Corrosion of Metals in Category B&C Waste

Home , Galvanic corrosion

SCK•CEN/27548174 ER-0412

Corrosion of metals in category B&C waste

A review of literature data

Frank Druyts and Sébastien Caes

Publication date: May 2019

Contract name: ONDRAF/NIRAS, Contrat de R&D "gestion à long terme des déchets radioactifs" (2015-2020)

Contract number: SCK•CEN: CO-90-14-3690-00; ONDRAF: CCHO 2015-0304/00/00; Specification sheet 15-SCK-EBC-12

SCK•CEN Boeretang 200 BE-2400 Mol Belgium

Phone +32 14 33 21 11 Fax +32 14 31 50 21

http://www.sckcen.be

Contact: Knowledge Centre [email protected]

COPYRIGHT RULES All property rights and copyright are reserved to SCK•CEN. In case of a contractual arrangement with SCK•CEN, the use of this information by a Third Party, or for any purpose other than for which it is intended on the basis of the contract, is not authorized. With respect to any unauthorized use, SCK•CEN makes no representation or warranty, expressed or implied, and assumes no liability as to the completeness, accuracy or usefulness of the information contained in this document, or that its use may not infringe privately owned rights.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 2 of 77 SCK•CEN/27548174 ISC: Public

Table of Content

Glossary of Abbreviations...... 6 Abstract ...... 10 Keywords ...... 10 1 Introduction ...... 14 1.1 Metals present in category B and C waste ...... 14 1.2 Engineered barrier system for category B and C waste ...... 14 1.2.1 Monolith for category B waste...... 14 1.3 Geochemical evolution of the engineered barrier system ...... 15 2 Qualification of techniques for the determination of corrosion rates ...... 16 3 Corrosion of zirconium alloys ...... 20 3.1 Introduction ...... 20 3.2 Corrosion mechanisms of zirconium alloys ...... 20 3.2.1 Corrosion of zirconium alloys in reactor conditions ...... 22 3.2.2 Uniform corrosion ...... 24 3.2.3 Microstructure of the oxide (under uniform corrosion) ...... 24 3.2.4 Effect of ion irradiation on the microstructure of the oxide ...... 25 3.2.5 Hydrogen absorption ...... 25 3.2.6 Corrosion of zirconium alloys in anoxic highly alkaline conditions ...... 26 3.3 Corrosion kinetics of zirconium alloys in geological disposal conditions ...... 28 3.3.1 Influence of pH ...... 29 3.4 Conclusions ...... 30 4 Corrosion of metallic uranium ...... 31 4.1 Introduction ...... 31 4.2 Metallic uranium corrosion ...... 31 4.2.1 Corrosion of metallic uranium in air or oxygen ...... 31 4.2.2 Corrosion of metallic uranium in water vapour ...... 32 4.2.3 Corrosion of metallic uranium in water ...... 34 4.2.4 Mechanism of uranium hydride formation ...... 35 4.2.5 Influence of irradiation and water radiolysis ...... 36 4.2.6 Influence of the counter ions and the pH on the corrosion mechanism ...... 36 4.2.7 Electrochemical uranium corrosion mechanism in water ...... 37 4.2.8 Corrosion of metallic uranium in cement-based materials ...... 39 4.2.9 Summary of the corrosion rates for uranium in alkaline conditions ...... 41 4.3 Conclusions ...... 42 5 Corrosion of beryllium ...... 43 5.1 Introduction ...... 43 5.2 Mechanism of beryllium corrosion ...... 43 5.2.1 Corrosion mechanisms of beryllium: general remarks ...... 44 5.2.2 Corrosion in highly alkaline, anoxic conditions ...... 44 5.3 Kinetics of beryllium corrosion in alkaline solutions ...... 4 5 5.4 Conclusions ...... 47

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 4 of 77 SCK•CEN/27548174 ISC: Public

6 Corrosion of aluminium ...... 48 6.1 Introduction ...... 48 6.2 Corrosion of metallic aluminium in alkaline conditions ...... 48 6.2.1 Corrosion in water ...... 48 6.2.2 Corrosion of metallic aluminium encapsulated in cement-based materials ...... 52 6.2.3 Summary of the corrosion rates for aluminium in alkaline conditions...... 54 6.3 Conclusions ...... 56 7 Corrosion of stainless steel ...... 57 7.1 Introduction ...... 57 7.2 Mechanism of stainless steel corrosion in geological disposal conditions ...... 58 7.2.1 Alloying additions ...... 59 7.2.2 Properties of the passive film ...... 61 7.2.3 Passivity breakdown ...... 62 7.3 Kinetics of stainless steel corrosion ...... 63 7.4 Conclusions ...... 65 8 Corrosion of Inconel alloys ...... 66 8.1 Introduction ...... 66 8.2 Corrosion of nickel and chromium ...... 66 8.3 Corrosion of Inconel in alkaline conditions ...... 69 8.4 Conclusions ...... 69 9 General conclusions ...... 70 10 References ...... 72

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 5 of 77 SCK•CEN/27548174 ISC: Public

Preface

Report SCK•CEN-ER-0412 “Corrosion of Metals in B&C waste” presents the deliverable in specification sheet 15-SCK-EBC-12 of the research package (RP) “EBC – Engineered Barriers Characterization (process identification & properties)” that forms part of the ONDRAF/NIRAS research programme on the geological disposal of high-level radioactive waste and spent fuel for the period 2015-2020. The topic covered by this report is mainly focused on the corrosion rate as a function of pH of metals relevant for B&C waste, other than carbon steel, and in particular Zirconium alloys, metallic uranium, aluminium, beryllium, stainless steel, and inconel.

This report summarizes corrosion rate data for the above mentioned metals in alkaline (pH 8- 13.5) and anoxic conditions. The ultimate goal of the report is to describe the dependence of the corrosion rate on pH.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 6 of 77 SCK•CEN/27548174 ISC: Public

Glossary of Abbreviations

APIMS Atmospheric pressure ionization mass spectrometer APT Atomic probe tomography BFS Blast furnace slag BR1 Belgian Reactor 1 BR2 Belgian Reactor 2 CAC Calcium aluminate cement EBSD Electron backscattering diffraction

Ecorr Corrosion potential

Ep Pitting potential

Er Repassivation potential ESHG Experimental system for high-accuracy evaluation of gas generation FIB Focused ion beam GC Gas chromatography HR-SEM High resolution scanning electron microscopy IAEA International Atomic Energy Agency INEEL Idaho National Engineering and Environmental Laboratory LILW Low- and intermediate-level waste MOX Mixed uranium and plutonium oxide MS Mass spectrometry OCP Open circuit potential OPC Ordinary Portland cement PFA Pulverized fuel ash QCM Quartz crystal microbalance RH Relative humidity SHE Standard hydrogen electrode SSE Saturated sulfate electrode TEM Transmission electron microscope UOX Uranium oxide XPS X-ray photoelectron spectroscopy

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 7 of 77 SCK•CEN/27548174 ISC: Public

List of Figures

Figure 1. (a) Monolith for category B waste (200 L drums). (b) Supercontainer for category C waste: 3D representation of a supercontainer containing 2 vitrified waste canisters (left) and a supercontainer containing 4 UOX spent fuel assemblies (right) [NIROND 2017]...... 15 Figure 2. Comparison of corrosion rates of pickled carbon steel versus pre-corroded carbon steel in various environments (Smart et al., 2014)...... 17 Figure 3. Schematic representation of the Japanese setup for online hydrogen gas measurements (Kaneko et al., 2004)...... 19 Figure 4. Potential-pH diagram of zirconium at 25 °C (Pourbaix, 1974)...... 22 Figure 5. Schematic drawing showing the three Zircaloy corrosion regions: pre-transition, transitory, and post-transition. The dashed lines indicate that early models recognized only the pre-transition and post-transition regimes (reproduced from Hillner (Hillner et al., 2000)). .... 22 Figure 6. Weight gain of zirconium alloys as function of time at (a) 300 °C, and (b) 400 °C (Allen et al., 2012)...... 23 Figure 7. Oxide layer thickness as a function of burn-up for Zircaloy-4 and M5 (Allen et al., 2012)...... 24 Figure 8. Formation of (a) Circumferential and (b) radial hydrides (Allen et al., 2012)...... 26 Figure 9. Corrosion rate of zirconium alloys as a function of pH in highly alkaline, anoxic conditions...... 30 Figure 10. Variation of the corrosion rate of uranium with RH in presence of water vapour and oxygen. Curve A, B and C were recorded with different uranium batches. At P, significant amounts of yellow UO3 were observed (Baker et al., 1966b)...... 33 Figure 11. Evolution of the corrosion rate in function of the pH of the solution at 100 °C (Baker et al., 1966a)...... 37 Figure 12. Potential-pH diagram of uranium (Pourbaix, 1974)...... 38 Figure 13. Polarisation curves for metallic uranium in KOH solution at pH 13.7 (Bullock et al., 1974)...... 39 Figure 14. Corrosion rates of metallic uranium in highly alkaline, anoxic conditions, as a function of pH at 25 °C...... 42 Figure 15. Equilibrium potential – pH diagram for the beryllium/water system at 25 °C (Pourbaix, 1974)...... 43 Figure 16. Potentiodynamic polarization curves (scan rate 1 mV/s) for beryllium in solutions of different pH (Gulbrandsen and Johansen, 1994)...... 44

Figure 17. Quasi-steady state passive current density (jp) for beryllium electrodes as a function of pH (Gulbrandsen and Johansen, 1994)...... 45 Figure 18. Plot of corrosion rate as a function of pH for beryllium in alkaline solutions (omitting outliers at pH 12 and pH 15)...... 46 Figure 19. Solubility of aluminium oxides in function of pH (Pourbaix, 1974)...... 49 Figure 20. Potential-pH diagram of aluminum (Pourbaix, 1974)...... 52 Figure 21. Corrosion rate evolution of metallic aluminium in NaOH solutions in function of pH and temperature. In blue, corrosion rate obtained at 25 °C. In red, corrosion rate obtained at 30 °C. In green, corrosion rate obtained at 50 °C. In orange, corrosion rate obtained at 60 °C...... 56 Figure 22. Overview of the composition and property links between the most important stainless steel alloys (Sedriks, 1996)...... 58

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 8 of 77 SCK•CEN/27548174 ISC: Public

Figure 23. Potential (E in V vs. SHE) – pH equilibrium diagrams for the system iron-water at 25 °C assuming (a) oxides and (b) hydroxides as solid substances (Pourbaix, 1974)...... 60 Figure 24. Potential (E in V vs. SHE) – pH equilibrium diagrams at 25 °C for the systems (a) chromium-water assuming Cr(OH)3 as solid substance, (b) chromium-water assuming Cr2O3 as solid substance, (d) nickel-water, and (e) molybdenum-water (Pourbaix, 1974)...... 60 Figure 25. Oxide layer thickness on a stainless steel as a function of potential for a Fe15Cr alloy in 0.5 M H2SO4, and for Fe10Cr and Fe20Cr alloys in 1 M NaOH, estimated using XPS. The film growth region is considerably wider in the alkaline medium, which also gives thicker films (Olsson and Landolt, 2003a)...... 62

Figure 26. Schematic of a polarization curve of a metal prone to pitting, with EP the pitting potential, ER the repassivation potential, and Ecorr the corrosion potential...... 6 3 Figure 27. Corrosion rate of stainless steel in highly alkaline, anoxic conditions at 50 °C...... 64 Figure 28. Pourbaix (potential-pH) diagram of the nickel-water system at 25°C (adapted from (Chivot, 2004))...... 67 Figure 29. Potential (E in V vs. SHE) – pH equilibrium diagrams at 25 °C for the systems (a) chromium-water assuming Cr(OH)3 as solid substance, (b) chromium-water assuming Cr2O3 as solid substance, (Pourbaix, 1974)...... 68 Figure 30. Theoretical conditions of corrosion, immunity and passivation of chromium, at 25°C (Pourbaix, 1974)...... 68 Figure 31. Isocorrosion diagram for Inconel 200 and 201 in NaOH (Metals, 2000)...... 69

List of Tables

Table 1. Overview of corrosion rates of zirconium alloys obtained in highly alkaline, anoxic conditions ...... 30 Table 2. Overview of corrosion rates of metallic uranium obtained in highly alkaline, anoxic conditions ...... 41 Table 3. Approximate corrosion rates for beryllium in alkaline solutions...... 46 In this section, the influence of pH on the corrosion rate is compiled from various authors (Table 4). Finally, these data are also presented in Figure 21 to illustrate the logarithm increase of the corrosion rate in function of the pH. This figure shows that the increase of the corrosion rate as a function of the pH is similar whatever the temperature used during the study...... 54 Table 5. Corrosion rates for aluminium in alkaline conditions...... 55 Table 6. Corrosion rates for stainless steel in highly alkaline, anoxic conditions...... 64

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 9 of 77 SCK•CEN/27548174 ISC: Public

Abstract

In Belgian B&C waste, many metallic waste streams are present, the most important being zirconium alloys, metallic uranium, beryllium, aluminium, stainless steel, and Inconel. The current report investigates the corrosion mechanisms and kinetics for these metals in an alkaline and anaerobic environment. The six metals all have in common that they are covered by a thin oxide film, which in principle protects them from accelerated corrosion. The properties of these oxide films are function of the irradiation history and of the environment. At high pH, the corrosion rate of zirconium alloys is in the order of 1 nm/y or less. These rates were measured on pre-transition oxides that may not be representative for real out-of-reactor zirconium alloys. The corrosion rate of metallic uranium is initially high in cement-based materials, dropping to the order of magnitude of a few µm/y after a few weeks and eventually reaching 0.1 µm/y. The corrosion of beryllium increases at high pH (above pH 8) reaching several µm/y at pH 13 and higher. Aluminium is an amphoteric material with corrosion rates of tens of mm/y at pH 13 and higher. Stainless steels exhibit low corrosion rates at high pH, in the order of a few nm/y to below 1 nm/y. Inconel is predicted to have corrosion rates lower than 10 nm/y.

Keywords

Corrosion, alkaline, cement, concrete, nuclear waste disposal, zirconium, metallic uranium, beryllium, aluminium, stainless steel, Inconel.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 10 of 77 SCK•CEN/27548174 ISC: Public

Executive Summary

In Belgium, radioactive waste is classified in three categories: A, B, and C. Category A is the short-lived (half-life less than 30 years) low- and intermediate- level waste (LILW), category B is the low- and intermediate-level waste contaminated by long-lived radionuclides and category C is the high-level conditioned waste containing large quantities of long-lived radionuclides. For the Belgian case, this corresponds basically to vitrified high-level waste and non-reprocessed spent fuel declared as waste. The origin of Belgian category B and C waste is multifold: fuel fabrication, electricity production in nuclear power plants, research reactors, pilot plants, … The main metals present in this waste are zirconium alloys (from the fuel claddings), metallic uranium (from the BR1 fuel), beryllium (from the BR2 moderator), aluminium (from the BR1 fuel cladding) and stainless steel (from the nuclear power plant reactor internals). Category B waste will be disposed of in cement-based monoliths, while category C waste will be placed in supercontainers, involving the use of a cement buffer. The disposal environment of both categories is based on Portland cement, with an initial pH of around 13.5, and maintaining a pH of 12.5 or higher for several tens of thousands of years. However, the purpose of this report is to describe the corrosion behavior of the considered metals in the pH range of 8-13.5, in order to provide input for all future disposal scenarios.

In general terms, the measurements of corrosion rates of metal can be grouped into three families:

 weight loss measurements,  electrochemical measurements, and  hydrogen gas measurements.

Weight loss measurements have the disadvantage that they can only present an average corrosion rate over the exposure duration of the specimens. In other words, no kinetic law can be derived from these measurements, whereas this is important to have an indication of the long-term corrosion behaviour. There exists a plethora of electrochemical techniques to determine the corrosion rate. The advantage of electrochemical techniques is that they can give real-time information and thus provide a kinetic law for corrosion processes. On the other hand, they are not as precise as hydrogen gas measurements. During the anaerobic corrosion of metals, hydrogen gas is produced. Direct measurement of the evolved hydrogen gas is by far the most accurate method of determining the corrosion rate and it also offers the advantage of low detection limits in comparison with weight loss measurements and electrochemical techniques. Interesting experimental setups are the Japanese ESHG (‘experimental system for high-accuracy evaluation of gas generation’) and the setup used in the work of Roger Newman and Nick Senior in Canada.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 11 of 77 SCK•CEN/27548174 ISC: Public

Corrosion of zirconium alloys

Three zirconium alloys have been used in Belgian nuclear power plants: Zircaloy-4 (a Zr-Sn alloy containing Sn, Fe and Cr), ZIRLO (a Zr-Nb-Sn-Fe-O alloy), and M5 (a Zr-Nb-Fe-O alloy without tin). In the in-pile corrosion of zirconium alloys three phases can be distinguished:

1. The early pre-transition phase, characterized by the formation of a thin, black, dense and tightly adherent corrosion film that grows thicker in accordance with a cubic rate law. 2. The midlife transition, or transitory state, that lies between the pre-transition and post- transition phase. This stage is comprised of a series of successive cubic curves, similar to the initial cubic kinetic curve, but initiating at shorter and shorter intervals. 3. The linear post-transition regime.

Data on the corrosion kinetics under alkaline and anoxic conditions, are scarce and are all derived from pre-transition oxides, whereas zirconium alloys would be expected to have post- transition oxides after their service life in the reactor. Zirconium alloys exhibit a long-term corrosion rate in the range 0.2-6 nm/y.

Corrosion of metallic uranium

In contact with water, metallic uranium corrodes to form uranium oxide and hydrogen, with the formation of uranium hydride as an intermediate compound. The corrosion rate of metallic uranium is higher in water than in air. However, the presence of oxygen in water leads to the formation of a protective layer of oxygen molecules at the surface of the metal, resulting in a reduction of the corrosion rate by a factor of 40. However, after only a few hundreds of hours, this inhibition disappears and the corrosion rate in both aerobic and anaerobic conditions is the same. The formation of the pyrophoric uranium hydride happened at the interface between metallic uranium and uranium oxide to reach a thickness of 3-5 nm. Moreover, this UH3 formation seems to form preferentially at grain boundaries or inclusion sites. No influence of irradiation was observed on the corrosion rate. However, water radiolysis could create H2O2, which is able to dissolve uranium in water. Increasing the pH of the water solution from 7 to 13.5 decreases the corrosion rate by only 10-15%, while at pH 2-3, the corrosion rate is decreased by a factor of 10. In cement-based materials at 25 °C, initial corrosion rates of 60- 150 µm/y were obtained. However, the corrosion rate rapidly dropped to 5-12 µm after only a few weeks. Moreover, values even lower than 0.1 µm/y have been measured if the cement- based material is no longer saturated with water.

Corrosion of beryllium

Beryllium is an amphoteric material whose corrosion behaviour can be compared to that of aluminium. The literature data on beryllium corrosion rates in highly alkaline, anoxic conditions is scarce. The corrosion rate seems to be a linear function of pH. In the pH range between 9 and 13, the corrosion rate is in the order of magnitude of several µm/y.

Corrosion of aluminium

Due to the amphoteric behaviour of the oxide/hydroxide layer formed at the aluminium surface, metallic aluminium corrodes in both acidic and alkaline media. However, this layer is protective against corrosion in the pH range between 4 and 8.5. In alkaline media, aluminium - corrodes to form aluminate ions (Al(OH)4 ) and hydrogen gas. If the temperature and the pH

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 12 of 77 SCK•CEN/27548174 ISC: Public

of the system increase, the corrosion rate increases too. At lower hydroxide concentrations (0.5 – 1 N), the corrosion rate is proportional to the cube root of the concentration of NaOH and

KOH, while it is proportional to the square root of the concentration of Ca(OH)2. At higher hydroxide concentrations, the corrosion rate is directly proportional to the hydroxide concentration. In alkaline solutions, corrosion rates as high as 500000 µm/y have been recorded at pH ~14 in some specific conditions. If encapsulated in cement-based materials, at the early stage of the encapsulation, when the cement-based material is still wet, the corrosion rate is high. However, the rate decreases fast to reach values as low as 0.5 µm/y after a few months/years. The corrosion rate can be reduced even further by adding inhibitors, such as

LiNO3 or sulphates, to the cement-based material. The cement composition can also be altered (e.g. addition of BFS) to decrease the pore water pH.

Corrosion of stainless steel

The corrosion resistance of stainless steel relies on the presence of a thin passive film on the surface of the metal. This oxide film is rich in chromium and forms and heals itself in the presence of oxygen. The basic mechanism of stainless steel corrosion is formed by the reactions of the main constituent, namely iron. The anoxic corrosion of stainless steel leads to the production of hydrogen gas. The main effect of an increased pH, as is the case in geological disposal, is a lower dissolution rate of the oxide. This leads to a thicker passive film and a lower corrosion rate. The corrosion rates measured in anoxic, alkaline conditions are in the order of several nm/y, with a maximum rate of 10 nm/y.

Corrosion of Inconel alloys

The most important Inconel alloys for the nuclear industry are alloys 600, 690, 718 and 800. There are no data found in the literature for the corrosion of Inconel alloys under anaerobic and highly alkaline conditions. Therefore, we suggest a maximum corrosion rate value of 10 nm/y, corresponding to the approximate corrosion rate of stainless steel, which also relies on a chromium-based passive film.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 13 of 77 SCK•CEN/27548174 ISC: Public

1 Introduction In Belgium, radioactive waste is classified in three categories: A, B, and C. Category A is the short-lived (half-life less than 30 years) low- and intermediate- level waste (LILW), category B is the low- and intermediate-level waste contaminated by long-lived radionuclides and category C is the high-level conditioned waste containing large quantities of long-lived radionuclides. For the Belgian case, this corresponds basically to vitrified high-level waste and non-reprocessed spent fuel declared as waste. The purpose of this report is to describe the corrosion behavior of the considered metals in the pH range of 8-13.5, in order to provide input for all future disposal scenarios.

1.1 Metals present in category B and C waste The origin of Belgian category B and C waste is multifold: fuel fabrication, electricity production in nuclear power plants, research reactors, pilot plants,… The main metals present in this waste are zirconium alloys (from the fuel claddings), metallic uranium (from the BR1 fuel), beryllium (from the BR2 moderator), aluminium (from the BR1 fuel cladding), stainless steel (from the nuclear power plant reactor internals), and Inconel (from reactor internals). Another important metallic waste stream is carbon steel, but its corrosion is treated in a separate report (Kursten, 2015).

1.2 Engineered barrier system for category B and C waste The current reference design of the engineered barrier system is often referred to as the “supercontainer design”, which in fact is the name of the category C waste disposal package. It contains specific designs for the disposal of category B waste and category C waste.

1.2.1 Monolith for category B waste The primary waste packages containing category B waste are immobilised in mortar in concrete caissons to form monoliths (Figure 1a). Several monolith B designs exist to accommodate the large variety of primary waste packages, which differ according to the source of the waste.

In the supercontainer design (Figure 1b), the primary waste packages of high-level waste (vitrified waste and spent nuclear fuel) are surrounded by a carbon steel overpack, a buffer made of concrete containing Portland cement, and a stainless steel envelope. Carbon steel has been chosen as the overpack material because in the high pH environment of the supercontainer it is covered by a passive film and prone to general corrosion rather than the less predictable localized corrosion. The purpose of the buffer is to act as a radiation shield and as a pH controller in order to create favourable conditions with regard to the passivation of carbon steel. A stainless steel envelope will surround the buffer. The envelope will serve as a mould for construction of the buffer, serve as a first barrier against the ingress of aggressive species, provide mechanical strength and confinement, and could facilitate retrievability.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 14 of 77 SCK•CEN/27548174 ISC: Public

(a)

(b)

1.3 Geochemical evolution of the engineered barrier system The monolith B and supercontainer designs have in common that they rely on a cement-based buffer or embedding matrix that will maintain the pH at high levels for several thousands of years. As the type of waste does not have a direct influence on the evolution of the Portland cement (although the heat dissipation of the waste will change the temperature of the cement and hence the diffusion parameters), a general description of the engineered barrier evolution suffices for both waste categories. In this discussion, we will focus on the evolution of pH.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 15 of 77 SCK•CEN/27548174 ISC: Public

2 Qualification of techniques for the determination of corrosion rates In the determination of corrosion rates of metals, several techniques have been used. Roughly speaking, the methods reported in the literature can be divided into three groups:

 Weight loss measurements  Electrochemical measurements  Direct measurement of the evolved hydrogen gas due to anaerobic corrosion

Before we discuss these three groups however, we will diverge on a common and important step in corrosion studies (prior to the actual corrosion rate measurement), namely the surface preparation.

Surface preparation

Ideally, the surface of the investigated sample should be identical to the surface of the actual waste form. In many cases, the metallic samples (in the waste inventory) are covered with an oxide or hydroxide layer whose properties are a function of the environmental and irradiation history of the metal. This implies that either ‘real’ waste samples have to be used or that the surface layer of the samples has to be engineered to closely resemble the surface of real samples. Because this is often impossible and because in many cases standardization of tests is desired, often a clean sample surface is used for the corrosion tests. Such a clean surface is acquired in most cases by chemical cleaning and/or polishing. This involves removal of oxide layers and polishing the samples to, in some cases, a high finish. A common preparation method for metals is acidic pickling. The purpose of pickling is to remove impurities, stains, inorganic contaminants and rust and scale. The basis of the pickling is mostly a strong acid, such as hydrochloric acid and sulphuric acid. The same cleaning degree can be obtained by mechanical polishing on SiC paper. As said, the advantage of these procedures is the production of a reproducible and well described surface, which enables to compare the results from different laboratories. However, the question has to be raised in how far these standardized surfaces represent reality. For example, on passivating metals, pickling will lead to an excessively high initial corrosion rate due to the formation of a fresh passive layer in the beginning of the exposure to the corroding environment (Figure 2). This means that tests have to last long enough for the corrosion rate to drop to realistic values. Another example is the corrosion of zirconium alloys, which exhibit a different kinetic regime for samples irradiated in a nuclear reactor (i.e. post-transition kinetics) compared to fresh samples (i.e. pre-transition kinetics). This will be explained in Chapter 3, but for now it suffices to mention that the use of pre-transition samples may lead to an erroneous measurement of the corrosion rate. The inconvenience with surface preparation is that the experimental error it produces is not easy to measure.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 16 of 77 SCK•CEN/27548174 ISC: Public

Figure 2. Comparison of corrosion rates of pickled carbon steel versus pre-corroded carbon steel in various environments (Smart et al., 2014).

Weight loss measurements

The determination of the corrosion rate by weight loss consists in comparing the weights of a given sample before and after the corrosion test. After the surface preparation step (which may also be omitted in the use of ‘as-received’ samples) the metallic samples are weighed and their dimensions are measured in order to determine the exposed surface area. The original area is used to calculate the corrosion rate during the test. Therefore, any error in measuring the surface area will be introduced into the calculation of the corrosion rate. A second source of error is that the corrosion rate determination is based on the total weight of oxide produced, including not only the oxide produced through corrosion, but also through pre-oxidation of the sample by contact with air before the experiment, insufficient degassing or leakage of the corrosion cells resulting in residual oxygen (in the case of anoxic tests), and contact of the sample with air after the corrosion test but before the dissolution of the oxide. The main shortcoming of weight loss measurements however is that they only produce an average corrosion rate over the duration of the corrosion test, in other words no kinetic law can be derived from this method, whereas in most cases (and certainly in the safety assessment of nuclear waste disposal systems), it is important to have a good idea of the long-term corrosion rate.

Electrochemical techniques

There exists a plethora of electrochemical techniques that enable to determine, indirectly, the corrosion rate: polarization curves, Tafel extrapolation, linear polarization resistance, galvanostatic pulse, electrochemical impedance spectroscopy and electrochemical noise. A detailed description of these methods can be found in e.g. (Bard and Faulkner, 2001; Kursten,

2015). Most of them yield a value of the corrosion current density (icorr), based on the measurement of the polarization resistance RP:

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 17 of 77 SCK•CEN/27548174 ISC: Public

(1) where B is a proportionality constant. This constant is estimated to be between 5 and 20 mV although it is difficult to measure its exact value. Andrade estimates this value to be 26 mV for steel in the active state and 52 mV for steel in the passive state (Andrade and González, 1978). Therefore, the uncertainty on this value introduces an error on the obtained corrosion rate. Some techniques, notably polarization curves and Tafel extrapolation, obtain a direct measurement of icorr. This current density is then transformed to a corrosion rate using Faraday’s law.

When comparing electrochemical techniques, one has to distinguish between potentiodynamic and potentiostatic methods. Potentiodynamic methods (which involve sweeping the voltage and recording the corresponding current density) tend to overestimate the corrosion rates while potentiostatic methods (holding the voltage at a constant value and recording the corrosion rate) are more precise. Also, the total current measured is the sum of all surface reactions involving electron transfer. Part of this is the corrosion process, but there can also be contributions from other electrochemical reactions. On the other hand, the advantage of electrochemical techniques is that they can give real-time information on the total current and thus provide a kinetic law for corrosion processes.

Hydrogen gas measurements

During the anaerobic corrosion of the metals discussed in this report, hydrogen gas is produced. Direct measurement of the evolved hydrogen gas is by far the most accurate method of determining the corrosion rate and it also offers the advantage of low detection limits in comparison with weight loss measurements and electrochemical techniques. A recent overview of hydrogen measurement methods is given in (Druyts and Jobbágy, 2017). There are essentially two technical approaches: by (small) real-time hydrogen sensors or by large analytical tools that are (mainly) used batch-wise (although there exists a Japanese application that uses mass spectrometry for real-time measurements, which is described below). The small hydrogen sensors are essentially portable and allow for real-time and online measurements of the hydrogen concentration, but they show the disadvantage of a relatively high limit of detection (in the order of tens of ppm’s). On the other hand, analytical tools such as gas chromatographs (GC) and mass spectrometers (MS) offer a (much) lower limit of detection but they are not so readily applicable for real-time measurements. However, as these tools have a very low limit of detection, this allows for a high gas sampling frequency, allowing the observation of the evolution of the corrosion rate with time. A particularly interesting experimental setup is the Japanese ESHG (‘experimental system for high-accuracy evaluation of gas generation’) described in (Kaneko et al., 2004) (Figure 3). This system is based on the use of mass spectrometry. The MS used is an atmospheric pressure ionization mass spectrometer (APIMS), in which the ionization method is atmospheric ionization and the analytical section is based on a quadrupole mass spectrometer. The detection limit is reported to be 1 µg/L (1 ppb). This setup was recently used to investigate the anaerobic corrosion of zirconium alloys and stainless steels, as described in sections 1 and 7 of this report.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 18 of 77 SCK•CEN/27548174 ISC: Public

A weak point in any hydrogen gas detection is the possible presence of leaks. All connections have to be hydrogen-tight in order to avoid an underestimation of the corrosion rate.

Figure 3. Schematic representation of the Japanese setup for online hydrogen gas measurements (Kaneko et al., 2004).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 19 of 77 SCK•CEN/27548174 ISC: Public

3 Corrosion of zirconium alloys

3.1 Introduction Zirconium alloys are widely used in nuclear reactors as fuel cladding, because they offer a low neutron cross section, reasonable mechanical properties and adequate corrosion resistance in high temperature waters. Several zirconium alloys exist for nuclear applications, most of which are grouped in either Zr-Sn or Zr-Nb alloys. Historically, Zircaloy-2 and Zircaloy-4 played an important role. For example, from the startup of the Belgian nuclear power plants up to the year 2000, Zircaloy-4, a Zr-Sn alloy containing 1.5 wt.% Sn, 0.22 wt.% Fe and 0.10 wt.% Cr, was the only zirconium alloy used in Belgian reactors. To improve the in-pile corrosion resistance, new alloys were developed in the late 1990’s and early 2000’s, to which niobium was added (Cox, 2005; Duan et al., 2017). The addition of niobium, to improve the strength, ductility and corrosion resistance of zirconium alloys, was a Russian initiative, which led to the development of new alloys (E110 – Zr-1%Nb; E365 – Zr-1%Nb- 1%Sn- 0.4%Fe). Afterwards, ZIRLO, a Zr-Nb- Sn-Fe-O alloy, was developed by Westinghouse. The purpose of this new alloy was to combine the beneficial characteristics of both the Zr-Sn alloys (high strength and creep resistance) and the Zr-Nb alloys (high ductility and corrosion resistance). It has been progressively introduced in the Belgian assemblies from 2000 onwards. AREVA in its turn developed the M5 material, which is a Zr-Nb-Fe-O alloy without tin. The incorporation of niobium and the elimination of tin has resulted in higher in-pile corrosion resistance and reduced irradiation-induced growth, relative to standard Zircaloy-4. In Belgium, two waste management routes for zirconium alloys are envisaged. The first is the direct disposal of spent fuel elements. In this case, the zirconium cladding is disposed together with the fuel rods in a supercontainer, offering a disposal environment with pH 13.5 which will evolve from oxic and warm to anoxic and low temperature. The second route is reprocessing of the fuel. In this case, the zirconium alloy claddings are separated from the fuel rods, which are reprocessed resulting in vitrified waste.

3.2 Corrosion mechanisms of zirconium alloys The corrosion behavior of zirconium alloys relies on the properties of the oxide film that covers the metal when it comes into contact with an oxidizing environment. Indeed, Zr and its alloys are highly reactive and have a high affinity for oxygen, and therefore readily develop an ‘air- formed’ oxide film. The structure and properties of the oxide film depend on the history of the zirconium, both in-pile and out-of-pile. Therefore, this chapter is structured as follows. First we discuss the general corrosion mechanism of zirconium alloys. Then we investigate the in-pile corrosion of zirconium and the influence of alloying elements to come to a comparison between Zircaloy-4, ZIRLO, and M5. The in-pile corrosion of zirconium alloys is part of their corrosion history and as such will determine the properties of the oxide film at the moment of disposal. Finally, we will discuss the corrosion mechanisms of zirconium alloys in an environment relevant for the Belgian disposal design, i.e. highly alkaline and anoxic conditions, including the effect of pH.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 20 of 77 SCK•CEN/27548174 ISC: Public

Zirconium alloys in general have a high corrosion resistance. This is due to the formation of a protective layer. The corrosion behavior of zirconium is well described in literature (Allen et al., 2012; Gras, 2014; Hillner et al., 2000; Lefebvre and Lemaignan, 1997; Mogoda, 1999; Motta et al., 2015). The basis of the discussion of the corrosion of zirconium is the Pourbaix diagram as shown in Figure 4 (Pourbaix, 1974). It is clear from this potential – pH equilibrium diagram (in water at 25 °C) that the range of thermodynamical stability of zirconium is between pH 3.5 and pH 12.5. In this area, zirconium is readily passivated by ZrO2 and corrodes extremely slowly.

The stability domain of ZrO2 can be extended to higher pH values when the concentration of zirconium ions in solution increases. The corrosion rate determining step is the migration of charged species (oxygen ions and electrons) across the oxide. Therefore, initially, kinetics should be parabolic, but it is experimentally determined that the kinetics are in fact cubic (Allen et al., 2012), according to the following rate law:

∆ (2) where W is the weight gain in g, k1 is a kinetic constant in g/s, and t is time in s. As corrosion progresses however, a critical oxide thickness is reached (approximately 2-3 µm) at which the protection of the oxide layer breaks down and a rapid increase in oxidation rate is observed (Figure 5). The breakdown is known as ‘transition’ and is followed by a reduction in the oxidation rate as a new protective layer forms. The time between transitions is strongly dependent on the alloy composition and microstructure. This phenomenon is observed in reactors or autoclaves, but to our knowledge it does not occur at room temperature. Summarizing, three phases in the oxidation of zirconium alloys can be distinguished (Hillner et al., 2000):

1. The early pre-transition phase, characterized by the formation of a thin, black, dense and tightly adherent corrosion film that grows thicker in accordance with a cubic rate law. 2. The midlife transition, or transitory stage, that lies between the pre-transition and post- transition states. As mentioned above, this stage is comprised of a series of successive cubic curves, similar to the initial cubic kinetic curve, but initiating at shorter and shorter intervals. 3. The linear post-transition regime.

The influence of radiolysis on the corrosion rate and growth of the oxide is reported to be low (Lefebvre and Lemaignan, 1997). After irradiation in the reactor, most zirconium alloys are in the post-transition regime, with linear kinetics. In the next paragraphs, we will first discuss the in-reactor corrosion of zirconium alloys, followed by a description of the out-of-pile corrosion in the expected geological disposal conditions.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 21 of 77 SCK•CEN/27548174 ISC: Public

Figure 4. Potential-pH diagram of zirconium at 25 °C (Pourbaix, 1974).

Figure 5. Schematic drawing showing the three Zircaloy corrosion regions: pre-transition, transitory, and post-transition. The dashed lines indicate that early models recognized only the pre-transition and post- transition regimes (reproduced from Hillner (Hillner et al., 2000)).

3.2.1 Corrosion of zirconium alloys in reactor conditions There exists a vast amount of literature on the corrosion of zirconium alloys in reactor conditions. It is not the purpose of this report to give an exhaustive review of the literature, as this has already been done in several papers (Cox, 2005; Garzarolli and Garzarolli, 2012; Motta, 2011; Yan et al., 2015). We will however, focus on the differences between Zircaloy-4, M5, and ZIRLO, as these are alloys of importance to the Belgian nuclear power plants. Also, our focus

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 22 of 77 SCK•CEN/27548174 ISC: Public

will be on those oxide film modifications that are important for the subsequent geological disposal of zirconium alloys.

In the core of the nuclear power plant, zirconium alloys are exposed to a high neutron flux. This induces irradiation damage in the cladding material and the oxide layer on the surface of this material. When referring to the general corrosion mechanism in Figure 5, Zirlo and M5 behave rather differently, as can be seen in Figure 6 and Figure 7.

Figure 6. Weight gain of zirconium alloys as function of time at (a) 300 °C, and (b) 400 °C (Allen et al., 2012).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 23 of 77 SCK•CEN/27548174 ISC: Public

Figure 7. Oxide layer thickness as a function of burn-up for Zircaloy-4 and M5 (Allen et al., 2012).

3.2.2 Uniform corrosion Uniform corrosion is a process that occurs with approximately the same speed over the entire surface of an object. The anodic process consists of oxidation of the metal:

→ (3)

0 where VO indicates a lattice vacancy in the ZrO2 layer. The corresponding cathodic reaction at the oxide/coolant interface can be the reduction of water:

→ (4)

Uniform corrosion is a passivating event since a protective layer of zirconium oxide is formed. In the early stages of the oxide formation (cfr. Figure 5), the oxide layer is indeed dense and protective, but with time, and at higher burn-up, cracking of the oxide layer occurs and the corrosion is accelerated.

3.2.3 Microstructure of the oxide (under uniform corrosion) In the early stages of oxidation, the oxide grains are predominantly tetragonal or cubic (Allen et al., 2012). As the grains grow, columnar grain growth occurs and the tetragonal grains tend to transform to monoclinic oxide. The size of the columnar grains and their grain-to-grain disorientation have been related to the transition thickness.

The microstructure of the oxide layer depends on the burn-up of the cladding. Bossis compared the oxides on zircaloy-4 and M5 at high burn-up (Bossis et al., 2006). In Zircaloy-4, with a fluence of 10.9 x 1025 n/m², he observed cracks in the oxide layer at an oxide thickness of 30 µm. The cracks were distributed circumferentially (i.e. parallel with the metal/oxide interface) with a periodicity of 1.8-1.9 µm. In addition, crack-free veins were present on the circumference and at the metal/oxide interface. Moreover, it seemed that these crack-free paths constituted preferential paths for the propagation of radial cracks through the layer thickness.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 24 of 77 SCK•CEN/27548174 ISC: Public

On M5, Bossis observed much less cracking. Also, the oxide thickness is a factor five less for the same burn-up compared to Zircaloy-4. The cracks that are present are distributed through the oxide layer with a periodicity of 2 µm. There are few radial cracks and the metal/oxide interface seems very stable with only a few cracks located at this interface.

Figure 7 compares the oxide thickness of Zircaloy-4 and M5 as a function of burn-up (Allen et al., 2012). From this graph it is clear that M5 is much less prone to high burn-up corrosion acceleration than Zircaloy-4. Several hypotheses exist to explain the onset of the high burn-up corrosion acceleration:

 Hydride precipitation and accumulation at the metal/oxide interface

 Dissolution of the Zr(Fe,Cr)2 precipitates  Sn content  Li effect

De Gabory used Transmission Electron Microscopy (TEM) to compare the structure of the oxide films formed on Zircaloy-4 and ZIRLO, respectively (de Gabory et al., 2015). The differences between the two alloys in terms of oxide microstructure and kinetics were not as pronounced as those between Zircaloy-4 and M5. The oxide layer was predominantly constituted of columnar grains which extended to the oxide/metal interface. The oxide was mainly composed of ZrO2, but intermetallics were also present.

3.2.4 Effect of ion irradiation on the microstructure of the oxide Irradiation by fast neutrons in the reactor yields microstructure changes in the zirconium base material, notably dislocation loops, second phase precipitates, and gas bubbles (Yan et al., 2015). This has consequences for the corrosion properties of the material. It is found that point defects and dislocation loops, induced by neutron irradiation, provide a fast diffusion channel for oxygen atoms, leading to an increase in oxidation rate. This would mean that the post- transition stage is reached more rapidly under the influence of neutron irradiation.

3.2.5 Hydrogen absorption The IAEA Technical Document ‘Waterside Corrosion of Zirconium Alloys in Nuclear Power Plants’ (Agency, 1998) provides a detailed description of the process of the absorption of hydrogen into zirconium-based materials. Because hydrogen absorption can impact the final mechanical properties and the long-term behavior of zirconium alloys under the expected geological disposal conditions, we will provide a brief discussion of the involved phenomena.

The reaction of Zr with water to form zirconium oxide produces atomic hydrogen:

→ (5)

The proton released in this reactor either recombines with another proton to form gaseous hydrogen or diffuses into the zirconium where it can form zirconium hydrides. The amount of absorbed hydrogen can vary as a function of the oxide thickness (Allen et al., 2012). Some researchers have shown that the initially high hydrogen absorption rate decreases as the oxide grows thicker, to accelerate again after the oxide reaches the transition region. Other researchers have observed an increase in hydrogen uptake just before the oxide transition. It is

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 25 of 77 SCK•CEN/27548174 ISC: Public

likely however, that the development of porosity after transition facilitates the pickup of hydrogen.

The absorbed hydrogen can form hydrides, leading to the embrittlement of the zirconium alloy cladding (Suman et al., 2015). The formed hydrides can be circumferential or radial (Figure 8), with the radial hydrides presenting the greater concern, as they can result in through-the-wall cracks of the cladding and an increased surface exposed to corrosion.

The hydrogen solubility in zirconium alloys decreases with decreasing temperature. As the overall hydrogen content increases as a result of increased corrosion at lower temperatures, eventually the outer layer of the cladding reaches saturation and a hydride rim will start to form, with hydrides building up circumferentially. This may lead to increased corrosion (Cox, 2005), but there is still research needed to come to a full understanding of this phenomenon.

Figure 8. Formation of (a) Circumferential and (b) radial hydrides (Allen et al., 2012).

3.2.6 Corrosion of zirconium alloys in anoxic highly alkaline conditions The corrosion of zirconium alloys in anoxic highly alkaline conditions is determined by the following processes:

 The corrosion of zirconium alloys is mainly governed by the presence of a passive layer

of ZrO2. The structure and composition of this passivating oxide layer are determined

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 26 of 77 SCK•CEN/27548174 ISC: Public

by the in-reactor history of the cladding. What also has to be taken into account is that, in the case of reprocessing, the cladding tubes are cut up and washed with nitric acid to dissolve the spent fuel (Poulard et al., 2001). Also, during the irradiation in the reactor, fission products are implanted by recoil in the oxide layer and remain implanted in the hulls even after reprocessing.  Modes of localized corrosion have to be taken into account, although this is only the case in oxidizing conditions (i.e. under anoxic conditions, no localized corrosion occurs,

except in the presence of oxidizing species other than O2, for example due to radiolysis).  The formation of the passive film could be prevented in penalizing circumstances, for example in the case of galvanic corrosion.

We will discuss the uniform corrosion rate (governed by a passive film) later in this chapter. First we will look at the probability of localized corrosion.

3.2.6.1 Localized corrosion of zirconium alloys Four main modes of localized corrosion could occur on zirconium alloys: pitting and crevice corrosion, galvanic corrosion, and microbiologically influenced corrosion.

3.2.6.1.1 Pitting and crevice corrosion

Pitting or crevice corrosion are initiated when the free corrosion potential (Ecorr) of the zirconium alloys rises above the pitting potential (Ep) or crevice potential (Ecr). Amongst other parameters, the pitting potential increases with halide concentration, as described by the following simple relationship (Szklarska-Smialowska, 1986):

. (6) where x is the aggressive (halide) ion. To exclude pitting or crevice corrosion, it has to be proved that the repassivation potential Erp (i.e. the potential below which there is neither initiation nor propagation of localized corrosion) is higher than the free corrosion potential of the material.

The corrosion potential is moved into the anodic direction by the presence of oxygen or of oxidizing species coming from the radiolysis of the water by the activated zirconium alloy.. , . Nevertheless, we cannot neglect localized corrosion under anoxic highly alkaline conditions. Pitting of zirconium alloys can cause hydrogen evolution and the disintegration of the oxide layer (Postlethwaite and Onofrei, 1979). According to a review by Gras (Gras, 2014), the probability of pitting and crevice corrosion, however, is very low in alkaline conditions. Data for

M5 show that in all investigated solutions, at pH 12, Ecorr is far more cathodic than Erp, which in some cases lies above the potential of the oxygen evolution reaction (Gras, 2014). Kurashige et al. conducted corrosion tests of Zircaloy-4 at 30 and 45 °C in alkaline (pH 10.5 or 12.8) and high chloride (0.55 M) solutions, under anoxic conditions. Pitting corrosion was not observed after exposing the metal for 400 or 500 days to the alkaline solutions at the free corrosion potential (Kurashige, 1999).

3.2.6.1.2 Galvanic corrosion In the final disposal environment, the zirconium alloys may be in contact with other metallic materials: austenitic stainless steel, nickel-based alloys,… If the environment is conductive enough and the potential difference between the two metals large enough (a rule of thumb is > 50 mV) then galvanic corrosion can occur. The risk of galvanic corrosion with pre-corroded

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 27 of 77 SCK•CEN/27548174 ISC: Public

zirconium alloys (as is the case for irradiated Zr alloys from nuclear power plants), however, can probably be neglected according to a review by Gras (Gras, 2014).

3.2.6.1.3 Microbiologically influenced corrosion Microbial activity may act as a precursor to pitting: microbes could produce a local acidic environment that may provoke multiple penetrations through the hulls. This type of corrosion, however, is considered extremely unlikely in anoxic highly alkaline environments.

3.2.6.2 Uniform corrosion of zirconium alloys The different corrosion regimes of zirconium alloys were described above. Most tests to determine the corrosion rate of zirconium alloys in geological disposal conditions start with a sample surface in the metallic condition (i.e. polished, pickled,…) which means that during the tests the oxides are in the pre-transition stage. Indeed, at low temperatures the film growth proceeds too slowly to ever reach a critical oxide thickness for transition to occur. However, the actual zirconium alloy waste resulting from the reactors will be in the post-transition phase due to high-temperature corrosion in the nuclear power plant.

The microstructure of the oxide has an influence on the corrosion mechanism and rate. Hillner states that when zirconium alloys are transferred from an aggressive environment to one that is considerably less aggressive, the corrosion process may continue at the more aggressive rate for some time (Hillner et al., 2000). This behaviour has been termed the ‘memory’ effect. However, Garzarolli conducted experiments that showed that after approximately 100 days this memory effect has vanished (Garzarolli et al., 1982).

3.3 Corrosion kinetics of zirconium alloys in geological disposal conditions In considering the corrosion kinetics of zirconium alloys in a highly alkaline and anoxic environment, we will assume that the oxide is in the post-transitory phase (Figure 7). At that point, multiple circumferential layers will constitute the oxide due to subsequent cracking/reoxidation sequences, as explained above.

The post-transition kinetics are approximately linear in time (Hillner et al., 2000), which can be described by an expression of the form:

∆ (7) where W is the specimen weight gain, in units of mg/dm², t the exposure time, in units of days, KL the empirical constant, usually termed the linear (or post-transition) rate constant, in units of mg/dm²/day, and C another constant, in units of mg/dm². The temperature dependence of the linear rate constant KL obeys an Arrhenius law of the form:

/ (8) where B is an empirical constant, in units of mg/dm²/day, QL the activation energy for the post- transition (linear) corrosion region, in units of J/mol, R the universal gas constant, 8.31 J/mol K, and T the absolute temperature, in units of K.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 28 of 77 SCK•CEN/27548174 ISC: Public

Kato investigated the validity of high-temperature kinetic equations at low temperatures (Kato et al., 2014). He found that at low temperatures the zirconium oxidation obeyed a parabolic- cubic law.

In the determination of the expected corrosion rate ranges of Zircaloys end-of-life irradiated material should ideally be used to obtain the linear post-transition corrosion rate. However, this information is not readily available.

To determine the corrosion kinetics of zirconium alloys at the low temperatures foreseen under geological disposal conditions, two different approaches are used (Allen et al., 2012). The first approach is to extrapolate the corrosion rates obtained at higher temperatures to the temperature range relevant for geological disposal. This is due to the observation that at near- ambient temperatures corrosion rates are too low to be determined with traditional techniques such as weight loss measurements. In this approach, equations such as described above are used. The second approach is the direct measurement of corrosion rates at low temperatures. In the following discussion, we focus on corrosion rates obtained using the latter approach.

3.3.1 Influence of pH The influence of pH on the corrosion rate can be determined by compiling the results from various authors (see Table 1). Kato et al (Kato et al., 2014) found a corrosion rate of 17 nm/y after 90 days at pH 12.5. The rate law was initially almost parabolic but gradually approached a cubic dependence. Adler Flitton retrieved Zircaloy-4 coupons that had been buried for six years at INEEL (Adler Flitton et al., 2004). The soil had a pH of 8.1-8.3 and the reported corrosion rate ranged from ‘non measurable’ to 0.5 nm/y. Weight gain measurements have a low accuracy (see chapter 1). Also, the kinetic data based on weight gain measurements represent an average corrosion rate over the investigated period, without providing an instantaneous corrosion rate. Hansson used potentiodynamic techniques to measure the corrosion rate of Zircaloy-2 in anaerobic concrete in the pH range 12-13.8 (Hansson, 1984). The reported rates range from 20 to 90 nm/y, depending on pH. It has to be remarked that potentiodynamic techniques tend to overestimate the corrosion rate. Therefore, these values are excluded from our survey. Kurashige (Kurashige, 1999) determined the corrosion rate of Zircaloy in alkaline groundwater at pH 12.5, using gas generation measurements. He obtained an average corrosion rate of 1 nm/y for the first 500 days. This result seems highly reliable due to the employed method and the long term of the experiments. A remark is that the data represent pre-transition kinetics. Wada also used gas generation measurements on Zircaloy in the pre- transition state at pH 12.5 and obtained long-term corrosion rates of 0.2 nm/y after 300 days (Wada, 1999). Based on the fact that he used a real-time measurement of the corrosion rate and that the tests duration was long enough to obtain a steady state, the reported rate seems reliable. Gad Allah (Gad Allah et al., 1989) and Huot (Huot, 1999) used electrochemical impedance spectroscopy to determine corrosion rates of zirconium at pH 14. They obtained values of 100 nm/y. However, these rates were obtained on oxide layers of only a few hours old (and thus not representative of the long-term corrosion behavior of zirconium, taking into account the experimentally observed cubic/parabolic rate law) and therefore are not withheld in this survey. Sakuragi (Sakuragi et al., 2012, 2013) used hydrogen measurements to determine the corrosion rate of Zircaloy-4 at pH 12.5. He obtained corrosion rates after 1500 days of 6 nm/y. Looking at the applied experimental method, and taking into account the relatively long duration of the measurements, these data can be considered as being of high quality.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 29 of 77 SCK•CEN/27548174 ISC: Public

Figure 9 shows a graphic representation of the corrosion rate as a function of pH for zirconium alloys, omitting the data provided by Kato, because the duration of the test was too short to produce reliable long-term corrosion rates.

Table 1. Overview of corrosion rates of zirconium alloys obtained in highly alkaline, anoxic conditions

pH T (°C) Duration Corrosion rate Technique Reference (days) (µm/y) 12.5 30 90 0.017 hydrogen (Kato et al., 2014) 8.1-8.3 15 6 years 0.0005 gravimetric (Adler Flitton et al., 2004) 12.5 30 500 0.001 hydrogen (Kurashige, 1999) 12.5 30 300 0.0002 hydrogen (Wada, 1999) 12.5 30 1500 0.0006 hydrogen (Sakuragi et al., 2013)

Figure 9. Corrosion rate of zirconium alloys as a function of pH in highly alkaline, anoxic conditions.

3.4 Conclusions The corrosion rate of zirconium alloys depends on the oxidation and irradiation history of the metal. In the reactor, an oxide thickness is reached (approximately 2 µm) at which a change in kinetics occurs, changing from cubic/parabolic to approximately linear. This is called the transition, and it is accompanied by a breakaway (spalling) of the oxide layer.

Data on the corrosion kinetics in highly alkaline and anoxic conditions are scarce and are all derived from pre-transition oxides, whereas zirconium alloys would be expected to have post- transition oxides after their service life in the reactor.

In the pH range 8-12.5, no discernible influence of pH on the corrosion rate is observed.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 30 of 77 SCK•CEN/27548174 ISC: Public

4 Corrosion of metallic uranium

4.1 Introduction Due to its high density (19.05 g/cm³) and its nuclear properties (fissionability of 235U), metallic uranium possesses favourable characteristics for different military applications, such as military tank armour, armour piercing ammunition, and civilian uses (e.g. fuel for nuclear reactors, radiation shielding,…). At SCK•CEN, metallic uranium is used as fuel in two nuclear research reactors. Natural unalloyed metallic uranium is used in the BR1 reactor while uranium alloyed with aluminium is used in the BR2 reactor.

Metallic uranium is highly reactive in presence of water to form uranium oxide (UO2), hydrogen gas (H2) and uranium hydride (UH3). Different problems could appear during the storage due to these products: (i) UO2 and UH3 possess a lower mass density than metallic uranium, which can lead to stress in the encapsulated matrix thereby causing damage of the matrix, (ii) the production of H2 could produce an extra pressure deepening the stress problem, (iii) UH3 is a pyrophoric product that can ignite in contact with oxygen to form uranium oxide and hydrogen.

At this moment, only little information is available on the mechanism and the corrosion rate of metallic uranium in high pH conditions. This report summarises metallic uranium corrosion in the presence of water under various conditions.

4.2 Metallic uranium corrosion

4.2.1 Corrosion of metallic uranium in air or oxygen At low temperatures (representative of geological disposal conditions) the corrosion of metallic uranium can be considered as the same as the uranium-oxygen reaction (Ritchie, 1981).

In oxygen, the oxidation reaction of metallic uranium is given in equation (9) (Ritchie, 1981):

→ (9) with values of x in the range of 0.06 to 0.4 (Hilton, 2000). At a temperature higher than 275 °C, the formation of U3O8 is also observed (Colmenares, 1984).

Initially, the corrosion rate follows a parabolic law due to the formation of a thin and adherent oxide layer. This layer plays a diffusion-limiting role. Afterwards, the corrosion mechanism occurs in three steps: (i) oxygen gas is adsorbed on the uranium surface, (ii) ionisation of oxygen to form O2- and (iii) oxygen ions diffusion through the oxide layer to reach the metal- oxide interface, where it forms a new oxide (Colmenares, 1984; Hilton, 2000). While the oxide film thickness increases, tensile stresses lead to fractures and spalling-off of this film, limiting the thickness of the adherent film to approximately one micron. At this time, the corrosion rate becomes linear because the spalling-off is counterbalanced by the creation of a new oxide layer, and the thickness of the adherent oxide film stays consistent. The Arrhenius expression of the uranium corrosion rate in dry air, and in the temperature range 40 °C to 300 °C, is given by equation (10) (Hilton, 2000):

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 31 of 77 SCK•CEN/27548174 ISC: Public

. . (10) where R is the gas constant (J·mol-1·K-1) and T is the temperature (K).

This corrosion rate (equation (10)) is slightly dependent on the oxygen pressure up to 5.3 kPa. Above this threshold, it becomes independent of the oxygen pressure (Hilton, 2000).

4.2.2 Corrosion of metallic uranium in water vapour In water vapour, uranium corrodes at a higher rate than in dry air. Moreover, the oxidation mechanism and kinetics strongly depends on the oxygen concentration in the water vapour. Therefore, two distinct sections are presented: corrosion in aerobic and anaerobic conditions.

4.2.2.1 Aerobic conditions The oxidation reaction of metallic uranium in water vapour, in the presence of oxygen, is given in equation (11) (Hilton, 2000; McGillivray et al., 1994):

→ (11) where a and b are the fractional contributions of oxidant in the oxide from water vapour and oxygen, respectively. Values of x are in the range of 0.17 to 0.24 (Hilton, 2000).

Depending of the relative humidity (RH) of the system, the shape of the corrosion rate curve can be split in three regions (Figure 10): (i) between 0 and 2% RH, the rate increases with the water pressure, (ii) between 2 and 90% RH, the rate is approximately constant and (iii) above 90% RH, the rate increases rapidly until 100% RH. In the third region, the reaction is similar to the one observed in liquid water (see Section 4.2.3) and the reaction products change from

UO2+x to UO3 (yellow), with an accompanying evolution of hydrogen gas (McGillivray et al.,

1994). After further exposure, the yellow product (UO3∙0.8H2O (Baker et al., 1966b)) could become brown-black or green-black. This product is compact and adherent to the surface of the uranium. However, as for the corrosion in dry air, when the thickness of the oxide layer increases, stresses build up and spalling is likely to occur (Wilkinson, 1962).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 32 of 77 SCK•CEN/27548174 ISC: Public

Figure 10. Variation of the corrosion rate of uranium with RH in presence of water vapour and oxygen.

Curve A, B and C were recorded with different uranium batches. At P, significant amounts of yellow UO3 were observed (Baker et al., 1966b).

Both water and oxygen contribute to the oxidation process of metallic uranium. Between 80 °C and 90 °C, approximately 70 to 75% of the oxygen from the oxide layer comes from the water vapour, probably as OH-, while ~25-30% comes from the oxygen from the atmosphere, probably as O2- (McGillivray et al., 1994).

Depending on temperature and the RH, two corrosion rate equations can be derived (Hilton, 2000). Equation 12 gives the corrosion rate equation between 2-90% RH and at a temperature between 20-200 °C. Equation 13 gives the corrosion rate equation at 100% RH and at a temperature between 20-100 °C.

. . (12)

. . (13) where R is the gas constant (J·mol-1·K-1) and T is the temperature (K).

Corrosion rate equations show that the reaction rate is one order of magnitude higher in aerobic water vapour (100% RH) than in dry air.

4.2.2.2 Anaerobic conditions In anaerobic water vapour conditions, the general oxidation reaction of metallic uranium is given by equation (14):

→ (14)

Values of x are in the range between 0.0 and 0.2 (Baker et al., 1966a; Colmenares, 1984; Hilton, 2000).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 33 of 77 SCK•CEN/27548174 ISC: Public

The reaction kinetics of equation (14) is linear for temperatures up to 300 °C (Hilton, 2000). The corrosion rate equation can be derived (Hilton, 2000). Equation 15 gives the corrosion rate equation at 100% RH and at a temperature between 20-100 °C.

. . . . (15) where p is the vapour pressure, R is the gas constant (J·mol-1·K-1) and T is the temperature (K).

This corrosion rate is much higher than the one observed in dry oxygen (2-3 orders of magnitude) or aerobic water vapour conditions (1-2 orders of magnitude). This difference could be explained by the formation of a corrosion protective monolayer of oxygen atoms at the surface of the metal in aerobic conditions. In anaerobic conditions, such a layer is not formed and the corrosion rate is higher (Hilton, 2000).

When the amount of hydrogen gas is measured, a deficiency is observed compared to what is expected from equation (13) (Baker et al., 1966a; Hilton, 2000). This is due to the formation of a very thin interface composed of uranium hydrides (UH3) at the oxide-metal interface (Martin et al., 2016). This hydride is an intermediate product of equation (13). Indeed, during the corrosion of uranium, hydrogen radicals are produced (Baker et al., 1966b) and, due to their high reactivity, they can directly react with the metallic uranium to form uranium hydrides

(Tyfield, 1988). Finally, UH3 can react with water to produce uranium oxides (equation (16)):

→ (16)

This reaction is not complete as it reaches an equilibrium after 83% of completion (Baker et al.,

1966a), meaning that 17% of UH3 is still remaining at equilibrium. However, completion is approached if the temperature is increased up to 240 °C and the RH is decreased to 4% (Haschke, 1995).

The reaction kinetics of equation (16) is given in equation (17) (Haschke, 1995):

. (17) where R is the gas constant (J·mol-1·K-1) and T is the temperature (K).

It has to be mentioned that this kinetics can only be considered if a small amount of uranium hydride is produced. Indeed, if a large amount of UH3 comes into contact with water, a violent pyrophoric reaction may occur. Furthermore, UH3 can also react violently with oxygen to form uranium oxide, as mentioned in equation (18) (Solbrig et al., 1994):

→ (18)

4.2.3 Corrosion of metallic uranium in water The corrosion of metallic uranium in water is similar to the one in water vapour with a RH of 100%. Up to 300 °C, the corrosion kinetics is linear and the rate and reaction equations can be assumed being the same than in equations (10), (12), (13) and (14). However, the reaction product UO2+x is almost stoichiometric and x is lower than 0.1 (Hilton, 2000).

In the presence of oxygen and at temperatures below 70 °C, corrosion occurs but only after an induction period (Tyfield, 1988). This induction period is due to the adsorption of molecular oxygen, which leads to the formation of a layer at the surface of the metal, inhibiting the

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 34 of 77 SCK•CEN/27548174 ISC: Public

uranium-water reaction. After 42 hours at 70 °C in aerated water, the measured corrosion rate was 40 times less than the one measured in anaerobic conditions (Hilton, 2000). However, after a few hundreds of hours, this inhibition disappears and the corrosion rate in both aerobic and anaerobic conditions are the same (Hilton, 2000).

Such as for the corrosion of uranium in water vapour, the hydrogen production is not stoichiometric in regard to equation (14) and only 70 to 95% of the hydrogen is detected, partly due to the production of UH3 and the non-stoichiometry of the UO2+x (x < 0.1) (Hilton, 2000; Tyfield, 1988). However, when accumulating, this hydrogen does not influence the corrosion rate up to 6 atm overpressure (Baker et al., 1966a). If the overpressure still increases, hydrogen can be incorporated in the oxide film, resulting in an increase of the film conductivity and consequently an increase of the corrosion rate (Wanklyn and Jones, 1962).

In contact with water in anaerobic conditions, metallic uranium undergoes a rapid darkening due to the production of an amorphous black powder. When in contact with oxygen, Ward et al. (Ward and Waber, 1962) showed that this amorphous powder transforms into UO2.2 and

UH3, while Haschke (Haschke, 1995) showed the formation of UO1.2H0.6. This corrosion is uniform and no pitting holes are present. However, after a prolonged corrosion, a rough surface is obtained due to the spalling of the oxide layer (Haschke, 1995).

Atomic Probe Tomography (APT) investigation of uranium exposed to air showed the presence of hydroxyl species within the oxide layer (Martin et al., 2016). From this, the authors suggested that the corrosion mechanism is driven by the diffusion of these hydroxyl species through the oxide layer.

4.2.4 Mechanism of uranium hydride formation Many researches focused on the uranium hydride formation. Martin et al. (Martin et al., 2016) used APT to analyse the oxide layer formed during corrosion of metallic uranium. During the oxidation reaction, a hydride layer of 3-5 nm thick is formed at the interface the interface between the metal and the oxide layer. This was revealed by using both normal and deuterated water. Other authors showed the same results by using a Focused Ion Beam (FIB) milling with a HR-SEM, coupled with Electron Backscattering Diffraction (EBSD) (Jones et al., 2013). Moreover, hydrides are preferentially formed at low energy nucleation sites, such as grain boundaries and inclusion sites. This was proven by Harker et al. (Harker et al., 2013), who studied the hydride-forming behaviour on metallic uranium containing inclusions of carbonitrides. It was clear that hydride grew almost exclusively around exposed inclusions on mechanically polished samples. However, the use of an electropolishing step decreased the number of hydride growth sites by a factor 150000. The surface preparation method is then extremely important on the hydride formation. The surface topography of uranium has also a huge importance on the hydride formation. Stitt et al. (Stitt et al., 2015c) induced irregularities at the surface of the uranium metal. If oxidation is realised at 220 °C and 0.5 bar in deuterated water, deuterated uranium hydride (UD3) nucleated earlier on steeper defects than on more gently sloping sides. Consequently, the oxide surface is pierced earlier and the corrosion rate increases rapidly. This is important for the nuclear industry as the decladding uranium surface is irregular and mechanically strained. Finally, at lower temperature, relevant to nuclear waste storage (between 30 °C and 200 °C), the UH3 phase composition has a strong dependence on the temperature (Orr et al., 2016), increasing the -UH3 fraction at lower temperature. However, the hydride morphology showed nearly no variation with preparation conditions.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 35 of 77 SCK•CEN/27548174 ISC: Public

4.2.5 Influence of irradiation and water radiolysis The corrosion rate of irradiated uranium is greater than the one of the unirradiated samples (Hilton, 2000). However, this increase is due to an increase of the surface area of the uranium sample as a result of swelling and void formation. The corrosion mechanism is not affected by the sample irradiation.

The effect of water radiolysis was also studied. Baker et al. (Baker et al., 1966a) studied uranium corrosion in water under -irradiation from a 60Co source at 30 °C. The corrosion rate was not enhanced even if the sample was exposed to ~72 Mrad. However, water radiolysis could produce reactive species such as free radicals, hydrogen peroxides, etc., which could increase the kinetics of metallic uranium corrosion (Walters, 1997). Indeed, Dong (Dong and Vandegrift, 1996) showed that it is possible to dissolve metallic uranium in alkaline hydrogen peroxide solution. They showed that the dissolution rate of metallic uranium foil increased with the alkaline concentration to reach a maximum at 1.5 M sodium hydroxide. Then it decreased as the alkaline concentration is further increased. An increase of the dissolution rate was also observed with an increase of the hydrogen peroxide concentration. By using a solution of 1M

NaOH/4M H2O2 at 40-65 °C, a metallic uranium foil of 0.2 g is almost completely dissolved within an hour.

4.2.6 Influence of the counter ions and the pH on the corrosion mechanism Baker et al. (Baker et al., 1966a) studied the corrosion rate in different acids (sulphuric, nitric and acetic acids) and bases (potassium and sodium hydroxides and aniline). At a fixed pH, corrosion rates are identical, even if the counter ion is different. This means that the corrosion rate of metallic uranium is determined by the balance of hydrogen ions/hydroxide ions. Other compounds present in the solution have no influence, except for carbonates. When carbonates are present, the corrosion rate decreases to about 50%. This is believed to be due to the formation of uranyl carbonates (Baker et al., 1966a).

Little research was conducted on the influence of the pH on the corrosion rate of metallic uranium in water. Baker et al. (Baker et al., 1966a) showed results for the corrosion of uranium immersed in aqueous media at different pH at 100 °C (Figure 11). The corrosion rate was nearly independent of pH between the values of 3 and 7. Between 7 and 13.5, a constant slight decrease of 10 to 15% is observed. Finally, the slowest corrosion rates are obtained at pH lower than 2. Indeed, when the pH decreases from pH 3 to pH 2 , the corrosion rate decreased by a factor of about 10.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 36 of 77 SCK•CEN/27548174 ISC: Public

Figure 11. Evolution of the corrosion rate in function of the pH of the solution at 100 °C (Baker et al., 1966a).

4.2.7 Electrochemical uranium corrosion mechanism in water When immersed in water, metallic uranium is expected to corrode. Initially, just after immersion of the electrode in de-aerated water, a potential of -0.54 V vs. SHE (Standard Hydrogen Electrode) was measured (Ward and Waber, 1962). Looking at the Pourbaix diagrams (Figure

12), this potential lies in the UO2 domain. However, over a period of three hours, the open circuit potential (OCP) decreased to reach a steady-state value of -1.34 V vs. SHE, which is at the UO2/UH3 boundary in the Pourbaix diagram. Indeed, the production of hydrogen by the reduction of water happens below -0.4 V to -0.8 V vs. SHE at pH 10 to 14, respectively. Ward et al. (Ward and Waber, 1962) studied the effect of the addition of different salts to the solution. No change of the OCP was found. However, changing the H+ or OH- concentration caused competing side reactions to occur.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 37 of 77 SCK•CEN/27548174 ISC: Public

Figure 12. Potential-pH diagram of uranium (Pourbaix, 1974).

The anodic reaction is:

° → (19)

While the cathodic reaction, in de-aerated conditions, is the reduction of water:

→ (20)

And, in aerated conditions, the cathodic reaction is:

→ (21)

Note that hydrogen peroxide was observed to be an intermediate species (Ward and Waber, 1962). The oxygen was reduced into hydrogen peroxide during the passivation of the electrode. This phenomenon was pointed out due to the presence of a higher redox exchange current than the passive corrosion current. The metal potential was also slightly more active than the reversible redox potential. Churchill (1939) already showed that hydrogen peroxide could be produced during the corrosion of many metals due to the oxidation of atomic hydrogen by oxygen.

Ward (Ward and Waber, 1962) investigated the corrosion potentials in aerated conditions at pH 7. In aerated 0.1 M KClO4 solutions (T = 35 °C), corrosion potentials of -0.1 V vs. SHE were measured. Increasing the pH to 14 in air equilibrated solutions decreased the OCP to -0.64 V vs. SHE (Bullock et al., 1974) and no hydrogen evolved at these potentials.

In de-aerated conditions, the OCP decreased with the decrease of the oxygen partial pressure (Jenks, 1971) to reach -1.34 V vs. SHE in fully de-aerated water (Ward and Waber, 1962). It was also shown that the OCP varied with the pH of the water solution (Bullock et al., 1974; Jenks, 1971; Ward and Waber, 1962). The OCP decreased linearly with increasing pH according to the Nernst equation:

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 38 of 77 SCK•CEN/27548174 ISC: Public

. (22)

Finally, the nature of the oxide layer formed on the surface of metallic uranium in water at pH 14 depends on the applied potential. When the applied potential is more negative than -0.24 V vs. SHE, the oxide layer is composed exclusively of UO2, while UO2, U3O8, U3O7 and probably

U(OH)4 are present if the applied potential is higher than -0.24 V vs. SHE (Bullock et al., 1974; Leach and Nehru, 1964).

The polarisation curve of metallic uranium was investigated in de-aerated KOH solution at pH 13.7 (Figure 13) (Bullock et al., 1974). The polarisation curve of metallic uranium revealed two regions of active-passive behaviour, such as for steels. Between -0.54 V and -0.14 V vs. SHE, the observed peak is attributed to the oxidation of UO2 into UO3. Between -1.34 V and -0.64 V vs. SHE, the observed peak is attributed to the oxidation of U(IV) species in solution produced by the UO2 chemical dissolution. While the first peak is quite reproducible, the second one is more sensitive to the solution composition and the uranium surface preparation method.

Figure 13. Polarisation curves for metallic uranium in KOH solution at pH 13.7 (Bullock et al., 1974).

4.2.8 Corrosion of metallic uranium in cement-based materials Corrosion rate data found in the literature of metallic uranium in a cementitious environment is scarce.

Nuttall (Nuttall and Curwen, 1990) studied the corrosion of uranium in (unsaturated) calcium hydroxide (pH = 11 to 13) and in BFS/OPC material by means of electrochemical methods. In

Ca(OH)2, a free corrosion potential of -0.49 V vs. SHE was measured, while a value of -0.39 V vs. SHE was found in BFS/OPC. According to the Pourbaix diagrams, these corrosion potentials should lead to the formation of UO2, UO3 and UH3. Moreover, a corrosion rate of around 100 µm/y at a pH varying from 11 to 13 was calculated.

The corrosion rate of uranium in an aerobic cementitious grout was also electrochemically investigated by Thomas and Naish (Thomas et al., 1994). Initially, a corrosion rate of 60 µm/y was measured. However, the corrosion rate rapidly decreased in time and reached a value of 12 µm/y after only 2 months.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 39 of 77 SCK•CEN/27548174 ISC: Public

Curwen (Curwen, 1994) investigated the corrosion rate of uranium in packaged waste from the Sellafield Drypack Process plant. He concluded that the corrosion rate may be driven by oxygen rather than water reduction, until anaerobic conditions arise. He observed an influence of water on the corrosion rate in BNFL Big Box package (Curwen and James, 1995). While a corrosion rate of 150 µm/y was measured at the beginning of the experiment, it decreased rapidly to values below the detection limit. This was explained by the limited water supply at the surface of the encapsulated uranium. However, in presence of enough water supply such low corrosion rates are not expected to persist.

Electrochemical techniques were also used by Blackwood (Blackwood and Farmilo, 1996) and Farmilo (Farmilo and Simmons, 1997) to measure the corrosion rate of the metallic uranium in a 3:1 PFA/OPC cement grout. In these studies, the corrosion rate was also found to depend on the availability of water at the metallic uranium surface. Indeed, while the corrosion rate fell below 0.1 µm/y after the initiation rate, irrigation of the grout led to an increase of this corrosion rate to 5 µm/y at room temperature. Moreover, increasing the temperature, increased also the uranium corrosion rate. Thus, a value of 160 µm/y was obtained for tests realised at 80 °C.

An effect of temperature on metallic uranium encapsulated in BFS/OPC cement grout with different BFS/OPC ratio and water content was also analysed by Hayes and Brogden (Hayes et al., 2003). At 25 °C, a corrosion rate of 60-150 µm/y was measured, while the corrosion rate increased to 250-580 µm/y at 35 °C. Godfrey and Curwen (Godfrey et al., 2004) concluded that the corrosion rate of uranium in such BFS/OPC cement grout was similar to the one measured in water.

Delegard (Delegard and Schmidt, 2009) studied the influence of different grouts on the corrosion rate of uranium by attempting to limit water access at the surface of uranium. Four different Portland cement-based grouts and two magnesium phosphate cement-based grouts were tested in this study. The grout composition was found to have an influence on the corrosion rate. However, all rates were close to the rate observed in anoxic water. The lowest rate was obtained for Portland cement containing bentonite, where the corrosion rate was found to decrease by a factor 2 compared to the corrosion rate observed for the metal alone. These results were explained by an insufficient sequestration or bounding of water by the grout to prevent reaction with the uranium metal. Finally, they concluded that grouts might only decrease the metallic uranium corrosion rate by altering the metal/oxide surface or by decreasing the water vapour pressure.

Finally, more recent research, conducted by Jones et al. (Jones et al., 2016), showed the possibility to analyse the corrosion products formed at the surface of metallic uranium encapsulated in grout without destruction of this grout. This was realised by absorption contrast radiography, using a single pulse exposure from X-ray source driven by a high-power laser. This team also used synchrotron X-rays to perform micro-scale in-situ observation and characterisation of uranium encapsulated in grout (Stitt et al., 2015a; Stitt et al., 2015b). In the first research (Stitt et al., 2015b), metallic uranium was encapsulated in a 3:1 mixture of blast furnace slag (BFS) and ordinary Portland cement (OPC) with a water/cement ratio of 0.4. The early stage of uranium corrosion in grout was examined after 3 days in a moist atmosphere and 1 week in atmospheric atmosphere. Other samples were further dried and exposed to hydrogen to evaluate the effect of hydrogen on the corrosion behaviour. When encapsulated uranium was exposed to water, the corrosion rate followed the anoxic water oxidation regime

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 40 of 77 SCK•CEN/27548174 ISC: Public

to form uranium dioxide, which was also confirmed during the study of Delegard (Delegard and Schmidt, 2009). This oxidation led to the development of strain in the uranium through a fast growing corrosion layer. Moreover, grout seemed to promote hydride growth in selective spots instead of uniform hydride growth on a bare uranium surface.

The second study analysed the oxidation of uranium and uranium hydride metallic uranium grouted in a BFS/OPC 3:1 cement-based material stored in de-ionised water for 10 months (Stitt et al., 2015a). Uranium hydride was deliberately formed on the surface of the sample before encapsulation. Initially, the rate of hydride oxidation was rapid. However, it slowed down rapidly. After only two weeks, only 15-20% of the hydride had reacted and hydrides were still present after 10 months. These results showed that artificial -UH3 persists during storage in water for at least 10 months at room temperature. This also proved that UH3 is less susceptible to rapid and continuous oxidation in grout than in air. This lowering of the oxidation rate was attributed to the barrier properties of the grout mixtures against water diffusion and the highly reducing conditions created by the BFS (Macphee et al.). On the contrary, corrosion products would not be able to diffuse through the grout. Then the hydrogen produced during corrosion would be confined at the uranium-grout surface. However, tomographic reconstructions also indicate a loss of corrosion products into the pore volume of the grout. However, this seemed to be more due to vibrations during transportation of the system uranium/grout than because of uranium oxidation.

4.2.9 Summary of the corrosion rates for uranium in alkaline conditions In this section, the influence of pH on the corrosion rate is compiled from various authors (Table 2). This table confirms that an increase in pH (in alkaline conditions) slightly decreases the corrosion rate of the metallic uranium (cfr. Figure 12). This is further illustrated in Figure 14, which shows a graph of the corrosion rate as a function of pH. The checkered rectangle embraces the data for which the pH is not specified in Table 2 but is estimated to be between 11 and 13.

Table 2. Overview of corrosion rates of metallic uranium obtained in highly alkaline, anoxic conditions

pH T (°C) Medium Corrosion rate (µm/y) Reference ~7 25 °C Distilled water 175 (Haschke, 1995) ~7 25 °C Sea water 6 (Haschke, 1995)

11-13 RT Ca(OH)2 ~100 (Nuttall and Curwen, 1990) n.i. RT BFS/OPC ~100 (Nuttall and Curwen, 1990) n.i. RT BNFL BigBox 150 (Curwen, 1994) n.i. RT aerobic cementitious grout 60 (Thomas et al., 1994) n.i. RT aerobic cementitious grout 12 (after 2 months) (Thomas et al., 1994) n.i. RT 3:1 PFA/OPC 5 (Blackwood and Farmilo, 1996) n.i. RT 3:1 PFA/OPC <0.1 (after initiation rate) (Blackwood and Farmilo, 1996) n.i. 80 °C 3:1 PFA/OPC 160 (Blackwood and Farmilo, 1996) n.i. 25 °C 3:1 BFS/OPC 60-150 (Hayes et al., 2003) n.i. 35 °C 3:1 BFS/OPC 250-580 (Hayes et al., 2003) n.i. = not indicated

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 41 of 77 SCK•CEN/27548174 ISC: Public

Figure 14. Corrosion rates of metallic uranium in highly alkaline, anoxic conditions, as a function of pH at 25 °C.

4.3 Conclusions In contact with water, metallic uranium corrodes to form uranium oxide and hydrogen, with the formation of uranium hydride as an intermediate compound.

The corrosion rate of metallic uranium is higher in water than in air. However, the presence of oxygen in water leads to the formation of a protective oxide film at the surface of the metal, resulting in a reduction of the corrosion rate by a factor of 40. However, after only a few hundreds of hours, this inhibition disappears and the corrosion rate in both aerobic and anaerobic conditions is the same.

The formation of the pyrophoric uranium hydride happened at the interface between metallic uranium and uranium oxide to reach a thickness of 3-5 nm. Moreover, this UH3 formation seems to form preferentially at grain boundaries or inclusion sites.

No influence of irradiation was observed on the corrosion rate. However, water radiolysis could create H2O2, which is able to dissolve uranium in water.

Increasing the pH of the water solution from 7 to 13.5 decreases the corrosion rate by only 10- 15%, while at pH 2-3, the corrosion rate is decreased by a factor of 10.

In cement-based materials, initial corrosion rates of 60-580 µm/y were obtained. However, the corrosion rate rapidly dropped to 5-12 µm after only a few weeks. Moreover, values even lower than 0.1 µm/y have been measured if the cement-based material is no longer saturated with water.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 42 of 77 SCK•CEN/27548174 ISC: Public

5 Corrosion of beryllium

5.1 Introduction Beryllium has a high specific heat and thermal conductivity, and high specific strength and stiffness. It is an attractive engineering material for nuclear applications because of its low neutron cross section. Inconveniences however are its toxicity in particulate form (leading to berylliosis and chronic beryllium disease), its low fracture toughness, and its high cost.

The literature on beryllium corrosion in highly alkaline, anoxic conditions is scarce. Few data on the corrosion rate in these conditions are available. In this chapter we will focus on the different forms of beryllium corrosion and propose a mechanism and an initial corrosion rate in highly alkaline conditions.

5.2 Mechanism of beryllium corrosion In air, beryllium is readily oxidized and forms a protective oxide layer, much like other metals such as aluminum, titanium, zirconium, etc. In fact, the corrosion behaviour of beryllium resembles that of aluminum, as can be seen in Figure 15, which shows the thermodynamic behaviour of beryllium as a function of potential and pH. Between pH 4 and 10.7, the formation of the insoluble species Be(OH)2 is thermodynamically favourable. This species, or its hydrate

BeO۰H2O, likely accounts for the observed passivity of beryllium in this pH range. In highly acid or alkaline solutions however, corrosion occurs.

Figure 15. Equilibrium potential – pH diagram for the beryllium/water system at 25 °C (Pourbaix, 1974).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 43 of 77 SCK•CEN/27548174 ISC: Public

5.2.1 Corrosion mechanisms of beryllium: general remarks In the pH range 2-12.5, the main corrosion mechanism of beryllium is uniform corrosion through passive dissolution. In this pH range, beryllium metal is covered by a protective oxide film. The basic reactions are given by:

→ (23)

→ (24)

In the presence of halides, pitting corrosion can occur. Chloride and sulfate ions are the most critical contaminants in aqueous corrosion. The pitting of beryllium is mainly due to the presence of impurities in the base metal (Punni and Cox, 2010). Indeed, pitting occurs as the result of a galvanic cell between the impurities and beryllium (Haws, 2005). Because beryllium is anodic to all common metals except magnesium, zinc and manganese, the beryllium area surrounding the impurities (other than Mg, Zn and Mn) is rapidly attacked, creating a pit, while the impurity is cathodically protected.

Galvanic corrosion can occur when beryllium is in contact with other metals, such as e.g. stainless steel in the case of the BR2 beryllium. In that case, the stainless steel is cathodically protected and the corrosion of the beryllium will accelerate. Galvanic corrosion is also possible in welded beryllium.

5.2.2 Corrosion in highly alkaline, anoxic conditions The literature on the corrosion of beryllium at alkaline pH values is scarce. Figure 16 shows the polarization curves of beryllium in acidic and alkaline media (Gulbrandsen and Johansen, 1994). At high pH (lower half of the figure), the polarization curve has the typical shape for a metal prone to pitting, with the pitting potential (i.e. the potential value at which the current density sharply increases due to the onset of pitting) decreasing with increasing pH.

Figure 16. Potentiodynamic polarization curves (scan rate 1 mV/s) for beryllium in solutions of different pH (Gulbrandsen and Johansen, 1994).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 44 of 77 SCK•CEN/27548174 ISC: Public

Figure 17 shows the electrochemically quasi-steady state passive current density of beryllium as a function of pH for various electrolyte solutions (Gulbrandsen and Johansen, 1994). It is clear that in alkaline solutions, the corrosion rate (which is proportionate to the current density) increases exponentially with pH. The open circuit potential decreases approximately 0.13 V per unit increase in pH in alkaline NaOH solutions, reaching a value of ca -1.7 V (SSE, saturated sulfate electrode) at pH 14 according to Gulbrandsen. This corresponds to approximately -1 V versus the standard hydrogen electrode (SHE), which falls well into the passive region and indicates that the risk of pitting is very low.

Figure 17. Quasi-steady state passive current density (jp) for beryllium electrodes as a function of pH (Gulbrandsen and Johansen, 1994).

5.3 Kinetics of beryllium corrosion in alkaline solutions The main source of kinetic data for the corrosion of beryllium in alkaline solutions is Figure 17 reproduced from Gulbrandsen (Gulbrandsen and Johansen, 1994). From this figure, corrosion rates are derived from current densities by the following equation

1 mA/cm² = 3.28 M/nd (expressed in mm/y) (25) where M is the atomic mass in g, n is the number of electrons exchanged in the corrosion reaction, and d is the density in g/cm³. The corrosion rates given in Table 3 cannot be considered as exact numbers, but rather provide an order of magnitude. From Table 3 it is clear that corrosion rates are rather high in the higher pH range. These values however will have to be confirmed by independent tests. The actual corrosion rates in a cementitious matrix will probably be lower due to mass transport restrictions; therefore the values reported here are overestimated. Another source for kinetic data is the work by Hill et al (Hill et al., 1996, 1997). They found that the passive current density exhibits a minimum in the pH range between 7

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 45 of 77 SCK•CEN/27548174 ISC: Public

and 9, while Gulbrandsen places this minimum in the vicinity of pH 11. Figure 18 shows a plot of the corrosion rates versus pH, omitting the obvious outliers at pH 12 and 15, together with the best linear fit. This yields the following relationship between corrosion rate and pH under alkaline conditions:

. . . (26) where rcorr is the corrosion rate in µm/y. This equation is only valid in the pH range 9-14.

Table 3. Approximate corrosion rates for beryllium in alkaline solutions.

pH Corrosion rate (µm/y) Reference 9 1 (Gulbrandsen and Johansen, 1994) 10 1 (Gulbrandsen and Johansen, 1994) 12 0.5 (Gulbrandsen and Johansen, 1994) 13 5 (Gulbrandsen and Johansen, 1994) 14 10 (Gulbrandsen and Johansen, 1994) 15 500 (Gulbrandsen and Johansen, 1994) 10.7 3 (Hill et al., 1996) 12.5 6 (Hill et al., 1996)

Equation y = a + b*x 8 Plot Corrosion rate Weight No Weighting Intercept -14.99893 Slope 1.67621 Residual Sum of Squares 6.97979 Pearson's r 0.93934 R-Square(COD) 0.88236 6 Adj. R-Square 0.85295

4 Corrosionrate (µm/y)

0 9 1011121314 pH

Figure 18. Plot of corrosion rate as a function of pH for beryllium in alkaline solutions (omitting outliers at pH 12 and pH 15).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 46 of 77 SCK•CEN/27548174 ISC: Public

5.4 Conclusions The literature data on beryllium corrosion rates in highly alkaline, anoxic conditions is scarce. The corrosion rate seems to be a linear function of pH. In the pH range between 9 and 13, the corrosion rate is in the order of magnitude of several µm/y.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 47 of 77 SCK•CEN/27548174 ISC: Public

6 Corrosion of aluminium

6.1 Introduction Although aluminium and aluminium alloys are very reactive materials, they have found application in the building construction domains, such as windows and door frames, roof structures, façade claddings,… The main reason for this wide range of applications is the formation of a very adherent and protective oxide and hydroxide film when they are in contact with water (Endtinger and Weber, 1967).

Aluminium has also been used as cladding for nuclear fuels and targets due to their low thermal neutron absorption cross-section behaviour. After irradiation, aluminium and aluminium claddings are stored in water-filled basins in order to cool down before being encapsulated within cementitious grout drums.

However, the oxide layer formed on its surface is amphoteric and so, not resistant to corrosion in acidic and alkaline conditions, because the protected film is known to dissolve (Deltombe and Pourbaix, 1958).

For these reasons, the corrosion behaviour of aluminium in alkaline conditions has been investigated. Here below, a literature review of the aluminium corrosion in water and in cementitious grout is summarised.

6.2 Corrosion of metallic aluminium in alkaline conditions

6.2.1 Corrosion in water Metallic aluminium is very reactive to water or air, forming a passive oxide and hydroxide film at the surface of the metal (Endtinger and Weber, 1967). This layer is protective against corrosion at pH between 4 and 8.5. Below and above those values, the protective layer dissolves and the metal is activated (Endtinger and Weber, 1967; Pourbaix, 1974; Szklarska-Smialowska, 1986).

When metallic aluminium is immersed in alkaline solutions, aluminium is oxidised following the anodic reaction presented in equation (27) (Bard et al., 1985):

° → (27)

Then, aluminium cations react with hydroxide ions to form aluminium hydroxides according to equation (28):

→ (28)

However, as mentioned earlier, in alkaline condition, this hydroxide layer dissolves to form soluble aluminate ions (equation (29)):

→ Deltombe and Pourbaix, 1958 (29)

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 48 of 77 SCK•CEN/27548174 ISC: Public

To counterbalance the anodic reaction, the reduction of water occurs as the cathodic reaction, which takes place at the aluminium surface (equation (30)) (Bard et al., 1985):

→ (30)

However, in oxic conditions, the anodic reaction can also be counterbalanced by the reduction of oxygen (equation (31)):

→ (31)

Finally, equation (32) gives the overall reaction of the aluminium dissolution in alkaline solution:

° → (32)

Pryor (Pryor and Keir, 1957) confirmed that the cathodic reaction is both the oxygen and water reduction by analysing the increase of the aluminium corrosion rate by increasing the amount of dissolved oxygen in water solutions with a pH varying from 2 to 10. Godard (Godard, 1967) showed that the corrosion rate of metallic aluminium was significantly reduced when oxygen was entirely eliminated from the test solutions. The absence of oxygen also decreased the risk of pitting corrosion because in the absence of oxygen, the metal surface is not polarised to its pitting potential (Szklarska-Smialowska, 1986). Finally, Guo et al. (Guo et al., 2016) also found that, for the corrosion of aluminium alloy 1100 in water containing boric acid, sodium borate and sodium hydroxide at a pH of 8.2, the cathodic reaction is dominated by the oxygen reduction below 55 °C, while the reduction of water is more pronounced at higher temperature.

6.2.1.1 Influence of pH on the corrosion rate of metallic aluminium Due to the amphoteric behaviour of the passive film present at the surface of the metallic aluminium, the corrosion rate of aluminium passes through a minimum at pH 5 (Pourbaix, 1974; Pryor and Keir, 1958; Vujičić and Lovreček, 1985), when the solubility of the passive layer is minimum (Figure 19).

Figure 19. Solubility of aluminium oxides in function of pH (Pourbaix, 1974).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 49 of 77 SCK•CEN/27548174 ISC: Public

Increasing the pH leads to an increase of the corrosion rate (Tabrizi et al., 1991). Indeed, there is a relationship between the corrosion rate and the hydroxide concentration (equation (33)) (Mansour et al., 1992):

. (33) where k is the corrosion rate, K is a constant, C is the hydroxide concentration and n is a value, which increases with increasing temperature.

The corrosion of aluminium in alkaline solutions was mainly studied in sodium, potassium and/or calcium hydroxides. At low concentrations (0.5 – 1 N), the corrosion rate is proportional to the cube root of the concentration for NaOH and KOH, while it is proportional to the square root of the concentration for Ca(OH)2. At higher concentrations, the corrosion rate is directly proportional to the concentration (Straumanis and Brakšs, 1949a, b). McKee (McKee and Brown, 1947) studied the corrosion of aluminium in NaOH solutions at different pH in lab conditions (room temperature under air) during 48 hours. They found that the corrosion rate increased logarithmically to reach ~25 µm/y at pH 10, ~250 µm/y at pH 11, ~5500 µm/y at pH 12 and ~35000 µm/y at pH 13.

More recently, Prabhu (Prabhu and Rao, 2017) evaluated the influence of both pH (12.7 to 13.7) and temperature (30 °C to 50 °C) in NaOH solutions in oxic conditions on freshly polished (mirror surface) 6063 aluminium alloy. This was realised by using electrochemical measurements such as Electrochemical Impedance Spectrometry (10 mV signal over a frequency range of 100kHz – 0.01 Hz) and polarisation studies (0.01 V/s between -250 mV and 250 mV vs. OCP) allowing to obtain information on the initial corrosion rate, measured during the first hour. He found even higher corrosion rates than McKee (McKee and Brown, 1947): at a pH of 12.7, 13.4 and 13.7, corrosion rates in the range of 44000-65000 µm/y, 228000-361000 µm/y and 355000-496000 µm/y were measured, respectively.

The corrosion rate also depends on the counter-ion. McKee (McKee and Brown, 1947) showed that aluminium is more corrosion resistant in ammonia solutions than in potassium or sodium hydroxide solutions possessing the same pH. In addition, the corrosion rate is higher in NaOH than in KOH at the same pH (Godard, 1967; Straumanis and Brakšs, 1949a).

Long-term (up to 80 days) corrosion rate measurements were also realised, using pure commercial aluminium in sodium hydroxide (Tabrizi et al., 1991) at 30 °C and 60 °C in oxic conditions. It seems that at pH 10 and 11, the corrosion rate decreases slightly as a function of the immersion time. However, no data is provided at pH 12 due to the high corrosion rate.

Takatani et al. (Takatani et al., 1982, 1983) studied the corrosion rate of different aluminium alloys in saturated Ca(OH)2 (pH ~12.5) at 30 °C in air. The same behaviour than the one observed in sodium hydroxide solutions is observed; the corrosion rate decreases with immersion time. After 40 days of immersion, the calculated corrosion rate is ~22 µm/y.

In a study performed by Cook et al. (Cook et al., 1989), different solutions were used to analyse the corrosion rate. Pore water coming from OPC, 3:1 PFA/OPC and 9:1 BFS/OPC and simulated pore water composed of a mixture of Ca(OH)2 and NaOH in the pH range between 10 and 14 were tested. As expected, the corrosion rate increased by increasing the pH, irrespective of the composition of the solution. Moreover, for saturated Ca(OH)2 solution, 3:1 PFA/OPC and 9:1 BFS/OPC pore water solutions, the corrosion rate decreased as a function of the immersion

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 50 of 77 SCK•CEN/27548174 ISC: Public

time. Authors explained this reduction by the creation of a corrosion product film at the surface of the sample.

6.2.1.2 Influence of temperature on the corrosion rate of metallic aluminium Increasing the temperature leads to an increase of the aluminium corrosion rate. This is due to the faster dissolution of the oxide formed on the surface at higher temperature (Mansour et al., 1992). Tabrizi et al. (Tabrizi et al., 1991) showed that after 40 days exposure in sodium hydroxide at pH 10, the aluminium corrosion rate increased from ~50 µm/y to ~100 µm/y when the temperature was increased from 30 °C to 60 °C, respectively. A similar result was observed at pH 11: the corrosion rate increased from ~100 µm/y to ~300 µm/y for the same temperature increase (i.e. from 30 °C to 60 °C).

6.2.1.3 Influence of impurities present in the test solution on the corrosion rate of metallic aluminium The influence of the addition of different compounds to the test solutions on the corrosion rate of metallic aluminium was also studied. In pH neutral solutions, some compounds are found to be able to penetrate the oxide film and consequently causing pitting corrosion. Halide ions, in particular chloride, are able to break down the passive oxide film and therefore inhibit repassivation (Sverepa, 1958). Bicarbonate ions are also aggressive ions with respect to pitting corrosion. If they are alone in solution, they do not cause corrosion. However, in presence of chlorides or copper ions, the pitting rate is enhanced (Godard, 1967, 1979; ROWE and M.S., 1961). On the contrary, the presence of sulphate ions seems to reduce the corrosion rate of aluminium (Pathak and Godard, 1968), as well as oxygen, as mentioned earlier at the end of Section 6.2.1.

In alkaline conditions, Onuchukwu (Onuchukwu, 1988) observed that di-sodium tetraborate (Borax) inhibited partly the corrosion of aluminium in 0.5 M KOH solution. Moreover, the efficiency of this additive increased with increasing time.

6.2.1.4 Nature of corrosion products formed in alkaline solutions The nature of the corrosion products formed during the corrosion of metallic aluminium depends on pH, duration, temperature and composition of the solution (Godard, 1967; Tabrizi et al., 1991). Based on the Pourbaix diagrams (Figure 20), the corrosion products in sodium or potassium hydroxide should be composed of aluminium hydroxide gel. This gel is then transformed in pseudo-boehmite, boehmite, bayerite and/or hydrargillite (Pourbaix, 1974). Below pH 10, the oxide film formed is composed of a thin pseudo-boehmite barrier layer at the metal-oxide interface, while a bayerite layer is present at the outer surface (Tabrizi et al., 1991). Walkner et al. (Walkner et al., 2016) also showed that at pH 11, the aluminium oxide layer present at the surface of the metal, formed in atmospheric conditions before the study, does not protect the metal against further corrosion. When immersed in such solution, the oxide film is dissolved to form a thick non-protective hydroxide film at the aluminium surface. When the pH is higher than 12, the solubility of aluminium hydroxide increases so that metallic aluminium is not entirely covered by a protective film (Tabrizi et al., 1991).

In Ca(OH)2 solutions, the film is mainly composed of aluminium hydroxide and calcium aluminate (Cook et al., 1989). Walton (Walton et al., 1957) identified calcium aluminate as

3CaO.Al2O3.8-12H2O.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 51 of 77 SCK•CEN/27548174 ISC: Public

Figure 20. Potential-pH diagram of aluminum (Pourbaix, 1974).

6.2.2 Corrosion of metallic aluminium encapsulated in cement-based materials

6.2.2.1 Corrosion rate Initially, aluminium corrodes in cement-based materials at a fairly high rate. Depending on the cementitious material, it can fluctuate from 1600 to 22000 µm/y (Lee and Wilding, 1989). However, the corrosion rate decreases very rapidly to reach values of 16 to 800 µm/y after a few tens of days. This decrease can be attributed to the depletion of water at the surface of aluminium or by passivation (White, 1987). The lowest reported values are obtained for 9:1 BFS/OPC and 9:1 PFA/OPC materials due to their lower pH than OPC cements. So, increasing the quantity of OPC increases the corrosion rate of aluminium.

Other studies showed that the corrosion rate decreases even more in time and reaches values of 0.1 to 0.5 µm/y after less than two years (Thomas and Naish, 1994), while the initial corrosion rate varies between 1000 and 13000 µm/y.

Kinoshita et al. (Kinoshita et al., 2013) studied the corrosion of aluminium in OPC and calcium aluminate cement (CAC) based materials. The generation of hydrogen gas was observed in all studied media. The H2 concentration was measured after 7 and 28 days. The 3:1 BFS/OPC system produced a huge amount of hydrogen gas during the first 7 days (2.07 mL/cm²/day), while only 0.02 mL/cm²/day was produced in the period from 7 to 28 days. The authors suggested that this corrosion rate reduction is caused by a depletion of hydroxide ions in the local environment close to the aluminium surface due to the initial intensive corrosion. For the PFA/CAC system, the hydrogen gas production slightly increases from 0.14 mL/cm²/day after 7 days to 0.20 mL/cm²/day after 7 to 28 days. In such a system, the hydroxide ion concentration

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 52 of 77 SCK•CEN/27548174 ISC: Public

in the surroundings of the aluminium surface remains high enough for aluminium corrosion, even after a certain amount of aluminium had reacted. This should due to the formation of

Al(OH)3, Ca3Al2(OH)12 and water from the metastable CaAl2O14H10 and Ca2Al2O13H16 phases present in the PFA/CAC matrix.

Long-term experiments, lasting for several years, were also performed using building materials such as cements, mortars or concretes. Low corrosion rates were reported, as corrosion mostly happened directly after encapsulation, when the concrete was still wet. Walton (Walton et al., 1957) studied the corrosion of the aluminium alloy 3003 in an OPC-based mortar up to 27 years. During the first hours, the corrosion rate was higher than 1000 µm/y. However, this rate quickly dropped to reach 50 µm/y after 6 months. Finally, the average of the corrosion rate over 27 years was 1.3 µm/y, meaning that, according to the very high initial rate, the corrosion rate decreased to much lower values at the end of the test.

6.2.2.2 Inhibition method of the corrosion of metallic aluminium More recent literature focused on the inhibition of corrosion of encapsulated aluminium by using different methods. Collier et al. (Collier et al., 2014; Collier et al., 2010) published several works on the effect of the addition of sulphate salts. First, they added near-neutral sulphate salt (anhydrite and gypsum) to the BFS/OPC and PFA/OPC cement in order to decrease the pH, which also resulted in the decrease of the corrosion rate. These additions changed the cement composition, but also the composition of the corrosion layer formed at the interface between the cement and the aluminium. Ettringite and Al(OH)3 were formed around the metallic aluminium. Moreover, the hydrogen formation rate, and consequently the corrosion rate, decreased even if only little difference was observed on the pH. The use of super-sulphated cement powder decreased the pH of the cement up to 2 units. The corrosion rate obtained with this cement appeared to be very small and was lower than the one measured with BFS/OPC paste (Collier et al., 2014).

The addition of phosphates to cement-based materials and especially to calcium aluminate cement (CAC), which possess a great potential for the encapsulation of aluminium due to their near-neutral internal pH, was also investigated. Kinoshita et al. (Kinoshita et al., 2013) showed that the addition of phosphate in such a system lowered the pH of the cement leading to a decrease of the corrosion rate. Moreover, this addition of phosphate improved the strength of the CAC matrix (Chavda et al., 2014).

Another method to inhibit the corrosion of metallic aluminium in cement-based material is to mix the cement with lithium nitrite. Matsuo et al. (Matsuo et al., 1996) found that the addition of LiNO3 inhibits the hydrogen gas generation produced during corrosion under alkaline conditions. This inhibition is attributed to the formation of a LiH(AlO2)2.5H2O (Li-Al) preservation film on the aluminium surface during the solidification process of the cement- based material. Corrosion was then inversely proportional to the amount of LiNO3 added (Matsuo et al., 1997). At least 1.5 wt.% lithium nitrite should be added to decrease the corrosion rate to 10 µm in 1000 hours. Due to water penetration, dissolution of the protective Li-Al layer could happen. However, this should only happen when the concentration of Na2O and K2O, and thus the pH of the cement, is decreased, leading to lower corrosion rates (Matsuo et al.,

1996). Moreover, when the Li-Al film is dissolved, LiNO3 present in the waste should restore the protection properties, increasing the time aluminium is protected. Another study showed that if the temperature of the system increases, the Li-Al film solubility is increased, lowering the time aluminium is protected from corrosion (Matsuo et al., 1999).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 53 of 77 SCK•CEN/27548174 ISC: Public

6.2.2.3 Corrosion products and interface aluminium/encapsulation material Setiadi et al. (Setiadi et al., 2006) studied the corrosion of aluminium, and the corrosion products formed, in different cement-based materials. They found that due to the production of hydrogen, aluminium corroded to produce a porous layer filled with white corrosion products. Moreover, the surrounding cement was affected by the aluminium corrosion up to 2 mm deep. Aluminium hydroxide and strätlingite (2CaO.Al2O3.SiO2.8H2O) were the main corrosion products. Initially, the aluminium hydroxide was the metastable bayerite phase that evolved to form the monoclinic gibbsite. After 180 leaching days, calcite (CaCO3) and some monosulphate (4CaO.Al2O3.SO3.12H2O) were also observed in OPC samples, while gehlenite

(2CaO.Al2O3.SiO2) was present in the BFS/OPC system. After 360 days in OPC, gibbsite was the predominant phase, while residues of bayerite were still observed. However, strätlingite was no longer identified. In the BFS/OPC system, the X-ray diffractogram was similar to the one observed after 180 days, indicating that the corrosion rate of aluminium in the BFS/OPC system is lower than that in pure OPC systems.

6.2.3 Summary of the corrosion rates for aluminium in alkaline conditions In this section, the influence of pH on the corrosion rate is compiled from various authors (Table 4). Finally, these data are also presented in Figure 21 to illustrate the logarithm increase of the corrosion rate in function of the pH. This figure shows that the increase of the corrosion rate as a function of the pH is similar whatever the temperature used during the study.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 54 of 77 SCK•CEN/27548174 ISC: Public

Table 4. Corrosion rates for aluminium in alkaline conditions.

Test pH Temperature Medium Corrosion Technique Reference material (°C) rate (µm/y) 2S-1/2H 10.0 25 (RT) NaOH 25 Weight loss (McKee and Brown, 1947) 2S-1/2H 11.0 25 (RT) NaOH 250 Weight loss (McKee and Brown, 1947) 2S-1/2H 12.0 25 (RT) NaOH 5500 Weight loss (McKee and Brown, 1947) 2S-1/2H 13.0 25 (RT) NaOH 35000 Weight loss (McKee and Brown, 1947) 6063 12.7 30 NaOH 44000 EIS + polarisation (Prabhu and Rao, 2017) 6063 12.7 50 NaOH 65000 EIS + polarisation (Prabhu and Rao, 2017) 6063 13.4 30 NaOH 228000 EIS + polarisation (Prabhu and Rao, 2017) 6063 13.4 50 NaOH 361000 EIS + polarisation (Prabhu and Rao, 2017) 6063 13.7 30 NaOH 355000 EIS + polarisation (Prabhu and Rao, 2017) 6063 13.7 50 NaOH 496000 EIS + polarisation (Prabhu and Rao, 2017) Commercial 10 30 NaOH 50 Weight loss (Tabrizi et al., Al (>99.5%) 1991) Commercial 10 60 NaOH 100 Weight loss (Tabrizi et al., Al (>99.5%) 1991) Commercial 11 30 NaOH 100 Weight loss (Tabrizi et al., Al (>99.5%) 1991) Commercial 11 60 NaOH 300 Weight loss (Tabrizi et al., Al (>99.5%) 1991)

Commercial 12.5 30 Ca(OH)2 22 Weight loss + (Takatani et al., Al polarisation 1983) 3003-H14 ~12 25 (RT) OPC- > 1000 Weight loss (Walton et al., based (early stage 1957) mortar corrosion) solution Commercial n.i. 25 9:1 13000 Hydrogen (Thomas and Al BFS/OPC (early stage evolution + EIS Naish, 1994) corrosion) Commercial n.i. 25 9:1 0.1 Hydrogen (Thomas and Al BFS/OPC evolution + EIS Naish, 1994) n.i. = not indicated

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 55 of 77 SCK•CEN/27548174 ISC: Public

Figure 21. Corrosion rate evolution of metallic aluminium in NaOH solutions in function of pH and temperature. In blue, corrosion rate obtained at 25 °C. In red, corrosion rate obtained at 30 °C. In green, corrosion rate obtained at 50 °C. In orange, corrosion rate obtained at 60 °C.

6.3 Conclusions Due to the amphoteric behaviour of the oxide/hydroxide layer formed at the aluminium surface, metallic aluminium corrodes in both acidic and alkaline media. However, this layer is protective against corrosion in the pH range between 4 and 8.5. In alkaline media, aluminium - corrodes to form aluminate ions (Al(OH)4 ) and hydrogen gas. If the temperature and the pH of the system increase, the corrosion rate increases too. At lower hydroxide concentrations (0.5 – 1 N), the corrosion rate is proportional to the cube root of the concentration of NaOH and

If encapsulated in cement-based materials, at the early stage of the encapsulation, when the cement-based material is still wet, the corrosion rate is high. However, the rate decreases fast to reach values as low as 0.1 to 0.5 µm/y after a few months/years.

The corrosion rate can be reduced even further by adding inhibitors, such as LiNO3 or sulphates, to the cement-based material. The cement composition can also be altered (e.g. addition of BFS) to decrease the pore water pH.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 56 of 77 SCK•CEN/27548174 ISC: Public

7 Corrosion of stainless steel

7.1 Introduction Stainless steels are iron-based alloys that contain a minimum of approximately 11% Cr (Sedriks, 1996). They owe their ‘stainless’ characteristics to the formation of an invisible and adherent chromium-rich oxide surface film. This oxide film forms and heals itself in the presence of oxygen. Other elements added to improve particular characteristics include nickel, molybdenum, copper, titanium, aluminum, silicon, niobium, nitrogen, sulfur, and selenium. The advantages of these additions are shown in Figure 22.

There are several families of stainless steels, depending on the microstructure of the alloys. This microstructure in turn determines the mechanical properties of the steels and thus the workability to fabricate containers. In general terms, ferritic, austenitic and martensitic steels can be distinguished. In addition, the so-called duplex grades contain approximately 50% ferrite and 50% austenite. In terms of microstructure, ferrite is a body-centered cubic crystal with a limited capacity to dissolve carbon. Austenite on the other hand, is a face-centered cubic crystal with a high capacity to dissolve carbon. In the iron-carbon phase diagram, the stability domain starts at 910 °C for pure iron and drops to lower temperatures with increasing carbon content. Slow cooling will transform the austenite to ferrite, which is stable at lower temperatures. If the alloy however is quenched, the carbon is very rapidly thrown out of solution to produce a very hard bcc crystal structure which is called martensite.

Martensitic stainless steels are ferromagnetic and are hardenable by heat treatment. They have a chromium content of 11-18% and a carbon content up to 1.2%. These alloys are difficult to weld (requiring extensive pre- and post-weld treatment), have a high susceptibility to hydrogen-assisted cracking and have a poor low-temperature impact resistance. Because of these properties, martensitic steels are not very suitable for nuclear applications.

Ferritic stainless steels are ferromagnetic and cannot be hardened by heat treatment. They have a chromium content in the range of 10.5-30%. They have a higher corrosion resistance than the martensitic grades, due to their generally higher chromium content. The weldability of these alloys is poor, and unless proper heat treatment is employed, may lead to brittleness and diminished corrosion resistance.

Austenitic stainless steels are essentially non-magnetic and can only be hardened by cold working. They acquire their austenitic structure by adding austenitizing elements, such as nickel, manganese, and nitrogen. Their chromium content is in the range 16-26%, their nickel content up to 35%, and their manganese content up to 15%. The austenitic stainless steels combine a good corrosion resistance, high-temperature strength, ease of fabrication and weldability, good ductility and impact resistance. Therefore, most research programs focused on austenitic stainless steels as candidate container materials.

In the past, stainless steel designations were formulated by the American Iron and Steel Institute (AISI) and these designations are still used up to now, for example for the low-carbon, high-molybdenum austenitic stainless steel AISI 316L hMo. More recently, the AISI designations have been replaced by a Unified Numbering System, in which the AISI 316L hMo is named S31700. Throughout this chapter, both designations will be used, with the AISI designation followed by the UNS designation between brackets.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 57 of 77 SCK•CEN/27548174 ISC: Public

Figure 22. Overview of the composition and property links between the most important stainless steel alloys (Sedriks, 1996).

7.2 Mechanism of stainless steel corrosion in geological disposal conditions The basic mechanism of stainless steel corrosion is formed by the reactions of the main constituent, namely iron. In highly alkaline and anoxic conditions, iron reacts with water to form iron hydroxide and hydrogen gas (Diomidis, 2014):

→ ↑ (34)

Fe (OH)2 has a low solubility and therefore forms a film on the surface of the steel (even on carbon steel this is the case). Ferrous hydroxide may transform via the following reaction, known as the Schikorr reaction, into magnetite:

→ ↑ (35)

This is reflected in Figure 23, which shows the thermodynamically stable states as a function of potential and pH (Pourbaix diagrams) for iron, assuming oxides and hydroxides respectively as

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 58 of 77 SCK•CEN/27548174 ISC: Public

solid substances. In the case of stainless steels, a passive film forms on the surface, which is mainly an oxy-hydroxide film, based on selective oxidation of Fe and Cr during anodic polarization (Olsson and Landolt, 2003a).

7.2.1 Alloying additions The properties of the passive film are a function of the presence of many possible alloying elements, leading to additional corrosion reactions. In our discussion, we will limit ourselves to the main constituents next to iron, namely chromium, nickel and molybdenum. The Pourbaix diagrams of these elements are shown in Figure 23 and Figure 24. We will now discuss their influence on the corrosion behaviour of stainless steel.

Chromium

Chromium is, after iron, the main constituent of austenitic stainless steels. It is added to increase the corrosion resistance, which increases as more chromium is added. In acid solutions the passive film is mainly composed of chromium III. In alkaline solutions, the solubility of chromium increases, resulting in a higher fraction of iron in the passive film (Olsson and Landolt, 2003a). The main oxidation reactions of chromium in anoxic alkaline environments are (Smart et al., 2004):

→ (36)

→ (37)

Nickel

Nickel is an austenite former. It is less readily oxidized than iron and chromium and therefore there is an enrichment of Ni in its metallic state in the metal closest to the oxide/metal interface. Nickel, together with chromium, is also responsible for the formation of a passive film on stainless steel. Its main oxidation reactions in anoxic alkaline environments are:

→ (38)

→ (39)

Molybdenum

Molybdenum, when added to chromium-nickel austenitic steels, has a strong beneficial influence on the pitting and crevice resistance of the steel in environments containing chlorides and sulphur species. Molybdenum does not form a passive film itself, but rather is incorporated into the passive film, thus stabilizing it (Frankel, 1998).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 59 of 77 SCK•CEN/27548174 ISC: Public

(a) (b)

Figure 23. Potential (E in V vs. SHE) – pH equilibrium diagrams for the system iron-water at 25 °C assuming (a) oxides and (b) hydroxides as solid substances (Pourbaix, 1974).

(a) (b)

Figure 24. Potential (E in V vs. SHE) – pH equilibrium diagrams at 25 °C for the systems (a) chromium-

water assuming Cr(OH)3 as solid substance, (b) chromium-water assuming Cr2O3 as solid substance, (d) nickel-water, and (e) molybdenum-water (Pourbaix, 1974).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 60 of 77 SCK•CEN/27548174 ISC: Public

7.2.2 Properties of the passive film An important aspect in the corrosion behaviour of stainless steels is passivity. The composition and structure of the passive film change continuously in a process of exchange of species with the environment. Cr plays an important role in the establishment and stability of passive films on stainless steels (Schmuki, 2002a). The passive film on stainless steel in acidic media is mainly composed of Cr(III)-rich oxides and hydroxides (Marcus and Maurice, 2017). The enrichment in chromium, especially at the film/electrolyte interface, is believed to be caused by preferential dissolution of iron (Olsson and Landolt, 2003a). In alkaline media however, iron cations accumulate in the outer part of the passive film because iron oxide does not dissolve in alkaline solutions (Schmutz and Landolt, 1999). The high corrosion resistance of stainless steels relies on the existence of a few nanometer thick film on the steel surface. The properties of this passive film depend strongly on the alloy composition. This implies that the corrosion resistance of stainless steels can be increased by adding suitable alloying elements. In particular the austenitic alloys have excellent corrosion properties, because the most commonly used austenite stabilizers, i.e. Ni, Mn, and N, all contribute to passivity.

Numerous investigations have focused on passive film structure. Overviews of these studies can be found in (Olsson and Landolt, 2003b), (Fischmeister and Roll, 1984), (Schmuki, 2002b), and (Clayton and Olefjord, 1995). In the early days, mainly ex situ techniques were used, such as XPS and electron microscopy. In combination with electrochemical techniques, these surface analytical methods give a good indication of the role of alloying elements in the passivation. It is found that the passive film on austenitic steels generally has a duplex structure, with an inner oxide-based barrier film, and an outer hydrated layer (deposit layer) on top of it. The inner oxide film is the primary diffusion barrier against egressing cations and ingressing aggressive anions such as chloride. Cr3+ is found in greater abundance in the passive film in both the inner and outer layers than Fe, despite the fact that Fe is the main element in the alloy. Clearly, Cr is the main passivating species in stainless steels.

Olsson (Olsson and Landolt, 2003b) reviewed recent in situ studies on passive films on stainless steels. In situ techniques offer the advantage that the passive film is not altered while transferring samples from solution to ultra-high vacuum. In addition, they make it possible to investigate the kinetics of film alteration in response to a change in environment. It is found that the factors influencing the passive film, include the potential, the presence of halides in the electrolyte, the pH, and the temperature.

7.2.2.1 Potential The thickness of passive films on stainless steels is a linear function of the applied potential. This is illustrated in Figure 25 for a Fe15Cr alloy in an acidic medium, and Fe10Cr and Fe20Cr alloys in basic solutions. From this figure it is also clear that the passive film is thicker in alkaline media than in acidic solutions. The composition and chemistry of the film also vary with potential (Olsson and Landolt, 2003a).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 61 of 77 SCK•CEN/27548174 ISC: Public

Figure 25. Oxide layer thickness on a stainless steel as a function of potential for a Fe15Cr alloy in 0.5 M

H2SO4, and for Fe10Cr and Fe20Cr alloys in 1 M NaOH, estimated using XPS. The film growth region is considerably wider in the alkaline medium, which also gives thicker films (Olsson and Landolt, 2003a).

7.2.2.2 pH The main effect of an increased pH is a lower dissolution rate. This leads to a thicker passive film and a lower corrosion rate. Schmutz used an electrochemical quartz crystal microbalance (QCM) to compare the passive behavior of iron-chromium alloys in acidic and alkaline media (Schmutz and Landolt, 1999). In alkaline media, the passive film consists of two parts: an inner chromium-rich part, and an outer iron-rich part (due to the low dissolution of iron oxide in alkaline media).

7.2.2.3 Temperature The influence of temperature on film thickness and dissolution rate seems to be negligible. Jin immersed stainless steels in 0.1 M NaCl at 25 °C, 60 °C and 90 °C and found little difference in the thickness and structure of the passive film (Jin and Atrens, 1988).

7.2.3 Passivity breakdown The main advantage of stainless steel from a corrosion protection point of view is the presence of a passive film with a very low dissolution rate. There are however circumstances which cause the passive film to break down locally without (immediate) repassivation. In particular, the presence of aggressive species such as chloride, can cause localized corrosion (Dillon, 1995; Marcus and Oudar, 1995; Soltis, 2015; Szklarska-Smialowska, 1986).

The susceptibility of stainless steels to localized corrosion (pitting, or crevice corrosion in the case of occluded cells) can be measured with electrochemical techniques. In particular, cyclic potentiodynamic polarization measurements yield polarization curves from which two characteristic pitting potentials can be derived in a straightforward manner, as illustrated in

Figure 26 (Szklarska-Smialowska, 1986): EP – the pitting potential or pit nucleation potential – is the potential above which pits nucleate and grow, ER – the protection potential – is the potential below which no pitting occurs and above which pits already nucleated can grow. A comparison with the free corrosion potential Ecorr indicates the susceptibility of the metal to pitting. In the case that the corrosion potential is more noble than the pit nucleation potential:

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 62 of 77 SCK•CEN/27548174 ISC: Public

(40) immediate pitting corrosion is to be expected. Another case occurs when the corrosion potential is below the pit nucleation potential but above the protection potential:

(41)

The protection potential, because it is related to steady state conditions inside pits or crevices, can in this sense be used for the long term prediction of the value of the pitting potential and as such as an indicator for the long term susceptibility to pitting corrosion, as put forward by A. Pourbaix (Pourbaix, 2003). So, a possible approach to predict pitting corrosion can be to use

ER as a critical limit for the corrosion potential instead of EP. It also needs to be mentioned that, at potentials below the pit nucleation potentials, metastable pits can be formed (Baroux, 1995).

Figure 26. Schematic of a polarization curve of a metal prone to pitting, with EP the pitting potential, ER the repassivation potential, and Ecorr the corrosion potential.

7.3 Kinetics of stainless steel corrosion The literature on corrosion data for stainless steel in highly alkaline and anoxic conditions is scarce. Fujisawa et al. (Fujisawa et al., 1999) investigated the corrosion of AISI 304 type stainless steel in synthetic ground waters of pH 12.8 (simulating the early stage of the disposal environment in concrete structures, resulting in a high initial pH) and pH 10.5 (simulating the long-term disposal environment after ten thousand years according to the Japanese concept). They determined the corrosion rate by measuring the displacement of a mercury column in a vertical glass tube. After 200 days, the corrosion rates had decreased to 8 nm/y at pH 12.8 and to 3 nm/y at pH 10.5. Smart et al conducted tests in 0.1 M NaOH for different test durations (up to approximately ten years of exposure). They determined the corrosion rate by gas generation rate measurements based on the displacement of a liquid in a vertical gas tube. The glass cell consisted of two compartments. The first compartment, containing the test pieces, was connected via a gas line to the second compartment, which was a reservoir for a low vapor

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 63 of 77 SCK•CEN/27548174 ISC: Public

pressure liquid, di-butyl phtalate. They reported data below 10 nm/y, which was the detection limit of their method (Smart et al., 2004). The real value is probably lower. Wada et al used mass spectrometry to measure the gas generation by anaerobic corrosion of 304 stainless and measured corrosion rates between 6 and 9 nm/y in the pH range 10.0-13.5 (Wada, 1999). A result based on gas chromatography and with a test duration of up to 230 days qualifies as reliable. However, because in this measurement the gas was accumulated in ampoules over the entire duration of the test, the result presents an average value over 230 days and therefore an overestimation of the long-term corrosion rate. Nishimura et al. derived corrosion rates on 304 stainless steel using hydrogen gas measurements (Nishimura et al., 2003). Again, the value reported represents an average rate over 650 days of exposure and therefore may represent an overestimation of the instantaneous corrosion rate after 650 days. The data at 50°C from 5 were used to construct the graph in Figure 27, which shows the dependence of the corrosion rate of stainless steel on pH at 50°C. It can be seen that the corrosion rate decreases with increasing pH.

Table 5. Corrosion rates for stainless steel in highly alkaline, anoxic conditions.

Test pH Temperature Test duration Corrosion Reference material (°C) (days) rate (µm/y) 304 SS 10.0 50 230 0.009 (Wada, 1999) 304 SS 12.5 50 230 0.0055 (Wada, 1999) 304 SS 13.5 50 230 0.0063 (Wada, 1999) 304 SS 12.5 30 200 0.0003 (Fujisawa et al., 1999) 304 SS 12.4 35 650 0.010 (Nishimura et al., 2003)

Figure 27. Corrosion rate of stainless steel in highly alkaline, anoxic conditions at 50 °C.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 64 of 77 SCK•CEN/27548174 ISC: Public

7.4 Conclusions The literature data on corrosion rates of stainless steel in highly alkaline anoxic conditions is scarce. In the pH range 10-12.5, the reported corrosion rates are in the order of magnitude of several nm/y, with an upper limit of 10 nm/y.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 65 of 77 SCK•CEN/27548174 ISC: Public

8 Corrosion of Inconel alloys

8.1 Introduction Inconel is a family of high-performance alloys based on nickel-chromium. At high temperature, Inconel forms a passivating oxide layer protecting the base metal from further attack. Inconel is especially attractive for high temperature applications because it retains high strength over a wide range of temperatures, thus outperforming aluminum and steel in creep resistance. The main reason for passive film development is the high content in chromium. Inconel exhibits high resistance to chloride-induced phenomena, such as pitting, crevice corrosion, and stress corrosion cracking. In the nuclear industry, Inconel has been mainly used in steam generators, with alloys 600 and 800 being the preferred materials. In later years, alloy 600 has been gradually replaced by alloy 690, due to caustic and high-purity water failures of alloy 600. Other PWR applications are nozzles and baffle plates. Alloy 718 has been used for high- strength bolts and springs. The effect of the alloying additions from a corrosion point of view are largely the following, according to Crook (Crook, 2005):

 Chromium: the role of chromium in Inconel is the same as that in stainless steels, that is, to participate in the formation of passive films that enhance the corrosion resistance (see Chapter 7).  Molybdenum: the addition of molybdenum to nickel greatly enhances its nobility under active corrosion conditions. In particular, it provides high resistance to reducing chemicals, such as hydrochloric acid. In combination with chromium, it produces alloys that are extremely versatile (resistant to both oxidizing and reducing chemicals) and that can withstand chloride-induced pitting and crevice corrosion.  Iron: the main purpose of adding iron is to reduce the cost of Inconel, but it also contributes to the formation of passive films.

As the Inconel grades are largely based on nickel and chromium, we will first discuss the corrosion properties of these metals, before proceeding to a description of the corrosion of Inconel itself.

8.2 Corrosion of nickel and chromium The Pourbaix (potential-pH) diagram for the nickel-water system at 25°C is shown in Figure 28. Nickel can be considered to be a slightly noble metal, as its domain of thermodynamic stability has a small zone in common with that of water (Pourbaix, 1974).

According to the thermodynamic diagram, the corrosion resistance of nickel should depend on the pH and the presence of oxidising agents, for non-complexing solutions, as follows: Ni should be immune to corrosion in neutral or alkaline solutions free from oxidising agents, slightly corrodible in acid solutions free from oxidising agents, and very corrodible in acid or very alkaline solutions containing oxidising agents. In neutral or slightly alkaline oxidising solutions, Ni should be covered with a layer of oxide (Pourbaix, 1974).

These predictions are only part in agreement with the experimental facts. When there is no oxidising action, i.e. notably in the case of solutions not containing oxidising agents and in the

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 66 of 77 SCK•CEN/27548174 ISC: Public

absence of any anodic polarisation, nickel is hardly corroded at all, not only in neutral or alkaline solutions, but also in many acid solutions. This favourable behaviour in non-oxidising acid media is probably due, on the one hand, to the great irreversibility of the corrosion reaction Ni → Ni2+ + 2e- and, on the other hand, to the large hydrogen overpotential on nickel (Pourbaix, 1974).

Figure 28. Pourbaix (potential-pH) diagram of the nickel-water system at 25°C (adapted from (Chivot, 2004)).

In near-neutral (pH ~ 8) to alkaline (pH ~ 13) solutions, nickel precipitates as nickelous hydroxide, Ni(OH)2. The position of the stability domain of Ni(OH)2 in Figure 28 shows that it is a thermodynamically stable substance in the presence of water or neutral or slightly alkaline solutions free from oxidising or reducing agents. It readily dissolves in acid solutions with the formation of nickelous ions Ni2+. The oxidation of nickelous hydroxide in alkaline media can 2- give rise to the formation of Ni3O4, Ni2O3, and NiO2 (Pourbaix, 1974).

Ni3O4 is thermodynamically stable in the presence of aerated water due to the fact that its stability domain has a large zone in common with that of water (Pourbaix, 1974).

Ni2O3 is an oxidising agent, since its stability domain lies completely above line (b) (oxidation of H2O). It is therefore unstable in the presence of water which it tends to decompose with the evolution of oxygen. It dissolves in acid solutions with the formation of Ni2+ ions and the evolution of oxygen. It is insoluble in alkaline solutions (Pourbaix, 1974).

NiO2 is an unstable substance that decomposes rapidly into Ni2O3, Ni3O4, and oxygen (Pourbaix, 1974).

The anaerobic corrosion of nickel is governed by reaction:

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 67 of 77 SCK•CEN/27548174 ISC: Public

→ (42)

Figure 29 shows the stability domains of chromium compounds. From this graph, it appears that chromium’s stability domain is significantly below that of water. However, Cr(OH)3 and

Cr2O3 form passive films that protect the base metal from corrosion in a large E-pH domain (see Figure 30).

(a) (b)

Figure 29. Potential (E in V vs. SHE) – pH equilibrium diagrams at 25 °C for the systems (a) chromium-

water assuming Cr(OH)3 as solid substance, (b) chromium-water assuming Cr2O3 as solid substance, (Pourbaix, 1974).

(a) Figure established considering Cr(OH)3. (b) Figure established considering Cr2O3.

Figure 30. Theoretical conditions of corrosion, immunity and passivation of chromium, at 25°C (Pourbaix, 1974).

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 68 of 77 SCK•CEN/27548174 ISC: Public

8.3 Corrosion of Inconel in alkaline conditions Nickel alloys are very corrosion-resistant in alkaline conditions. For example, Figure 31 shows the isocorrosion diagrams for nickel 200 and 201 in NaOH as a function of temperature in aerobic conditions. It can be seen that at ambient temperatures, the corrosion rate is less than 2.5 µm/y. The corrosion behaviour under reducing anoxic conditions, however, is not clear. In reducing conditions, the passive film will dissolve, but the kinetics of this process are not known. However, we can assume that the corrosion behaviour of Inconel can be approached by the one of stainless steel, as both metals rely for their corrosion resistance on the presence of a chromium-based passive film. Therefore, we suggest the same upper corrosion rate limit of 10 nm/y. At the time of writing this report, we have no indication of a value for the lower limit of the corrosion rate range.

Figure 31. Isocorrosion diagram for Inconel 200 and 201 in NaOH (Metals, 2000).

8.4 Conclusions Inconel is a family of superalloys based on nickel-chromium. For their corrosion resistance, they rely on the formation of a passivating oxide layer. The most important Inconel alloys for the nuclear industry are alloys 600, 690, 718 and 800. There are no data found in the literature for the anaerobic and highly alkaline conditions. Therefore, we suggest a maximum corrosion rate of 10 nm/y, like for stainless steel, which also relies on a chromium-based passive film.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 69 of 77 SCK•CEN/27548174 ISC: Public

9 General conclusions

This report describes the corrosion behaviour of the main metals present in Belgian category B and C waste under anoxic and highly alkaline conditions. The effect of lower pH conditions on the corrosion rate was also discussed in this report. The metals investigated were zirconium alloys, metallic uranium, beryllium, aluminium, and stainless steel. The following corrosion rates can be set forward for the metals under concern:

 Zirconium alloys exhibit a long-term corrosion rate in the range 0.2-6 nm/y. These corrosion rates were obtained for Zircaloy-4 in aqueous solutions. It is expected that the corrosion rates for M5 and ZIRLO are lower, but this still has to be proven. In a solid cementitious matrix, the corrosion rates will probably also be lower.  In highly alkaline aqueous solutions, metallic uranium exhibits corrosion rates in the order of hundreds of µm/y. In cement-based solid matrices, however, initial corrosion rates of 60-150 µm/y were obtained, with the corrosion rate rapidly dropping to 5-12 µm/y. Values lower than 0.1 µm/y were detected in non-saturated concrete.  The literature data on beryllium corrosion in highly alkaline, anoxic conditions is scarce. In these conditions, the corrosion rate is in the order of magnitude of several µm/y. These values were obtained in aqueous solutions and will probably be lower in a cement-based solid matrix.  In highly alkaline solutions, the corrosion rate of aluminium is a logarithmic function of pH. Corrosion rates up to 500000 µm/y have been recorded at pH 14. These values have been measured in aqueous solutions, while the corrosion rate will probably be lowered by mass transport restrictions in a cement-based solid matrix.  The literature data on corrosion rates of stainless steels in highly alkaline, anoxic conditions is scarce. In the pH range of 10-12.5, the reported corrosion rates are in the order of magnitude of several nm/y, with an upper limit of 10 nm/y.  There are no literature data found on the corrosion rate of Inconel grades in highly alkaline, anoxic conditions. As the corrosion behaviour of Inconel is based on the same mechanism as stainless steel (formation of a passive film by chromium), we put forward an upper limit of the corrosion rate of 10 nm/y.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 70 of 77 SCK•CEN/27548174 ISC: Public

9 Acknowledgments

This work is undertaken in close co-operation with, and also with the financial support of ONDRAF/NIRAS, the Belgian Agency for the Management of Radioactive Waste and Enriched Fissile Materials, as part of its programme on the geological disposal of high- level radioactive waste.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 71 of 77 SCK•CEN/27548174 ISC: Public

10 References

Adler Flitton, M.K., Bishop, C.W., Delwiche, M.E. and Yoder, T.S. (2004) Long term corrosion/degradation test: six year results. Idaho National Engineering and Environmental Laboratory. Report INEEL/EXT‐04‐02335. Agency, I.A.E. (1998) Waterside corrosion of zirconium alloys in nuclear power plants. IAEA‐TECDOC‐ 996. Allen, T.R., Konings, R.J.M. and Motta, A.T. (2012) Corrosion of zirconium alloys, in: Konings, R.J.M. (Ed.), Comprehensive Nuclear Materials. Elsevier, Amsterdam, pp. 49‐68. Andrade, C. and González, J.A. (1978) Quantitative measurements of corrosion rate of reinforcing steels embedded in concrete using polarisation resistance measurements. Werkstoffe und Korrosion 29, 515‐519. Baker, M.M., Less, L.N. and Orman, S. (1966a) Uranium + water reaction. Part 1.‐Kinetics, products and mechanism. Transactions of the Faraday Society 62, 2513‐2524. Baker, M.M., Less, L.N. and Orman, S. (1966b) Uranium + water reaction. Part 2.‐Effect of oxygen and other gases. Transactions of the Faraday Society 62, 2525‐2530. Bard, A.J. and Faulkner, L.R. (2001) Electrochemical Methods: Fundamentals and Applications. John Wiley, New York. Bard, A.J., Parsons, R. and Jordan, J. (1985) Standard Potentials in Aqueous Solutions. Baroux, B. (1995) Further insights on the pitting corrosion of stainless steels, in: Marcus, P., Oudar, J. (Eds.), Corrosion mechanisms in theory and practice. Marcel Dekker, New York, pp. 265‐309. Blackwood, D. and Farmilo, N.M. (1996) Corrosion of Uranium in Cement. AEA technology, AEAT‐0025. Bossis, P., Pécheur, D., Hanifi, K., Thomazet, J. and Blat, M. (2006) Comparison of the high burn‐up corrosion on M5 and low Tin Zircaloy‐4. Journal of ASTM International 3, 494‐524. Bullock, J.S., Brown, A.D. and Dobbins, A.G. (1974) Electrochemistry of Aqueous Corrosion. Uranium and its Alloys. Oak Ridge Y‐12 Plant, Y‐1949. Chavda, M.A., Kinoshita, H. and Provis, J.L. (2014) Phosphate modification of calcium aluminate cement to enhance stability for immobilisation of metallic wastes. Advances in Applied Ceramics 113, 453‐459. Chivot, J. (2004) Thermodynamique des produits de corrosion. Foctions thermodynamiques, diagrammes de solubilité, diagrammes E‐pH des systèmes Fe‐H2O, Fe‐CO2‐H2O, Fe‐S‐H2O, Cr‐H2O et Ni‐ H2O en fonction de la température. ANDRA, Châtenay‐Malabry, France. Clayton, C.R. and Olefjord, I. (1995) Passivity of austenitic stainless steels, in: Marcus, P., Oudar,J. (Ed.), Corrosion mechanisms in theory and practice. Marcel Dekker. Collier, N.C., Milestone, N.B., Gordon, L.E. and Ko, S.C. (2014) The suitability of a supersulfated cement for nuclear waste immobilisation. Journal of Nuclear Materials 452, 457‐464. Collier, N.C., Milestone, N.B. and Swift, P.D. (2010) Immobilisation matrices for intermediate level nuclear wastes using sulphate activated BFS/OPC and PFA/OPC composite cements. Advances in Applied Ceramics 109, 269‐274. Colmenares, C.A. (1984) Oxidation mechanisms and catalytic properties of the actinides. Progress in Solid State Chemistry 15, 257‐364. Cook, A., Fox, T. and Wilding, C.R. (1989) The Effect of pH on Aluminium Corrosion. AERE‐G5156. Cox, B. (2005) Some thoughts on the mechanisms of in‐reactor corrosion of zirconium alloys. Journal of Nuclear Materials 336, 331‐368. Crook, P. (2005) Corrosion of nickel and nickel‐base alloys, in: Cramer, S.D., Covino, B.S. (Eds.), Corrosion: Materials. ASM International, Materials Park, Ohio. Curwen, S. (1994) Assessment of the Effect of Uranium and Mild Steel Corrosion on the Stability of Cemented SSSF Wasteforms. BNFL R&D Curwen, S. and James, P. (1995) The Corrosion of Uranium in Big Box Treatment Plant Grout. BNFL R&D de Gabory, B., Motta, A.T. and Wang, K. (2015) Transmission electron microscopy characterization of Zircaloy‐4 and ZIRLO oxide layers. Journal of Nuclear Materials 456, 272‐280.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 72 of 77 SCK•CEN/27548174 ISC: Public

Delegard, C.H. and Schmidt, A.J. (2009) Uranium metal reaction behavior in water, sludge, and grout matrices. Pacific Northwest National Laboratory, PNNL‐17815, Rev 1. Deltombe, E. and Pourbaix, M. (1958) The electrochemical behavior of aluminum: potential pH diagram of the system Al‐H2 at 25 C, Corrosion, pp. 496‐500. Dillon, C.P. (1995) Corrosion resistance of stainless steels. Marcel Dekker, New York. Diomidis, N. (2014) Scientific basis for the production of gas due to corrosion in a deep geological repository, Nagra Arbeitsbericht NAB 14‐21. Dong, D. and Vandegrift, G.F. (1996) Kinetics of Dissolution of Uranium Metal Foil by Alkaline Hydrogen Peroxide. Nuclear Science and Engineering 124, 473‐481. Druyts, F. and Jobbágy, V. (2017) Current status of hydrogen measurement techniques in gases: feasibility study on low level corrosion application. SCK‐CEN Internal Report I‐0540, SCK•CEN, Mol, Belgium. Duan, Z., Uang, H., Satoh, Y., Murakami, K., Kano, S., Zhao, Z., Shen, J. and Abe, H. (2017) Current status of materials development of nuclear fuel cladding tubes for light water reactors. Nuclear Engineering and Design 316, 131‐150. Endtinger, F. and Weber, H. (1967) Stability of aluminium in contact with alkaline building materials. Anti‐Corrosion Methods and Materials 14, 8‐24. Farmilo, N.M. and Simmons, R.F. (1997) Corrosion of Uranium in Cement and Simulated Pore Water. AEA Technology, APEC(97)P290. Fischmeister, H. and Roll, U. (1984) Passivschichten auf rostfreien Stählen: Eine Übersicht über oberflächenanalytische Ergebnisse. Fresenius Z Anal Chem 319, 639‐645. Frankel, G.S. (1998) Pitting corrosion of metals: a review of the critical factors. Journal of the Electrochemical Society 145, 2186‐2198. Fujisawa, R., Kurashige, T., Inagaki, Y. and Senoo, M. (1999) Gas generation behavior of transuranic waste under disposal conditions. Materials Research Society Symposium Proceedings 556, 1199‐1206. Gad Allah, A.G., Mazhar, A.A., El‐Taib Heakal, F. and Ameer, M.A. (1989) Formation and dissolution of anodic films on zirconium in NaOH: kinetic studies. Journal of Applied Electrochemistry 19, 213‐218. Garzarolli, F. and Garzarolli, M. (2012) PWR Zr alloy cladding water side corrosion. Advanced Nuclear Technology International, Mölnlycke, Sweden. Garzarolli, F., Jung, W., Schoenfeld, H., Garde, A.M., Parry, G.W. and Smerd, P.G. (1982) Waterside corrosion of Zircaloy fuel rods. EPRI, NP‐2789. Godard, H.G. (1967) The corrosion of light metals. Wiley, New York. Godard, H.G. (1979) Corrosion of metals by waters. Materials Performance 18, 21‐27. Godfrey, I.H., Brogden, M. and Curwen, S. (2004) BNFL Historical Data on the Corrosion of Uranium in BFS/OPC Cement. NSTS (04) 4992. Gras, J.‐M. (2014) State of the art of 14C in Zircaloy and Zr alloys ‐ 14C release from zirconium alloy hulls, CArbon‐14 Source Term ‐ CAST. ANDRA. Gulbrandsen, E. and Johansen, A.M.J. (1994) A study of the passive behaviour of beryllium in aqueous solutions. Corrosion Science 36, 1523‐1536. Guo, S., Leavitt, J.J., Zhou, X., Lahti, E. and Zhang, J. (2016) Corrosion of aluminium alloy 1100 in post‐ LOCA solutions of a nuclear reactor. RSC Advances 6, 44119‐44128. Hansson, C. (1984) The corrosion of zircaloy 2 in anaerobic cement pore solution, SKB Technical Report. Svensk Kärnbränslehantering AB. Report 84‐13 Harker, N.J., Scott, T.B., Jones, C.P., Petherbridge, J.R. and Glascott, J. (2013) Altering the hydriding behaviour of uranium metal by induced oxide penetration around carbo‐nitride inclusions. Solid State Ionics 241, 46‐52. Haschke, J.M. (1995) Reactions of Plutonium and Uranium with Water: Kinetics and Potential Hazards. Los Alamos National Laboratory. Report LA‐13069‐MS. Haws, W.J. (2005) Corrosion of beryllium and aluminium‐beryllium composites, Corrosion: Materials. ASM International, pp. 360‐369. Hayes, M., Godfrey, I.H. and Brogden, M. (2003) Initial Studies to Determine the Acute and Intermediate Corrosion Rate of Uranium in BFS/OPC Grouts. Report RAT (03) 3854.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 73 of 77 SCK•CEN/27548174 ISC: Public

Hill, M.A., Butt, D.P. and Lillard, R.S. (1996) The corrosion/electrochemistry of beryllium and beryllium weldments in aqueous chloride environments. Los Alamos National Laboratory, Los Alamos, New Mexico. Report LAUR‐96‐3669. Hill, M.A., Butt, D.P. and Lillard, R.S. (1997) The passivity and breakdown of beryllium in aqueous solutions. Los Alamos National Laboratory, Los Alamos, New Mexico. Report LAUR‐97‐5007. Hillner, E., Franklin, D.G. and Smee, J.D. (2000) Long‐term corrosion of Zircaloy before and after irradiation. Journal of Nuclear Materials 278, 334‐345. Hilton, B.A. (2000) Review of oxidation rates of DOE spent nuclear fuel : Part 1 : nuclear fuel. ; Argonne National Lab., IL (US). Report ANL‐00/24. Huot, J.‐Y. (1999) Electrochemical behaviour of zirconium metal and its oxide film in LiOH and HF solutions. Journal of Applied Electrochemistry 22, 852‐858. Jenks, G.H. (1971) Electrochemistry of uranium and uranium‐alloys in aqueous solutions. Oak Ridge National Lab., Tennessee. Report ORNL‐‐4651. Jin, S. and Atrens, A. (1988) ESCA‐studies of the structure and composition of the passive film formed on stainless steels by various immersion temperatures in 0.1 M NaCl solution. Applied Physics A 45, 83‐91. Jones, C.P., Brenner, C.M., Stitt, C.A., Armstrong, C., Rusby, D.R., Mirfayzi, S.R., Wilson, L.A., Alejo, A., Ahmed, H., Allott, R., Butler, N.M.H., Clarke, R.J., Haddock, D., Hernandez‐Gomez, C., Higginson, A., Murphy, C., Notley, M., Paraskevoulakos, C., Jowsey, J., McKenna, P., Neely, D., Kar, S. and Scott, T.B. (2016) Evaluating laser‐driven Bremsstrahlung radiation sources for imaging and analysis of nuclear waste packages. Journal of Hazardous Materials 318, 694‐701. Jones, C.P., Scott, T.B., Petherbridge, J.R. and Glascott, J. (2013) A surface science study of the initial stages of hydrogen corrosion on uranium metal and the role played by grain microstructure. Solid State Ionics 231, 81‐86. Kaneko, M., Miura, N., Fujiwara, A. and Yamamoto, M. (2004) Evaluation of gas generation rate by metal corrosion in the reducing environment, RWMC Engineering Report RWMC‐TRE‐03003. Kato, O., Tanabe, H., Sakuragi, T., Nishimura, T. and Tateishi, T. (2014) Corrosion tests of Zircaloy hull waste to confirm applicability of corrosion model and to evaluate influence factors on corrosion rate under geological disposal conditions. Materials Research Society Symposium Proceedings 1665, 195‐ 202. Kinoshita, H., Swift, P., Utton, C., Carro‐Mateo, B., Marchand, G., Collier, N. and Milestone, N. (2013) Corrosion of aluminium metal in OPC‐ and CAC‐based cement matrices. Cement and Concrete Research 50, 11‐18. Kurashige, T. (1999) Gas generation behaviour of Zircaloy‐4 under waste disposal conditions, 7th Interantional Conference on Radioactive Waste Management and Environmental Remediation (ICEM'99), Nagoya, Japan. Kursten, B. (2018) Uniform corrosion rate data of carbon steel in cementitious environments relevant to the supercontainer design: 'best estimate' from available literature data, supplemented with data coming from O/N funded R&D programmes. SCK•CEN report ER‐094, SCK•CEN, Mol, Belgium. Leach, J.S.L. and Nehru, A.Y. (1964) The temperature dependence of the corrosion of uranium in aqueous solutions. Journal of Nuclear Materials 13, 270‐274. Lee, D.J. and Wilding, C.R. (1989) The Corrosion of Aluminium in Cement – A Review of the Current Position. Report CPDG(89)P263. Lefebvre, F. and Lemaignan, C. (1997) Irradiation effects on corrosion of zirconium alloy claddings. Journal of Nuclear Materials 248, 268‐274. Macphee, D.E., Atkins, M. and Glassar, P.P. Phase Development and Pore Solution Chemistry in Ageing Blast Furnace Slag‐Portland Cement Blends. Materials Research Society Symposium Proceedings 127, 475‐480. Mansour, H., El‐cheikh, F.M., Moustafa, H.M. and El‐wafa, A. (1992) Corrosion of Al and Al‐Mg Alloys in Aggressive Media of HCl and NaOH Solutions. Trans. of the SAEST 27, 25.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 74 of 77 SCK•CEN/27548174 ISC: Public

Marcus, P. and Maurice, V. (2017) Atomic level characterization in corrosion studies. Philosophical Transactions of the Royal Society A 375, 2098. Marcus, P. and Oudar, J. (1995) Corrosion mechanisms in theory and practice. Marcel Dekker, New York. Martin, T.L., Coe, C., Bagot, P.A.J., Morrall, P., Smith, G.D.W., Scott, T. and Moody, M.P. (2016) Atomic‐ scale Studies of Uranium Oxidation and Corrosion by Water Vapour. Scientific Reports 6, 25618. Matsuo, T., Izumida, T., Hironaga, M., Horikawa, Y. and Shiomi, T. (1997) LiNO₃ Effect on Corrosion Prevention of Aluminum with Complex Shapes. Journal of Nuclear Science and Technology 34, 823‐828. Matsuo, T., Matsuda, M., Hironaga, M. and Horikawa, Y. (1996) Effect of LiNO3 on Corrosion Prevention of Aluminum Wastes after Their Land Disposal. Journal of Nuclear Science and Technology 33, 852‐862. Matsuo, T., Nishi, T., Izumida, T. and Matsuda, M. (1999) Influence of Increased Temperature from Cement Hydration on Aluminum Corrosion Prevention When LiNO3 Is Added to the Cement. Nuclear Technology 125, 332‐336. McGillivray, G.W., Geeson, D.A. and Greenwood, R.C. (1994) Studies of the kinetics and mechanism of the oxidation of uranium by dry and moist air A model for determining the oxidation rate over a wide range of temperatures and water vapour pressures. Journal of Nuclear Materials 208, 81‐97. McKee, A.B. and Brown, R.H. (1947) Resistance of Aluminum to Corrosion in Solutions Containing Various Anions and Cations. Corrosion 3, 595‐612. Mogoda, A.S. (1999) Electrochemical behaviour of zirconium and the anodic oxide film in aqueous solutions containing chloride ions. Thin Solid Films 357, 202‐207. Motta, A.T. (2011) Waterside corrosion in zirconium alloys. The Journal of the Minerals, Metals and Materials Society 63, 59‐63. Motta, A.T., Couet, A. and Comstock, R. (2015) Corrosion of zirconium alloys used for nuclear fuel cladding. Annual Review of Materials Research 45, 18.11‐18.33. Nishimura, T., Wada, R. and Fujiwara, K. (2003) Evaluation of gas generation rates caused by metal corrosion under the geological repository conditions. Kobe Steel Engineering Reports 53, 78‐83. Nuttall, P.A. and Curwen, S. (1990) An Assessment of the Effects of Uranium Fuel Carry Over on the Stability of Encapsulated Waste Products. Report PETF(90)P2. Olsson, C.‐O.A. and Landolt, D. (2003a) Passive films on stainless steels ‐ chemistry, structure and growth. Electrochimica Acta 48, 1093‐1104. Olsson, C.O.A. and Landolt, D. (2003b) Passive films on stainless steels ‐ chemistry, structure and growth. Electrochimica Acta 48, 1093‐1104. Onuchukwu, A.I. (1988) Corrosion inhibition of aluminum in alkaline medium. I: Influence of hard bases. Materials Chemistry and Physics 20, 323‐332. Orr, R., Godfrey, H., Broan, C., Goddard, D., Woodhouse, G., Durham, P., Diggle, A. and Bradshaw, J. (2016) Formation and physical properties of uranium hydride under conditions relevant to metallic fuel and nuclear waste storage. Journal of Nuclear Materials 477, 236‐245. Pathak, B.R. and Godard, H.P. (1968) Equation for predicting the corrosivity of natural fresh waters to Aluminium. Nature (London) 218, 893‐894. Postlethwaite, J. and Onofrei, M. (1979) Hydrogen evolution and anodic disintegration during electrochemical pitting of zirconium in alkaline‐chloride solutions. Corrosion 35, 185‐189. Poulard, K., Chevarier, A., Moncoffre, N., Trocellier, P. and Crusset, D. (2001) Study of zircaloy‐4 fuel cladding corrosion using ion beams. Application to long‐term disposal of nuclear wastes. Nuclear Instruments and Methods in Physics Research B 181, 640‐643. Pourbaix, A. (2003) Long term prediction of corrosion of passive materials for high‐level waste containers based on protection potential, in: Féron, D., MacDonald, D.D. (Eds.), Prediction of long term corrosion behaviour in nuclear waste systems. Maney, pp. 225‐237. Pourbaix, M. (1974) Atlas of electrochemical equilibria in aqueous solutions. NACE, Houston, Texas, USA.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 75 of 77 SCK•CEN/27548174 ISC: Public

Prabhu, D. and Rao, P. (2017) Corrosion behaviour of 6063 aluminium alloy in acidic and in alkaline media. Arabian Journal of Chemistry 10, S2234‐S2244. Pryor, M.J. and Keir, D.S. (1957) Galvanic Corrosion: I . Current Flow and Polarization Characteristics of the Aluminum‐Steel and Zinc‐Steel Couples in Sodium Chloride Solution. Journal of The Electrochemical Society 104, 269‐275. Pryor, M.J. and Keir, D.S. (1958) Galvanic Corrosion: II . Effect of pH and Dissolved Oxygen Concentration on the Aluminum‐Steel Couple. Journal of The Electrochemical Society 105, 629‐635. Punni, J.S. and Cox, M.J. (2010) The effect of impurity inclusions on the pitting behaviour of beryllium. Corrosion Science 52, 2535‐2546. Ritchie, A.G. (1981) A review of the rates of reaction of uranium with oxygen and water vapour at temperatures up to 300° c. Journal of Nuclear Materials 102, 170‐182. ROWE, L.C. and M.S., W. (1961) Effect of Mineral Impurities in Water On the Corrosion of Aluminum and Steel. Corrosion 17, 353t‐356t. Sakuragi, T., Miyakawa, H., Nishimura, T. and Tateishi, T. (2012) Corrosion Rates of Zircaloy‐4 by Hydrogen Measurement under High pH, Low Oxygen and Low Temperature Conditions. Materials Research Society Symposium Proceedings 1475, 311‐316. Sakuragi, T., Miyakawa, H., Nishimura, T. and Tateishi, T. (2013) Long‐term Corrosion of Zircaloy‐4 and Zircaloy‐2 by Continuous Hydrogen Measurement under Repository Condition. Materials Research Society Symposium Proceedings 1518, 173‐178. Schmuki, P. (2002a) From Bacon to barriers: a review on the passivity of metals and alloys. Journal of Solid State Electrochemistry 6, 145‐164. Schmutz, P. and Landolt, D. (1999) In‐situ microgravimetric studies of passive alloys: potential sweep and potential step experiments with Fe‐25Cr and Fe‐17Cr‐33Mo in acid and alkaline solution. Corrosion Science 41, 2143‐2163. Sedriks, A.J. (1996) Corrosion of stainless steels. Wiley‐Interscience, John Wiley & Sons, Inc. Setiadi, A., Milestone, N.B., Hill, J. and Hayes, M. (2006) Corrosion of aluminium and magnesium in BFS composite cements. Advances in Applied Ceramics 105, 191‐196. Smart, N.R., Blackwood, D.J., Marsh, G.P., Naish, C.C., O'Brien, T.M., Rance, A.P. and Thomas, M.I. (2004) The anaerobic corrosion of carbon and stainless steels in simulated cementitious repository environments: a summary review of NIREX research. AEA Technology, Harwell. Report AEAT/ERRA‐0313. Smart, N.R., Rance, A.P., Fennell, P.A.H. and Kursten, B. (2014) Effect of sulphur species on anaerobic corrosion of carbon steel in alkaline media. Corrosion Engineering, Science and Technology 49, 473‐ 479. Solbrig, C.W., Krsul, J.R. and Olsen, D.N. (1994) Pyrophoricity of Uranium in Long‐term Storage Environments, Spent Nuclear Fuel Meeting: Challenges and Initiatives. Argonne National Lab. Report ANL/TD/CP‐84441. Soltis, J. (2015) Passivity breakdown, pit initiation and propagation of pits in metallic materials ‐ Review. Corrosion Science 90, 5‐22. Special Metals. (2000) High‐performance alloys for resistance to aqueous corrosion: wrought nickel products. Special Metals Corporation. Special Metals Corporation, SMC‐026. Stitt, C.A., Harker, N.J., Hallam, K.R., Paraskevoulakos, C., Banos, A., Rennie, S., Jowsey, J. and Scott, T.B. (2015a) An Investigation on the Persistence of Uranium Hydride during Storage of Simulant Nuclear Waste Packages. PLOS ONE 10, e0132284. Stitt, C.A., Hart, M., Harker, N.J., Hallam, K.R., MacFarlane, J., Banos, A., Paraskevoulakos, C., Butcher, E., Padovani, C. and Scott, T.B. (2015b) Nuclear waste viewed in a new light; a synchrotron study of uranium encapsulated in grout. Journal of Hazardous Materials 285, 221‐227. Stitt, C.A., Paraskevoulakos, C., Harker, N.J., Jones, C.P. and Scott, T.B. (2015c) The effects of metal surface geometry on the formation of uranium hydride. Corrosion Science 98, 63‐71. Straumanis, M.E. and Brakšs, N. (1949a) The Rate of Solution of High Purity Aluminum in Various Bases. Journal of The Electrochemical Society 96, 21‐26. Straumanis, M.E. and Brakšs, N. (1949b) The Rate of Solution of Highest Purity Aluminum in Sodium Hydroxide Solutions. Journal of The Electrochemical Society 95, 98‐106.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 76 of 77 SCK•CEN/27548174 ISC: Public

Suman, S., Khan, M.K., Pathak, M., Sing, R.N. and Chakravartty, J.K. (2015) Hydrogen in Zircaloy: mechanism and its impacts. International Journal of Hydrogen Energy 40, 5976‐5994. Sverepa, O. (1958) Korrosion des Aluminiums und seiner Legierrungen in Wässern verschiedener Zusammensetzung. Werkstoffe und Korrosion 9, 533‐536. Szklarska‐Smialowska (1986) Pitting corrosion of metals. National Association of Corrosion Engineers, Houston, Texas. Tabrizi, M.R., Lyon, S.B., Thompson, G.E. and Ferguson, J.M. (1991) The long‐term corrosion of aluminium in alkaline media. Corrosion Science 32, 733‐742. Takatani, Y., Yamakawa, K. and Yoshizawa, S. (1982) Corrosion Behaviour of Aluminium in Alkaline Solution. Journal of the Society of Materials Science, Japan 31, 1024‐1030. Takatani, Y., Yamakawa, K. and Yoshizawa, S. (1983) Corrosion Behaviour of Aluminum Alloys in Saturated Calcium Hydroxide Solution. Journal of the Society of Materials Science, Japan 32, 1218‐ 1222. Thomas, M.I. and Naish, C.C. (1994) Corrosion of Aluminium in Replacement Cement Grouts at Elevated Temperatures: Further Tests. Report APEC(94)P166. Thomas, M.I., Rance, A.P. and Naish, C.C. (1994) Galvanic Corrosion of Uranium and Aluminium in Cement Grouts at Elevated Temperatures. Report APEC(94)P170. Tyfield, S.P. (1988) Corrosion of reactor grade uranium in aqueous solutions relevant to storage and transport. Nuclear Energy 27, 91‐98. Vujičić, V. and Lovreček, B. (1985) A Study of the influence of pH on the corrosion rate of aluminium. Surface Technology 25, 49‐57. Wada, R. (1999) Experimental study on hydrogen gas generation rate from corrosion of zircaloy and stainless steel under anaerobic alkaline conditions, 7th International Conference on Radioactive Waste Management and Environemental Remediation (ICEM'99), Nagoya, Japan. Walkner, S., Hafner, M. and Hassel, A.W. (2016) Film dissolution kinetics of aluminium at precisely adjusted pH‐values. physica status solidi (a) 213, 1410‐1416. Walters, W.S. (1997) The effect of Radiolysis on the Corrosion of Radioactive Waste Containers. AEA Technology, AEAT‐0374. Walton, C.J., F.L., M. and E.T., E. (1957) Compatibility of Aluminum With Alkaline Building Products. Corrosion 13, 41‐50. Wanklyn, J.N. and Jones, P.J. (1962) The aqueous corrosion of reactor metals. Journal of Nuclear Materials 6, 291‐329. Ward, J.W. and Waber, J.T. (1962) The Electrochemical Behavior of Uranium. Journal of The Electrochemical Society 109, 76‐81. White, J. (1987) Generic Studies on the Encapsulation of Aluminium in Cement. Report PETF(87)P61. Wilkinson, W.D. (1962) Uranium Metallurgy. Interscience Publishers, New York, London. Yan, C., Wang, R., Wang, Y., Wang, X. and Bai, G. (2015) Effects of ion irradiation and properties of zirconium alloys ‐ a review. Nuclear Engineering and Technology 47, 323‐331.

Corrosion of metals in category B&C waste in highly alkaline, anoxic conditions Page 77 of 77 SCK•CEN/27548174 ISC: Public