&STUK The influence of modified water chemistries on metal ...

FI9900103&STUKS T U K - Y T O - T R 1 52 March 1999

The influence of modifiedwater chemistries onmetal oxide films,activity build-up andstress corrosion crackingof structural materialsin nuclear power plants

3 0 - 2 4

K. Makela, T. Laitinen, M. BojinovVTT Manufacturing Technology

In STUK this study was supervised by Seija Suksi

S T U K » S A T E I L Y T U R V A K E S K U S • S T R A L S A K E R H E T S C E N T R A L E NR A D I A T I O N A N D N U C L E A R S A F E T Y A U T H O R I T Y

The conclusions presented in the STUK report series are those of the authorsand do not necessarily represent the official position of STUK.

ISBN 951-712-295-0ISSN 0785-9325

Oy Edita Ab, Helsinki 1999

STUK-YTO-TR 1 52

MAKELA, Kari, LAITINEN, Timo, BOJINOV, Martin (VTT Manufacturing Technology). The influenceof modified water chemistries on metal oxide films, activity build-up and stress corrosion cracking ofstructural materials in nuclear power plants. STUK-YTO-TR 152. Helsinki 1999. 55 pp.

ISBN 951-712-295-0ISSN 0785-9325

Keywords: oxide films, high temperature aqueous environments, steels,activity incorporation, zinc water chemistry, noble metal coatings

ABSTRACTThe primary coolant oxidises the surfaces of construction materials in nuclear power plants. Theproperties of the oxide films influence significantly the extent of incorporation of activated corrosionproducts into the primary circuit surfaces, which may cause additional occupational doses for themaintenance personnel. The physical and chemical properties of the oxide films play also an importantrole in different forms of corrosion observed in power plants.

This report gives a short overview of the factors influencing activity build-up and corrosion phenomenain nuclear power plants. Furthermore, the most recent modifications in the water chemistry to de-crease these risks are discussed. A special focus is put on zinc water chemistry, and a preliminarydiscussion on the mechanism via which zinc influences activity build-up is presented. Even though theexact mechanisms by which zinc acts are not yet known, it is assumed that Zn may block the diffusionpaths within the oxide film. This reduces ion transport through the oxide films leading to a reducedrate of oxide growth. Simultaneously the number of available adsorption sites for 60Co is also reduced.

The current models for stress corrosion cracking assume that the anodic and the respective cathodicreactions contributing to crack growth occur partly on or in the oxide films. The rates of these reactionsmay control the crack propagation rate and therefore, the properties of the oxide films play a crucialrole in determining the susceptibility of the material to stress corrosion cracking.

Finally, attention is paid also on the novel techniques which can be used to mitigate the susceptibility ofconstruction materials to stress corrosion cracking.

STUK-YTO-TR 1 52

MÄKELÄ, Kari, LAITINEN, Timo, BOJINOV, Martin (VTT Valmistustekniikka). Modifioitujenvesikemioiden vaikutus metallioksidifilmeihin, aktiivisuuden kerääntymiseen ja rakennemateriaalienjännityskorrosioon ydinvoimalaitoksissa. STUK-YTO-TR 152. Helsinki 1999. 55 s.

ISBN 951-712-295-0ISSN 0785-9325

Keywords: oksidifilmi, korkealämpötilavesi, teräs, aktiivisuuden kerääntyminen,sinkkivesikemia, jalometallipinnoite, dielektrinen pinnoite

TIIVISTELMÄYdinvoimalaitosten primääripiirin rakennemateriaalien pinnat hapettuvat jäähdytysveden vaikutuk-sesta. Syntyvien oksidifilmien ominaisuudet vaikuttavat merkittävästi radioaktiivisten osaslajien ke-rääntymiseen primääripiirin pinnoille, mikä taas vaikuttaa henkilökunnan saamiin säteilyannoksiinhuoltoseisokkien aikana. Oksidifilmien fysikaalisilla ja kemiallisilla ominaisuuksilla on myös suurivaikutus eri korroosioilmiöihin ydinvoimalaitoksissa.

Tässä kirjallisuustyössä on esitetty lyhyt yhteenveto tekijöistä, jotka vaikuttavat aktiivisuuden ke-rääntymiseen ja tiettyihin korroosioilmiöihin ydinvoimaloissa. Lisäksi työssä käsitellään uusimpiatapoja modifioida voimaloiden vesikemiaa tarkoituksena pienentää edellä mainittujen ilmiöiden riskiä.Sinkkivesikemiaa on painotettu erityisesti. Vaikka sinkin tarkka vaikutusmekanismi aktiivisuudenkerääntymisen pienentämisessä on tuntematon, sinkin voidaan olettaa tukkivan diffuusioreittejä oksi-difilmissä. Tämä hidastaa ionien kuljetusta filmissä ja johtaa hitaampaan filmin kasvuun. Samallamyös saatavilla olevien 60Co:n adsorptiopaikkojen määrä pienenee.

Nykyisissä jännityskorroosiomalleissa oletetaan, että särönkasvuun vaikuttavat anodiset ja vastaavatkatodiset reaktiot tapahtuvat osittain oksidifilmeissä tai niiden pinnoilla. Näiden reaktioiden nopeudetvoivat kontrolloida särönkasvunopeutta, joten oksidifilmien ominaisuuksilla voi olla ratkaiseva merki-tys materiaalien jännityskorroosiokestävyyden kannalta. Tässä työssä käsitellään uusimpia tekniikoi-ta, joiden avulla materiaalien jännityskorroosiokestävyyttä voidaan parantaa muuttamalla metallienpinnoille muodostuvien oksidifilmien sähköisiä ominaisuuksia.

STUK-YTO-TR 1 52

CONTENTS

ABSTRACT 3TIIVISTELMA 4ACKNOWLEDGEMENTS 6

1 INTRODUCTION 7

2 WATER CHEMISTRIES IN LIGHT WATER REACTORS 8

3 OXIDE FILMS ON THE CONSTRUCTION MATERIALS IN NUCLEAR POWER PLANTS 12

3.1 Structure of metal oxide films in light water reactors 123.2 Oxide films on stainless and carbon steels in BWRs 133.3 Oxide films on alloyed steels and nickel based alloys in PWRs 14

4 OXIDE FILMS AND MECHANISMS OF ACTIVITY INCORPORATION AND STRESSCORROSION CRACKING 154.1 Activity incorporation 154.2 Oxide films and stress corrosion cracking (SCC) 17

5 NOVEL WATER CHEMISTRIES 185.1 Zinc water chemistry 18

5.1.1 Zinc water chemistry in BWRs 195.1.2 Zinc water chemistry in PWRs 215.1.3 Effect of zinc on activity incorporation into oxide films 225.1.4 Correlation between role of Zn and oxide structure 255.1.5 Zn water chemistry and stress corrosion cracking 325.1.6 Detrimental effects of zinc 355.1.7 Alternatives for zinc injection 37

5.2 Noble metal water chemistry 385.2.1 Principles of noble metal water chemistry 385.2.2 Different types of noble metal coatings 405.2.3 Long term stability 425.2.4 Effect of the operational environment 425.2.5 Possible side effects of noble metal coating 43

5.3 Application of dielectric oxides on construction materials 43

6 SUMMARY AND CONCLUSIONS 46

REFERENCES 48

STUK-YTO-TR 1 52

ACKNOWLEDGEMENTS

The authors are grateful to the Radiation and Nuclear Safety Authority (STUK), Teollisuuden VoimaOy (TVO), Imatran Voima Oy (IVO), the Ministry of Trade and Industry (KTM) and OECD HaldenReactor Project for the funding of this project.

STUK-YTO-TR 1 52

1 INTRODUCTION

The interaction between the construction materi-als and the aqueous coolant leads to the oxidationof the primary circuit piping surfaces in nuclearpower plants. As a result of the susceptibility ofmaterials to oxidise, different forms of corrosionmechanisms may pose serious hazards to the ope-ration of the plant. Oxide films formed on materialsurfaces play an important role in uniform corro-sion attack, as well as on localised corrosion orstress corrosion cracking. In addition, the proper-ties of oxide films influence the extent of incorpo-ration of active species into the primary circuitmaterials, which may result in increased occupa-tional doses of radiation for the personnel of theplant.

The main objective of radiation field control ina nuclear power plant is to maintain personnelradiation exposures as low as reasonable achieva-ble (ALARA). The first attempts to achieve expo-sure savings were mainly obtained by minimisingthe time spent by workers in radiation fieldsduring maintenance, inspections and refuelling.Another approach is to develop and modify thewater chemistry of the plant and thus achieveconditions, under which risks for activity build-upand simultaneously for detrimental corrosion phe-nomena are minimised. The report by Interna-tional Atomic Energy Agency (IAEA) co-ordinatedresearch program entitled "Investigation on WaterChemistry Control and Coolant Interaction withFuel and Primary Circuit Materials in WaterCooled Power Reactors (WACOLIN)" summariesthe present philosophy on good coolant chemistry

as follows: "Good reactor coolant chemistry, corro-sion control and minimum of activity build-up areindispensable for the optimum performance ofnuclear power plants. Without these the systemintegrity may be jeopardised and the activitytransport may create various problems."[l]

In order to maintain good coolant chemistry,extensive water chemistry guidelines have beendeveloped for the pressurised and boiling waterreactors (PWR and BWR). Properly controlledwater chemistry during steady state operationand shutdowns has led to low corrosion rates ofconstruction materials. Nevertheless, some plantdata have shown that despite of following thestrict water chemistry guidelines, certain undesir-able phenomena, such as increased activity build-up on the primary loop piping surfaces and stresscorrosion cracking of in-core components, can stilloccur. Although further developments in waterchemistry in NPPs, e.g. zinc water chemistry,have recently been introduced, it is evident that aproper understanding of the interaction of thecoolant and the oxide films on material surfaces inNPPs is not yet available.

The aim of this report is to give a short over-view of the factors influencing activity build-upand corrosion phenomena in nuclear power plants.Moreover, the most recent modifications of waterchemistries to decrease these risks are discussed.A special focus is on zinc water chemistry, and apreliminary discussion on the mechanism viawhich zinc influences activity build-up is alsoreviewed.

STUK-YTO-TR 1 52

2 WATER CHEMISTRIES IN LIGHT WATERREACTORS

Thermal reactors are most commonly used in nuclear power plants. They arecategorised according to the coolant, which is used to cool down the reactoritself and also according to the medium which slows down the neutronsformed during the fission reaction. The most common thermal reactor typesare the light water reactors which can be either boiling or pressurised waterreactors. There are some fundamental differences in the water chemistries ofthese two types of reactors and therefore a short introduction is given infollowing chapters.

2.1 Boiling Water reactors (BWR) tive than -O.23O VSHE in high purity water (con-

Steam/water cycle is essentially a closed loop inboiling water reactors (BWR). The water is heatedin the reactor core, where a fraction of it is con-verted into steam. In the upper part of the reactorcore steam passes through the steam drier and istransported in main steam line to high and lowpressure turbines. Steam starts to condense alrea-dy in the turbines but condensation mainly occursin the condensers. After the condensers the wateris cleaned by demineralisers and is fed back to there-circulation loop. To minimise the corrosion pro-duct deposition on the fuel cladding surfaces, im-purity concentrations in the water circulating inBWRs is kept as low as possible. Therefore, thewater conductivity is typically below 0.13 nScm-1

during steady-state operation.Despite of the strict water chemistry guide-

lines, sensitised microstructure of the primarycircuit stainless steels, coupled with residualstresses produces susceptibility to stress corrosioncracking (SCC) in the presence of oxidising spe-cies in the coolant.[2] The sensitised microstruc-tures and stresses can not be eliminated in exist-ing plants and therefore, the obvious remedy toprevent intergranular stress corrosion cracking(IGSCC) is to optimise the operational environ-ment further. Laboratory tests and actual powerplant measurements indicate that to prevent crackformation and propagation in the BWR stainlesssteel parts, the corrosion potential (ECP) of pri-mary loop components has to be kept more nega-

ductivity less than 0.3 uScrcr1). Despite the ratherhigh volatility of dissolved oxygen and hydrogenperoxide, their contents under normal waterchemistry (NWC) conditions in BWR coolant arestill about 100-300 ppb (ugkg-1). These oxidisingspecies can shift the corrosion potential of stain-less steel parts up to a value of+0.150 VSHE, thussupporting stress corrosion cracking. On the otherhand, if the water contains chromates or otherpowerful oxidative impurities, ECP may reachcritical values even though O2 concentrations areas low as 5 ppb.[3] In addition to chemical compo-sition of the coolant, the water flow rate has asignificant impact on the corrosion potentials ofconstruction materials as shown in Figure 1.

A possible remedy for the risks caused by thedissolved oxygen and hydrogen peroxide can besuppressed by hydrogen additions to the feedwater (hydrogen water chemistry, HWC).[5] HWCwas first introduced to protect re-circulating pip-ing welds, but is more recently used to protectinternal structures and welds in pressure vessels.The amount of hydrogen needed to decrease thecorrosion potential below the threshold potentialfor SCC is plant- and site-specific depending onthe coolant flow rate, construction of the core anddose rate in the down comer. [6]

As a drawback some plants have experiencedan increase in the 16N concentration in mainsteam during the HWC operation, causing elevat-ed dose rates in turbine building. Under normalwater chemistry (NWC) operation 16N forms pri-

S T U K - Y T O - T R 152

marily N03 , which remains in water phase,whereas under more reducing conditions NH3 isformed. NH3 is volatile, gets easily into the steamphase and is transported into the turbines.[7]

Another side effect of hydrogen water chemis-try is a significant increase in shutdown doserates which has been observed in some boilingwater reactors. This phenomena is related to thedeposition of the corrosion products on fuel clad-ding surfaces where they become radioactive. Thesubsequent dissolution of these activated corro-sion products and incorporation into the oxidelayers on the out-of-core surfaces of the primarycircuit is the major source for activity build-up. Ifthe BWR plant operating under NWC starts toinject hydrogen into the reactor water, the struc-tures of the oxide films will change to adopt intothe new environment. This has led to increasedactivity pickup of the oxide films in such plants. Inaddition, due to operational realities, hydrogeninjection system is not always on-line, which cre-ates a periodic cycling between the reducing(HWC) and oxidising (NWC) conditions, which hasshown to further increase 60Co incorporation intothe oxides.

2.2 Pressurised water reactors(PWR)

A pressurised water reactor consists of primaryand secondary sides. The primary side operates inconditions under which the water passing thereactor core does not boil. This is obtained bymaintaining high enough pressure in the primarycircuit. The heated water passes through the tu-bes of the steam generator, transferring its heat tothe secondary side to produce steam, which drivesthe turbines. After the turbines water is conden-sed and returned by feed water lines back to thesteam generator.

Due to the low oxidation rate of the primaryloop surfaces and low rate of transport of ionsthrough the existing oxide films, corrosion productconcentrations in the coolant during the steady-state operation are low. The solubility studies ofdifferent metal oxides have shown that releaserates of corrosion products are material specificand vary also as a function of temperature. There-fore, an improved analysis and a narrow rangecontrol of primary water chemistry, especiallypHT, in PWRs can be successful in reducing the

0.4

02.

Range o( LaboratoryMeasurements

-0.8 _ 3.89 x K 1.95 x 10s

1ppt> 100 ppO 1 ppm 10 pfxn

Figure 1. The effect of flow rate (expressed in Reynolds numbers) on the corrosion potential of stainlesssteel as a function of oxygen and hydrogen peroxide content.[4J.

STUK-YTO-TR 1 52

activity build-up.[35,9] The control of pHT in thePWR primary circuit systems is a rather compli-cated task. Boric acid is used as a chemical shimto control nuclear reactivity resulting in a need toadjust pHT by adding lithium or potassium hy-droxide to the primary coolant. During the 1980'smost plants kept the primary coolant pH300 «c at6.9 (called co-ordinated Li-B chemistry), based onthe magnetite solubility studies by Sweeton et al,Tremaine et al. [10, 11, 12, 13] These studiesshowed that 6.9 was the optimum pHTto preventdeposition of dissolved corrosion products on hotcore surfaces due to the positive temperaturecoefficient of solubility. The oxide examinations inpower plants have confirmed that as a result ofusing this pHT value, the deposition on hot coresurfaces has been reduced.[10] However, morerecent solubility studies have shown that theminimum pHT for a positive temperature coeffi-cient ranges from 6.9 to 7.7. Despite of the largescatter of the data, all the solubility studies indi-cate that an optimum pH300 oC is greater than 6.9.Therefore, most of the power plants have startedto use so called modified water chemistry, inwhich a steady-state pH300 oC of 7.2 is achieved assoon as possible after start-up. This pHT is keptconstant throughout the fuel cycle without exceed-ing Li concentration of 2.2 mgkg-1 to avoid en-hanced fuel cladding corrosion.[ll,16]

Incorporation of activated corrosion productsinto the growing oxides can be further decreasedby pre-conditioning the replacement parts (like inthe primary side channel heads of new steamgenerators, SG) by using electropolishing or chro-mium passivation techniques. Both techniqueshave shown their effectiveness in plant applica-tions (Millstone Point-2 and Doel) by decreasingthe surface activities by a factor of two. These twotechniques can be easily applied on the new partsto be installed into existing nuclear power plants.However, replacement of large components in op-erating plants is not done often. Therefore, the

optimisation of a plant specific water chemistrymay be a more useful way to proceed in operationpower plants. One way to optimise water chemis-try is to inject a blocking agent into the primarycoolant, which decreases incorporation of activityinto the oxides. This topic is discussed in moredetail in chapters below.

In addition to pHT control, the PWR primarychemistry guidelines state that the hydrogen con-tent should be kept between 35-55 mlkg-1 tominimise the concentration of oxidising speciescreated by radiolysis of water and to reduce thepossible traces of oxygen in the make-up water.Because the PWR primary coolants contain highconcentrations of dissolved hydrogen, the corro-sion potentials of in-core construction materialsare likely to be low enough to prevent stresscorrosion cracking to occur in thermally sensitisedstainless steels observed in BWRs. On the otherhand, the use of high hydrogen levels in theprimary circuit of PWRs is believed to be a possi-ble cause of cracking of Inconel 600 steam genera-tor tubes (primary water stress corrosion crack-ing, PWSCC) and other components manufac-tured from nickel-based alloys. If there is a con-nection between oxide film properties andPWSCC, then both pHT and hydrogen concentra-tion in the coolant should also have an impact onPWSCC.[19] The effects of these two chemistryparameters on ion solubilities from the syntheticNi0 50Co0 05Fe2 45O4, are shown in Figure 2. As thepHT increases the solubility of iron and cobaltdecrease, whereas the solubility of iron and cobaltfrom the oxide film increases when increasing thehydrogen concentration. The dissolution of theoxide film partly affects the extent of ion transportthrough the film. Therefore, the dissolution andgrowth of oxide film has an effect on the vacancyproduction in the base metal. How this phenome-na is related to SCC of construction materials, isdiscussed below.

10

STUK-YTO-TR 1 52

1.0

cr.TOU 0.1

oCM

-2- 0.01

©

0.001

6.25

PH,

6.35 6.47

450 500 550

Temperature

6.74

600

Figure 2. Average solubility of iron and cobalt from synthetic Ni0 S0Co0 0SFe2 4B04 as a function of temper-ature, pHTand different hydrogen concentrations.[14]

11

STUK-YTO-TR152

3 OXIDE FILMS ON THE CONSTRUCTIONMATERIALS IN NUCLEAR POWER PLANTS

The properties of oxide films of construction materials influence the behaviour ofthe components in operational conditions. The structure of the oxide film determi-nes how different ions are transported into and through the oxides. However,different types of operation environments leads to the formation of different typesof oxide film compositions. Therefore, to understand the phenomena occurring onand within the oxide films the most typical oxide film structures are presented inthe next chapter.

3.1 Structure of metal oxide filmsin light water reactors

From a thermodynamic point of view, the stabilityof an oxide depends on the temperature, pH, oxidi-sing/reducing power of the environment as well asthe type and concentration of dissolved ions in thesolution. However, an oxide growing on the metalsurface cannot be considered as a system in ther-modynamic equilibrium. Its properties and com-position are to a great extent affected by the ki-netic factors determining its growth rate and sta-tionary state thickness. This leads often to theformation of a film in which the composition andstoichiometry changes gradually with distancefrom the film/environment interface. The oxidefilms formed at high temperatures in variouskinds of environments generally consist of a com-pact inner layer and of a more porous outer layer.Although this duplex film concept is certainly asimplification, it has proven to be useful in themodelling of the behaviour of the film.

The inner layer consists of fine-grained oxidebecause it grows in a confined space. The outerlayer consists of loosely packed, larger grainsbecause it grows without volume constraint. Theboundary between the layers has been found to lieat the position of the original metal surface, whichindicates that the inner layer grows at the metal/oxide interface and that the outer layer grows atthe oxide/solution interface. It is possible to dividethe duplex oxide film further into different sub-layers, in which the composition and stoichiome-try change gradually with distance. [40,41] Thealloying elements are distributed into these two

layers differently depending on the primary cool-ant conditions (pHT, hydrogen concentration, tem-perature, etc.) as well as their original concentra-tion in the corroding metal. In addition, the differ-ent transport rates of metal ions (Fe2+ > Co2+ >Ni2+ » Cr 3+) determine their distribution withinthe oxides. Slower moving ions tend to be retainedin the inner part of the duplex oxide, increasingthe chromium content of the inner oxide.

The outer oxide film is partly formed by deposi-tion via the coolant as a result of a high concentra-tion of dissolved species in the bulk solution and—even more likely—a flux of cation species throughthe oxide film to the solution.[34]. This outermostpart of the oxide film has a porous structure,which consists of non-uniform crystallite agglom-erates. These crystallites can form as thick layersor exist as single crystallites at scattered sitesdepending whether mass transport in the systemfavours re-precipitation. The coolant flow velocity,pH, temperature, water saturation and presenceof reducing or oxidising agents influence mainlythe properties and behaviour of this outer oxidelayer.[37,38,39] Therefore, this part of the oxidefilm changes most significantly, when the opera-tional environment is changed from oxidising toreducing conditions. Under oxidising conditions,the outer oxide film is either a-Fe2O3 (hematite) ory-Fe2O3 (maghemite) independently of the con-struction material in question, but it stronglydepends on the water chemistry. Under reducingconditions, the film changes into magnetite type ofspinel. It is possible that below the deposited partof the porous film is formed as a result of break-down of the inner oxide layer.

12

STUK-YTO-TR 1 52

On stainless steels and nickel based alloys, theinner, dense part of the duplex oxide film consistsof Cr-rich layer, whereas on carbon steels theinner layer is composed of magnetite. This part ofthe oxide film is often called grown-on oxide,because it grows at the metal/oxide interface. Amore comprehensive view can be obtained, if thispart of the oxide is divided into two different oxidelayers. The composition of the oxide film next tothe base metal, does not change due to the chang-es in operational environments. It is covered bythe outer oxide film layers. Therefore, the struc-ture of the film remains similar in both reducingand oxidising conditions. However, it is possiblethat the composition of the outer part of the innerdense oxide has lower chromium concentrationunder oxidising environments than under reduc-ing conditions. This is due to higher chromiumoxidation rate at higher corrosion poten-tials.[34,35,36] The formation of the Cr-rich spi-nels in the dense part of the duplex oxide has beenreported to result in a low corrosion rate of stain-less steels when compared to the oxidation rate ofcarbon steels. The overall corrosion rate of stain-less steels is likely to be controlled totally by iontransport through this dense oxide layer and isnot influenced by fluid flow conditions, by thepresence or absence of the outer layer, nor by thesaturation degree of the solution. However, thecurrently used models still need experimentalsupport, specially the preferential paths and driv-ing forces for ion transport through the oxidefilms, as well as the nature of mobile species/defects.[34,35,36]

3.2 Oxide films on stainless andcarbon steels in BWRs

To understand the corrosion processes in boilingwater environments, the characteristics of oxidefilms on AISI 316 SS and AISI 304 SS have beeninvestigated thoroughly. As discussed above, typi-cally these films have a duplex structure. The filmcomposition of the outer part formed under NWC,i.e. under oxidising conditions, differs from thatformed under reducing conditions (HWC). Normalwater chemistry in BWR conditions leads to theformation of Y-Fe2O3,

a-Fe2O3 anc* NiFe2O4 (orNixFe3_xO4) in the outer part of the duplex oxidefilm on AISI 316 SS, while hydrogen water che-mistry results in the formation of Fe3O4 andNiFe2O4 (or NixFe3_xO4). A typical oxide composi-tion in different operating environments is shownin Figure 3. Pores are likely to be present in theouter part of the duplex oxide film, but to a lesserextent in the inner, Cr rich oxide layer.

The inner part of the duplex oxide films on Fe-Cr-Ni alloys may also become partly depleted inchromium if high enough oxygen concentrationsexist in the coolant. This behaviour is due to theoxidation of Cr(III) to soluble Cr species.[44]

The composition of the oxide film is rathercomplicated because, depending on the environ-ment the outer part of the inner, the dense oxidenext to the base metal is either NixFeyCr3_x_yO4 orNixFe3_xO4. It is hypothesised by Asakura et al.[40]that this lowest part of the porous oxide film isformed by the breakdown of the inner layer, whichoccurs at oxygen concentrations > 100 ppb. This

Under NWC:Large Particles= a .<w y-FeSmall Parlicles= a-Fe2OaUnder HWC:Panicles=Mey Me* Fea-x-y

Grain Boundarlcs-Porcs ModelJ. Robertson. "The Mechanism of High TcrnperatvAqueous Corrosion of Stainless SieeU."Corrosion Sclence.Vel. 32. No. 4 (March 1991).

JAt liquid layerinterfacey - FeOOH

3T*£A I Under HWC^ - i —-j MeY Me , Crj.^.yO** . I Under NWC

Under HWC or NWC

Figure 3. A schematic diagram of the oxide films formed on AISI 316 SS in BWR water under differentconditions.[42]

13

STUK-YTO-TR 1 52

agrees well with the model presented by Her-mansson for films grown on stainless steels inBWR coolant under normal water chemistry con-ditions.[40,41] However, the part of the chromiumrich inner oxide, which is next to the base metal,is not affected by the primary coolant chemis-try.[8,43]

3.3 Oxide films on alloyed steelsand nickel based alloys inPWRs

Corrosion of stainless steel in PWR coolant, i.e. inreducing conditions, results also in the formationof a duplex oxide film on the metal surface if thewater above the growing oxide is saturated withcorrosion products. On stainless steels the inneroxide layer consists of a chromite (NixFe(1_x)Cr204),usually containing some elemental nickel, whilethe outer layer is a non-stoichiometric nickel ferri-te (NixFe(3_x)O4). On top of this nickel ferrite layer,new single crystals of nickel ferrite can form inc-reasing the total oxide thickness.[45,46,47] Simi-lar duplex oxide films form also on Incoloy 800 (I-800) surfaces. The similarities in the oxide filmcompositions are not a surprise due to the smalldifferences in base metal compositions.

The oxide films on nickel-based alloys, Inconel600 (1-600) and Inconel 690 (1-690), are alsoformed via the same mechanisms as those onstainless steels. The majority of the chromium isretained in the inner oxide film, whereas iron andnickel are mainly present in the outer oxide layer.However, there are some distinct differences inthe distribution of ions in the oxides forming onthese materials in the high temperature water asdiscussed below.[48]

The oxides on Inconel alloys can not be solelyspinels, because the fixed valences of Ni and Cr(in PWR environments) can only form a spinel offixed composition NiCr2O4, whereas the II and IIIvalences of iron allow it to form a continuousseries of spinels. Thus, according to the model ofRobertson, the inner oxide layer on 1-600 mayconsist of NiCr2O4 (48%) and NiO (52%), while theouter layer probably consists of NiO with a smallquantity of NiFe2O4.[34] The presence of a sub-stantial proportion of NiO throughout the oxidefilm allows the oxidation rate to be controlled bythe faster growth rate of NiO rather than theslower oxidation rate of Ni,Cr spinel. Therefore,Cr will not lower the non-selective oxidation rateof Ni-Cr alloys until the Cr content increases to33%, since there will always be some NiO phasepresent to short circuit the Cr spinel.[34]

The higher Cr content in 1-690 when comparedto 1-600 can lead to a higher concentration ofNiCr2O4 in the inner oxide layer with little NiOincorporation whereas the outer layer shouldmainly be NiO. The presence of nickel oxide in theinner layer will control the corrosion rate, but therather low concentration should reduce the oxidegrowth rate considerably below that of 1-600.[34]

However, due to the high hydrogen concentra-tion in the primary circuits of PWRs, NiO is notthermodynamically stable. [46] Schuster et al.have shown that the prevailing oxidation state ofnickel was found to be Ni°, except for a fewnanometers at the oxide/solution interface.[49]Furthermore, the water chemistry environmentswithin the cracks and pores differ significantlyfrom the bulk solution conditions, which makes itdifficult to estimate the stabilities of the oxides.

14

STUK-YTO-TR 1 52

4 OXIDE FILMS AND MECHANISMSOF ACTIVITY INCORPORATION ANDSTRESS CORROSION CRACKING

4.1 Activity incorporation

The growth of oxide films is influenced by theextent of saturation of the coolant next to the oxi-de layer by corrosion products and by the rate ofion transport through the existing oxide. The de-posited corrosion products simultaneously in-cor-porate the activated corrosion products into theoxide films. Additional activity incorporation canoccur through deposition of suspended particleson the top of these films. Laboratory tests haveshown that 60Co incorporation into the oxide filmsis directly proportional to the soluble 60Co concent-ration in the coolant. The accumulation rate intothe outer layer increases linearly with the exposu-re time.

The basic mechanisms of the incorporation ofradioactive cobalt into oxide films on iron andnickel-based materials, involve soluble metal ions.The incorporation can basically proceed via atleast three different mechanisms:[52]

• surface adsorption/re-crystallisation• ion-exchange• direct reaction/crystallisation

The adsorption of radioactive cobalt species on theouter part of the duplex oxide film will result inthe formation of nickel ferrite with a basic formulaof (CoxNiyFe(1_x_y)O4). The reaction can be initiatedby an exchange reaction between a cobalt ion anda surface hydroxyl group of the oxide. Two pos-sible reaction routes have been suggested:Co** + -FeO-H = FeO-Co+ + H+ (1)Co2- + 2 H2O = CoOH+ + H3O (2a)CoOH+ + -FeO-H = FeO-Co+ + H,0 (2b)

A scheme of the adsorption reaction (1) is shownin Figure 4. The adsorption equilibrium can bedescribed by means of the equation:

E_[FeO-Co+-]-[H*][Co2+]-[-FeO-H]

ADSORPTION/RECRYSTALLIZATION:

ION-EXCHANGE REACTION:

Fe Fe

O O O

Fe Fe 2 " +

Fe

\

/1 \O O O

\l/Fe

Figure 4. Illustration of the adsorption and ion-exchange mechanisms for activity incorporation intooxide films.[52]

15

STUK-YTO-TR 1 52

The value of K increases about 600-fold with anincrease of temperature from 20 to 285 °C.

It has been suggested that cobalt adsorptiontakes place only at sites in which the -FeO-Hgroup is available. The rate of Co adsorption isrelatively slow compared to the rate of directreaction/crystallisation. Its rate may increase sig-nificantly with increasing the pH and the porosityof the film. [52]

Kelen and Hermansson have discussed theapplications of the surface complexation theorywith respect to the incorporation of radioactivecobalt into oxide films.[53] The concept of surfacecomplexation refers to the adsorption of dissolvedmetal species onto the oxide surface as a surfacecomplex, for which surface complex constants canbe determined. Accordingly, this approach is anal-ogous to the adsorption mechanism.

The coolant is able to fill the areas between theseparate crystals of nickel ferrite on the top of theduplex oxide film. Therefore, a fast solution-diffu-sion pathway is created for the activated solubletransition metal ions to incorporate into the outeroxide surfaces. If the outer oxide layer containspores or cracks, the diffusion of metal ions intothis oxide layer is much faster than any of thepossible solid state diffusion routes. In typicalprimary coolant environments, a hydrated metalion has a diffusion coefficient of 10-* cm2s-1 asreviewed by Hermansson et al.[41] If there is asufficient number of the pores and cracks in theouter oxide layer, they provide also an efficientmeans for the metal ions to be transported trans-versely throughout the outer nickel ferrite oxidesurface. This leads to higher probability for theactivated corrosion products to be adsorbed on thesurfaces of the porous oxide layer.

The other way for the activated corrosion prod-ucts to be incorporated into the oxide films isthrough ion exchange reactions. In this reactionthere is a direct exchange between divalent cati-ons in the spinel structure and cobalt ions in thesolution. The difference between the ion exchangeand adsorption mechanisms is shown in Figure 4.The direct ion exchange requires a high activationenergy and therefore it is relatively unlikely inmost cases. However, some degree of exchangemay be possible if the oxide cracks and/or under-goes morphology transformation resulting in ex-

posure of excessive divalent cations in the oxide tothe solution.[52,54]

After diffusing down the pores of the oxidefilm, the radioactive cobalt species may reacteither with soluble Ni, Cr and Fe species or withthe solid products to form oxide mixtures. This isthe third possible mechanism of activity incorpo-ration, that is the direct reaction/crystallisation. Itis assumed to proceed at the interface between thedense oxide and the porous part of the oxide filmresulting in an oxide structure containing cobalt,possibly according to formula ofCoxNiyFe(1_x_y)Cr2O4.[52] The rate of activity incor-poration by means of direct reaction/crystallisa-tion is believed to be controlled by the corrosionrate of the base metal, which produces the solubleand solid oxidation products to react with cobaltspecies. Therefore, incorporation of the radioac-tive cobalt via this mechanism can be expected toproceed fast on a new surface.[52]

The inner chromite oxide layer of a steadystate oxide film does not usually have many poresor cracks. Therefore, the only pathway for theactivated or non-activated transition metal ions tobe incorporated throughout the oxide is by solidstate transport mechanisms. Actually, the samemechanism applies to the parts in the outer oxidelayer where no pores or cracks exist. This longitu-dinal metal ion transport can occur through thegrain boundaries or through the crystal lattice ofthe oxide. The estimated grain-boundary diffusioncoefficients are of the order of 10-13 cm2s4 com-pared to lO-18 cm2s-1 for lattice diffusion.[41]

According to Lister [47], the inner chromiumrich oxide film incorporates 60Co strongly, butsince it is rather thin, a relatively small amount ofactivity is involved. The outer ferrite layer istypically thicker, but incorporates less 60Co on aunit mass basis than the inner layer. [47] Similar-ly, if the outer oxide film is hematite, it will be lessprotective than the inner parts of the oxide filmresulting in increased thickness. However, hema-tite has no crystal sites for divalent ions and thusactivity build-up is lower than expected solely onthe basis of the oxide thickness. This suggeststhat the extent of activity build-up and its distri-bution in the oxide film is largely determined bythe operational environment.

16

STUK-YTO-TR 1 52

4.2 Oxide films and stresscorrosion cracking (SCC)

Stress corrosion cracking (SCC) can be understoodas a localised corrosion process, which is causedand/or accelerated by mechanical stresses. Diffe-rent models developed to explain SCC are, to agreat extent, based directly or indirectly on localanodic reaction processes in the crack. Once initi-ated, the SCC of steels and nickel based alloys hasbeen suggested to proceed in increments (so-calledslip-dissolution model).[96] Each increment con-sists of the following steps at the crack tip:a) activation, i.e. exposure of new metal surface

by mechanical fractureb) dissolution of the active surfacec) re-passivation

According to the coupled environment fracturemodel (CEFM) formulated by Macdonald and Ur-quidi-Macdonald, the crack advance may occur viathe slip-dissolution-repassivation mechanism.This model requires charge conservation as a star-ting point for the calculations.[97] Other compe-

ting SCC mechanism models are based on thecombination of selective dissolution and vacancycreep, internal oxidation, cleavage and surface-mobility.[98,99,100]

It is assumed in all these models that theanodic and the respective cathodic reactions con-tributing to crack growth occur partly on or in theoxide films. Thus, the rates of these reactions maycontrol the crack propagation rate, in which casethe properties of the oxide films play a crucial rolein determining the susceptibility of the materialto SCC.

In addition, the surface mobility SCC mecha-nism [103] and the recently introduced selectivedissolution-vacancy creep (SDVC) model for SCC[103,104] include a postulate that the crackgrowth takes place by capture of vacancies in themetal close to the crack tip. Because the vacanciesin the metal lattice can be generated as a result ofthe dissolution of the metal through the oxide, thecrack growth rate may again be controlled by thetransport rate of species through the oxide film.Both models require a supply of metal vacanciesfor the crack growth to occur.

17

STUK-YTO-TR 1 52

5 NOVEL WATER CHEMISTRIES

5.1 Zinc water chemistry

The first attempts to reduce the exposure wereobtained mainly by minimising the time spent bythe workers in the radiation fields during themaintenance, inspections and refuelling. The ra-diation fields were further decreased by optimi-sing start-up, shutdown and steady-state waterchemistry conditions as well as applying other re-medies such as surface pre-treatments and chemi-cal decontamination of the primary circuit surfa-ces. The correlation between BWR radiation fieldbuild-up and the ionic zinc concentration in thecoolant was identified in early 1983. It was obser-ved that the plants, which had brass tubes(sourcefor soluble zinc) in condensers and a powderedresin condensate polishing system, had the lowestactivity incorporation into piping surfaces.[59,60]Because of this design combination, these plants

were known to have a continuous level of 5 to 15ppb soluble zinc in the reactor water. The plantsusing deep bed demineralisers did not have so-luble zinc in the coolant, because these deminera-lisers are superior to the powdered resin filterswith respect of the filtration efficiency of ionic im-purities. The long flow path through the deep beddemineraliser permits excellent opportunity forion exchange with the resins resulting usually inremoval of over 98% of the ionic species from thecoolant, whereas the powered resin filter demine-raliser is typically 80% efficient.[61] The plantshaving stainless steel condensers do not have anatural source of zinc and therefore no solublezinc in the reactor water. This has been shown toresult in higher dose rates on the piping surfaces.The comparison of the various BWR radiation be-haviour is shown in Figure 5.

Most of the natural zinc plants have nowadays

800

600

tStainless steel

10Effective Full Power Years

Figure 5. BWR radiation behaviour plants when sorted by condenser material and condensate systemtype.[60]

18

STUK-YTO-TR 1 52

either replaced their brass condensers or addeddeep bed demineralisers to the condensate treat-ment system to eliminate copper and other harm-ful ions from the reactor water. In order to main-tain low dose rates characteristic to these plants,zinc addition has been implemented. Marble wasthe first to suggest that zinc addition could beused as a remedy to control activity incorporationinto the oxides on the construction materials.[55]It was hypothesised that soluble zinc inhibits thecorrosion of stainless steel and thereby reducesthe build-up of 60Co into the oxide films. Sincethen several BWR plants have adapted zinc dos-ing and have observed clear decrease in activitybuild-up.

5.1.1 Zinc water chemistry in BWRs

5.1.1.1 Effects of zinc dosing on activity levelsin BWRs

Fourteen boiling water reactors were using zincinjections at the end 1994 for the dose rate cont-rol.[58] Lin et al.[52] have shown that zinc additi-on is effective under both NWC and HWC in boi-ling water reactors, as shown in Figure 6. The

NWC-Zn 115 ppb] ( O )

HWC-Zn 115 ppbl ( A )A A ^

500 1000 1500

EXPOSURE TIME (HOURS)

Figure 6. Effects of zinc ions on 60Co deposition onas-received AISI 304 SS samples under normal andhydrogen water chemistry conditions.[52]

plant experience has shown that changes in waterchemistry (NWC, HWC, NWC, etc.) leads to inc-reased activity levels in the oxide films. [8] Howe-ver, the expectations for the zinc addition inplants, in which perturbation in the amount of H2

in the coolant occur, are that the effects of waterchemistry changes will be lessened and that thedose rates remain only moderately higher than ifthe plant is continuously operated under NWC.[61]

The zinc additions can be carried out in differ-ent ways. Plants have been using low flow, posi-tive displacement pumps to inject zinc oxide sus-pension into a re-circulation loop. This system hasbeen further modified by direct zinc injections intothe feed water pipe by using higher flow rateinjections. The third possibility is to use a passivesystem without moving parts. In this system, abed of sintered zinc oxide pellets is contained in asmall pressure vessel, through which the primarywater is circulated. Sufficient amount of zinc isdissolved from the pellets to maintain the desiredconcentration of zinc in the reactor water. Thistechnique is attractive, because the operating andmaintenance requirements of zinc additions canbe minimised. Furthermore, this pellet bed pas-sive system is designed in such a way that it willlast at least one full fuel cycle.[60,63]

As discussed later in chapter 5.1.6, the plantswhich have been dosing natural zinc have experi-enced one side effect: the activation of 64Zn to form65Zn, which adds to the inventory of activatedcorrosion products of the plant. Iron adsorbs zincand then deposits on the fuel cladding surface,where Zn becomes activated. However, the "natu-ral" zinc plants have had very little problems,because they all have used powdered resin con-densate system and thereby have a very littlecrud input into the reactor feed water. Severalplants have nowadays deep bed demineralisers inthe condensate system and, as a consequence,have also higher crud input resulting in higheractivation of 64Zn.[62] Some plants have started touse Zn, which is depleted in 64Zn to avoid theproblem of 65Zn. The content of 64Zn in the deplet-ed zinc oxide (DZO) is reduced from 49% to 1%.However, the cost of DZO is high due to therequired processing. Therefore, reduction of ironinput in a high crud plants should significantlyreduce the cost of using DZO.[60] Moreover, the

19

STUK-YTO-TR 1 52

impact of DZO will not be seen in the plants,which have earlier had a natural zinc source, forseveral cycles because of the large natural zincinventory present in the oxides throughout thereactor surfaces.

Each of the zinc adding plants is, for differentreasons, currently using lower zinc concentrationsthan the recommended 10 ppb as shown in Table I.For the plants using natural zinc oxide the result-ing S5Zn has been a concern because of its impacton piping dose rates, radwaste, etc. The plantsusing DZO are running at reduced zinc concentra-tions for cost saving reasons.

5.1.1.2 Changes in BWR coolant chemistryduring Zn dosing periods

Since soluble zinc is ionic and acts as a weak base,it is to be expected that the addition of zinc willalter both the conductivity and the pH of the reac-tor water in boiling water reactors. The extent ofthis effect depends on the concentration of otherimpurities in the reactor water. If the coolant isacidic without the presence of zinc, then the im-pact would be to decrease conductivity as a resultof a neutralisation reaction. The root cause wouldbe the lower specific ionic conductance of Zn2+

when compared to H+. If the coolant is alreadybasic, the zinc addition would increase both theconductivity and the pH, but only slightly. Theplant observations confirm these conclusions. Ho-

Table I. Typical zinc contents in the reactor waterin the plants which are dosing Zn.[61]

Plant

FitzPatrick

Hatch 1

Hatch 2

Hope Creek

Leibstadt

Limerick 1

Limerick 2

Millstone Pt 1

Monticello

Nine Mile Pt 2

Peach Bottom 2

Peach Bottom 3

Perry

RxW Zn (ppb)

3

6

6

2

3

5

2

4

7

3

2

2

4

wever, small increases in the conductivity and thepH are not considered to have any adverse effectsfor the BWR.[61,95]

One consequence, which was not anticipatedprior to the zinc addition, was the suppression ofconcentration of the soluble 60Co in the reactorwater. This decrease has been found at each BWR,which has started to use zinc after operating forone or more cycles as a non-zinc plant. The com-parison of 60Co content in the reactor water beforeand after Zn addition in four plants is shown inFigure 7.

0.50

0.00

HPre-ZincEGEZIP

Milistone FitzPatrick Monticello Leibstadt

Figure 7. Reduction of reactor water 60Co as a result ofZn addition.[61]

20

STUK-YTO-TR 1 52

Zn is assumed to act in two ways to lower thesoluble 60Co concentration in the reactor water.First, it suppresses the corrosion release ratesfrom in-core cobalt alloys, such as Stellite rollersand pins. Second, as zinc is incorporated into theiron based fuel deposits, the release rate of 60Co islower. This is probably due to the lower solubilityof the zinc rich crud, reducing 60Co incorporationinto the oxides on primary system piping andcomponents. In addition, due to the lower solubili-ty, the amount of 58Co and 60Co released to thereactor water during a shutdown period is re-duced. Therefore, the activity remains within ox-ides on the in-core materials and in the fueldeposits. As such, part of the activated corrosionproducts are removed when fuel is replaced. How-ever, the degree of 60Co suppression is a functionof the zinc concentration in the reactor water,similarly as the effect of zinc on the reduction inthe corrosion rate.[61,62] There are also otherexplanations: The pHj, may increase significantlyclose to the fuels cladding surface due to theboiling. This leads to decreased solubility of thefuel crud. Similarly, the pHT increase would lowerthe solubility of oxides on in-core Stelliteparts. [53,64]

Operational experience has indicated that zincdemand is increased after each refuelling outageas a result of the zinc is dissolved from the oxidesduring the shutdown period and there are newsurfaces inserted during the refuelling.

5.1.2 Zinc water chemistry in PWRs

5.1.2.1 Effects of zinc dosing on activity levelsin PWRs

As discussed in the previous chapter, zinc hasbeen used widely in BWRs to obtain lower activityin-corporation into oxides on the primary systemsurfaces. Several studies have also been carriedout in out-of-core loops to evaluate the effect of Znunder typical PWR environments.[18,75,56] Theresults have shown that dissolved zinc in the ran-ge of 10 to 40 ppb reduces the pickup of 60Co by afactor of 8 to 10. Zinc injections have resulted alsoin thinner oxide films. These studies suggestedthat zinc addition may be a cost effective methodto reduce the rate of activity build-up also on the

primary circuit surfaces in PWRs.[76]Even though only two operating PWR plants

have reported preliminary results from the zincinjection tests, the results so far look rather prom-ising. Investigations at Farley-2 unit have shownthat out-of-core exposure rates can be reduced byinjecting low concentrations of natural zinc intothe primary coolant. The zinc content in the cool-ant was maintained in a range of 35-45 ppb forapproximately nine months. Measured radiationdose rates decreased significantly at steam gener-ator channel heads and at the main coolant pipingcompared to the cycles without Zn. An encourag-ing observation was also that 65Zn was less than10% of the radioisotope mix and only a minorcontributor to the increasing dose rates. [57,17]

The second PWR which has performed Zndosing into the primary coolant, is Biblis NPP inGermany. In Biblis, the dosed Zn concentrationwas very low (< 5 ppb) and therefore rather smallreduction of activity incorporation into the oxidefilms was observed.[78]

Some controversy exists in the published re-sults. The tests which were carried out at MITtest reactor under PWR conditions using Zn injec-tions showed no beneficial effects on the activityincorporation. In addition, no changes in theamount of deposited corrosion products on themetal surfaces were observed.[78] However, thetests performed at OECD Halden test reactor intypical PWR environments have shown that theactivity incorporation during zinc injection peri-ods was lower than in the coolants without Znaddition.

5.1.2.2 Changes in PWR coolant chemistryduring Zn dosing periods

The first plant experiments of zinc additions (aszinc acetate) at Farley-2 unit were started in midJune 1994, about six months after the cycle 10start-up. Zinc was injected for 11 days before itwas detected in the primary water. [76] However,during the temperature decreases, the zinc con-centration increased even though the Zn injectionwas stopped. The decrease in the zinc concentrati-on in the coolant was much slower than expectedindicating that zinc was being released from theoxide films. After the zinc injection was re-started

21

STUK-YTO-TR 1 52

and the plant returned to full power operation, thezinc concentrations returned to the nominal le-vel.[77]

The activity of soluble 60Co in the coolant didnot change after the Zn injection. A similar trendwas also observed at Biblis during the Zn dosingperiod. [78] A review of the data showed thatnearly all of the cobalt activity was in insolubleform in the primary coolant whereas 65Zn activitywas predominantly in the soluble form. The calcu-lations showed that some of the 65Zn must haveoriginated from the core crud. This indicates thatsome of the zinc in the coolant was exchangingwith the zinc in the core crud or causing theactivated zinc to be released.[76,77]

5.1.3 Effect of zinc on activityincorporation into oxide films

5.1.3.1 Zinc and oxide films under oxidisingconditions

The incorporation of zinc can be assumed to de-pend on the oxide film structures but not on theplant type. Therefore, the effects of zinc dosingdiscussed in these following chapters are based onthe oxidising (BWR, NWC) or reducing (BWR,HWC and PWR) conditions of the primary coolant.

The oxide films on the carbon steel (CS) surfac-es under oxidising conditions consist of hematiteand spinels. Some laboratory experiments haveshown that the greater the content of hematite,the less protective the film is resulting in thickeroxide films. However, hematite has no crystalsites for divalent ions. This explains why Penneret al. results showed that as the outer oxide filmconsisted mainly of hematite, the amount of Znand cobalt incorporated into outer oxide layer waslow. [66] Thus activity build-up is not solely con-trolled or explained by the total oxide thickness.

Hanzawa et al. studied the incorporation ofcobalt and zinc into the oxide films on carbon steelsurfaces using fairly high ionic concentrations(max. Zn = 590 ppb, max. Co =1200 ppb). Theyfound out that the cobalt content in the oxide filmincreases with increasing Co content in the aque-ous phase even though the coolant contained 300ppb of zinc. They also showed that the increase ofZn in solution decreases the cobalt incorporationinto the oxide film. This indicates that zinc ionscompete with soluble cobalt ions of the adsorption

sites on the oxide film surface and consequentlysuppresses the incorporation of cobalt into theoxide films. Another interesting observation wasthat zinc did not have any effect on the filmgrowth rate on carbon steel in oxidising environ-ments. [72] Similarly, Lister et al. found out thatthe oxide growth rate of the CS sample exposed tozinc containing coolant did not differ significantlyfrom the growth rate of the oxide on the carbonsteel sample which was grown without Zn.[46]

When effects of zinc dosing on behaviour ofstainless steel samples were studied, the observa-tions of the oxide growth rates and ion releaserates from SS oxide films turned out be totallydifferent. Lister et al. has shown that dissolved Znhas a profound effect on the release rates ofcorrosion products from stainless steel even inNWC conditions. His experiments demonstratedthat addition of Zn resulted in thinner and moreprotective oxide films. In addition, the release rateof cobalt from the base metal and oxide was farlower than that observed in the normal BWRconditions. This indicates that zinc somehow de-creased the transport rate of cobalt ions throughthe oxide.[46]

Lin et al. verified in their zinc injections tests,(10 ppb of Zn) that eoCo incorporation into theoxide film on both AISI 304 SS and AISI 316 SSwas low as shown in Figure 6.[52] They wasobserved also that the higher Zn concentration inthe coolant, the less activity is incorporated intothe oxide, but addition of Zn did not replace thecobalt from the oxide as shown in Figure 8. BothCo and Zn had highest concentrations on the oxidesurface. However, some tests showed that thecobalt accumulation rate into the inner layer de-creased by zinc addition even though cobalt wasdosed into the coolant continuously. Permer et al.have also reported that Zn had a maximum con-centration in these parts of the oxide in whichchromium concentration was also highest.[46,66]

These results seem to indicate that zinc addi-tion reduces significantly the 60Co incorporationinto the oxide films on stainless steels as well asreduces the ion transport through the oxide lead-ing to thinner oxide layers. However, the zincdosing seems to have rather small impact on thebehaviour of oxide films on carbon steels underoxidising conditions.

22

STUK-YTO-TR 1 52

5.1.3.2 Zinc and oxide films under reducingconditions

An important aspect of surface oxide films formedunder oxidising and reducing conditions is theirspinel structure and distribution of metal cationstherein, because activity incorporation into theoxide films can not be only related to film thick-ness but is also strongly affected by film characte-ristics and composition.

As shown in the previous chapter, Zn additiondid not have a significant effect on oxide growthrate on carbon steel surfaces in oxidising condi-tions. Similarly Allsop et al.[73] have shown thatthose carbon steel coupons, which were exposed toZinc containing coolant in reducing conditions hadonly slightly thinner oxide films on the surface(2.7 |J.m with zinc: 3.7 \im without Zn). Duringthese tests in reducing environments it was shown

10°

10"

10- 2

10-3

Fe

CrB-164 11/15/85

ATOM RATIO TO Fe vs. SPUTTER DEPTH (A)

Ni

Mn

Mo

Co

0-16

200 400 600 800 1000 1200 1400 1600 1800

10°

10"

10-2

10i 3

A146 11/25/85ATOM RATIO TO Fe vs. SPUTTER DEPTH (A)

Mn

500 1000 1500 2000 2500 3000 3500Figure 8. Elemental depth profiles in oxide films (AISI 304 SS, AR and AISI 316 SS,AR) exposed toNWC-Zn(15 ppb) and NWC-without Zn conditions.[52]

23

STUK-YTO-TR 1 52

that zinc seemed to have very little effect onactivity in-corporation into the oxide films on theCS coupons whereas a significant difference wasobserved in stainless steel coupons. Based onthese results it is clear that also in reducingenvironments Zn has smaller effect on oxidationbehaviour of carbon steel when compared to theresults obtained for stainless steel.[8,73,74]

Riess et al. [80] studied the influence of zinc onthe oxide films and activity incorporation of thesample which had been exposed prior the tests inthe steam generator channel head of a PWR overtwo years. They reported that the oxides on stain-less steel specimens which were exposed to zincenvironments incorporated zinc up to 16%. Theseresults also showed that after the Zn injection ca.50% lower y-activity was measured on the samplesdue to the activity release into the coolant.[80]Therefore Riess et al. concluded that zinc is veryefficient in displacing 60Co from the contaminatedsystem surfaces. The initial rate of displacementof cobalt and other elements from existing oxide isapparently very high.[80] The spinel compositionon the specimens ranged from (FeNiZn)Cr2O4 to(FeNiZn)FeCrO4. In addition, some of the parti-cles on the coupon surfaces were analysed andthey corresponded roughly to the spinelNi0SZn05Fe2O4.[80] However, Fe,Ni-spinels weredetected only in isolated cases, which is not typi-cal for oxides grown in high temperature waterunder PWR conditions. This could indicate thatfor some reason some of the outermost depositedlayer had been removed during the zinc injectionperiods. The fact that the oxide thickness of SSdecreased during the exposure to the zinc contain-ing coolant was also partly supported by Riess et al.

The studies carried out at OECD Halden Reac-tor Project have shown that zinc injection to theprimary coolants results in thinner oxide layerson new metal surfaces (SS, 1-600, 1-690, Incoloy800) with low visible porosity. In addition Zninjection hinders the incorporation of activity intoalready existing oxides.[82,83,84,85] During thezinc injection tests, activity pickup of the oxidefilms was low in all the studied samples. Inaddition, the oxides did not grow in thickness,because the corrosion product deposition from thesolution was minimal. The lack of thick depositedoxide layer partly explained also the observedreduction in activity incorporation during Zn in-

jection periods.[82,83,84,85] The deposition ratesof corrosion products on primary circuit surfaceswere low, since their concentrations in the coolantdecreased due to the restricted diffusion of ionsthrough the oxide films. The effect of Zn on theactivity build-up is partly due to the fact that asmaller amount of corrosion products are releasedto the coolant, i.e. less corrosion products, whichcan become active and be redeposited on thetubing.[18,20] However, the tests at Halden haveshown that the rate of activity build-up on out-of-core surfaces can be reduced only with continuouszinc injection. Once the injection of zinc into thecoolant was stopped, the activity incorporationinto the stainless steel started to increase.[84,85]

In some tests zinc concentrations have beenhighest on the oxide surface and decreased rapid-ly deeper in the oxide. The zinc concentrationthroughout the oxide film appeared to increasewith exposure time, both in terms of the maxi-mum surface concentration and the depth of thezinc enriched layer. In this process, time appearedto be more determining than the zinc concentra-tion in the coolant, probably due to complex kinet-ics associated with combined diffusion and ionexchange reactions. In one of the test performedin Halden, the surface enrichment of Zn wasrelated to high Cr concentration on the outeroxide film.[85] During tests done by Korb et al., itwas observed that ZnO was formed on the oxidesurface, due to the very high soluble zinc concen-trations in the coolant (up to 281 ppb).[81] Thesefindings are supported by Beverskog et al. whoreported that the solubility limit of ZnO to beroughly 60 ppb.[71,86] In some cases the highestzinc concentration have been measured betweenthe inner and outer oxide films, where also thehighest Cr concentration existed. [66,85] These re-sults seem to indicate that zinc has a fairly highaffinity for Cr rich oxide layers.

Some results show that the decrease in 60Codeposition caused by zinc seems to be greater thanthe decrease in the film thickness, indicating thatZn may modify the propensity of the oxide to in-corporate 60Co.[18,68] Similar results were ob-tained also in Halden. Those test showed thateven though the oxide thickness did not change onthe studied coupons, the total amount of incorpo-rated activity remained lower than during theperiods without Zn.[85]

24

STUK-YTO-TR 1 52

Table II. Distribution of ions in tetrahedral and octahedral positions in normal and inverse spinels.

Normal spinel AB2O4

Inverse spinel B(AB)O4

Occupied close-packedpositions

32 O2

32 02

Occupied tetrahedralpositions

8 A2+ ions

8 B3+ ions

Occupied octahedralpositions

16 B3+ ions

8 A2+ and 8 B3+ ions

5.1.4 Correlation between role of Znand oxide structure

The oxides formed on the primary circuit construc-tion materials in light water reactors are usuallyspinels. On carbon steel, the oxide can be either amagnetite or maghemite spinel depending on thereducing or oxidising power of the coolant. On al-loyed steels, the alloying elements are incorpora-ted in the growing spinel oxides more or less inthe proportion to their concentration in the basemetal. The structure of spinel oxides, the cationdistribution in them as well as defects in oxidesare presented in this chapter. Moreover, the pos-sible mechanisms of the zinc cation incorporationinto the oxide films are discussed on the basis ofthese concepts.

5.1.4 Structure of spinel oxidesBulk oxides are crystalline compounds, which con-sists of ions packed regularly in a three dimensio-nal arrangement. For any points in the pattern, itis possible to find other points possessing exactlythe same surroundings. Such points define a regu-lar lattice. One lattice point can be reached fromanother point by taking a suitable number of stepsalong each of three directions. This small volumeof atoms/ions is called a unit cell. The whole latti-ce can be built up by these unit cells.[21]

The most efficient packing of uniform spheres(ion/atom) forms a close-packed layer of spheres.

Figure 9. Construction of the basic close-packed unit(twopacking layers for any closed-packed structure).P = oxygen ion, O = octahedral site, T = tetrahedralsite, A,B,C, the three relative packing positions.[22]

Spinel structures are based on close-packed ar-rays of oxygen ions, which are on a face-centredcubic lattice. Each unit cell of a spinel latticecontains eight (8) formula units of AB2O4. Thismeans that each unit cell contains 32 O2~ ions.Because the spinel must be electrically neutral,each unit cell must contain 64 positive charges. Inthe face-centered cubic close-packing, each sphereis surrounded by twelve nearest neighbours. Thevolume of the structure containing n spheres withthe radius r is 5.66*r3*n. If there are n spheres inthe system, there are 2*n tetrahedral sites and noctahedral sites. This means that in a unit cellcontaining 32 O2- ions, there are 64 tetrahedralsites and 32 octahedral sites. [22] The positions ofthe sites lie as shown in Figure 9.

5.1.4.2 Classification of spinel oxidesSpinels can be classified as normal, inverse andintermediate, depending on the location of A2+ andB3+ ions in the tetrahedral and octahedral posi-tions. Table II shows how the lattice sites are oc-cupied in normal and inverse spinel:

Higher valency cations tend to prefer the octa-hedral holes.[23] When all the eight A2+ cationsexist in the tetrahedral positions and all thesixteen B3+ cations in the octahedral positions, anormal spinel structure prevails. In an inversespinel eight higher valency B3+ ions are located inthe tetrahedral positions, and the remaining eightB3+ ions and the eight A2+ ions with a lower chargeoccupy the sixteen octahedral positions in a statis-tically disordered manner.

The normal and inverse spinels correspond tothe extreme values of distribution of B3+ ionsbetween tetrahedral and octahedral sites. In in-termediate spinels the A2+ and B3+ cations arearranged in a disordered manner both in tetrahe-dral and octahedral positions.[23]

A2+ and B3+ ions can be arranged in threedifferent ways to obtain an electrically neutral

25

STUK-YTO-TR 1 52

Table III. Examples of typical normal and inverse spinels. The results are based on experimental results,not on predictions by the crystal field theory.[24,25,69]

Mineralname

Franklinite

Danathite

Chromite

Mineralname

CoFerrite

Trevorite

Magnetite

Jacobsite

ZnFe2O,

ZnCr?O4

CoCr2O,

NiCr2O4

FeCr2O4

MgCr2O4

MnCr2O4

CoFe2O4

NiFe2O4

F e 3 O 4

MgFe2O4

MnFe2O4

Tetrahedralsite

8

8

8

8

8

8

8

" Z n ^

* Z n 2 '11 Co?-

* N i 2 ' ii

* F e 2 +

* M g 2 +

* M n 2 +

Tetrahedralsite

8

8

8

8

8

* F e 3 +

* F e 3 + •

* F e 3 +

* F e 3 +

* Fe3+

Normal spinel

Octahedral site

8 * (Fe3-)2

8* (Cr 3 - ) ?

8* (Cr 3 - ) 2

8 * (Cr3-)2

8 * (Cr3+)2

8 * (Cr3+)2

8 * (Cr3+)2

Close-packed

8

8

8

8

8

8

8

Inverse spinel

Octahedral site

8

8

8

8

8

* (Co2+ + Fe3+)

* (Ni2+ + Fe3+)

* (Fe2+ + Fe3+)

* (Mg2 + + Fe3+)

* (Mn 2 + + Fe3+)

* [0%

* i o 2 l* [ 0 z 1 4

* [0 2 1 4

* [0 2 1 4

Close-packed

8

8

8

8

8

* [02]4

* [0%

* [0 2 ] 4

* [02"]4

* [ 0 2 1 4

Empty tetrahedral/octahedralsites in a unit cell

56/16

56/16

56/16

56/16

56/16

56/16

56/16

Empty tetrahedral/octahedralsites in a unit cell

56/16

56/16

56/16

56/16

56/16

spinel structure:(i) [A (̂B3+)2](ii) [AHB^)2

(iii)

2:3 spinel4:2 spinel

4:2 spinels are found in an inverse structure only,one example being [Ti4+(Zn2+)2]8. Spinels of theform 2:3 are found both in the normal and inverseclass. Some examples of these structures areshown in Table III.

Several factors, such as hydration energy andelectrostatic forces between different ions contrib-ute to the stability of an ion in an octahedral ortetrahedral site.[26] However, an estimation ofthe crystal field stabilisation energy of a cationalone gives a reliable indication whether the ionwill occupy an octahedral or tetrahedral site de-termining which type of spinel structure is fa-voured.

5.1.4.3 Metal ion distribution in the spinelsaccording to crystal field theory

The structure of spinels can be predicted in termsof the valence bond [28] and the molecular orbitaltheories [28], but they are rather complicated tobe applied on spinels. An other possibility to esti-mate spinel structures and how different ions af-

fect the stabilities of different structures, is to usecrystal or ligand field theory. In spite of the appro-ximations, these theories yield quantitative re-sults which are more in accordance with experi-mental values than are the results of calculationsbased on valence and molecular orbital theories.The approximations used in crystal field theory,are also included to the ligand field theory, whichdiffers from crystal field theory only in two ways.In the ligand field theory, the transition metal ionis described to be surrounded by ligands which areeither negative ions or highly polar molecules; at agreater distance are the ions of opposite sign tothat of the transition metal complex ion. The li-gand field also persists so long as the complexpersist and is therefore present not only when thecomplex is crystalline, but also when it is a va-pour, a melt or in solution. In the following chap-ters, the stability of octahedral and tetrahedralcomplexes is discussed on the basis of crystal fieldtheory.

In an oxide consisting of a close-packed struc-ture of negative ions (like O2-), the transitionmetal ions are in the sites between them. Their dorbitals are exposed to the field between metalnucleus and the negative charge of the surround-ing ions. This field is known as the crystal field.

26

STUK-YTO-TR 1 52

The crystal field theory assumes that the interac-tion between the transition metal ion and sur-rounding oxygen ions is purely electrostatic andno covalency is included (no combination of orbit-als).[27] The field has also a definite symmetryand produces similar effects on all of the d orbit-als.

The five d orbitals (of the transition metalsfrom Cr to Zn) in the isolated, gaseous metal ionare degenerate (not split) as shown in Figure 10b.If a spherically symmetric field of negative charg-es is placed around the transition metal ion, theenergies of all five orbitals will be raised as aresult of the repulsion between the negative oxy-gen ion and the negative electrons in the orbitals.The five d orbitals remain still degenerate. If thefield results from the influence of the real crystalsystem, the symmetry must be less than sphericalbecause of the finite number of oxygen ions in-volved. This leads to splitting of the energy levelsof d orbitals, which, on the other hand, alsostabilises the system. The magnitude of splittingvaries from 10-6 eV to 5 eV depending on theanions causing the splitting. In the crystal fieldapproach, the used splitting energy is denotedwith 10 Dq by definition. In ligand field theory Ao

is used usually to describe the same parameter.The five d orbitals are split to the eg and t2g

orbitals as shown in Figure 10a.In order to understand the interactions that

are responsible for crystal field effects, it is neces-sary to understand the geometrical relationshipsof d orbitals. In the case of six O2- ions approach-ing a transition metal ion to form an octahedralcompound, the six oxygen anions enter along the

axes of the co-ordinate system, i.e. from the direc-tions z, -z, x, -x, y and -y as shown in Figurell.[28] Under these conditions, the anions willinteract strongly with two eg orbitals (shaded)lying in the x, y z axis. The remaining three t2g

orbitals (unshaded) will be interacted to a lesserextent, because they are directed between theapproaching anions as shown in Figure 11.

The energy of the orbitals may be raised by therepulsion of the oxygen anions. However, by rear-ranging the O2- ions from a hypothetical sphericalfield to an octahedral field does not alter theaverage energy of the set of d orbitals. Hence tomaintain the barycenter it is necessary for the 2 eg

orbitals to be repelled by an energy of 6 Dq tobalance the stabilisation of the 3 t2g orbitals to theextent of 4 Dq as shown in Figure 11. Thisconstancy of the barycenter of the d orbitals islinked to the conservation of the orbital energy(2*6 Dq + 3*(-4 Dq) = 0). So the energy of the t2g

level relative to the barycenter of unperturbed dorbitals is -4 Dq which is the crystal field stabili-sation energy (CFSE) in the case of d1. For d3

(three electrons in d orbitals) the CFSE is -12 Dq.In these configurations, the electrons remain un-paired and enter different degenerate orbitals.However, the fourth electron can not be placed into the t2g levels, without electron pairing. In theweak field limit (low CFSE) the splitting of theorbitals (10 Dq) is small with respect to the energynecessary to cause electron pairing in a singleorbital and the fourth electron to enter one of theeg levels (leading to a high spin complex). Theoxide ions in spinels provide a moderately weakcrystalline field and therefore the weak field situ-

>

L

= eg orbitals

VsA,

2A-

orbitals

(a) Octahedral field (b) No field;degenerate d orbitals

Figure 10. Diagrammatic presentation of the splitting of d orbitals caused by electrostatic octahedralfield as well as no field situation.[27]

27

STUK-YTO-TR 1 52

Table IV. Crystal field effects for weak octahedral and tetrahedral fields.

d1

d2

d3

d*

d5

d6

d7

d8

d9

Octahedral field

Configuration

V1.+3 p2T 2g e 9t4 e 2

r2gegt5 e2

X29e9

T29e9

V.

Unpairedelectrons

1

2

3

4

5

4

3

2

1

0

CFSE

^4Dq

-8 Dq

-12 Dq

~6Dq

0

-4Dq

-8Dq

-12Dq

-6 Dq

0

Tetrahedral field

Configuration

e1

e2

e>t\

e¥2

e 2 ^

e 3 ^

e^2

eH*a

e 4 ^

CFSE

- 6 D q

-12 Dq

- 8 D q

-4Dq0

-6Dq-12 Dq-8Dq-4Dq

0

ations are only considered in this report. The netCFSE energy is then:

CFSE = (3*-4 Dq) + (1*6 Dq) = -6 Dq ( t ^ )

The addition of a fifth electron results in a halffilled d orbitals and the CFSE is zero. The pres-ence of two electrons in the unfavourable eg levelexactly balances the stabilisation resulting fromthe three electrons in the t2g level. The remainingelectrons are placed to the orbitals in the similarway and the resulting CFSE are listed in Table IV,together with the number of unpaired electronsexpected for each configuration.

Two of the most common geometries for a 4 co-ordinate compound are the tetrahedral and squareplanar arrangements. The square planar geome-try is a special case and not discussed in thisreview. Energy level scheme for tetrahedral sym-

metry is qualitatively similar to that for cubicsystem, only the splitting (10 Dq) is half as large.The d orbitals and their splitting in tetrahedralcomplexes are shown in Figure 12.

In Figure 12, eight oxygen anions are ap-proaching the central transition metal ion. If fourligands are removed from the alternate corners ofthe cube as shown in Figure 12, the remainingoxygen anions form a tetrahedron around themetal. In this arrangement the O2- ions do notdirectly approach any of the metal d orbitals, butcome closer to the t2 orbitals directed to the edgesof the cube than to the e levels directed to thecentres of the faces of the cube. Hence the t2 levelsare raised in energy and the e levels stabilised.Furthermore, since the barycenter rule holds, thethree t2 levels are raised by 4 Dq and the two elevels lowered by 6 Dq from the barycenter. The

Figure 11. Complete set ofd orbitals in an octahedral field. The eg orbitals are shaded and the t2g orbitalsare unshaded (a). Splitting of the degeneracy of the five d orbitals by an octahedral field (b).[28]

28

S T U K - Y T O - T R 152

energy levels for the tetrahedral symmetry areexactly the inverse of that for octahedral symme-try. The electron pairing energy is larger than 10Dq and the electrons enter the five orbitals re-maining unpaired until the sixth electron forcespairing. The d4 case results in an e2t22 configura-tion with crystal field stabilisation of-4 Dq:

CFSE = (2 * - 6 Dq) + (2 * 4 Dq) = -4 DqThe number of the unpaired electrons, the confi-gurations and CFSEs are given in Table IV. Sincethe absolute value of 10 Dq (in kJmoH) is less intetrahedral complexes than in octahedral ones (be-cause of the indirect effect of the O2~ ions and theirsmaller number), the total crystal field stabilisati-on is much less important than with octahedralcomplexes. There are several factors affecting theextent of splitting of the d orbitals. Some valuesfor 10 Dq of aqua complexes of the metal ions of thefirst transition metal series are listed in Table V.

Some trends in the table become apparent. Theionic charge on the metal ion has a direct effectupon the magnitude of Dq. This is expected, sincethe increased charge of the metal ion will attractthe O2- ions more closely, hence they will have agreater effect on perturbing the metal d orbitals.Secondly, the splitting in an octahedral field ismore than twice as strong as for a tetrahedralfield from the same metal ion:

10 Dqtd = 4/9 * 10 Dqoh

in which 4/9 is based on the square of the ligandinteractions (42/62).

There are several factors that influence theadoption of tetrahedral or octahedral co-ordina-tion, and occasionally the balance between oppos-

ing factors is a delicate one. From a purely electro-static viewpoint, octahedral co-ordination is fa-voured because six ligands are approaching in-stead of four (O2-). On the other hand, if theligands are bulky then the ligand-ligand interac-tion may cause some opposing effects.

Tetrahedral complexes are always in the highspin. As a result, the maximum CFSE can be 12Dq which converted to octahedral field equiva-lents, is only 5 Dq (12 Dq*4/9). Therefore, whencomparing a given ion in a tetrahedral field withthe same ion in an equivalent octahedral site, theion is always at least as stable in the octahedralhole, usually more so. The difference in energywhich always favours the octahedral case, istermed octahedral site stabilisation energy(OSSE), since the term was originally applied tocationic preference for octahedral holes in anioniclattices. Some estimates of OSSE in kJmoH areshown in Table VI.

For some configurations, such as d1, d2, d6

(Zn2+), d6, d7 and d10 the advantage of the octahe-dral arrangement in spinels is little or nothing.Others, such as d3 (Cr3+) and d8 (Ni2+), are stronglyfavoured to be octahedral as shown in Table VI.

Fe3+ in trevorite (NiFe2O4), is located both intetrahedral and octahedral sites and Ni2+ in octa-hedral sites as shown in Table III. For the d5 Fe3+

ion the CFSE is zero for both tetra and octahedralsites, but the d8 Ni2+ ion has an OSSE of 8.45 Dq(Table VI) or approximately 96 kJmol-1 (Table V).This CSFE advantage for Ni2+ ion in the octahe-dral holes is sufficient to invert the structure.Similarly also in magnetite, Fe3O4, the d6 Fe2+ ion

+ *

— _A

5 ̂ ~"\

> ^

a)

W

b)

/

\\\

\\\

1'a1

Figure 12. Complete set ofd orbitals in a tetrahedral field. The eg orbitals are shaded and the t2g orbitalsare unshaded (a). Splitting of the degeneracy of the five d orbitals by an tetrahedral field (b).[28]

29

STUK-YTO-TR 1 52

Table V. Crystal field theory data for metal ions of the first transition metal series in aqua complexes.[28]

Numberoft/electrons

1

2

3

4

5

6

7

8

9

10

1.2.3.4.5.6.7.

8,9.10.

c . Octahedral TetrahedralFree ion . . . . ,. . .

. . field fieldIon ground . .

stat ground groundstate state

Ti r3 'D t\ e1

V-3 JF t2^ e2

V + 2 4/~ t3 e2tf1

Cr+3 AF P2q e2t\

C r + 2 ^D t3 t?1 ©2?2

Mp+3 5Q *3 g1 t?2f2

M n + 2 6S if3 e 2 e2f32

F e + 3 6S f3 e2 e2fi2g a 2

Pg+2 BQ ^4 g2 e^t3

Q Q + 3 B£J *6 I?3?3

C o + 2 4/T ^ ^ 2 ^ p ^

Nr 2 *F F2ge*g e^z

Cu+2 2D f2 e3 eAtz2

Zn+2 1S f6^4^ e^62

The data given are:Number of d electronsTransition metal ionsFree ion Russell-Saunders ground (spin-orbit coupling isElectron configuration of octahedral ground stateElectron configuration of tetrahedral ground stateDq values for octahedral hydrates of the ionsDq calculated for tetrahedral coordination

Dq{cm-1)

oct.2030

1800

1180

1760

1400

2100

750

1400

1000—

1000

860

1300

0

neglected

The thermodynamic stabilization in octahedral or tetrahedra! fields

Dq(cm1)

tetr.900

840

520

780

620

930

330

620

440

780

440

380

580

0

Stabilization(kJ moM)

oct.96.6

174.5

168.0

250.8

100.3

150.1

0

0

47.6

188

71.5

122.4

92.8

0

tetr.64.4

120.0

36.4

55.6

29.3

44.3

0

0

31.4

107

62.7

27.2

27.6

0

n the term designation)

The octahedral site preference, or the difference between columns 8 and 9* The octahedral site stabilization of Co+3was estimated

Oct. sitepreferenceenergy(kJ mol"1)

32.3

54.5

131.6

195.2

71.0

105.8

0

0

16.3

8 1 *

8.8

95.2

65.2

0

from the heat of hydration increment caused bythe crystal field, and the tetrahedral site stabilization was taken to be the same as for Cr+3

SOURCE: T. M. Dunn, D. S. McClure, and R. G. Pearson, "Some Aspects of Crystal Field Theory,"Harper & Row, New York, 19965, p. 82. Used with permission.

is stabilised to the extent of 1.33 Dqoh to cause aninverse structure to be formed.

Not all metal ions of the first transition metalseries form a spinel which has an inverse struc-ture. All the chromium spinels ACr2O4 have thenormal structure as a result of the strong octahe-dral site preference of Cr3+. Further examples ofspinels are listed in Table III. When applying thecrystal field theory, one must remember that thestructure of the oxide films, forming on the con-struction materials, is mainly governed by theoperation environment. However, as soon as thetransition metal ion is incorporated into the spinelstructure, by using crystal field theory, it is possi-ble to obtain an estimate how it will affect thecrystal system. The oxide growth itself dependsalso on the ion transport through the oxide whichis governed significantly by the defect concentra-

tion in the oxide film. This is discussed in somedetail in the next chapter.

5.1.4.4 Defect structures in oxidesSolid materials are never perfect at temperaturesabove absolute zero but contain imperfections ordefects in their structure. In spite of the often lowconcentration of defects, many important proper-ties of solids are governed by their presence in thelattice. For example, the diffusion of or conductionby ions and electrons in crystalline compoundsdepend on the defect concentration in the latti-ce.[31]

In general, vacancies in an oxide lattice canresult from imperfect packing during the originalcrystallisation or they may arise from thermalvibrations at elevated temperatures. Vacanciescan be single or two or more of them may conden-

30

STUK-YTO-TR 1 52

Table VI. Relative crystal field stabilisation energies (CFSE) and resulting octahedral site stabilisationenergies (OSSE) for various electron configurations.[28]

Configuration

d1

d2

d3

d4

d5

d6

d7

d8

d9

d i o

CFSE, tetrahedral complex

Dq* Dq0,,6

12

8

4

0

6

12

8

4

0

2.67

5.33

3.55

1.78

0

2.67

5.33

3.55

1.78

0

CFSE, octahedralcomplex, (Dqoh)

4

8

12

6

0

4

8

12

6

0

OSSEkJ/mol

1.33

2.67

8.45

4.22

0

1.33

2.67

8.45

4.22

0

sate into a multi-vacancy. The presence of a va-cancy does not mean that one lattice site is vacantwhile all the surrounding lattice particles remainin their original positions. However, it is possiblethat the neighbouring particles are moved fromtheir original positions so that a certain distortionof lattice occur near the .vacancy but at a greaterdistance the lattice is unperturbed. [29,32]

Behaviour of the lattice atoms on the surface ofthe crystalline compound differs also from that inthe bulk. The surface atoms have neighbouringatoms only on one side. Therefore, they have ahigher energy and they are less firmly bondedthan the internal atoms. If additional atoms wereto be deposited on the surface atoms, energywould be released just as it is released when twoindividual atoms are combined.[33] Stackingfaults, internal surfaces (i.e. grain boundaries),twin boundaries and different types of disloca-tions belong to line or plane defects. They mayalso offer preferential pathways for the transportof species in the film.[30,33] A comprehensiveconsideration of line and plane defects is beyondthe scope of this survey.

5.1.4.5 Mechanisms by which zinc affectsactivity incorporation

A general observation in all the reported zinc testshas been that Zn injection to high temperaturewater results in thin oxide layers with low visibleporosity on new metal surfaces. In addition, thealready existing oxide films do not grow in thick-ness, partly because corrosion product deposition

from the solution is minimal.[67,82,83,84,85] Ithas been postulated that zinc somehow decreasesthe defect concentration in the spinel structure byoccupying existing holes in spinel lattice. Thisshould slow down the ion transport though theoxide, leading to a reduced rate of oxide growthand the formation of thinner oxide films. [46,59,66,67] The lack of thick deposited oxide film islikely to contribute to the observed low in activitylevels in the existing oxides during Zn injectionperiods. However, the exact mechanisms by whichzinc reduces activity incorporation into the oxidefilms are not yet known. It has been postulatedthat Zn may either replace the active and inactivecobalt from the oxide[46] or may block the adsorp-tion sites for 60Co pickup.[52,59,65]

As mentioned in the previous chapter cationscan be placed in the spinel structure either intetrahedral or octahedral positions. Co as well asZn can be considered as dopant ions in the oxidestructure. To calculate displacement energies re-quired to replace one ion with another, the follow-ing energies are required: (i) Madelung and short-range energies, (ii) the crystal field stabilisationenergy for an ion in a crystal site. These calcula-tions show that zinc has a very strong stabilisa-tion in tetrahedral sites, and in fact it should beable to displace all other divalent cations from thechromites. This could explain the function of zincin promoting thinner and more protective oxidefilms and inhibiting the incorporation of cobaltinto the chromium rich oxide film.[47] Some labo-ratory experiments have shown that displacement

31

S T U K - Y T O - T R 1 5 2

reaction is possible, but a chromium rich oxidefilm (throughout the film) is required for suitablesolid state reactions to occur. In addition, highzinc concentrations as ZnO should exist on theoxide surface. [85] If the chromium rich oxide filmhas incorporated cobalt (CoCr204) and zinc(ZnCr2O4), an addition of ZnO to a stoichiometricmixture of CoCr2O4 and ZnCr2O4 may result in theformation of CoO and Co. Due to the high solubili-ty of these oxide phases, the cobalt concentrationin the aqueous phase should increase by almostfour orders of magnitude. This would automatical-ly lead to a lower cobalt concentration in the oxide

An other way for zinc to affect is to competewith cobalt for occupancy of surface sites as theoxide is formed and thereby effectively block theroutes for cobalt incorporation. [18] This can be animportant intermediate stage in the incorporationof cobalt and zinc into the oxide film.

It has been suggested that the transition metalions at concentrations higher than 10 ppb wouldpractically saturate the available adsorption sitesfor 60Co. Although eoCo could compete with allions, the total amount of 60Co which could reachthe reaction/adsorption/ion-exchange sites wouldbe significantly diluted by the competing ions.Also the fact that zinc ions have shown to signifi-cantly retard the film growth leads automaticallyto a smaller number of available adsorption sitesfor 60Co which are limited by both the thinner filmand the competition from zinc ions.[52,59,65] Theblocking of the adsorption site may control, if thesolid state reactions explained above can not occuror are very slow. The different mechanisms couldexplain why some research groups have madetotally different observations.[85]

5.1.5 Zn water chemistry andstress corrosion cracking

As discussed in the chapter 4.2, the models deve-loped for SCC assume that the anodic and corres-ponding cathodic reactions contributing to crackgrowth occur partly on or in the oxide films. Thus,the rates of these reactions may control the crackpropagation rate, in which case the properties ofthe oxide films play a crucial role in determiningthe susceptibility of the material to SCC. Thecrack initiation always involves rupture of the oxi-

de film on the construction material. Therefore, adeliberate injection of some additional ions, suchas Zn, into the coolant affecting the behaviour ofoxide films is likely to have an impact of the sus-ceptibility of the materials to SCC.

5.1.5.1 Effect of Zn on PWSCC of Inconel600 in PWRs

Primary water stress corrosion cracking (PWSCC)of Inconel 600 components has become an increa-sing problem in western PWRs. Even though the-se cracks have not been a safety issue, they pose asignificant reliability and economic concern. Thefirst PWSCC indications were observed in U-bendsteam generator tubes. The SG tubing is a physi-cal boundary between the primary and the secon-dary waters. Hence it is essential that the integri-ty of the tube remains secured. During the lastyears cracking has also occurred in various typesof penetrations, such as instrument and pressurevessel penetrations, pressuriser heaters and nozz-les. The cracking in PWSCC is controlled by acombination of stresses, environment (temperatu-re, pHT, [H2]gaseous etc.) and material microstructu-re (distribution of carbides, grain size, etc.). [19,101] The initiation needs an induction time, i.e.the time needed for an apparently smooth surfaceto develop a crack. This is followed by crack propa-gation, which finally leads to tube rupture.

Early studies by Esposito et al. showed thatzinc injections (50 ppb doses as zinc borate) ap-peared to significantly affect the stress corrosioncracking behaviour of Inconel 600 MA steam gen-erator tubing material. [20] These tests were car-ried in typical PWR environments with 25 cc/kg ofhydrogen using highly stressed U-bend specimens(RUBs). In the tests where Zn was injected intothe water formed oxide layer on all specimens thewas thinner. The surface film analysis showedthat Zn existed throughout the film. In addition,there were significant changes in the PWSCCinitiation times in all materials between the sam-ples which were exposed to waters with andwithout Zn as shown in Figure 13.

The time for PWSCC initiation of the speci-mens made from the heats 1019 MA and 96834MA in zinc containing water was 2.5 times longerthan for the specimens exposed to the coolantwithout zinc. The thermally treated specimenmade of heat 752537 TT did not experience

32

STUK-YTO-TR 1 52

PWSCC initiation at all in zinc containing water,even though 50% of the specimens exposed to hightemperature water without Zn experienced theSCC initiation. Zinc seemed also to reduce therelative extent of the cracking, in terms of lesscrack density and shorter crack lengths.[20]

A more detailed analysis of the same speci-mens done by Byers et al. revealed that iron andchromium concentrations in the oxide of Inconel600 were higher when also Zn was incorporatedinto the oxide film. The analysis of chromiumbinding energies indicated that a chromium-zincphase was possibly formed.[68]

Due to the potential benefits of zinc additionson mitigating PWSCC of Alloy 600, the followingdemonstration was carried out at Farley 2 unit.The zinc concentration (added as zinc acetate) waskept at 40 ppb in the primary coolant for 9 monthsduring the cycle 10. Eddy current measurementswere made from all hot leg tube ends at the top ofthe tube sheet during the outage. These measure-ments showed that the rate of PWSCC had de-creased during the presence of Zn. The detailedanalysis of the formed oxide layers asserted thatzinc was incorporated into the oxide in the sameextent as in the laboratory tests during whichdecrease of PWSCC had been observed. However,due to the rather short zinc injection period andpossible contributions from other actions taken,

5000 -,-

s>~ 4000co

£ 3000 - -ooCO

2000 - -

1000 - -

1200ppmB,Z2ppmLi

5 Primary Water Environments

1200 ppm B. 22. ppm U, 20 ppb 2n

i I1019 MA24% Strain

96834 MA12% Strain

752537 TT24% Strain

Figure 13. Mean PWSCC initiation times for dif-ferent conditions of Inconel 600 reverse U-bend spec-imens. [20]

no firm conclusion on the benefit of zinc additioncould be drawn. These referred actions takenincluded shotpeening of the hot leg tube ends in1987 which most likely had resulted in a decliningnumber of new indications of PWSCC since 1990.[76,77]

Bergmann et al. have reported results from thecrack growth rate tests using passively loaded CTspecimen made of Inconel 600.[77] No crack prop-agation was observed in the specimens exposed tothe coolant containing zinc, whereas all specimensin the coolant without zinc exhibited crack propa-gation. However, quite different results were ob-served by Airey et al.[105] They dosed 40 ppb ofzinc (as zinc acetate) into typical PWR primarycoolant but did not see any effect of zinc on theSCC initiation times with the used specimens(reverse u-bend and bent beam specimens). Thecrack growth rate measurements (using compactwedge open loading specimens) did not indicateany benefit from the zinc injections either. Howev-er, the test results showed that the formed oxideswere thinner and enriched with chromium due tothe zinc addition.

5.1.5.2 Effect of Zn on IGSCC in BWRs

Andresen et al. [107,108] have studied the effectsof Zn on stress corrosion cracking of different allo-ys under the BWR conditions using CT fracturemechanics specimens. The specimens were expo-sed to high purity water containing differentamounts of oxygen and impurities under study at288 °C. Zn concentrations of 5 to 100 ppb (as ZnO)were found to reduce crack growth rates in allstudied materials. However, the factor of improve-ment with Zn addition was usually lower thanthat obtained when changing from NWC to HWC.Some of the results are summarised in Figure 14,which shows the crack growth rate (CGR) as afunction of corrosion potential and zinc concentra-tion.

The results clearly indicate that zinc had themost significant effect on the CGR when thecorrosion potentials of the studied materials werearound 0 mVSHE. This was the low limit of theinvestigated potential range during the Zn tests.At higher potentials the difference between crackgrowth rates in non-Zn and in Zn containingcoolants was less obvious, Andresen et al. related

33

STUK-YTO-TR 1 52

this observation to two issues: first, at highercorrosion potentials, the potential gradient withinthe crack becomes larger which increases theanion concentration within the crack but drivesmore cations (including Zn2+) out of the crack.Second, at higher potentials the crack growthrates in general are higher therefore providingless time for Zn to have effect on IGSCC. Thelargest effect of Zn resulted from long term expo-sure, low corrosion potentials and long hold timesbetween unloading cycles.

Similar test results were also obtained by Het-tiarachchi et al.[109] They reported that if theoxygen concentration was reduced below 100 ppb,the probability of IGSCC to occur could be signifi-cantly reduced in the presence of zinc (dosed asZnO). This improvement existed even at very lowzinc concentrations (3 ppb), and no additionalbenefit was observed using higher Zn concentra-tions (100 ppb). They measured also the corrosionpotential of the specimens as a function of oxygenconcentration with and without Zn. In the pres-ence of zinc, the corrosion potential of specimensdecreased approximately 70 mV in solutions con-taining high O2 concentrations. With low oxygenconcentrations, the decrease in corrosion potentialwas even larger (roughly 120 mV), as shown inFigure 15.

It was concluded that addition of Zn providesthe benefit by moving the corrosion potential intomore negative values similar to hydrogen. [109]However, the effect of zinc was significantly small-

CO

1E

Rat

e

"5<5

1o

1 n• k/

10"

-6

7

10"8

10- 9

Sens. 304 Stainless SteelConstant K = 30288°C Water

• 304SS• A-182• A-600Open = NoZnClosed = +Zn

Maximum benefit of Zn(see arrows) appearsto increase at low $c >(or low growth rate) /

y.—-—

-0.6 -0.4 -O.2

Corrosion

ksiVirt

//

/ l >

/ 4// ./ /' //

A-182

/

sA

A20-100 ppb Zn

0

Potential, V

,°AL6 | A^OO/ • S-iOppb

t # 304SS20-100 ppb

3O4SS(est)20-100 ppb Zn

• . —

0.2 0.

iha

103

-5"

102

101

4

er. This was clearly shown in the test in which thewater contained 200 ppb of oxygen, 100 ppb Znand 10 ppb hydrogen. The crack growth rate ofAlloy 182 weld metal was reduced only whenexcess of hydrogen was present in the water.[110]This agrees with the results reported by Andresenet al. (see above).

According to the slip-dissolution model, an in-crease in oxide film rupture strain or re-passiva-tion kinetics of a new surface at the crack tip willimprove the resistance to IGSCC or reduce thecrack growth rate. The results of Angeliu et al.show that as the zinc concentration in the coolantincreases, the oxide film rupture strength of AISI304L SS increases as shown in Figure 16.[111]

50 100 150 200 250 30C02/ppb «*<•""•'

Figure 15. The influence of zinc on the corrosionpotential of the specimen at different dissolved oxy-gen levels.[109]

u.Olb

0.014

% 0.012

1OS

3 0.010

B0.008

r

I0.006 0

water purity without Zn, <0.1 microS/cm Mand with 60 ppb Zn, <0.5 microS/cm - /

s .-•'•'-

1 deaerated• 200 ppb oxygen

•

•''/ aU except 40 ppb Zn are/ an average of 2 tests

20 40 60 80Zn (ppb)

Figure 14. Overview of the crack growth rate re-sponse vs. corrosion potential and Zn addition.[108]

Figure 16. Oxide rupture strain as a function ofZnfor type AISI 304L SS exposed to 200 ppb O2 anddeareated high purity water up to 166 h at 288 eC.[Ill]

34

STUK-YTO-TR 1 52

In addition, long term tests showed that anexposure over 300 hours to a solution containingonly 20 ppb of Zn resulted in similar oxide rupturestrain as with 60 ppb of zinc in short term tests.High zinc concentrations provide a greater drivingforce and availability for Zn incorporation. Howev-er, similar effects can be obtained with low Znconcentrations if long enough exposure times areallowed.

Angeliu et al. have also measured the re-passivation kinetics of Fe-12 Cr steel in hightemperature water. In all the studied environ-ments, the measured re-passivation current den-sities were similar immediately after the filmrupture with and without Zn.flll] This has beenverified also with more recent measurements,which showed that re-passivation kinetics of In-conel 600 (up to -100 s) were not affected by Znaddition.[112,113] However, after 104 s the meas-ured current densities of Fe-12 Cr steel werelower in the presence of Zn. This indicates thatthe study of the mechanisms by which zinc ischanging or affecting electrical and mechanicalproperties of the oxide films, requires long enoughexposure time.

5.1.5.3 Comments of zinc effects on SCCAs discussed in the chapter 4.2, the properties ofthe oxide films on construction material surfacesmay have a strong influence on SCC. The thick-nesses of the oxide films have been lower in allzinc injection tests than in the tests withoutzinc.[67] It has been postulated that Zn2+ slowsdown the iron transport though the oxide, leadingto a reduced rate of oxide growth and to the for-mation of thinner oxide films. It is assumed thatthis kind of thinner oxide is less likely to breakand expose the base metal to the environmentthan a thicker oxide with a higher defect concent-ration. [106] Therefore, the influence of zinc onSCC is likely to be due to its incorporation into theoxide films. To be able to affect cracking processzinc has to reach the oxide surface in the crackand have enough time to be incorporated into theoxide. The probability for these processes to takeplace depends also on the corrosion potential asdescribed below.

At higher corrosion potentials, the potentialgradient within the crack becomes larger andincreases the anion concentration within the crackand drives more cations (including Zn2+) out of the

crack. Therefore, the laboratory results whichshow that at higher corrosion potentials the differ-ence between crack growth rates in non-Zn andwith Zn containing coolants was less substantial,are consistent with the ion distribution within thecrack.[107,108] It has been shown both in labora-tory and in operating power plants that additionof Zn (as ZnO) increases pHT of the high puritywater moving the ECP of the materials slightly inthe negative direction. This decreases the poten-tial difference in the water between the crackmouth and crack tip and enables more zinc ions tobe transported into the crack surfaces.[109]

Secondly, at higher corrosion potentials thecrack growth rates in general are higher providingtherefore less time for Zn to have an effect. Thelargest effect of Zn resulted from long term expo-sure, lower corrosion potentials and long holdtimes between unloading cycles.[107,108] The re-passivation kinetics of fresh crack surface in hightemperature water were similar with and withoutZn immediately after film rupture. This indicatesthat the mechanisms by which zinc is changing oraffecting the electrical and the mechanical proper-ties of the oxide films, requires long enough expo-sure time in zinc containing coolant. On the otherhand, higher zinc concentrations provide a greaterdriving force and availability for Zn incorporationresulting accordingly, shorter exposure times toobserve the benefits.[Ill]

5.1.6 Detrimental effects of zinc

As discussed in the chapter 5.1.1.1, the plants do-sing natural zinc have observed that the activati-on of 64Zn to form 65Zn adds to the inventory ofactivated corrosion products in the plant. Zn injec-tions have also resulted in the deposition of hard,tenacious crud on fuel cladding surfaces. For somereason eddy current measurements have given ex-traordinary high values for oxide thickness on fuelcladding surfaces. In addition, the restructuredcrud deposits on fuel cladding surfaces have beenreported to be resistant to brushing with stainlesssteel bristles. The changes in crud loading duringthe zinc injection periods at Hatch-1 unit areshown in Figure 17. The zinc concentration in theoxides on cladding material has increased from5% up to 20%. Prior to zinc injections, the Zn sour-ce had been the brass condenser tubes. The depo-sits were analysed to have a structure typical for

35

STUK-YTO-TR 1 52

magnetite, but based on the chemical analysis thecomposition of the spinel was most likely ZnFe2O4.Crystallographic analysis also showed that theformation of spinels in the fuel deposits increasedwith increasing zinc concentrations in the reactorwater.[91]

The rod average crud loading on the Hatch 1rods was relatively low, as shown in Figure 17,partly due to low feed water Fe concentrations. Onthe other hand, the crud loading on the HopeCreek rods was relatively high, because of a high-er feed water iron input (6-8 ppb). In general, thethermal conductivity of iron based oxides in solidform is higher than that of the zirconium oxide onfuel cladding surface. However, if dry steam formswithin a delaminated oxide or tenacious deposits,the thermal conductivity could decrease by a fac-tor of 8, increasing the temperature of the fuelcladding. Thus, the combined effect of increasedtenacity and crud loading with increasing burn-upcould become significant.[61,91,92]

In PWRs, the fuel performance has been lesssensitive to water chemistry conditions, mainlybecause of the negligible surface boiling. Additionof Zn to the primary coolant has been suggested topresent a potential risk for fuel reliability. Howev-er, no detectable effects on cladding corrosionhave been observed at Farley-2 unit after Znadditions for 9 months. In addition to this plantdemonstration, an in-core loop test was carriedout at Halden test reactor to provide an earlywarning of any deleterious effects of zinc injec-tions on the fuel cladding behaviour. Both freshand re-irradiated fuel segments were exposed to

Table VII. Chemical decontamination results inzinc addition plants.

Plant

Millstone Pt 1

FitzPatrick

Monticello EOC-14

Monticello EOC-15BRAC Pts

All Points

Hatch 1

BRACPtDF

22

11

25

7

2

8

Pre/PostDose rate

(mR/hr)

217/10

113/10

613/25

3 pts - 400/604th pt - 3000/450

1250/430

180/22

PWR conditions with 50 ppb of Zn. The resultsshowed no apparent effects of zinc on either corro-sion or hydriding of the low tin and standardZircaloy cladding. Although no effect of Zn isanticipated based on the experience from Haldentests and Farley-2 results, continuous monitoringof fuel is recommended if the zinc injections arecarried out.[76,77,93,94]

One of the early concerns in the plants usingzinc dosing was whether the oxide films formedduring the presence of zinc could be decontami-nated using currently available chemical process-es. In the past years, decontaminations have beencarried out in several plants and they have beenhighly successful in all but one case, as shown inTable VII. However, the precise cause for theineffective decontamination is not yet known. [61]

One Cycle Bundles Three Cycle Bundles

Zn.

11 12 12 13 14 15 16Cycle*

Cycles Cycles Cycles10-12 12-14 14-16

Figure 17. Crud loading in Hatch-1 fuel bundles under HWC chemistry with Zn and no-Zn addition.[91]

36

STUK-YTO-TR 1 52

5.1.7 Alternatives for zinc injection

To overcome the problems associated with the useof zinc water chemistry, several research groupshave studied alternatives for zinc by evaluatingthe general properties and characteristics of diffe-rent metal ions. A possible way to prevent deposi-tion of cobalt into the oxide surfaces is to maintainits concentration in the aqueous phase as low aspossible compared with the concentrations of ot-her ions which compete with Co for adsorption orlattice sites. According to Niedrach and Stadderthe order of incorporation of transition metal ionsinto the existing oxides is Ni = Co > Zn > Mn.[52]This was supported by the results of Lin et al.,who showed that both zinc and nickel can effecti-vely suppress the S0Co build-up in the oxide filmas shown in Figures 18 and 19. It has been sugge-sted that other transition metal ions such as Mg2+,Mn2+ may behave similarly. [52]

The results of Osato et al. indicated that theeffects of Ni2+ dosing into the high temperaturewater decreases the up-take of cobalt into theoxides under HWC environments. The activitybuild-up on electropolished and as-received stain-less steel samples became lower during Zn and/orNi addition when compared to the samples, whichwere exposed to the high purity coolant. However,the activity in-corporation into the carbon steel

2.0

"E

3

ION

CEN

TRA

Co-

60 S

URFA

CE C

ON

P

-

0c

-

HWC-REFERENCE ^*

^ NWC-REFERENCE

jr S C u l A l

/ /yT Ni ( O )I ff ^ r-j Q A

/ _ ^ ^ - " Z n ' D )

^ ^ , , ,

600 1000 1500 2000

EXPOSURE TIME (HOURS)

samples was not inhibited to the same extent.[87]Korb et al. have also studied alternatives for

zinc. The remaining elements of interest left aftertheir evaluation were Ca, Mg, Mn and Sr. Fur-thermore, the results showed that for trivalentcations the chromites have higher stability thanthe ferrites. With regard to bivalent cations, Zn,Co, Ni, Fe, Mg and Mn, the trends are not sosystematic. In addition, the reaction conditionshad a considerable influence on the order of stabil-ity of the spinels involved. However, some conclu-sions were drawn: Co spinels appered to be morestable than the spinels containing Fe and Ni.Moreover the affinity of Zn for spinels was sug-gested to be even higher than that of cobalt.[81]

Uetake et al. studied also how different ionsaffect the deposition of activity into the oxide filmson AISI 316 SS. They found out that Al has almostsame reducing effect as Zn as shown in Figure 20.Mg2+ was also studied with same molar concentra-tion, but no inhibition of activity incorporationwas seen.[90]

Iron injection in BWRs has been widely used inJapan as well as in some plants in USA andEurope. The idea is to control the composition ofcrud on fuel by means of controlling the concen-trations of Ni and Fe in the coolant, in order tominimise the concentrations of soluble 60Co and58Co. The crud on the fuel consists mainly of Fe

HWC-REFERENCE ^

Ni I O I Z n l D I

•y500 1000 1500

EXPOSURE TIME (HOURS)

Figure 18. Comparison of e0Co deposition on as- Figure 19. Comparison of 60Co deposition on as-received AISI 304 SS samples under NWC condi- received AISI 304 SS samples under HWC condi-tions with metallic ions at 15 ppb.[52] tions with metallic ions at 15 ppb.[52]

37

STUK-YTO-TR152

and Ni, which have been deposited as NiFe2O4. Ifthe iron to nickel ratio in the coolant is two to one,Fe and Ni will form nickel ferrite, from whichcobalt can replace nickel to some extent. Thisleads to lower levels of activated cobalt in thecoolant due to the low solubility of the spinels. Toosmall amount of Fe in the coolant results in theformation of NiO and subsequently to a higherrelease of 60Co and 58Co. The effects of iron injec-tion on the concentrations of activated cobaltisotopes and nickel in the reactor water are shownin Figure 21.[88]

However, some Swedish results have shownthat iron injection has led to increased activitybuild-up in the oxides on SS piping surfaces.[88]This has been reported to be due to the changedoxide structure in the primary circuit. During thelow iron concentration in the coolant, typical BWRoxide loses the outermost hematite layer andleaves behind a nickel ferrite oxide layer, whichreduces the tendency of the surface to take up Sb-124.

c

q/cr

ivit

ted

radi

oact

epos

4 0 0

3UU

200

100

4

O :D:• :A :

_

E 1

no additionZn additionMg addition QAl addition Q^T

/

ex—Ar-^jpi ^ t r i r tF

50 100

time(h)

A

r

150

—

2 0 0

Figure 20. The effect of metal ion additions on Coincorporation into existing oxides.[90]

In some Japanese plants 60Co concentrations inthe coolant have increased even though the Fe/Niratio has been optimised. The reasons are stillunknown but several possible mechanisms havebeen proposed, such as the adverse effect of notpre-oxidised fuel surfaces and higher than normalCr concentrations in the coolant.[89]. Uetake et al.confirmed that the fuel surfaces must have anexisting Fe crud before the iron injections remaineffective. The metal ions deposit as hematite andnickel and Co monoxides on fuel cladding surfacesas a result of growing steam bubbles. As thebubble reaches a critical size it will leave thesurface. A part of the deposits re-dissolve and therest of them form ferrites. This process allowsincorporation of Co into the oxides. If enough Feexists in the solution more Co will be incorporatedinto the stable spinel.[90]

Nevertheless, the overall understanding of theprocesses within the oxides is still lacking. Thiscomplicates the optimisation of primary coolantconditions to further decrease dose rates at oper-ating power plants.

5.2 Noble metal water chemistry

5.2.1 Principles of noble metal waterchemistry

As discussed in the preceding chapters 4 and 4.2,the susceptibility for the stress corrosion crackingof construction materials in BWRs can be mitiga-ted by adding hydrogen into the feed water. Thisdecreases the levels of oxidising species and thuslowers the corrosion potential (ECP) of theconstruction materials. This has been shown toapply both under irradiated (IASCC) and unirra-diated (IGSCC) conditions as shown in Figures22a and 22b.[114]

ZO0&O7

1.50&O7

I.1.0C&-07

5.0C&06

0.O0EK»:

3XX&07

200&07

1.0C&07

accp*m

* \A

-o-FWFe

y

0.8

0.6

0.4

0.2

Jarv94 Feb-94 Mar-94 Apf-94 May-94 Jun-94 Jul-94 Jan-&4 JJ-94

Figure 21. Effect of iron injection on cobalt isotopes in reactor water at Forsmark 3.[88]

38

STUK-YTO-TR 1 52

The low corrosion potentials of constructionmaterials are obtained only if all oxygen whichdiffuses onto the material surface is consumed bythe formation of water. The minimum amount ofhydrogen needed (CH:C0 = 2:1) corresponds to a1:8 CH/C0 weight ratio. Therefore, the hydrogenthreshold concentration is > l/8th of the oxygenconcentration expressed for example in ppb. Labo-ratory experiments have shown that even lowerhydrogen concentrations (CH:C0 = 1:12) are suffi-cient to decrease the measured corrosion poten-tials, because the diffusivity of hydrogen in thewater layer next to the oxide surface is higherthan that of oxygen or hydrogen peroxide. Themechanism by which catalysed surfaces behave isshown in Figure 23.[116]

The measured or calculated ECP depends on:

(1) the exchange current densities (io) for differentreactions (i0, reversible reaction rate at equilibri-um), (2) the rate constants for reactions whichdetermine the activation controlled polarisationresponse and (3) the diffusion coefficients forgases and ions, which determine the diffusioncontrolled response (or the limiting current densi-ties, iL shown in Figure 23). These parameters aredependent on the studied system. They are affect-ed by O2, H2, H2O2 concentrations, the surfacecondition, the surface composition (stainless steel,noble metal modified alloys, etc.) and the flow rateof bulk solution.[117]

The corrosion potential is governed by a bal-ance between the total oxidation and reductionreaction rates that occur on the material surface.In a simplified situation as shown in Figure 23,

10*

304 Stainless Steel

25 mm CT Specimen' Constant Load

288° WaterTest Conditions:«15 C/cm2 EPRo27.5 MPaVm0.1 -0.3/ iS/cm

O 10"8

a)10"9

Predicted CuivesFrom PLEDGE Code

Post-irrad. SSRT 2-3Comm. Purity 304SS (••) & 316SS (A)• • A 42 ppm O2-sat'd vs. oo&0.02ppmO 2

Data Shifted Right by Init. GB Cr Enrichment

-06 -0.+ -0.2 0Corrosion Potential,

0.2 0.4

1020 1 Q 2 ,

Neutron Ruence, n/cm2 (E>1MeV)10B

Figure 22. Observed and predicted crack growth rate as a function of corrosion potential.[114]

t_J

.NTI/

POTE

Reversible O;/H,O

N "\ N >,

T* ^

/

/

^ E 2 P , E,p,

'"A

E4P,

Esss

v ^ ^ ^ H2 Oxidation on SS

E6SS ^ " ~ \ ^ ^

/ H, Oxidation on Pt

^ v H2O Reduction on Pi

Reversible H,/H,O >v. . \ ^\ , , e uctionon ">v

Log 1 i 1 •

Figure 23. Schematic Evans diagram showing the intersection points for the curves of oxygen reductionand hydrogen oxidation on stainless steel and Pt.[116]

39

STUK-YTO-TR 1 52

the corrosion potential (EiSS/pt ) is fixed to thepotential, where the O2 reduction curve intersectsthe H2 oxidation curve. Typically, the constructionmaterials have a low exchange current density forO2 reduction and H2 oxidation. This means thateven with moderate oxygen concentrations thecorrosion potentials of different materials can riseto fairly positive values as shown in Figure 23 bythe points E3SS,...,E7SS. Low potentials can beachieved only with very low oxygen concentra-tions where O2 reduction becomes diffusion limit-ed (like the curve for iL1 in Figure 23) and inter-sects the H2 oxidation curve at a current density,lower than the exchange current density for H2

oxidation. However, catalytic surfaces have a highexchange current density for H2-H2O oxidationand reduction. Therefore, the corrosion potentialof Pt stays low even at fairly high oxygen concen-trations as shown in Figure 23 by the pointsE2Pt,...,E6Pt . This is the basic phenomenon whichexplains the efficiency of noble metal coatings inobtaining low corrosion potentials of the in-corecomponents even with low hydrogen dosing intothe feed water.

Noble metals have been used as catalytic sur-faces in science for a long time. If the constructionmaterial surface could be made to behave like anoble metal surface, it would catalyse the recom-bination reaction between hydrogen and oxygen,decreasing the corrosion potentials with reasona-bly low hydrogen concentrations. This would also

6

Main 4

SteamLineRad 2

Increase

0

0

fn -0.2Vessel

ECP-0.4

-0.6

-

Protection Potential

Feedwater H2 Addition Rate

Figure 24. Schematic basis for the use of noblemetal technology in BWRs.[115]

result in lower main steam line dose rates asshown in Figure 24. In plant applications theamount of hydrogen needed to reduce potentialsenough has shown to be plant specific and isalways in excess of the stoichiometric amountrequired for the recombination with oxygen toform water. In the out-of-core regions, hydrogenadditions have resulted in low enough corrosionpotentials, but the materials in in-core locationscan be protected only with much higher hydrogenconcentrations.

An additional benefit of a lower corrosion po-tential is higher acceptable impurity levels in thecoolant without effect on the crack growth rates asshown in Figure 25. During this test the conduc-tivity was changed from 0.11 iiScm-1 to 0.86 |j.Scm-1 without any effect on crack growth rate. Thisshows that low corrosion potentials provide hightolerance to severe water chemistry transients(e.g. impurity in-leakage).

5.2.2 Different types of noble metalcoatings

A variety of different types of noble metal coatingtechniques have been developed for improving thecatalytic properties of oxide surfaces on structuralmaterials. The noble coating on the metal surfacecan be obtained using electro- or electroless pla-ting and vapor deposition. However, these techni-ques can be applied only in autoclaves or in otherlaboratory environments. Surface analysis has in-dicated that typically several weight percent of

£ 300x

Narrow Sens 450C. 304 SS AJ9139 c57 [38x]Const K=33MPo/m + R=0.7, O.OIHz every 1000s400 ppb 02, 78 ppb H2. 0.109 uS/cm H2S04

(excess H2 with Pd-ized CT specimen)

Linear RecessionSlopes » 0.07 uro/h 2.8 um/h/ 0.D5 um/til

fro M2504 J/cin).863 uS/ci

To 7S ppo Hj400 ppb 02(excess H2)

iom« H2SO4

0-5900

Jo 4 8 ppBHZ1000 ppb 0 2(excess 0 2 )

w m e H2SO4

6000 GIOO an

Time, hours6300

1.6

14

1.2 -^in

1 if*>

0.8 §

0 , 1§

O.J I

0.2

0

Figure 25. Response ofPd coated CT specimen atlow corrosion potentials and high conductivity val-ues.[117]

40

STUK-YTO-TR 1 52

the amount of noble metal was incorporated intothe oxide film to the depths of several tens of nm.However, it is also likely that some areas had thin-ner coverage of noble metal.[117,118]

More dilute noble metal layers can be producedby using thermal spray coating of noble metalalloy powders or by direct alloying of the materialitself. Under water thermal spray coating tech-niques can be carried out in high purity watereither in shallow or deep waters. The catalyticproperties of metal surfaces have been obtainedby using hyper-velocity oxy-fuel (HVOF) and plas-ma spray (PS) coating techniques. The used coat-ings consisted of AISI 309L SS powder with 0.42%of Pd and Alloy 82 with 0.4% of Pd. In the HVOFprocess, gas mixtures of propylene/oxygen or hy-drogen/oxygen were used. In the PS process, a dcplasma arc is created in an inert gas (helium and/or argon) between anode and cathode, both in thetorch head. The powder is fed into the ionised gasstream and is heated and accelerated towards themetal surface. [119] The results of Kim et al. showthat HVOF and PS coatings responded fully cata-lytically in the presence of a stoichiometric excessof hydrogen in different oxygen concentrations. Ithas been reported that the corrosion potential ofstainless steel increases more significantly by theaddition of H2O2 when compared to the effect ofoxygen. However, Kim et al. have shown that PSand HVOF coatings exhibited good catalytic be-haviour in water containing 500 ppb of hydrogenperoxide when excess of hydrogen was added tothe water. [120] These techniques have some limi-tations when applied in operating BWRs, because

300

n

£250

C-SC0.2

J 150

g toou

SO'

BPV1111 3O4SS+O.15SPd C63 [10y]Kmox»=30 ksi/in + unloading cycle

of R-0.7. 0.01 Hz every 1000sExcess H2. 0.435 uS/cm H2SO4100 ppb 02. 48 ppb H2

ITo H 2504C.8S3 uS/or

TOO 750 800 BO 900Time, hours

I.S

1.5

1.2

0.9

0.6 i

0.3

0 \

-0.3 1

1000 UK 1100-0.6

Figure 26. Crack length, corrosion potential andoutlet conductivity vs. time for CT specimen.[122]

of the poor accessibility of a large number of in-core components. For some components like thetop guide, parts of the core shroud etc. thesetechniques seem to be very attractive.

Direct alloying of the construction materialscan be used only in replacement parts. Neverthe-less, the technique has been found to improve thecatalytic efficiency for hydrogen/oxygen recombi-nation in high temperature water with relativelysmall amounts of hydrogen.[121] Andresen et al.studied the crack growth rates of the noble metalalloyed stainless steels in high temperature waterin different conductivity ranges and in differentO2 and H2 concentrations. One set of results isshown in Figure 26.[122] The specimen was notsensitised and therefore higher conductivitieswere needed to increase crack growth rates, butthe benefit of noble metal coating can be clearlyseen from Figure 26.

A system wide approach could be the applica-tion of the noble metal chemical addition (NMCA)technique, in which the reactor coolant is used asthe medium of transport for depositing smallamounts of noble metal onto the oxide surfaces.The first verification tests with NMCA were car-ried out in autoclaves. The noble metal compoundwhich was added into the water was either palla-dium acethylacetonate or palladium nitrate. Dur-ing this high temperature water exposure thenoble metal concentration in the water variedbetween 10 and 100 ppb. The effect of Pd NMCAon the corrosion potential of AISI 304 SS specimenin different hydrogen/oxygen molar ratios isshown in Figure 27.[110]

0.1

0

-0.1

304 SS

304 SS / Pd

2 4 6 8 10 12 14Molar Ratio (H2/O2) »«.«««

Figure 27. Corrosion potential response of a Pddoped and a reference AISI 304 SS specimen toH2/O2 molar ratio.[110]

41

STUK-YTO-TR 1 52

According to the AES results of Kim et al., Pdadded as acethylacetonate was present in theoxide surface at a few atomic percents to thedepths of about 400 A. [124]

5.2.3 Long term stability

To be a viable process in operating BWRs, theapplied coating should retain the catalytic activityover an extended period of time under high flowrates of the BWR coolant. In addition to this, thecoating should withstand the changes from NWCto HWC. The results from the test, which wasperformed to verify the operation of the catalyticproperties over the period to 12 months with wa-ter chemistry changes from normal to hydrogenwater chemistry are shown in Figure 28.[123]

The ECP increases with increasing flow rate asshown earlier (Figure 1). The high flow rate itselfcould pose a risk in promoting erosion corrosion ofthe noble metal coatings. Kim et al. studied theeffects of high flow rate and ultrasonic exposure tothe catalytic properties of the exposed surfaces.They found out that one week exposure to highwater velocities increased the ECP from -500mVSHE to -250 mVSHE.[124] A similar increase inpotential was also observed after the ultrasonictests.

To further improve the long-term stability ofthe noble metal coatings, a mixed noble metalapplication has been tested. Initial data showsthat the durability of noble metal coatings can beimproved by doping simultaneously several noblemetals (e.g. Pt, Ir, Ru, Rh, etc.). The first laborato-

0.1

0

-0.1

-0.2

-0.3

-0.4

304SS 10 Mos exposure

SS/Pd(NMCA) 9 Mos exposui

SSZPd(NMCA) 7 Mos exposure

SS/Pd(NMCA) 3 Mos exposure

ii , , , , i i , , i i

0 1 2 3 4 5Fluid Velocity ft/s

Figure 28. Corrosion potential response ofPd/SSafter 9 months durability test under high flow.[123]

ry results have indicated even better catalyticactivity of the mixed coated surfaces for the sameflow velocities, indicating also better durability inhigh water flow rates.[124]

5.2.4 Effect of the operationalenvironment

The plant observations have indicated that thechange from NWC to HWC causes reduction inthe oxide film thicknesses. This is mainly due tothe enrichment of chromium in the film, whichleads to lower corrosion rates of the base metal.However, results reported by Hettiarachchi showthat film thinning during HWC operation does notaffect the catalytic activity of the surface in pro-longed exposure to HWC.[123]

Kim et al. studied the effects of Zn (as ZnO)and Cu (as CuSO4) on the catalytic behaviour ofnoble metal coated surfaces. [125] They used 0.1%Pd alloyed AISI 304 SS steel samples. The resultsshowed that both zinc (at 100 ppb range, notbelow) and copper increased the corrosion poten-tials of the specimens with some tens of millivolts.The conclusion was that Zn or Cu deposited on thesurface film blocked either of the half-reactionsoccurring on the noble metal coated surface: O2

reduction or H2 oxidation. Because the ECP in-creased, they concluded that the rate of H2 oxida-tion reaction was somehow decreased. The meas-urements confirmed that the recombination effi-ciency of O2 and H2 decreased by 10% when zinc orcopper was injected into the high temperaturewater under excess of H2. The analysis of theoxide showed that both Zn and Cu concentrationswere highest on the outermost surface layers, butno significant difference was observed in the oxideparticle density and size after Zn and Cu injec-tions.

To find out how the NMCA would work on thesurfaces in plants, in which crud deposition ontothe core surfaces was high, a set of tests werecarried out in laboratory environments. The re-sults of Hettiarachchi et al. indicated that typicalcrud layers (sample which had been in operatingplant for 10 years) did not affect the NMCAprocess and proper catalytic behaviour of thesurface was obtained.[123] The test also verifiedthat it is possible to use the NMCA technique to

42

STUK-YTO-TR 1 52

successfully dope all in-core components. The elec-troless coating process has also been shown towork efficiently both on bare surfaces and onsurfaces with old, representative oxide layers. Noloss of catalytic activity was observed by Andresenet al.[117]

After the NMCA had been applied in the simu-lated crudding processes, the surfaces of the speci-mens remained catalytically active even thoughthe crud layer formed on top of the existing oxidefilm had a thickness of 1 |xm. Hettiarachchi et al.have studied whether the crudding process iscatalysed on noble metal coated AISI 304 SSsurfaces, but they did not observe any effect.[123]

5.2.5 Possible side effects of noblemetal coating

One of the concerns related to the noble metalapplications has been that the corrosion potentialof Pd/Pt containing surfaces might be higher thanthe corrosion potential of uncoated materials un-der excess of oxygen/hydrogen peroxide concent-rations in the reactor water. This would lead tohigher crack growth rates in BWRs under normalwater chemistry conditions during transients fromHWC to NWC or during some other intermediatewater chemistry conditions.[126] There is somecontroversy in the results, because the laboratoryresults done by Kim showed that Pt has higherpotentials than the stainless steel samples in ex-cess of oxygen. [116] However, the Pt plate electro-des never exhibited a higher corrosion potentialthan that of stainless steel samples in in-core ECPmeasurements,as reported by Andresen.[117] Thetests carried out at Halden test reactor have alsoshown that the crack growth rates of noble metaldoped specimens were not higher compared to un-doped specimens under NWC conditions.[110]

Another concern has been the corrosion behav-iour of fuel cladding material once the NMCA hasbeen applied on the surfaces of reactor core com-ponents. Laboratory and some in-core resultsshow that noble metal coating on Zircaloy fuelcladding does not affect the corrosion nor theintegrity of the cladding material.[117] However,there are no results if noble metal coating willhave an effect on hydride formation in the fuelcladding material. Laboratory tests and modelling

work has also shown that 16N formation on cata-lytic surface is an insignificant factor. [117]

Once NMCA has been applied, the oxidesformed during NWC will start to restructure. Thiscould lead to increased 60Co incorporation into thechanging oxides. Plant measurements have shownincreased dose rates during shutdowns due to theaccumulation of 60Co into the oxides on the re-circulation piping after operation under HWC.The reason has been related mainly to the restruc-turing of the oxide films. On the other hand, thelower potentials cause reduction of oxide filmsformed under NWC operation, leading possiblyagain to increased 60Co incorporation into theoxide. However, activity incorporation into theoxide films can be mitigated efficiently by usingZn injections as shown by Lin.[8]

5.3 Application of dielectric oxideson construction materials

There are several unknown parameters in the coreof BWR conditioons, such as water flow rates, ra-diation flux, dissolved hydrogen/oxygen concent-rations, etc. A typical location with a poorly defin-ed conditions is the core channel boiling region,where both hydrogen and oxygen are strippedfrom the liquid to the vapour phase. The concent-ration of hydrogen peroxide remains high as aresult of the water radiolysis and therefore increa-sing the ECP of the construction materials in thisregion. Therefore, it is possible that neither HWCnor NMCA can provide sufficient decrease in theECP of construction materials simply because ofvery high oxidant concentrations or difficulty inachieving stoichiometric excess of hydrogen. The-se technologies may also be unattractive to theoperators.

Another possibility to reduce susceptibility ofstainless steel component to stress corrosioncracking has been proposed by Yeh et al. Theyhave modelled two real cases, in which the corro-sion potentials of materials can be decreased inoperating BWRs by reducing the exchange currentdensities of the major redox couples by usingdielectric coatings onto the metal surface.[126]Their calculations show that by adopting a gener-al inhibition technique, the reactor componentscan be protected from IGSCC even without HWC,

43

STUK-YTO-TR 1 52

as shown in Figure 29.The calculations shown in Figure 29 simulate

the situation in Dresden 2 unit. The top figureclearly shows that the noble metal coating can notprotect any parts of the heat transport systemunder NWC. However, if the oxides on materialsurfaces are insulating, only the components incore channel (1), core bypass (2) and upper ple-num (3) have potentials higher than -230 mVSHE.All the other parts are protected from IGSCC. The

situation changes when HWC is applied. By usingsome of the noble metal coating techniques mostof the components have low enough potentials.However, due to the intense radiolysis of thewater in core channel and upper plenum, thepotentials in these regions remain high. A generalinhibition method seems to decrease potentialsthroughout the heat transport system, providingprotection against IGSCC even with low hydrogenconcentrations. Calculations reported by Yeh et al.

ICO

>

otu

CO

o

111

CO

a.oui

8 I 91 1 1 1 i 1 r

General Catalysis

Noimal HWC '

General Inhibition

JL _JL J_ _L J _

0.5

0.3

0.1

-0.1

-0r3

-0.5

-0.7

0.5

0.3

0.1

-0.1

-0.3

-0.5

-0.7

0 600 1200N4|6| 6

1800 24007

3000 36008

r~i 1 1 1 1 1 1 1 1 1 1 i 1 1 1 rnI — -~- Normal HWC

General Catalysis

"**••- General Inhibition

i i I i i I I i I i i I I I I0 600 1200 1800 2400 3000 3600

-

-

J

I 11

1

1

^ 4 | 5 |

1

, ^ ^

— J J*.

\ \

6 | 7 |i l 1 i l 1 l l I 1 i

• —"""" ~ " -~- ^ Normal HWC

General Catalysis

. . ._ General Inhibition

i i 1 i i 1 i i 1 ? i

8

1 I

- - - - - -

i i

9

I1

\ -

! -

600 1200 1800 2400 3000

Flow Path Distance from Core Inlet (cm)

3600

[H2]FW0.0 ppm

[H23FW0.5 ppm

1.0 ppm

Figure 29. Corrosion potential variation as a function of feed water hydrogen concentration along theheat transport circuit at Dresden-2.[126]

44

STUK-YTO-TR 1 52

are only qualitative in nature. However, theyshow that by adopting a general inhibition tech-nique, the reactor components can be protectedfrom IGSCC.

A similar approach was used by Kim et al., whomeasured the ECPs of materials, which formed anelectrically non-conducting oxide film onto themetal surface.[127] This type of films in hightemperature water could be obtainable by coatingthe construction material surfaces or by alloyingcertain elements, such as Zr and Y, into the basemetal. The behaviour of the corrosion potentials ofthe specimens coated either by plasma spray withyttrium stabilised zirconia (YSZ) or immersed in

water containing 1 mM ZrO(NO2)2 are shown inFigures 30 and 31. The former specimens behavedlike pure Zr having low potentials even at ratherhigh oxygen concentrations.

At oxygen levels higher than 300 ppb ECPstarted to increase rather rapidly, most likely dueto the presence of excessive pores and cracks inthe coating. However, when the oxygen concentra-tions were lower than 180 ppb, the measured ECPwas somewhat lower than with untreated sam-ples. At the moment the thermal spray coatingseems to provide the most promising approach inlowering the corrosion potential of constructionmaterials below the IGSCC protection potential.

S1/3

400

200

uHOCu

otoO£" -60005

-200

-400

-800

e Zr

IPC(YSZ) on 3O4SS288C, 200cc/mln.Electrodes faav« been I2SSC water for 3 months.

J.01 102 103

OXYGEN CONCENTRATION IN WATER, ppb

Figures 30. Corrosion potentials ofAISI 304 SS, pure Zr and YSZ coated AISI 304 SS electrodes.[127]

400288C, 200cc/mln.

-800

IOppm ZrO2, 60CultrasonlcallyIPC-1: 10 daysIPC-2: 20 days

lO1 102

OXYGEN CONCENTRATION IN WATER, ppb

Figure 31. Corrosion potentials ofAISI 304 SS, pure Zr and ZrO2 doped AISI 304 SS electrodes.[127]

45

STUK-YTO-TR 1 52

6 SUMMARY AND CONCLUSIONS

The primary coolant oxidises the surfaces ofconstruction materials in nuclear power plants.The oxide films formed in high temperatureaqueous solutions generally consist of a compactinner layer and of a more porous outer layer. Alt-hough this duplex-film concept is a simplification,it can be used to describe the behaviour of oxidefilms in different operational environments. In ad-dition to physical differences, the composition andstoichiometry of the oxide films changes graduallywith distance from the film/environment interfa-ce. The nature of the oxide films influence signifi-cantly the extent of incorporation of activated cor-rosion products into the primary circuit surfaces,which may cause additional occupational doses forthe maintenance personnel.

The growth of oxide films and the oxide thick-ness is determined by the rate of ion transportthrough the existing oxide and by the extent ofsaturation of the coolant next to the oxide layer bysoluble corrosion products. During deposition ofcorrosion products, activated species are also in-corporated into the oxide films. The incorporationof radioactive cobalt into oxide films can basicallyproceed via at least three different mechanisms:surface adsorption/complexation, ion-exchange ordirect reaction/crystallisation. In addition, diffu-sion along the pores and cracks in the outer partof the duplex oxide film, as well as diffusion alongthe grain boundaries in the dense part of theoxide, enables the activity to be spread through-out the oxide film.

Zinc and activity build-up

Injection of zinc into the primary coolant has beenshown to decrease the incorporation of activatedcorrosion products into the existing oxides. Eventhough the exact mechanisms by which zinc actsare not yet known, it is assumed that Zn may

block one of the above mentioned diffusion paths.Zinc may also decrease the defect concentration inthe spinel oxide structure by occupying existingdefects in the oxide lattice. This should slow downthe ion transport though the oxide, leading to areduced rate of oxide growth and the formation ofthinner oxide films. The fact that zinc ion signifi-cantly retards the film growth leads automaticallyto a smaller number of available adsorption sitesfor 60Co. It is also possible that zinc competes withcobalt for occupancy of surface sites for adsorptionand therefore prevents additional 60Co adsorptiononto the oxide surface. Zn may also inhibit the ionexchange reactions between inactive cobalt and60Co on the surfaces leading to lower activitybuild-up in the oxide. However, this reaction re-quires a high activation energy and therefore it isrelatively unlikely in most cases.

Most of the studies on the effect of Zn haveconcentrated on observing the physical changes inthe oxide film structure. Only little attention hasbeen paid to changes in the electronic and electro-chemical properties of the oxide films caused byzinc injections. Even though a significant amountof testing has been carried out, the comparisonbetween the results is tedious, because the experi-mental arrangements (environments, pH, materi-als) differ significantly from one test to another.

Zinc and stress corrosion cracking (SCO

The current models for SCC assume that the ano-dic and the respective cathodic reactions contri-buting to crack growth occur partly on or in theoxide films. The crack growth has been explainedto take place in the metal by capture of vacanciesin the metal close to the crack tip. If most of thevacancies in the metal lattice are generated as aresult of the dissolution of the metal through theoxide, the crack growth rate may again be control-

46

S T U K - Y T O - T R 152

led by the transport rate of species through theoxide film. The decreasing influence of Zn on iontransport may lead to a reduced rate of vacancyproduction and thus to low crack growth rates.Furthermore, the resulting thinner oxide film isless likely to break and expose the base metal tothe environment than a thicker oxide with ahigher defect concentration.

The lack of an unambiguous explanation of theeffects of zinc on the crack growth rates can beascribed to the fact that control of the test envi-ronments (Zn concentration, pH, conductivity, etc.)differ from one experiment to another. This com-plicates the interpretation of the results, becauseusually more than one factor is different betweenseparate tests. On the other hand, the impact ofZn may also be relatively small and longer expo-sure times may be needed to observe real effectson crack growth rates.

Hydrogen and noble metal 'water chemistries

Clear decrease in susceptibility of constructionmaterials for the stress corrosion cracking in boi-ling water reactors can be observed when hydro-gen is added into the feed water. Hydrogen additi-on decreases the levels of oxidising species andthus lowers the corrosion potential (ECP) of theconstruction materials. Recent plant experimentshave shown that by applying the noble metal che-mical addition- technique, oxide films on construc-tion materials behave like noble metal surfaces,catalysing the recombination reaction betweenhydrogen and oxygen. This reaction leads to corro-sion potentials low enough to avoid stress corro-sion to proceed even with reasonably low hydro-gen concentrations. An additional benefit of a lo-wer corrosion potential is higher acceptable impu-rity levels in the coolant without an effect on thecrack growth rates. This provides high toleranceto severe water chemistry transients (e.g. impuri-ty in-leakage). However, to be an economicallyviable process in operating BWRs, the applied co-ating should retain the catalytic activity at leastfor two fuel cycles under the high flow rates of theBWR coolant.

One of the concerns related to the noble metalapplications has been that the corrosion potentialof Pd/Pt containing surfaces might be higher than

the corrosion potential of uncoated materials un-der excess of oxygen/hydrogen peroxide concen-trations in the reactor water. This would lead tohigher crack growth rates in BWRs during tran-sients from hydrogen water chemistry (HWC) toNWC or during some other intermediate waterchemistry conditions. Therefore, it is possible thatneither HWC nor noble metal chemical addition-technique can provide sufficient decrease in theECP of construction materials.

Dielectric coating on metal surfaces

Another possibility to reduce susceptibility ofstainless steel components to stress corrosioncracking is to apply dielectric coatings onto themetal surface. This type of films in high tempera-ture water could be obtained by coating theconstruction material surfaces using chemical ad-ditions into the coolant or by alloying certain ele-ments, such as Zr and Y, into the base metal. Byusing dielectric coatings techniques, the reactorcomponents can be protected from intergranularstress corrosion cracking even without HWC. It isnot yet clear if these techniques can be developedinto the stage where real plant experiments willbe performed.

Concluding remarks

Although further improvements in water che-mistry in NPPs, e.g. zinc dosing and application ofnoble metal coating technology, have recently beenintroduced, it is evident that a proper understan-ding of the interaction of the coolant and the oxidefilms on material surfaces in NPPs is not yet avai-lable. The fundamental questions related to ionicand electronic conduction in the oxide films aswell as effects of different reaction rates on theobserved phenomena in the oxide films have re-mained unanswered. Therefore, it is impossible topredict material behaviour in different environ-mental conditions such as novel water chemistrieswhich involve injection of new chemicals into theprimary coolant. More experimental work and mo-delling in carefully controlled environments isneeded to find out the key processes related to thephenomena in oxide films during application ofnovel water chemistries.

47

STUK-YTO-TR152

REFERENCES

Chantoin P. Coolant technology of watercooled reactors. Volume 3: Activity transportmechanisms in water cooled reactors. IAEATechnical Reports Series No. 667. Vienna:IAEA, 1992.

Fox MJ. A review of boiling water reactorwater chemistry. NUREG/CR-5115, ANL-88-42, R5. Washington: US Nuclear RegulatoryCommission, NRC FINA2212, 1989.

Indig ME, Nelson JL. Electrochemical meas-urements and modelling predictions in boil-ing water reactors under various operatingconditions. Corrosion 1991;47:202-209.

Macdonald DD. Viability of hydrogen waterchemistry for protecting in-vessel componentsof boiling water reactors. Corrosion1992;48:194-204.

Cowan RL, Lin CC, Marble WJ, Ruiz CP.Hydrogen water chemistry in BWRs. Pro-ceedings of Fifth International Symposiumon Environmental Degradation of Materialsin Nuclear Power Systems—Water Reactors,Monterey, 1991: 50.

Macdonald DD, Song H, Makela K, YoshidaK. Corrosion potential measurements on Type304 SS and Alloy 182 in simulated BWRenvironments. Corrosion 1993;49:8-17.

Jones R. BWR Hydrogen Water ChemistryGuidelines: 1987 Revision. EPRINP-4947SR.Palo Alto, USA: EPRI, 1988.

8 Lin CC. Hydrogen water chemistry effects onBWR radiation build-up. Vol. 1: Laboratoryresults and plant data. EPRI TR-104605-V1.Palo Alto, USA: EPRI, 1994.

9 Lister DH. The transport of radioactive corro-sion products in high temperature water. II.The activation of isothermal steel surfaces.Nuclear Science and Engineering 1976;59:406-26.

10 Bergman CA, Durkosh DE, Linday WT, Roes-mer J. The role of coolant chemistry in PWRradiation field build-up. EPRI NP-4247. PaloAlto, USA: EPRI, 1985.

11 Polley MV, Garbett K, Pick WE. A survey ofthe effect of primary coolant pH on Westing-house PWR plant radiation fields. EPRI TR-104180. Palo Alto, USA: EPRI, 1994.

12 Sweeton FH, Baes CF Jr. The solubility ofmagnetite and hydrolysis of ferrous ion inaqueous solutions at elevated temperatures.Journal of Chemistry and Thermodynamics1970;2:479-500.

13 Tremaine PR, LeBlanc JC. Solubility of mag-netite and the hydrolysis and oxidation ofFe2+ in water to 300 °C. Journal of SolutionChemistry 1980;9:415-28.

14 Kunig RH, Sandier YL. The solubility of sim-ulated PWR primary circuit corrosion prod-ucts. EPRI NP-4248. Palo Alto, USA: EPRI,1986.

48

STUK-YTO-TR 1 52

15 Niedrach LW. Effect of palladium coatings onthe corrosion potential of stainless steel inhigh temperature water containing dissolvedhydrogen and oxygen. Corrosion 1991;47:162-169.

16 Wood CJ. PWR primary water chemistryguidelines: Revision 2. EPRI NP-7077. PaloAlto, USA: EPRI, 1990.

17 Rickersten D. PWR zinc injection experienceat Farley unit 2. Proceedings of EPRI Radia-tion Filed Control and Chemical Decontami-nation Seminar, Tampa, USA, November1995.

18 Lister DH, Godin MS. The effect of dissolvedzinc on the transport of corrosion products inPWRs. EPRI NP-6975-D. EPRI, 1990.

19 Rebak RB, Szklarska-Smialowska Z. Aboutthe mechanism of stress corrosion cracking ofAlloy 600 in high temperature water. Pro-ceedings of Seventh International Symposi-um on Environmental Degradation of Materi-als in Nuclear Power Systems—Water Reac-tors, Vol 1, Breckenridge Colorado, USA, Au-gust 1995: 855-866.

20 Esposito JN, Economy G, Byers WA, EspositoJB, Pement FW, Jacko RJ, Bergmann CA.The addition of zinc to primary reactor cool-ant for enhanced PWSCC resistance. Pro-ceedings of Fifth International Symposiumon Environmental Degradation of Materialsin Nuclear Power Systems—Water Reactors,Monterey, USA, August 1991: 495-501.

21 Heslop RB, Jones K. Inorganic chemistry; Aguide to advanced studies. Elsevier ScientificPubl. Comp., 1976.

24 Dunn TM, Mclure DS, Pearson RG. "Someaspects of Crystal Field Theory". New York:Harper & Row, 1965.

25 Wells AF. "Structural Inorganic Chemistry".London: Oxford University Press, 1962.

26 Harding JH. Modelling the effects of zincaddition on the uptake of cobalt by oxidefilms in PWRs. Proceedings of Water Chemis-try of Nuclear Reactor Systems, 7, Bourne-mouth, Great Britain, 1996: 308-316.

27 Durrant PJ. Durrant B. Introduction to ad-vanced inorganic chemistry. London: Long-man Group Ltd., 1970.

28 Huheey JE. Principles of structures and reac-tivity in inorganic chemistry. Harper Interna-tional SI Edition, 1983.

29 West A. Solid state chemistry and its applica-tions. Wiley, 1985.

30 Kofstad P. High temperature corrosion. Lon-don & New York: Elsevier Applied Science,1988.

31 Rickert H. Electrochemistry of Solids. Spring-er-Verlag, 1982.

32 Rosenberg HM. The solid state. Oxford Sci-ence Publ., Great Britain, 1988.

33 Van Vlack LH. Elements of materials scienceand engineering. Addison-Wesley Publ. Co.,1989.

34 Robertson J. The mechanism of high temper-ature aqueous corrosion of stainless steel.Corrosion Science 1991;32(4):443-65.

22 Douglas B, McDaniel DHJ, Alexander JJ. 35 Lister DH. The transport of radioactive corro-

23

Concepts and models of inorganic chemistry.New York: John Wiley & Sons, Inc., 1983.

Cotton FA, Wilkinson G. Basic InorganicChemistry. New York: John Wiley & Sons,Inc., 1976.

sion products in high temperature water I.Recirculation loop experiments. Nuclear Sci-ence and Engineering 1975;58:239-51.

49

STUK-YTO-TR 1 52

36 Tapping RL, Davidson RD, MacAlpine E,Lister DH. The composition and morphologyof oxide films formed on type 304 stainlesssteel in lithiated high temperature water.Corrosion Science 1986;26:563-577.

37 Smith-Magowan D. Evaluation of the applica-bility of colloid studies to Cobalt-60 deposi-tion in LWRs. EPRINP-3773. Palo Alto, USA:EPRI, 1984.

38 Lister DH, McAlpine E, Tapping RL, HockingWH. Corrosion-product release in LWRs:1984-1985 Progress Report. EPRI NP-4741.Palo Alto, USA: EPRI, 1986.

39 Berry WE, Diegle RB. Survey of corrosionproduct generation, transport, and depositionin light water nuclear reactors. EPRI NP-522.Palo Alto, USA: EPRI, 1979.

40 Asakura Y, Karasawa H, Sakagami M, Uch-ida S. Relationships between corrosion be-haviour of AISI 304 stainless steel in hightemperature pure water and its oxide filmstructures. Corrosion 1989;45:119-124.

41 Hermansson HP, Stigenberg M, Wikmark G.The KEMOX-2000 project: minimising radia-tion doses by optimising oxide conditions.Proceedings of water chemistry of nuclearreactor systems, 7, Bournemouth, Great Brit-ain, 1996:141-143.

42 Baston VF, Garbauskas MF, Ocken H. Mate-rial characterisation of corrosion films onBWR components exposed to HWC and zincinjections. Proceedings of Water Chemistry ofNuclear Reactor Systems, 7, Bournemouth,Great Britain, 1996: 558-565.

43 Lister DH, Davidson RD, McAlpine E. Themechanism and kinetics of corrosion productrelease from stainless steel in lithiated hightemperature water. Corrosion Science1987;27(2):113-140.

44 Bogaerts WF, Bettendorf C. Electrochemistryand corrosion of alloys in high-temperaturewater. EPRI NP-4705. Palo Alto, USA: EPRI,1986.

45 Sandier YL. Structure of PWR primary corro-sion products. Corrosion 1979;35:205-208.

46 Lister DH, Davidson RD. Corrosion productrelease in light water reactors. EPRI NP-6512. Palo Alto, USA: EPRI, 1989.

47 Lister DH. Activity transport and corrosionprocesses in PWRs. Proceedings of Sixth Con-ference on Water Chemistry of Nuclear Reac-tor Systems, 6, Bournemouth, Great Britain,1992:49-60.

48 Mclntyre NS, Zetaruk DG. Owen D. X-rayphotoelectron studies of the aqueous oxida-tion of Inconel 600 alloy. Journal of the Elec-trochemical Society 1979;126:750-760.

49 Schuster E, Neeb KH, Ahlanger W, Henkel-mann R, Jarnstrom RT. Analysis of primaryside oxide layers on steam generator tubesfrom PWRs and radiochemical issues on thecontamination of primary circuits. Journal ofNuclear Materials 1988;152:l-8.

50 Stellwag B. Growth mechanism of oxide filmson austenitic FeCrNi-alloys in high tempera-ture water. Proceedings of International Con-ference on Interaction of Iron Based Materi-als with Water and Steam. EPRI TR-102101.Palo Alto, USA, EPRI, 1993: 20/1.

51 Beverskog B, Puigdomenech I. Revised Pour-baix diagrams for chromium at 25-300 °C.Corrosion Science 1997;39:43-57.

52 Lin CC, Smith FR. BWR cobalt depositionstudies: Final report. EPRI NP-5808, May.Palo Alto, USA: EPRI, 1988.

50

STUK-YTO-TR 1 52

53 Kelen T, Hermansson HP. The KEMOX-2000project: minimising radiation doses by opti-mising oxide conditions. Proceedings of WaterChemistry of Nuclear Reactor Systems, 7,Bournemouth, Great Britain, 1996:156-161.

54 Ishigure K, Matsuura C, Mizuochi M, Taka-hashi M, Isotope exchange processes of cobaltions on the surfaces of erud particles, Pro-ceedings of Water Chemistry of Nuclear Reac-tor Systems, 4, Bournemouth, Great Britain,1986:145-152.

55 Marble WJ. Control of radiation-field build-up in BWRs. NP-4072, June. Palo Alto, USA:EPRI, 1985.

56 Ocken H, Wood CJ. Status report on radia-tion exposure reduction at US nuclear powerplants. Proceedings of Sixth Conference onWater Chemistry of Nuclear Reactor Sys-tems, 6, Bournemouth, Great Britain, 1992:1-8.

57 Rickersten D. PWR zinc injection experienceat Farley unit 2. Proceedings of EPRI Radia-tion Filed Control and Chemical Decontami-nation Seminar, Tampa, USA, November1995.

58 Stellwag B, Ruehle W, Staudt U. Overview ofthe VGB project activity build-up in lightwater reactors, Proceedings of Water Chemis-try of Nuclear Reactor Systems, 7, Vol. 2,Bournemouth, Great Britain, 1996: 544-551.

59 Marble WJ, Cowan RL, Wood CJ. Control ofcobalt-60 deposition in BWRs, Proceedings ofFourth Conference on Water Chemistry ofNuclear Reactor Systems, 4, Bournemouth,Great Britain, 1986:113-119.

60 Ocken H, Wood CJ. Radiation-field controlmanual—1991 Revision. EPRI TR-100265,March. Palo Alto, USA: EPRI, 1992.

61 Marble WJ, Diaz, Levin HA, Garcia SE. Eval-uation of recent experience using zinc addi-tion to reduce BWR primary system radiationbuild-up. EPRI TR-104606, December. PaloAlto, USA: EPRI, 1994.

62 Marble WJ, Cowen RL. Mitigation of radia-tion buildup in the BWR by feed water zincaddition. Proceedings of JAIF InternationalConference on Water Chemistry in NuclearPower Plants—Operational Experience andStrategy for Technical Innovation. Fukui City,Japan, April 22-25, 1991: 55-65.

63 Puyane R. Effectiveness of isotope depletedZnO to minimize radiation build-up in boilingwater nuclear reactors. Journal of MaterialsProcessing Technology 1996;56:863-872.

64 Kelen T, Lundgren K. Chemistry and materi-als impact on activity build-up in BWRs—general conclusions from BKM-crud compu-ter simulation studies. Proceedings of Confer-ence on Chemistry in Water Reactors: Oper-ating Experience & New Developments. Nice,France, 1994: 263-270.

65 Haginuma, Effect of zinc addition on cobaltion accumulation into the corroded surface oftype 304 SS in high temperature water. Pro-ceedings of Water Chemistry of Nuclear Reac-tor Systems, 7, Bournemouth, Great Britain,1996: 128-130.

66 Permer L, Osterlundh CG. Experimentalstudy of cobalt uptake by trevorite and chr-omite with and without the presence of zinc.Proceedings of Water Chemistry of NuclearReactor Systems, 7, Bournemouth, GreatBritain, 1996: 122-125.

67 Niedrach LW, Stoddard WH. Effect of zinc oncorrosion films that form on stainless steel.Corrosion;1986;42(9):546-549.

51

STUK-YTO-TR 1 52

68 Byers WA, Jacko RJ, The influence of zincaddition and PWR primary water chemistryon surface films that form on nickel basealloys and stainless steel. Sixth InternationalSymposium on Environmental Degradationof Materials in Nuclear Power Systems—Water Reactors, 1993: 837-844.

69 Baston VF, Garbauskas MF, Bozeman, J. De-contamination flange film characterisationfor a boiling water reactor under hydrogenwater chemistry, Journal of Nuclear Technol-ogy 1996;114:339-349.

70 Haginuma M, Ono S, Kumagai M, TakamoriK, Tachibana K, Ishigure K. Cobalt deposi-tion control by zinc and hydrogen injection inBWR environment. Proceedings of Confer-ence on Chemistry in Water Reactors: Oper-ating Experience & New Developments. Nice,France, 1994: 386-389.

71 Beverskog B, Puigdomenech I. Revised Pour-baix diagrams for zinc at 25-300 °C. Corro-sion Science 1997;39:107-113.

72 Hanzawa Y, Ishigure K, Matsuura C, HiroishiD. The effect of zinc addition on cobalt accu-mulation on steel surfaces and its thermody-namics. Proceedings of Water Chemistry ofNuclear Reactor Systems, 7, Vol. 2, Bourne-mouth, UK, 1996: 301-308.

73 Allsop HA. The benefits of zinc addition toprimary side coolant. Proceedings of Confer-ence on Chemistry in Water Reactors: Oper-ating Experience & New Developments. Nice,France, 1994: 321-327.

74 Walker ZH, Allsop HA, Godin MSL, SawickiJA, Turner CW, Klimas SJ. Effects of zincadditions under Candu pressurised heavy wa-ter reactor conditions. Proceedings of WaterChemistry of Nuclear Reactor Systems, 7,Vol. 2, Bournemouth, UK, 1996: 552-557.

75 Bennet PJ, Gunnerud P, Loner H, Pettersen,Harper A. The effect of zinc addition on cobaltdeposition in PWRs, Proceedings of WaterChemistry of Nuclear Reactor Systems, 7,Vol. 2, Bournemouth, UK, 1996: 293-300.

76 Pathania R, Yagnik S, Gold RE, Dove M,Kolstad E. Evaluation of zinc addition toPWR primary coolant. Proceedings of Sev-enth International Symposium on Environ-mental Degradation of Materials in NuclearPower Systems—Water Reactors, Vol. 1. Bre-ckenridge, Colorado, USA, August 1995: 163-173.

77 Bergmann, RE, Gold RE, Sejvar, JD, PerockJD, Dove M. Overview of zinc addition in theFarley 2 reactor. Proceedings of Water Chem-istry of Nuclear Reactor Systems 7, Vol. 2,Bournemouth, UK, 1996: 287-292.

78 Nieder D, Wolter D. Einfuhrung der Zink-Dosierung in KWB-B. Proceedings of VGBMeeting on Water Chemistry in NPPs. Essen,Germany, 1997: 1-7.

79 Hsueh R, Kohse G, Harling OK. In-reactorstudy of zinc injection to reduce radioactivecorrosion product transport in PWRs. Am.Nuclear Society Transactions: 151-152.

80 Riess R, Stellwag B. Effects of zinc on thecontamination and structure of oxide layers.Proceedings of Water Chemistry of NuclearReactor Systems, 7, Vol. 2, Bournemouth,Great Britain, 1996: 573-582.

81 Korb J, Stellwag B. Thermodynamics of zincchemistry in PWRs-effects and alternativesto zinc. Proceedings of Water Chemistry ofNuclear Reactor Systems, 7, Bournemouth,Great Britain, 1996: 147-49.

82 Bennet P. The longer term effects of zincaddition on cobalt deposition in PWRs. HWR-477. Halden, Norway, 1996.

52

STUK-YTO-TR 1 52

83 Bennet P. The effects of zinc on cobalt deposi- 91tion in PWRs: Summary Report. HWR-450.Halden, Norway, 1996.

84 Bennet P. Incorporation of zinc into oxidefilms in PWRs. HWR-489. Halden, Norway,1996.

9285 Makela K. Activity incorporation into zinc

doped PWR oxides. HWR-557. To be pub-lished at Enlarged Halden Programme GroupMeeting at Lillehammer, March 1997.

86 Gold RE, Byers WA, Jacko RJ. Zinc-oxidecorrosion film interactions in PWR primarycoolant. SFEN, Fontevraud, France: 300-309. 93

87 Osato T, Hemmi Y. Corrosion and Co uptakebehaviour on structural material in BWRprimary coolants at Zn and Ni addition. Pro-ceedings of Water Chemistry of Nuclear Reac-tor Systems, 7, Bournemouth, Great Britain,1996:125-127. 94

88 Ullberg M. Iron to nickel ratio of BWR fuelcrud - effects and interpretation. Proceedingsof Water Chemistry of Nuclear Reactor Sys-tems, 7, Vol. 2, Bournemouth, Great Britain,1996: 496-501.

Cheng BC, Turnage KG, Hudson M, BrownJM, Armstrong EA. Fuel performance andwater chemistry variables in LWRs. Proceed-ings of 1997 International Topical Meeting onLight Water Reactor Fuel Performance, Port-land, Canada: 379-388.

Levin HA, Gercia SE, BWR fuel experiencewith zinc injection, Proceedings of SeventhInternational Symposium on EnvironmentalDegradation of Materials in Nuclear PowerSystems—Water Reactors. Vol. 2. Brecken-ridge, Colorado, USA, August, 1995: 1217-1230.

Cheng BC, Turnage KG, Hudson M, BrownJM, Armstrong EA. Fuel performance andwater chemistry variables in LWRs. Proceed-ings of 1997 International Topical Meeting onLight Water Reactor Fuel Performance, Port-land, Canada: 379-388.

Aomi M, Kogai T, Shimada S, Ichikawa N, IbeE, Ishii Y, Cheng B, Lutz D. Evaluation ofZircaloy corrosion under various water chem-istries in a BWR simulation loop, Proceedingsof Water Chemistry of Nuclear Reactor Sys-tems, 7, Bournemouth, Great Britain, 1996:284-286.

89 Aizawa M, Ohsumi K, Asakura Y, Morikava 95Y, Hirahara Y, Sakai T, Harguchi K, Operat-ing experience of Japanise improvement andstandardisation BWRs and behaviour of radi-oactivity in reactor water, Proceedings of 96Sixth Conference on Water Chemistry of Nu-clear Reactor Systems, 6, Bournemouth,Great Britain, 1992: 39-44.

90 Uetake N, Hosakawa H, Uchida S, Ohsumi K, 97Tone T, Suzuki N. Development of waterchemistry optimisation technologies in recentJapanese BWRs. Proceedings of Water Chem-istry of Nuclear Reactor Systems, 7, Bourne-mouth, Great Britain, 1996: 168-175.

Baston VF, Indig ME, Skarpelos JM, BWRchromium chemistry. EPRI TR-100792, Octo-ber. EPRI: Palo Alto, USA, 1992.

Ljungberg L, Hallden E. BWR water chemis-try impurity studies: literature review of ef-fects on stress corrosion cracking: InterimReport EPRI NP-3663. EPRI: Palo Alto, USA.

Macdonald DD, Urquidi-Macdonald M. A cou-pled environment model for stress corrosioncracking in sensitised type 304 stainless steelin LWR environments. Corrosion Science1991;32:51-81.

53

S T U K - Y T O - T R 152

98 Galvele JR. Electrochemical aspects of stresscorrosion cracking. In: White, RE et al (Eds.)Modern Aspects of Electrochemistry, No. 27.New York: Plenum Press.

99 Andresen PL. SCC growth rate behaviour inBWR water of increased purity. 8th Interna-tional Symposium on Environmental Degra-dation of Materials in Nuclear Power Sys-tems—Water Reactors, Amelia Island, USA,August 10-14, 1997: 603-614.

100 Turnbull A. Modelling of environment assist-ed cracking. Corrosion Science 1993;34:921-960.

101 Sarver JM, Pathania RS, Stuckey K, FyftchS, Gelpi A, Foucault M, Hunt ES. Stresscorrosion cracking of welded Alloy 600 pene-tration mockups. Proceedings of Seventh In-ternational Symposium on EnvironmentalDegradation of Materials in Nuclear PowerSystems—Water Reactors, Vol. 1, Brecken-ridge, Colorado, USA, August 1995: 13-24.

102 Rebak RB, Szklarska-Smialowska Z. Effect ofpartial pressure of hydrogen on IGSCC ofalloy 600 in PWR primary water. Corrosion1991;47:754-757.

103 Galvele JR. Electrochemical aspects of stresscorrosion cracking. In: White, RE et al. (Eds.)Modern Aspects of Electrochemistry, No 27.New York: Plenum Press, 1995: 233-358.

104 Lagerstrom, J, Ehrnsten, U, Saario, T, Laitin-en, T, Hanninen, H. Model for environmental-ly assisted cracking of Alloy 600 in PWRprimary water. Proceedings of the EighthInternational Symposium on EnvironmentalDegradation of Materials in Nuclear PowerSystems—Water Reactors, Florida, August10-14,1997: 349-356.

105 Airey GP, Allan SJ, Angell MG. The effect ofzinc on the stress corrosion cracking of In-conel alloys. Proceedings of Water Chemistryof Nuclear Reactor Systems, 7, Bournemouth,Great Britain, 1996: 478-482.

106 Congleton J, Yang W. The effect of appliedpotential on the stress corrosion cracking ofsensitized type 316 stainless steel in hightemperature water. Corrosion Science 1995;37:429-444.

107 Andresen PL, Diaz TP. Effects of zinc addi-tions on the crack growth rate of sensitisedstainless steel and alloys 600 and 182 in 288°C water. Proceedings of Sixth Conference onWater Chemistry of Nuclear Reactor Sys-tems, 6, Vol. 1, Bournemouth, Great Britain,1992: 169-175.

108 Andresen PL, Angeliu TM. Effects of zincaddition on the stress corrosion crack growhtrate of sensitised stainless steel, alloy 600 &alloy 182 weld metal in 288 °C water. Corro-sion 95, USA. Paper 95409.

109 Hettiarachchi S, Wozadlo GP, Diaz TP. Influ-ence of zinc additions on the intergranularstress corrosion crack initiation and growthof sensitised stainless steel in high tempera-ture water. Corrosion 95, USA, Paper 410.

110 Hettiarachchi S, Wozadlo GP, Andresen PL,Diaz TP, Cowan RL. The concept of noblemetal chemical addition technology for IG-SCC mitigation of structural materials. Pro-ceedings of Seventh International Symposi-um on Environmental Degradation of Materi-als in Nuclear Power Systems—Water Reac-tors, Vol. 1, Breckenridge Colorado, USA,August, 1995: 735-746.

111 Angeliu TM, Andresen PL. The effect of zincadditions on the oxide rupture strain andrepassivation kinetics of Fe-based alloys in288 °C water. Corrosion 1996;52:28-35.

112 Angeliu TM, Andresen PL, Pollick ML. Re-passivation and crack propagation of Alloy600 in 288 °C water. Corrosion 1997;53:114-119.

54

STUK-YTO-TR 1 52

113 Angeliu TM, Andresen PL. The effect of zincadditions on the oxide rupture strain andrepassivation kinetics of Fe-based alloys in288 °C water. Corrosion 95, Paper 95411,USA.

114 Andresen PL. Effect of noble metal coating &alloying on the stress corrosion crack growthrate of stainless steel in 288 °C water. Pro-ceedings of Sixth International Symposiumon Environmental Degradation of Materialsin Nuclear Power Systems—Water Reactors,1993:245-253.

115 Cowan RL. The mitigation of IGSCC of BWRinternals with hydrogen water chemistry.Proceedings of Water Chemistry of NuclearReactor Systems, 7, Bournemouth, GreatBritain, 1996:196-206.

116 Kim YJ, Niedrach LW, Andresen PL. Corro-sion potential behaviour of noble metal modi-fied alloys in high temperature water. Corro-sion 95, Paper 95099, USA.

117 Andresen PL. Application of noble metal tech-nology for mitigation of stress corrosioncracking in BWRs. Proceedings of SeventhInternational Symposium on EnvironmentalDegradation of Materials in Nuclear PowerSystems—Water Reactors, Vol. 1, Brecken-ridge Colorado, USA, August 1995: 563-577.

118 Andresen PL, Angeliu T. The effect of in-situnoble metal chemical addition on crackgrowth rate behaviour of structural materialsin 288 °C water. Corrosion 96, Paper 96084,USA.

119 Kim Y-J, Andresen PL, Gray DM, Lau YC,Offer HP. Corrosion potential behaviour inhigh temperature water of noble metal dopedalloys coatings deposited by underwater ther-mal spraying. Proceedings of Seventh Inter-national Symposium on Environmental Deg-radation of Materials in Nuclear Power Sys-tems—Water Reactors, Vol. 1, BreckenridgeColorado, USA, August 1995: 723-734.

120 Kim YJ, Andresen PL, Gray DM, Lau YC,Offer HP. Corrosion potential behaviour inhigh temperature water of noble metal dopedalloys coatings deposited by underwater ther-mal spraying. Corrosion 1996;52:440-446.

121 Kim YJ, Niedrach LW, Andresen PL. Corro-sion potential behaviour of noble metal-modi-fied alloys in high temperature water. Corro-sion 1996;52:738-743.

122 Andresen PL. Mitigation of stress corrosioncracking by underwater thermal spray coat-ing of noble metals. Corrosion 95, USA, Paper95412.

123 Hettiarachchi S, Wozadlo GP, Diaz TP. Long-term performance of noble metal doped sur-faces in simulated BWR environments. Cor-rosion 96, USA, Paper 96099.

124 Kim YJ, Andresen PL, Angeliu TM. In-situnoble metal deposition and its durability onthe catalytic response of stainless steel sur-faces in high temperature water. Corrosion96, USA, Paper 96109.

125 Kim YJ, Andresen PL. Effect of zinc andcopper additions on catalytic response of no-ble metal alloyed 304 SS in high temperaturewater. Corrosion 97, USA, Paper 97112.

126 Yeh TK, Macdonald DD. Predictions of en-hancing hydrogen water chemistry for boilingwater reactors by general catalysis and gen-eral inhibition. Corrosion 96, USA, Paper96124.

127 Kim YJ, Andresen PL. Application of insulat-ed protective coatings for reduction of corro-sion potential in high temperature water.Corrosion 96, USA, Paper 96105.

55