Influence of water chemistry on the corrosion mechanism of a zirconium–niobium alloy in simulated light water reactor coolant conditions

Influence of water chemistry on the corrosion mechanism of a zirconium–niobium alloy in simulated light water reactor coolant conditions

Corrosion Science 52 (2010) 54–67 Contents lists available at ScienceDirect Corrosion Science journal homepage: www.elsevier.com/locate/corsci Influ...

1MB Sizes 12 Downloads 69 Views

Corrosion Science 52 (2010) 54–67

Contents lists available at ScienceDirect

Corrosion Science journal homepage: www.elsevier.com/locate/corsci

Influence of water chemistry on the corrosion mechanism of a zirconium–niobium alloy in simulated light water reactor coolant conditions Martin Bojinov a,*, Vasil Karastoyanov a,b, Petri Kinnunen b, Timo Saario b a b

Department of Physical Chemistry, University of Chemical Technology and Metallurgy, Kl. Ohridski Blvd. 8, 1756 Sofia, Bulgaria VTT Materials and Building, VTT Technical Research Centre of Finland, P.O. Box 1000, Kemistintie 3, FIN-02044 VTT, Espoo, Finland

a r t i c l e

i n f o

Article history: Received 25 June 2009 Accepted 18 August 2009 Available online 23 August 2009 Keywords: A. Zirconium alloy B. EIS B. SEM C. Passive films C. Kinetic parameters

a b s t r a c t The oxidation of the zirconium alloy E110 inf simulated nuclear power plant coolant at 310 °C is characterised using in situ Electrochemical Impedance Spectroscopy (EIS) and ex-situ microscopic observations. EIS data have been fitted to a transfer function derived from the Mixed Conduction Model. The kinetic parameters characterising the oxidation process – interfacial rate constant of oxidation, diffusion coefficient of oxygen vacancies, and field strength in the inner layer – have been estimated. The dependence of their values on LiOH/KOH/NaF content are discussed in terms of an enhanced rate of dissolution of the barrier layer at higher level of alkali and fluoride. Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction The good irradiation stability, corrosion resistance, and mechanical properties of zirconium-based alloys in a reactor environment have promoted their wide use as fuel cladding and structural materials for nuclear reactors. Recently, enhanced operating conditions such as an increased burn-up and higher reactor temperatures have called for the development of advanced Zr-based alloys [1–5]. The tendency to select Nb as the major alloying element in such alloys is a common characteristic for the newly developed fuel claddings [5]. Corrosion of the Zr-alloy fuel cladding in Pressurised Water Reactors (PWRs) has become more important due to a higher fuel discharge burnup to reduce fuel cycle costs, a higher coolant inlet temperature to increase plant thermal efficiency, and an increase of the coolant pH and lithium concentration to reduce plant radiation levels. Even though the corrosion mechanism of zirconium alloys is still not fully elucidated, the main governing factors of the process are presumed to be the metallurgical characteristics of the cladding, the irradiation environment, water chemistry, and the thermo-hydraulic condition of the coolant [6]. Concerning the water chemistry, an increase in the corrosion rate of 10–30% has been observed when the maximum coolant lithium content was increased from 2.2 to 3.5 ppm [7], the effect of higher lithium concentrations being much more pronounced [8]. On the other * Corresponding author. Tel.: +359 2 816 3 430; fax: +359 2 868 20 30. E-mail address: [email protected] (M. Bojinov). 0010-938X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.corsci.2009.08.045

hand, there are also reports that do not suggest a discernible oxidation enhancement in the presence of an elevated lithium concentration [9]. Elevated lithium concentration combined with local subcooled nucleate boiling accentuated by high burn-up may result in formation of crud onto fuel cladding surfaces and further in the phenomenon called Crud Induced Power Shift (CIPS) or localised cladding corrosion [10]. Several studies are being carried out to elucidate the mechanism of CIPS (e.g. the IAEA’s Optimisation of Water Chemistry to ensure Reliable Water Reactor Fuel Performance at High Burnup and in Aging Plant (FUWAC) 2006-2010 – programme). In most part of the studies published so far, it has been proposed that there may be a connection between the alkali and boron concentrations, the composition of the fuel cladding material and the occurrence of CIPS [11–14]. Even if there seems to be a general agreement on the nature of the oxidation process of zirconium alloys in high-temperature water, the mechanism of conduction through the formed oxide has not been fully explored [15,16]. In order to evaluate quantitatively the effect of water chemistry on the oxidation process of zirconium alloy fuel cladding materials, in situ techniques are needed to complement the ample amount of information stemming from ex-situ characterisation of the oxide with surface analytical means. In recent years, Electrochemical Impedance Spectroscopy (EIS) has been extensively used for the characterisation of corrosion films on Zr-based alloys [17–35]. However, only a few studies attempt at modelling the full obtained spectra [32–35], the majority of works concentrating on the high-frequency part of the diagrams in order to estimate the thickness of the different layers of oxide

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

formed at variable oxidation times [27–29]. In a previous paper, some of us presented a quantitative comparison between the impedance data for Zircaloy-4 and E110 alloys in Water cooled Water moderated Energy Reactor (WWER) coolants with different KOH contents [35]. The results reported in that paper indicated that the oxidation is enhanced at higher KOH concentrations. Based on the interpretation of the impedance spectra using the Mixed Conduction Model (MCM) [34], it has been found that the rate constant of Zr oxidation (generation of oxygen vacancies) at the alloy/film interface increases, the diffusion coefficient of oxygen vacancies in the barrier layer increases, whereas both the defect induced resistivity of the inner layer and the overall resistivity of the outer layer decrease with increasing KOH concentration [35]. In the present paper, those results are compared with measurements in simulated PWR coolant with normal and elevated Li concentrations (2.2, 5.4 and 10 weight ppm). In addition, the effect of fluoride ion concentration (120 and 1000 weight ppb) in normal beginning-of-cycle WWER chemistry has been assessed. The EIS results obtained during exposure of E110 at 310 °C are interpreted using the MCM and the main kinetic and transport parameters are estimated by fitting the spectra to the model equations. The obtained parameter values are discussed in terms of the effect of LiOH, KOH and fluoride concentration on the kinetics of oxidation and the conduction mechanism of the formed oxide on E110 alloy.

55

2. Experimental 2.1. Material and conditions The fuel cladding material used in this study was E110 (Zr– 1%Nb, Chapetsky Mechanical Plant, Russia, nominal alloy composition, wt.%: Nb 1.00, Sn < 0.01, Fe 0.014, Cr < 0.003, Ni < 0.004, C < 40–70 weight ppm, Si 46–90 weight ppm, N < 30–40 weight wppm, Hf 300–400 weight ppm, balance Zr). The E110 alloy was not anodised, i.e. the outer surface of the tube has not been preoxidised. The disc-shaped specimens were cut from as-fabricated tubes by a diamond saw. The edges of the discs (5.5 mm in diameter) were rounded with a side cutter. To prepare the working electrodes for the electrochemical measurements, a zirconium wire was spot-welded to the backside of the specimens and mechanically connected to a nickel wire. All the electrical connections were insulated using multilayer PTFE tape. The area of the samples exposed to the simulated coolant was estimated to be 0.5 cm2. The experiments have been carried out in the following environments: (1) Normal PWR conditions – 2.2 weight ppm Li as LiOH, 1150 weight ppm B as H3BO3, 2.35 weight ppm of dissolved H2 (26 cm3 kg1 STP), inlet oxygen content <5 weight ppb, pH300  6.9.

Fig. 1. Cross sections of the E110 samples after 120 h oxidation in simulated PWR coolant with 2.2 ppm (a) and 5.4 ppm (right) LiOH at 310 °C.

56

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

(2) PWR with elevated Li concentrations – 5.4 and 10 weight ppm Li as LiOH, other parameters similar to those in environment 1, pH300  7.3..7.4). (3) Beginning-of-cycle WWER water chemistry – 11, 28 and 56 weight ppm K as KOH, 1150 weight ppm B as H3BO3, 2.35 weight ppm of dissolved H2 (26 cm3 kg1 STP), inlet oxygen content <5 weight ppb, pH300  6.9. (4) Similar environment as in test 3 with 11 weight ppm K as KOH but with 120 and 1000 weight ppb of fluoride as NaF. All experiments were performed at 310 °C in an autoclave connected to a re-circulation loop. All the measurements were at least triplicated in order to ensure the reproducibility of the obtained results. 2.2. Apparatus and procedure Electrochemical impedance spectra were measured by a Solartron 1287/1260 system controlled by ZPlot/ZView software (Scribner Associates) in a semi continuous fashion during 1. . .5 days of exposure to the simulated coolant. Ir (99.9%) was used as both counter and pseudo-reference electrode in analogy to previous measurements [31]. The contribution of the Ir electrode to the

impedance has been found to be less than 3% over the whole frequency range as estimated by measuring the impedance of an Ir– Ir couple in similar conditions. The frequency range in the measurements was from 80 kHz to 0.001 Hz and the amplitude of the a.c. perturbation was 50 mV (rms). The validity of the impedance spectra was ensured by checking the linearity condition, i.e. measuring spectra at signal amplitudes between 5. . .50 mV (rms), and by checking the causality using a Kramers–Kronig transform (KK) test included in the ZView software. The reproducibility of the impedance spectra was ±2% by magnitude and ±3° by phase angle. After the completion of the impedance measurements, the specimens were removed from the autoclave to estimate the total thickness of the oxides by Scanning Electron Microscopy (SEM). For the simulation and fitting of impedance spectra to the transfer function derived from the kinetic model, Microcal Origin-based software was employed. 3. Results 3.1. Microscopic estimation of film thickness Estimates of oxide film thickness have been obtained from scanning electron microscopic examinations of sample cross sections.

Fig. 2. Microscopic estimates of the oxide film thickness on E110 after 140 h of exposure to simulated coolant as depending on the LiOH content (a), KOH content (b) and fluoride content (c).

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

This method was preferred to surface analytical techniques associated with sputter depth profiling due to the fact that these types of measurements make use of external standards (such as tantalum or silicon oxide) to convert sputtering time to penetration depth. In fact, a search of the relevant literature showed that use of such techniques (e.g. X-ray photoelectron spectroscopy, Auger electron

57

spectroscopy and Secondary Ion Mass Spectrometry) for the determination of corrosion layer thickness on zirconium alloys is rather limited [36–40]. As an example, Fig. 1 shows SEM micrographs of the cross-sections of E110 samples oxidised for 120 h in simulated PWR coolant with 2.2 and 5.4 ppm LiOH. At the magnifications used, the oxides

Fig. 3. Electrochemical impedance spectra of E110 in simulated PWR coolant with (a) 2.2 ppm, (b) 5.4 ppm and (c) 10 ppm LiOH at 310 °C as depending on the exposure time. Left – impedance magnitude vs. frequency, right – phase angle vs. frequency. Points – experimental values, solid lines – best-fit calculation according to the proposed model.

58

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

Fig. 4. Electrochemical impedance spectra of E110 in simulated WWER coolant with (a) 11 ppm, (b) 28 ppm and (c) 56 ppm KOH at 310 °C as depending on the exposure time. Left- impedance magnitude vs. frequency, right – phase angle vs. frequency. Points – experimental values, solid lines – best-fit calculation according to the proposed model.

seem rather homogeneous, their thickness increasing to a measurable amount with LiOH content.

Fig. 2 summarises the effect of LiOH, KOH and fluoride content on the thicknesses of oxide films formed on E110 in simulated PWR

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

59

Fig. 5. Electrochemical impedance spectra of E110 in simulated WWER water (11 ppm KOH) with (a) 120 ppb and (b) 1000 ppb NaF at 310 °C as depending on the exposure time. Left – impedance magnitude vs. frequency, right – phase angle vs. frequency. Points – experimental values, solid lines – best-fit calculation according to the proposed model.

and WWER coolant conditions for 120 h at 310 °C. The thickness values are in broad agreement with literature data, although thickness measurements for such relatively short exposure times are scarce [16]. There is a noticeable increase in film thickness with increasing alkali concentration, the trend of this increase being supralinear. On the other hand, the addition of 120 ppb of fluoride to the simulated WWER coolant containing 11 ppm of KOH leads to a noticeable increase in film thickness, whereas the oxide film formed in a solution containing 1000 ppb of F has been found to be somewhat thinner. Summarising, increasing both alkali and fluoride content of the simulated coolant leads to an increase in the thickness of the oxide film formed for identical exposure periods, the trends not leading to simple relationships between film thickness and concentration. Explanations for these features were sought by conducting in situ EIS measurements during exposure of the alloy to the simulated coolant. 3.2. Impedance spectra The electrochemical impedance spectra of E110 at 310 °C in simulated PWR coolant with different LiOH content (2.2, 5.4 and

10 ppm) are shown in Fig. 3 for several exposure times. The corresponding spectra measured in simulated WWER coolant with different KOH contents (11, 28 and 56 ppm) as depending on the time of exposure are presented in Fig. 4, whereas the effect of fluoride addition on the spectra measured in simulated WWER coolant (11 ppm KOH) is illustrated in Fig. 5. Following our previous work [34,35], Bode representations of the spectra were adopted in order to be able to discern the impedance features at both high and low frequencies. The magnitude of the impedance at low frequencies (ca. 0.001 Hz), Zf?0, can be interpreted as the inverse of the oxidation rate of the alloy/oxide/electrolyte system. This quantity increases in general with exposure time, which can be interpreted as a decrease of the oxidation rate with increasing oxide thickness. The increase of Zf?0 is not very pronounced, indicating that the corrosion process is slow and the system remains in a quasi-steady state with respect to the measurement time of a single spectrum. This conclusion was corroborated by the fact that all the presented impedance spectra passed the Kramers–Kronig transform test. On the other hand, Zf?0 decreases with increasing LiOH, KOH and F content in the coolant (Figs. 3–5.), the decrease being more

60

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

pronounced at the highest concentrations of the respective additives (10 ppm LiOH, 56 ppm KOH and 1000 ppb fluoride). This observation is in accordance with the increase of estimated oxide layer thickness after 120 h of exposure and demonstrates the accelerating effect of the investigated additives on the oxidation rate. A possible explanation to this observation is that the increase in LiOH, KOH and fluoride concentration leads to an increase of the solubility of zirconium oxide at its interface with the electrolyte, therefore a quasi-steady state is reached at a larger respective rate of oxide growth to compensate for dissolution. This hypothesis implies that either solubility or dissolution rate of the oxide are corrosion rate limiting. In fact, analogous ideas have been put forward by Markworth et al. [41] for the corrosion of Zr–2.5%Nb alloy in Canadian Deuterium Uranium (CANDU) coolants. In the model devised by these authors, the zirconium oxide layer is treated as a bilayer with an inner protective and an outer porous non-protective layer. The transformation of the barrier layer into a non-protective porous layer at the outer side of the barrier layer ultimately controls the reaction, because the thickness of the barrier layer is dependent on this transformation. Similar ideas have been expressed by Macdonald and co-workers using the framework of the Point Defect Model (PDM) [32,33,42]. Three time constants are detected in the phase angle vs. frequency curves (Figs. 3–5) in a certain analogy to the results published by Schefold et al. in simulated PWR coolant at 360 °C [29] and by some of us in a simulated WWER coolant [35]. Following our earlier work [35], the time constant situated at the highest frequencies (ca. 10–30 kHz) can be related to the electric properties of the outer layer of oxide. The characteristic frequencies of the highest-frequency time constant increase with the time of exposure and also with LiOH/KOH/NaF content in the electrolyte, indicating a larger thickness and/or an increase in the conductivity of the outer layer. On the other hand, the time constant observed at intermediate frequencies (100–500 Hz) most probably depicts the electric properties of the inner, protective barrier layer. A certain increase in the characteristic frequency of this time constant with exposure time is also observed, although it is to a certain extent masked by the overlap of the two time constants and is not so easily discernible. A noteworthy observation is that the intermediate frequency time constant is well defined in the case of spectra measured in WWER coolant with the addition of fluoride (Fig. 5), which may mean that the barrier layer formed in the presence of this additive is rather thin in comparison to those formed in the absence of fluo-

ride. Last, the lowest frequency time constant situated below 0.1 Hz is most probably due to a solid state diffusion-migration process (e.g. transport of oxygen vacancies in the barrier layer). No meaningful evolution of the characteristic frequency of the time constant associated with the transport process with exposure time or concentration of alkali/fluoride could be observed. This could be due to the fact that the transport properties of the barrier layer are not so significantly affected by either the exposure time or LiOH/ KOH/NaF concentration. However, a certain increase of this characteristic frequency could be observed when the KOH content is increased from 28 to 56 ppm, which could be related to an increase in the transport rate. On the other hand, a certain flattening of the low-frequency response has been detected in the solution containing 10 ppm Li, which could possibly be related to some alteration of the transport process in that particular case. Summarising, the in situ impedance measurements point to a well noticeable effect of LiOH, KOH and fluoride concentration on the initial stages of the oxidation process of zirconium alloys in simulated PWR/WWER coolant. These measurements allow to discern the contributions of the electric properties of two layers of oxide – a thin barrier layer and a thicker (more defective, and hence probably less protective) outer layer, as well as a solid-state transport process of point defects probably occurring in the barrier sublayer. In the next section, an attempt to rationalise these findings using a version of the MCM developed for the corrosion of zirconium alloys will be presented with the ultimate goal to determine the kinetic and transport parameters of the oxidation process.

4. Discussion 4.1. Physical model A scheme of the reactions considered in the proposed model for the initial stages of oxidation of E110 alloy is shown in Fig. 6. It is based both on the evidence for the bilayer structure of the formed oxide obtained from in situ impedance measurements in the present study and previous reports by a number of authors [29,32–35,42]. At the alloy/barrier layer interface oxidation of Zr (and also of Nb and other alloying elements in the Zr-matrix [43]) proceeds. This reaction is written in complete analogy to the PDM and MCM as follows: k1

m ! Zrzr þ 4e0 þ 2Vo

Fig. 6. A scheme of the proposed model for the corrosion of E110 in a PWR/WWER coolant.

ð1Þ

61

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

The oxidation reaction is considered to be irreversible since the corrosion potential of zirconium alloys in simulated PWR and WWER coolants (determined in this case by the dissolved hydrogen content) is much higher than the equilibrium potential of Zr oxidation. In what concerns the behaviour of the main alloying element, niobium, very recently, the effective valency state of its substitute ions in the Zr sub lattice of the oxide have been estimated by X-ray Adsorption Near Edge Spectroscopy (XANES) to be between 2 and 4 [5]. This means that Nb can be considered either as hypovalent or isovalent defect, at variance to previous discussions that consider the presence of Nb(V) as a hypervalent defect [30]. The barrier layer is considered to be a mixed conductor in which both electrons and oxygen vacancies are transported under the influence of both concentration and potential gradients, in a certain analogy to the coupled currents model proposed by Fromhold [44]. Since moving oxygen vacancies can act as effective traps for electrons, it can be assumed that the diffusivities of ionic and electronic defects are comparable, and their transport fluxes are described by the equation:

J O ðx; tÞ ¼ DO

@cO ðx; tÞ F~ E  2 DO cO ðx; tÞ @x RT

ð2Þ

In this equation, cO ðx; tÞ is the time and spatially dependent concentration of oxygen vacancies, ~ E is the electric field strength in the oxide and DO is the diffusion coefficient of oxygen vacancies. At the barrier layer/outer layer interface, oxygen vacancies can react with water incorporated in the outer layer via pores and cracks to achieve barrier layer growth: k2

2Vo þ 2H2 O þ 4e0 ! 2Oo þ 4Hbl

ð3Þ

where Hbl denotes an oxygen atom incorporated in the barrier layer. The process is accompanied with atomic hydrogen incorporation in the barrier layer and its further transport within this layer to reach the underlying alloy. The transport of hydrogen through this layer is, however, considered as a slow reaction in parallel to oxidation, since several reports have indicated that the rate of incorporation of hydrogen into zirconia is much lower than the rate of oxide growth [6]. The steady state solution of this equation with the boundary conditions JO (L) = k1, JO(0) = k2cO(0) in a coordinate system in which x = 0 is at the barrier oxide/water interface at the bottom of the pores of the outer layer and x = L at the alloy/barrier layer interface, L being the barrier layer thickness, gives the concentration profile of oxygen vacancies in the barrier layer:

" cO ðxÞ ¼ 2k1

F~ E

e2RT x RT þ k2 2F~ EDO

# ð4Þ

At the barrier layer/water interface, restructuring of that layer occurs resulting in the formation of an outer layer. This restructuring could proceed via dissolution, the rate of which is influenced by the concentration of alkali and fluoride in analogy to what has been proposed by Markworth et al. [41]: Kd

ZrO2 þ OH ! HZrO3 Kd

ZrO2 þ 2F ! ZrO2 F2 2

ð5Þ

The transport of matter and charge through the defective outer layer is assumed to take place via short circuit paths like imperfections and cracks, or interconnected porosity [6,16]. Thus no concentration gradient of oxygen vacancies is thought to be present in this layer and the transport mode through it is presumed to be pure migration.

4.2. Impedance response Following the line of reasoning in the previous section, the overall impedance of the Zr alloy/oxide/coolant system can be written as:

Z ¼ Rel þ Z b þ Z out

ð6Þ

For the impedance of the outer layer, trials of different distributed functions [35] demonstrated that the so-called Havriliak–Negami impedance [45] gave the best fit to the experimental data:

Z out ¼ 

Rout 1 þ ðjxRout C out Þu

ð7Þ

n

where C out and Rout are the capacitance of the outer layer and the apparent resistance of defect migration through that layer, whereas u and n are fractional exponents. The Havriliak–Negami response represents a generalisation of the Constant Phase Element (CPE) to account for asymmetric capacitive loops. It is also closely related the Williams–Watts distribution derived from the continuous random walk model of Scher and Lax for transport in disordered and composite media, which has already been used to describe charge transport in zirconium oxides [31]. The Havriliak–Negami element has been proposed originally for polymer dispersions [45] and can be regarded according to the original authors as the impedance of a two phase mixture, which seems to be a good approximation for the outer layer of oxide. The impedance of the barrier layer is given by the sum of the impedances characterising its electric properties and ionic transport through it in parallel: 1 1 Z b ¼ ðZ 1 e þ Z ion Þ

ð8Þ

The electronic contribution to the impedance, Ze, is related to the spatial variation of the steady-state concentration of oxygen vacancies in the oxide, which creates a positive ionic space charge that requires electronic compensation to achieve electroneutrality. Using Eq. (4) as a starting point, an expression for Ze has been derived by some of us [46]:

2 3 RT RT 1 þ jxqd ee0 e2F~ELb 5 4 Ze ¼ ln 1 þ jxqd ee0 Eee0 2jxF~

ð9Þ

1 where qd ¼ F 2RTk , e is the dielectric constant of zirconium oxide, asDe k 2 sumed to be equal to 22 [16], e0 is the dielectric permittivity of free space, x is the angular frequency and De is the diffusion coefficient of the electronic current carriers. On the other hand, the impedance due to the motion of oxygen vacancies, Zion, can be obtained as the small amplitude ac solution of Eq. (2):

Z ion ¼ Rt þ

RT rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi    jxðRTÞ2 F~ E 4F RT DO cO ðLÞð1  aÞ 1 þ 1 þ ðF ~ EÞ2 D

ð10Þ

2

O

where cO(L) is the concentration of oxygen vacancies at the alloy/ barrier layer interface and a is a parameter that characterises the part of the overall potential drop in the alloy/barrier layer/water system which is located at the barrier layer/water interface as opposed to film bulk. In the present case, it is assumed that a = 0, i.e. all the potential drop is located within the barrier layer [35]. F~ E F~ E On the other hand, under the assumption that k2 e2RT Lb >> 2 RT DO [34,35] and using Eq. (4) the concentration cO(L) can be approxi1 . Thus the equation for the impedance due to mated as cO ðLÞ  FRTk ~ EDO the transport of oxygen vacancies becomes:

62

Z ion ¼ Rt þ

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

RT rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  jxðRTÞ2 4F k1 1 þ 1 þ ðF ~ EÞ2 D

ð11Þ

2

O

4.3. Estimation of kinetic and transport parameters The kinetic and transport parameters, namely, the rate constant of oxidation at the alloy/barrier layer interface, k1 , the diffusion coefficient of oxygen vacancies, DO , the field strength in the barrier layer, ~ E, the charge transfer resistance at the alloy/oxide interface,

Rt , the qd e parameter defined in Eq. (9), as well as the parameters characterizing the outer layer (C out and Rout ) were estimated using non-linear least square fitting of the experimental spectra to the transfer function derived in the previous paragraph. Statistical weighting was used for the experimental data set and the errors of parameter estimation were multiplied by the square root of the reduced v2-value resulting from the fit. In spite of the relatively large number of parameters, this resulted in a sufficient number of degrees of freedom in the system in order to obtain statistically reliable values of the kinetic parameters. The calculated spectra are presented in Figs. 3–5 with solid lines and demonstrate the ability of the proposed model to account for

Fig. 7. Dependences of the (a) inner (full symbols) and outer (open symbols) layer thickness, (b) the parameter qde, (c) the resistance of the outer layer Rout, (d) the rate constant of zirconium oxidation at the alloy/oxide interface k1, (e) the diffusion coefficient of oxygen vacancies DO and (f) the field strength in the inner layer on the time of exposure of alloy E110 to the simulated PWR coolants with different LiOH content (in weight ppm).

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

both the magnitude and the frequency distribution of the experimental impedances. Due to the strong overlap of the time constants, the charge transfer resistance Rt could only be calculated with a relatively large error and is thus not commented further. Assuming a dielectric constant of 22, the thickness of the outer layer was computed from the respective capacitance value and was found to be much larger than the thickness of the barrier layer Lb. A collection of all the estimated parameters as depending on exposure time is presented in Fig. 7 (PWR coolant with different LiOH content), Fig. 8 (WWER coolant with different KOH content) and Fig. 9 (WWER coolant with different fluoride content). As an

63

example, Fig. 10 shows the dependences of the steady-state values of k1 ; DO ; qd and qout ¼ Rout =Lout on LiOH content of the simulated coolant. The following main conclusions can be drawn from the calculated parameter values: 1. The calculated estimates of the thickness of the whole oxide film after 120 h of exposure agree very well with the microscopic estimates of the total oxide thickness, indicating the validity of the proposed model for the oxide film. The thickness of the inner layer is several times smaller than that of the outer

Fig. 8. Dependences of the (a) inner (full symbols) and outer (open symbols) layer thickness, (b) the parameter qde, (c) the resistance of the outer layer Rout, (d) the rate constant of zirconium oxidation at the alloy/oxide interface k1, (e) the diffusion coefficient of oxygen vacancies DO and (f) the field strength in the inner layer on the time of exposure of alloy E110 to the simulated WWER coolant with different KOH content (in weight ppm).

64

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

layer, which indicates that the corrosion properties of the zirconium alloys in the initial stage of oxidation are controlled by a thin layer close to the alloy/oxide interface. It is noteworthy to mention that the thickness of the inner oxide layer is comparable to that measured during oxidation of Zr in dry oxygen at 573 K for 100 h [47–49]. 2. The thickness of the outer layer increases with the concentration of LiOH and KOH in the electrolyte, whereas the values of the inner layer thickness for the highest concentrations of LiOH and KOH are the smallest. This could be explained by an increased defectiveness of the oxide formed in high LiOH/

KOH/NaF water chemistries and furnishes some support to the hypothesis that the restructuring process that leads to the transformation of the inner layer into the outer oxide is assisted by dissolution. 3. The values of the qd e parameter are in general also lower in the electrolytes containing higher amounts of hydroxides. As 1 qd ¼ F 2RTk , and the values of the rate constant of oxidation at D e k2 the alloy/inner layer interface in general increases with alkali concentration, a decrease in qd ecould translate into an increase of either the diffusion coefficient of electronic carriers or the rate constant of reaction (3), or both. Unfortunately, the present

Fig. 9. Dependences of the (a) inner (full symbols) and outer (open symbols) layer thickness, (b) the parameter qde, (c) the resistance of the outer layer Rout, (d) the rate constant of zirconium oxidation at the alloy/oxide interface k1, (e) the diffusion coefficient of oxygen vacancies DO and (f) the field strength in the inner layer on the time of exposure of alloy E110 to the simulated WWER coolant with 11 weight ppm KOH and variable fluoride content (in weight ppb).

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

65

Fig. 10. Dependences of the steady-state values of the kinetic and transport parameters for E110 in PWR water chemistry on LiOH content in the electrolyte (in weight ppm).

calculation procedure does not allow for the determination of the diffusion coefficient of electronic carriers as an independent parameter. 4. As mentioned already above, the rate constant of zirconium oxidation at the inner interface increases with the concentration of LiOH/KOH/NaF. The order of this reaction vs. LiOH is close to 2 (as indicated by the slope of the straight line, Fig. 10), whereas the estimated order vs. KOH was of the order of 1. On the basis of this difference it could be speculated that the effect of increased LiOH content is due not only to the increase in alkali concentration, but there is an additional effect of Li itself. This could be in principle related to some hypotheses on the incorporation of Li in the oxide film on zirconium alloys proposed in the literature, as discussed e.g. in Ref. [6]. The observations in the present work are in line with previously reported differences in the corrosion rates of zirconium alloys in PWR and WWER water chemistries [50]. 5. The estimated values of the diffusion coefficients of oxygen vacancies are in broad agreement with values published in the literature [33,48,49]. It is noteworthy that the diffusion coefficients calculated for the oxidation of zirconium alloys in PWR and WWER coolants are considerably smaller than those estimated in BWR conditions (in the presence of dissolved oxygen) [34] or for anodically polarized zirconium in deaerated high-temperature water solutions [32]. This observation can be correlated to the higher oxide film thicknesses obtained for similar exposure times in oxygenated high-temperature solutions. 6. The increase of the diffusion coefficients for films formed in solutions with higher alkali (Fig. 10) and/or fluoride concentrations can be once more correlated to the higher rate of film

growth needed to compensate for an increase in dissolution rate. The decrease of the diffusion coefficient with time of exposure in the solution with 2.2 ppm Li (Fig. 7) is to a certain extent analogous to what has been proposed in Refs. [49,50] and can be explained by a decrease of the available easy paths for oxygen diffusion via grain boundaries as the oxide grows. 7. The values of the field strength in the barrier layer preserves rather low values (around 10 kV cm1), which demonstrates the viability of the low-field approximation of the transport equations used in the present model. A certain decrease of the field strength with increasing LiOH concentration is detectable, and can be correlated to the increase of the defectiveness of the barrier layer e.g. via incorporation of Li ions in it. 8. The resistivity of the outer layer, which can be regarded as an inverse of the transport rate in this layer by easy path migration, decreases with increasing alkali and fluoride content, indicating that this layer becomes more defective and thus indirectly supporting the hypothesis that the restructuring process that leads to the formation of that layer is dependent on the solution composition, i.e. water chemistry. Clearly, to obtain further support for such a hypothesis, more compositional, structural and morphological data on the oxides are needed. Such data are planned to be produced and analyzed in the near future.

5. Conclusions The experimental results obtained in the present work indicate that for short term exposures of alloy E110 in simulated PWR and WWER water, the rate of oxidation is increased at the higher LiOH/

66

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67

KOH concentration. Based on the quantitative interpretation of in situ impedance spectra by the Mixed Conduction Model the rate constant of Zr oxidation (generation of oxygen vacancies) at the alloy/film interface is found to increase, its apparent order vs. alkali concentration being ca. 2 in LiOH-containing water and ca. 1 in KOH-containing water. The diffusion coefficient of oxygen vacancies in the barrier layer also increases, which can be understood as an increase of the transport of point defects in the inner barrier layer to compensate for enhanced dissolution/restructuring of this oxide at its interface with the coolant water. The obtained calculation results also show that both the defect induced resistivity of the inner layer and the overall resistivity of the outer layer decrease with increasing alkali concentration, which can be interpreted as an increase in the defectiveness of these layers. In addition, it has been found that both the rate of oxidation, the transport in the barrier layer, as well as the outer layer conductivity increase in the presence of small amounts of fluoride, a wellknown complexing agent for zirconium ions. The hypothesis put forward to explain such a behaviour is that the effect of increasing LiOH/KOH/NaF content of the simulated coolant is to enhance the chemical dissolution rate of the oxide, which in turn influences the rates of transport in the barrier layer and oxidation of Zr at the alloy/barrier layer interface. The process of dissolution at the oxide/ electrolyte interface itself is not discernable in the impedance spectra because it can be regarded as a chemical process that is not directly influenced by the potential drop through the oxide film. As an overall result, the changes in the above mentioned parameters lead to a film thickness increase with increasing LiOH and KOH concentration, which could lead to an increase of the oxide film thickness also for prolonged exposure of E110 to coolants with elevated alkali contents. Acknowledgement The funding of this work by SAFIR2010 – The Finnish Research Programme on Nuclear Power Plant Safety 2007 – 2010 is gratefully acknowledged. References [1] H.G. Kim, S.Y. Park, M.H. Lee, Y.H. Jeong, S.D. Kim, Corrosion and microstructural characteristics of Zr–Nb alloys with different Nb contents, J. Nucl. Mater. 373 (2008) 429. [2] J.-Y. Park, S.J. Yoo, B.-K. Choi, Y.H. Jeong, Corrosion and oxide characteristics of Zr–1.5Nb–0.4Sn–0.2Fe–0.1Cr alloys in 360 °C pure water and LiOH solution, J. Nucl. Mater. 373 (2008) 343. [3] H.-G. Kim, B.-K. Choi, J.-Y. Park, Y.-H. Jeong, Influence of the manufacturing processes on the corrosion of Zr–1.1Nb–0.05Cu alloy, Corros. Sci. (2009), doi:10.1016/j.corsci.2009.06.023. [4] A. Yilmazbayhan, E. Breval, A.T. Motta, R.J. Comstock, Transmission electron microscopy examination of oxide layers formed on Zr alloys, J. Nucl. Mater. 349 (2006) 265. [5] A. Froideval, C. Degueldre, C.U. Segre, M.A. Pouchon, D. Grolimund, Niobium speciation at the metal/oxide interface of corroded niobium-doped Zircaloys: a X-ray absorption near-edge structure study, Corros. Sci. 50 (2008) 1313. [6] B. Cox, Some thoughts on the mechanisms of in-reactor corrosion of zirconium alloys, J. Nucl. Mater. 336 (2006) 331. [7] P. Billot, J. Robin, A. Giordano, J. Peybernes, J. Thomazet, H. Amaulrich, in: A.M. Garde, E.R. Bradley (Eds.), Zirconium in the Nuclear Industry: Tenth International Symposium, ASTM STP 1245, 1994, p. 351. [8] W. Liu, B. Zhou, Q. Li, M. Yao, Detrimental role of LiOH on the oxide film formed on Zircaloy-4, Corros. Sci. 47 (2005) 1855. [9] T. Karlsen, C. Vitanza, in: A.M. Garde, E.R. Bradley (Eds.), Zirconium in the Nuclear Industry Tenth International Symposium, ASTM STP 1245, 1994, p. 779. [10] P.L. Frattini, J. Blok, S. Chauffriat, J. Sawicki, J. Riddle, Axial offset anomaly: coupling PWR primary chemistry with core design, Nucl. Energy 40 (2001) 23. [11] L.O. Actis-Dato, L. Aldave de Las Heras, M. Betti, E.H. Toscano, F. Miserque, T. Gouder, Investigation of mechanisms of corrosion due to diffusion of impurities by direct current glow discharge mass spectrometry depth profiling, J. Anal. Atom. Spectrom. 15 (2000) 1479. [12] J.S. Moya, M. Diaz, J.F. Bartolome´, E. Roman, J.L. Sacedon, J. Izquierdo, Zirconium oxide film formation on zircaloy by water corrosion, Acta Mater. 48 (2000) 4749.

[13] J. Sawicki, Evidence of Ni2FeBO5 and m-ZrO2 precipitates in fuel rod deposits in AOA-affected high boiling duty PWR core, J. Nucl. Mater. 374 (2008) 248. [14] Jim Henshaw, John C. McGurk, Howard E. Sims, Ann Tuson, Shirley Dickinson, Jeff Deshon, A model of chemistry and thermal hydraulics in PWR fuel crud deposits, J. Nucl. Mater. 353 (2006) 1–11. [15] Y. Ding, D.O. Northwood, A study of the interphase structure at the oxidemetal interface in the corrosion of zirconium-based alloys, Corros. Sci. 36 (1994) 259. [16] IAEA-TECDOC-996 Waterside corrosion of zirconium alloys in nuclear power plants, International Atomic Energy Agency, Vienna, 1998, p. 29. [17] L. Durand-Keklikian, G. Cragnolino, D.D. Macdonald, Ex-situ a.c. impedance studies of oxide film growth on zircaloys in high-temperature, high-pressure steam, Corros. Sci. 32 (1991) 347. [18] L. Durand-Keklikian, G. Cragnolino, D.D. Macdonald, In-situ a.c. impedance studies of oxide film growth on zircaloys in high-temperature, high-pressure steam, Corros. Sci. 32 (1991) 361. [19] C. Bataillon, S. Brunet, Electrochemical impedance spectroscopy on oxide films formed on zircaloy 4 in high temperature water, Electrochim. Acta 39 (1994) 455. [20] B. Cox, F. Gascoin, Y.-M. Wong, Properties of thin anodic oxide films on zirconium alloys, J. Nucl. Mater. 218 (1995) 113. [21] B. Cox, Y.-M. Wong, Simulating porous oxide films on zirconium alloys, J. Nucl. Mater. 218 (1995) 324. [22] J.J. Vermoyal, A. Frichet, L. Dessemond, A. Hammou, AC impedance study of corrosion films formed on zirconium based alloys, Electrochim. Acta 45 (1999) 1039. [23] P. Barberis, A. Frichet, Characterization of Zircaloy-4 oxide layers by impedance spectroscopy, J. Nucl. Mater. 273 (1999) 182. [24] J.J. Vermoyal, L. Dessemond, A. Hammou, A. Frichet, In situ characterization of Zircaloy-4 oxidation at 500 °C in dry air, J. Nucl. Mater. 288 (2001) 297. [25] M. Oskarsson, E. Ahlberg, U. Andersson, K. Petersson, Characterisation of pretransition oxides on Zircaloys, J. Nucl. Mater. 297 (2001) 77. [26] S. Forsberg, E. Ahlberg, M. Limbäck, Studies of corrosion of cladding materials in simulated BWR environment using impedance measurements, J. ASTM Int. 4 (2007) JAI101123. [27] G. Nagy, Z. Kerner, T. Pajkossy, In situ electrochemical impedance spectroscopy of Zr–1%Nb under VVER primary circuit conditions, J. Nucl. Mater. 300 (2002) 230. [28] G. Nagy, R. Schiller, Dispersive charge carrier mobility in a surface oxide layer, Phys. Chem. Chem. Phys. 4 (2002) 791. [29] J. Schefold, D. Lincot, A. Ambard, O. Kerrec, The cyclic nature of corrosion of Zr and Zr–Sn in high-temperature water at 633 K. A long-term in situ impedance spectroscopic sstudy, J. Electrochem. Soc. 150 (2003) B451. [30] J.J. Vermoyal, A. Hammou, L. Dessemond, A. Frichet, Electrical characterization of waterside corrosion films formed on ZrNb(1%)O(0.13%), Electrochim. Acta 47 (2002) 2679. [31] R. Schiller, J. Balog, G. Nagy, Continuous-time random-walk theory of interfering diffusion and chemical reaction with an application to electrochemical impedance spectra of oxidized Zr–1%Nb, J. Chem. Phys. 123 (2005) 094704. [32] J. Ai, Y. Chen, M. Urquidi-Macdonald, D.D. Macdonald, Electrochemical impedance spectroscopic study of passive zirconium. I. High-temperature, deaerated aqueous solutions, J. Electrochem. Soc. 154 (2007) C43. [33] J. Ai, Y. Chen, M. Urquidi-Macdonald, D.D. Macdonald, Electrochemical impedance spectroscopic study of passive zirconium. II. High-temperature, hydrogenated aqueous solutions, J. Electrochem. Soc. 154 (2007) C52. [34] M. Bojinov, L. Hansson-Lyyra, P. Kinnunen, T. Saario, P. Sirkia, In-situ studies of the oxide film properties on BWR fuel cladding materials, J. ASTM Int. 2 (2005) JAI12820. [35] M. Bojinov, W. Cai, P. Kinnunen, T. Saario, Kinetic parameters of the oxidation of zirconium alloys in simulated WWER water – Effect of KOH content, J. Nucl. Mater. 378 (2008) 45–54. [36] O. Gebhardt, A. Hermann, Microscopic and electrochemical impedance spectroscopic analyses of zircaloy oxide films formed in highly concentrated LiOH solution, Electrochim. Acta 41 (1996) 1181. [37] J.G. Han, J.S. Lee, W. Kim, D.S. Sun, K.H. Chung, Zirconium oxide formation and surface hardening by nitrogen implantation under oxygen atmosphere in Zircaloy 4, Surf. Coat. Technol. 97 (1997) 492. [38] D.Q. Peng, X.D. Bai, X.W. Chen, Q.G. Zhou, X.Y. Liu, R.H. Yu, Effect of self-ion bombardment on the corrosion behavior of zirconium, Nucl. Instrum. Methods Phys. Res. B 215 (2004) 394. [39] Y.H. Jeong, K.H. Kim, J.H. Baek, Cation incorporation into zirconium oxide in LiOH, NaOH, and KOH solutions, J. Nucl. Mater. 275 (1999) 221. [40] L.O. Actis-Dato, L. Aldave de Las Heras, M. Betti, E.H. Toscano, F. Miserque, T. Gouder, Investigation of mechanisms of corrosion due to diffusion of impurities by direct current glow discharge mass spectrometry depth profiling, J. Anal. At. Spectrom. 15 (2000) 1479. [41] A.J. Markworth, A. Sehgal, G. Frankel, Oxidation of Zr–2.5 Nb nuclear reactor pressure tubes. A new model, J. Electrochem. Soc. 146 (1999) 3672. [42] J. Ai, Y. Chen, M. Urquidi-Macdonald, D.D. Macdonald, Electrochemical impedance spectroscopic study of passive zirconium, J. Nucl. Mater. 379 (2008) 162–169. [43] Y.H. Jeong, K.O. Lee, H.G. Kim, Correlation between microstructure and corrosion behavior of Zr–Nb binary alloy, J. Nucl. Mater. 302 (2002) 9–19. [44] A.T. Fromhold Jr., Theory of Metal Oxidation – Vol. I. Fundamentals, NorthHolland, Amsterdam, 1976.

M. Bojinov et al. / Corrosion Science 52 (2010) 54–67 [45] S. Havriliak, S. Negami, A complex plane analysis of a-dispersions in some polymer systems, J. Polym. Sci. C 14 (1966) 99. [46] M. Bojinov, G. Fabricius, P. Kinnunen, T. Laitinen, K. Makela, T. Saario, G. Sundholm, Electrochemical study of the passive behaviour of Ni–Cr alloys in a borate solution—a mixed-conduction model approach, J. Electroanal. Chem. 504 (2001) 29. [47] R.A. Ploc, Transmission electron nicroscopy of a-ZrO2 films formed in 573 K oxygen, J. Nucl. Mater. 61 (1976) 79–87.

67

[48] J. Kovacs, E.A. Garcia, Diffusion model for the oxidation of zirconium at 573 and 623 K, J. Nucl. Mater. 210 (1994) 78. [49] E.A. Garcia, Dynamical diffusion model to simulate the oxide crystallization and grain growth during oxidation of zirconium at 573 and 623 K, J. Nucl. Mater. 224 (1995) 299. [50] Y.H. Jeong, J.H. Baek, S.J. Kim, H.G. Kim, H. Ruhmann, Corrosion characteristics and oxide microstructures of Zircaloy-4 in aqueous alkali hydroxide solutions, J. Nucl. Mater. 270 (1999) 322.