Corrosion Science 53 (2011) 30753085

Contents lists available at ScienceDirect

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

Dissolution mechanism of 316L in leadbismuth eutectic at 500 C

Eriko Yamaki , Kevin Ginestar, Laure Martinelli
CEA, DEN, DPC, SCCME, Laboratoire dEtude de la Corrosion Non Aqueuse, Bt. 458 PC 50, F-91191 Gif-sur-Yvette, France

a r t i c l e

i n f o

a b s t r a c t
In the frame of the Accelerator Driven System (ADS) cooled by liquid leadbismuth eutectic (LBE), the austenitic stainless steel 316L is considered as a possible structural material for the reactor. However, the corrosion of 316L in this liquid alloy environment can be substantial, especially when a dissolution process occurs. In order to understand the dissolution process and to obtain a modelling of the 316L corrosion rate by LBE, an experimental dissolution kinetics of 316L is carried out in stagnant LBE at 500 C up to 3000 h. A Ni preferential dissolution of the 316L is observed, leading to the formation of a ferritic layer at the 316L surface. A discussion on the various steps occurring in dissolution process leads to the conclusion that only the Ni dissolution reaction rate can control the 316L dissolution kinetics. The dissolution reaction rate constant, kd, calculated from this study experimental points is equal to 4.2 1011 mol cm2 s1. 2011 Elsevier Ltd. All rights reserved.

Article history: Received 8 November 2010 Accepted 8 May 2011 Available online 13 May 2011 Keywords: A. Stainless steel B. Modelling studies C. High temperature corrosion

1. Introduction Leadbismuth eutectic (LBE) is expected as a reactor coolant and a spallation target material for an Accelerator Driven nuclear transmutation System (ADS). One of the main issues for the application is the compatibility of the structural steel with lead-alloy. Austenitic stainless steel 316L is one of the candidates of the structural materials for ADS. The chemical composition of the 316L steel is given in Table 1. In order to proportion the system, the corrosion behaviour of 316L in LBE must be studied. The corrosion behaviour of 316L has been investigated not only in leadbismuth eutectic (LBE) but also in leadlithium (Pb7Li) which is expected as a tritium breeder and a coolant for fusion reactors. Lots of similarities are observed between the corrosion behaviour in LBE and in Pb-17Li. The results of this study are compared to that of literature for 316L dissolution in LBE and in leadlithium alloy. After corrosion test in Pb17Li, the surface of 316L exhibits a ferritic layer caused by the preferential dissolution of Ni into the Pb17Li. In LBE environment, the corrosion behaviour can proceed by two processes depending on the oxygen concentration dissolved in lead-alloy. One is the steel dissolution which leads to the formation of a ferritic layer, and the other is the steel oxidation by the oxygen dissolved in LBE. The 316L can thus be damaged by dissolution

process in low oxygen concentration containing LBE. As the 316L steel is one of the chosen structural materials for lead-alloy cooled facility, the understanding of its dissolution mechanism in lead alloy is necessary to the facility proportioning. In order to understand the dissolution process and to obtain a modelling of the corrosion rate in LBE, more detailed investigations which are performed in LBE at the low oxygen content are needed. Experimental data on the 316L steel corrosion in stagnant LBE at 500 C are reported in this paper. The results are compared with all steps of dissolution process in LBE, and the limiting step of the 316L dissolution rate in LBE is discussed.

2. Literature review on dissolution of austenitic steels in LBE and Pb17Li The corrosion behaviour of 316L in LBE has been studied by several authors [13]. However most of these studies focus on the oxidation process and few concerns the dissolution process. Furthermore, 316L corrosion behaviour in Pb17Li presents many similarities that in LBE. Consequently, in this literature review, data derived from 316L dissolution in PbBi are presented together with data derived from corrosion in Pb17Li. In both cases of corrosion in Pb17Li above 400 C and in LBE at low oxygen content above 450 C, the 316L exhibits a transformation from austenite to ferrite caused by the Ni preferential dissolution within the liquid metal. The dissolution is driven by the solubility of the elements in the liquid alloy at 500 C (as shown in Table 2). In order to compare the present results with the previous data reported from the literature, the following characteristics of the 316L dissolution in lead alloys are discussed in this

Corresponding authors. Address: Tokyo Institute of Technology, Graduate school of Science and Engineering, Department of Nuclear Engineering, N1-18, 212-1, Ookayama, Meguro-ku, Tokyo 152-8550, Japan. Tel./fax: +81 3 5734 2957 (E. Yamaki), tel.: +33 1 69 08 16 13, fax: +33 1 69 08 15 86 (L. Martinelli). E-mail addresses: yamaki.e.aa@m.titech.ac.jp (E. Yamaki), laure.martinelli@cea.fr (L. Martinelli).
0010-938X/$ - see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.corsci.2011.05.031

3076

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085

Nomenclature Symbols Acorr surface area of the corrosion zone at the surface of the sample (cm2) Asample initial surface area of the sample (cm2) ANi Ni atomic weight (g mol1) aS Ni activity in the solid pure Ni Ni aLBE Ni; eq aLBE Ni a316L Ni C Ni C LBE Ni C Ni;0 C LBE Ni C 316L Ni DLBE Ni h h1 h2 Jcorr J316L diff ; Ni Jferr diff ; Ni Jdiss; Ni JLBE diff ; Ni J316L diff ; Ni
LBEFerr Jdiff ; Ni

Ni activity in the liquid LBE at equilibrium Ni activity in the liquid LBE Ni activity in the solid 316L Ni concentration (mol cm3) Ni concentration in the liquid LBE (mol cm3) dissolved Ni concentration at the ferritic layer/LBE interface (mol cm3) Ni concentration in the bulk LBE (mol cm3) Ni concentration of 316L (wt.% or mol cm3) Ni diffusion coefcient in the liquid LBE at 500 C (cm2 s1) thickness of the ferritic layer (lm) mean thickness of the uniform ferritic layer (lm) mean thickness of the tortuous ferritic layer (lm) corrosion rate (g h1, mol h1 or lm year1) Ni diffusion ux in the solid 316L matrix (mol cm2 s1) Ni diffusion ux in the solid ferritic layer (mol cm2 s1) Ni dissolution reaction rate into the liquid LBE at the liquid alloy/ferritic layer interface Ni diffusion ux in the diffusion boundary layer of the liquid LBE (mol cm2 s1) Ni diffusion ux in the 316L (mol cm2 s1) Ni diffusion ux in the LBE contained in the porous ferritic layer (mol cm2 s1)

K k kd kpr M mNi nNi R rNi SLBE Ni T t tW Vcorr x316L Ni y

mass transfer coefcient (m s1) Boltzmann constant (J K1) dissolution rate constant (mol cm2 s1) precipitation rate constant (mol cm2 s1) concentration of the element M (wt.%) Ni mass number dissolved into the liquid LBE (g) Ni mole number dissolved into the liquid LBE (mol) gas constant (J mol1 K1) Ni atomic diameter (nm) Ni solubility in LBE (wt.% or mol cm3) temperature (K) corrosion duration (h or year) duration needed for sample wetting (h or year) volume of the corrosion layer (cm3) Ni mole fraction in the solid 316L distance from the surface (cm) ratio of the ferritic layer P total length divided by the total sample length a Lferr =Ltotal thickness of the LBE diffusion boundary layer (cm) Ni activity coefcient in the solid 316L Ni activity coefcient in the liquid LBE density of 316L (g cm3) viscosity of LBE (Pa s) Ni chemical potential (J mol1) Ni chemical potential in the solid pure Ni Ni chemical potential in LBE at equilibrium, Ni standard chemical potential of formation in the solid pure Ni (J mol1) Ni standard chemical potential of formation in the liquid LBE (J mol1)

a
d

c316L Ni cLBE Ni q316L: gLBE lNi lS Ni lLBE Ni; eq  lS Ni

lLBE Ni

Table 1 Chemical composition of 316L (wt.%). Fe Bal. Cr 16.82 Ni 10.17 Mo 2.08 Mn 1.84 Si 0.64 P 0.026 N 0.025 C 0.016 S 0.003

Table 2 Solubility of metallic elements into liquid lead-alloy at 500 C (wt.%). Liquid metal LBEa Pb-17Li
a b c

Fe 2.2 104 4.7 103b

Ni 2.7 0.36c

Cr 1.1 103 1 10 104b

Si 1.69 105

Mn 2.3 8.5 102c

The values were given by the Ref. [4]. The values were given by the Ref. [5]. The values were given by the Ref. [6].

section:(i) the nature and the structure of the corrosion layer which is formed at the 316L surface in LBE and in Pb17Li, (ii) the corrosion kinetics in LBE and Pb17Li and (iii) the corrosion mechanisms that have been proposed.

2.1. Nature and structure of corrosion layer 316L corrosion in LBE containing low oxygen concentration and in Pb17Li, leads to a porous corrosion layer which has strong Ni depletion and which is constituted of porosities lled with the liquid metal. Simon et al. [7] performed X-ray diffraction (XRD) on

the corrosion layer formed at the surface of 316L in Pb17Li at 500 C. Their results showed that the corrosion layer was identied as a ferritic phase (a structure). This fact indicates that the transformation from the austenitic phase of 316L to the ferritic phase occurs by a preferential dissolution of Ni from the 316L matrix into the liquid metal. Table 3 summarized the ferritic layer compositions available in literature. Depletions of Ni, Mn and Cr in the ferritic layer were found in the ferritic layer in all reports. The Mn and Ni concentrations in the ferritic layer are nearly zero. However, as the Cr concentration in the ferritic layer is different from each experimental condition (liquid metal temperature, corrosion duration, liquid metal ow), it cannot be concluded on which condition affects the element concentration in the ferritic layer. The inuence of the corrosion time to the element concentration of the ferritic layer is discussed in this article. According to Simon et al. [7], the element concentrations in the ferritic layer is homogeneous in case of 316L corrosion in Pb17Li. Barker et al. [13] (corrosion in Pb 17Li at 495 C) and Deloffre et al. [8] (corrosion in LBE at 450 C) noticed, considering the results of microprobe analyses, that the boundary between the ferritic layer and the sound 316L was well dened without intermediate diffusion zone. This fact indicates that the concentration of the elements changes suddenly as a step at the 316L/ferritic layer interface without any diffusion zone, and that the dissolution of elements may start at the 316L matrix/ferritic layer interface. The nature and structure of the dissolution corrosion layer formed on the 316L both in Pb17Li and in LBE seems to have same features and they can be summarized as follows;

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085 Table 3 Mean concentrations of elements in ferritic layer formed at the 316 surface and which have been reported in the literatures (wt.%). Ref. [8] [9] [7] [10] [11] [12] [13] [14] [15] Liquid alloy (LA) LBE LBE PbLi PbLi PbLi Li PbLi PbLi PbLi PbLi PbLi PbLi Test type (velocity of LA (m s1)) Stagnant Semi-stagnant (No data) Stagnant Flowing (0.025) Flowing (0.142) Stagnant Semi-stagnant (0.01) Flowing (0.1) Flowing (0.12) Flowing (0.1) Semi-stagnant (0.01) Temperature  ( C) 450 600 500 600 500 500 600 450 495 450 400 450 400 400 450 Oxygen concentration in the LA (wt.%) 5 107 8 1010 Bubbling Ar-50%O2 Corrosion duration (h) 3000 3000 3000 500 1500 2640 1900 750 6130 4000 3000 3000 3000 3000 900 3000 3000 Fe Ni 0.2 0.2 1.5 0.15 1.9 1.5 3.1 1.4 0.9 2.6 0.7 2.7 0.8 0.37 0.55 0.56 Cr 1 8.0 10 6.6 10.5 16.2 19.5 15.6 16.1 9.8 5.3 5.0 5.5 5.3 2.5 3.7 7.5 Si Mo

3077

Mn

90.0 76.8 74.4 79.7 78.3 85.5 84.1 87.1 88.7 91.3 93.8 93.0 86.4

0.6 1.6 0.7 0.9 1.1 0.5 0.5 0.5 0.5 0.5 0.5 0.52

2.7 3.6 2.4 2.2 3.2 2.3 2.5 2.4 2.0 2.7 2.2 2.4

0.1 0.1 0.4 0.2 0.2 0.01 0.07 0 0 0.03 0.04 0.1

Thickness, h / m

1. the austenite phase is transformed into a ferritic phase due to a the preferential dissolution of Ni; 2. the ferritic layer is porous and lled with the liquid metal; 3. the ferritic layer presents a strong depletion of Ni, Mn and Cr; 4. the concentrations of the elements in the layer are constant and change as a step at the 316L/ferritic layer interface.

50 45 40 35 30 25 20 15 10 5 0 0 1000 2000 3000 4000

Barker 495C [10] Simon 500C [7] Broc 400C [14] Broc 400C_2 [14] Tortorelli 430C [21] Tortorelli 450C [21] Tas 500C loop Tab[11] Tas 500C loop Tube [11] Tas 495C loop [11] Sannier 400C [15]

2.2. Dissolution kinetics The corrosion kinetics of 316L has been determined by two kinds of methods in literature: (i) it is determined with the weight loss of the samples before and after the corrosion test (ii) or with the depth of the ferritic layer after the corrosion tests. In case of the weight loss, the level of removing the residual Pb17Li or LBE after the corrosion tests affects the real weight loss value of the sample. Moreover, since the corrosion layers are porous, it is difcult to remove completely the residual Pb17Li or LBE. Moreover some layers can easily spall off from the matrix. For these reasons, only the dissolution corrosion rate determined with the depth of the ferritic layer will be considered. Figs. 1 and 2 show that the thicknesses of the ferritic layers as a function of the corrosion duration in case of corrosion of 316 type stainless steels in LBE (Fig. 1) and Pb17Li (Fig. 2) around for 450 600 C temperature range. All thicknesses were measured by optical microscope or by SEM.

5000

6000

Corrosion time, t / h
Fig. 2. Thicknesses of ferritic layer as a function of corrosion for corrosion of 316L in Pb-17Li for 450600 C temperature range.

200

Thickness, h / m

150

100

Martin 600C [16] Soler 600C [17] Deloffre 600C [18] Deloffre 500C [19] Deloffre 450C [8] Sapundjiev 570C [20] Sapundjiev 565C [20] Sapundjiev 535C [20] Sapundjiev 490C [20] Sapundjiev 445C [20] Sapundjiev 425C [20]

50

0 0 500 1000 1500 2000 2500 3000

Corrosion time, t / h
Fig. 1. Thicknesses of ferritic layer as a function of corrosion duration for corrosion of 316L in LBE for 450600 C temperature range.

Sapundjiev et al. [20] performed 316L dissolution experiments in static LBE at various temperatures: 425, 445, 490, 535, 565 and 570 C. Their experimental points are presented in Fig 1. They measured the corrosion rate considering either the maximum corroded thickness determined from SEM X-ray images either the liquid metal composition of Ni, Cr or other high soluble elements after the corrosion test. They reported that the corrosion rate is strongly temperature dependent. The obtained corrosion rate was about 5 lm/year at 490 C, considering a linear kinetics for the ferritic layer growth. Barker et al. [13] performed corrosion experiments in Pb17Li thermal convection loop for duration up to 4000 h at 495 C. They reported that the corrosion rate showed linear kinetics. Broc et al. [14] and Sannier et al. [15] made the same observation for corrosion of 316L in Pb17Li at 400 C with a thermal convection loop. Fig. 2 shows that in the 400500 C temperature range, the dissolution of the 316 type stainless steel in Pb17Li shows linear kinetics. As most of authors concluded that the growth kinetics of the ferritic layer is linear, and as it seems also observed on Figs. 1 and 2, a linear regression is made on the experimental data of each experiment: h J corr t t W with h the ferritic layer thickness (lm), Jcorr the corrosion rate (lm year1), t the corrosion duration (year) and tW the duration needed for wetting (year). The corrosion rates derived from literature data (Figs. 1 and 2) and obtained from previous relation are given in Table 4. Fig. 3

3078

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085

Table 4 Corrosion rates obtained by linear regression on literature data presented on Figs. 1 and 2. Ref. [13] [14] [21] Liquid alloy Pb17Li Pb17Li Pb17Li Pb17Li Pb17Li LBE LBE LBE LBE LBE LBE LBE LBE LBE Test type (velocity of LA (m s1) Semi-stagnant (0.01) Flowing (0.1) Thermal convection (0.02) Semi-stagnant Stagnant Temperature (C) 495 400 430 450 400 570 565 535 490 600 600 600 500 450 Corrosion rate (lm year1) 86.7 52.3 29.7 58.4 19.7 502.8 341.6 183.1 5.4 963.6 367.9 613.2 204 65.7

[15] [20]

[16] [17] [18] [19] [8]

Stagnant Stagnant Stagnant Thermal convection

of 316L in LBE. However, some corrosion mechanisms are proposed for dissolution of 316L in Pb17Li. Tortorelli presents corrosion data for the type 316 stainless steel in lithium (at 600 C) and lithiumlead alloy (at 500 C) for durations up to 10,000 h with a thermal convection loop [21]. After exposures to the both liquid metal, the preferential dissolution of Ni and Cr from the steels and the transformation of austenite to ferrite in the surface layers were observed. The weight loss of the samples exposed to both environments shows a paralinear rate law: nonlinear weight loss during the rst few thousand hours, which indicates, according to the author, a diffusion control, and nearly linear behaviour at extended times. They concluded that the 316L corrosion rate in the lithium alloy in 500600 C temperature range is controlled by a mixed process based on reaction kinetics and diffusion. Simon et al. [7] performed corrosion experiments at 500 and 440 C in an anisothermal container containing liquid Pb17Li. They identied three steps that can control the corrosion rate: (i) the supercial dissolution of Ni and Cr; (ii) the phase transformation from austenite to ferrite containing high level of porosities; (iii) the diffusion of the dissolved species in the liquid alloy and in the diffusion boundary layer of liquid Pb17Li. The authors also described that the thickness of diffusion boundary layer is much more important than the corrosion layer one, a linear kinetics is observed, and on the opposite, the growth of the corrosion layer with time leads to a parabolic dependence with the data reported in the reference [14]. However, they didnt discuss that the element dissolution reaction rate to the liquid metal is one of the possibility for controlling the whole corrosion process. The both processes of the diffusion in the diffusion boundary layer and the element dissolution reaction to the liquid metal are discussed in this article. 3. Experiment 316L austenitic steel was used as test materials. Its composition is presented in Table 1. The size of specimens was 10 15 1.5 mm. The surface of specimens was ground using up to #1200 of SiC paper. The corrosion experiment was performed in stagnant LBE at 500 C in the COLIMESTA device [22] which allows to immerse some specimens in pots containing about 7 l of LBE under a controlled atmosphere in a glove box. The radius of the LBE crucible is 10.5 cm and LBE was dispensed to the height of up to 20 cm. Corrosion durations were 170, 430, 1005, 1510, 2010, 2540 and 3045 h. The oxygen concentration in LBE was controlled to be lower than 108 wt.% by using Ar4%H2 sweeping gas and monitored by a specic oxygen sensor made with yttrium stabilized zirconia and containing a Bi/Bi2O3 reference electrode [23,24]. After the corrosion tests, the specimens were immersed in a solution, containing in volume 1/3 of 100% vol CH3COOH, 1/3 of 30% vol H2O2, and 1/3 of 100% vol C2H5OH to remove residual LBE on the surface of the specimens. The samples were examined by optical microscope and scanning electron microscope (SEM) to measure the thickness of corrosion layer and the length of the corrosion layer. The corrosion layers were analysed by energy dispersive X-ray spectrometer (EDS) and wavelength-dispersive spectrometry (WDS). The phase of the corrosion layer was identied by X-ray diffraction (XRD). The concentration of dissolved elements (Fe, Ni, Cr) in the LBE bulk have been measured by ICP-AES, using LBE taken at 500 C after the corrosion test. The results of these measurements are [Ni] = 2.1 103 wt.%, [Fe] = 7.8 104 wt.% and [Cr] = 2.4 104 wt.%.

1000

100

Corrosin rate in Pb-17Li, Jcorr (Pb-17Li) / m year-1

Corrosion rate in LBE, Jcorr (LBE) / m year-1

900 800 700 600 500 400 300 200 100 0 1.1 1.15 1.2 1.25 1.3 1.35 1.4

LBE Pb-Li

90 80 70 60 50 40 30 20 10 0 1.5

1.45

1000/T /K-1
Fig. 3. Corrosion rates of Table 4 versus temperature inverse.

presents these corrosion rates versus the temperature inverse. Table 4 and Fig. 3 emphasize the important discrepancy between the various experimental results. The corrosion rate obtained by Deloffre et al. [19] at 500 C in LBE is three times higher than that of Simon et al. [7] at 500 C in Pb17Li that is more than one order of magnitude higher than that of Sapundjiev et al. [20] at 490 C in LBE. The results coming from Sapundjiev et al. [20] at low temperatures (from 425 C up to 490 C) are much lower than the others. Moreover, Sapundjiev et al. [20] explained that an oxide scale covers the surface of the samples until 500 h of corrosion test for 450 and 490 C. Moreover, the samples were not wetted below 450 C. The corrosion rates derived from Sapundjiev et al. [20] results are thus not considered for temperatures below 500 C. Consequently, the lead alloy nature (Pb17Li or LBE) does not seem to notably affect the corrosion rate of the 316 type stainless steel. Without taking into account the experimental data from Sapundjiev et al. [20] at temperatures lower than 500 C, the corrosion rate as the function of the temperature can be obtained from the overall other data. It leads to:

log J corr lm year1 6:5

2.3. Corrosion mechanism

27705 RT

There are no report that describes the detailed dissolution mechanism of 316L in LBE or which step limits the dissolution rate

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085

3079

Fig. 4. XRD results of the samples: (a) shows the spectrum of sample after 3045 h and (b) shows the close spectrum of 316L before the corrosion test and after various corrosion durations: 170, 2540 and 3045 h.

4. Results 4.1. Nature and structure of corrosion layer The sample exhibited few oxidation zones, dissolved zones and no damaged zones. The thin oxide layer, which thickness was less than 2 lm, was formed on some sample surface for test duration higher than 1005 h. According to the results of XRD performed on the sample after the corrosion test (Fig. 4(a)), the oxide scale has a spinel structure: it is (Fe,Cr)3O4. Fig. 5 shows the SEM image of the sample cross sections after 3045 h. The darker porous areas in the sample are the corrosion zones. Fig. 4(b) presents XRD spectra of 316L surface before corrosion test and after 170, 2540 and 3045 h of corrosion. It shows the presence of peaks of FeCr ferritic phase (fcc) after 170 h of corrosion whereas the peaks are not observed on the 316L sample before the corrosion test (Fig. 4(b)). The peaks corresponding to the ferritic phase increased when the corrosion duration increases.

Therefore, the corrosion layer observed in Fig. 5 can be identied as a FeCr ferritic phase. Fig. 6 shows the WDS concentration prole of the ferritic layer after 3045 h of corrosion. The concentration of each element changes suddenly as a step at the 316L/ferritic layer interface and these concentrations are constant in the ferritic layer. The Ni and the Mn are strongly depleted in the ferritic layer: their concentrations are almost equal to zero. The concentrations of Cr and Mo are lower in the ferritic layer than in the 316L matrix. It means that these elements (Cr and Mo) dissolved in LBE. WDS proles are performed across the ferritic layer for three corrosion time (430, 1050 and 3045 h) and EDX proles are performed for each corrosion duration. The concentrations of Cr and Mo in the ferritic layer are respectively around 5 and 1 at.% whatever the corrosion durations from 170 to 3045 h. Fig. 7 presents the element concentrations in the ferritic layer, determined with WDS measurement, as a function of the corrosion duration. Fig. 7 shows that the concentrations of Fe, Cr, Ni, Mn, Mo and Si are, respectively about 93, 5, 0, 0, 1 and 0.7 at.% whatever the corrosion duration.

3080

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085

BSE
LBE Ferritic layer

80

Ratio of elements in ferritic layer

70 60 50 40 30 20 10 0 0

Fe/Si Fe/Cr

Uniform layer Uniform layer Tortuous layer

316L

100 m

(a)
LBE
L1 L2 L3

500

1000

1500

2000

2500

3000

Time, t / h
No corrosion Ferritic zone 316L matrix

(b)
Fig. 5. SEM image of cross sections of sample after 3045 h: (a) shows the shape of the ferritic layer and (b) shows how to measure the length of the ferritic layer.

Fig. 8. Ratios of Fe/Cr and Fe/Si (at.%) in ferritic layer, which were calculated with the mean concentration of the elements in the ferritic layer, as a function of corrosion duration.

0.9

Ratio of corroded surface area,

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 500 1000 1500 2000 2500 3000

Ferritic layer
100 90 80 70 60 50 40 30 20 10 0 0 5 10 15 20 25
Cr Ni Mn Fe Si Mo

316L
2.5

Concentration (Si, Mo and Mn) / at%

Concentration (Fe, Cr and Ni) / at%

1.5

Time, t / h
Fig. 9. Ratio of total ferritic layer length (measured on the sample cross section), RLferr , divided by the sample cross section length , Ltotal, a = RLferr/Ltotal, as a function of the corrosion duration.

0.5

0 30

Distance / m
Fig. 6. Microprobe WDS proless of cross section of at 316L sample immersed in oxygen free LBE for 3045 h at 500 C.

Concentration of Si, Cr, Mn, Ni and Mo / at%

Si Ni

Cr Mo

Mn Fe

95 94 93

The ratios of Fe/Cr and Fe/Si (at.%) in the ferritic layer, which are calculated with the mean concentration of the elements in the ferritic layer, determined with EDX and WDS measurements, as a function of corrosion duration are shown in Fig. 8. They are roughly constant whatever the corrosion duration. This fact indicates that either no dissolution of the ferritic layer occurs after its formation or the dissolution of the ferritic layer is uniform. All the elements dissolve with the same rate. 4.2. Corrosion rate

Concentration of Fe / at%

5 4 3 2 1

92 91 90 89 88 87 86

0 0 500 1000 1500 2000 2500 3000

85 3500

Time, t / h
Fig. 7. Element concentrations in uniform ferritic layer, which were determined with the WDS measurement, as a function of the corrosion duration.

Fig. 5(a) shows two different types of ferritic layers. The rst one is a ferritic layer with a uniform thickness (called uniform ferritic layer). The 316L/ferritic layer interface is nearly parallel to the initial surface of the sample. This uniform layer is observed on all the samples after corrosion tests. The other one is a ferritic layer which thickness is not uniform and is thicker than the rst one; the 316L/ferritic layer interface is tortuous in that case (called tortuous ferritic layer). In this corrosion area, some cracks are observed. As the thickness and the shape of these two types of ferritic layers are much different, their thicknesses were individually measured. Fig. 5(b) shows that the ferritic layer is not continuous. Some zones are not corroded. On a sample cross section, the length of each corroded zone is measured (Fig. 5(b)). The total length of the ferritic layer is then calculated and noticed RLferr . The ratio,

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085

3081

70 60

f rm lay a er h1 unifo h2 not un u ifo f rm lay a er

(A)
LBE Diffusion boundary layer Ferritic layer

(B)
LBE Jd if f , M LBE(Ferr) Jd iff, M

Thicknness, h / m

50 40 30 20 10 0 0 1000 2000 3000

LBE Jd iff, M J diss, M

ferr Jd iff, M

h2 (t ) = 1.7 10 2 t 1.9
316L

J diss, M
316L Jd iff, M

h1 (t ) = 7.3 10 3 t 0.1

Fig. 11. Schematic gure of 316L dissolution process.

Time, t / h
Fig. 10. Mean thickness of uniform ferritic layer, h1, and maximum thickness of not uniform ferritic layer, h2, as a function of corrosion duration.

, of the total ferritic layer length, RLferr , over the sample length , Ltotal , is shown in Fig. 9 as a function of the corrosion duration. Assuming the one dimension approximation, this ratio, a, corresponds to the ferritic area divided by the total sample area. It corresponds thus to the relative corroded area of the sample. This relative corroded area, a, increases with increasing the corrosion duration. This fact indicates that the ferritic zone grows laterally. The mean thickness of the uniform ferritic layer, h1, and the maximum thickness of the tortuous ferritic layer, h2, are shown in Fig. 10 as a function of the corrosion duration. The results show that both thicknesses, especially h1, increased linearly as a function of time. A linear t performed on the experimental points leads to the following linear regressions: h1 = 7.3 103t0.1 and h2 = 1.7 102t1.9 (with h in lm and t in h). It means that the 316L matrix begins to dissolve after an incubation time which corresponds to the time needed for wetting and for reduction of the native oxide layer. This incubation time is equal to 14 h (0.1/ 7.3 103) considering the linear regression for h2, and equal to 112 h (1.9/1.7 102) considering the linear regression for h2. The corrosion rates, dh/dt, are 64 lm/year in case of h1 and 147 lm/year in case of h2. These values and linear kinetics agree with the values which have been reported in case of the corrosion in Pb17Li as described previously in Fig. 2. Indeed, the corrosion in Pb17Li at 495500 C showed linear kinetics and the corrosion rates are 80 lm year1 by Barker et al. [13] and Simon et al. [7]. Moreover, Fig. 3 shows that the obtained experimental corrosion rate is in agreement with the others. Indeed, the corrosion rate obtained from the linear regression of Fig. 3 derived from literature corrosion rate is 154 lm year1 at 500 C.
Ltotal

Lferr

5. Discussion The different uxes involved in the 316L dissolution can be schematized in Fig. 11. Two distinct dissolution processes, presented on Fig. 11(A) and (B) can occur. The process (A) considers that the diffusion uxes across the ferritic layer are performed via solid phase through the solid lattice or the solid grain boundaries. The process (B) considers that the diffusion uxes across the ferritic layer are performed via liquid penetrations of lead and bismuth within the ferritic layer porosities. The process (A) has four steps. The rst one is the diffusion of the metallic elements (M) in the solid 316L matrix, J 316L diff ; M . The second is the diffusion of the metallic element in the solid ferritic layer, J ferr diff ; M . The third is the dissolution of the metallic elements into the liquid LBE at the liquid alloy/ferritic layer interface, J diss; M . The forth is

the diffusion of M in the concentration boundary layer of liquid LBE, J LBE diff ; M . On the other hand, the process (B) has also four steps; the rst one is the diffusion of M in the 316L, J 316L diff ; M . The second is the dissolution of M from the 316L into the LBE contained in the porous ferritic layer, J diss; M . This step occurs at the 316L/ferritic layer interface. The third is the diffusion of M in the LBE contained LBEFerr in the porous ferritic layer, J diff ; M . The forth is the diffusion of M in the boundary layer of the liquid LBE, J LBE diff ; M . Whatever the corrosion process, as there is no concentration gradient in the 316L near the 316L/ferritic layer interface, it can be concluded that the element diffusion ux in the 316L, J 316L diff ; M , does not control the corrosion process. Considering that the process (A) occurs, the corrosion rate can then be limited either by the diffusion ux in the ferritic layer, LBE J ferr diff ; M , or by one ux J diss; M or J diff ; M at the ferritic layer/liquid alloy interface. It was previously mentioned that the element concentration prole decreases as a step at the 316L/ferritic layer interface without any concentration gradient in the ferritic scale. This observation leads to assume that there is no diffusion ux in the solid ferritic layer, J ferr diff ; M . Moreover the nickel concentration is equal to zero in the ferritic layer. This fact indicates that the Ni dissolution starts at the 316L/ferritic layer interface. Since the ferritic layer is very porous, LBE can easily penetrate within the ferritic layer until the 316L/ferritic layer interface. Therefore, it can be considered that the process (B) is dominant in the case of the dissolution corrosion of 316L. The corrosion rate can then be limited either by the dissolution ux of M from the 316L into the LBE contained in the porous ferritic layer, J diss; M or by the diffusion ux of M in the LBEFerr LBE contained in the porous ferritic layer, J diff ; M or by the diffusion ux of M in the boundary layer of the liquid LBE, J LBE diff ; M . According to Balbaud et al. [25] if the corrosion rate is controlled by the diffusion ux of M in the liquid Pb-Bi penetrations in the porous ferritic layer, the corrosion kinetics must be parabolic whereas if it is controlled by the dissolution reaction rate, J diss; M , or by the diffusion ux of M in the boundary layer of the liquid LBE, J LBE diff ; M , the corrosion kinetics must be linear. The results shown in Fig. 10 indicate that the 316L dissolution kinetics is linear. Consequently, the dissolution rate cannot be limited by the diffusion in the LBE penetration within the ferritic layer but it can be limited by the dissolution reaction rate or by the diffusion of the dissolved elements in the LBE diffusion boundary layer (Figs. 11 and 12). In order to determine the limiting step of the 316L dissolution process, the nickel diffusion rate in the LBE diffusion boundary layer and the nickel dissolution reaction rate from the 316L at the 316L/LBE penetrations interface are calculated and discussed as described below.

5.1. Ni diffusion in LBE diffusion boundary layer Considering that the corrosion kinetics is controlled by the Ni diffusion in the LBE diffusion boundary layer (Fig. 12), the diffusion coefcient of Ni in LBE, DLBE Ni , is calculated and then compared to the

3082

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085

CNi

Diffusion boundary layer

Bulk LBE

where Asanple is the initial surface area of the sample (1.5 cm2). Consequently,

J corr

CNi,0

LBE Jd iff, Ni

1 C 316L 1 daht Ni wt:% q316L ANi 100 at dt

10

And then using the Eqs. (2), (5), (8) and (10), the DLBE Ni can be calculated as:
LBE C Ni

Ferritic layer LBE

DLBE Ni

J corr C Ni;0 C LBE Ni

316L 1 C Ni wt:% 100 ANi

daht q316L a1 t d t

C Ni;0 C LBE Ni

11

Fig. 12. Schematic gure of Ni diffusion in diffusion boundary layer of liquid LBE [5].

reference values in order to estimate the reliability of this hypothesis. After Ni dissolves from the 316L to the liquid LBE penetrations, Ni diffuses very quickly in the liquid LBE penetrations across the ferritic layer and then diffuses through the boundary layer to the bulk LBE. The corrosion rate, Jcorr, which is considered to be equal to the Ni diffusion ux in the boundary layer, J LBE diff ; Ni , is expressed with Ni mole number dissolved into the liquid LBE according to the following relation:

J corr

1 dnNi J LBE diff ; Ni Acorr dt

Moreover, the diffusion ux can be calculated according to the second Ficks law:

In the following all the values necessary to the calculation of 1 daht DLBE has to be estimated. Ni are determined. Firstly, the ratio at dt Therefore, the experimental product ah is represented as a function of time in Fig. 13. It is noticed that Fig. 13 is the product of the values a and h, respectively, presented in graphics Fig. 9 and Fig. 10. Fig. 11 shows that ah1(t) is a line, which can be tted by the following equation ah1 = 5.7 103 (lm h1) t 1.9. ah2(t) presents more scattered points, but 4 points out of 6 are close the ah1 line. Consequently, for the calculation of DLBE Ni , the ah1 valdaht ues are taken. As the ratio a1 is roughly constant when the t dt corrosion duration increases (Fig. 13) the maximum value of this ratio is taken for the calculation of DLBE Ni in order to the maximize Fig. 13 DLBE calculated value. Consequently it will be considered Ni 1 daht for the calculation of DLBE = 8.6 103 lm h1. Ni that: at dt Secondly, the Ni concentration in the bulk LBE, C LBE Ni increases with the corrosion duration because Ni dissolves into LBE from the corroded region of 316L. Therefore, Ni concentration in LBE at the time t2 can be obtained using the experimental results as;
LBE C LBE Ni t 2 C Ni t 1 dah C 316L Ni mol cm t 2 t 1 nAsample dt V LBE 3

C Ni DNi ~ J LBE gradlNi lNi lo RT ln cC Ni diff ;Ni RT

LBE

12

The y axial component (perpendicular to the sample surface) of ~ JLBE diff ;Ni is:
LBE J LBE diff ; Ni DNi

@ C Ni @y

Considering that a steady state is reached, the concentration of Ni in the boundary layer does not depend on time but only on the y C Ni C Ni C Ni variable. Consequently, @@ dd , and the derivative dd can be y y y written as the difference between the dissolved Ni concentration at the ferritic layer/LBE interface, CNi, 0 ,and the Ni concentration in the bulk LBE, C LBE Ni , over the thickness of the LBE diffusion boundary layer d. The Eq. (4) can be written as follows:
LBE J LBE diff ; Ni DNi
DLBE Ni d

where t2 and t1 are time (h), n is the number of specimens which are ah exposured in LBE during the duration from t1 to t2. The ratio of dd t can be estimated as 5.7 107 cm h1 by Fig. 13. When t is equal to zero, the Ni concentration in LBE is 5.7 108 mol cm3 which was measured in the LBE before experiment by ICP-AES. Using the equation (13), the Ni concentrations in LBE during the corrosion test were determined as listed in Table 5. If the corrosion kinetics is limited by the Ni diffusion in the LBE diffusion boundary layer, the Ni concentration at the 316L/LBE penetrations in the ferritic layer is the concentration at equilibrium between the Ni in the solid 316L matrix and the dissolved Ni into

C LBE Ni C Ni;0 d

The ratio is called mass transfer coefcient and noted K (ms1). JLBE can be written with K as: diff ; Ni
LBE J LBE diff ; Ni K C Ni; 0 C Ni

On the other hand, the corrosion rate is given by:

J corr

1 dnNi 1 1 dmNi 1 gh Acorr dt Acorr ANi dt

where A(Ni) is the Ni atomic weight, Acorr is the surface area of the corrosion zone, which is equal to aAsample and mNi is the weight loss only due to the Ni dissolution. mNi can be determined by:

mNi t

C 316L Ni wt:% q316L V corr t 100

Vcorr is dened with the experimental values as follows:

V corr Acorr h Asample ah

Fig. 13. The experimental product ah and the ration dah1/adt as a function of the corrosion duration.

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085 Table 5 Ni concentration in LBE during corrosion test. Corrosion time (h) 0 170 430 1000 1510 2010 2540 3054 Number of specimens 12 11 9 8 7 6 4
3 C LBE Ni (mol cm )

3083

5.68 108 5.91 108 6.24 108 6.82 108 7.27 108 7.67 108 8.03 108 8.26 108

Table 6 List of numerical values for the calculations of DLBE Ni . The Ni concentration in LBE was used the value determined by ICP-AES after the corrosion tests. In order to the calculation the Ni concentration in 316L in mol cm3, 7.9 g cm3 was used as the density of 316L. C Ni;0 (mol cm3) 4.3 104 C LBE Ni (mol cm3) 3.6 106 C 316L Ni (wt.%) 10.17
at
1 daht dt 1

(lm h

d (cm) 10.5

DLBE Ni (cm2 s1) 8.2 108

8.6 103

the LBE penetrations, C LBE Ni; eq . Moreover, since the Ni diffusion in the LBE penetrations in not limiting process, this diffusion is thus very quick in comparison with the Ni diffusion in the LBE diffusion boundary layer. Consequently, the dissolved Ni concentration at the ferritic layer/LBE interface, CNi,0 can be considered equal to C LBE Ni; eq which can be estimated by the following relations. At the 316L/LBE penetrations interface, the equilibrium of the dissolution reaction Nis ( ) Nid is reached. The Ni chemical potential in the solid pure Ni, lS Ni , is then equal to the one in LBE, lLBE : Ni; eq
LBE lS Ni lNi; eq

of Ni in the diffusion boundary layer in LBE is a limiting factor of the whole corrosion rate of 316L can be obtained as 8.2 108 cm2 s1. The experimental value of DLBE Ni has not been reported in the literature. The reference values given by StokesEinstein equation, SutherlandEinstein equation and the experimental Fe diffusion coefcient in pure lead reported by Robertson et al. [26] can be compared. StokesEinstein equation and SutherlandEinstein equation are expressed by:

DLBE Ni

kT Npr Ni gLBE

20

13

And considering that the reference states for the solid Ni is the pure solid Ni and for the dissolved Ni is the Ni solubility limit, the previous equation becomes:
S LBE lS RT ln aLBE Ni; eq Ni RT ln aNi lNi
 

where N is a constant equal to 6 for the StokesEinstein equation and to 4 for the SutherlandEinstein equation. Nhe viscosity of LBE, gLBE, is given by [4]:

gLBE Pa s 4:94 104 exp

14

 754:1 T

21

In case of the pure Ni, aS Ni = 1 and the dissolved Ni concentration in LBE is equal to its solubility limit. Consequently, according to the S LBE considered reference states aLBE Ni; eq = 1 and then lNi lNi . In case of 316L, at equilibrium, the following equality can thus be written:

The experimental Fe diffusion coefcient in pure lead as a function of the temperature is given by [26]:

  43932 2 DPb s-1 4:9 103 exp Fe cm RT

22

a316L aLBE Ni Ni; eq

15

The Ni activity in the solid 316L can be written with the Ni activity coefcient in the solid 316L, c316L and the Ni mole fraction Ni in the solid 316L, x316L Ni .
316L 316L a316L aLBE Ni Ni; eq cNi xNi

16

The reference values of DLBE Ni at 500 C are listed in Table 7. The reference values are much higher than the Ni diffusion coefcient calculated with the experimental results on the assumption that the diffusion of Ni in the diffusion boundary layer in LBE is limiting factor of the whole corrosion rate of 316L. Consequently, the Ni diffusion rate in the diffusion boundary layer is too fast to control the 316L corrosion kinetics.

Considering the Ni solubility limit as reference state for the dissolved Ni, the Ni activity in the liquid LBE is written by:

aLBE Ni c

LBE LBE C Ni Ni SLBE Ni

5.2. 316L dissolution mechanism and calculation of Ni dissolution rate constant The proposed 316L dissolution mechanism is described in paragraph 6, as shown in Fig. 11(B), and summarized by three successive steps: rstly Ni dissolves from the 316L matrix into the liquid LBE penetrations at the 316L/ferritic layer interface, then Ni diffuses in the LBE penetrations across the ferritic layer and nally Ni diffuses in the liquid LBE diffusion boundary layer. For the following reasons, it has been concluded that the 316L dissolution is limited by the Ni dissolution reaction rate into the liquid LBE penetrations, at the 316L/ferritic layer interface:

with cLBE Ni 1
SLBE Ni , is given by [27]:

17

The Ni solubility in LBE,

log SLBE Ni wt:% 1:7 0:08

1009 T K

18

Assuming that the activity coefcient c316L and cLBE Ni Ni are equal to 1, the dissolved Ni concentration at equilibrium with the 316L matrix is given by
316L LBE C Ni;0 C LBE Ni; eq xNi SNi

19
Table 7 List of reference values of DLBE Ni .
2 1 DLBE Ni (cm s )

Finally, since the thickness of the diffusion boundary layer is considered to be less than the radius of the LBE crucible (10.5 cm), this value has been used as the maximum of the thickness of the diffusion boundary layer. The numerical values used for the calculation of DLBE Ni are listed in Table 6. The diffusion coefcient of Ni in LBE calculated on the assumption that the diffusion

StokesEinstein equation SutherlandEinstein equation DPb Fe by Robertson

3.5 105 5.2 105 5.3 106

3084

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085

Dissolution rate constant, kd / mol cm-2s

1. the 316L experimental dissolution kinetics is linear which indicates that the dissolution rate cannot be limited by the diffusion in the LBE penetration within the ferritic layer; 2. the Ni diffusion in the LBE diffusion boundary layer is too fast to control the 316L corrosion kinetics according to the calculations presented in paragraph 5.1. Consequently, the kinetics constant of the Ni dissolution reaction can be calculated considering our experimental corrosion results.

7E-11 6E-11 5E-11 4E-11 3E-11 2E-11 1E-11 0 0 500 1000 1500 2000 2500 3000 3500

J diss; Ni J corr

23

The Ni dissolution reaction into liquid LBE is shown as follows:

Nis ( ) Nid

24

where Ni(s) is Ni in the solid 316L and Ni(d) is the dissolved Ni in the liquid LBE.The dissolution rate, Jdiss,Ni is given by:

Time, t / h
Fig. 14. Calculated kd as a function of corrosion duration.

J diss; Ni kd a316L kpr aLBE Ni Ni

25

At the equilibrium, since the dissolution ux is equal to zero and aLBE is equal to the Ni activity in LBE at the equilibrium, Ni aLBE Ni; eq , the Eq. (25) becomes:

kd a316L kpr aLBE Ni Ni; eq

26

Introducing the Eqs. (26) into (25), the dissolution ux can be written as:

J diss; Ni kd a316L 1 Ni

aLBE Ni aLBE Ni;eq

! 27

Using the same reference states for Ni activities in solid and in LBE as proposed in paragraph 5.2, aLBE Ni; eq is given by the Eq. (16):
316L 316L aLBE c316L Ni; eq aNi Ni xNi

28

Considering the Eq. (19) and assuming that the activity coefcient c316L and cLBE Ni Ni are equal to 1, the Eq. (27) is simplied into:

J diss; Ni kd x316L 1 Ni

C LBE Ni
LBE x316L Ni SNi

! : 29

Fig. 15. Arrhenius plot of log kd. If the ratio a is equal to 1 as a function of 1000/T with the reference values calculated with the thickness of the ferritic layer measured by the microscope observation considering that the corrosion kinetic is linear.

As the corrosion rate is controlled by the solid Ni dissolution reaction rate into the liquid LBE, Jdiss,Ni can also be determined with the present results by:

J diss; Ni C 316L Ni

at

1 daht -3 with C 316L inmol cm Ni dt

30

In order to compare the calculated value with that obtained from the literature kinetics, the calculation considering that the ratio a is always equal to 1 is described below. If a is equal to 1, the Eq. (31) becomes:

Therefore, the dissolution rate constant, kd, can be calculated by.

kd x316L Ni

1 1 x316L SLBE
Ni Ni

C LBE Ni

 C 316L Ni

dht dt

32

kd

1 1 daht 3   C 316L mol=cm Ni at dt C LBE 316L Ni 1 x316L SLBE xNi

Ni Ni

31

The numerical values used for the calculation of kd are listed in daht Table 8. The ratio a1 is given in Fig. 13. Considering the data in t dt Table 8 and Fig. 13, the calculated kd as a function of time is shown in Fig. 14. The mean calculated kd for h1 is 4.2 1011 mol cm2 s1.

Table 8 List of numerical values for calculation of kd. x316L Ni 0.1

3 C LBE Ni (mol cm ) 3 SLBE Ni (mol cm )

Considering that the ratio a is equal to 1 in our study, the kd value is equal to 2.8 1011 mol cm2 s1 for h1 and 6.5 1011 mol cm2 s1 for h2 . These values are very close to the mean kd value previously calculated. The Arrhenius plot of ln kd calculated with Eq. (32) as a function of temperature with the literature values are shown in Fig. 15. The literature values for dh(t)/dt are calculated with the thickness of the ferritic layer formed in the liquid LBE measured by microscope observations and considering that the corrosion kinetic is linear [8,1321]. The considered corrosion rates are listed in Table 4 and represented (in lm year1) in Fig 3. According to Fig. 15, the Arrhenius equation can be given as:

(mol cm3) C 316L Ni 1.4 102

3.6 106

4.2 102

ln kd 11:6

7:83 104 RT K

33

E. Yamaki et al. / Corrosion Science 53 (2011) 30753085

3085

The activation energy of the Ni dissolution reaction from the 316L can be determined as 7.83 104 J mol1. 6. Conclusion Corrosion tests of 316L in stagnant LBE at 500 C under low oxygen concentration (108 wt.%) were performed. The 316L corrosion leads to the formation of a discontinuous ferritic layer which is obtained by a preferential dissolution of Ni. Some following observations and calculations permit to propose a 316L dissolution mechanism: 1. the Ni concentration in the ferritic layer is equal to zero, therefore Ni dissolution occurs at the interface between the 316L and the liquid LBE penetrations observed within the ferritic layer; 2. there is no Ni concentration gradient in the 316L close to the 316L/ferritic layer interface; the corrosion rate is thus not controlled by the Ni diffusion in the 316L matrix; 3. the dissolution kinetics is linear; the corrosion rate is thus not controlled by the Ni diffusion across the ferritic layer into the liquid LBE penetrations; 4. the Ni diffusion in the diffusion boundary layer of liquid LBE is too fast to control the 316L dissolution kinetics; 5. therefore, only the Ni dissolution reaction rate can control the 316L dissolution kinetics. 6. considering this proposed mechanism, the dissolution reaction constant, kd, is calculated from this study experimental points. It is roughly equal to 4 1011 mol cm2 s1 at 500 C. Taking into account the values derived from the literature in LBE and in Pb17Li, the dissolution rate constant is estimated to 5 1011 mol cm2 s1 at 500 C.

References
[1] F. Barbier, G. Benamati, C. Fazio, A. Rusanov, J. Nucl. Mater. 295 (2001) 149. [2] G. Mller, A. Heinzel, J. Konys, G. Schumacher, A. Weisenburger, F. Zimmermann, V. Engelko, A. Rusanov, V. Markov, J. Nucl. Mater. 335 (2004) 163. [3] A. Aiello, M. Azzati, G. Benamati, A. Gessi, B. Long, G. Scaddozzo, J. Nucl. Mater. 335 (2004) 169. [4] Handbook on leadbismuth eutectic alloy and lead properties, materials compatibility, thermal-hydraulics and technologies, OECD, Nuclear Science, 2007, 106, (chapter 3). [5] T. Flament, P. Tortorelli, V. Coen, H.U. Borgstedt, J. Nucl. Mater. 191194 (1992) 132. [6] M.G. Barker, V. Coen, H. Kolbe, J.A. Lees, L. Orecchia, T. Sample, J. Nucl. Mater. 155157 (1988) 732. [7] N. Simon, A. Terlain, T. Flament, Corros. Sci. 43 (2001) 1041. [8] Ph. Deloffre, A. Terlain, F. Barbier, J. Nucl. Mater. 301 (2002) 35. [9] Ph. Deloffre, A. Terlain, J. Nucl. Mater. 335 (2004) 244. [10] M.G. Barker, T. Sample, Fusion Eng. Des. 14 (1991) 219. [11] H. Tas, P. Lemaitre, Ja. Dekeyser, F. De Schutter, P. Lemaitre, Fusion Eng. Des. 14 (1991) 321. [12] V. Coen, P. Fenici, H. Kolbe, L. Orecchia, T. Sasaki, J. Nucl. Mater. 110 (1982) 108. [13] M.G. Barker, J.A. Lees, T. Sample, P. Hubberstey, J. Nucl. Mater. 179181 (1991) 599. [14] M. Broc, P. Fauvet, T. Flament, J. Sannier, J. Nucl. Mater. 141143 (1986) 611. [15] J. Sannier, M. Broc, T. Flament, A. Terlain, Fusion Eng. Des. 14 (1991) 299. [16] F.J. Martn, L. Soler, F. Hernndez, D. Gmez-Briceo, J. Nucl. Mater. 335 (2004) 194. [17] L. Soler, F.J. Martn, F. Hernndez, D. Gmez-Briceo, J. Nucl. Mater. 335 (2004) 174. [18] Ph. Deloffre, F. Balbaud-Clrier, J. Nucl. Mater. 335 (2004) 180. [19] Ph. Deloffre, F. Barbier, C. Delisle, C. Blanc, CEA report, RT-SCECF 555, Commissariat a lEnergie Atomique, 2000. [20] D. Sapundjiev, S. Van Dyck, W. Bogaerts, Corros. Sci. 48 (2006) 577. [21] F. Tortorelli, J. Nucl. Mater. 191194 (1992) 965. [22] J.L. Courouau, P. Deloffre, R. Adriano, J. Phys. IV France 12 (2002) Pr8141. [23] V. Ghetta, F. Gamaoun, J. Fouletier, M. Hnault, A. Lemoulec, J. Nucl. Mater. 296 (2001) 295. [24] Handbook on leadbismuth eutectic alloy and lead properties, materials compatibility, thermal-hydraulics and technologies, OECD, Nuclear Science, 2007, (chapter 4). [25] F. Balbaud-Clrier, L. Martinelli, J. Eng. Gas Turbines Power 132 (2010) 1. [26] W.M. Robertson, Trans. TMS-AIME 242 (1968) 2139. [27] L. Martinelli, F. Vanneroy, J.C. Diaz Rosado, D. LHermite, M. Tabarant, J. Nucl. Mater. 400 (2010) 232.

Acknowledgments The authors want to thank the GEDEPEON French project for its nancial contribution to this work.