You are on page 1of 74

Hydrazine substitutes for use as

oxygen scavengers in the


secondary circuits of pressurized
water reactors
Frej Lindfors

School of Chemical Engineering

Thesis submitted for examination for the degree of Master of


Science in Technology.
Espoo 8.1.2020

Supervisor

D.Sc. (Tech.) Jari Aromaa

Advisor

M.Sc. (Tech.) Konsta Sipilä


Copyright ⃝
c 2020 Frej Lindfors
Aalto University, P.O. BOX 11000, 00076 AALTO
www.aalto.fi
Abstract of the master’s thesis

Author Frej Lindfors


Title Hydrazine substitutes for use as oxygen scavengers in the secondary circuits of
pressurized water reactors
Degree programme Chemical, Biochemical and Materials Engineering
Major Functional Materials Code of major CHEM3025
Supervisor D.Sc. (Tech.) Jari Aromaa
Advisor M.Sc. (Tech.) Konsta Sipilä
Date 8.1.2020 Number of pages 59+8 Language English
Abstract
Proper chemical treatment of the secondary water is vital for the safety and perfor-
mance of pressurized water reactors (PWR) and has conventionally incorporated
hydrazine as an oxygen scavenger. Oxygen scavengers consume dissolved oxygen and
thereby mitigate corrosion and other degradation mechanisms of structural materials
found in the PWR secondary circuits. However, since hydrazine has been classified as
possibly carcinogenic, the use of hydrazine within the EU may become restricted in
the near future. This has stressed the need for an appropriate substitute to hydrazine.
The aim of this thesis is to evaluate the feasibility of the two hydrazine substitutes
erythorbic acid and diethylhydroxylamine for use as oxygen scavengers in PWR
secondary circuits. For this purpose, the hydrazine substitutes were evaluated
with respect to their effectiveness to consume oxygen and their ability to protect
carbon steel from corrosion in an experimental setting resembling the environment
of Loviisa WWER-440 secondary circuits. The experiments relied on electrochemical
measurement techniques including chronopotentiometry, linear polarization resistance
and electrochemical impedance spectroscopy.
The results show that erythorbic acid consumed oxygen as quickly and effec-
tively, and in some measurements even more efficiently than hydrazine, whereas
diethylhydroxylamine reduced the oxygen content at rates of approximately one
third of hydrazine. In addition, the results indicated that all three oxygen scavengers
temporarily increase the corrosion rate of carbon steel to comparable extents when
administered at high concentration ratios (C/CO2 values greater than 8).
The results suggest that erythorbic acid and diethylhydroxylamine can offer safe
approaches for chemical treatment of PWR secondary waters, provided that any
secondary effects of the substances can be accounted for. These secondary effects
include the generation of carbon dioxide and various decomposition products that
may acidify the secondary waters and increase the electrolytic conductivity.
Keywords Hydrazine, erythorbic acid, diethylhydroxylamine, oxygen scavengers,
corrosion inhibitors, PWR, secondary circuit, WWER
Aalto-universitetet, PB 11000, 00076 AALTO
www.aalto.fi
Sammandrag av diplomarbetet

Författare Frej Lindfors


Titel Ersättande ämnen till hydrazin för använding som avluftningsmedel i
tryckvattenreaktorers sekundärkrets
Utbildningsprogram Kemi-, bio- och materialteknik
Huvudämne Funktionella material Huvudämnets kod CHEM3025
Övervakare TkD Jari Aromaa
Handledare DI Konsta Sipilä
Datum 8.1.2020 Sidantal 59+8 Språk Engelska
Sammandrag
Den kemiska behandlingen av sekundärvattnet i tryckvattenreaktorer är väsentlig för
god reaktorsäkerhet och termodynamisk prestanda och har konventionellt inbegripit
användandet av hydrazin som ett avluftningsmedel. Avluftningsmedlet förbrukar
i vattnet upplöst syre och lindrar därmed korrosion och andra former av nedbryt-
ningsprocesser av byggnadsmaterial i tryckvattenreaktorns sekundärkrets. Dock kan
användandet av hydrazin som avluftningsmedel inom en snar framtid begränsas inom
EU eftersom hydrazin kan vara cancerframkallande, vilket har lett till ett ökat behov
av ersättande avluftningsmedel för hydrazin.
Avhandlingens mål är att utvärdera användandet av substanserna isoaskorbinsyra
och dietylhydroxylamin som avluftningsmedel i tryckvattenreaktorns sekundärkrets.
I detta syfte utvärderades substanserna med hänsyn till deras effektivitet att förbuka
syre och förmåga att skydda kolstål från korrosion i experiment vars förhållanden
efterliknar sekundärkretsen i Lovisa kärnkraftverks tryckvattenreaktor av typen
VVER-440. Experimenten bestod av elektrokemiska mätningar i form av kronopo-
tentiometri, linjär polarisationsresistans och elektrokemisk impedansspektroskopi.
Resultaten visar att isoaskorbinsyra förbrukar syre lika snabbt och effektivt, och i
vissa mätningar även mer effektivt än hydrazin medan dietylhydroxylamin förbrukar
syre med en hastighet som motsvarar en tredjedel av hydrazins. Därtill visar resultaten
att samtliga avluftningsmedel tillfälligt ökar korrosionshastigheten för kolstål i jäm-
förbar utsträckning när de tillfördes i stora överskott (när koncentrationsförhållandet
C/CO2 var 8 eller större).
Därmed tyder resultaten på att isoaskorbinsyra och dietylhydroxylamin kan er-
sätta hydrazin som avluftningsmedel i tryckvattenreaktorers sekundärkrets, under
förutsättningen att hänsyn fästs vid eventuella indirekta effekter av de nya sub-
stanserna. Dessa indirekta effekter innefattar bildandet av koldioxid och diverse
nedbrytningsprodukter som kan sänka sekundärvattnets pH och öka dess elektroly-
tiska konduktivitet.
Nyckelord hydrazin, isoaskorbinsyra, dietylhydroxylamin, antioxidant,
korrosionsinhibitor, tryckvattenreaktor, sekundärkrets, VVER-reaktor
iii

Acknowledgements
This thesis was carried out in the facilities of VTT Technical Research Centre of
Finland between August and December 2019. The thesis has been part of the
VTT project “Extended lifetime of structural materials through improved water
chemistry” ELMO, which has received funding as part of the Finnish national research
programme on the safety of nuclear power plants SAFIR2022. The support from the
VTT organization and the SAFIR2022 programme including its financiers has been
acknowledged.
I would like to thank Konsta Sipilä for his active involvement as an advisor
and whom I could bounce off ideas for the thesis. In addition, I want to thank my
supervisor Jari Aromaa for sharing his expertise in electrochemistry, for proof-reading
the thesis and inspecting the balancing of the chemical equations.
Gratitude is due to Tiina Ikäläinen and Timo Saario, who have together with
Konsta tested the experimental methodology, introduced me to the experimental
setup and provided entertainment during lab sessions and coffee breaks. A special
expression of gratitude is directed to Timo for his support on the intepretation of
results and for introducing me to Tibetan throat singing.
The support of Prof. Martin Bojinov on the experimental sensor calculus was
sincerely endorsed. I acknowledge the help of Johanna Lukin for preparing the
characterization samples and carrying out the SEM and EDS characterization. Thanks
are also due to Aki Toivonen and Pasi Väisänen for promptly rebooting the automation
system whenever it decided to malfunction.
Finally, I want to thank my family, friends and significant other for putting up
with me and supporting me throughout the thesis.

Otaniemi, 8.1.2020

Frej Lindfors
iv

Contents
Abstract i

Abstract (in Swedish) ii

Acknowledgements iii

Contents iv

1 Introduction 1

2 The corrosion environment in NPP secondary circuits 3


2.1 Pressurized water reactors . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Secondary water chemistry . . . . . . . . . . . . . . . . . . . . . . . . 4
2.3 WWER steam generator . . . . . . . . . . . . . . . . . . . . . . . . . 5

3 Corrosion fundamentals 7
3.1 Metallic corrosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.2 Thermodynamics of corrosion . . . . . . . . . . . . . . . . . . . . . . 10
3.3 Corrosion rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

4 Secondary circuit corrosion and its mitigation 15


4.1 Flow assisted corrosion of carbon steel . . . . . . . . . . . . . . . . . 15
4.2 Degradation of steam generator tubes . . . . . . . . . . . . . . . . . . 17
4.3 Hydrazine as oxygen scavenger and corrosion inhibitor . . . . . . . . 19
4.4 Alternatives to hydrazine in oxygen scavenging . . . . . . . . . . . . . 21

5 Electrochemical techniques 25
5.1 Electrochemical cells . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.2 Electrochemical oxygen sensors . . . . . . . . . . . . . . . . . . . . . 27
5.3 Linear polarization resistance . . . . . . . . . . . . . . . . . . . . . . 29
5.4 Electrochemical impedance spectroscopy . . . . . . . . . . . . . . . . 31

6 Experimental 36
6.1 Equipment and materials . . . . . . . . . . . . . . . . . . . . . . . . . 36
6.2 Oxygen reduction kinetics . . . . . . . . . . . . . . . . . . . . . . . . 37
6.3 Electrochemical measurements of carbon steel . . . . . . . . . . . . . 38

7 Results and discussion 39


7.1 Oxygen reduction kinetics . . . . . . . . . . . . . . . . . . . . . . . . 39
7.2 Electrochemical measurements of carbon steel . . . . . . . . . . . . . 44

8 Conclusions 50

References 51
v

Appendices 60
A Experimental sensor calculus . . . . . . . . . . . . . . . . . . . . . . . 60
B Pump rate correction . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
C Injection tests with blank solutions . . . . . . . . . . . . . . . . . . . 63
D Characterization of the iron oxides . . . . . . . . . . . . . . . . . . . 65
E The effect of OSs on the electrolytic conductivity . . . . . . . . . . . 67
1

1 Introduction
A significant part of Finland’s environmental strategy for reducing carbon emissions
relies on nuclear energy to meet a larger share of the power demand [1, p. 71].
Currently, four nuclear reactors operate in Finland: two boiling water reactors
(BWR) in Olkiluoto and two WWER pressurized water reactors in Loviisa, which
conjointly produce roughly 25% of the annual power supply [2]. It is estimated that
this share will increase to 40% by 2030 [1, p. 78] as two new nuclear reactors are
commissioned. Unit 3 of the Olkiluoto plant (OL3) and Unit 1 of the Hanhikivi power
plant (HK1) are scheduled to begin energy production in 2020 [3] and in 2028 [4],
respectively. As the nuclear power capacity increases, the use of fossil fuels will be
phased out, as Finland aims to become coal-independent by 2030 [1, p. 34].
To ensure that nuclear power can fulfill the role of coal and other fossil fuels, it
is important that the future operation and safety of the existing nuclear reactors
at Olkiluoto and Loviisa is assured. These reactors were commissioned between
1977 and 1982 but were originally planned to operate only until the end of 2007
(Loviisa) [5, p. 2] and 2018 (Olkiluoto) [6, p. 3]. Nevertheless, the reactors have been
granted extended operating licenses until 2030 for Loviisa [5, p. 10] and 2038 for
Olkiluoto [6, p. 16], provided that the plants meet the safety guidelines set by the
Radiation and Nuclear Safety Authority (STUK). STUK monitors the operation of
the power plants through periodic safety reviews, in which the authority extensively
assesses the nuclear safety of the nuclear power plants (NPP).
In the latest safety review [7], STUK found the Loviisa power plant to be compliant
with nuclear safety regulations. However, the review raised concern about the use
of the chemical substance hydrazine for controlling the water chemistry in Loviisa
secondary coolant circuits [7, p. 40]. This is a common practice for pressurized water
reactors (PWR), where hydrazine is introduced into the secondary coolant circuits to
maintain a low content of dissolved oxygen, thereby mitigating corrosion and other
material degradation mechanisms [8]. However, the use of hydrazine may become
restricted in the near future under European legislation [9], as hydrazine may be
carcinogenic [10]. Moreover, hydrazine is known to be hazardous to the environment
and its use should thus be avoided [11]. Therefore, it is important for both safety
and environmental reasons that a safe substitute for hydrazine be secured in order to
manage the secondary circuit water chemistry of nuclear pressurized water reactors,
not only in the two Loviisa units but also in OL3 and HK1.
Although several promising hydrazine substitutes have been proposed, finding
a qualified substitute has proved to be challenging [12]. One of the challenges
lies in hydrazine being a very versatile chemical that carries out several corrosion
preventative functions. Hydrazine simultaneously removes dissolved oxygen (i.e., it
acts as an oxygen scavenger), helps to form protective oxides on metal surfaces (i.e.,
it acts as a passivator), and its decomposition products maintain appropriate pH
levels [13]. So far, few reported alternatives to hydrazine have been able to fulfill all
three functions, and none has done so adequately [12].
One suggested approach for fulfilling these three functions is to combine multiple
substitutes that collectively provide the properties of hydrazine [14]. However, to
2

determine the feasibility of such an approach, it would have to be comprehensively


validated through experimental testing. In particular, the substitutes that are to
fulfil the role as oxygen scavengers must be tested with respect to their overall impact
on the coolant chemistry, their compatibility with coexisting materials, and their
effectiveness in removing dissolved oxygen. These tests should ideally be carried out
under conditions resembling those present in the coolant circuits of an operating
nuclear reactor.
The aim of this thesis is to evaluate the feasibility of two hydrazine substitutes
for use as oxygen scavengers in pressurized water reactors. In order to compare
the oxygen reduction kinetics of the two hydrazine substitutes, a high temperature
water loop experiment is devised resembling the environment of Loviisa secondary
circuits. Under these conditions and oxygen contents adjusted to approximately
100 ppb and 10 ppm, hydrazine and the two hydrazine substitutes are administered
at varying concentrations into the water loop. The effectiveness of the oxygen
scavengers in reducing oxygen were deduced using two methods for measuring oxygen
concentration. In addition, the electrochemical properties of a carbon steel (22K)
commonly found in Loviisa secondary circuits are monitored during administration
of scavengers using two electrochemical techniques: linear polarization resistance
(LPR) and electrochemical impedance spectroscopy (EIS).
Accordingly, this thesis is limited to benchmarking the oxygen reduction kinetics
of the two studied oxygen scavengers. In addition, the thesis is limited to depicting
the electrochemical properties of only one of the many structural materials in the
coolant circuits of nuclear pressurized water reactors that may interact with the
oxygen scavengers. Nevertheless, the final outcome of this thesis provides estimates
on the effectiveness of the hydrazine substitutes as oxygen scavengers at elevated
temperatures. These results can be seen as an important step in the direction of
replacing hydrazine in pressurized water reactors.
The thesis is divided into eight chapters. Chapter 2 introduces the reader to the
nuclear reactors as an environment where corrosion can take place. Chapter 3 reviews
the literature on fundamental concepts of corrosion, and Chapter 4 applies these to
the secondary circuit of nuclear PWR. Chapter 4 reviews the use of hydrazine as an
oxygen scavenger and plausible substitutes to hydrazine for use as oxygen scavengers.
Chapter 5 assesses the applicability of the measurement techniques utilized in the
thesis experiments. Chapter 6 describes the experiments, while Chapter 7 summarizes
and discusses the results obtained from the experiments. Finally, Chapter 8 concludes
the thesis by presenting the main findings and suggesting future work.
3

2 The corrosion environment in NPP secondary


circuits
This chapter outlines the operating principle of a nuclear power plant (NPP) together
with the power plant structure and the key components of PWRs. By doing so, the
chapter introduces the reader to concepts, vocabulary and abbreviations that will
be used throughout the thesis. More importantly, the chapter defines the corrosion
environment of the secondary circuit in terms of water chemistry, present materials
and physical conditions. Furthermore, the chapter highlights the structure of the
steam generator (SG) since it is a component considered particularly susceptible to
corrosion. The actual corrosion mechanisms are detailed in Chapter 4.

2.1 Pressurized water reactors


A commercial NPP is a thermal power station that converts energy from controlled
nuclear fission reactions into electricity. In general, the energy from these controlled
nuclear reactions are transferred as heat via a coolant (commonly water) to produce
steam that ultimately drives a turbine generator [15, Ch. 3]. The general layouts
of NPPs are quite similar but vary with respect to their components and reactor
concepts.
Nuclear reactors can be classified based on reactor medium into gas-cooled and
water-cooled reactor types [16, p. 634]. Water-cooled reactors constitute over 95%
of the conventional reactors used for power production [17, p. 74]. The water-
cooled reactors can further be sub-classified into two types: light water and heavy
water reactors, which use ’normal’ and deuterium oxide 2H2 O waters as cooling
mediums, respectively. For every power producing reactor using heavy water, there
are approximately eight reactors using light water [17, p. 74].
Nuclear reactors cooled by light water are classified into two main concepts:
boiling water reactors (BWR) and pressurized water reactors (PWR) [18, p. 3].
One of the main distinction between the two is that boiling occurs at the core in
a BWR while in a PWR, water is turned into steam in a separate heat exchanger
or so-called steam generator (SG) [18, p. 7]. As of December 2016, the BWR and
PWR concepts represented 20.8% and 75.4% of the 448 operating nuclear reactors,
respectively [17, p. 74]. The Olkiluoto reactors 1 and 2 are BWRs while the Loviisa
reactors are PWRs. The Loviisa reactors are PWRs of former Soviet WWER-440
designs (WWER stands for ’water water energetic reactor’ [19, p. 23] or ’water-
cooled and water-moderated energy reactor’ while 440 refers to the original 440 MW
electrical design output). Some writers prefer the use of VVER before WWER,
whereas this thesis will henceforth use explicitly WWER.
The operating principle of the PWR relies on three coolant circuits as shown in
Figure 1.1. The coolant in the primary circuit, which is kept under approximately
15 MPa pressure for the water to remain as liquid at high temperatures, absorbs
heat from the reactor core [20]. The primary coolant passes through the steam
generator and transfers heat to the secondary circuit which operates at a lower,
roughly 7 MPa, pressure [18, p. 5]. The transferred heat causes the secondary circuit
4

water to boil into steam, which is subsequently dried and fed to the steam turbine
generator where electricity is produced. As the secondary circuit steam passes the
turbines, the tertiary circuit condenses the steam back into water, commonly referred
to as “feedwater”, which is pumped back to the steam generator and thereby closes
the secondary circuit cycle [18, p. 5]. Since hydrazine is associated with the water
chemistry of the secondary circuit, the discussion will henceforth be limited to the
secondary circuit.

Figure 1: Schematic of a PWR nuclear power plant with highlighted cooling circuits.
Image adapted from [21]

2.2 Secondary water chemistry


The secondary circuits of the Loviisa WWER-440 reactors are quite extensive and
incorporate a series of structural materials. Out of these, the most abundant materials
included carbon steel (in the steam and feedwater pipes), stainless steel (in the
preheater and steam generator tubes) and high-alloyed stainless steel (in the condenser
tubes) [22]. At the time of writing this, the last components made of copper should
have been removed from Loviisa secondary circuits [7, p. 40].
The conditions that these materials face vary considerably between different
sections of the circuit considering that both water and steam phases are present. For
instance, the temperature and pressure of the water phase range from 24 to 225 ◦ C
and 24.1 to 60 bar between the condenser exit and the entry to the steam generator,
respectively [23]. Under these conditions, the fast-flowing water and steam place
stress on the structural materials which may result in their degradation in the form
of wear and corrosion.
In order to reduce the risk of corrosion, the water chemistry of the secondary circuit
is constantly monitored and adjusted to values shown in Table 1. The feedwater
is kept alkaline at a pH value of 9.6 using ammonia, and hydrazine is injected to
5

maintain low levels of dissolved oxygen. The feedwater also contains trace amounts
of impurities such as chloride Cl – and sodium ions Na+ . Furthermore, the water
can be assumed to contain dissolved ions from metals such as Fe, Cr, Ni and Cu in
concentrations on the order of ppb originating from corrosion of various parts of the
secondary circuit.

Table 1: Secondary circuit feedwater chemistry in terms of species concentration


(weight-pbb or μg/l) and physical quantities [22, 24, 25].
Species or quantity Feedwater Unit
NH3 (ammonia) 500 ppb
N2 H4 (hydrazine) 2-20 ppb
Cl – <50 ppb
Na+ <5 ppb
O2 (dissolved oxygen) <1 ppb
Acidity, pH25◦ C C 9.6
Conductivity 10 μS/cm

Although the water chemistry is controlled properly, the PWR secondary circuit
contains a collection of structural materials that may be prone to corrode [26]. These
include carbon steels, low-alloy steels, stainless steels and nickel based alloys [26].
In case of Loviisa and other WWER-440 reactors, the materials that have been the
most prominent to degrade include [27, p. 10] carbon steels in the feedwater and
steam pipes and stainless steels in the steam generators. The corrosion mechanisms
of either material are outlined in sections 4.1 and 4.2. In order to understand the
corrosion process of stainless steel in the steam generator, the following section will
describe the characteristics of the WWER steam generator.

2.3 WWER steam generator


The NPP component which is considered the most sensitive to corrosion is the
steam generator (SG) [28, p. 126]. The SG component that is the most susceptible
to fail in WWERs and other types of PWRs are the SG heat exchange tubes,
which have historically failed mainly due to stress corrosion cracking (SCC) [16,
p. 648]. Moreover, the replacement of the WWER-440 SG can be very expensive
and challenging, which is why it is considered the most critical component for the
lifetime extensions of WWER NPPs [16, p. 648]. The corrosion-sensitive nature
of the SG can be attributed to its combination of materials and complex internal
structure paired with the challenging conditions present inside of it. These aspects
are necessary to depict in order to understand the associated corrosion processes.
Characteristic for WWER-440 reactors are six primary circuits with one horizontal
PGV-440 steam generator each [18, p. 8]. The PGV-440 steam generators, which are
conceptually very similar to the PGV-1000 steam generator shown in Figure 2, are
shell-and tube heat exchangers in which the primary coolant flows through the tubes
and the secondary coolant resides outside the tubes. The inlet temperatures of the
primary and secondary waters are 297 ◦ C [19, p. 108] and 228 ◦ C [24], respectively.
6

Figure 2: The WWER steam generator PGV-1000 with internal components. Image
modified from [16, p. 644]

The 16 mm diameter heat exchange tubes, 5536 in total [19, p. 108], form u-
bends between the primary coolant entry and exit (hot and cold collector, shown in
Figure 2). The tubes and collectors are made from titanium stabilized austenitic
stainless steel 08Ch18N10T [19, p. 108] with 18% Cr and 10% Ni content [16, p. 643].
The tubes are arranged in line and evenly spaced by 30 mm carbon steel (grade 22K)
support plates in vertical direction and carbon steel wave plates (presumably also
22K) in the horizontal direction [19, p. 107].
The 3.2 m diameter and 11.6 m long SG body consists of 22K carbon steel and
holds the secondary coolant [19, p. 108]. The secondary coolant evaporates into
steam at a rate on the order of 100 l/s [19, p. 108] and make-up water is distributed
at the center of the steam generator through the feedwater pipe. The feedwater
pipes were originally carbon steel 22K [19, p. 109] but were later replaced in some
Loviisa steam generators by stainless steel designs [29]. Under operating conditions,
the water level is kept above the tube bundle at a constant height of approximately
2.12 m [30, p. 109]. Since water is constantly evaporating, the concentration of
dissolved impurities increases in the SG that precipitate as solids which sediment
on the bottom of the SG. Therefore, the SG is equipped with ’blowdown valves’
through which the sedimentary sludge is flushed out at periodic intervals [16, p. 643].
This chapter has introduced the corrosion environment of the PWR secondary
circuit and the steam generator. The following chapter will outline the corrosion
processes taking place in the secondary circuit from a theoretical perspective while
Chapter 4 depicts commonly observed corrosion forms for carbon steels and stainless
steels.
7

3 Corrosion fundamentals
In order to understand the corrosion mechanisms that the use of hydrazine aims
to mitigate, it is first necessary to understand the main fundamental concepts of
metallic corrosion. This chapter reviews corrosion of metals from the literature by
emphasizing main concepts of thermodynamics and kinetics with respect to the
metal-solution interface.

3.1 Metallic corrosion


Metallic corrosion has been described [31, p. 1] as a natural process by which a metal
destructively reacts with its environment so that the metal returns into its oxidized
form and the properties of the metal deteriorate. In general, these corrosion reactions
occur through electrochemical half-cell reactions as opposed to the metal interaction
directly with the environment through chemical reactions [31, p. 15].
The electrochemical corrosion process occurring in an aqueous solution comprises
of an anodic reaction coupled to one or more cathodic reactions [32, p. 77]. The
anodic reaction involves the dissolution of the solid metal as a metal ion and the
release of electrons (i.e., the anodic reaction is an oxidation reaction). For an anodic
reaction to be scientifically defined as a corrosion reaction, it must include both
mass and electron transfer across a metal and solution interface [31, p. 16]. The
anodic reactions for iron [31, p. 18] are shown in equations (1) and (2) and since they
involve both mass and electron transfer (metal ion dissolution and electron release,
respectively) they can be considered as valid corrosion reactions.

Fe(s) −−→ Fe2+ (aq) + 2 e− (1)


2+ 3+ −
Fe −−→ Fe (aq) + e (2)

In the cathodic reaction, a given species captures the electrons from the anodic
reaction and thereby undergoes reduction [31, p. 16] (i.e., the cathodic reaction
is a reduction reaction). In neutral and alkaline solutions, the cathodic reaction
is commonly depicted to be that of dissolved oxygen as in [31, p. 177], shown in
equation (3), but another plausible cathodic reaction under the conditions of the
secondary circuit could be the reduction of water [31, p. 102], shown in equation (4).
However, neither reaction can be considered a corrosion reaction as defined in [31,
p. 16], since they do not involve mass transfer across the metal-solution interface.
Nevertheless, since the oxygen from the resulting hydroxide OH – is consumed in
subsequent reactions to form metal oxide, and thus integrating in the metal and
crossing the metal-solution interface, the reactions can be considered indirect corrosion
reactions.

O2 (g) + 2 H2 O(l) + 4 e− −−→ 4 OH− (3)


2 H2 O + 2 e− −−→ H2 (g) + 2 OH− (4)
8

On a corroding piece of steel (e.g. the feedwater pipe carbon steel), both anodic
and cathodic half-cell reactions occur simultaneously on different sites of the steel
surface [31, p. 17], as demonstrated in Figure 3. The resulting overall reaction [31,
p. 19], shown in equation (5), can be considered dependent on both half-cell reactions.
This means that reducing the reaction rate of one of the reactions would decrease
the rate of the other reaction as well. This principle can be applied to the secondary
circuit feed water; if the content of dissolved oxygen is reduced, the cathodic oxygen
reduction reaction is impeded (assuming that no other cathodic reaction compensates
for that of oxygen) and consequently the oxidation of iron slows down. Under these
assumptions, the corrosion rate of iron can be reduced due to the codependency of
the anodic and cathodic half-cell reactions by removing oxygen with e.g. hydrazine
or other oxygen scavenger.

2 Fe(s) + O2 (g) + 2 H2 O(l) −−→ 2 Fe2+ (aq) + 4 OH− (aq) (5)

Figure 3: Simplified model of the coexisting half-cell reactions in equations (1) and (3)
for iron oxidation in a neutral or basic solution. [31, p. 19]

The two half-cell reactions can occur on the same surface because the metal has
an imperfect nature [31, p. 17]. These imperfections cause different parts of the
metal to become more energetically favorable for facilitating anodic and cathodic
reactions. Once the metal has dissolved from an anodic site, another part of the
surface becomes more energetically favorable and thus develops into a new anode. If
the anodic and cathodic sites randomly change locations, the entire surface ultimately
corrodes evenly [31, p. 19]. This process is called uniform or general corrosion.
On the contrary to general corrosion, if the local anodic and cathodic sites are
fixed to predisposed locations, corrosion will favor these locations while others remain
unharmed. The result is then localized corrosion. General and localized corrosion are
the two major types of corrosion [31, p. 25]. General corrosion can be considered to
affect carbon steel in the secondary circuit whereas localized corrosion forms applies
to both carbon and stainless steels.
For corrosion to occur as has been depicted in Figure 3, four constituents are
required and they comprise of: an anodic reaction, a cathodic reaction, a path for
conducting the electrons and an electrolyte [31, p. 18]. The first two constituents
have already been discussed. The electron conducting path is the third constituent
and it can be part of the corroding metal as shown in Figure 3, or the conducting path
9

can be in the form of wires as utilized in the electrochemical techniques discussed in


Chapter 5. The fourth constituent is the electrolyte which is a solution capable of
conducting a current as it contains charged species [31, p. 33]. These four constituents
facilitate corrosion reactions over a solution-metal interface, the characteristics of
which can be considered to determine the likelihood of aggressive corrosion to occur
and the rate at which the corrosion progresses. Therefore, this interface is often
the object of interest in corrosion studies in order to understand and describe the
underlying corrosion mechanisms. This interface is described next and later on
applied to the corrosion environment of the Loviisa secondary circuit.
In the Loviisa secondary circuits, the feedwater acts as the electrolyte and it is
an alkaline solution with a pH adjusted by ammonia to approximately 9.6. As the
electrolyte meets a metal, the various chemical species in the electrolyte interact with
the metal surface and some of them may adsorb as a layer onto it. If charged species
of either negative or positive sign are preferentially adsorbed, a structure of two
layers of oppositely charged species is formed at the metal surface. This structure
is described as a double layer or electrical double layer (EDL) which gives rise to
an electrochemical potential (i.e., the corrosion potential) over the metal-solution
interface. The EDL has been simulated using several models, out of which the
Bockris-Devanathan-Müller (BDM) model, shown in Figure 4, is considered to be
the most sophisticated [31, p. 42].

Figure 4: The electrical double layer and its absolute potential E (V) over the
metal-solution interface. The abbreviations IHP and OHP represents the inner and
outer Helmholtz planes, respectively. GC stands for Guoy-Chapman. The image was
reproduced from [31, p. 42].
10

The EDL as depicted in the BDM model comprise of three components: the
Guoy-Chapman diffusion layer, the outer Helmholtz plane (OHP), and the inner
Helmholtz plane (IHP). The three layers differ in composition of charged species
and therefore contribute to varying extent to the electrochemical potential over
the metal-solution interface [31, p. 43], as shown in Figure 4. The electrochemical
potential can be determined against a reference electrode and it can be considered
to serve as a thermodynamic measure of the metal-solution interface. The following
section describes how the electrochemical potential and the thermodynamics of the
metal-solution interface may determine the corrosion behavior of a metal.

3.2 Thermodynamics of corrosion


Depending on the thermodynamic conditions of the metal and its environment,
the metal may either corrode (i.e., dissolve indefinitely), be immune to corrosion
(i.e., remain in its solid state), or the corrosion products may form oxide layers
that ‘passivate’ the underlying metal surface and thus reduce the corrosion rates
to acceptable levels. With respect to iron, these three scenarios are presented in
Figure 5 as regions of aqueous solution acidity (pH) and electrochemical potential
(E) of the environment in so-called Pourbaix diagrams [31, p. 95]. The Pourbaix
diagrams show that iron corrodes indefinitely in acidic and very alkaline solutions in
the regions marked by ions Fe2+ , Fe3+ and HFeO2 – . In contrast, iron stays immune
within the region marked by Fe (solid iron) and passivates in the regions marked by
iron oxides Fe3 O4 (magnetite) and Fe2 O3 (hematite).

Figure 5: Pourbaix diagrams for iron at temperatures A) 25 ◦ C and B) 200 ◦ C. The


environment of the secondary circuit is highlighted by blue fields (pH25 around 9.7
and 6.5, for A and B, respectively). Under these conditions, the iron oxides magnetite
(Fe3 O4 ) and hematite (Fe2 O3 ) can be considered thermodynamically stable. The
unmarked region in B between Fe, HFeO2 – and Fe3 O4 are due to the uncertainty in

∆G calculations with respect to temperature. The diagrams were adapted from [31,
p. 113].
11

For this thesis, it is useful to know which forms of iron are thermodynamically
stable under the conditions of the WWER secondary circuit. An acceptable acidity
of the secondary circuit feedwater is generally considered to be approximately pH25◦ C
9.5 to 9.7 [16, p. 760]. Consequently, when the feedwater is heated to 228 ◦ C, the
pH228◦ C was estimated to 6.5 (calculated with MULTEQ software). These pH values
are highlighted in Figure 5 by blue fields. The blue fields are limited in terms of
potential by the water stability window between the dashed lines “a” and “b”. As
shown in Figure 5, either passivating oxide magnetite (Fe3 O4 ) or hematite (Fe2 O3 ) are
thermodynamically stable in the secondary circuit, depending on the electrochemical
potential.
The passive oxide film grown on carbon steel under secondary circuit conditions
has been previously characterized [33] as a magnetite (Fe3 O4 ) bi-layer film consisting
of a uniform and protective inner layer and a porous, non-protective outer layer of
irregularly distributed particles, as shown in Figure 6. The presence of magnetite
in the passive film has been confirmed with XRD and FTIR analyses, which were
also able to identify maghemite (γ-Fe2 O3 ) [34, 35] and tiny amounts of lepidocrocite
(γ-FeOOH) [35]. Although the studies did not specify the locations of the different
oxide phases in the passive film, the observations can be considered to correspond
with the Bilayer model [36, p. 787] embedded in Figure 6B, which in general terms
describes the passive film of iron as a bilayer composed of an inner and outer layer
of Fe3 O4 and γ-Fe2 O3 , respectively.

Figure 6: A) Surface and B) cross-section SEM images of low alloy steel (LAS) with
a 290 nm thick oxide film. The outer layer is abundant in particles and has a porous
structure, whereas the inner layer is compact. B contains also the corresponding
Bilayer model. SEM images from [24] and the bilayer model was reproduced from [31,
p. 225].

The oxide layers have been described by investigators to grow via several mech-
anisms. These include solid-state and diffusion [37], metal dissolution and oxide
precipitation [38] or a combination of both mechanisms in which the inner and outer
layers are formed by a solid-state and precipitation processes, respectively [39]. Since
the formed iron oxides allocate a larger volume than the dissolved metal (Pilling-
12

Bedworth ratios of 2.14 and 2.10 for Fe2 O3 and Fe3 O4 , respectively [31, p. 236]),
part of the oxide can be thought to form at the metal-oxide interface and part at the
oxide-solution interface. Another consideration is that the inner part of the film is
more compact and therefore more probable to grow via a solid-state mechanism, while
the outer porous layer can be reasoned to form via a dissolution and precipitation
mechanism as described in [40]. In general, the thickness of the resulting oxide has
been estimated to range from 0.1 μm to 1.5 μm depending on the temperature, pH
and the electrochemical potential as well as the time the metal has been exposed [34].
The characteristics of the passivating oxide are important since they may dominate
the rate at which corrosion reactions occur. The following section will describe the
corrosion rate and how it is affected by the presence of a passivating iron oxide at
the metal-solution interface.

3.3 Corrosion rate


The rate at which the corrosion reactions proceed is commonly called corrosion rate.
Since the corrosion reaction can be considered a process with many reaction steps,
its overall rate is dictated by the slowest reaction step, which is commonly referred
to as the rate-determining or rate-limiting step [41, p. 77]. The rate-limiting step of
an electrochemical reaction is in general defined [42, p 12] as either (i) the kinetics
of electron transfer reactions, adsorption reactions or preceding chemical reactions,
or (ii) the mass transfer of reagents to and from the electrode surface. That is, the
slowest step of an electrochemical corrosion reaction can either be a reaction step
(the one with the largest free energy barrier ΔG as defined by the absolute reaction
rate theory [31, p. 132]) or the transportation of reagents to and from the reaction
sites. If the kinetics of the reactions are fast, then the reaction is most likely limited
by mass transfer. Mass transfer may occur through three modes: diffusion (reagents
move down a concentration gradient), migration (charged species are forced into
movement by an electric field) and convection (reagents move due to stirring or flow
of the solution) [42, p. 17].
In case of steel covered by a passive oxide film, the electrochemical reactions can
be considered to occur at either side of the oxide film so that the anodic reactions
occur at the metal-oxide interface while the cathodic reactions take place at the
oxide-solution interface. This spatial separation of the electrochemical reactions
means that mass transfer occurs in both the solution and inside the oxide film. In the
solution, the mass transfer is presumably a combination of diffusion and convection
modes while migration can be considered negligible due to the absence of electric
fields. In the oxide film, the main mass transfer modes are migration and diffusion,
as shown in Figure 7, since solid material cannot facilitate convection. It is safe to
assume that the mass transfer through the oxide film is much slower than in the
solution since the film is a solid material. If the corrosion rate is limited by mass
transfer, the rate-limiting step can therefore be tied to the mass transfer properties
of the oxide film.
The mass transfer of reagents through the film, which is shown in Figure 7, include
the outward movement of metal species (Fe2+ and Fe3+ cations) and the inward
13

Figure 7: The corrosion interface of a passivated steel including both the BDM model
of the electrical double layer and the Bilayer model of the passive film structure. The
schematic characterizes the mass transfer modes of iron and oxygen species according
to the point defect model as described in [40].

movement of oxygen. The metal can be considered to move via both diffusion and
migration modes through local defects in the oxide film as assumed in the point defect
model [40]. To specify, the metal has been described [40] to move through the oxide
film in part as interstitial cations (Fe2+ , from equation (1)) and in part as lattice
point cations (Fe3+ or iron (III), from equation (2)). The outward movement of
Fe3+ occurs by solid diffusion driven by the dissolution of Fe3+ at the oxide-solution
interface and the formation of Fe3+ vacancies. The Fe3+ vacancies are filled with
adjacent Fe3+ cations which creates a net inward movement of the vacancies (and a
corresponding net outward movement of the Fe3+ cations). As the vacancies reach
the metal-oxide interface, they are filled with new Fe3+ ions from the oxidation
reaction of iron.
The inward movement of oxygen has been attributed [40] to the presence of
oxygen vacancies, which are formed as a byproduct of the iron oxidation reactions
from iron to iron (III). The oxygen vacancies move outward by solid diffusion and
consequently, oxygen moves inward. As the oxygen vacancies reach the oxide-solution
interface, they react with adsorbed water and are filled with oxygen atoms [40].
This movement of oxygen through the film allows for oxide film growth at both
the metal-oxide and oxide-solution interfaces as described in Section 3.2. As the
oxide film grows in thickness, the corrosion reactions can be assumed to proceed at
14

declining reaction rates [32, p. 89] since the flux from diffusion decreases with an
increase in layer thickness according to the Nernst-Planck equation [42, p. 17].
Although mass transfer is limited through the oxide film, the film conducts
electrons reasonably well as magnetite has been acknowledged to have n-type semi-
conducting properties [43]. The transfer of electrons at the metal-oxide and the
oxide-solution interfaces in the electrochemical reactions creates a Faradaic current.
This current can be characterized as a function of the electrochemical potential
over the metal-oxide-solution interface through various electrochemical techniques to
experimentally determine the corrosion rate of the system. Correspondingly, other
electrochemical techniques can be employed to characterize other properties of the
corrosion interface and to deduce the rate-limiting step of the corrosion process. Two
such electrochemical techniques: linear polarization resistance (LPR) and electro-
chemical impedance spectroscopy (EIS) are employed in this study and are reviewed
in Chapter 5.
This chapter has reviewed some of the most fundamental concepts of metallic
corrosion relevant for the conditions of the secondary circuit in terms of electrochem-
istry, thermodynamics and reaction kinetics. In the following chapter, these concepts
are applied as the chapter reviews the corrosion forms of carbon steel and stainless
steel PWR components from the literature. In addition, some of the concepts and
models included in this chapter will be applied in Chapter 5.
15

4 Secondary circuit corrosion and its mitigation


In the secondary circuit of PWRs, the water chemistry affects the degradation rate
of pipelines and other physical barriers that separate radioactive species from the
surroundings. Therefore, proper water chemistry in the secondary circuit is an
essential part of nuclear safety and has conventionally incorporated hydrazine as an
oxygen scavenger. In order to evaluate the feasibility of substitutes for hydrazine, it
is important to understand the corrosion forms that the use of hydrazine aims to
mitigate.
This chapter presents two materials particularly susceptible to corrosion in the
PWR secondary circuit and their associated corrosion mechanisms. Furthermore, the
chapter identifies the effect of oxygen on these corrosion mechanisms and presents
methods for controlling these corrosion phenomena using an oxygen scavenger such
as hydrazine. Finally, the chapter concludes by introducing hydrazine substitutes,
including two such compounds selected for experimental evaluation.

4.1 Flow assisted corrosion of carbon steel


As mentioned in Chapter 2.2, the structural components most susceptible to corrosion
in the WWER secondary circuit include carbon and stainless steels. The corrosion
processes of these materials can be considered as interlinked. First, the carbon
steel in the feedwater pipes dissolves through general or flow assisted corrosion
(FAC) [26]. Thereafter, the dissolved corrosion products are transported to the steam
generator where they deposit on various SG components through a process frequently
described as fouling [26, 44, 45]. Fouling decreases the thermal performance of the
steam generator and can facilitate local corrosion forms, including pitting and stress
corrosion cracking (SCC). This section reviews the FAC mechanism of carbon steel
from the literature while the following section describes the consequences of fouling
on the SG components.
Dissolution of carbon steel piping through flow-assisted corrosion (FAC, which is
also called flow accelerated corrosion) has been suggested as the most significant issue
in PWRs [26]. FAC can be described as the dissolution of the normally protective
oxide film under conditions of fast flowing water or steam. As the protective oxide
reduces in thickness, the underlying metal oxidizes at increasing rates. However,
FAC is sometimes used as a synonym to erosion corrosion, which is a common
misconception [46, p. 5]. FAC differs from erosion-corrosion in that it excludes
the mechanical wear mechanisms of cavitation and the impingement of bubbles
and particles [46, p. 5]. Instead of comparing FAC to erosion-corrosion, FAC can
be considered as an extension of general corrosion assisted by convective diffusion
processes provided by the flow of water or steam [46, p. 5].
The rate of FAC under secondary circuit conditions has been found to depend
on five factors: temperature, pH, material composition, hydrodynamics and the
corrosion potential of the metal [46, p. 3]. With respect to temperature, FAC of
carbon steel in fossil power plants has generally occurred between temperatures of
120 ◦ C to 280 ◦ C [47, p. 88]. The rate of FAC has in these cases been measured
16

to peak at a certain temperature, such as 150 ◦ C [47, p. 88] as the increase of


temperature allows for faster mass transfer but also reduces the solubility of iron
oxides [46, p. 21] and the rate of FAC can be expected to peak when the sum of
these two factors reaches a maximum value.
Regarding the solution pH, the FAC rate increases with reducing values of solution
pH as the solubility of iron oxides increases with reducing values of pH [48, p. 81].
Although carbon steel is susceptible to FAC, stainless steels are not. This
demonstrates that the material composition affects the rate of FAC. Alloying
elements, including chromium, have shown to significantly reduce FAC rates when
present in concentrations on the order of 1% or higher [48, p. 81]. The reduced rate of
FAC for alloys containing chromium has been attributed to the iron oxide containing
chromium having a much lower solubility than oxides containing no chromium [46,
p. 24].
The rate of FAC is affected also by hydrodynamical factors, by which the rate
increases with flow velocity, turbulence and surface roughness since they provide for
better mass transfer via convectional transport [46, p. 26] and increase the mechanical
stresses.
The most influential factor on the rate of FAC in the secondary circuit has been
recognized to be the electrochemical potential (ECP), which is directly affected
by the presence of oxidising and reducing species including dissolved oxygen and
hydrazine [46, p. 14]. As shown by [49, p. 128], the electrochemical potential increases
with oxygen content and FAC rates are generally lower in the presence of more than 5
ppb oxygen and oxidising conditions of ECP values greater than -0.3V (vs. standard
hydrogen electrode, SHE) [50]. The lower rates of FAC in the presence of oxygen has
been attributed to an oxide structure abundant in hematite (Fe2 O3 ) and iron oxide
hydroxide (FeOOH) that provides for a lower overall oxide solubility than an oxide
only consisting of magnetite (Fe3 O4 ) [46, p. 17], which is formed under reducing and
deoxygenated conditions as shown in the Pourbaix diagrams in Figure 5.
FAC has previously caused issues in the Loviisa secondary circuits. The cases
of FAC wall thinning have been concentrated to pipe fittings and welds found in
flanges, T-pieces, elbows and reducers in both the steam and water phase parts
of the circuits [22, p. 77]. In the most severe cases of FAC, the turbulent flow
of steam and water combined with insufficient chromium content in the steel was
believed to have accelerated the wall thinning [22, p. 77]. An example is the FAC
damages of T-connections and nozzles of the steam generator feedwater distribution
pipes which prompted the replacement of the entire distribution pipe in many steam
generators [29]. At the time of the above-mentioned FAC issues (up until 1994 [51]),
the acidity of the secondary water was held at neutral (i.e., pH25◦ C of approximately 7)
and involved no chemical additives. Currently, FAC seems to be under control in
Loviisa secondary circuits as the acidity of the secondary coolant is adjusted to higher
values of pH (as detailed in Section 2.2), and the components most vulnerable to
FAC have been replaced by more corrosion resistant materials [22].
Even though carbon steel components may not be subjected to FAC, which is
generally considered the main cause for release of corrosion products to the feedwater
system [34], general corrosion of carbon steel and other structural materials continues
17

to take place and produces low concentration of metal species in the secondary
circuit [52, p. 1]. These corrosion products, which can be in the form of dissolved
ions or dispersed particles [26], are transported downstream with the feedwater and
reach eventually the steam generators, where they accumulate in concentration and
subsequently precipitate or deposit. The following section depicts the deposition
of these corrosion products and the accumulation of impurities within them, which
has been the major cause for various degradation issues in the steam generators [45,
p. 17].

4.2 Degradation of steam generator tubes


The environment and the challenging conditions inside the steam generator (Sec-
tion 2.3) can result in the degradation of SG materials through various forms of
corrosion including FAC of the carbon steel (e.g., of the shell and tube support
plates TSP), stress corrosion cracking (SCC) of the stainless steel heat exchange
tubes, and clogging of the spaces between the TSPs and the tubes [44]. However,
other causes and forms of corrosion of SG components has been reported in various
reviews [27, 53, 44, 26, 16], which emphasizes that the mode of corrosion can vary
greatly between different types of SGs depending on the SG materials, the materials
in other components of the secondary circuit, and the methods for controlling the
secondary circuit water chemistry.
The degradation forms that are most essential for this thesis are those that the
use of hydrazine as an oxygen scavenger aims to mitigate. These are mainly pitting,
stress corrosion cracking (SCC) and denting of the SG heat exchange tubes [54], as
the presence of oxygen can be associated with an increased risk of all these three
corrosion forms.
Pitting is a localized form of corrosion that involves aggressive anions, such as
chloride ions Cl – , that break down the passive oxide film which allows for anodic
dissolution of the metal [31, p. 277]. The passive film on stainless steel, such as
titanium stabilized austenitic stainless steel 08Ch18N10T, is similar to that of carbon
steel (Figure 6) except for that part of the iron in the oxide has been replaced
by chromium and nickel atoms, thus producing a mixed oxide of general structure
Crx Niy Fe3 – x – y O4 [16, p. 650].
Corrosion pitting of metal initiates once the potentials reach critical values, or
the so-called pitting potential [31, p. 278], which for stainless steels are typically in
the range of 0.3 to 0.5 V in 0.1M NaCl at room temperature [55]. Since the pitting
potential is inversely dependent on the concentration of chloride, as demonstrated
in [56], the risk of pitting is decreased the lower the chloride concentration. Cor-
respondingly, the electrochemical potential of a metal increases with the oxygen
content, as demonstrated by [49, p. 128], and the presence of dissolved oxygen may
thus increase the risk of potentials approaching the pitting potential.
In case of the secondary circuit, which has very low concentrations of chloride and
oxygen as shown in Table 1 (i.e., less than 50 ppb and practically null, respectively),
the risk of pitting corrosion is by no means alarming. However, undesired ingress of
air may result in transient states during which an increased concentration of dissolved
18

oxygen may reside in the SGs. Moreover, chloride and other non-volatile corrosive
compounds may concentrate in flow-restricted locations of the SG where continuous
boiling occurs, such as the crevices between the SG tubes and TSPs [57].
The corrosion products from the feedwater and condensate systems form deposits
on top of the stainless steel mixed oxide film through particulate and crystallization
fouling [58]. As the water evaporates into steam at the proximity of these deposits,
corrosive impurities concentrate inside the deposits [53]. These impurities may alter
the pH beneath the deposits into regions where the oxide layer is no longer stable or
they may promote corrosion through other mechanisms.
Impurities that are considered to promote pitting and other forms of corrosion
include oxygen [53, 54] (from either deliberate on non-deliberate ingress of air in
the secondary circuit), sodium [54] and chloride [16, 26, 53, 54] (which may remain
unfiltered from e.g. a leaking ion exchanger), sulfate [26, 16, 54] (from e.g. ion
exchange resins), lead [54] (which may dissolve as it is an impurity found in most
metals) and copper [53, 54] (from heat exchange tubes in the condensers or preheaters).
Once corrosion pits have formed, they propagate autocatalytically and can act as
initiation sites for stress corrosion cracking [31, p. 277].
Stress corrosion cracking (SCC) is a mechanically assisted form of corrosion
that ultimately yields in macroscopic ruptures due to a combination of a corrosive
environment and mechanical stress [31, p. 315]. The stress corrosion crack may
initiate from a corrosion pit as the stresses from an external load are intensified at
the pit bottom [31, p. 321]. The crack then propagates through the metal by the
action of an external load, which may be assisted by hydrogen embrittlement, as the
environment in the crack may allow the generation of hydrogen ions [31, p. 231]. In
WWERs, stress corrosion cracking of the SG tubes have been determined [27, p. 4]
to have primarily initiated from the secondary side beneath deposits of corrosion
products and have been found to propagate both intergranularly or transgranularly
(IGSCC and TGSCC, respectively) [16, p. 647].
The locations for which SCC and pitting corrosion have occurred in WWER SGs
have been in the crevices between the tubes and carbon steel tube support plates
(TSP) but also on tube spans where there are no adjacent metals [27]. At the TSP,
the build-up of corrosion products can fill the entire crevice which can result in the
mechanical deformation of the SG tubes, commonly referred to as denting [59]. That
is, denting is not a corrosion form, but a result of corrosion and build-up of corrosion
products. For denting to occur, both oxidizing and acidic conditions are required, of
which the oxidizing conditions are provided by the presence of oxygen and copper in
the deposits [45, p. 15].
The degradation issues of WWER-440 SG tubes have been less extensive compared
to early PWR counterparts in western countries. Up until 2007, some 300 Western
PWR SGs had been replaced due to corrosion damage of the heat exchanger tubes [60]
whereas not a single WWER-440 SG has been in need of replacement [61]. The
main reason for the wast degradation issues of SGs in western countries has been
attributed [45, p. 60] to the use of the mill annealed nickel based Inconel Alloy 600,
which was found particularly sensitive to the corrosion mechanisms described above.
To summarize, the effect of removing oxygen on the SG degradation mechanisms
19

can be considered to be the following. With regards to pitting, the absence of oxygen
ensures that the potential does not reach the critical pitting potential and pitting can
thus not initiate. Due to the absence of corrosion pits, the removal of oxygen also
reduces the risk of SCC whose cracks commonly initiate from sites within corrosion
pits. In addition, a low oxygen content reduces the risk of denting but the mechanism
through which this occurs was not identified from literature. However, the removal
of oxygen can be considered to mitigate denting by either reducing the rates of TSP
carbon steel dissolution or the precipitation and attachment of corrosion products at
the deposits, or both.
This section has reviewed three major degradation routes of SG tubes that the
use of hydrazine as an oxygen scavenger aims to mitigate: pitting, SCC and denting.
The following section reviews the properties of hydrazine and substitutes of hydrazine
as oxygen scavengers and inhibitors in the secondary circuit.

4.3 Hydrazine as oxygen scavenger and corrosion inhibitor


This section describes hydrazine and the following section describes two hydrazine
substitutes with respect to their corrosion inhibiting properties and in particular,
their ability to remove dissolved oxygen from aqueous solutions.
Hydrazine (chemical formula N2 H4 , CAS number 302-01-2, EC number 206-114-
9) is an inorganic compound which as of 2011 has been labelled by the European
Chemical Agency as a substance of very high concern since it may be carcinogenic [62,
p. 7]. Within the EU, the main use of hydrazine (which amounts to approximately
80% of the annual consumption) include polymer production and other forms of
chemical synthesis, in which hydrazine acts as a precursor or reagent [62, p. 17].
The second largest application within the EU (which amount to about 20%) include
chemical refining, metal reduction and corrosion inhibitors [62, p. 17].
Hydrazine is used as a corrosion inhibitor under boiler and steam generator
conditions as it reduces the corrosion rates of metals through three reactions as
described by [13]. The first reaction is that with oxygen, which produces nitrogen
and water, shown in equation (6). The removal of oxygen prevents corrosion as
discussed in Section 4.2. Although the reaction has a theoretical ratio of 1 between
hydrazine and oxygen, the amount of introduced hydrazine needs to be 2-3 times
that of oxygen in order to remove oxygen entirely [13, p. 15].

N2 H4 (aq) + O2 (g) −−→ N2 (g) + 2 H2 O(l) (6)


The reaction rate of oxygen with hydrazine has been estimated [63] as a quasi-
first order reaction, as shown in equation (7), with reaction orders of 1 and 0.5
for hydrazine and oxygen, respectively. These reaction orders can be considered
indicative as other investigators have reported reaction orders of 0.5 [64, 65] for
hydrazine, and 0.5 [64] or 1 [65] for oxygen, depending on the nature of the metal
surfaces facilitating the reactions.
dCN2 H4 dCO2
− =− = kCN2 H4 CO0.52 (7)
dt dt
20

where:
Ci = concentration of species i
k = rate constant

The rate constant k in equation (7) has been demonstrated [63] to be temperature
dependent according to the Arrhenius expression shown in equation (8).
Ea
(︃ )︃
k = k0 exp − (8)
RT
where:
k = rate constant
k0 = frequence factor
Ea = activation energy, 51.8±0.4 kJmol−1 for carbon steel [63]
R = gas constant, 8.314 Jmol−1 K −1
T = absolute temperature in K

The second reaction, by which hydrazine reduces corrosion rates [13, p. 3], is
the thermal decomposition of hydrazine, which produces ammonia and nitrogen as
shown in equation (9). Compared to the oxygen scavenging reaction in equation (6),
this reaction occurs much slower and can in the presence of oxygen be considered
negligible under Loviisa secondary circuit conditions since the reaction rate constant
for oxygen scavenging is approximately 260 times that of the thermal decomposition
(calculated using estimates from [63] and [64] at 228 ◦ C on carbon steel). Nevertheless,
when no oxygen is present, the thermal decomposition reaction dominates and the
resulting ammonia aids in maintaining alkaline pH, which ensures the passivity of
steels as described in Section 3.2.

3 N2 H4 (aq) −−→ 4 NH3 (l) + N2 (g) (9)


The third corrosion impeding reaction of hydrazine [13, p. 4] is the reduction of
hematite to magnetite as shown in equation (10). Since magnetite layers are generally
considered more protective than hematite (except against FAC) [26], hydrazine can
be considered a passivating inhibitor [31, p. 359] (or passivator) as it promotes the
formation of a more protective passive film on steel surfaces through this reaction.

N2 H4 (aq) + 6 Fe2 O3 (s) −−→ N2 (g) + 2 H2 O(l) + 4 Fe3 O4 (s) (10)


The reaction kinetics for the passivating reaction could not be identified from the
literature, but are presumably slow in comparison to the oxygen scavenging reaction.
Nevertheless, with a constant concentration of hydrazine present in the SG, this
reaction ensures that the passive film remains in its magnetite form, which may help
to protect the SG tubes against impurities or should an excessive amount of oxygen
unexpectedly reach the SG [14, p. 7].
Hydrazine has three other favorable properties as an oxygen scavenger in the
secondary circuit as identified by [14, p. 7], which an alternative method for oxygen
scavenging would have to provide for. The first property is that hydrazine is a volatile
21

liquid, which means that it simultaneously protects sections of the secondary circuit
where liquid, steam, and both phases interact with metals. The second property is
that hydrazine does not produce solids that could cause problems in the circuit. The
third property is that hydrazine does not produce, or decompose into, species that
decrease the solution pH.
This section has reviewed the chemical properties of hydrazine as an inhibitor and
oxygen scavenger. To summarize, hydrazine can be considered a versatile corrosion
inhibitor with many favorable properties. The following section reviews the literature
on promising alternatives to hydrazine.

4.4 Alternatives to hydrazine in oxygen scavenging


Although finding an alternative to hydrazine has proved to be challenging, researchers
have identified several promising substitutes for oxygen scavenging [14]. These include
methyl-ethyl-ketoxime (MEKO), di-ethyl-hydroxylamine (DH) carbohydrazide (CH)
and erythorbic acid (EA). The oxygen reduction kinetics of these compounds under
conditions resembling maintenance and revision periods have been measured [66,
67] to be in order i) EA, ii) CH, iii) DH, iv) MEKO (from fastest to slowest).
Correspondingly, this section reviews the properties of CH, EA and DH. From these
three compounds, two are then chosen for experimental evaluation.

Carbohydrazide
Carbohydrazide (CH, (N2 H3 )2 CO, CAS number 497-18-7, EC number 207-837-2) is a
hydrazine derivative that was introduced to the US utility industry in the 1980’s as a
replacement to hydrazine [68]. At temperatures greater than 135 ◦ C [68], CH reacts
with oxygen via two reaction routes. The first reaction route is the direct reaction
shown in equation (11) and it governs the kinetics at lower reaction temperatures.
The rate of oxygen removal was determined to follow the expression in equation (12)
at temperatures of 25 and 50 ◦ C [66, 67]. The rate constant k for this reaction can
be assumed to follow the Arrhenius equation shown in equation (8).

(N2 H3 )2 CO (aq) + 2 O2 (g) −−→ 2 N2 (g) + 3 H2 O (l) + CO2 (g) (11)

1 dCCH dCO2
− =− = kCCH CO2 (12)
2 dt dt
The second reaction route is an indirect reaction, by which CH first decomposes
to hydrazine, as shown in equation (13), which subsequently reacts with oxygen
according to equation (6). The rate constant k for the thermal decomposition reaction
has been determined [69] to follow an Arrhenius expression.

(N2 H3 )2 CO (aq) + H2 O (l) −−→ 2 N2 H4 (aq) + CO2 (g) (13)


However, this reaction route occurs somewhat slower than the direct reaction
when under Loviisa secondary circuit conditions, considering that the reaction rate
constant was estimated to be approximately four times greater in the direct reaction.
22

Here, the rate constant for the direct reaction was calculated at 228 ◦ C by use of
linear regression from results of [66, 67] and assuming an Arrhenius relationship over
the temperatures range of 25 ◦ C and 228 ◦ C. Correspondingly, the rate constant via
the indirect route was estimated at 228 ◦ C using the results from [69] and assuming
that thermal decomposition acts as the rate-limiting step. That is, at temperatures
of 228 ◦ C, oxygen scavenging by CH can be expected to occur primarily (80%) via
the direct reaction and (20%) via the indirect reaction.
Similarly to hydrazine, carbohydrazide also reduces hematite into magnetite as
shown in equation (14) and has been shown [68] to passivate carbon steel surfaces more
effectively than hydrazine at temperatures below 138 ◦ C. In addition, carbohydrazide
is a volatile compound that does not generate solids to the system [14]. However,
when CH reacts it produces carbon dioxide (CO2 ), which may impact the conductivity
and pH of the coolant since dissolved carbon dioxide forms carbonic acid (H2 CO3 )
in water [70, p. 310].

(N2 H3 )2 CO (aq)+12 Fe2 O3 (s) −−→ 8 Fe3 O4 (s)+3 H2 O (l)+2 N2 (g)+CO2 (g) (14)

Although CH acts similarly to hydrazine, it is not hazardous at room temperature


to the environment nor acutely toxic to people [71] and can thus been considered
safer in terms of handling and storing than hydrazine. However, this safety benefit
has been argued [14] to be rather small since much of the handling and storing of
the chemical would occur by the use of various engineered and closed systems, which
already limits the exposure to people. Moreover, since CH decomposes into hydrazine
at temperatures over 180 ◦ C [12], it still poses a remote, but potential toxic risk in
case of exposure during operation.
Although CH can technically be considered a reasonable alternative, hydrazine
should ideally be replaced by a method or chemical that poses no toxic risk. The
following sections presents two organic oxygen scavengers that meet this objective.

Erythorbic acid
Erythorbic acid (EA, D-⟨−⟩) -isoascorbic acid, C6 H8 O6 CAS number 89-65-6 and EC
number 201-928-0) is a stereoisomer of ascorbic acid (AA), which is more commonly
known as vitamin C [72]. AA is an antioxidant found naturally in food and used
as a food suppliment [73] but also as an oxygen scavenger in food packaging [74].
For this study, the reaction of EA and AA with oxygen can be considered close to
identical, and is in its simplified version [66, p. 8] shown in equation (15). However,
the actual reaction mechanisms [75] for these organic molecules are quite complex
and their understanding is not necessary for the aim of this thesis.
1
C6 H8 O6 (aq) + O2 (g) −−→ C6 H6 O6 (aq) + H2 O (l) (15)
2
The kinetics of the oxygen scavenging properties of erythorbic acid have been
determined [66, 67] and the reaction rate was found to follow the expression shown
in equation (16) at a temperature of 50 ◦ C. In the measurements conducted at room
23

temperature [66], the reaction rate order with respect to EA was estimated to be zero
(i.e, the the reaction rate was independent of the EA concentration). Needless to say,
the rate constant k for EA can be assumed to also follow the Arrhenius relationship
shown in equation (8).
dCEA dCO2
− =− = kCEA CO2 (16)
dt dt
EA passivates steels by converting hematite and iron hydroxide to magnetite [76,
p. 7] according to the overall reactions [67, p. 10] in equation (17). However, EA was
determined an ineffective passivator of 22K carbon steel at a temperature of 50 ◦ C
when compared to hydrazine and carbohyrazide [67, p. 32].

C6 H8 O6 (aq) + 3 Fe2 O3 (s) −−→ C6 H6 O6 (aq) + 2 Fe3 O4 (s) + H2 O (l)


(17)
C6 H8 O6 (aq) + 6 FeOOH (s) −−→ C6 H6 O6 (aq) + 2 Fe3 O4 (s) + 4 H2 O (l)

The reactions of EA produce dehydroerythorbic acid (C6 H6 O6 ), which subse-


quently undergoes further reactions. Depending on the reaction temperature, pH and
oxygen content, erythorbic acid finally decomposes into several organic compounds,
including organic acids and salts, water and carbon dioxide [14, p. 17]. The corrosion
effect of these compounds has not been discussed in the literature. In addition, since
EA is not volatile [77], it does not follow the steam and thus cannot protect steam
components.
Nevertheless, EA does not contribute solids to the system [77], it is generally
recognized as safe (GRAS) by the US food and drug administrationa [78] and the
toxicity [72] is insignificant compared to hydrazine.

Diethylhydroxylamine
N-N-Diethylhydroxylamine (DH, (CH3 CH2 )2 NOH, CAS number 3710-84-7 and EC
number 223-055-4) is an organic compound that finds uses in paint and polymer
manufacturing but is primarily used as an oxygen scavenger in water treatment [79].
DH reduces dissolved oxygen according to the reaction [77, p. 3] shown in equa-
tion (18). Similarly to EA, the actual reaction mechanisms for DH with oxygen
are quite complicated and include several reaction steps [76, p. 6]. Therefore, the
theoretical ratio of DH to oxygen has been estimated to be 1.24 [77, p. 3] although
equation (18) shows a ratio of 0.44.

4 (CH3 CH2 )2 NOH (aq) + 9 O2 (g) −−→ 8 CH3 COOH (aq) + 2 N2 (g) + 6 H2 O (18)

The rate of oxygen reduction has been measured [66, 67] for DH and was found
to follow the expression in equation (19). The reaction rate was found to increase
considerably with temperature [67, p. 21], which suggests that the rate constant k
follows an Arrhenius relationship, shown in equation (8). However, the reaction rate
for DH was considerably lower than hydrazine, CH and EA, which may limit its
24

applicability. On the other hand, the oxygen reduction rate of DH can be increased
by applying organic catalyst as demonstrated by [80].
dCO2
− = kCDH CO2 (19)
dt
DH reacts with iron oxides as shown in equation (20) and has thus been considered
to passivate steels [76, p. 6]. However, the passivating effect of DH on 22K carbon steel
was measured to be quite ineffective compared to that of hydrazine and carbohydrazide
at a solution temperature of 50 ◦ C [67, p. 32].

DH + 6 Fe2 O3 −−→ CH3 CH−NOH + CH3 CHO + 4 Fe3 O4 + H2 O


(20)
DH + 12 FeOOH −−→ CH3 CH−NOH + CH3 CHO + 4 Fe3 O4 + 7 H2 O

The reaction products of DH undergo subsequent reactions and the final degrada-
tion products include acetaldehyde, dialkyl amines and acetaldoxime acetic acid [77].
Depending on what compounds are formed, the reaction products of DH may either
increase or decrease the pH of the environment. That is, the pH increases should
diethylamine be produced [76, p. 6], whereas the pH decreases if acetic acid forms [77].
However, the literature does not mention the requisite for either product to form
nor the rate of the decomposition reactions. Regardless of decomposition route and
kinetics, the final degradation products should include carbon dioxide, which again
may acidify the coolant.
Although DH is volatile, some of its reaction products are not. E.g., in an alkaline
solution containing sodium hydroxide, acetic acid (CH3 COOH, from the oxygen
reduction in equation (18)) reacts with sodium to form sodium acetate which would
remain in the SG [76, p. 6]. That is, DH contributes probably with solids (such as
sodium or calcium acetate [77]) that may deposit in the SG. Other reaction products,
such as diethylamine, are volatile and can thus proceed to the steam-containing
sections of the secondary circuit and protect these by controlling the pH [76, p. 6].
The toxicity of DH [81] is minor in comparison to hydrazine and can thus be used
more safely. However, DH has been suspected to be carcinogenic and mutagenic
and thus included [82, p. 8] in the community rolling action plan (CoRAP, which is
the first step of chemical regulation via REACH). That is, DH may face a similar
regulation as hydrazine in the future.
To summarize, this section has introduced three organic compounds that may
provide effective oxygen scavenging properties to the secondary circuit. Although
they do not fulfill the properties of hydrazine on an individual level, they may be
combined with other chemicals or techniques to do so. In this study, the rate of
oxygen reduction for EA and DH are measured and compared against hydrazine.
The following section presents two methods for determining the oxygen reduction
rates.
25

5 Electrochemical techniques
This chapter evaluates the electrochemical techniques used to determine the rate
of oxygen reduction and the corrosion properties of carbon steel as an effect of
concentration of oxygen scavengers. This chapter aims to evaluate the techniques with
respect to if they produce reliable outcomes under the conditions of the experiments.
Section 5.1 introduces the electrochemical cell and the three-electrode set-up which the
measurement techniques rely on and the subsequent sections review the measurement
techniques from the literature.

5.1 Electrochemical cells


An electrochemical cell used for experimental purposes is shown in Figure 8. The
electrochemical cell commonly comprises of three electrodes: i) the working electrode
(WE), which is the sample material and object of study, ii) the reference electrode
(RE, such as Ag/AgCl), against which the potential of the WE is measured, and iii)
the counter electrode (CE, commonly platinum), which provides a path for current
and closes the cell circuit [31, p. 168].

Figure 8: A glass cell used for electrochemical measurements including a three


electrode set-up of working, counter and reference electrodes. Image adapted from [31,
p. 168].

The electrodes are connected to a control and measuring device, commonly called
a potentiostat, which is connected to the electrochemical cell as shown in Figure 9.
The potentiostat can carry out various electroanalytical measurements in both direct
current (DC) and alternating current (AC) modes.
In DC mode, the potentiostat can be used to conduct measurements that are either
current or potential controlled [31, p. 168]. In potential controlled measurements,
such as linear polarization resistance (LPR), the potentiostat supplies a current at
26

Figure 9: The three-electrode set-up connected to a potentiostat under a potential


controlled measurement. E stands for the controlled WE potential, CA for control
amplifier (which is a type of operational amplifier), Um and Rm for the current
measuring voltage and resistor, Icell and Rcell for the cell current and resistor, and
Ua for the measured actual voltage between the RE and WE. The image was drawn
based on [83, p. 254].

the CE in order to maintain the potential of the WE (with respect to the RE, Ua )
at a desired value E and records the current Icell [31, p. 168]. The LPR method is
reviewed from the literature in Section 5.3.
In AC measurement mode, such as electrochemical impedance spectroscopy
(EIS), the potentiostat applies an alternating current or potential perturbation of
small amplitude on the WE over a range of frequencies and measures the resulting
impedance of the sample [31, p. 427]. Section 5.4 reviews EIS from the literature.
The results from electrochemical measurements can be used to depict the kinetics
of the electrochemical cell by the use of an equivalent circuit. The most simple of these
equivalent circuits is considered to be the Randles circuit, shown in Figure 10 [84].
The current Icell in an electrochemical cell corresponding to the Randles circuit
results from two processes. The first process is the electron transfer of a redox
reaction [42, p. 639] (e.g., the corrosion reaction of iron according to equation (5)),
which generates a current that passes the charge transfer resistor component Rct .
This process is also termed [42, p. 22] a Faradaic process and the resulting current a
Faradaic current. The value of Rct for a corroding metal is inversely proportional to
the reaction rate of the redox reaction [42, p. 639]. That is, Rct can be considered
inversely proportional to the corrosion rate.
The second current generating process is the charging of the EDL [42, p. 639],
represented by the capacitor Cedl , by which charged species in the electrolyte approach
the WE surface and are adsorbed onto it. This charging of the EDL results in a
so-called non-faradaic current [42, p. 22].
27

Both Faradaic and non-faradaic currents (i.e., charge transfer and capacitive
currents, respectively) pass through the solution resistor Rs [42, p. 639]. The solution
resistance causes a voltage drop on the measured voltage Ua equivalent to Icell Rs
according to Ohm’s law should the RE and CE be positioned at the same distance
from the WE [42, p. 24]. By placing the RE as close as possible to the WE, the
measurement error from the voltage drop can be reduced by an amount Icell Rc to a
value Icell Ruc . The Rc and Ruc are the compensated and uncompensated part of the
solution resistance, respectively [42, p. 24].

Figure 10: Suggested equivalent circuit for the carbon steel metal-solution interface.
WE is the working electrode, CE is the counter electrode, Cedl is the equivalent
capacitor for the electrical double layer, Rct is the equivalent resistor for charge
transfer and Rs is the equivalent resistor for the solution. Image adapted from [31,
p. 432].

5.2 Electrochemical oxygen sensors


The rate at which oxygen scavengers reduce dissolved oxygen from a liquid can be
determined by measuring the oxygen concentration as a function of time, as shown
in Section 4.3. To this end, this thesis utilizes two types of electrochemical sensors
for measuring the oxygen concentration: i) a commercial membrane sensor and ii) an
experimental redox sensor. This section discusses the operating principles for these
two sensors.

Commercial sensor
The first oxygen sensor is an "Orbisphere model 3660 analyzer" produced by Hach
Company which comprises of a two electrode setup, shown in Figure 11. The
electrodes are immersed in an electrolyte and covered by an oxygen permeable
membrane, which allows for diffusion of dissolved oxygen between the cathode and
the analyzed solution [85, p. 508].
28

Figure 11: Top-view of the Orbisphere 3660 analyzer sensor head. Image adapted
from [86, p. 54].

According to the operational manual [86, p. 53], the sensor operates by applying a
constant voltage to the anode and measuring the current at the anode, which fits the
description of a polarographic oxygen sensor described by [87, p. 17]. A polarographic
sensor operates by applying a constant potential at the anode, causing the oxygen to
reduce at the cathode according to equation (3) [85, p. 508]. The resulting current
through the sensor is then directly proportional to the oxygen concentration via the
relation of Fick’s first law of diffusion as discussed in [88, p. 260].
The Orbisphere sensor measures the oxygen concentration over a range of 1
ppb to 80 ppm at an accuracy of ±1% [86, p. 51]. In addition, the analyzer
compensates the measured current with respect to the temperature of the solution
by the use of an external temperature sensor. Furthermore, the highest operating
temperature and pressure are 100 ◦ C and 20 bar, respectively. The response time of
the sensor after an 90% change in oxygen concentration is 30 seconds. Considering
these operating conditions, the sensor can be regarded applicable for use as an
concentration measurement device for the experiments, provided that the sensor
is placed in a location where the temperature and pressure are within the sensor
specifications.

Experimental sensor
The second sensor is a three electrode set-up with WE and CE consisting of platinum
and a Ag/AgCl RE. The in-house sensor allows for estimate measurement of the
concentration of dissolved oxygen from the potential of the platinum WE assuming
the reaction in equation (21) using the relationship in equation (22) as demonstrated
by [89, p. 13].
1
O2 (g) + H2 O (l) + 2 e− −−→ H2 O2 (l) + 2 OH− (aq) (21)
2
zF (︂
(︃ )︂)︃

CO2 (t) = exp E(t) − E (22)
RT
29

where:
z = number of transferred electrons, 2 in this case
F = Faraday constant, 96485 C mol−1
R = gas constant, 8.314 J mol−1 K −1
T = absolute temperature in K
E(t) = measured cell potential at time t in V

E = standard cell potential, 0.37 V at 228 ◦ C and pH 7 [89, p. 13]

In previous measurements [89, p. 13], the sensor was shown to generate good esti-
mates of oxygen concentrations at 10 ppm using the relationship above. However, at
lower oxygen concentrations, such as 100 ppb, the sensor was not able to measure the
oxygen content accurately. Therefore, the equation (23) was derived for determining
the rate of oxygen reduction. The origin of equation (23) is given in Appendix A.
RT ′
E(0) − E(t) = kt (23)
4αrF
where:
E(t) = measured cell potential at time t in V
αr = transfer coefficient for the oxygen reduction reaction, 0.043 [90]
k ′ = apparent rate constant in s−1

The benefit of this sensor is that it is able to operate at high temperatures and
pressures whereas the membrane sensor is limited to 100 ◦ C and 20 bar. That is,
under conditions resembling the secondary circuit, the experimental sensor can be
directly placed at the location of oxygen scavenging, whereas the commercial sensor
has to be placed in a cooled segment on the return line. Therefore, the experimental
sensor can be considered to gather more representative data.

5.3 Linear polarization resistance


Linear polarization resistance (LPR) is a non-destructive polarization method carried
out with a three electrode set-up. Polarization essentially means that the potentiostat
supplies current via the CE to shift the WE potential an amount η into either positive
or negative direction from the open circuit potential EOCP [31, p. 127]. When the
amount of polarization η is low (on the order of ±30 mV), the measured current
correlates linearly with the potential as shown in Figure 12.
The inverse of the slope for the linear region in Figure 12 is then equivalent by
Ohm’s law to a resistance, or the so-called polarization resistance Rp [31, p. 161],
as shown in equation (24). The polarization resistance Rp can be considered the
sum of the charge transfer resistance Rct and the uncompensated resistance Ruc
of the equivalent circuit shown in Figure 10 [91, p. 1145]. If the uncompensated
resistance is much smaller than that of charge transfer, the polarization resistance
can be considered approximately that of the charge transfer. In case of steels covered
with an oxide film such as WWER secondary circuits, the polarization resistance
30

Figure 12: Polarization curve in the proximity of the open circuit potential showing
a linear dependency between current and potential. Image adapted from [31, p. 160].

consists, in addition to the charge transfer resistance, of the resistance of the oxide
film.

= Rp = Rct + Ruc (24)
dinet
where:
η = amount of polarization in V
inet = net current density in Acm−2
Rp = polarization resistance in Ωcm2
Rct = charge transfer resistance in Ωcm2
Ruc = uncompensated resistance in Ωcm2

The electrode set-up used in the LPR measurements were similar to that for the
experimental oxygen sensor and the LPR measurements can be considered applicable
also at high temperatures and pressures. However, LPR as a method relies on
several assumptions in order to produce reliable results [91, p. 1144] which have to
be considered. One assumption is that the system has reached a steady state so
that the open circuit potential remains constant. In the measurements in this thesis,
there was a risk of the system not reaching steady-state due to practical reasons
prior to starting the measurements in particular after and during the injection of
oxygen scavengers. Therefore, these measurements must be analyzed cautiously.
Another assumption is that the measured potential and current show a linear
relationship as shown in Figure 12, which would have to be confirmed with linear
regression analysis. However, due to the large amount of LPR data gathered in the
31

thesis experiments, linear regression analysis was not conducted on every single data
set. This has to be considered in the discussion of the results.
A third assumption is that the uncompensated resistance is much smaller than
the charge transfer resistance. In our experiments, placing the RE as close as possible
to the WE was not possible due to the arrangement of the measurement cell and
therefore the uncompensated resistance can be expected to be significant and must
thus be accounted for accordingly. One method for discriminating the uncompensated
and charge transfer resistances is by measuring the voltage drop separately or by
complementary EIS measurements [91, p. 1145], of which the latter is reviewed in
the following section.

5.4 Electrochemical impedance spectroscopy


Electrochemical impedance spectroscopy (EIS) is an electrochemical technique that
applies an sinusoidal perturbation of either current or potential to the electrochemical
system of interest, such as the metal-solution interface of carbon steel under secondary
circuit conditions, and measures the impedance response of that system [31, p. 427].
The experimental setup for EIS comprises of a three electrode set-up inside an
electrochemical cell and a potentiostat [31, p. 430].
In EIS measurements, the presence of an alternating electric field causes different
parts of the metal-solution interface to respond at characteristic frequencies [31,
p. 427], which the EIS potentiostat measures as a value of impedance. Considering
the equivalent circuit in Figure 10, the resistors Ruc and Rct have impedances (Zuc and
Zct , respectively) equivalent to their resistances as shown in equation (25) whereas
the capacitor Cedl has an impedance Zedl expressed as a complex number as shown
in equation (26) [31, p. 431].

Zuc = Ruc
(25)
Zct = Rct

where:
Zi = impedance of component i in Ωcm2
Ri = resistance of component i in Ωcm2

1 1
Zedl = = (26)
jωCedl j2πf Cedl
where:
2
Zedl = electrical double layer
√ impedance in Ωcm
j = imaginary number −1
ω = angular frequency 2πf in Hz or s−1
Cedl = electrical double layer capacitance in µF cm−2
f = frequency in Hz or s−1
32

Circuit analysis and a bit of algebra [92, p. 302] gives the total impedance of the
equivalent circuit in Figure 10 as a complex number shown in equation (27).
2
Rct ωCedl Rct
Z = Ruc + 2 2
− j 2 2
(27)
1 + ω 2 Cedl Rct 1 + ω 2 Cedl Rct
The total impedance Z consists of a real (Z′ ) and imaginary part (Z′′ ) as shown
in equations (28),(29) and (30) [31, p. 433]. The absolute impedance |Z| is the
Euclidean distance as shown in equation (31)[31, p. 430].

Z = Z ′ + jZ ′′ (28)

Rct
Z ′ = Ruc + 2 2
(29)
1 + ω 2 Cedl Rct
2
ωCedl Rct
Z ′′ = − 2 2
(30)
1 + ω 2 Cedl Rct

⎡ (︄ )︄2 )︄2 ⎤ 21

(︄
2
Rct ωCedl Rct
|Z| = Z ′ 2 + Z ′′ 2 = ⎣ Ruc + 2
+ 2 2
⎦ (31)
1 + ω 2 Cedl Rct 1 + ω 2 Cedl Rct

Equation (31) shows that the absolute impedance can have two limit values.
At very high values of angular frequency ω, the impedance approaches that of the
uncompensated resistance Ruc as shown in equation (32). On the other hand, when
the angular frequency approaches zero, the impedance approaches the sum of the
uncompensated and charge transfer resistances as shown in equation (33).


)︃2 (︄ )︄2 ⎤ 12
2
Rct Cedl Rct
(︃
lim |Z| = ⎣ Ruc + + 1
⎦ = Ruc (32)
ω→∞ 1+∞ ∞
+∞
⎡ ⎤1
)︃2 )︃2 2
Rct 0
(︃ (︃
lim |Z| = ⎣ Ruc + + ⎦ = Ruc + Rct (33)
ω→0 1+0 1+0

These limits appear as horizontal lines in the Bode plot in Figure 13A [31, p. 435].
Between the horizontal lines, the spectra appears as a sloped line, which represents
the capacitor of the EDL. The impedance spectrum can also be plotted as the
imaginary impedance against the real impedance in a so-called Nyquist plot shown in
Figure 13B [31, p. 433]. Here, the limit values for the absolute impedance are shown
intercepting the x-axis at either end of a half-circle with a radius Rct /2. At the point
of the half-circle apex (i.e., when dZ′′ /dω= 0), the value of angular frequency can be
used to determine the capacitance of the EDL according to equation (34).
1
ωZmax
′′ = (34)
Rct Cedl
33

Figure 13: Impedance spectrum shown as A) Bode plot and B) Nyquist plot. Images
adapted from [31, p. 434].

By the use of EIS, all three components of the equivalent circuit for the metal-
solution interface can be determined as described above. However, an actual electro-
chemical system is usually more complex and the measured impedance responses
may deviate from that of the Randles circuit. In that case, the equivalent circuit
may have to be adjusted by adding or replacing components, thus constructing a
more elaborate equivalent circuit.
Finding a representative equivalent circuit for a complex system may prove
challenging as numerous circuits may equally well model the same impedance response
as shown by [93]. Therefore, the equivalent model should be constructed so that
each component represents a plausible chemical or physical process of the measured
electrochemical system. The use of an equivalent circuit should also be justified by
e.g., changing the physical properties of a component and verifying its effect on the
impedance response [94, p. 49].
The impedance spectra of carbon steel under secondary circuit conditions, shown
in Figure 14, significantly deviate from the ideal response of the Randles circuit.
The deviation was attributed to [24] the properties of the inner oxide layer rather
than the capacitance of the EDL. The model for analysing these spectra did not
rely on an equivalent circuit, but was derived [95, 96] from properties of the oxide
layers according to the mixed conduction model, some of which were introduced in
Section 3.3.
Considering the thesis scope, it would be excessive to fit impedance spectra to
the mixed conduction model. Instead, this thesis assumes a modified Randles circuit
shown in Figure 15 in which the EDL capacitor Cedl has been replaced by a constant
phase element (CPE), which is an imperfect capacitor. The CPE has an impedance
ZCPE described [97] in general by the expression in equation (35), which is similar to
the complex impedance of a capacitor. Correspondingly, the CPE has a capacitance
CCPE according to equation (36) [98].
34

Figure 14: Absolute impedance (A) and phase angle (B) bode plots for carbon steel
under secondary circuit conditions. The arrows correspond to the ionic conduction
of the inner oxide layer (1.), the processes at the inner oxide-solution interface (2.)
and the electronic properties of the inner oxide layer (3.) [24, p. 111].

The CPE capacitance and impedance are described by the nameless parameters
T and P. The parameter T is directly related to the CPE capacitance as shown in
equation (36), whereas the parameter P is a measure of the capacitative proportion
of the CPE. For P values of 1, the CPE is considered an ideal capacitor whereas for
P values of 0 it is considered a perfect resistor [98]. For P values between 0 and 1,
the CPE can be considered a semi-capacitor.

1
ZCP E = (35)
T (jω)P
1
CCP E = T P (36)

where:
ZCP E = constant phase element impedance in Ωcm2
T = parameter in µF cm−2
P = parameter [0 : 1]
CCP E = capacitance of the CPE in µF cm−2
35

Figure 15: Suggested equivalent circuit for fitting impedance spectra. The CPE is
an imperfect capacitor, which may represent either the EDL or the iron oxide.
36

6 Experimental
This chapter presents the methodology, materials and equipment used in the thesis ex-
periments. This chapter thereby provides necessary information for other researchers
to reproduce the experiments and possibly confirm or debunk any drawn conclusions.

6.1 Equipment and materials


The experiments were conducted in a water circuit shown schematically in Figure 16.
The water was purified with a mixed-bed ion exchanger (Veolia SD2000) and the
pH25◦ C was adjusted using ammonia to 9.6±0.2, resulting in a conductivity of 13±5
μS/cm2 . The content of dissolved oxygen was adjusted to 100±20 ppb and 10±2
ppm by bubbling nitrogen and oxygen in the mixing tank.
The circuit is divided into a low pressure, cold section and a high pressure, hot
section. The cold section, which had a temperature of 26±1 ◦ C and a pressure of
1.0±0.3 bar above the ambient, provided the experiments with water of consistent
quality in terms of conductivity, pH and oxygen content. The cold section also
accomodated the commercial oxygen sensor.
The condition of the hot section was adjusted to a temperature of 228±1 ◦ C
and pressure of 65±1 bar, resembling the feedwater at the entry point to Loviisa
steam generators but at a comparatively small flow rate of 220±10 ml/min. The
6 mm outer diameter pipeline of the hot section was made of titanium, while the
flow-through cell consisted of AISI 316 stainless steel.

Figure 16: Water circuit comprising of a cold (blue) and hot (red) section. The
measurements took place at locations marked by O2 (t), LPR and EIS. γ indicates a
conductivity sensor while LP and HP are low and high pressure pumps, respectively.
Image reproduced from [24].

The 0.3 l volume flow-through cell housed a 0.05 M KCl Ag/AgCl RE (-0.0051 V vs
SHE at 225 ◦ C [99, p. 655]) and two WE-CE pairs for electrochemical measurements.
The first electrode pair was a 7 cm2 platinum mesh CE wrapped around a 1 cm2
37

cuboid-shaped 22K carbon steel WE at a separating distance of 5 mm. Grade 22K


carbon steel has a nominal composition of 0.19-0.26% C, 0.75-01.0% Mn, 0.2-0.4% Si,
≤0.03% P, ≤0.03% S, ≤0.3% Cr, ≤0.3% Ni, ≤0.3% Cu and balance Fe [19, p. 109].
The second electrode pair comprised of a 10 cm2 platinum wire CE coiled around a 1
cm2 platinum wire WE at a separating distance of 5 mm. The flow velocity in the
flow-through cell was approximately 3 mm/s.
The studied oxygen scavengers were erythorbic acid (EA, 98% Alfa Aesar) and
diethylhydroxylamine (DH, 97%, Alfa Aesar). In addition, hydrazine (HZ, hydrazine
monohydrate, 98% Sigma-aldrich) was used as a reference oxygen scavenger. The
oxygen scavengers were dissolved into deoxygenated and deionized water producing
stock solutions of concentrations between 0.15 mM and 60 mM.
The oxygen scavenger stock solutions were injected at the entry point to the
flow-through cell using a liquid chromatography pump (Shimadzu LC-20AT). The
oxygen scavengers were injected to achieve molecular concentrations between one to
ten times that of the initial concentration of dissolved oxygen. The rate of injection
was adjusted for the actual oxygen content to a value between 2 ml/min to 5 ml/min,
having corrected for the pumping rate error described in Appendix B. The oxygen
scavengers were finally removed from the water by filtering through the ion exchanger.

6.2 Oxygen reduction kinetics


The concentration of dissolved oxygen was continuously measured prior to, during and
after administration of oxygen scavengers using two oxygen sensors. The experimental
Pt-Pt sensor, which was located in the hot section flow-through cell, measured the
oxygen concentration close to the injection point. The commercial sensor, which was
placed in the cold section return line, measured the oxygen concentration roughly
three minutes downstream from the injection point.
In case of the commercial sensor, the rate of oxygen reduction was determined from
the general equation [63] shown in (37). The expression was simplified to equation (38)
by substituting with the apparent rate constant k ′ = kCOS m
and assuming a unity
value for n as previous measurements have shown [66, 67] a pseudo-first reaction
rate order with respect to oxygen. The simplified expression can be reorganized
with a bit of algebra and finally integrated to give the linear relationship shown in
equation (39).

dCO2 m
− = kCOS COn2 (37)
dt
k′ =kCOS
m
dCO2
===== ⇒ − = k ′ CO 2 (38)
n=1 dt

algebra CO2
======⇒ ln = k′t (39)
integration CO2

Using the relationship in equation (39), linear regression analysis was made to
determine the apparent rate constant k ′ . Based on the determined k ′ values, a second
38

linear regression analysis was applied to determine the rate constant k from the
relationship k ′ = kCOS
m
.
In case of the experimental sensor, the apparent rate constant k ′ was determined
using linear regression of the expression in equation (40). The resulting k ′ values were
then feed into a second linear regression analysis to determine the rate constant k
from the relationship k ′ = kCOSm
. The origin of equation (40) is given in Appendix A.
RT
E(0) − E(t) = (k ′ − k ′′ )t (40)
4(αo + αr )F

6.3 Electrochemical measurements of carbon steel


The effect of OS addition on the linear polarization resistance and electrochemical
impedance spectra were measured for a carbon steel sample conditioned for 72
hours in 228 ◦ C water with an oxygen content of 100±20 ppb using an Autolab
PGSTAT302F potentiostat and the electrochemical cell described above. Additional
measurements were conducted in oxygen concentrations of 10±2 ppm. The samples
were ground with P1200 grit emery paper prior to installation to the water circuit.
In the LPR measurements, the potential was swept from -20 mV to +30 mV with
respect to a stable open circuit potential at a rate of 10 mV/min after held for 60 s
at -20 mV. The intent was to use the standardized potential window of -30 mV to
+30 mV relative to the corrosion potential but the initial 10 mV (from -30 mV to -20
mV) produced non-linear relationships and were thus omitted in all measurements.
The LPR measurements were repeated three times and the polarization resistances
were determined using linear regression.
In the EIS measurements, a 30 mV alternating current was applied over a broad
frequency range of 10 kHz to 1 mHz in the absence of oxygen scavengers and over a
narrower frequency range 200 Hz to 10 mHz in the presence of oxygen scavengers.
The narrow frequency range in the presence of oxygen scavengers was applied to
speed up the impedance measurements since the prepared stock solution of oxygen
scavengers would otherwise have run out during the measurement. The impedance
spectra were analyzed and fit to equivalent circuits using Zview software.
39

7 Results and discussion


This chapter presents the experimental results and compares them to the results
of previous studies. Should the results deviate from previous studies, this chapters
considers procedural and theoretical reasons that could explain such deviations.

7.1 Oxygen reduction kinetics


The rate at which the oxygen scavengers (OS) react with dissolved oxygen (DO) was
determined by measuring the change in oxygen concentration using a commercial
sensor and the change in electrochemical potential of an experimental sensor as
the oxygen scavengers were administered into the experimental water circuit. The
experimental Pt-Pt sensor was located close to the injection point and the commer-
cial oxygen sensor was located in the cold section return line. The initial oxygen
concentration was adjusted in the majority of the measurements to 100±20 ppb but
additional measurements were also carried out in 10±2 ppm.

Commercial sensor
The typical measurement procedure is presented in Figure 17A, which shows the
decline in oxygen concentration as the oxygen scavenger was administered into the
water circuit. The measurements were repeated up to seven times by periodically
injecting the oxygen scavengers with a dosing pump. Prior to injection, the pumping
rate was adjusted to the oxygen concentration at the time to ensure a consistent
concentration ratio of oxygen scavengers and dissolved oxygen during each injection.

Figure 17: Exemplary results from A) the concentration measurements (EA, OS/DO

ratio 6, 80-90 ppb CO2 ) and B) the corresponding linear regression analysis from
the first four injections. The linear regression analysis was used to determine the
apparent rate constant k ′ . The k ′ value of the first injection was usually found to be
an outlier as shown in B).
40

The apparent rate constants k ′ were determined from the linear relationships
shown in Figure 17B. For each injection and corresponding instance of analysis,

the initial oxygen concentration CO2 was set to the last measured stable value of
oxygen concentration prior to the decline. The k ′ values were tested using the Q test
method [100] and the resulting outliers were excluded if the certainty was 90% or
more.
The oxygen reduction kinetics determined using the commercial sensor are shown
with respect to the injected amount of OS in Figure 18. The apparent rate constants
k ′ showed a semi-linear relationship with respect to the OS concentration as shown
in Figure 18B which corresponds to a unity value for the exponent term m. The
apparent rate constants k were determined over the linear regions using linear
regression analysis as shown in Figure 18B and the rate constant values are shown
in Table 2.
The results show that HZ and EA have considerably greater reaction rate constants
than that of DH, which is a trend also shown in previous measurements [67] and
can thus be considered an expected result. On the other hand, the reaction rate
constant was found greater for EA than for HZ whereas previous measurements
have shown that HZ is considerably faster than EA. Considering that the previous
measurements [66, 67] were conducted at low temperatures, one explanation could
be that the reaction kinetics of EA are accelerated to a greater extent with the rise
in temperature, thus allowing EA to approach the kinetics of hydrazine. However, a
beneficial temperature dependency cannot singlehandedly explain why EA would
have faster kinetics than HZ, considering that EA is an organic chemical with a
significantly larger molecular structure and that it may undergo many reaction steps
prior to the oxygen reduction reaction compared to HZ.

Figure 18: The apparent rate constant k ′ as a function of A) the ratio of OS to DO


and B) the OS concentration.

The indication that EA was measured to have faster kinetics than HZ can
instead be attributed with greater confidence to a systematic difference in the
41

measurements. The measurements differed in the initial oxygen concentration, which


for measurements of EA were in the range of 80-100 ppb and for HZ in the range of
100-120 ppb. Since the injected amount of OS was dictated by the concentration ratio
of OS to oxygen, the lower oxygen concentrations during EA measurements resulted
in lower OS concentrations in Figure 18B. As a consequence, the linear regressions
produce a seemingly steeper slope for EA than for HZ although the concentration
ratios were in fact the same as indicated in Figure 18A.
Although the automatic adjustment of the oxygen concentration provided consis-
tent oxygen levels between consecutive days of measurements, the oxygen level could
vary between weeks. Since the tests with a specific OS usually extended over one or
two weeks, the kinetics of the following OS was measured in waters having a slightly
different level of dissolved oxygen.
That is, Figure 18B and the rate constants k would have been suitable as
comparative measures if the initial oxygen concentration had been equal between
the measurements. However, since the concentrations did differ (although only
moderately), Figure 18A is more applicable for comparing the oxygen reduction
kinetics of these experiments. Figure 18A shows that the reaction rate is marginally
faster for hydrazine than for EA. This trend was also found in the measurements
conducted in waters saturated with oxygen, having an initial oxygen concentration
of 10±2 ppm.
Although the oxygen reduction kinetics was only measured at two oxygen levels
(and at the same concentration ratio OS to oxygen), the results indicate that the
assumption of a unity rate order with respect to oxygen was within reason. This
is because the ranking of the apparent rate constants k ′ between OSs remained the
same at both oxygen concentrations (100 ppb and 10 ppm) and k ′ values increased
with the same proportions with an increase in oxygen concentration.
The dosing pump was found to cause an undesired ingress of air which has affected
the results of the kinetic measurements as discussed in Appendix C. Nevertheless,
the entered amount of air was considered quite insignificant and is believed to have
only slightly distorted the results by producing somewhat slower reaction kinetics.

Experimental sensor
The measurement data from the experimental sensor was analysed analogously to that
of the commercial sensor as described previously. An exemplary set of measurement
data from the experimental sensor is shown in Figure 19A. The figure shows that
the potential decreases as the oxygen scavengers were injected. The decrease in
potential was considered proportional to the decrease in oxygen concentration. Under
this assumption, the apparent rate constants k ′ were determined using the linear
relationship between the change in potential and time as shown in Figure 19B.
The oxygen reduction kinetics determined from the experimental sensor data are
shown as functions of the injected amount of OS in Figure 20. The apparent rate
constant k ′ displayed linear relationships with the OS concentration, as shown in
Figure 20B. This observation justifies the assumption of first rate orders with respect
to the OSs (i.e., the exponent m can be considered to have a value of 1). The rate
42

Figure 19: Measurement results from A) the Pt-Pt cell potential (EA, OS/DO ratio
8) and B) the corresponding linear regressions of the first four injections.

constants k were determined over the linear regions using the expression embedded
in Figure 20B and the rate constants are shown in Table 2.

Figure 20: The apparent rate constant k ′ as a function of A) the ratio of OS to DO


and B) the OS concentration. The initial oxygen concentration was about 80-90 ppb
prior to OS injection.

The results suggest that EA and HZ have faster reaction kinetics than DH, which
corresponds well with the results from the commercial sensor. However, the results
also suggest that the reaction kinetics of EA are faster than HZ, which deviates
from the results of previous studies [67, 66]. In comparison with the results from
the commercial sensor, the results here deviate by indicating faster kinetics for EA
than HZ in both Figures 20A and B. Therefore, the deviations cannot be attributed
to minor differences in the initial oxygen concentration nor to EA having kinetics
that benefit more from an increase in temperature. Instead, the deviation has to
originate from a difference in the measurement methods.
43

Comparing to the results from the commercial sensor, the experimental sensor
produces higher rate constants. Since the Pt-Pt sensor cannot discriminate between
redox reactions (poor selectivity), this observation suggests that the change in the
electrode potential may result from some complimentary redox reactions in addition
to the oxygen reduction reactions. These complimentary reactions are discussed
more thoroughly in the LPR and EIS result sections. Considering this deviation, the
thesis finds the results of the commercial sensor more representative.
In addition, the rate constants for the organic substitutes had increased to a
larger extent than for HZ as they were 57% and 118% higher for EA and DH, but
only 23% higher for HZ. This finding suggests that EA and DH may contribute with
complimentary reactions to a larger extent than HZ.
The measurements with a concentration ratio of 3 were repeated in waters with
an oxygen content of 10±2 ppm. The results indicate that the assumption of
unity reaction rate order with respect to oxygen was within reason. However, the
results corresponded poorly wit the 100 ppb measurements as DH and EA displayed
significantly faster reaction kinetics than HZ in the 10 ppm measurements. Moreover,
the results corresponded poorly with the data from the commercial sensor. These
findings demonstrate again that greater confidence should be put in the commercial
sensor.

Summary of the kinetic measurements


The results from the Pt potential and oxygen concentration measurements verified
semi first order reaction rates with respect to both oxygen and the oxygen scavengers.
More importantly, the results from the kinetics measurements demonstrated that the
reaction kinetics of EA are as fast as those of hydrazine, whereas the reaction rates
of DH are approximately one third of hydrazine. In addition, the results demonstrate
that the experimental sensor was able to fairly well estimate the reaction rates of
the three oxygen scavengers when the initial oxygen content was about 100 ppb.

Table 2: Reaction rate constants of hydrazine (HZ), erythorbic acid (EA) and diethyl-
hydroxylamine (DH) determined using a commercial sensor (CS) and experimental
sensor (ES) in secondary circuit water (228 ◦ C, 66 bar and buffered with NH3 to
pH25◦ C 9.6±0.2).

OS Rate equation CO2 (ppb) Method k (mM-1s-1) R2
CS 1.34±0.03 0.94
HZ kCHZ CO2 100-120
ES 1.65±0.03 0.97
CS 1.65±0.03 0.96
EA kCEA CO2 80-100
ES 2.59±0.05 0.94
CS 0.40±0.01 0.99
DH kCDH CO2 90-120
ES 0.87±0.04 0.85
44

7.2 Electrochemical measurements of carbon steel


Linear polarization resistance
The polarization resistance was measured for type 22K carbon steel in the presence
of the oxygen scavengers using the LPR method. The results from the LPR mea-
surements in secondary circuit water containing about 100 ppb of oxygen are shown
in Figure 21.

Figure 21: The polarization resistance for carbon steel decreased with A) an increasing
ratio of OS to oxygen and B) a decreasing concentration of oxygen. Repeated tests
with hydrazine after an extended exposure time did in part not reproduce the same
results.

Figure 21A shows that the polarization resistance decreased with an increasing
concentration of oxygen scavengers. The figure shows that the substitutes EA and
DH reduce the resistance to a greater extent than hydrazine. Correspondingly, the
polarization resistance was found to decrease with a declining oxygen concentration,
as shown in Figure 21B, regardless whether oxygen scavengers were present or not.
However, in the presence of EA and DH, the polarization resistance decreased further
than if the oxygen was removed by bubbling nitrogen. HZ did not reduce the
polarization resistance beyond that achieved by scavenging with nitrogen gas.
The LPR measurements were also conducted in water saturated with oxygen,
having an initial oxygen concentration of 10±2 ppm. However, the measurements were
limited to one injection series, where the OSs were administered at a concentration
ratio OS to oxygen of approximately 3. Nevertheless, similar trends of decreasing
polarization resistances were found in these measurements. Although the chemicals
did not differentiate as distinctively as in the measurements in 100 ppb oxygen, EA
and DH were again found to decrease the polarization resistance to a larger extent
than HZ.
These results may seem counter-intuitive since a lower polarization resistance
45

would mean a higher corrosion rate, but previous researchers [67, 24] have also
reported decreases of polarization resistance and it can therefore be considered an
expected result. Indeed, the decrease in polarization resistance with the addition of
hydrazine was found to correlate with an increased corrosion rate as confirmed by
weight loss measurements (100 h exposure time, same environment) [24].
Another explanation for the decrease in polarization resistance may be parallel
redox reactions taking place at the oxide-solution interface in addition to the corrosion
reactions as discussed in [67]. These complimentary reactions may include oxidation
reactions of the OS [67], such as intermediate reaction steps for the reduction of
oxygen. In addition, once most of the oxygen has been consumed, the oxidation
reactions of the OS may couple to the reduction of the outermost layer of the oxide
film, such as the reduction of FeOOH to Fe3 O4 which results in a more protective
outermost oxide layer of magnetite as discussed in [67]. Under this assumption, the
decrease in polarization resistance can be considered a measure of how much the OS
interacts with the oxide film in a beneficial manner, rather than an indication of
increased corrosion rates.
Since the measurements cannot discriminate between the corrosion reactions
and the suggested complimentary and ’beneficial’ redox reactions, both explanations
should be considered plausible. Nevertheless, the thesis finds more reason in the
explanation that the ’instantaneous’ decrease in polarization resistances would result
in a ’long-term’ increase in corrosion rate since the weight-loss measurements in
hydrazine [24] verified such a correlation. However, this reasoning means that the
thesis assumes that this correlation also extends to EA and DH, although it is a biased
assumption that requires confirmation with corresponding weight-loss experiments
for EA and DH.
Under these assumptions, the results indicate that the injections of EA, DH
and HZ cause instantaneous increases in the corrosion rates of carbon steel, which
may extend into the long-term but which are considered more likely to normalize
with time as the oxide chemistry transitions and becomes more protective. More
importantly, the results show that the injections of EA and DH increase the corrosion
rates to greater extents than HZ when administered at concentration ratios COS /CO2
between 1-10.
This is an unexpected outcome compared to the previous study [67], which showed
that EA and HZ affected the steel surface to similar extents whereas the effect of
DH was significantly lower. One reason for this deviation could be the temperature
of the experiments, which in [67] were room temperature and since the chemicals
can be considered to ’benefit’ from the temperatures to different degrees.
The polarization resistances measured in this work with LPR for blank solutions
for both levels of oxygen (0.76 MΩcm2 in 100±20 ppb oxygen and 0.42 MΩcm2 in
10±2 ppm) were in general on the order of one magnitude greater than those reported
by [67] and two orders of magnitude greater than those measured by [24]. Although
many attempts were made to attribute this unexpected deviation to procedural and
theoretical explanations, no definite source of error could be identified.
The deviation could not be explained by the characterization of the oxide film with
scanning electron microscopy (SEM) in Appendix D, since the oxide film thicknesses
46

and surface morphology were found to be similar to those depicted in [24]. Moreover,
the deviation could not be attributed to differences in environmental aspects since
the experimental circuit and environmental parameters were close to identical to
those specified in [24].
In addition, the uncompensated resistance Ruc was determined by EIS measure-
ments to be on the order of 5 kΩ and its effect on the measured polarization resistances
was thus considered insignificant. Therefore, this deviation was most likely caused
by a procedural or systematic issue in the measurements. One reason could be the
cell arrangement, which differed significantly in terms of the distance between the
WE and RE being about 50 mm in this study compared to the <50 μm in [24]. The
arrangement used in [24] could possibly eliminate the electrical interference caused
by the low-conductivity electrolyte.

Electrochemical impedance spectroscopy


The effect of oxygen scavengers on the electrochemical properties of carbon steel
grade 22K were measured with EIS. A typical EIS measurement in 100±20 ppb
oxygen concentration is shown in Figure 22. The Nyquist plots indicated only one
time constant appearing as a depressed semi-circle between frequencies of 1 Hz and
0.1 Hz, which corresponds to the processes at the oxide film-solution interface, as
discussed in [24, 67]. These processes include charge transfer and the charging of the
EDL which are represented in the equivalent circuit by the charge transfer resistance
Rct and the semi-capacitative CPEedl component, respectively.

Figure 22: EIS data and best-fit calculations of 22K carbon steel in water with 100
ppb oxygen shown as A) Nyquist and B) Bode plots.

In addition, the phase angle Bode plots hinted at the presence of two additional
processes. The first process is shown as a shoulder in the Figure 22B at a frequency
around 100 Hz, which could be related to the semiconducting properties of the oxide
as discussed in [24]. The second process can be barely recognized at the low frequency
end, which would be related to the diffusion processes within the oxide [24]. However,
47

since the processes were not considered recognizable enough from the spectra, they
were omitted in the modelling.
The impedance spectra were also measured for carbon steel in oxygen concen-
trations of 10±2 ppm, and a characteristic set of measurement data is shown in
Figure 23. The Nyquist spectra indicated the presence of a small semi-circle fused
together with a larger semi-circle, as shown in Figure 23A. Since the larger semi-circle
dominated the impedance spectra (but also due to issues in the fitting of the data to
the elaborated equivalent circuit embedded in Figure 23A), a single time constant
equivalent circuit was assumed in the curve fitting and modelling.

Figure 23: The impedance spectra of 22K carbon steel in 10 ppm oxygen indicated
two processes as shown in the A) Nyquist and B) Bode pots. The two processes were
drawn by individually fitting the data to an equivalent circuit with a single time
constant.

The first and ’smaller’ time-constant appears in the 10 ppm spectra between
frequencies of 1 Hz and 0.1 Hz and can again be attributed to the interface processes
between the solution and the oxide film as described previously. The second and
significantly ’larger’ time-constant peaks at a frequency of approximately 10 mHz
and can be considered affiliated with the diffusion processes within the oxide film
as described by [24, 67]. The response due to the diffusion in the oxide film is
represented in the equivalent circuit by the resistor Rf joined in parallel to the
semi-capacitative component CPEf .
The addition of oxygen scavengers changed the 100 ppb impedance response
as demonstrated in Figure 24. The figure shows that the impedance magnitude
diminished with the addition of hydrazine, which is an observation also found for EA
and DH. This result correlates well with the LPR measurements, which showed that
the polarization resistance decreased significantly with an increasing concentration
of oxygen scavengers. The drastic change in the impedance spectra can be again be
attributed to either an increased rate of corrosion reactions or due to complementary
redox reactions associated with the presence of oxygen scavengers as discussed in the
LPR results, or both.
48

Figure 24: The impedance responses diminished with the addition of hydrazine
shown in the A) Nyquist and B) Bode plots. The initial oxygen concentrations were
100±20 ppb.

The same trend of diminishing impedance spectra were found in the 10 ppm
measurements after the addition of oxygen scavengers. In the 10 ppm measurements,
the impedance response of the oxide film disappeared entirely and only the inter-
face processes were recognized in the presence of oxygen scavengers. This clearly
demonstrated a flaw in the measurement method since the oxygen scavengers are
considered incapable of instantly affecting the diffusion processes in the oxide film
under the limited time between the injection and the impedance measurement (about
45 minutes).
The charge transfer resistance and the capacitance of the EDL were determined
for the 100 ppb impedance spectra by fitting the data to an equivalent circuit with
a single time constant. The results from the best-fit calculations of the 100 ppb
spectra are compiled in Figure 25. Figure 25A shows that the addition of OS reduces
the charge transfer resistances, which correlates with the observations of previous
studies [67, 24] and can thus be considered an expected result. Considering hydrazine,
the charge transfer resistance declines linearly from 100 kΩ cm2 to 10 kΩ cm2 with
an increasing concentration ratio of OS to oxygen whereas for EA and DH the charge
transfer resistances decrease exponentially and converge at around 5 kΩ cm2 . That
is, at low to moderate concentration ratios ( COS /CO2 between 1 and 8) EA and
DH reduce the charge transfer resistance significantly more than HZ, whereas at
high concentration ratios (COS /CO2 above 8) EA and DH reduce the charge transfer
resistance only slightly more than HZ.
These trends correspond well with the observations from the the polarization
resistances shown in Figure 21. However, the impedances from the EIS measurements
that correspond to that determined by the LPR are one the order of one magnitude
lower (i.e.,the impedance at 1.4 mHz, which corresponds to a potential sweep from
-30 mV to +30 mV relative to the corrosion potential at a rate of 10 mV min-1 ).
That is, the results from the LPR and EIS measurements show similar ’qualitative’
49

Figure 25: The results from 100 ppb impedance spectra indicate that A) the charge
transfer resistance decreased with the OS amount and B) the EDL capacitance
increased with the OS amount.

trends but the results are quantitatively not comparable. On the other hand, the
Rct values correspond well with those measured by [67]. Therefore, the EIS results
can be considered more representative in quantitative terms than those of the LPR
measurements.
The electrical double layer capacity was found to increase semi-linearly with the
addition of HZ whereas the capacity increased irregularly for EA and DH, as shown
in Figure 25B. The increase in the EDL capacity can be attributed to the change in
solution chemistry with the addition of oxygen scavengers and a subsequent change
in the chemistry at the solution-electrode interface.

Summary of LPR and EIS results


The LPR and EIS results demonstrated that the addition of OSs results in an
immediate decrease of the polarization and charge transfer resistances. The decrease
in resistances were considered primarily to represent an increase in corrosion rates
and in part to be the effect of complementary redox reactions associated with the OSs.
Under this assumption, the results indicate that EA and DH increase the corrosion
rates to greater extents than HZ when administered at low to moderate concentration
ratios (values of COS /CO2 between 1 and 8), but only to a slightly greater extent
when administered at high concentration ratios (COS /CO2 values greater than 8).
That is, the results suggest that the effects of EA, DH and HZ can be considered
comparable at high concentration ratios.
50

8 Conclusions
This thesis set out to evaluate the feasibility of the two hydrazine substitutes ery-
thorbic acid and diethylhydroxylamine for use as oxygen scavengers in the secondary
circuit of PWRs. To this end, an experimental water circuit was devised to resem-
ble the Loviisa secondary circuit and the proposed substitutes were experimentally
evaluated with respect to their effectiveness to consume oxygen. In addition, the
experiments evaluated the effect of oxygen scavengers on the corrosion resistance of
22K carbon steel, which is one of the most abundant structural materials in Loviisa
and corresponding WWER-440 secondary circuits.
The results from the measurements of the oxygen reduction kinetics showed that
erythorbic acid consume oxygen as quickly and efficiently as hydrazine. In addition,
diethylhydroxylamine was found to consume oxygen at rates of approximately one
third of hydrazine and erythorbic acid. However, the injections of oxygen scavengers
was found to increase the electrolytic conductivity as shown in Appendix E. Never-
theless, the results indicate that erythorbic acid and diethylhydroxylamine can offer
safe approaches for oxygen removal in PWR secondary circuits, provided that the
generation of decomposition products are accounted for since they may increase the
electrolytic conductivity and reduce the pH of secondary waters.
The results from the LPR and EIS measurements of carbon steel grade 22K
demonstrated that the addition of oxygen scavengers resulted in an immediate de-
crease of the polarization and charge transfer resistances. The decrease in resistances
were considered primarily to represent an increase in corrosion rates and in part to
be the effect of complementary redox reactions associated with the OSs. Under this
assumption, the results indicate that EA and DH increase the corrosion rates to
comparable extents when administered at high concentration ratios (COS /CO2 values
greater than 8).
In this study, carbon steel was exposed to the oxygen scavengers for only a few
hours and the measured increases in corrosion rates can thus be considered to be
temporary. With time, the corrosion rates are expected to decrease and normalize as
the oxide chemistry transitions and the oxide becomes more protective. Nevertheless,
this assumption should ideally be experimentally evaluated and field-tested in future
work in order to determine the long-term effects of the oxygen scavengers on the
corrosion resistance of secondary circuit materials. A suggested duration for this
long-term experiment could be one year, which would correspond to the full duration
between the annual maintenances.
In addition, the effort of future work should be put on comparing additional
hydrazine substitutes, including carbohydrazide and methylethylketoxime. As for
suggestions on improvements of the experimental methodology, future studies should
ensure that the initial oxygen concentration remains identical between measure-
ments and emphasize the results gathered with the commercial sensor and from EIS
measurements. Finally, the method for determining the oxygen reduction kinetics
using the experimental sensor could be tailored better for the organic substitutes by
increasing the sensor selectivity or modifying the analysis calculus.
51

References
[1] Ministry of Economic Affairs and Employment. Government report on the
National Energy and Climate Strategy for 2030. Tech. rep. Helsinki: Ministry
of Economic Affairs and Employment, 2017, p. 121. url: http://urn.fi/
URN:ISBN:978-952-327-199-9.
[2] Supplies and total consumption of electricity, GWh. Tech. rep. Statistics
Finland, 2019. url: http://pxnet2.stat.fi/PXWeb/pxweb/en/StatFin/.
[3] Anni Lassila and Jarno Hartikainen. Olkiluoto 3:n käynnistyminen myöhästyy
jälleen, Areva joutuu maksamaan korvauksia. Helsinki, 2019. url: https:
//www.hs.fi/talous/art-2000006176296.html.
[4] Anna Juvonen. “Fennovoima: Hanhikivi 1 kaupalliseen käyttöön 2028
26.12.2018”. In: Kauppalehti (2018). url: https://www.kauppalehti.
fi/uutiset/fennovoima- hanhikivi- 1- kaupalliseen- kayttoon- 2028/
c666e0fa-8fbc-46b5-a11a-d76d0b9547e2.
[5] Ministry of Trade and Industry. Decision on operating license of Loviisa 1
and 2. Tech. rep. Helsinki: Ministry of Trade, Industry (2019: the Ministry of
Economic Affairs, and Employment), 2007. url: https://tem.fi/loviisa-
1-ja-2-kayttolupa-2007-2030.
[6] Ministry of Economic Affairs and Employment. Decision on operating license
of Olkiluoto 1 and 2. Tech. rep. Ministry of Economic Affairs and Employment,
2018. url: https://tem.fi/en/loviisa-1-and-2-operating-licence.
[7] Safety assessment 1 (107) PSR2015 Appendix 1 Nuclear Reactor Regulation
An assessment by the Radiation and Nuclear Safety Authority on the periodic
safety review of Loviisa NPP. Tech. rep. Helsinki: STUK - Radiation and
Nuclear Safety Authority, 2015. url: https://www.stuk.fi/documents/
88234/254201/1694438-stuks-assessment-on-loviisa-npp-periodic-
safety- review- safety- assessment.pdf/05543da6- 3dd2- 2a79- f8b8-
7ddb9ba4d8bd.
[8] Hirotaka Kawamura et al. “PWR secondary water chemistry guidelines in
Japan - Purpose and technical background”. In: Progress in Nuclear Energy
114 (2019), pp. 121–137. issn: 01491970. doi: 10.1016/j.pnucene.2019.
01.027.
[9] European Chemical Agency. Inclusion of Substances of Very High Concern
in the Candidate List for eventual inclusion in Annex XIV. 2011. url:
https://echa.europa.eu/documents/10162/c5b972a9-f57f-4fd5-8177-
04b4e46c5e93.
[10] Työterveyslaitos. OVA-ohje: Hydratsiini. 2017. url: https://www.ttl.fi/
ova/hydratsi.pdf.
52

[11] Environment Canada. Canadian Environmental Protection Act, 1999 Federal


Environmental Quality Guidelines - Hydrazine. Tech. rep. 2013. url: http:
//www.ec.gc.ca/ese- ees/2FE86525- C439- 4BC8- B20A- 315AF1CB6F6A/
FEQG_TBBPA_EN.pdf.
[12] Mersiha Spahic et al. “Control and management of the chemical risk linked
with hydrazine hydrate storage, unloading and injection across french nuclear
fleet”. In: The 18th International Conference on Water Chemistry in Nuclear
Reactor Systems. 2012. url: http://www.inchem.org/documents/jecfa/
jecmono/v28je03.htm.
[13] Harbhajan S Mahal and Katta Venkateswarlu. Water chemistry studies - if
use of hydrazine for scavenging dissolved oxygen in high temperature water - a
review. Tech. rep. Bombay: Government of India Atomic Energy Commission,
1971, pp. 3–4.
[14] Iva Betova, Martin Bojinov, and Timo Saario. Hydrazine replacement in
nuclear power plants – alternative substances and techniques, VTR-R-03426-
16. Tech. rep. Espoo: VTT Technical Research Centre of Finland, 2016,
p. 30.
[15] Paul A. Breeze. Nuclear power. Academic Press, 2016. isbn: 9780128095126.
url: https://learning.oreilly.com/library/view/nuclear- power/
9780128095126/?ar.
[16] K. Varga, E.H. Deák, and J. Schunk. “Corrosion issues in water-cooled
water-moderated energetic reactor (WWER) systems”. In: Nuclear Corrosion
Science and Engineering. Elsevier, 2012, pp. 634–678. isbn: 9781845697655.
doi: 10.1533/9780857095343.5.634.
[17] Nuclear Power Reactors in the World. Vienna: International Atomic Energy
Agency, 2017, p. 74. isbn: 9789201040176.
[18] John. N. Lillington. The Future of Nuclear Power. Dorchester: Elsevier, 2004.
Chap. Present Ge, pp. 3–12. isbn: 9780080444895. doi: 10.1016/B978-0-
08-044489-5.X5000-1.
[19] L. Papp and J. Vacek. “WWER steam generators”. In: Steam Generators
for Nuclear Power Plants. Elsevier, 2017, pp. 107–124. isbn: 9780081009284.
doi: 10.1016/B978-0-08-100894-2.00007-8.
[20] Marcin Karol Rowinski, Timothy John White, and Jiyun Zhao. “Small and
Medium sized Reactors (SMR): A review of technology”. In: Renewable and
Sustainable Energy Reviews 44 (2015), pp. 643–656. issn: 13640321. doi:
10.1016/j.rser.2015.01.006.
[21] Jose San and Niabot. Kernkraftverk mit Druckwasserreaktor (ohne Turm).png.
2009. url: https://de.wikipedia.org/wiki/Datei:Kernkraftwerk_
mit_Druckwasserreaktor_(ohne_Turm).png.
53

[22] Ritva Korhonen and Ossi Hietanen. Erosion Corrosion of Parallel Feed
Water Discharge Lines At the Loviisa Vver 440. Tech. rep. Vantaa: IVO
international, 1994, pp. 75–85. url: https://inis.iaea.org/search/
search.aspx?orig_q=RN:28008731.
[23] M. Huhtinen et al. Voimalaitostekniikka. 1st editio. Helsinki: Opetushallitus,
2008. isbn: 978-952-13-3476-4.
[24] Sari Järvimäki et al. “Effect of hydrazine on general corrosion of carbon
and low-alloyed steels in pressurized water reactor secondary side water”. In:
Nuclear Engineering and Design 295 (2015), pp. 106–115. issn: 00295493.
doi: 10.1016/j.nucengdes.2015.08.033.
[25] Private communication on 29.11.2019 with Head of chemistry S. Buddas at
Fortum, Loviisa NPP.
[26] Yi Xie and Jinsuo Zhang. “Corrosion and deposition on the secondary circuit
of steam generators”. In: Journal of Nuclear Science and Technology 53.10
(2016), pp. 1455–1466. issn: 0022-3131. doi: 10.1080/00223131.2016.
1152923.
[27] G. Saji et al. “Fundamental Mechanisms of Component Degradation by
Corrosion in Nuclear Power Plants of Russian Design”. In: Volume 1: Plant
Operations, Maintenance, Installations and Life Cycle; Component Reliability
and Materials Issues; Advanced Applications of Nuclear Technology; Codes,
Standards, Licensing and Regulato. Vol. 1. ASME, 2008, pp. 499–518. isbn:
0-7918-3820-X. doi: 10.1115/ICONE16-48308.
[28] Vladimír Slugeň. Safety of VVER-440 Reactors. London: Springer London,
2011, pp. 1–178. isbn: 978-1-84996-419-7. doi: 10.1007/978-1-84996-420-
3.
[29] S. Savolainen and B. Elsing. “Feed water distribution pipe replacement
at loviisa npp”. In: Third international seminar on horizontal steam gen-
erators. 1994, p. 16. url: https : / / inis . iaea . org / collection /
NCLCollectionStore/_Public/31/018/31018414.pdf.
[30] Antoaneta E. Stefanova, Rositsa V. Gencheva, and Pavlin P. Groudev. Use of
main loop isolating valves (GZZS) in VVER 440. Tech. rep. Sofia: Institute
for Nuclear Research and Nuclear Energy, Bulgaria, 2002. url: https :
//inis.iaea.org/collection/NCLCollectionStore/_Public/34/043/
34043603.pdf.
[31] Edward McCafferty. Introduction to Corrosion Science. New York, NY:
Springer New York, 2010. isbn: 978-1-4419-0454-6. doi: 10.1007/978-1-
4419-0455-3.
[32] F. King. “General corrosion in nuclear reactor components and nuclear
waste disposal systems”. In: Nuclear Corrosion Science and Engineering.
Elsevier, 2012. Chap. 4, pp. 77–103. isbn: 9781845697655. doi: 10.1533/
9780857095343.2.77.
54

[33] A.M. Olmedo, R. Bordoni, and M. Strack. “Characterization of the Oxide


Films Grown at 260C in a Simulated Secondary Coolant of a Nuclear Reactor”.
In: Procedia Materials Science 1.54 11 (2012), pp. 528–534. issn: 22118128.
doi: 10.1016/j.mspro.2012.06.071.
[34] S. Delaunay et al. “Formation and Deposition of Iron Oxides on Stainless
Steel and Carbon Steel in Conditions of Secondary Circuits of Pressurized
Water Reactors”. In: CORROSION 67.1 (2011), pp. 015003–1–015003–10.
issn: 0010-9312. doi: 10.5006/1.3546849.
[35] S. Nasrazadani et al. “Characterization of oxides on FAC susceptible small-
bore carbon steel piping of a power plant”. In: International Journal of
Pressure Vessels and Piping 86.12 (2009), pp. 845–852. issn: 03080161. doi:
10.1016/j.ijpvp.2009.10.003.
[36] Masaichi Nagayama and Morris Cohen. “The Anodic Oxidation of Iron in a
Neutral Solution I. The Nature and Composition of the Passive Film”. In:
Journal of The Electrochemical Society 109.9 (1962), p. 787. doi: 10.1149/1.
2425555.
[37] J. Robertson. “The mechanism of high temperature aqueous corrosion of
stainless steels”. In: Corrosion Science 32.4 (1991), pp. 443–465. issn:
0010938X. doi: 10.1016/0010-938X(91)90125-9.
[38] R Winkler, F Huettner, and F Michel. “Reduction of corrosion rates in the
primary circuit of pressurized water reactors in order to limit radioactive
deposits”. In: VGB Kraftwerkstechnik (1989). url: https://inis.iaea.
org/search/search.aspx?orig_q=RN:20061153.
[39] B. Stellwag. “The mechanism of oxide film formation on austenitic stainless
steels in high temperature water”. In: Corrosion Science 40.2-3 (1998),
pp. 337–370. issn: 0010938X. doi: 10.1016/S0010-938X(97)00140-6.
[40] B. Beverskog et al. “A mixed-conduction model for oxide films on Fe, Cr and
Fe–Cr alloys in high-temperature aqueous electrolytes—II. Adaptation and
justification of the model”. In: Corrosion Science 44.9 (2002), pp. 1923–1940.
issn: 0010938X. doi: 10.1016/S0010-938X(02)00009-4.
[41] Paul L. Huoston. Chemical Kinetics and Reaction Dynamics. Dordrecht:
Springer Netherlands, 2006. isbn: 978-1-4020-4546-2. doi: 10.1007/978-1-
4020-4547-9.
[42] Cynthia G. Zoski. Handbook of Electrochemistry. Elsevier, 2007. Chap. 1,
p. 17. isbn: 978-0-08-046930-0.
[43] J. S. Kim, E. A. Cho, and H. S. Kwon. “Photoelectrochemical study on the
passive film on Fe”. In: Corrosion Science 43.8 (2001), pp. 1403–1415. issn:
0010938X. doi: 10.1016/S0010-938X(00)00159-1.
[44] Guangze Yang et al. “A review on clogging of recirculating steam generators
in Pressurized-Water Reactors”. In: Progress in Nuclear Energy 97 (2017),
pp. 182–196. issn: 01491970. doi: 10.1016/j.pnucene.2017.01.010.
55

[45] Suat Odar and Francis Nordmann. PWR and VVER secondary system
water chemistry report. Tech. rep. Skultuna: Advanced Nuclear Technology
International, 2010. url: https://www.antinternational.com/docs/
samples/CCC/02/04-sswc_sample_report.pdf.
[46] Mikko Vepsäläinen and Timo Saario. Magnetite dissolution and deposition in
NPP secondary circuit, VTT-R-09735-10. Tech. rep. Espoo: VTT Technical
Research Centre of Finland, 2010. url: https : / / www . vtt . fi / inf /
julkaisut/muut/2010/VTT-R-09735-10.pdf.
[47] R.B Dooley and V.K Chexal. “Flow-accelerated corrosion of pressure vessels in
fossil plants”. In: International Journal of Pressure Vessels and Piping 77.2-3
(2000), pp. 85–90. issn: 03080161. doi: 10.1016/S0308-0161(99)00087-3.
[48] R Barry Dooley. “Flow-Accelerated Corrosion in Fossil and Combined Cycle
/ HRSG Plants”. In: Power Plant Chemistry 10.2 (2008), pp. 68–89.
[49] U. R. Evans. The corrosion and oxidation of metals. London: Edward Arnold,
1971, p. 128. isbn: 978-0713120547.
[50] Shunsuke Uchida et al. “Evaluation of flow accelerated corrosion by coupled
analysis of corrosion and flow dynamics. Relationship of oxide film thick-
ness, hematite/magnetite ratio, ECP and wall thinning rate”. In: Nuclear
Engineering and Design 241.11 (2011), pp. 4585–4593. issn: 00295493. doi:
10.1016/j.nucengdes.2010.09.018.
[51] M. Halin et al. “Influence of feed water distribution pipe replacement on the
water chemistry in the steam generator at Loviisa NPP”. In: 1998. url:
https://inis.iaea.org/collection/NCLCollectionStore/_Public/31/
018/31018414.pdf.
[52] Jonathan J Morrison et al. “Effect of Water Chemistry on Corrosion of
Stainless Steel and Deposition of Corrosion Products in High Temperature
High Pressure Water”. In: Nuclear Plant Chemistry (2012). url: https:
/ / inis . iaea . org / Search / searchsinglerecord . aspx ? recordsFor =
SingleRecord&RN=46071844.
[53] N. B. Trunov et al. “Present state of the problem of managing the service
life of steam generators used at nuclear power stations equipped with VVER
reactors”. In: Thermal Engineering 58.3 (2011), pp. 184–189. issn: 0040-6015.
doi: 10.1134/S0040601511030141.
[54] Francis Nordmann, Suat Odar, and Dewey Rochester. “Issues and remedies
for secondary system of PWR/WER”. In: Revue Générale Nucléaire 6 (2012),
pp. 45–55. issn: 0335-5004. doi: 10.1051/rgn/20126045.
[55] E. McCafferty. “Corrosion Behavior of Laser-Surface Melted and Laser-Surface
Alloyed Steels”. In: Journal of The Electrochemical Society 133.6 (1986),
p. 1090. issn: 00134651. doi: 10.1149/1.2108792.
[56] H. P. Leckie and H. H. Uhlig. “Environmental Factors Affecting the Critical
Potential for Pitting in 18–8 Stainless Steel”. In: Journal of The Electrochemi-
cal Society 113.12 (1966), p. 1262. issn: 00134651. doi: 10.1149/1.2423801.
56

[57] D You et al. “Experimental study of concentrated solutions containing sodium


and chloride pollutants in sg flow restricted areas”. In: International confer-
ence on water chemistry in nuclear reactors systems - operation optimisation
and new developments. International Atomic Energy Agency, 2002, p. 1. url:
https://www.osti.gov/etdeweb/servlets/purl/20386586.
[58] S.M. Peyghambarzadeh, A Vatani, and M Jamialahmadi. “Influences of
bubble formation on different types of heat exchanger fouling”. In: Applied
Thermal Engineering 50.1 (2013), pp. 848–856. issn: 13594311. doi: 10.
1016/j.applthermaleng.2012.07.015.
[59] D.R. Diercks, W.J. Shack, and J. Muscara. “Overview of steam generator tube
degradation and integrity issues”. In: Nuclear Engineering and Design 194.1
(1999), pp. 19–30. issn: 00295493. doi: 10.1016/S0029-5493(99)00167-3.
[60] V D Bergunker. “Integrity of heat-exchange tubes of vertical and horizontal
steam generators”. In: 7th International Seminar on Horizontal Steam Gen-
erators. Podolsk, 2006, pp. 70–87. url: http://www.gidropress.podolsk.
ru/files/proceedings/seminar7/documents/f53.pdf.
[61] N. B. Trunov et al. “Steam generators – horizontal or vertical (which type
should be used in nuclear power plants with VVER?)” In: Atomic Energy 105.3
(2008), pp. 165–174. issn: 1063-4258. doi: 10.1007/s10512-008-9090-1.
[62] Annex XV dossier - Identification of hydrazine as SVHC. Tech. rep. European
chemical agency, 2011, p. 7. url: https://echa.europa.eu/documents/
10162/2f29c6b2-e043-4a13-927e-c25a04cd4abe.
[63] Kazushige ISHIDA et al. “Hydrogen and Hydrazine Co-injection to Mitigate
Stress Corrosion Cracking of Structural Materials in Boiling Water Reactors,
(VI) The Effect of Ammonia on Intergranular Stress Corrosion Cracking”. In:
Journal of Nuclear Science and Technology 44.12 (2007), pp. 1550–1556. issn:
0022-3131. doi: 10.1080/18811248.2007.9711405.
[64] N. L. Dickinson. “An experimental investigation of hydrazine-oxygen reaction
ratios in boiler feed water”. In: Proceedings of the American Power Con-
ference Volume XIX. 1957, p. 692. url: https://ci.nii.ac.jp/naid/
10016732390/.
[65] D. Feron. “Hydrazine, oxygen and materials interactions in a PWR secondary
system”. In: JAIF INT. Conf. on Water Chemistry in Nuclear Power Plants.
Tokyo, 1988, p. 636. url: https://inis.iaea.org/search/search.aspx?
orig_q=RN:20037831.
[66] Iva Betova, Martin Bojinov, and Timo Saario. Kinetics of oxygen removal by
hydrazine alternatives in conditions of steam generator preservation during out-
age, VTT-R-05209-17. Tech. rep. Espoo: VTT Technical Research Centre of
Finland, 2017. url: https://cris.vtt.fi/en/publications/kinetics-
of-oxygen-removal-by-hydrazine-alternatives-in-condition.
57

[67] Iva Betova, Martin Bojinov, and Timo Saario. Hydrazine alternatives as
oxygen scavengers and passivation agents in steam generator preservation
conditions, VTT-R-01908-18. Tech. rep. Espoo: VTT Technical Research
Centre of Finland, 2018.
[68] A Banweg, D G Wiltsey, and B N Nimry. “Carbohydrazide - A Hydrazine
Replacement: 10 Years of Utility Experience”. In: EPRI International
Conference on Cycle Chemistry in Fossil Plants. Nalco Chemical Company.
Baltimore, 1991. url: https://vdocuments.site/carbohydrazide- a-
hydrazine-replacement-10-years-a-hydrazine-replacement.html.
[69] Kazutoshi Fujiwara et al. “Evaluation of Thermal Decomposition Rate of
Carbohydrazide And Its Reducing Effect on Carbon Steel Corrosion”. In:
NACE International corrosion conference. Tokyo: NACE International, 1997.
url: https://www.onepetro.org/conference-paper/NACE-97089.
[70] N. N. Greenwood and A. Earnshaw. Chemistry of the Elements. 2nd. Elsevier,
1997, p. 310. isbn: 978-0-0805-0109-3.
[71] Safety data sheet on 1H-pyrrole-2-carbohydrazide, 97% from Thermo Fisher
Scientific. Loughborough, 2019.
[72] R. Walker. Erythorbic acid and its sodium salt. Guildford, 1991. url:
www.inchem.org/documents/jecfa/jecmono/v28je03.htm.
[73] Eunok Choe and David B. Min. “Mechanisms of Antioxidants in the Oxidation
of Foods”. In: Comprehensive Reviews in Food Science and Food Safety 8.4
(2009), pp. 345–358. issn: 15414337. doi: 10.1111/j.1541-4337.2009.
00085.x.
[74] Aishee Dey and Sudarsan Neogi. “Oxygen scavengers for food packaging
applications: A review”. In: Trends in Food Science & Technology 90.August
2018 (2019), pp. 26–34. issn: 09242244. doi: 10.1016/j.tifs.2019.05.
013.
[75] R.J. Wilson, A.E. Beezer, and J.C. Mitchell. “A kinetic study of the oxidation
of L-ascorbic acid (vitamin C) in solution using an isothermal microcalorime-
ter”. In: Thermochimica Acta 264.C (1995), pp. 27–40. issn: 00406031. doi:
10.1016/0040-6031(95)02373-A.
[76] John D Zupanovich. Oxidation And Degradation Products Of Common Oxygen
Scavengers. Tech. rep. Association of Water Technologies, 2002. url: https:
//www.awt.org/pub/0149322F-0C20-5CEC-AE62-1E826AF61A4C.
[77] Arkema S.A. Selection Guide for Oxygen Scavengers. Philadelphia, 2001.
url: http : / / www . subsportplus . eu / wp - content / uploads / 2012 / 05 /
alternative-to-hydrtazine-USA-2001-k.pdf.
[78] USFDA. Sec. 182.3041 Erythorbic acid. 2019. url: https : / / www .
accessdata . fda . gov / scripts / cdrh / cfdocs / cfcfr / CFRSearch . cfm ?
fr=182.3041.
58

[79] N,N-Diethylhydroxylamine. Tech. rep. Pubchem, 2019. url: https : / /


pubchem.ncbi.nlm.nih.gov/compound/19463.
[80] Raphaël Lebeuf, Véronique Nardello-Rataj, and Jean-Marie Aubry. “Dramatic
synergistic effects between hydroquinone and resorcinol derivatives for the
organocatalyzed reduction of dioxygen by diethylhydroxylamine”. In: Chem.
Commun. 50.7 (2014), pp. 866–868. issn: 1359-7345. doi: 10 . 1039 /
C3CC47261B.
[81] Sigma-Aldrich. Safety data sheet on N,N-Diethylhydroxylamine. 2015. url:
https://www.sigmaaldrich.com/catalog/product/aldrich/471593.
[82] Justification Document for the Selection of a CoRAP Substance. Tech. rep.
Swedish Chemicals Agency KEMI, 2018. url: https://echa.europa.eu/
documents/10162/6dd43077-f450-c22f-fe6e-c2bc2612ca3c.
[83] J. A. von Fraunhofer and C. H. Banks. Potentiostat and its applications.
London: Butterworth & Co, 1972, p. 254.
[84] Mari Lundström. Lecture 5 in CHEM-E6185 Applied electrochemistry and
corrosion. Espoo, 2019.
[85] Jacob Fraden and Lawrence G. Rubin. “AIP Handbook of Modern Sensors”.
In: Physics Today 47.6 (1994), pp. 74–75. issn: 0031-9228. doi: 10.1063/1.
2808535.
[86] Operator’s manual for the 3660 Analyzer for Oxygen or Ozone. Neuchâtel,
1998.
[87] The Dissolved Oxygen Handbook. 2009. url: https://www.fondriest.com/
pdf/ysi_do_handbook.pdf.
[88] P. P. L. Regtien et al. Measurement science for engineers. 1st. Twente:
Elsevier Science and Technology, 2004, p. 260. isbn: 9780080536019.
[89] Tiina Ikäläinen, Timo Saario, and Konsta Sipilä. Alternatives for hydrazine
in PWRs. Tech. rep. Espoo: VTT Technical Research Centre of Finland,
2019.
[90] Farzin Arjmand and Lefu Zhang. “Solution Resistivity, Ohmic Drop and
Oxygen Reduction Rate at High Temperature Pressurized Water”. In: Elec-
trochimica Acta 216.October (2016), pp. 438–448. issn: 00134686. doi:
10.1016/j.electacta.2016.08.136.
[91] D. C. Silverman. “Practical Corrosion Prediction Using Electrochemical
Techniques”. In: Uhlig’s Corrosion Handbook. Hoboken, NJ, USA: John
Wiley & Sons, Inc., 2011. Chap. 85, pp. 1129–1166. isbn: 9780470080320.
doi: 10.1002/9780470872864.ch85.
[92] Florian Mansfeld. “Recording and Analysis of AC Impedance Data for
Corrosion Studies”. In: CORROSION 37.5 (1981), pp. 301–307. issn: 0010-
9312. doi: 10.5006/1.3621688.
59

[93] Stephen Fletcher. “Tables of Degenerate Electrical Networks for Use in the
Equivalent-Circuit Analysis of Electrochemical Systems”. In: Journal of
The Electrochemical Society 141.7 (1994), p. 1823. issn: 00134651. doi:
10.1149/1.2055011.
[94] Mari Lundström. Lecture 8 in CHEM-E6185 Applied electrochemistry and
corrosion. Espoo, 2019.
[95] Martin Bojinov et al. “Characterisation of the oxide layer on carbon steel
during hot conditioning of primary heat transport systems in heavy-water
reactors”. In: Corrosion Science 51.5 (2009), pp. 1146–1156. issn: 0010938X.
doi: 10.1016/j.corsci.2009.02.006.
[96] Martin Bojinov et al. “Effect of Chloride on the Oxides on Low-Alloyed Steel
in Conditions of a Light Water Reactor Pressure Vessel Cladding Flaw”. In:
Journal of The Electrochemical Society 161.4 (2014), pp. C177–C187. issn:
0013-4651. doi: 10.1149/2.004404jes.
[97] S. Kochowski and K. Nitsch. “Description of the frequency behaviour of
metal–SiO2–GaAs structure characteristics by electrical equivalent circuit
with constant phase element”. In: Thin Solid Films 415.1-2 (2002), pp. 133–
137. issn: 00406090. doi: 10.1016/S0040-6090(02)00506-0.
[98] Zview 2 software Help documentation Equivalent Circuits - Circuit Elements.
[99] Richard S. Greeley et al. “Electromotive Force Studies in Aqueous Solutions
at Elevated Temperatures. I. The Standard Potential of the Silver-Silver
Chloride Electrode 1”. In: The Journal of Physical Chemistry 64.5 (1960),
pp. 652–657. issn: 0022-3654. doi: 10.1021/j100834a031.
[100] R. B. Dean and W. J. Dixon. “Simplified Statistics for Small Numbers of
Observations”. In: Analytical Chemistry 23.4 (1951), pp. 636–638. issn:
0003-2700. doi: 10.1021/ac60052a025.
[101] Martin Bojinov. Private communication on 18.12.2019 with Prof. Martin
Bojinov and colleagues at University of Chemical Technology and Metallurgy,
Bulgaria, including calculation instructions "Analysis of E-t curves". Sofia.
60

Appendices
A Experimental sensor calculus
Since the experimental oxygen sensor was found to overestimate the reaction rates for
oxygen reduction in waters with 100 ppb oxygen concentrations, a new method for
determining the reaction rates was developed. This appendix outlines this approach,
which was developed by Prof. Martin Bojinov and colleagues [101].
The method was derived from the oxygen reduction reaction for hydrazine, shown
in (A1). This reaction can be considered to occur on the Pt-electrode via the two
half-cell reactions for oxidation of hydrazine and the reduction of oxygen shown in
equations (A2) and (A3), respectively.

k
total reaction: N2 H4 + O2 −−−→ 2 H2 O + N2 (A1)
oxidation: N2 H4 −−→ N2 + 4 H+ + 4 e− (A2)
reduction: O2 + 2 H2 O + 4 e− −−→ 4 OH− (A3)
The reaction rate was assumed to be first order with respect to oxygen and
hydrazine, as shown in equation (A4). By substituting with the formal rate constants
k ′ and k ′′ and integrating over the time interval t = 0 and t = t and corresponding
concentrations CO2 (t = 0), CO2 (t = t), CN2 H4 (t = 0) and CN2 H4 (t = t), the reaction
rate expressions were converted to equations (A5) and (A6).

dCO2 dCN2 H4
− =− = kCO2 CN2 H4 (A4)
dt dt
k′ =kCN2 H4 CO (0)
====== ⇒ ln 2 = k′t (A5)
integration CO2 (t)
k′′ =kCO2 CN H (0)
======⇒ ln 2 4 = k ′′ t (A6)
integration CN2 H4 (t)
Since no net current flows at the open circuit potential, the sum of the oxidation
and reduction currents is equal to zero as shown in equation (A7). Here the kinetics
of the two half-cell reactions were assumed to be dictated by charge transfer.

4αo F 4αr F
(︃ )︃ (︃ )︃
4F ko CN2 H4 (t)exp E(t) − 4F kr CO2 (t)exp E(t) = 0 (A7)
RT RT
where:
F = Faraday constant, 96485 C mol−1
k ′′ , k ′ = oxidation and reduction rate constants in s−1
Ci (t) = concentration of species i at time t in mol l−1
αi = transfer coefficient for ox. and red. reactions
R = gas constant, 8.314 Jmol−1 K −1
T = absolute temperature in K
E(t) = measured cell potential at time t in V
61

By taking natural logarithms and using a bit of algebra, the expression can be
simplified to equation (A8).

4(αo + αr )F
lnkr CO2 − lnko CN2 H4 = E(t) (A8)
RT
Since the current sum equals to zero regardless of time, the difference between
the current sums at two given times should also be zero. The difference between the
current sums between times t = 0 and t = t is shown in equation (A9).

CO2 (0) CN H (0) 4(αo + αr )F


ln − ln 2 4 = (E(t) − E(o)) (A9)
CO2 (t) CN2 H4 (t) RT
By substituting with the apparent rate constants k ′ and k ′′ (shown in equa-
tions (A5) and (A6)) and solving for the potential difference, the expression in
equation (A10) was formed.
RT
E(0) − E(t) = (k ′ − k ′′ )t (A10)
4(αo + αr )F
By then assuming that αo «αr and k ′′ «k ′ (which were validated by comparing to
experimental data), the expression can be simplified to equation (A11). The value
αr was determined to be 0.043 using Pt electrode oxygen reduction curves measured
in 11.5 ppm ammonia at a temperature of 220 ◦ C [90].
RT ′
E(0) − E(t) = kt (A11)
4αr F
The same method was applied to erythorbic acid and diethylhydroxylamine using
the oxidation reactions shown in equations (A12) and (A13), respectively.

EA oxidation: 2 C6 H8 O6 −−→ 2 C6 H6 O6 + 4 H+ + 4 e− (A12)


DH oxidation: 2 (C2 H5 )2 NOH −−→ 2 C2 H5 NOC2 H4 + 4 H+ + 4 e− (A13)

Using the same procedure as described above, the relationship for the change in
open circuit potential shown in equation (A11) was found to be applicable also to
erythorbic acid and diethylhydroxylamine. Equation (A11) was used to determine the
apparent rate constants k ′ and to estimate the rate constant k from the measurement
data in the thesis result section.
62

B Pump rate correction


During the initial measurements for determining the kinetics of oxygen reduction
there were indications that the dosing pump supplied the oxygen scavenger solution
at a faster rate than the desired rate promted into the pump user interface. To verify
these indications, measurements were carried out to determine the actual pump rate.
The purpose of this appendix is to describe the measurements used to determine the
the actual pump rate.
The pump rate of the Shimadzu LC-20AT chromatography pump was determined
by injecting distilled water drawn from a 25 ml graduated cylinder and measuring
the time at intervals of a certain volume of water. The desired pumping rate was set
to values between 2.0 ml min-1 and 4.0 ml min-1 .
The measurement results are shown in Figure B1. The results show that the
actual pump rate is approximately 25% greater than the desired pump rate. The
actual pump rates were used in the results analysis for calculating new concentration
ratios COS /CO2 .

Figure B1: Results from the pump rate measurements. The slopes from A) are
shown as separate values in B). The actual pump rate is approximately 25% larger
than the desired.
63

C Injection tests with blank solutions


In some oxygen concentration measurements, the commercial oxygen sensor would
indicate a sudden increase of oxygen shortly before the oxygen concentration started
to decline, as shown in Figure C1A. That is, undesired oxygen had entered the circuit
during the time when the oxygen scavengers were injected. This indicated that the
dosing pump or the pipe line between the dosing pump and the injection point was
leaking, thus allowing external oxygen to enter the circuit from an ingress of air.
Since this ingress of air could affect the results of the kinetic measurements, an
injection test with deoxygenated and deionized water containing no OS was carried
out to determine the amount of air ingress and its significance. This appendix
describes this injection test and discusses the measurement error caused by the
undesired ingress of air.
The ingress of air was most noticeable during the very first injection if the pipeline
between the dosing pump and the injection point had not been prefilled. That is, the
intermediate pipeline seemed to allow some air to enter the stagnant solution which
would increase the oxygen concentration of the initial amount of injected solution,
which would then appear as a peak in the measurements.
When the intermediate pipeline was prefilled with the OS solution, oxygen peaks
would not appear in the measurements. However, the ingress of air could still occur
in a continuous manner but go unnoticed since the OS would instantaneously remove
the entered oxygen. In order to determine the amount of continuous oxygen ingress,
deoxygenated and deionized water containing no OS (i.e., ’blank solution’) was
injected at a rate of 2.6 ml min-1 . The results from the injection tests with blank
solutions are shown in Figure C1B.

Figure C1: The ingress of air was shown as A) peaks during OS injection and B) as
regions of peaks and continuous ingress when a blank solution was injected (pump
rate).

Figure C1B indicate that the ingress of air initially increased the oxygen concentra-
tion by some 30 ppb shown as peaks, which were presumably caused by the stagnant
64

water in the intermediate pipeline having accumulated a higher oxygen concentration.


However, after the oxygen peak, the oxygen concentration decreased and stabilized
at a value of 15-20 ppb above the oxygen concentration prior to pumping. This
continuous ingress of air can be attributed to leaks in both the intermediate pipeline
and the dosing pump. When the dosing pump was stopped, the oxygen concentration
decreased again to an oxygen level same as prior to injection of the blank solution.
These results show that the dosing pump contributed with a 15-20 ppb continuous
increase in the oxygen concentration at a pump rate of 2.6 ml min-1 . Although this
test was not repeated under additional pump rates, the amount of the continuous
ingress of air can be considered to increase with the pump rate. Since the same
pump and similar pump rates have been used in all measurements, the determined
reaction kinetics are still comparable within this thesis.
This undesired ingress of air has affected the oxygen reduction kinetics. The
ingress oxygen has most likely reacted with the OS already in the dosing pump and
the intermediate pipeline and can therefore be considered to have lowered the OS
concentration of the injected solution. For a given pump rate, the effect of the air
ingress has been greater the lower the concentration of the initial OS stock solution.
To summarize, the effect of the air ingress on the results of the oxygen reduction
kinetics are dependent on the initial OS stock solution and the pump rate. Although
the magnitude of this error was not determined in quantitative measures, the air
ingress cannot be considered significant enough to entirely debunk any of the thesis
results. The air ingress has in the worst case scenario caused the determined reaction
kinetics to be slightly slower than what can be expected. That is, if the air ingress
can be eliminated in repeated tests, faster oxygen scavenging kinetics can be expected
for hydrazine, erythorbic acid and diethylhydroxylamine.
65

D Characterization of the iron oxides


The oxide films on the carbon steel samples were characterized with Zeiss Gemini
Ultra Plus scanning electron microscopy (SEM) and energy-dispersive X-ray spec-
troscopy (EDS). The purpose of this appendix is to highlight the findings from the
characterizations.
The electron micrographs of the 100 ppb and 10 ppm samples are shown in
the figures D1 and D2. The surface morphology differs significantly between the
two samples as the 100 ppb oxide surface consists of irregularly shaped particles of
seemingly similar sizes whereas the 10 ppm oxide consists of sphere-shaped particles
of varying sizes.

Figure D1: A) surface and B) cross-section electron micrographs and elemental


compositions (EDS, atom-%) of 22K carbon steel after 1200 hours of exposure to
recirculating water with an CO2 of 100±20 ppb, a temperature of 228 ◦ C and a
pH25◦ C of 9.6±0.2.

Figure D2: A) surface and B) cross-section electron micrographs and elemental


compositions (EDS, atom-%) of 22K carbon steel after 200 hours of exposure to
recirculating water with an CO2 of 10±0.2 ppm, a temperature of 228 ◦ C and a
pH25◦ C of 9.6±0.2.
66

The oxides differed also in thickness and composition. The 100 ppb oxide had a
thickness of 0.41±0.12 μm and the 10 ppm oxide had a thickness of 0.25±0.05 μm.
The 100 ppb oxide cross-section was also found to consist to a larger degree of oxygen
(oxygen to iron ratio of 1.25) than the oxide of the 10 ppm sample (oxygen to iron
ratio of 0.72) as determined by the EDS analysis.
These observations may seem counter-intuitive as a higher oxygen concentration
in the water can be expected to result in a thicker oxide and possibly also a higher
oxygen content in the oxide. However, these deviations can be attributed to the
exposure time as the 10 ppm sample was only part of the circuit for 200 hours
whereas the 100 ppb sample was for a total of 1200 hours which has allowed for a
thicker oxide to form but possibly also oxygen to occupy a larger share of the film.
Although the oxide film should consist of an inner and an outer layer, the
micrographs were not able to convincingly discriminate between inner and outer
layers. Nevertheless, the surface morphology and the measured thicknesses correspond
generally with oxides described in similar studies [24, 33, 34].
Although the EDS analysis could semi-quantitatively determine the ratio of
oxygen to iron atoms in the oxide films, the analysis was not able to discriminate
between iron oxide forms, and thus not identify what oxides the films consisted of.
However, the electrochemical potentials in LPR and EIS measurements were found to
have been for the most part inside the thermodynamically stable region of hematite.
Therefore, the oxide films can be considered to have consisted primarily of hematite.
67

E The effect of OSs on the electrolytic conductivity


During measurements in waters containing approximately 10 ppm oxygen content, the
conductivity was found to increase with the addition of erythorbic acid. This appendix
presents the effect of adding oxygen scavengers on the electrolytic conductivity.
The electrolytic conductivity was continuously measured with an electrochemical
conductivity sensor (HACH polymetron 8310) located in the cold section return line
as shown in Figure 16.
The electrolytic conductivity during administration of oxygen scavengers are
shown as a function of time in Figure E1. In the 100 ppb measurements, erythorbic
acid decreased the conductivity slightly, whereas hydrazine and diethylhydroxylamine
had no recognizable effect on the conductivity. In 10 ppm measurements, all three
oxygen scavengers increased the electrolytic conductivity.
The increase in conductivity can be attributed to the oxygen scavengers decom-
posing into or participating in reactions that produce charged species, which in turn
increase the solution conductivity. An increase of the conductivity can be expected
to result in higher corrosion rates, which is undesirable. The threefold increase in
conductivity caused by erythorbic acid is particularly concerning.
The decrease in conductivity of erythorbic acid in the 100 ppb measurements
seems counter-intuitive. A decrease in conductivity would mean that the activities
of erythorbic acid would result in the consumption of ammonia, since the charged
species in the solution can be considered associated with the ammonia ions NH4 +
and NH2 – . A decrease in conductivity would be concerning should it mean that
ammonia, which acts as a pH buffer, would be consumed by the oxygen scavenger.

Figure E1: The conductivity of the return line during administration of oxygen
scavengers in waters with oxygen concentrations of A) 100±20 ppb and B) 10±2 ppm
measurements. Erythorbic acid caused the most significant change of conductivity.
The linear declines are due to filtering the water through ion-exchangers.

You might also like