this support for the empirical rule stated earlier, the fundamental significance of the correlation between the severity of embrittlement and electronegativity has not been understood.

Figure 7.5 Embrittlement of polycrystalline pure aluminum by various mercury solutions [4].

Liquid Metal Attack

421

Requirements for Embrittlement The general requirements for LME to occur in a ductile metal are as follows: 1. There must be intimate contact or wetting of the solid metal by the liquid metal. 2. The solid metal must be sufficiently stressed to produce plastic deformation. 3. There should be an adequate supply of liquid metal. An intimate contact between the solid metal and the liquid metal is the most critical and mandatory condition for LME. This is required to initiate embrittlement as well as to ensure the presence of the liquid metal at the tip of the propagating crack to cause brittle failure. A thin oxide film on the surface of the solid metal may intervene in the process of embrittlement. Local breakdown of the film by the acting stress or plastic deformation would expose freshly created surface and establish a true interface between the solid and the liquid metals. A few metals wet oxides, e.g., molten gallium will wet aluminum oxide and molten lithium will wet iron oxides; the specificity of the environment may be attributed to this factor. The surface wetting by a liquid metal can lead to intergranular penetration originating at the interface. In general, this can only happen when the grain boundary dihedral angle contained by two liquid–solid surface tension vectors approaches zero. On the other hand, wetting over large areas can be achieved as long as the dihedral angle is less than 90°. Grain boundary penetration is therefore not a natural consequence of wetting and is also not relevant to the occurrence of LME, as has been pointed out earlier. The wettability can be enhanced by alloying additions to the liquid metal. For example, it is difficult to make pure mercury embrittle aluminum alloys because of the large contact angle between mercury and aluminum oxide. Additions of a few percent of zinc or gallium reduces the contact angle and renders the wetting action more effective. Some amount of plastic deformation is required for the cracking process (discussed later in Section 7.2.4). A general yielding of the solid metal is not envisaged. The yielding or deformation may be localized in a few grains. For this purpose some stable obstacle to slip serving as stress concentrator should be present. The grain boundaries,a twin or a kink band produced during deformation, second phase particles, and notches can provide such a situation. An adequate supply of liquid metal is necessary to adsorb at the obstacle and subsequently at the propagating crack tip. The gross amount of liquid need not be large; a few monolayers of liquid metal atoms are necessary for LME and it has been reported that even micrograms of liquid lead can cause LME in 75mm-thick tubes [3]. If the solid metal is notch-brittle, liquid metal is not required to be present at the propagating crack tip because the crack, once initiated, will propagate in a brittle manner.

422

7.2.2

Chapter 7

Factors Influencing LME

Effect of Grain Size The yield stress and the fracture stress of a metallic material normally bear a linear relationship with the inverse square root of grain diameter, which is known as the Hall-Petch relationship: σ ⫽ σ1 ⫹ kd⫺1/2 where σ ⫽ yield stress or fracture stress d ⫽ grain diameter di, k ⫽ constants The same relationship holds true for LME as well. A linear decrease of fracture strength as a function of d⫺1/2, where d is the average grain diameter, has been observed for copper and iron in molten lithium, 70–30 brass in mercury, and zinc in mercury, indicating that coarser grained materials are more susceptible to LME. Figure 7.6 shows the variation of flow and fracture stresses of amalgam-

Figure 7.6 Variation of flow and fracture stresses of amalgamated zinc specimens with grain size at 25°C [8].

Liquid Metal Attack

423

ated zinc with grain size. In this case, the fracture is nucleation-controlled in region I but propagation-controlled in region II. The grain size dependence of LME is indicative of a reduction in cohesive strength of the material rather than an effect of the penetration or dissolution of liquid into the grain boundary. Effect of Temperature LME takes place at temperatures above the melting point of the liquid metal component, except for the few cases of embrittlement caused by the vapor phase (SMIE). In the vicinity of the melting point of the liquid metal, LME is relatively temperature-insensitive. At higher temperatures, a brittle-to-ductile transition occurs in many systems over a temperature range and the ductility is restored (Fig. 7.7). The fracture stress in air is regained at and above the transition temperature. The effect is generally ascribed to the increased ductility of solid with increase in temperature. The brittle-to-ductile transition temperature is dependent on the presence of notch, grain size, and strain rate. The transition temperature is raised in the presence of notches. An increase in strain rate and a decrease in grain size increases the transition temperature. Effect of Strain Rate In addition to the effect on brittle-to-ductile transition temperature, as mentioned above, the strain rate of test may be an important factor for the occurrence of LME. It has been reported [4] that in precracked aluminum single crystals tested in liquid gallium at the crack tip, embrittlement was not observed at a slow strain

Figure 7.7 Temperature dependence of strain at fracture for (a) unamalgamated and (b) amalgamated zinc single crystals of approx. 1 mm diameter [8].

424

Chapter 7

rate (10⫺4 cm/s) of testing, but failure in a brittle manner was observed when the strain rate was increased by several orders of magnitude. The effect of strain rate appears to be related to the increase in yield strength and this corresponds to an increase in the LME susceptibility. Effect of Alloying Some solid metals are embrittled in their pure state, e.g., zinc by mercury and cadmium or aluminum by liquid gallium. On the other hand, pure iron is not embrittled by mercury and pure copper is relatively immune in liquid mercury (coarse-grained copper is embrittled). However, iron becomes susceptible to embrittlement in mercury if alloyed with more than 2% Si, 4% Al, or 8% Ni (Fig. 7.8). The susceptibility of copper to embrittlement in mercury is increased many fold when alloyed with Zn, Al, Ge, or Si. The same has been observed for zinc alloyed with a small amount of copper or gold when embrittled in mercury. The increase in the yield strength of the material on alloying is considered responsible for the increased susceptibility. It is generally observed that the high-strength alloys are more severely embrittled than low-strength alloys based on the same metal. This has been shown true for a series of commercial aluminum alloys when wetted with Hg ⫹ 3% Zn amalgam and for AISI 4340 steel quenched and tempered to different hardness when wetted with molten lithium [1]. The factors contributing to the increase in yield strength, i.e., reduced tendency for cross-slip due to lowering of stacking-fault energy, occurrence of coarse slip, dislocation source locking by impurities or precipitations, have been correlated

Figure 7.8 Effects of nickel additions to iron on the LME susceptibility of smooth tensile specimens in liquid mercury [8].

Liquid Metal Attack

425

with the susceptibility to LME for some alloy systems. For example, in copper alloys (alloyed with Al, Si, Zn, or Ge) the ratio of fracture stress in mercury to flow stress in air, σF /σY, has been shown to decrease progressively with increase in yield stress; at the same time this ratio increases linearly with increasing energy over the range 0–25 ergs, as well as with the increasing electron/atom ratio in these alloys (Fig. 7.9). In iron, a nickel addition of more than 8% gives rise to martensite with coarse slip lines. In precipitation-hardening aluminum and copper alloys, maximum susceptibility to LME coincides with the peak strength of the alloys. All of these point to the generation of stress concentrators as a result of alloying. It is interesting to note that precracked or notched samples of iron-nickel alloys having 2– 8% Ni are readily embrittled in mercury. Effect of Solute Additions to the Liquid Metal It can be seen in Fig. 7.1 that the progressive addition of lead to liquid bismuth reduces the embrittlement of copper [5]. Similar effects have been reported for additions of Sb, Cd, Tl, or Zn. Several minor additions have been made to mer-

Figure 7.9 Embrittlement of copper alloys as a function of stacking-fault energy and electron/atom ratio. (After Stoloff et al. [5].)

426

Chapter 7

cury at room temperature to investigate their effect on the LME of a number of solid metals. The results are varied. For example, zinc and gallium enhance the embrittlement of aluminum significantly but decrease the severity in α brass. The enhancement of severity may be a consequence of improved wetting by the liquid metal.

7.2.3

Delayed Failure

Delayed failure is a generic term applied to failures under sustained load after a period of time. In liquid metal environments the embrittlement and failure of some metals are time-dependent. The term ‘‘static fatigue’’ is also applied to such phenomena as the time to failure increases with a decrease in applied stress and the stress versus time-to-fracture curve assumes the nature of a typical S-N curve encountered in fatigue exhibiting a ‘‘static endurance list.’’ Figure 7.10 shows the delayed failure curve of Cu-2% Be alloy at room temperature wetted with Hg–2% Na amalgam. Attempts to induce delayed failure in notch-sensitive metals like zinc, cadmium, and iron-aluminum alloys in appropriate environments have not been met with success. In these systems crack nucleation and propagation occur instantaneously. On the other hand, notch-insensitive aluminum-copper and copper-beryl-

Figure 7.10 Delayed failure of copper–2% Be alloy wetted with Hg–2% Na amalgam at room temperature [8].

Liquid Metal Attack

427

lium alloys in liquid mercury environments exhibit delayed failure. Delayed failure has also been reported for AISI 4130 steel in molten lithium. Investigations have shown that a significant period of incubation or inactivity exists during which the exposed specimens suffer no permanent change of mechanical properties, which indicates that the role of diffusion is insignificant in this type of embrittlement. Age-hardable alloys exhibit the lowest time of fracture in the maximum hardened state. The susceptibility increases with prior strain or cold work. It is generally believed that grain boundary penetration of the solid metal by the liquid metal occurs in the presence of applied stress and some critical depth of penetration gives rise to brittle fracture propagation. However, the mechanism for this has not been established.

7.2.4

Mechanisms of LME

There are three models proposed for the mechanism of liquid metals embrittlement. These are: 1. Reduction in surface energy model 2. Stress-assisted dissolution model 3. Adsorption-induced reduction in tensile cohesion/shear strength model Reduction in Surface Energy Model According to this model, the liquid metal embrittlement is associated with a reduction in the surface energy of the solid metal as a result of the adsorption of liquid metal species [1,6]. The interfacial energy plays a big role in the generation of fracture and a truly brittle fracture should be associated with energies of the order of 103 ergs/cm2 or less. Values of the similar order have been calculated in several embrittling systems, e.g., Cu–liquid Bi, brass–liquid Bi, Fe–liquid Li, Cu–liquid Li, and brass–liquid Hg. The model envisages that the fracture propagates simply by the progressive rupture of atomic bonds and the creation of new surface. The opponents of this model argue that the energy involved in crack propagation are greater than the surface energy of the solid by several orders of magnitude and the proposed model does not suggest how this energy involved in plastic deformation is lowered so that a brittle crack propagation can be accomplished. Stress-Assisted Dissolution Model In this model, embrittlement has been envisaged to be an outcome of a very rapid localized dissolution process occurring at the crack tip under the influence of an applied stress. Volume diffusion of the dissolved solute through the liquid controls the propagation or the crack. Crack velocities of the order of tens of centimeters per second have been claimed to be achieved through the increased solubility of the solid at the crack tip [7].

428

Chapter 7

This model suffers from the simple fact that a majority of the embrittlement complex is mutually insoluble and it is unlikely that the simple application of stress would change the solubility drastically. The embrittlement process also in most cases is instantaneous and not influenced by time-dependent kinetic process. According to this model, higher dissolution would be expected at a higher temperature with a concurrent increase in embrittlement, but the practical evidences are to the contrary; a brittle-to-ductile transition is encountered with an increase in temperature in a number of systems. Adsorption-Induced Reduction in Tensile Cohesion/Shear Strength Model The tensile decohesion model [8] treats the liquid metal embrittlement as a special case of brittle fracture that is normally encountered in metals at low temperatures in inert environments. The reduction of the shear strength model has been extended to explain the occurrence of ductile fracture in LME. Both of these models are based on chemisorption of the embrittling species at the crack tip or at the sites of stress concentration at the surface of the solid metal, bringing about a localized reduction in the strength of the atomic bonds. The situation at the crack tip has been schematically represented in Fig. 7.11. The bond A-A0 constitutes the crack tip and B is the liquid metal atom. σ represents the largest tensile fracture stress and τ the largest shear stress on the most

Figure 7.11 Schematic representation of the displacement of atoms at the crack tip. (After Kamdar [8].)

Liquid Metal Attack

429

favorably oriented slip plane S-P near the tip. F represents the increasing applied force. Chemisorption may occur as such or only after the A-A0 bonds have been strained to some critical value. The bonds get weaker because of the electronic rearrangement and the crack propagates if the applied stress exceeds the reduced breaking strength. The liquid metal then gets stably chemisorbed on the fracture surface and surface diffusion of the liquid metal atoms over the chemisorbed layer, which is a low-activation energy process, feeds the crack tip with the necessary embrittling species. The crack propagates through repetition of the process. The shear strength of the atomic bonds at the crack tip may also get reduced because of chemisorption. The reduced shear strength facilitates nucleation of dislocations or slip at low stresses at or near the crack tip. The increased plasticity produces a large localized plastic strain ahead of the crack tip that is effective in nucleating voids at precipitations, inclusions, or at subboundaries in single crystals. The coalescence of the voids produces a ductile fracture. The localized increased plasticity results in an overall reduction in the strain at failure. The process has been schematically represented in Fig. 7.12.

7.2.5

Preventive Measures

Prevention or reduction in the occurrence of liquid metal embrittlement can be achieved by the following means: 1. Introduction of impurity atoms in the solid having more affinity for sharing electrons with the liquid metal atoms than for the solid metal atoms has been successful in reducing embrittlement in some cases. Examples are the addition of phosphorus to Monel to reduce embrittlement in liquid mercury or the addition of lanthanides to leaded steels. The addition of gold to silver inhibits its embrittlement by gallium, presumably because of the formation of strong Au-Ga or Ag-Au bonds at the surface which counters the weakening influence of gallium on Ag-Ag bonds. 2. As discussed in the section, ‘‘Effect of Solute Additions to the Liquid Metal,’’ the addition of a second metal to the embrittling liquid decreases the embrittlement in some cases. 3. Providing an effective barrier in between the solid and liquid metals is an effective measure. The barrier may be a metallic one that is not embrittled by the liquid metal concerned, or a ceramic or a covalent material coating. 4. Cladding with a soft, high-purity metal is at times employed. An example would be zircaloy clad with pure zirconium to resist embrittlement in liquid cadmium. 5. Elimination of the embrittling liquid metal, particularly where a chance contamination produces LME, is advised (see example given in Section 7.2.6). 6. Reduction in the level of applied or residual stress below the static endurance limit, where possible, is an obvious measure to reduce embrittlement.

430

Chapter 7

Figure 7.12 Schematic representation of the mechanism of crack growth by microvoid coalescence. (a) Inert environment. (b) Embrittling liquid metal environment. (After Kamdar [3].)

7.2.6

LME Failures in Practice

The occurrence of LME has been encountered during processing of the metal (e.g., galvanizing and hot rolling), during fabrication of the material (e.g., in the nuclear industry and in smelters), and during chance melting down of the contaminating embrittling metals. During the hot dipping process, intimate contact between the molten metal and the component to be coated must occur and if the latter is stressed conditions for LME develop. In a continuous galvanizing operation of steel wires, snapping

Liquid Metal Attack

431

of the wire was encountered at the emerging point from the molten bath. The occurrence was more in the case of steel wires with higher carbon content or above a certain diameter of the wire. The incidence of such failures could be minimized by lowering the slant of the emerging wire over the pulley, apparently because of the reduction in stress. The intrinsic higher strength of the high carbon steels made them susceptible to LME, whereas under the identical conditions the low-carbon-steel wires did not suffer from LME. The hot shortness of copper-containing steels during hot rolling is an example of LME. Iron oxides form preferentially at the surface below which a copperenriched phase forms. This phase melts at the hot working temperatures and penetrates along the grain boundary under the action of rolling stresses. Similar hot shortness have been reported in lead- or tellurium-containing steels where these elements are added for the improvement of machinability. The combined presence of a molten metal and stress makes the processes of welding, brazing, and soldering ideal for liquid metal embrittlement. The condition of good wellability is enhanced by the use of fluxes in these operations. In one reported case [9], where the operating temperatures of aircraft radiations were raised by a replacement of lead-tin solders with a higher melting point lead-silver solder, extensive cracking of radiations followed and the practice was discontinued. The embrittlement resulted from the improved wettability of steel by molten lead with the addition of silver. Zircaloy-2, used as cladding material for fissionable fuel rods in water-cooled nuclear reactors, has been reported to fail by LME in contact with liquid cadmium, which appears as a fission product in the operation. Cadmium plating on steel and titanium bolts and nuts has been reported [3] to cause failure of the fasteners used in military jet engine compressor ducts or in solid fuel rocket engines where aerodynamic heating melted the cadmium. Failure of titanium jet compressor discs by LME in contact with cadmium-plated bolts has also been reported. The incidence of a disaster from LME failure because of a chance contamination of liquid metal took place in the cyclohexane plant at Flixborough in 1974 [10]. The unsupported length of a pressurized piping collapsed, killing 28 people and causing widespread destruction. Many of the stainless steel pipes recovered after the fire showed extensive cracking. Investigations revealed that zinc from galvanized walkways, staircases, and girders had been transported to the external surfaces of pipes, either as molten droplets or vapors, and had caused the embrittlement.

7.3 CORROSION BY LIQUID METALS Corrosion by liquid metals becomes a matter of concern when they have to remain in contact with the solid metal over a long period. For example, due to their

432

Chapter 7

excellent heat transfer properties liquid metals are being extensively used in nuclear power generation plants, e.g., molten sodium in liquid metal fast-breeder reactors (LMFBR) or molten sodium-potassium as static heat sinks in automotive and aircraft valves. Heat transfer systems utilizing heat pipes use a liquid metal, e.g., lithium, sodium, or sodium-potassium, as the working fluid. These and the other applications of liquid metal make the containment material vulnerable to corrosive attack, which sometimes reduces its life considerably. Liquid metal corrosion can take place through any one or combination of the following process: 1. 2. 3. 4.

Direct dissolution Corrosion product formation Elemental transfer Alloying

Direct dissolution is the release of atoms of the containment material into the melt. As the adjacent melt becomes saturated with the dissolving metal, the dissolution reaction decreases or ceases altogether. However, in a nonisothermal liquid metal system such a situation may not be attained because of the convection from hotter to colder regions. As a consequence, the dissolved metal from the ‘‘hot leg’’ is carried to the ‘‘cold leg’’ where it gets deposited. Plugging of coolant pipes as a result of such depositions of dissolved species in the colder zones has been encountered. The dissolution may be uniform or selective. For example, preferential leaching of silicon from stainless steels in molten sodium and preferential dissolution of nickel from a type 316 stainless steel in molten lithium sodium or bismuth have been reported [11,12]. The selective leaching sometimes proceeds to such an extent that voids are left in the steel. The solubilities of refractory metals, as well as iron and nickel base alloys, are very low (of the order of a fraction of a ppm) in alkali metals. However, the interstitials in the metals such as carbon, oxygen, and nitrogen, and impurities in the liquid metal, particularly oxygen, dominate in the corrosion process either by the formation of corrosion products or by elemental transfer. The solubility of oxygen as Na2O in sodium is 3 ppm O2⫺ at 150°C, which enhances to 1000 ppm O2⫺ at 500°C. Na2O reacts with iron to form (Na2O)2FeO, which increases the apparent solubility of iron in molten sodium. In chromiumcontaining steels, Na2O reacts with chromium to form NaCrO2 as corrosion product. Niobium forms double oxides with Na2O, sometimes accompanied by a change in valency state, e.g., Nb2O5 ⫹ Na2O → Na2O, Nb2O3 ⫹ O2 In flowing sodium, the double oxides are removed from the metal surface. The attack is aggravated under a thermal gradient. Reduction of the oxygen concentration of the sodium to less than 3 ppm provides protection. This is achieved by

Liquid Metal Attack

433

installing a cold trap of steel wool in a bypass to the main sodium circuit. Oxides precipitated from the solution preferentially deposit on steel wool, which is periodically replaced. Nitrogen has been observed to increase the corrosion of steel in liquid lithium whether it is in the liquid metal or in the steel. The formation of a corrosion product such as Li3CrN5, or an equivalent one with iron, is held responsible for the aggravated attack. Oxygen-induced intergranular and transgranular penetration of niobium and tantalum by lithium, sodium, potassium, and sodium-potassium eutectic NaK has been reported [13]. The penetration has been observed in metals containing oxygen above a ‘‘threshold level’’ and is independent of oxygen in the lithium. The oxygen accumulated at the grain boundaries and habit planes reacts with lithium forming oxide which in turn reacts with the refractory metal to form ternary oxide corrosion products. Figure 7.13 illustrates such an attack of niobium by molten lithium. The problem of lithium penetration of niobium was eliminated by the

Figure 7.13 Intergranular and transgranular penetration of niobium containing 1500 ppm O and exposed to lithium for 100 h at 816°C. 750⫻. Etched with HF-HNO3-H2SO4H2O.

434

Chapter 7

addition of small amounts of zirconium 1 (ⱕ1%) to the niobium. Zirconium reacts with the oxygen dissolved in the metal to form ZrO2, thereby lowering the chemical activity of oxygen in solid solution. The attack on tantalum is likewise reduced by the addition of hafnium. The corrosion or reaction products sometimes form protective layers on the containment metal surface, thus reducing further attack. For example, in molten lead of high oxygen activity steel develops such a protective layer. An addition of aluminum or silicon to steel helps in forming surface-protecting corrosion products in the melts of low oxygen activity. The addition of zirconium to liquid bismuth or mercury has an inhibiting effect or the corrosion of steel in these liquid metals. The nitrogen present in steel forms a surface layer of ZrN that is thermodynamically a very stable compound and is an effective diffusion barrier. Elemental transfer refers to the net transfer of interstitials or impurities to or from a liquid metal. In such a case the liquid metal atoms do not react with the atoms of the containment metal atoms. Carburization of refractory metals and of austenitic stainless steels has been observed in liquid sodium contaminated with carbon. Decarburization of iron-chromium-molybdenum steels, particularly lower chromium steels, in liquid sodium or lithium is another example of element transfer. When two solid metals are in contact with a liquid metal, elemental transfer can lead to the intermetallic compound formation. For example, for aluminum and molybdenum exposed to molten bismuth, intermetallic compounds Al3Mo, Al5Mo, and Al12Mo have been found on the molybdenum surface. An alloying action can be observed between the atoms of the liquid metals and the constituents of the material. It is advisable to avoid systems that form alloys or stable intermetallic compounds, e.g., nickel in molten aluminum.

REFERENCES 1. W. Rostoker, J. M. McCaughey, and H. Markus, Embrittlement by Liquid Metals, Reinhold, New York, 1960. 2. D. W. Fager and H. F. Spurr, Corrosion, Vol. 24, (209, 1969); Corrosion, Vol. 27, p. 72, 1971. 3. M. H. Kamdar, Liquid-Metal Embrittlement in Metals Handbook, 9th ed., Vol 11, American Society for Metals, Metals Park 1986, pp. 225–238. 4. A. R. C. Westwood, C. M. Preece, and M. H. Kamdar, Am. Soc. Metals Trans. Quart., Vol. 60, p. 723, 1967. 5. N. S. Stoloff, R. G. Davies, and T. L. Johnston, Environment Sensitive Mechanical Behavior, A. R. C. Westwood and N. S. Stoloff (eds.), Gordon and Beach, New York, 1966. 6. V. I. Likhtman, P. A. Rebinder, and G. V. Karpenko, Physico-Chemical Mechanics of Metals, Acad. Sci. USSR, Moscow, 1962. 7. W. M. Robertson, Trans. AIME, Vol. 236, p. 1478, 1966.

Liquid Metal Attack

435

8. M. H. Kamdar, Embrittlement by liquid metals, in Progress in Material Science, B. Chalmers, J. W. Christian and T. B. Massalski (eds), Pergamon Press, Oxford, Vol. 15, p. 287, 1973. 9. R. Chadwick, J. Inst. Metals, Vol. 97, p. 93, 1969. 10. P. J. L. Fernandes, R. E. Clegg, and D. R. H. Jones, Failure Analysis, Vol. 1, p. 51, 1994. 11. P. F. Tortorelli, in Metals Handbook, 9th ed., Vol. 13, American Society for Metals, Metals Park, 1987, pp. 56–60. 12. P. Roy, D. Dutina, and F. Comprelli, in Corrosion by Liquids Metals, Joseph E. Draby and John R. Weeks (eds.), Plenum Press, New York, 1970, pp. 1–20. 13. R. K. Kluch, ibid., pp. 177–196.

This Page Intentionally Left Blank

8 Hydrogen Damage

8.1 INTRODUCTION Hydrogen damage refers to the degradation of physical and mechanical properties of metals resulting from the action of hydrogen, which may be initially present inside the metal or accumulated through absorption. Most often the damage is associated with residual or applied tensile stress. The damage may manifest itself in several ways: 1. Loss of ductility and/or tensile strength 2. Internal damage due to defect formation 3. Sustained propagation of defects at stresses well below those required for mechanical fracture, and 4. Macroscopic damage, such as internal flaking, blistering, fissuring, and cracking The term hydrogen embrittlement has long been in use to describe some of these forms, but the generic term hydrogen damage should be preferred as many of the damages do not conform to the classical features of embrittlement, namely, the reduced load bearing capacity or fracture below the yield strength. Hydrogen damage has been encountered in many metals and alloys of engineering interest. The first report on hydrogen embrittlement of steels appeared in 1873 [1]. The high-strength steels are particularly vulnerable, and there have been many incidents of failure of oil drilling and other equipment made of high437

438

Chapter 8

strength steels working in ‘‘sour’’ oil fields as a result of hydrogen damage. Stainless steels of all types, aluminum, nickel, copper and their alloys, titanium and zirconium alloys, and refractory metals like tungsten, niobium, vanadium, and tantalum suffer from hydrogen damage. Hydrogen-induced subcritical crack growth has been suggested as the dominant stress corrosion cracking (SCC) mechanism for ferritic steels in particular, and for metastable stainless steels, nickel-base alloys, titanium alloys, and aluminum alloys.

8.2 SOURCES OF HYDROGEN Hydrogen may enter the metal from various sources. Atomic hydrogen, rather than molecular hydrogen, is considered to be the detrimental species in metals which may, however, be absorbed from a molecular hydrogen gas atmosphere. The hydrogen is readily available in environments such as water, water vapor, moist air, hydrocarbons, acids, hydrogen sulfide, and in various liquids and gases involved in chemical process operations. A major source of hydrogen in liquid metals is from water contained in the scrap used as charge material in the furnace, from the slag ingredients, or from the refractory materials of the furnace linings. The solubility of hydrogen decreases with decreasing temperatures in metals such as iron, nickel, cobalt, copper, chromium, and aluminum, and hydrogen gets entrapped in these metals during solidification after melting or welding operations. The entrapped hydrogen is responsible for solidification porosity and the formation of hydrogen ‘‘flakes’’ or ‘‘fish-eyes’’ (Section 8.3.3) in rolled or forged steel products. High-strength steels are more prone to hydrogen pickup when melted under high-humidity conditions. As a result, their vacuum degassing becomes necessary before pouring. Liquid steels, in general, show rapid absorption of hydrogen after deoxidation. As such, the killed steels become more susceptible to flaking or blistering than the semikilled steels. Moisture contamination is a particular problem with electroslag refining (ESR) process, and special care must be taken to avoid contamination of slag materials. Hydrogen may be introduced during several stages of equipment or component manufacture, even before they are put in service. Welding, heat treatment in hydrogen-containing furnace atmospheres, acid pickling, or electroplating operations can each introduce hydrogen into the lattice of the metal. Underbead cracking is an embrittlement phenomenon in weldments that is associated with hydrogen pickup during welding operations. Moisture in electrode coating, high humidity in the atmosphere, and organic contaminants on the surface of the prepared joints are responsible for hydrogen entry in the metal. Upon rapid cooling of the weld, entrapped hydrogen can produce internal fissuring and other damages.

Hydrogen Damage

439

Hydrogen solubility is higher in the high-temperature fcc structure of iron than in the low-temperature bcc structure. Consequently, if the steel is cooled from the level of 1100°C in a furnace containing hydrogen atmosphere, e.g., cracked ammonia, entrapment of released hydrogen inside the metal results in hydrogen damage. During acid pickling or electroplating, and as a result of corrosion in service, atomic hydrogen is generated on the metal surface as a cathodic reduction product that diffuses in the bulk material. The diffusion rate is high particularly when the material is stressed. In the pickling of steel, the level of hydrogen absorption is dependent on both the bath temperature and the nature of the acid. The entry of hydrogen is promoted by cathodic poisons that inhibit recombination of the absorbed, discharged nascent hydrogen atoms. These poisons include cyanide and ionic species of sulfur, arsenic, selenium, bismuth, tellurium, phosphorus, iodine, and antimony. The role of sulfur as a poison is particularly important, as sulfur is commonly encountered in steels as well as in the environment. Cathodic protection of metals, by both galvanic coupling and impressed current, likewise facilitates the production of cathodic nascent hydrogen on the metal surface and can promote embrittlement of susceptible metals or alloys. Gases or liquids containing hydrogen sulfide can embrittle certain highstrength steels. Wet hydrogen sulfide environments are considered to be among the most effective in promoting hydrogen entry. In these cases, hydrogen sulfide reacts with the steel to form atomic hydrogen: Fe ⫹ H2S → FeS ⫹ 2H

(8.1)

The chemisorbed sulfur partially poisons the hydrogen recombination reaction and promotes hydrogen absorption. The damage is absent in solutions with pH values above 8 because the protective iron sulfide film formed on the metal surface stops the corrosion of steel. Cyanides, if present in the solution, destroy this protective film. The unprotected steel corrodes rapidly, and hydrogen damage is encountered. Embrittlement or cracking in steel can develop in the presence of only a few ppm of hydrogen sulfide. Hydrogen stress cracking is a serious problem in petrochemical equipment used to store and handle the sour or hydrogen sulfide–containing oils. Exposure to process fluids bearing hydrogen, as in catalytic cracking, can cause hydrogen entry into the material. Exposure to hydrogen gas or molecular hydrogen under high pressure and temperature facilitates hydrogen entry and induces damage in iron alloys, nickel alloys, and titanium alloys, Hydrogen gas even at one atmospheric pressure has been reported [2] to cause cracking in highstrength steel. The effect of hydrogen on the behavior of metal is the same irrespective of the source of hydrogen. However, there are some basic differences between gaseous and cathodic hydrogen absorption processes. Cathodic hydrogen absorbs on

440

Chapter 8

the surface of the metal as atomic hydrogen, whereas gaseous hydrogen absorbs in molecular form. Molecular hydrogen must dissociate to diffuse into the metal. At the same time, desorption of the loosely bound molecular hydrogen is also relatively easy. Consequently, the hydrogen buildup inside the metal varies substantially in the two processes for the equal hydrogen activities at the surface.

8.3 TYPES OF HYDROGEN DAMAGE Hydrogen damage encountered under different conditions has been described by numerous terminologies. The specific types of hydrogen damage may be categorized as follows: 1.

2. 3. 4.

Hydrogen embrittlement, which may be further subdivided as a) Loss in tensile ductility b) Hydrogen stress cracking c) Hydrogen environment embrittlement d) Embrittlement due to hydride formation Hydrogen blistering Flakes, fish-eyes, and shatter cracks Hydrogen attack

8.3.1

Hydrogen Embrittlement

Loss in Tensile Ductility Loss in tensile ductility is one of the earliest recognized forms of hydrogen damage. The entry of hydrogen into the metal results in significant decreases in elongation and reduction in area without the formation of any visible defects, chemical products, or cracking. The loss of ductility is only observed during slow-strain rate testing and conventional tensile tests. A drop of ductility in such tests from 42% to 7% has been reported for a carbon steel [3]. Tensile strength is also affected, but there is no loss in impact strength. As such, impact tests do not indicate susceptibility and are not recommended to determine whether embrittlement exists. The extent of loss in tensile ductility is a function of hydrogen content of the material, as shown in Fig. 8.1. The embrittling effect is caused by hydrogen atoms collecting interstitially between metal atoms causing local distortion of the metal lattice. The mobility of the dislocations is thus restricted; hence the ability of the lattice to deform. Hydrogen atoms diffuse preferentially along grain boundaries and zones where the lattice has already been distorted by cold working or hardening. The solubility of hydrogen in metals obeys Sievert’s law, with the concentration being directly proportional to the square root of the pressure or fugacity.

Hydrogen Damage

441

Figure 8.1 Loss of ductility in steel as a function of hydrogen content [23].

This type of damage is most often observed in lower strength alloys and has been encountered in steels, stainless steels, nickel-base alloys, aluminum alloys, and titanium alloys exposed to hydrogen. Figure 8.2 shows the ductility loss for several austenitic stainless steels in high-pressure hydrogen. A wide variation in hydrogen damage in these alloys is apparent. Type 304L is most susceptible and the stable austenitic alloys, such as 15Cr-25Ni, are minimally affected. The loss of ductility is temporary and can be reversed by the driving out of hydrogen from the metal. This is accomplished by heating the metal. The rate of recovery depends on time and temperature; lower is the time required at higher temperatures. However, heating above 315°C is not usually recommended due to the risk of high-temperature hydrogen attack. Hydrogen Stress Cracking Hydrogen stress cracking (HSC) refers to the brittle fracture of a normally ductile alloy under sustained load in the presence of hydrogen. This type of damage has been encountered in carbon and low-alloy steels, stainless steels, nickel alloys, and aluminum alloys. HSC has been studied most extensively in steels and has been described by various other names, i.e., hydrogen-induced cracking (HIC), hydrogen-assisted cracking (HAC), delayed failure, and static fatigue. The cata-

442

Chapter 8

Figure 8.2 Ductility loss of austenitic stainless steels in high-pressure hydrogen [21].

strophic cracking of high-strength steels in hydrogen sulfide environments, known as sulfide stress cracking, is a special case of HSC. The cracking of the embrittled metal is caused by static external stresses, transformation stresses (e.g., as a result of welding), internal stresses, cold working, and hardening. In the absence of a sharp initial crack, the hydrogen-induced fracture often initiates at subsurface sites where triaxial stress is highest. If a sharp crack is already present, as in the case of a fatigue or stress corrosion crack or a surface defect, the hydrogen cracking may initiate at the tip of the preexisting crack. High hydrogen concentration ahead of the crack tip then helps the subcritical crack to grow. The source of hydrogen for HSC may be gaseous hydrogen or the hydrogen generated in thermal processing, electrolysis, or corrosion. A total hydrogen content of as low as 0.1–10 ppm is sufficient to induce cracking. However, local concentrations of hydrogen are substantially greater than average bulk values.

Hydrogen Damage

443

Most often, the fracture occurs at sustained loads below the yield strength of the material. The fracture time increases as the load is decreased. The applied stress versus the fracture time plots resemble similar plots obtained in SCC or liquid metal embrittlement. Like the fatigue S-N curve, a threshold stress exists below which HSC does not occur. This threshold stress is a function of the strength level of the steel and the specific hydrogen-bearing environment. Generally, the threshold stress decreases as the yield strength and tensile strength of the material increase. Lower strength alloys below a minimum tensile strength are not usually affected. In ferrous alloys, HSC is generally restricted to those alloys having a hardness of 22 HRC or greater. The sensitivity increases with the increase in strength level, as shown in Fig. 8.3. A predominant feature of HSC is that the occurrence of the fracture is delayed, which implies that hydrogen diffusion in the metal lattice is important for the building up of sufficient hydrogen concentration at the regions of triaxial stresses for crack nucleation or at the crack tip for its propagation. The susceptibility to cracking, therefore, depends on the kinetic factors, such as hydrogen gas pressure and temperature, which influence the diffusion process. Increasing the hydrogen pressure reduces the threshold stress intensity for crack preparation and increases

Figure 8.3 Ductility versus hydrogen content for quenched and tempered steels at various strength levels. Figures in parentheses indicate the ultimate tensile strength in MPa [21].

444

Chapter 8

the crack growth rate for specific stress intensity value, as can be seen in Fig. 8.4 for AISI 4340 steel. The threshold stress intensity and crack growth rate are a function of the specific hydrogen environment. The threshold stress intensity is represented by K1H, which is lower than that measured in inert environments and is a measure of susceptibility to HSC. HSC is observed in the temperature range of ⫺100°C to ⫹100°C, the most severe being at about room temperature. At temperatures well below room temperature, the hydrogen diffusion rate is slow and the necessary hydrogen buildup at the crack tip for its propagation is also slow. At temperatures in excess of 120°C, the hydrogen in solid solution tends to be homogenized and the local buildup of hydrogen concentration needed to cause embrittlement may not occur. The deleterious effect of hydrogen on delayed failure is increased by the triaxiality of the stress state, i.e., with an increase in notch acuity, as shown in Fig. 8.5. Hydrogen concentration in the alloy is a function of the fugacity or the approximate concentration of hydrogen at the surface exposed to the environment. Therefore, hydrogen gas pressure and pH of the environment are controlling factors for the embrittlement. Certain constituents of the environment also may play a significant role. Hydrogen evolution poisons contained in the aqueous environments favor more hydrogen entry in the metal. On the other hand, traces of oxygen in hydrogen gas environment inhibit HSC in steels. The susceptibility of an alloy to hydrogen embrittlement is strongly dependent on hydrogen concentration in the metal (Fig. 8.4). The effect is reversible to the extent that if hydrogen is removed from the metal, brittle fracture is avoided. Figure 8.6 represents static fatigue curves for a 4340 steel, showing the reversibility of embrittlement with systematic baking out of the changed hydrogen. The term internal reversible hydrogen embrittlement is applied to describe this behavior. The reversibility is absent in case the hydrogen undergoes any type of chemical reaction after it has been absorbed within the lattice, or if any immediate and resolvable damage has taken place in the material structure. Hydrogen-induced cracking has also been observed as a consequence of mechanical testing at slow-strain rates in hydrogen environments. Like delayed failure, slow-strain rate embrittlement also requires a minimal level of applied stress for the failure to occur. The susceptibility is greater in stronger alloys and exhibits the similar temperature dependence. Additionally, it is strongly dependent on the strain rate, the embrittlement increasing with decreasing strain rate. At relatively high strain rates (⬃10⫺4 in./in./min) there may be little or no decrease in macroscopic plasticity because of hydrogen. Apparently, at these high strain rates, the alloy may fracture before any significant diffusion of hydrogen can take place. Figure 8.7 shows the variation in notch tensile strength of a high-strength steel as a function of strain rate and testing temperature. The susceptibility to embrittlement of steels depends to a large extent on their microstructure. Highly tempered martensitic structure with equiaxed ferrite grains

Hydrogen Damage

445

Figure 8.4 Dependence of crack growth rate in AISI 4340 steel on stress intensity at various hydrogen pressures at 24°C [21].

446

Chapter 8

Figure 8.5 Static fatigue curves for specimens of different notch sharpness [21].

and spheroidized carbides evenly distributed throughout the matrix possesses maximum resistance to embrittlement, compared to normalized or bainitic steels at equivalent strength levels. The resistance also increases with decreasing prior austenitic grain size. The presence of retained austenite is helpful, possibly because it either absorbs hydrogen or slows down crack growth. The effects of individual alloying elements on cracking susceptibility are associated with their effects on the heat treatment, microstructure, and strength of the steels. In general, carbon, phosphorus, sulfur, manganese, and chromium increase susceptibility and titanium decreases the sensitivity to HIC by decreasing the amount of hydrogen available for cracking. The behavior of stainless steels in hydrogen environments is dependent on their strength level. Ferritic stainless steels are extremely resistant to HSC because of their low hardness. However, in the cold-worked and as-welded conditions they are susceptible. Martensitic and precipitation hardening stainless steels are most susceptible because of their higher strength. Austenitic stainless steels are also highly resistant to hydrogen cracking in the annealed or lightly coldworked condition. A major factor for this resistance is the fcc structure of the

Hydrogen Damage

447

Figure 8.6 Static fatigue curves of a 4340 steel specimen showing the gradual recovery of ductility on baking at 570°C with increasing length of time [21].

alloy, which is relatively impermeable to diffusion of atomic hydrogen. In the heavily cold-worked condition, however, the austenitic stainless steels are prone to hydrogen cracking, which is attributed to the deformation-induced formation of martensite or the formation of a metastable iron-chromium-nickel hydride. Nickel and nickel alloys also are prone to HSC to a much lesser degree than steels because of their fcc structure. Cold working and aging treatment of alloys increase susceptibility to cracking. Cold working increases the strength of the materials, and aging leads to the segregation of sulfur and phosphorus to grain boundaries. The mode of fracture in hydrogen embrittlement can be both intergranular and transgranular. Intergranular cracking along the prior austenitic grain boundaries is generally observed in steels (Fig. 8.8), but the crack path depends to a large extent on the nature and degree of segregation or partitioning of alloying elements in the matrix and at the grain boundaries. For example, the integranular cracking

448

Chapter 8

Figure 8.7 Notch tensile strength of a high-strength steel as a function of testing temperature and strain rate [21].

Figure 8.8 Intergranular crack resulting from hydrogen embrittlement during chromium plating [23].

Hydrogen Damage

449

observed in an iron-carbon alloy has been reported [4] to have changed to transgranular mode with the addition of silicon. Similarly, a transition from transgranular to intergranular cracking in nickel alloys with the addition of antimony and tin has been reported [5]. There is evidence that the crack path depends on the stress intensity at the crack tip, being intergranular at lower stress intensities and transgranular at higher stress intensities. Fractures in low-strength steels show ductile dimple rupture, tearing, cleavage, and quasi-cleavage features. In high-strength steels, intergranular facets or quasicleavage features, depending on the stress intensity, are observed. Apparently flat cleavage fractures have often been reported [6] to exhibit dimples ruptures (microdimples) when viewed under the scanning electron microscope at magnifications higher than 5000⫻ (microdimples), which emphasizes the role of plastic deformation in the cracking process. The embrittlement may be highly localized at the cracked regions and the regions between the adjacent microcracks failing in a ductile manner may show evidence of dimpling. Figure 8.9 shows hydrogeninduced intergranular fracture region in a steel surrounding an MnS inclusion, surrounded by ductile fracture regions. The quasi-cleavage fracture surface

Figure 8.9 SEM fractograph of hydrogen-embrittled steel showing intergranular features near an MnS inclusion surrounded by regions showing dimpling [23].

450

Chapter 8

Figure 8.10 SEM fractograph of embrittled 304 stainless steel exposed to highpressure hydrogen showing quasi-cleavage features [23].

(Fig. 8.10) observed in 304 stainless steel bears the evidence of embrittlement due to deformation-induced martensite formation. The mode of fracture and fractographic features associated with HSC are often strikingly similar to those of SCC. Hydrogen-induced crack growth has been suggested by many authors as the dominant stress corrosion mechanism for ferritic steel in particular, and for metastable stainless steels, nickel-base alloys, titanium alloys, and aluminum alloys. However, there are certain distinguishing features between the two cracking processes: 1. 2.

The ‘‘specific ion’’ effect that characterizes SCC is absent in HSC. The application of cathodic potential or current, which retards or stops SCC, enhances the intensity of HSC.

Hydrogen Damage

451

3. Stress corrosion cracks generally originate at the surface, whereas hydrogen embrittlement failures originate internally. 4. HSC usually produces sharp singular cracks in contrast to the branching of cracks observed in SCC. Hydrogen Environment Embrittlement Hydrogen environment embrittlement refers to the embrittlement encountered in an essentially hydrogen-free material when plastically deformed or mechanically tested in gaseous hydrogen. This phenomenon has been observed in ferritic steels, nickel alloys, aluminum alloys, titanium alloys, and in some metastable stainless steels in hydrogen gas pressures ranging from 35 to 70 MPa. Embrittlement appears to be most severe near room temperature. The degree of embrittlement is maximum at low strain rates and when the gas purity is high. These characteristics are comparable with those observed for HSC. Therefore, there is marked disagreement as to whether this should be treated as a separate class of embrittlement. There is, however, a major exception: nickel alloys are very susceptible to hydrogen environment embrittlement, whereas they are relatively insusceptible to HSC. Embrittlement Due to Hydride Formation Embrittlement and cracking of a number of transition, rare earth and alkaline earth metals, such as titanium, zirconium, tungsten, vanadium, tantalum, niobium, uranium, thorium, and their alloys, result from hydride formation. Significant increases in strength and large losses in tensile ductility and impact strength are encountered. The brittleness is associated with the fracture of the hydride particle or its interface. The solubility of hydrogen in these metals is 103 –104 times greater than that of Fe, Ni, Cu, and Al and increases with decrease in temperature. The solubility tends toward saturation at low temperatures and at atmospheric pressure the composition of the solution approaches that of a finite compound hydride or a pseudohydride. Then either the crack may get arrested at the ductile matrix or its growth may continue by ductile rupture of the regions between the hydrides. Hydride formation is enhanced for some metal–hydrogen systems by the application of stress. In such cases, the stress-induced hydride formation at the crack tip leads to a continued brittle fracture propagation. Titanium and zirconium are known to form stable hydrides under ambient conditions when hydrogen is absorbed in excess of about 150 ppm. The size of the hydride particles is directly related to the kinetics of the nucleation process. Slow cooling from higher charging temperatures and high supersaturation tend to precipitate large particles. Absorption of hydrogen by these metals increases dramatically once the protective oxide film normally present on the metal is damaged through mechanical abrasion or chemical reduction. Surface contaminants, e.g., iron smears, enhance hydrogen intake

452

Chapter 8

and the absorption is accelerated at temperatures exceeding 70°C. Hydrogen is readily picked up during melting or welding, and hydride formation takes place during subsequent cooling. When sufficient hydrogen is present, the craking is attributed to the strain-induced formation of hydrides. With lower concentrations (⬃5 ppm), some other mechanism may be operative. Both titanium and zirconium have two allotropic forms: a lower temperature α phase and a high-temperature β phase. The α phase has a relatively low solubility for hydrogen and forms hydrides at low concentrations, whereas the β phase has a high hydrogen solubility and can form hydrides only at high concentrations. The embrittlement of α-titanium alloys is observed when tested at high strain rates and low temperatures. In contrast, the embrittlement in α-β titanium alloys is more pronounced when tests are conducted at low strain rates. In α-β alloys, hydrides form at the phase interfaces and cracking occurs by stepwise brittle cleavage of the thin hydride film, resulting in an integranular fracture at the αβ boundaries. In the refractory metals, i.e., tungsten, vanadium, tantalum, and niobium, hydrides formed are not stable; nevertheless, embrittlement and cracking are encountered.

8.3.2

Hydrogen Blistering

This type of damage is prevalent in low-strength unhardened steels and is caused by the pressure generated by the process of combination of atomic hydrogen into molecular hydrogen. The diffusing hydrogen atoms accumulate at internal macrodefects such as voids, laminations, or inclusion–matrix interfaces already present in the steel. At sufficiently high concentrations they tend to combine into molecular hydrogen, exerting an estimated pressure of several thousand atmospheres, which brings about the damage. Hydrogen blistering literally means the formation of surface bulgings resembling a blister (Fig. 8.11). The generation of gas in voids or other defect sites situated near the surface can lead to such a situation. The blisters often rupture producing surface crackings. Internal hydrogen blistering on a microscopic scale along grain boundaries (fissures) can lead to hydrogen-induced stepwise cracking. The interaction of accumulated hydrogen at an elongated inclusion–matrix interface may lead to delamination of a steel sheet or plate (Fig. 8.12). In ordinary rolled sheet or plate, banded structures containing elongated and flattened inclusions are common. Killed steels are more susceptible to blistering than semikilled steels because of greater hydrogen intake after deoxidation, but the nature and size of inclusions are overriding factors. Rimmed steels show a high susceptibility because of inherent presence of voids. Sulfur-bearing free-cutting steels are also specially prone because sulfur favors the hydrogen entry by acting as a cathodic poison. Hydrogen blistering is encountered mostly during acid pickling opera-

Hydrogen Damage

Figure 8.11 Blisters in a carbon steel plate formed by hydrogen [23].

Figure 8.12 Delamination of a steel plate as a result of hydrogen blistering.

453

454

Chapter 8

tions. Corrosion-generated hydrogen causes blistering of steel in oil well equipment and petroleum storage and refinery equipment.

8.3.3

Flakes, Fish-eyes, and Shatter Cracks

Flaking refers to small internal fissures that occur in steels when cooled from temperatures of the order of 1100°C in hydrogen atmospheres. These are also described by the terms fish-eyes, shatter cracks, or snowflakes and are common features of hydrogen damage in forgings, weldments, and castings. At higher temperatures of melting or welding or heat treatment in the austenite range, the solubility of hydrogen in steel is higher than in the solidified or low-temperature bcc state. The excess released hydrogen atoms accumulate at internal defects, combine to form hydrogen gas, and cause these types of damage. The mechanism of formation of these damages as well as many features are similar to those of hydrogen blistering and as such may be considered as a special class of hydrogen blistering. The extent of damage is dependent on the time of exposure to a hydrogencontaining environment. The cracks produced are readily detectable by radiographic or ultrasonic inspection, or by visual and microscopic observation of transverse sections (Fig. 8.13). On a fracture surface, the flakes appear as bright, highly reflective spots, which at higher magnifications (more than 6000⫻) often reveal striation patterns.

(a)

(b)

Figure 8.13 Hydrogen flake patterns in SAE 5145 steel (a) on unetched transverse section, 1⫻. (b) Microscopic features in eched condition, 240⫻ [23].

Hydrogen Damage

8.3.4

455

Hydrogen Attack

Hydrogen attack is a form of damage that occurs in carbon and low-alloy steels exposed to high-pressure hydrogen gas at high temperatures for extended time. The damage may manifest itself as loss in strength of the alloy or formation of cracks and fissures, and is prevalent at temperatures above 200°C. Under these conditions the reaction occurs between absorbed hydrogen and the iron carbide or the carbon in solution resulting in the formation of a hydrocarbons, generally 2H2 ⫹ Fe3 → CH4(g) ⫹ 3Fe

(8.2)

The methane produced does not dissolve in iron lattice and internal gas pressures lead to the formation of fissures or cracks. The generated defects or the decarburization itself may lower the strength and ductility of the steel. The decarburization may take place internally or at the surface. In the latter case, the decarburized layer grows to increasing depths as the reaction continues. Cracking may develop in the metal under tensile stress, or the progressive weakening of the metal may result in failure by some other mechanism. The damage is dependent on temperature and hydrogen partial pressures. Surface decarburization takes place at temperatures above 540°C and the internal decarburization at temperatures above 200°C. Hydrogen attack can take several forms within the metal structure depending on the severity of attack, stress, and the presence of inclusions in the steel. In the absence of stress, a component may undergo a general surface attack. The fissures formed are parallel to the metal surface. Areas of high-stress concentration are often the initiation point of hydrogen attack because of preferential diffusion of hydrogen to these areas. Isolated decarburized and fissured areas are often found adjacent to weldments. Blisters and laminations may also result from severe hydrogen attack. Resistance of steels to hydrogen attack is related to the stability of carbides. The addition of carbide-stabilizing elements such as vanadium, titanium, chromium, and molybdenum has beneficial effects. Hydrogen attack does not occur in austenitic stainless steels. The extent of hydrogen attack is a function of exposure time. Nelson [7] related the susceptibility of the failure of carbon and lowalloy steels from the formation of methane gas inside the steel, with the partial pressure of the hydrogen in contact with steels and temperature. The operating limits of these steels are often represented by Nelson curves (Fig. 8.14). Decarburization has also been reported in nickel alloys during heat treatment at 1100°C in hydrogen atmospheres. Hydrogen reacting with foreign elements other than carbon in the matrix at high temperatures and thus producing gaseous reaction products may also cause damage in some materials. Examples are the formation of steam (H2O) in welded steels, Cu, Ni, and Ag by reaction with oxygen; and formation of ammonia (NH3) in molybdenum by reaction with nitro-

456

Chapter 8

Figure 8.14 Nelson curves showing the operating limits of carbon and alloy steels in contact with hydrogen at high temperature and pressure.

gen. The disintegration of oxygen-containing copper in the presence of hydrogen is a typical example of hydrogen attack.

8.4 THEORIES OF HYDROGEN DAMAGE Several theories have been proposed for the mechanism of various types of hydrogen damage encountered. The prominent among these are as follows: 1. 2. 3. 4. 5. 6.

Hydrogen pressure theory Surface adsorption theory Decohesion theory Enhanced plastic flow theory Decarburization, and Hydride formation

Hydride formation is an established mechanism for embrittlement of titanium, zirconium, and several other metals and has been discussed in the section ‘‘Embrittlement Due to Hydride Formation.’’ However, the embrittlement of steels, nickel alloys, and aluminum alloys is not due to hydride formation. Decarburization is the cause of hydrogen attack of steels at high temperatures, which has been

Hydrogen Damage

457

discussed in the previous section. While these mechanisms are well understood, controversies exist regarding the mechanism of hydrogen embrittlement, particularly HSC, in steels and in other alloys. The absorbed and dissolved hydrogen could have a variety of effects, and the various theories proposed (1–4) differ on this count. Since the critical events associated with HSC occur on an atomic scale at inaccessible crack tips, the theories cannot easily be verified. Each of these theories can explain certain observations related to the phenomenon, whereas the others remain unexplained. The salient features of these theories have been discussed subsequently. The phenomenon of hydrogen trapping may be considered in this context. Diffusion studies of iron and steel have shown lag time for hydrogen diffusion through these materials before a steady-state diffusivity compatible with that expected theoretically is achieved. The lag time is attributed to the interaction of hydrogen with impurities, structural defects, or microstructural constituents in the metal, which is referred to as ‘‘trapping.’’ Hydrogen accumulates at these internal interfaces, called ‘‘traps.’’ These hydrogen traps may be mobile (such as dislocation and stacking faults) or stationary (such as solute atoms, particle interfaces, grain boundaries, cracks, and voids). The traps have been classified as reversible or irreversible [8]. Short-duration trapping of hydrogen is referred to as reversible, whereas a long residency time for hydrogen characterized by a high binding energy is termed irreversible trapping. Deep or reversible traps act as sinks for hydrogen and reduce the population of hydrogen at the crack tip, thus increasing the resistance to HIC.

8.4.1

Hydrogen Pressure Theory

This is the earliest mechanism of hydrogen embrittlement proposed by Zappfe and Sims [9]. According to this theory, the atomic hydrogen diffuses through the metal lattice and accumulates at preexisting voids and other internal surfaces in the alloy. As the concentration of hydrogen increases at these sites, it recombines to form molecular hydrogen. A high internal pressure is created that enhances void growth or initiates cracking. The sequence is represented schematically in Fig. 8.15. Pressure theory can successfully explain the phenomenon of hydrogen blistering and also the occurrence of internal defects such as flakes and shatter cracks. The unusual temperature and strain rate dependency of hydrogen embrittlement could be explained in terms of the diffusion rate of hydrogen. At high strain rates or low temperatures, the diffusion of hydrogen to the voids would not be sufficient and the susceptibility to embrittlement is thereby reduced. However, the theory has been less successful in explaining the brittle behavior of high-strength steels, or nonbrittle fracture with loss of ductility found in some low-strength materials.

458

Chapter 8

Figure 8.15 Schematic representation of hydrogen diffusion and recombination at internal voids.

Theoretical analyses have shown [10] that voids can become pressurized if hydrogen is transported by dislocations into voids faster than hydrogen escapes from voids, but there is no experimental evidence that such an effect occurs.

8.4.2

Surface Adsorption Theory

Surface adsorption theory is based on the well-known Griffith criterion for fracture in ideally brittle solids, which relates the energy for fracture to the energy required to produce new surfaces and is represented by σc ⫽ (2Eγs /πc)1/2

(8.3)

where σc is the fracture stress necessary to cause the propagation of an elliptical crack of length 2c, E is Young’s modulus, and γs is the surface energy. The fracture stress is proportional to the square root of the surface energy and any reduction in the term would increase the brittleness. The mechanism proposed by Petch and Stables [11] suggests that the reduction in surface energy is caused by the adsorption of hydrogen on the walls of microcracks. The fracture stress is thereby reduced and this enhances crack propagation at stress levels below those typically experienced for a particular alloy in an inert environment. In metals, however, the Griffith microcracks are not inherently present but are produced by plastic deformation. Accordingly, the crack should propagate when (σ ⫹ P)nb ⬎2(γs ⫹ γp)

(8.4)

Hydrogen Damage

459

where σ P n b γs and γp

⫽ ⫽ ⫽ ⫽ ⫽

applied stress hydrogen pressure in voids number of dislocations comprising cracks Burger’s vector surface energy and plastic work term, respectively.

Since γp is considerably greater than γs, any reduction in the latter through adsorption would not change the energy requirement for crack propagation significantly. Moreover, the discontinuous crack growth that has been observed for hydrogen cracking is not explained by this mechanism.

8.4.3

Decohesion Theory

The theory has been originally proposed by Troiano [12] and subsequently developed by Oriani [13] and others. The theory envisages that the effect of hydrogen is to lower the cohesive strength of the lattice. Hydrogen distributed uniformly throughout a metal lattice is nondamaging because its concentration is so small. Under the influence of an applied stress, hydrogen would diffuse to the regions of high triaxial stress provided by a crack tip or a network of internal voids. It has been suggested that the electrons from the hydrogen atoms would enter the d bands of the metallic cores and the increase of the electron concentration of these bands produces an increase in repulsive forces between metallic cores, i.e., a decrease in the cohesive strength of the lattice. Since a crack would propagate in a brittle manner if the ratio of the largest tensile fracture stress, σ, at the crack tip to the largest shear stress, τ, on slip phases intersecting crack tips was greater than the ratio of the ideal cleavage stress, σmax, to the ideal shear stress, τmax, the brittle fracture would be preferred by low value of σmax /τmax. The mechanism envisages the fracture as a perfect cleavage. However, by the Tresca criterion the group of atoms in a small element should slip apart, rather than separate in a tensile mode, because the stress required for shear is one-half that required for tensile fracture. The same considerations hold good at the tip of a growing crack and slip should occur at a lower stress than tensile or cleavage separation. Thus, some form of plastically based separation ought to occur at the tip of a growing cleavage crack. The fractographic evidence of plastic flow even in high-strength steels strengthens the postulate of shear separation. It has been suggested that in such cases the decohesion is facilitated by hydrogen and there is plastic tearing of ligaments unaffected by hydrogen between decohered regions. There have been suggestions [14,15] that plastic flow around crack tips or the stress concentrations produced by dislocation pile-ups near crack tip can indirectly influence the decohesion process.

460

8.4.4

Chapter 8

Enhanced Plastic Flow Theory

Based on the fractographic evidence that HSC produced fracture surfaces showing 100% microvoid coalescence (dimples) [15] and the observation of shallow dimples (microdimpling) in the flat cleavage facets at magnification higher than 5000⫻ in some steels [6], it has been proposed that hydrogen enhances the processes of plastic flow associated with the propagation of fracture by making dislocations move at reduced stresses at the crack tips. The movement of screw dislocations is generally considered to be favored. Further, it has been suggested that hydrogen is subsequently transported by dislocations to the regions ahead of cracks. If appreciable reductions or flow stress due to dissolved hydrogen were localized to regions just ahead of crack tips, then embrittlement due to strain localization would be expected. In a modified version of the model [6], the adsorption of hydrogen has been considered to ease the generation of dislocations at the crack tip which, in turn, decreases the surface tension at the crack tip spreading apart the surface atoms. The principal objection to this mechanism is that there has been evidence that hydrogen may induce decreases or increases in flow stress according to the circumstances. Furthermore, the plastic flow ahead of cracks is largely controlled by the large number of solute atoms and precipitates in steel whether hydrogen is present or not.

8.5 PRACTICAL EXAMPLES In the petroleum refining industry, hydrogen blistering is a problem in the vapor recovery section of catalytic cracking units and in the low-temperature areas of the reaction effluent section of hydrotreating and hydrocracking units. Hydrogen blistering has also been encountered in the overhead systems for sour water stripper towers and amine regenerator towers, as well as in the bottom of amine contactor towers [17]. Corrosion-generated hydrogen is the cause for blistering of steel in oil well equipment. A case of hydrogen blistering produced on the interior surface of a low-carbonsteel tank used to transport concentrated sulfuric acid has been reported [18]. Slight dilution of the acid had resulted in chemical attack on the 7/32-in.-thick tank wall and the diffusion of atomic hydrogen into the steel. The blistering was observed within 1/16 in. of the inside surface of the tank wall, which was attributed to the presence of internal discontinuities in an otherwise normal metal that had partially spheroidized structure and was free from defects. Cracks were observed during machining operations in a 6-in.-diameter alloy steel shafting made by forging [19]. The radiography of a 1/2-in. disc representing the steel shaft revealed a large number of cracks (Fig. 8.16). The microstructure indicated that homogenization of the cast structure had not been attained and

Hydrogen Damage

461

Figure 8.16 Radiograph of a forged shafting showing the presence of internal hydrogen flakes [19].

there had been a relatively rapid cool from forging temperatures. These factors contributed to the formation of shatter cracks or flakes in this steel with normal hydrogen conent. The failure of several aircraft components due to hydrogen embrittlement has been reported [20]. These include landing gear cylinders, main landing gear pivot pins, and main landing gear drag link bolt made of 4340 steel. These components were either chromium- or cadmium-plated and hydrogen pickup during the plating operation was considered to be the cause of embrittlement. The failure also occurred in a propeller retaining ring having the composition Fe-0.5C-2.5Ni0.75Cr. The component had been heat-treated to a strength of 1379 MPa and a bright electroplated cadmium coating had been applied. Failure by HSC has been reported [21] in AISI 4137 steel bolts having a hardness of 42 HRC. Figure 8.17 shows the multiple, branched cracking originating from the thread roots. Although the service temperature was too high (400°C) for hydrogen embrittlement, its occurrence was attributed to corrosion caused by acidic chlorides from a leaking polymer solution at ambient temperatures during the extended shutdown periods. The corrosive environment contained trace hydrogen chloride, acetic acid vapors, and calcium chloride, and the bolt surfaces

462

Chapter 8

Figure 8.17 Longitudinal section of a 4137 steel bolt failed by HSC showing crack emerging from thread grooves [21].

showed extensive corrosion deposits. The failed bolts were replaced with 17-4 PH stainless steel having a hardness of 22 HRC. Collapse of the Point Pleasant, West Virginia (USA) bridge began by failure of a 1060 steel eyebar [18]. Sulfur compounds were found on the fracture surface and are believed to have caused hydrogen sulfide cracking in the eyebar. An example of hydrogen attack is provided by the failure of a steel pipe (0.22% C), 0.31% Si) used as a hot-gas bypass line for hydrogen-rich gas at 34 MPa and 320°C [22]. After 15 months of service, the pipe ruptured, causing a serious fire. The microstructure showed interconnected grain boundary fissures and radially aligned voids as a result of internal methane formation.

8.6 PREVENTIVE METHODS Hydrogen blistering may be prevented by the applications of one or more of the following measures: 1.

Control of inclusions in steel. Since inclusions play a big role in blistering or flaking, inclusion-free ‘‘clean’’ steels are recommended. Elongated inclusions are particularly detrimental, as their presence induces delamination. Use of low-sulfur, calcium-treated, argon-blown steels reduces the incidence of hydrogen blistering, as the poisoning effect of sulfur or sulfides on hydrogen evolution is reduced at low sulfur levels. Also, a treatment with synthetic slag or the addition or rare earth metals can favor the formation of less detri-

Hydrogen Damage

2.

3.

4.

5.

6.

7.

463

mental globular sulfides. On the similar consideration, hot-rolled or annealed steel is preferred to cold-rolled steel. Material selection. Rimmed steels contain numerous voids and are much prone to blistering or flaking. Killed steels show much less susceptibility. Silicon-killed steels are preferable to aluminum-killed steels. The occurrence of corrosion-generated hydrogen blistering can be minimized by the use of metals that are chemically resistant to the environment. Nickel-containing steels, including austenitic stainless steels and nickel-base alloys, have very low hydrogen diffusion rates and are often recommended to prevent hydrogen blistening. Use of coatings. The use of inside coating or liner that is impervious to hydrogen penetration and resistant to the medium can avoid hydrogen blistering of steel tanks or containers. The coatings may be metallic, inorganic, or organic. Cladding of steel with austenitic stainless steel or nickel or a rubber lining on steel is often used. Removal of poisons. Removal of hydrogen evolution poisons such as sulfides, arsenic compounds, cyanides, and phosphorus-containing ions from the environment greatly reduces the incidence of hydrogen blistering. In petroleum process streams these poisons are quite prevalent. The basic approach to reduction of corrosion-induced hydrogen blistering in catalytic cracking units is to reduce the concentration of sulfur and bisulfide ions in water condensate. Use of inhibitors. The corrosion-generated hydrogen blistering can be minimized by use of inhibitors, as the cathodic reduction of hydrogen ions also thereby gets retarded. However, the use of inhibitor can be economical only in closed recirculating systems. Improvements in design. Modifications in design for improved performance of a component need careful consideration. For example, in services in which blistering is expected, external support pads should not be continuously welded to the vessel itself to prevent hydrogen entrapment at the interface. Proper heat treating procedure. The decreased solubility of hydrogen in bcc structured steel compared to the fcc structure leads to flaking and fish-eye formation in steels when they are cooled in hydrogen atmospheres from high temperatures (more than 1100°C). The damage is aggravated if the cooling is rapid because this results in hydrogen-sensitive martensitic microstructure. A reduced cooling rate inhibits the formation of martensite and also allows hydrogen to be slowly released from the steel, thereby eliminating the damage.

Hydrogen embrittlement may be prevented or minimized by the following measures: 1. Material selection. In general, the susceptibility of steels to hydrogen embrittlement increases with the tensile strength of the material. The threshold

464

2.

3.

4.

5.

6.

Chapter 8 tensile strength is 1000 MPa, which can be lower in acidic environments. Wet hydrogen sulfide environments are considered to be among the most aggressive in promoting hydrogen entry. Common metals and alloys are graded according to strength level and/or heat treatment in terms of their resistance to hydrogen-induced cracking. The steels are generally restricted to a maximum hardness of 22 HRC (35 HRC for other alloys). In aerospace applications, steel parts are restricted in strength to below 1380 MPa for tensile loading applications. Heat treatment. For the same stress level, the susceptibility to hydrogen embrittlement of steels is dependent on their microstructure. Untempered martensite is the most susceptible phase. Quenched and tempered microstructures are more resistant than normalized and tempered ones. Accordingly, the heat treatment procedure may be selected. The removal of hydrogen in steels can be carried out by heat treatment at low temperatures (up to 200°C), a process known as baking. In the absence of inversible damages inside the material, a baking treatment can restore the mechanical properties almost to its normal level (Fig. 8.6). Steel parts having tensile strengths greater than 965 MPa that are processed in potentially embrittling fluids must receive in-process and final baking for 23 h at 190°C in order to reduce hydrogen content. Longer bathing times may be required for parts that have large cross-sections, bright cadmium treatments, or very high strength. Hydrogen can be removed from titanium, zirconium, and their alloys by annealing in vacuum. Alloying additions. Alloying steels with strong hydride-forming elements such as titanium, molybdenum, and vanadium reduces susceptibility. However, the concentration of the alloying element is an important factor. For example, molybdenum up to about 0.75% reduces susceptibility of AISI 4130 steel to sulfide stress cracking, Beyond this concentration, a tempering treatment at and above 500°C leads to the precipitation of Mo2C and the resistance to sulfide stress cracking decreases. Proper plating conditions and coatings. Hydrogen pickup during plating can be controlled by the proper choice of plating parameters, i.e., the composition of the bath and the plating current. Cadmium plating and hot-dip galvanizing should be avoided in the case of very-high-strength steels. Hydriding of titanium can be minimized by anodizing or thermal oxidizing treatments to increase the thickness of the protective oxide film. Use of inhibitors. The source of hydrogen pickup by steels is often the pickling operation. Addition of inhibitors to reduce the corrosion of the base metal can largely decrease hydrogen pickup. Proper welding practice. Solutions to the hydrogen embrittlement problems associated with welding include proper cleaning and degreasing procedures for prepared weld joints, use of dry electrode, and the maintenance of dry

Hydrogen Damage

465

conditions during welding. The use of an appropriate preheat before welding and a postweld heat treatment are also recommended. For titanium, zirconium, and their alloys, inert gas shielding during welding is required to minimize hydrogen pickup. 7. Oxygen addition. An addition of 0.4–0.7 vol % oxygen effectively inhibits the embrittlement of steels in gaseous hydrogen environments. However, such additions are not effective to prevent cracking in hydrogen sulfide gas environments. Hydrogen attack can be minimized by the following measures: 1. Material selection. Carbide-forming elements, such as chromium and molybdenum, increase the resistance of steel to hydrogen attack. Steels containing 0.5 Mo or, preferably, 1–2.25 Cr and Mo are recommended for use in hydrogen atmospheres at high temperatures. Since increased carbon content decreases the resistance of steel to hydrogen attack, the carbon content should be low. Stainless steels, particularly austenitic stainless steels, are immune to hydrogen attack. However, atomic hydrogen will diffuse through these steels when used as thin cladding material and the nonresistant substrate steel may be prone to attack. 2. Use of Nelson Curves. Nelson curves (Fig. 8.14) provide the operating limits of various steels in high-temperature, high-pressure hydrogen service. The curves are based on long-term refinery experience rather than on laboratory studies. The curves are revised periodically by the American Petroleum Institute and the latest data should be consulted for the proper selection of steel.

REFERENCES 1. W. H. Johnson, Iron, Vol. 1, pp. 291 and 452–453, 1873. 2. G. G. Hancock and H. H. Johnson, Trans. Met. Soc. AIME, Vol. 236, pp. 513–516, 1966. 3. G. A. Nelson and R. T. Effinger, Welding J., Vol. 34, pp. 125–215, 1955. 4. I. M. Berstein, R. Garber, and G. M. Pressouyre, Effect of hydrogen on the behavior of materials, in Proc. of International Conference, TMS-AIME, New York, 1976, p. 39. 5. R. M. Latanision and H. Oppenhauser, Met. Trans. A., Vol. 5, p. 483, 1974. 6. S. P. Lynch and N. E. Ryan, Hydrogen in metals, Proc. of 2nd International Congress, Paris, June 1977; Pergamon Press, Oxford, 1978, Paper 3D12. 7. G. A. Nelson, Proc. Am. Petrol Inst., Vol. 29MIII, pp. 163–174, 1949. 8. G. M. Pressouyre, Met. Trans. A, Vol. 10A, p. 1571, 1979. 9. C. A. Zapffe and C. E. Sims, Trans. AIME, Vol. 145, pp. 225–261, 1941. 10. J. K. Tien, Effect of Hydrogen on the Behaviour of Materials, A. W. Thomson and I. M. Bernstein, (eds.), Met. Soc. AIME, New York, 1975, p. 309.

466

Chapter 8

11. N. J. Petch and P. Stables, Nature, Vol. 169, pp. 842–843, 1952. 12. A. R. Troiano, Trans. ASM, Vol. 52, p. 54, 1960. 13. R. A. Oriani, Berichte der Bunsen-Gesellschaft fur Physikalische Chemie, Vol. 76, pp. 848–857, 1972. 14. R. Thomson, J. Mat. Sci., Vol. 13, p. 128, 1978. 15. C. J. McMahon Jr., C. L. Briant, and S. K. Banerji, in Fracture, Vol. 1, D. M. C. Taplin (ed.), (Univ. of Waterloo Press, Canada, 1977) p. 363. 16. C. D. Beachem, Met. Trans. A, Vol. 3, p. 437, 1972. 17. Corrosion in petroleum refining and petrochemical operations, Metals Handbook, 9th ed., Vol. 13, American Society for Metals, Metals Park, 1987, p. 1262. 18. Hydrogen-damage failures, Metals Handbook, 8th ed., Vol. 10, American Society for Metals, Metals Park, 1973, p. 230. 19. R. D. Barer and B. F. Peters, Why Metals Fail, Gordon and Breach, New York, 1970, p. 332. 20. Corrosion in the aircraft industry, ASM Metals Handbook, 9th ed., Vol. 13, 1987, p. 1019. 21. Metals Handbook, 9th ed., Vol. 13, American Society for Metals, Metals Park, 1987, p. 332. 22. D. Warren, Hydrogen effects on steel, in Process Industries Corrosion, National Association of Corrosion Engineers, 1986, p. 31. 23. V. J. Colangelo and F. A. Heisen, Analysis of Metallergical Failures, John Wiley, New York, 1974.

9 Radiation Damage

9.1 INTRODUCTION Irradiation of metals and alloys with energetic particles generates point defect clusters, phase transformation, and chemical reaction (transmutation) products that individually or combinedly bring about changes in the mechanical properties of the materials and dimensional changes in the metallic components. Radiation damage manifests itself in the following forms: 1. 2. 3. 4.

Irradiation growth Void swelling Radiation-enhanced creep, and Irradiation strengthening and embrittlement

The study of radiation damage has assumed importance with the development of fast-breeder reactors for the production of nuclear energy. Whereas a thermal neutron diffuses through a metal without displacing lattice atoms, a fast neutron collides with the lattice atoms to produce energetic primary knock-on atoms (PKA), many with tens of thousands of electron volts of energy. Each of the energetic primaries produces a cascade of displacements, resulting in a number of vacancies and interstitials. The damage level is specified by the term ‘‘displacements per atom’’ (dpa), which is the average number of times that a lattice atom has been displaced during irradiation, and the damage rate is given as dpa-s⫺1. Some core components of a commercial fast-breeder reactor are expected to reach a damage level of 2000 dpa during their reactor life. 467

468

Table 9.1

Materials and irradiation conditions for components in various reactor systems [1]

Component

Material

Pressure vessel (PWR) Fuel cladding (PWR)

Carbon or low-alloy steel Zircaloy

Fuel cladding and ducts

Austenitic stainless steel

First wall at 1 MW/m Neutronic wall loading

Austenitic stainless steel Vanadium

Lifetime fluence, ⬎0.1 MeV (n/cm2) Light water reactors 1.6 ⫻ 1019 ⬃1 ⫻ 1022 Fast-breeder reactor ⬃3 ⫻ 1023 Fusion reactora 8.5 ⫻ 1021 8.5 ⫻ 1021

Displacements per atom (dpa)

Transmutationproduced He (appm)

Temp. range (°C) 290 315–370

150

⬃100

300–700

10 12

200 55

350–550 550–750

a Lifetime fluence, dpa, and helium production for fusion systems are per year of full-power operation. PWR, pressurized water reactor.

Chapter 9

Radiation Damage

469

Table 9.1 [1] lists some of the materials, irradiation temperatures, and projected end-of-life neutron fluences for various components in light water reactors (LWRs), liquid metal fast-breeder reactors (LMFRs), and controlled thermonuclear reactors (CTRs). The wide range of temperatures and fluences encountered in these reactors is detrimental to the mechanical properties of the components involved and such damages are directly related to the irradiation-induced point defects.

9.2 RADIATION-INDUCED DEFECT PRODUCTION Defect production in metals is associated with the displacement of atoms from lattice sites. It is difficult to achieve in research studies the irradiation conditions with fast neutrons encountered in reactors, and it takes several years of reactor irradiation to test a material to the target fluence. However, electron and ion irradiations are used to quickly achieve the equivalent of very long exposures to fast-fission reactor neutrons. Computer simulation and experimental field ion microscopy have provided a qualitative picture of the irradiation-induced defect production. The sequence of events is illustrated in Fig. 9.1 [2]. The PKA is dislodged by a neutron collision. It in turn dislodges other atoms. The net result ˚ ), termed a depleted zone or displacement is a localized region (typically ⬃10 A cascade, in which a large fraction of the atoms are missing, thereby resulting in a high density of vacancies. It is the displacement cascade which dissipates the

Figure 9.1 Schematic representation of radiation damage. P denotes the position where the primary knock-on atom comes to rest. (From Ref. 2)

470

Chapter 9

initial kinetic energy of the primary and in which the PKA comes to rest. The region is surrounded by more or less isolated vacancies and interstitials and a few small clusters. The cascade of collisions is completed on a time scale that is short as compared to the times required for more ordinary physical and chemical processes. This permits cascade generation to be treated separately from those processes which control the stability of the point defects first produced. A portion of the defects produced in cascades will be mechanically unstable and will annihilate even at absolute zero temperature. Only the defects that survive this step can be considered to alter the properties of the irradiated material. It has been shown [3,4] that the depleted zone can maintain its three-dimensional nature after low-temperature spontaneous recombination and relaxation events that occur following the cascade production. At higher temperature, due to the spatial arrangement some interstitials, once they become mobile, will be free to migrate from the depleted zone. At still higher temperatures, the vacancies that have escaped recombination become mobile and some can migrate by random walk. However, since a significant number of them are immobilized in the depleted zones, far fewer vacancies than interstitials become free to diffuse through the lattice and precipitate in various diffusion-controlled phenomena. Once sufficient radiation damage has occurred, the immobilized vacancies in the depleted zones will reach a saturation level and from that point on the number of vacancies and interstitials ‘‘freed’’ per damage cascade becomes equal.

9.3 IRRADIATION GROWTH The phenomenon of irradiation growth in materials is characterized by elongations (or contractions) suffered by them in the absence of any externally applied stresses. The elongations or contractions occur only along certain preferred directions in a crystal with little or no volume change. The growth is found to take place only in anisotropic materials, with the anisotropy being due either purely to crystallographic considerations or to an anisotropic dislocation distribution caused in crystallographically isotropic materials by plastic deformation. Irradiation growth was first observed and reported in α-uranium in 1956 [5]. Figure 9.2 shows the photographs of a single crystal of α-uranium before and after irradiation. The crystal was originally nearly a right circular cylinder 0.125 in. in diameter. A great deal of lengthening and shortening has occurred in the [010] and [100] directions, respectively, and the circular cross-section has become elliptical. The growth value showed that the length of a uranium rod would roughly double for about 0.2% atom burn-up. A similar phenomenon has been observed in single crystals of zirconium. Highly oriented polycrystalline specimens of α-uranium have been reported to have shown longitudinal growth rates more than double those measured in single crystals irradiated at the same temper-

Radiation Damage

471

Figure 9.2 Views of a cylindrical single crystal of alpha uranium before and after irradiation (0.1% atom burn up). Ref. 5.

ature. In nonfissile materials like stainless steel, irradiation growth has been observed in cold-worked specimens because of nonisotropic dislocation distribution.

9.3.1

Mechanism

A number of mechanisms have been proposed to account for the irradiation growth in single crystals as well as in polycrystalline metals, keeping in view the role played by anisotropy. Diffusion Mechanism The diffusion mechanism is based on anisotropic diffusion of interstitials and vacancies. It has been proposed [6] that in α-uranium interstitials migrate with some preference for the [010] direction, whereas vacancies would migrate in the [100] and [001] directions. It is assumed that diffusion of interstitials and vacancies essentially balances in the [001] direction, leading to a net shrinkage in the [100] direction. The theory predicts that the rate of growth in the [010] direction will vary as the 3/4 power of the neutron flux and the square root of the diffusion coefficient for interstitials. Since diffusion is temperature-dependent, the growth is predicted to be low at lower temperatures and this gets support from the greatly reduced deformation rate of cold-worked uranium foil irradiated at liquid air temperatures. Partitioning of point defects between interstitial and vacancy loops on different planes due to anisotropy of bonding on closed-packed planes in fissile materials

472

Chapter 9

has also been demonstrated [7]. However, a similar model to explain the growth in nonfissile metals in reactor environments could not be considered because of the incompatibility between the orientations of defect aggregates actually observed and those predicted by the model on the basis of observed growth rated in these metals [8]. Plastic Deformation Accompanying Fission Spikes Plastic deformation is a growth mechanism based on the thermal expansion in the fission spike [9]. The irradiation damage in α-uranium single crystals has been explained [10] as follows: The uniform compressive stress around the site of fission causes local preferential plastic yielding by twinning in the longitudinal direction. When the fission spike cools, the outer region is subjected to a uniform tensile stress and therefore yields plastically, this time by twinning in the [100] and [001] directions. The net result is a local increase in length in the [010] direction in the outer region. The local extension throws a stress on the surrounding matrix, which is relieved by equal amounts of slip on both [110] planes. Since tension in the [010] direction produces no resolved shear stress on the (010) plane, slip in this plane does not occur even though it is the major slip mode. The macrodeformation of the crystal is, therefore, by (110) [110] shears, which agrees with the experiment in that extension occurs in the [010] direction, contraction in the [100] direction, and no change in the [001] direction. Rate Theory Model In rate theory models [11], the growth strains are assumed to result from the climb of edge dislocations by a net flux of interstitials arriving at and annihilating them, and a corresponding vacancy flux arriving at the grain boundaries. Interstitials are biased for edge dislocations because of the Cottrell first-order size effect. Atoms are deposited on crystal planes normal to the Burgers vectors of the edge dislocations. The rate at which atoms deposit on any given crystal plane is proportional to the line density of edge dislocations with Burger’s vectors perpendicular to that plane. Because of an anisotropic distribution of dislocations in the material, the deposition of atoms on different planes takes place at different rates, resulting in a time-dependent deviatonic strain.

9.4 VOID SWELLING Void swelling in fast reactor materials was discovered during the planning and development of liquid metal fast-breeder reactors. As much as 7% void volume was observed in stainless steel fuel claddings that had been irradiated in a test reactor at a neutron fluence that was about one-fourth of the target fluence for a commercial reactor [12]. The large flux of high-energy neutrons in a fast-breeder reactor core results in a much higher rate of defect production (⬃10⫺6 dpa/s⫺1)

Radiation Damage

473

than in a thermal reactor, and this relatively high damage rate is responsible for void formation. The technical implication of this phenomenon is immense. The increase in volume of structural materials in the core of fast-breeder reactors would require a more open and hence less economical core structure, and nonuniform swelling would lead to distortions of the core structure. Some common metals and alloys swell almost 100% at damage levels corresponding to the target fluence, whereas others swell very little. It has now been established that the swelling in practical alloys can be reduced to a safe level for application as core structural members by control of metallurgical variables, particularly the alloy composition.

9.4.1

Mechanism

Irradiation-produced point defects, predominantly vacancies and interstitials, are mobile at elevated temperatures. The defects are annihilated by recombination with opposite types of defects or by reincorporation into the regular crystal structure at various sinks, such as dislocations, grain boundaries, and existing voids. An approximate balance between the production of mobile defects and their annihilation is set up in a relatively short time compared with that required for significant changes in microstructure to develop. If all sinks accept vacancies and interstitials in equal number and rate, the void will not grow. There must exist at least one sink that exhibits preferences for interstitials, thereby altering the relative steady-state concentrations of vacancies and interstitials. Under such conditions, the other sinks that normally exhibit no bias will trap more vacancies. Since dislocations interact more strongly with the strain field of an interstitial than that of a vacancy, there is preferential precipitation of interstitials to dislocations, and of vacancies to voids. The imbalance of the fluxes of the opposite types of defects is small but leads to the void swelling phenomenon. The dislocations that are produced during irradiation, rather than the dislocations originally present, act as preferential sinks for the interstitials. In the early stages of irradiation of a face-centered cubic metal, interstitial platelets are formed. These are essentially faulted dislocation loops that unfault and grow with further irradiation. The swelling can be viewed as the growth of new platelets of material, and the voids merely indicate how much swelling has occurred [13]. The temperature dependence of void swelling supports the above mechanism. The peak swelling occurs in the temperature range of 0.3–0.55 Tm, where Tm is the melting point of the metal in degrees Kelvin (Fig. 9.3). At low temperatures, there are high steady-state concentrations of vacancies and interstitials because these cannot move quickly to sinks. Defect loss by recombination dominates, and as a consequence void swelling ceases. At high temperatures, the limit of swelling is caused by the lack of void nucleation because of low-vacancy supersaturation.

474

Chapter 9

Figure 9.3 Temperature dependence of swelling in ion bombarded and neutron irradiated type 316 stainless steel, Ref. 12.

However, void swelling has been observed under conditions of damage rate, temperature, and sink density at which the vacancy supersaturation should be too low to nucleate voids. Helium gas that is produced by a transmutation reaction is believed to aid void nucleation in such cases.

9.4.2

Control Through Metallurgical Variables

Metallurgical variables such as cold work, grain size, and composition have a distinct influence on void swelling. Cold work brings about a reduction in void swelling, but this effect is lost at high temperatures and at high neutron fluences. Higher amounts of cold work will be necessary to bring about a reduction of the same level at higher damage levels. Cold work increases dislocation density in the material, and it has been

Radiation Damage

475

observed that the swelling is maximal at an intermediate dislocation density. At low dislocation densities there are few preferential sinks for interstitials, whereas high dislocation densities provide so many sinks for vacancies that the vacancy supersaturation remains too low for void nucleation and growth. At higher temperatures, the recovery of cold work is responsible for the loss of swelling inhibition. Since grain boundaries act as sinks for point defects, vacancy concentration remains low at grain interiors if the grain size is sufficiently small and void nucleation is hindered. It has been demonstrated that the swelling 0.45-µm grain size stabilized by the addition of a fine precipitate of Al2O3 was an order of magnitude less than at 3.77 µm grain size [14]. Major and minor alloying additions to the base metal have been found to counteract void swelling and this method has been accepted as an effective means to increase the swelling resistance of practical alloys. In Fe-Ni-Cr alloys, the void swelling is inversely proportional to their nickel content (Fig. 9.4). The swelling increases rapidly with increasing chromium. The lowest swelling base compositions are at low Cr and in the vicinity of 50 wt % Ni. A nimonic alloy, PE-16, has been found to be more resistance to swelling than the austenitic stainless steels. The swelling resistance of this alloys is attributed to very fine coherent precipitates of Ni3(Al, Ti), which prevent dislocation climb and also prevent dislocations from continuing to operate as preferential sinks. Minor additions of Si, Ti, Zn, and Nb have been found to appreciably reduce the peak swelling in stainless steels. Ferritic steels are more resistant to swelling than austenitic steels, but the peak swelling temperature is considerably lower than for austenitic steels with the same chromium content.

9.5 RADIATION-ENHANCED CREEP Creep is the deformation of a metal or alloy under sustained load that is normally noticeable at temperatures above about Tm /2 where Tm is the melting temperature of the metal or alloy in degrees Kelvin. Typical operating conditions in nuclear reactors produce very low thermally induced creep rates, but the effect of the damaging flux of particles is to greatly increase the creep rate of the material. This phenomenon is called radiation-enhanced creep, irradiation creep, or simply radiation creep. Radiation-enhanced creep may be defined as the component of the measured in-pile deformation resulting exclusively from the interaction of various creep mechanisms that become operative when the material is under irradiation.

9.5.1

Characteristics

The radiation-enhanced creep is known to occur at relatively low temperatures (⬍Tm /3) where the thermal creep is generally negligibly small. Like thermal

476

Chapter 9

Figure 9.4 Effect of nickel content on swelling in commercial and simple Fe-15CrX Ni alloys. Ref. 13

creep, radiation-enhanced creep shows a dependence on microstructure, texture, degree of cold work, and grain size. Additionally, it exhibits a weak temperature dependence and complex stress and flux dependencies. For a given material at fixed temperature, flux, fluence, and plastic strain level, the components of creep in metals before, during, and following irradiation are schematically represented in Fig. 9.5. A normal effect of irradiation on metals is radiation hardening, which is exhibited by an increase of yield stress, accompanied by a lower creep rate under the same conditions than before irradiation. The postirradiated curve thus assumes a higher position than the unirradiated curve (see Section 9.6). If various levels of fluence are considered, a family of curves will occur progressively from the unirradiated curve to the final or saturated postirradiated curve. The creep rate is given by: ε ⫽ Aσ n

Radiation Damage

477

Figure 9.5 Schematic representation of components of creep in metals before, during, and following irradiation, Ref. 16.

where A ⫽ constant, σ ⫽ applied stress, and n is the stress component. The complex stress dependence of creep rate can be immediately appreciated from the diagram. At low fluence, the irradiation creep rate is observed to be proportional to the damage rate. As the fluence increases, the creep rate increases, as shown in Fig. 9.6 [15].

9.5.2

Mechanism

A number of mechanisms for radiation-enhanced creep have been suggested. Whereas some derive the creep strain from a stress-induced preferred condensation of point defects, others derive it from either climb alone or climb plus glide of dislocations. The mechanism varies depending on the prevailing stress. At low stresses (below unirradiated yield), the reduction creep has been suggested to be primarily due to the formation of point defect loops, preferentially oriented with respect to the applied stress. Radiation growth is concurrently a contributing factor. At intermediate stresses, a process of flux-enhanced dislocation climb in series with glide has been proposed. At high stresses (above postirradiated yield stress), thermally activated cutting of radiation obstacles as alternative or parallel means of escaping the obstacles has been suggested [16].

478

Chapter 9

Figure 9.6 Dose dependence of stress normalised creep strain of stainless steel [15].

The principal mechanisms of radiation-enhanced creep are as follows: 1. 2. 3. 4.

Stress-induced preferred nucleation (SIPN) Stress-induced preferred absorption (SIPA) Climb-enabled glide mechanisms, and Stress-induced gas-driven mechanism

Stress-Induced Preferred Nucleation Nucleation of planar aggregates of self-interstitial atoms on crystallographic planes orthogonal to the externally applied stress axis would increase the length of the body in the direction of stress, leading to creep strain. A similar effect could be obtained by the condensation of vacancies on planes parallel to the stress axis. In the SIPN mechanism, it is envisaged that the stress induces or enhances such preferred nucleation of radiation-produced point defects [17]. Stress-Induced Preferred Absorption In this mechanism [18], the irradiation creep is derived purely from the dislocation climb that results from the preferred absorption of point defects. Preferential absorption of point defects is mainly attributable to two factors: (1) the Cottrell

Radiation Damage

479

first-order size effect, i.e., the self-interstitials atoms have a larger relaxation strain and hence a stronger size effect interaction with the edge dislocations, and (2) the Eshelby elastic strain effect, i.e., in the presence of an internal stress interstitials have a larger inhomogeneity interaction with those edge dislocations whose Burger vectors are aligned in the direction of the stress axis. Such interaction causes more interstitials to be attracted to, and annihilated at, dislocations aligned with their Burger vectors in the direction of the applied tensile stress resulting in their climb. For vacancies, the bias is just the opposite; therefore, the corresponding excess vacancies are annihilated at dislocations with their Burger vectors perpendicular to the applied tensile stress leading to their climb as well. Climb-Induced Glide Mechanisms In a crystallographically isotropic material with isotropic dislocation distribution, the climb of dislocations by itself causes no creep strain, but this enables dislocations to break loose of obstacles. Stress then causes slip of the glissile dislocations, resulting in creep strain. Under such circumstances, the creep rate is controlled by climb but the main source of the strain is the dislocation glide. A number of models based on the concept of climb-enabled glide have been proposed, i.e., I-creep model [19], PAG model [20], climb-controlled glide (CCG) model [21], and climb-induced yield (CIY) model [22]. Stress-Induced Gas-Driven Mechanism Coupled with the growth of gas bubbles on the grain boundaries orthogonal to the stress axis, extension of grains may result from suitably oriented dislocation loops. This has been termed a ‘‘jacking mechanism’’ [23], and is important whenever conditions suitable for gas bubble growth prevail.

9.6 IRRADIATION STRENGTHENING AND EMBRITTLEMENT Particle irradiation brings about large changes in some of the mechanical properties of metals. The yield stress increases, leading to the phenomenon of irradiation strengthening or hardening. The flow stress also shows an increase at least in the initial stages of plastic deformation. The fracture behavior changes, exhibiting a transition from ductile to brittle features, and the ductile-to-brittle transition temperature shows an increase on irradiation. The stress–strain curves of irradiated and unirradiated single crystals of copper are shown in Fig. 9.7. The increases in yield stress and flow stress at a lower deformation level are clearly visible, although after appreciable plastic deformation the difference in the curves essentially disappears. However, in some metals the whole stress–strain curve is appreciably altered by irradiation, as is shown for nickel in Fig. 9.8. In copper, iron, and zinc single crystals the critical shear stress has been observed to increase progressively with exposure to irradiation.

480

Chapter 9

Figure 9.7 Stress-strain curves of copper single crystals, Ref. 10.

Figure 9.8 The effect of irradiation on the stress strain curve of nickel, Ref. 10.

Radiation Damage

481

Figure 9.9 Effect of neutron irradiation on the critical shear stress of copper, Ref. 10.

Figure 9.9 shows the effect of neutron irradiation on the critical shear stress of copper single crystals. An increase in the value of the critical shear stress from 0.241 kg/mm2 to 2 kg/mm2 has been observed to take place with an exposure of about 2 ⫻ 1018 nvt. The rate of increase has fallen thereafter. The effect of irradiation in copper is somewhat similar to solid solution hardening. However, it has been estimated [24] that one interstitial atom is 40 times more effective than a zinc atom in raising the critical shear stress. The tensile fractographic features in type 304 tested at 370°C are shown in Fig. 9.10 [25]. A transition from ductile to brittle fracture is evident. In body-centered cubic metals, a ductile-to-brittle transition temperature is encountered and irradiation results in an increase in transition temperature.

Figure 9.10 Tensile fractographs of type 304 stainless steel tested at 370°C. (a) Unirradiated, (b) after about 2.8 ⫻ 1022 n/cm2, and (c) after about 10.7 ⫻ 1022 n/cm2. Ref. 25.

482

Chapter 9

Figure 9.11 Transition curves for unirradiated and irradiated impact specimens of mild steel, Ref. 23.

Figure 9.11 shows the results obtained in impact tests before and after irradiation with 18.6 MeV deuterons for an average total exposure of 6.7 ⫻ 1017 particles/ cm2 [26]. The transition temperature has changed from ⫺1°C to about 8°C. Similarly, in molybdenum, an increase from ⫺30°C to 70°C has been reported [27].

9.6.1

Mechanism

Irradiation strengthening of metals is a result of the barriers to dislocation motion that are produced from precipitation of radiation-induced defects, i.e., vacancies and interstitials. The barriers formed are voids and dislocation loops. In some alloy systems, such as nickel alloys and stainless steels, irradiation-induced precipitation of second phases such as carbides may also contribute to the observed strength increase. While the pinning of dislocations directly contributes to the increase in yield stress, the critical shear stress also increases because it depends on the breakaway of pinned dislocations. The controlling factor in the effects of irradiation on deformation behavior is the irradiation-produced microstructure, which is temperature-dependent. It has been demonstrated that [28] 1.

At low irradiation temperatures (below 0.35 Tm), irradiation-produced vacancies and interstitials collect into small dislocation loops.

Radiation Damage

483

Figure 9.12 Microstructures in type 304 stainless steel irradiated and tested near (a) 450°C and (b) 650°C.

484 2. 3.

Chapter 9 In the approximate temperature range of 0.35–0.6 Tm, voids and a dislocation structure are produced. Above about 0.6 Tm, displacement damage continuously anneals, and helium may precipitate to form equilibrium bubbles.

The increase in yield stress has been well accounted for in terms of strengthening from the three main components of the irradiation-produced microstructure, i.e., dislocation loops, network dislocation, and voids. In type 304 stainless steel, samples irradiated at about 450°C have been shown [1] to contain a high concentration of faulted interstitial loops and voids, with the deformation confined to narrow channels within the matrix, whereas at 600°C the structure consists of a dislocation network and voids, both present in relatively low concentrations, and the deformation is nearly homogeneous, as observed in unirradiated stainless steels. The dislocation structures as encountered are shown in Fig. 9.12. Irradiation embrittlement arises as a result of one or a combination of any of the following phenomena [1]: 1. 2. 3.

Changes in flow properties due the interaction of dislocations with irradiation-produced defects Precipitation of transmutation-produced gases such as helium at potential fracture sites such as grain boundaries, and possibly Irradiation-induced segregation due to vacancy fluxes to sinks such as grain boundaries, which are also potential fracture sites

REFERENCES 1. E. E. Bloom, Irradiation strengthening and embrittlement, in Radiation Damage in Metals, N. L. Peteson and S. D. Harkness (eds.), American Society for Metals, 1976, p. 295. 2. A. Seeger, Proc. Symp. Rediat. Damage Solids React. Mat., Vlenna, IAFA, 1962, p. 101. 3. J. R. Beeler, Phys. Rev., Vol. 150, p. 470, 1966. 4. L. A. Beavan, R. M. Scanlan, and D. N. Seidman, Acta Met., Vol. 19, p. 1339, 1971. 5. S. H. Paine and J. H. Kittel, Proceedings of the International Conference on the Peaceful Uses of Atomic Energy, Vol. 7, United Nations, 1956, p. 445. 6. L. L. Seigle and A. J. Opinsky, U. S. Atomic Energy Comm., Report SEP-160 (1954) quoted in Radiation Effects in Solids, G. J. Dienes and G. H. Vineyard (eds.), Interscience, New York, 1957, p. 187. 7. R. J. McElroy, J. Nucl. Mat., Vol. 90, p. 297, 1980. 8. D. O. Northwood, Atom. Energy Rev., Vol. 15, p. 547, 1977. 9. S. F. Pugh, Ref. 5, p. 441. 10. G. J. Dienes and G. H. Vineyard, Radiation Effects in Solids, Interscience, New York, 1957, p. 183. 11. S. R. MacEwen and G. J. C. Carpenter, J. Nucl. Mat., Vol. 90, p. 108, 1980.

Radiation Damage

485

12. C. Cawthorne and E. J. Fulton, Nature, Vol. 216, p. 575, 1967. 13. W. G. Johnston and T. Lauritzen, The effect of metallurgical variables on void swelling, in Ref. 1, p. 227. 14. B. N. Singh, Phil. Mag., Vol. 29, p. 25, 1974. 15. K. Ehrlich, J. Nucl. Mat., Vol. 100, p. 149, 1981. 16. F. A. Nichols, Radiation-enhanced creep, in Ref. 1, p. 267. 17. R. V. Hesketh, Phil Mag., Vol. 7, p. 417, 1962. 18. R. Bullough and J. R. Willis, Phil. Mag., Vol. 31, p. 875, 1975. 19. J. H. Gittus, Phil. Mag., Vol. 25, p. 345, 1972. 20. L. K. Mansur, Phil. Mag. A, Vol. 39, p. 497, 1979. 21. W. G. Wolfer and A. Boltax, European Conf. on Irradiation Embrittlement and Creep on Fuel Cladding and Core Components, London (1972), paper 31, quoted by I. S. Batra and P. Das Gupta, Trans. Indian Inst. of Metals, Vol. 41, p. 433, 1983. 22. W. G. Wolfer, J. Nucl. Mat., Vol. 90, p. 175, 1980. 23. J. E. Harris, Proc. of Conf. Vacancy 76, Univ. of Bristol, 1976, Metals Society, London, p. 170. 24. Quoted in Ref. 10, p. 182. 25. R. L. Fish, J. L. Straalsund, C. W. Hunter, and J. J. Holnes, Effects of Reduction on Substructure and Mechanical Properties of Metals and Alloys, ASTM-STP 529, 1973, p. 149. 26. R. A. Meyer, J. Appl. Phys., Vol. 25, p. 1369, 1954. 27. C. A. Bruch, W. E. McHugh, and R. W. Hockenbury, Trans. Am. Inst. Mining Met. Petrol. Engrs., Vol. 203, p. 281, 1955. 28. H. R. Brayer and J. L. Straalsund, J. Nucl. Mat., Vol. 26, p. 134, 1973.

SUGGESTED READING G. J. Dienes and G. N. Vineyard, Radiation Effects in Solids, Monographs in Physics and Astronomy, Volume 2, Interscience, New York, 1957. J. Koutsky, Radiation Damage of Structural Materials, Materials Science Monographs, Volume 79, Elsevier, New York, 1994. N. L. Peterson and S. D. Harkness (eds.), Radiation Damage in Metals, Americal Society for Metals, Metals Park, Ohio, 1976.

This Page Intentionally Left Blank

Index

Activation energy, values of, 261– 276 for hafnium oxidation, 270 for interstitial solute diffusion, 261 for nickel oxidation, 272, 276 for titanium oxidation, 265 for zirconium oxidation, 267 Active-passive metals, 32, 34 Adsorption, 180, 189, 228, 271 Aliovalent impurities, 288 Alloy oxidation, 283–412 Alloy oxidation, stages in, 304– 309, 312–330 breakaway stage, 304–306 steady state stage, 312–330 transient stage, 306–307 transient to steady state, 307– 309 Aluminide coatings, 389–391 high activity process, 390–391 low activity process, 390

Aluminum and aluminum alloys, 143 Amorphous phase, 252 Aqueous corrosion, 5–177 classification, 47 electrochemical cell analogy, 8 electrochemical nature of, 5 forms of, 47–130 fundamentals of, 5–45 homogeneous theory of, 12 kinetics of, 23–45 local cell theory of, 10 prediction from emf series, 18– 20 prevention, 131–177 thermodynamics of, 13–23 Anode, definition, 6 Anodic inhibitors (see Passivators) Anodic protection, 170–172 Anodic reaction, 6 Anodizing, 159 Atmospheric corrosion, 48 487

488 Bertholides, 197 Biologically-influenced corrosion, 124–129 description of, 124 practical examples, 128 (See also Macrobiofouling; Microbiological corrosion) Boltzmann constant, 229–231 Bond coating, 398–399, 408 Brass, 78–83, 92 (see also Copper and copper alloys) Breakaway oxidation (in zirconium), 266–269 Bronze (see Copper and copper alloys) Buckling (in spallation), 255– 259 Butler–Volmer equation, 30 Cabrera–Mott’s equations (see Tarnishing processes) Carbon steels, 138–139 Cast irons, 136–138 Catastrophic oxidation, 301–304 description of, 301–302 mechanisms, 302–304 Cathode, definition, 6 Cathode precipitates (see Cathodic inhibitors) Cathodic corrosion, 166, 168 Cathodic inhibitors, 153 Cathodic protection, 162–170 (see also Sacrificial anode) criteria for, 168 methods, 166 principles of, 164–166 stray current effects, 169 Cathodic reaction, 6 Caustic embrittlement, 92 Cavitation damage, 87–88 Cavitation erosion (see Cavitation damage)

Index Cement coatings, 159 Chemical conversion coatings, 160–161 Chemical vapor deposition, 396– 397 Chromate coatings, 160 Chrome plate, 157 Cladding, 157, 283, 288, 388, 393 Coatings, high temperature, 375– 409 coating methods, 387–397 pack aluminizing, 389–391 pack chromizing, 388–389 overlay coatings, 391–397 coating–substrate requirements, 382–383 degradation, 399–409 by diffusional interaction processes, 399–403 by reaction with environment, 404–407 of thermal barrier coatings, 407–409 diffusion in oxides, 384–387 history of, 375–380 principles of protection, 380– 382 stability and compatibility of oxides, 384 thermal barrier coatings, 397– 399 Coatings, inorganic, 159–161 cement coating, 159 chemical conversion coatings, 159–161 glass coating, 159 Coatings, metallic 156–158 (see also Coatings, high temperature) cladding, 157 diffusion coating, 157 electroplating, 157

Index [Coatings, metallic] flame spraying, 157 hot dipping, 156 surface modification, 158 vapor deposition, 157 Coatings, organic, 161–163 lacquers, 161 paints, 161–163 performance in various environments, table of, 163 varnishes, 162 Cobalt alloys, 320–330, 352, 370– 373 Coch–Cohen clusters, 202 Coextrusion, 392–393 Concentration cells, 10 Copper and copper alloys, 143– 145, 219–221, 229–231, 310–311, 317–319 Corrosion, definition, 2 allowance, 173 atmospheric, 48 biologically influenced, 124– 129 by liquid metals, 431–434 cracking, 91–124 general, 48–49 localized, 49–124 prevention, 132–147 rate expressions, 41–42 rate measurement, 41–45 Corrosion current density, 29 Corrosion fatigue, 91, 118–124 effect of environmental factors, 121 mechanism, 121–123 practical examples, 123–124 remedial measures, 123 stress factors in, 119 Corrosion potential, 29 Corrosion prevention, 132–147 Corrosion rate expressions, 41–42

489 Corrosion rate measurement, 41– 45 electrochemical methods, 43 (see also Tafel extrapolation method; Linear polarization method) exposure tests, 42 Corrosion resistance of materials (see specific materials; Material selection for corrosion prevention) Corrosive wear, 89 Creep strain, 243 Crevice corrosion, 63–68 description of, 63 design improvements against, 174 mechanism, 64–65 practical examples, 68 remedial measures, 66 Critical pitting potential, 59 Critical solute concentration (in internal oxidation), 300 Daltonides, 196 Dangerous inhibitors (see Anodic inhibitors) Deactivation, 149 Deaeration, 149 Dealloying of copper alloys, 81 Dealuminification, 82 Decarburization (see Hydrogen attack) De-electronation reaction (see Anodic reaction) Defect chemistry of oxides and inorganic compounds, 194– 210 classification of defects, 194– 196 defect formation reactions, 203– 210

490 [Defect chemistry of oxides and inorganic compounds] electrical conductivity dependence of, 209–210 defect structures of non-stoichiometric compounds, 200– 202 metal-deficient oxides, 200– 202 oxygen-deficient oxides, 200 defect structures of stoichiometric compounds, 198–200 deviation from stoichiometry, 196–197 electroneutrality conditions, 196, 205 Frenkel disorder, 199–200 Kro¨ger–Vink diagrams, 207 notation, 195 neutral and charged defects, 196 Delayed failure, 414, 426–427 Depleted zone, 469 Deposit corrosion (see Crevice corrosion) Design improvements for corrosion prevention, 172–177 Dezincification, 78–83 description of, 78 mechanism, 79–81 practical examples, 82–83 remedial measures, 81 Differential temperature cells, 11 Diffusion annealing, 387 Diffusion coatings, 157, 388–391 Diffusion coefficients in oxides and spinels, 384–386 Diffusional transport, 180 Displacement cascade, 469 Dissimilar electrode cells, 10 Dissolution and diffusion of oxidants, 259–271

Index [Dissolution and diffusion of oxidants] diffusion of interstitial atoms, 260–261 effect on oxidation behavior, 260, 262–271 grain boundary diffusion, 261 solubility of gases in metals, 260 temperature dependence of diffusion coefficient, 260 thermal diffusion, 261–262 Doping effect in alloy oxidation, 288–295, 339–340 in ionic conductors, 292–295 in n-type semiconductors, 289– 290 in p-type semiconductors, 290– 292 Dry corrosion, 194 (see also Metal oxidation) Durichlor (see Cast irons) Duriron (see Cast irons) Dynamic segregation theory, 341 Electrode kinetics (see Kinetics of aqueous corrosion) Electronation reaction (see Reduction reaction) Electronic charge, 229–230 Electron tunneling, 229 Electroplating, 157 Ellingham–Richardson diagrams, 181–185 of metal carbides, 185 of metal halides, 184 of metal oxides, 182 of metal sulfides, 183 Environmental degradation of metals, classification, 2

Index Epitaxial strain, 252 Epitaxy, 243 EPR technique, 74 Erosion corrosion, 83–91 description of, 83–85 design improvement against, 175 effect of environmental factors, 85–86 effect of metallurgical factors, 86–87 practical examples, 90 remedial measures, 87 Exfoliation corrosion, 76 Faraday number, 215 Faraday’s laws, 23 Fermi electrons, 230 Fick’s law, 297 Filiform corrosion, 66–67 Fish eyes, 454 Flakes, 454 Flame spraying, 157, 393 Free energy of formation for oxides, 286 Frenkel disorder, 199–200, 203– 206, 294 Fretting corrosion, 89–90 Galvanic corrosion, 49–56 area effect in, 50 (see also Sacrificial anode) description of, 49 practical examples, 55 prediction of, 49–52 remedial measures, 55 Galvanic series, 52 in seawater, 53 Galvanizing, 54, 162 (see also Hot dipping) Gasket corrosion (see Crevice corrosion)

491 General corrosion, 48–49 description of, 48 remedial measures, 48 Gettering effect (in alloy oxidation), 310–312 Gibbs–Duhem equation, 215 Glass coatings, 159 Graphitic corrosion, 78, 81–82 Hall coefficient, 211 Hastelloy (see Nickel and nickel alloys) Henry’s law, 203 High temperature alloys, development and selection, 283– 288 High temperature corrosion, definition, 3 High temperature oxidation behavior, 262–279 of chromium, 276–277 of hafnium, 270 of iron, 277–279 of nickel, 272–276 of niobium, 271 of tantalum, 271 of titanium, 262–266 of zirconium, 266–270 Hot corrosion, 3, 283, 288, 304, 347–375 classification, 348–349, 360– 362 of cobalt alloys, 370–373 kinetics of, 357–362 mechanisms of, 363–375 of pure nickel, 365–367 salt component induced, 361 scale morphology in, 362 thermodynamics of, 349–357 vanadate induced, 348 Hot corrosion, mechanisms, 363– 375

492 [Hot corrosion, mechanisms] acid fluxing, 368–373 basic fluxing, 365–368 electrochemical model, 363– 364 salt-fluxing reactions, table of, 374 Hot dipping, 156 Hot spot, 177 Huey test, 74 Hydride formation, embrittlement due to, 451 Hydrogen-assisted cracking (see Hydrogen embrittlement) Hydrogen-assisted stress corrosion cracking, 108 Hydrogen attack, 455–456, 465 Hydrogen blistering, 452–454, 462–463 Hydrogen damage, 3, 437–466 classification, 440 description of, 440–456 practical examples, 460–462 preventive methods, 462–463 sources of hydrogen for, 438 theories of, 456–460 Hydrogen embrittlement, 437, 440–452, 463–465 due to hydride formation, 451– 452 hydrogen environment embrittlement, 451 hydrogen stress cracking, 441– 451 loss in tensile ductility, 440– 441 mechanisms, 457–460 Hydrogen embrittlement, mechanisms, 457–460 decohesion theory, 459 enhanced plastic flow theory, 460

Index [Hydrogen embrittlement] hydrogen pressure theory, 457– 458 surface adsorption theory, 458– 459 Hydrogen environment embrittlement (see Hydrogen embrittlement) Hydrogen-evolution poisons, 153 Hydrogen-induced cracking, 91 (see also Hydrogen damage) Hydrogen stress cracking (see Hydrogen embrittlement) Hydrogen trapping, 457 Inhibitors, 150–155 classification, 151 definition of, 150 for steel components, table of, 151 Intergranular corrosion, 68–77 of austenitic stainless steels, 69–73 description of, 68 of ferritic stainless steels, 73–74 mechanism, 70–71 of other alloys, 75–77 practical examples, 77 remedial measures, 73 tests for, 74–75 Intergranular segregation, 346 Internal oxidation, 287, 295–301, 307, 309, 312 concentration profile, 297 of copper alloys, 296–297 criteria for occurrence, 296 kinetics of, 297–299 of silver-indium allys, 300–301 transition to external scaling, 299–301 zone (IOZ), 297

Index Internal reversible hydrogen embrittlement, 444 Intrinsic defect concentration, 294 Ion implantation, 158, 335, 393 Iron-chromium alloys, 320, 322– 329 (see also Stainless steels) Irradiation embrittlement (see Irradiation strengthening) Irradiation growth, 470–473 description of, 479 mechanisms, 471–473 Irradiation strengthening, 479–484 description of, 479 mechanisms, 482–484 Isocorrosion charts, 132–136 code for sulfuric acid, 134–135 of six alloys in sulfuric acid, 133 of steel in sulfuric acid, 133 Kinetics of aqueous corrosion, 23– 45 activation polarization, 26 Butler–Volmer equation, 30 combined polarization, 28 concentration polarization, 27 electrode kinetic parameters, 30 exchange current density, 24 Faraday’s laws, 23 importance in corrosion appraisal, 30–32 limiting-diffusion current density, 27 mixed potential theory, 28–30 passivity, 32–41 polarization, definition, 25 overvoltage, definition, 25 Tafel equation, 26 Kinetics of metal oxidation, 187– 194, 213–216, 227–232, 271–279

493 [Kinetics of metal oxidation] effect of surface preparation and pre-treatment, 271–279 in chromium oxidation, 276– 277 in iron oxidation, 277–279 methods of surface preparation, 271–272 in nickel oxidation, 272–276 procedure of oxidation test, 272 grain boundary diffusion, 191 kinetic laws, 213–216, 227–232 Cabrera–Mott parabolic law, 213–216 cubic law, 231–232 inverse logarithmic law, 228– 229 logarithmic law, 229–230 Wagner’s parabolic law, 213– 216 lattice diffusion, 191 partial processes, 187–191 rate equations for thin film growth, 193–194 stresses affecting, 191–192 Knife-line attack, 71 Kro¨ger-Vink diagrams, 207 notation, 195

Lacquers (see Coatings, organic) Langlier index, 148 Laser processing, 158, 396–397 Limiting-diffusion current density, 27 Limiting thickness in thin film growth, 229 Linear polarization method, 44 Linear thermal expansion coefficient ratio, 192–193, 331 Liquid metal attack, 413–435

494 Liquid metal embrittlement, 3, 413–431 definition, 3, 413 effect on stress strain curve, 414 embrittlement couples, table of, 417 factors influencing, 422–426 fracture morphology, 414–415 inert carriers for, 418 mechanisms, 427–429 practical examples, 430–431 preventive measures, 429 requirements for, 421 severity of environment, 419 specificity of environment, 415 Liquid metal induced embrittlement (see Liquid metal embrittlement) Low alloy steels, 138 Low and high temperature regimes, 180 Macrobiofouling, 127–128 description and mechanism, 127 practical examples, 128 remedial measures, 128 Marker study, 211, 266–267 Material selection for corrosion prevention, 49, 55, 62, 66, 73, 81, 87, 116, 132–147 Metallic coatings, 156–159, 387– 397 Metallizing (see Flame spraying) Metal oxidation, definition, 179 Microbiological corrosion, 125– 127 bacteria causing, table of, 126 description and mechanism, 125 practical examples, 128 remedial measures, 128 Monel (see Nickel and nickel alloys)

Index Negative charge carrier, 200 Nelson curves, 465 Nernst equation, 20, 37 Nickel and nickel alloys, 145–146, 272–276, 319–330, 350– 351, 358–360, 365–367 Nickelate, 356, 367 Ni-hard (see Cast irons) Ni-resist (see Cast irons) n-type semiconductors, 200, 206– 208, 210, 289–290 Octahedral site, 260 Oswald ripening, 296 Organic inhibitors, 153–155 Overvoltage, definition, 25 Oxide keying/pegging, 343, 346 Oxidation of metals and alloys, 3, 179–181, 283–412 (see also Alloy oxidation; Tarnishing and scaling processes) Oxidation reaction, 6, 180 Oxide coatings, 159, 397–399 Oxidizers (see Passivators) Oxyanion salts, 354 Oxygen scavengers, 149, 153 Paints, 161–163 function of pigments in, 161 performance of, table of, 163 Passivators, 33, 151–153 Passivity, 32–41 active-passive metals, 32, 34 critical current density for, 34 effect of environmental factors on, 36 effect of galvanic coupling on, 38–40 passivators, 33, 151 primary passive potential, 34 theories of, 40 transpassivity, 34

Index Permittivity, 232 Phase boundary reactions, 180, 187, 192 Phosphatizing, 159 Physical vapor deposition, 395– 396 Pickling inhibitors, 154 Pilling–Bedworth ratio, 192, 240– 241 limitations of, 192 Pitting, 56–58 characteristic features, 56 critical potential for, 59 description of, 56 evaluation of damage, 57 mechanism, 59–62 practical examples, 63 protection potential for, 59 remedial measures, 62–63 Pitting factor, 58 Plasma spraying, 394–395 (see also Flame spraying) Polarization, definition, 25 (see also Linear polarization) curve for active-passive metals, 35 diagram illustrating anodic protection, 171 diagram illustrating cathodic protection, 165 Polarization resistance method (see Linear polarization method) Positive charge carriers, 201 Potential-pH (Pourbaix) diagrams, 20–23 Precipitate in internal oxidation, 296 Protection potential for pitting, 59 p-type semiconductors, 200–202, 208–210, 272, 290–292

495 Radiation damage, 3, 467–485 classification, 467 materials and irradiation conditions, table of, 468 Radiation-enhanced creep, 477– 479 characteristics, 477 mechanisms, 477–479 Radiation-induced defect production, 469–470 Rate equations for thin film growth, 193–194, 228–232 cubic, 193, 231–232 inverse logarithmic, 193, 228– 229 normal logarithmic, 193, 229– 230 parabolic, 193, 230–231 Reactive element effects, 330–346 beneficial effects, 334–337 choice of reactive elements, 332–334 incorporation methods, 334–335 mechanisms of scale growth, 337–342 mechanisms for improved scale adherence, 342–346 Reduction reaction, 6 Reversible and nonreversible reactions, 180 Roth clusters, 201 Rusting, 8 Sacrificial anode, 54, 166–167 Scaling process, 179, 232–259 definition, 179 mechanisms, 210–225, 232–239 stresses and strains development in, 239–259 factors influencing, 240–253 failure of scales due to, 253– 259

496 Scaling process, mechanisms, 210–225, 232–239 consequences of void formation, 236–239 dependence of parabolic rate on defect concentration, 223– 224 driving forces for diffusion, 211–213 electrical conductivity of oxides, table of, 224 electrochemical steps involved, 216–217 growth by lattice and grain boundary diffusion, 232–234 rationale rate constant, 223–225 verification of Wagner’s parabolic law, 216–225 Wagner’s parabolic law, 213– 216 Schottky disorder, 198–199, 203 Season cracking of brass, 92 Selective leaching, 77–83 alloy environment combinations for, 79 of brass (see Dezincification) of cast iron (see Graphitic corrosion) description of, 77–79 of other copper alloys, 81 practical examples, 82 Self-diffusion coefficient, determination of, 225–227 Feuki and Wagner’s approach, 225–226 Mrowec’s approach, 226–227 Sensitization, 70 tests for, 74–75 Shatter cracks, 154 Silicide paste, 403 Silver-indium alloys, 300–301

Index Slushing compounds, 155 Solid metal-induced embrittlement, 419 Spallation, 253–259 buckling in, 255–259 critical temperature drop to initiate, 256–257 map for, 256, 258 mode of, 253–256 wedging in, 255–259 Stainless steels, 33, 139–143 age-hardening, 142 austenitic, 142 chemical composition, table of, 140–141 duplex, 142 effect of nickel on SCC of, 100 ferritic, 139 intergranular corrosion of, 69– 74 knife-line attack of, 71 martensitic, 139 sensitization of, 70 weld decay of, 70–71, 73 Standard hydrogen electrode, 16– 18 Static fatigue (see Delayed failure) Steady state scaling of binary alloys, 312–326 classification of alloys, 312– 317 comparison of Fe-Cr, Ni-Cr, and Co-Cr systems, 322– 326 formation of mutually insoluble oxides, 318–319 formation of oxide solid solutions, 319–320 morphology of scales, 316 parameters for steady state, 314

Index [Steady state scaling of binary alloys] partial miscibility and compound formation, 320–322 simultaneous formation of oxides in the external scale, 317–322 Steady state scaling of ternary alloys, 326–330 comparison of Fe-Cr-Al, Ni-CrAl, and Co-Cr-Al alloys, 328–330 description of oxidation behavior, 326–328 Strauss test, 74 Stray current corrosion, 169 Streicher test, 74 Stress corrosion cracking, 91–118 alloy environment combinations, 94 description of, 91 electrochemical aspects, 102– 106 general features of cracks, 93– 96 mechanisms, 106–114 metallurgical aspects, 99–102 practical examples, 117–118 remedial measures, 114–117 sources of stress for, 93 testing methods, 96–99 Stress corrosion cracking, mechanisms, 106–114 adsorption-induced cleavage, 111 atomic surface mobility, 111 corrosion tunnel model, 108 film-induced cleavage, 111 hydrogen-assisted cracking, 108 pre-existing active path, 107 strain-generated active path, 107 tarnish rupture, 111

497 Stress corrosion cracking, testing methods, 96–99 constant strain tests, 96 slow strain rate test, 99 static tensile tests, 97–98 Sulfur effect, 344 Surface modification for corrosion prevention, 158–159 (see also Ion implantation; Laser processing) Superalloys (Ni- and Co-based), 283, 288, 348–352, 359, 362 Tafel equation, 26 Tafel extrapolation method, 43 Tarnishing and scaling process, 179 thin film growth, 229–232 very thin film growth, 228–229 Tetrahedral site, 260 Thermal barrier coatings, 397– 399, 407–409 Thermodynamics of aqueous corrosion, 13–23 cell potential, 15 electrode potential, 15 emf series, 18 free energy change, 13 Nernst equation, 20, 37 potential-pH (Pourbaix) diagrams, 20–23 redox potential, 18 single-electrode potential (see Electrode potential) standard hydrogen electrode, 16 Thermodynamics of metal-single oxidant systems, 181–187 Ellingham–Richardson diagrams, 181–185

498 [Thermodynamics of metal-single oxidant systems] equilibrium oxygen pressure at interfaces, 186 Gibb’s free energy change, 181 single or multilayered oxide, 186 standard free energy change, 181 Thermoelectric power, 211 Thermoionic emission, 229–232 Tin cans, 54 Titanium and titanium alloys, 146–147, 262–266 Transpassivity, 34 Udimet 700, 287, 391–392

Index Vapor deposition, 157 (see also Chemical vapor deposition; Physical vapor deposition) Vapor phase inhibitors, 155 Varnishes (see Coatings, organic) Void swelling, 473–477 description of, 473 mechanisms, 473–474 metallurgical variables, 474–476 Volatile oxides, 286–287, 403–405 Wagner’s parabolic law, 213–216 Water treatment, 147–150 Wedging (in spallation), 255–259 Weld decay, 70–71, 73 Wet corrosion, 194 (see also Aqueous corrosion) Work function, 230