Corrosion Science 50 (2008) 3371–3377
Contents lists available at ScienceDirect
Corrosion Science journal homepage: www.elsevier.com/locate/corsci
Microstructural effects on the hydrogen permeation of an Inconel alloy 690 J.M. Zagal a,c, H.F. López b,*, O. Flores a, J.L. Albarran a, L. Martínez a a b c
Centro de Ciencias Físicas, UNAM, P.O. Box 48-3, C.P. 62251, Cuernavaca, Morelos, Mexico Materials Department, University of Wisconsin – Milwaukee, P.O. Box 784, Milwaukee, WI 53201, USA Facultad de Química, UNAM, Ciudad Universitaria, C.P. 04510, Mexico
a r t i c l e
i n f o
Article history: Received 28 February 2008 Accepted 22 August 2008 Available online 2 September 2008 Keywords: A. Nickel B. Hydrogen permeation C. Kinetic parameters C. Stress corrosion
a b s t r a c t In the present work, hydrogen permeability tests were carried out in Inconel 690 in the as-received (AR) condition and after heat treating. The heat treatments promoted total solid solution (SA) or a microstructure full of grain boundary (gb) coverage with M23C6 carbides (A800). first and second polarization hydrogen transients were determined and used in disclosing the role of reversible and irreversible hydrogen trapping. It was found that in the SA condition, the permeation rates were the highest but they were significantly reduced in the AR condition, particularly in the A800 due to the presence of gb M23C6 carbides. Published by Elsevier Ltd.
1. Introduction Inconel alloy 690 which contains 28 to 30 wt. pct. Cr (twice the Cr found in Inconel 600) has been found to exhibit superior stress corrosion cracking (SCC) resistance when exposed to pressurized water reactor (PWR) environments as compared with Inconel alloy 600 [1–8]. Accordingly, this alloy is currently being used or considered in replacing components made of alloy 600. Thus far, there are no reports of SCC failures in components made of alloy 690 exposed to PWR environments. In contrast, alloy 600 has been found to be susceptible to intergranular stress corrosion cracking (IGSCC) under PWR operating conditions, with the effect being exacerbated when the alloy is in the solid solution condition [9–15]. Moreover, the severity of IGSCC is significantly enhanced when hydrogen evolution (i.e., under cathodic conditions) is prevalent in the environment [9,12]. Since hydrogen evolution is generated or involved during crack initiation/growth in Inconel alloys exposed to PWR environments, it is essential to disclose the role of hydrogen in these environments. Extensive experimental evidence indicates that under these conditions, there is a build up of elevated hydrogen concentrations in Ni-based alloys [9,16], well beyond those predicted by Sieverts’s law (upto 20–80 ppm atomic hydrogen [16]). In turn, this suggests the presence of appreciable hydrogen trapping sites which are expected to play a key role on the effective hydrogen diffusivity and potential alloy susceptibility to hydrogen embrittlement.
* Corresponding author. Tel.: +1 414 229 6005; fax: +1 414 229 6958. E-mail address: Lopez@uwm.edu (H.F. López). 0010-938X/$ - see front matter Published by Elsevier Ltd. doi:10.1016/j.corsci.2008.08.042
Among the published works on hydrogen permeation in Inconel 690, Symons et al. [16] found two hydrogen desorption peaks corresponding to reversible and irreversible hydrogen trapping. In their work, they estimated trapping binding energies, Eb of 37 kJ/ mol for irreversible carbide traps of the M23C6 type. Moreover, they found that plastic straining influences Eb to some extent. In the case of irreversible traps, they found that the Eb associated with gb M23C6-matrix interfaces increased from 37 kJ/mol at 0 pct. strain to 41 kJ/mol at 10 pct. plastic strain. To date there is still not enough evidence available in the literature on Inconel alloy 690 to conclusively affirm that this alloy is immune to IGSCC. From the published literature [7], it is apparent that once a crack is initiated by mechanical means, crack propagation apparently occurs in this alloy under PWR environments. Moreover, Sui et al.’s [8] have found that alloy 690 can develop intergranular cracks in U-bend specimens exposed to hydrogensteam conditions. Hence, it is important to establish the interaction between any absorbed atomic hydrogen and the alloy microstructure including plastic straining effects. In this work, the hydrogen permeability properties of Inconel alloy 690 under various heat conditions were investigated. In particular, the apparent hydrogen diffusivities were linked to possible trapping effects associated with the exhibited microstructures. 2. Experimental The Inconel alloy 690 in the as-received condition (AR) was provided by Allvac Corporation as a one inch thick plate stock in a mill annealed condition. The grain structure was equiaxed with an average grain size of 29.6 lm. Table 1 shows the composition of
3372
J.M. Zagal et al. / Corrosion Science 50 (2008) 3371–3377
Table 1 Chemical composition of Inconel Alloy 690 (wt.%) Ni
Cr
Fe
Mn
Si
Cu
Ti
C
61.66
27.52
9.21
0.19
0.50
0.50
0.37
0.05
the Inconel 690 alloy used in this work. Inconel alloy bars were cut from this plate and heat treated at 1150 °C for 15 min (solid solution anneal, SA) followed by oil cooling. In addition, various SA alloy bars were subsequently annealed at 800 °C for 1 h and then oil cooled (A800). Table 2 gives the exhibited mechanical properties of the Inconel alloy in the as-received and heat treated conditions. Microstructural characterization of the Inconel alloy 690 samples (AR, SA and A800) was made by metallographic means followed by electrolytical etching in a 6% nital solution under 4 V for 3 min. Optical and scanning electron microscopy (SEM) including EDS were used for microstructural determinations. Fig. 1 shows the geometry and dimensions of the machined Inconel alloys used as electrochemical permeability membranes in this work. These membranes were prepared by electrical discharge machining all the way down to 500 lm thickness. Final preparation of the membrane surfaces was attained by surface grinding and polishing all the way to mirror finishing. 2.1. Hydrogen permeation testing Hydrogen permeation was electrochemically measured using two cells separated by the Inconel 690 membrane. The technique proposed by Devanathan and Stachurski [17] (G 148-97 standard) was used in this work. Fig. 2 shows schematically the electrochemical system employed for permeability testing. The electrolyte in the loading cell (cathodic side) consisted of 0.1 mol/l H2SO4 with 5 10 5 mol/l As2O3, whereas the oxidation cell side was under a 0.1 N NaOH solution. The membrane area exposed to the electrolyte solutions was 2.230 cm2.
Table 2 Exhibited mechanical properties of Inconel 690 as a function of the heat condition Material condition
UTS (MPa)
ry
rfract
(MPa)
(MPa)
Elongation (%)
Reduction in area (%)
As-received Solid solution 800 anneal
552 504 497
168 132 126
364 225 362
52.2 64.9 61.0
58.4 69.0 49.3
Prior to running any permeability tests, the electrolyte solutions were bubbled with nitrogen for approximately 2 h. A saturated Calomel reference electrode (SCE) and a graphite auxiliary electrode were placed in the oxidation cell (anode), including a nitrogen bubbling system. Both cell compartments were deoxygenated by bubbling nitrogen through the electrolyte solutions before and during testing. The electrochemical system was a 953 ACM Gill-AC computer driven potentiostat, which included a precise thermostatic temperature control at 25 °C ± 0.1. An oxidation potential of +300 mV versus the SCE electrode was established for hydrogen oxidation. Once a static passive state was reached in the oxidation cell, the 0.1 mol / l H2SO4 with 5 10 5 mol / l As2O3 electrolyte solution was introduced in the cathodic cell side and immediately polarized with a cathodic current of 10 mA/cm2. The temperature of the system was adjusted and the system pressurized to follow the hydrogen permeation current as a function of time until steady state was reached. In addition, transient hydrogen flow was monitored through measurements of current permeability transients.
3. Results and discussion 3.1. Exhibited microstructures The microstructures of the as-received and heat treated inconel alloy 690 are shown in Fig. 3. Notice that the grain structure is somewhat inhomogeneous as there are regions consisting of small and large grains. In addition, there are some twins and significant discrete grain boundary (gb) carbide coverage in the AR condition (Fig 3a). The present carbide phases were identified by electron diffraction as M23C6 and they were preferentially found along the gbs. In the SA condition, M23C6 gb carbides were almost totally dissolved resulting in essentially carbide-free gb structures (Fig. 3b). Heat treating at 800 °C for 1 h lead to copious gb precipitation of semi-continuous carbide films some of them of the dendrite/globular type [3,16] as shown in Fig. 3c. In addition, relatively coarse Ti containing phases were found to be present both, at the gbs and within the bulk (Fig. 3d). Fig 4a and b shows a coarse Ti-particle and corresponding EDS peaks in a fractured Inconel membrane. Notice that Ti is the main element in this precipitate. In addition, elemental composition mapping using EDS of the Ti-particles further confirmed that Ti is the main element (Fig. 5a–d). Apparently, these Ti-phases developed during alloy processing as heat treating did not promote any significant dissolution nor coarsening. From the published literature, it is apparent that these phases are Ti-nitrides and not Ti-carbonitrides as suggested by Venkatesh et al., [18] and as evidenced by the observations of this work of a lack of carbon in these phases (see Fig. 5). Table 3 gives the mean precipitate size as well as the estimated precipitate densities found for each of the heat conditions. 3.2. Permeability
Fig. 1. Geometry and dimensions of Inconel 690 hydrogen membranes.
Fig. 6a shows first polarization hydrogen transients for the as received and heat treated membranes obtained under identical electrochemical testing conditions. First permeation transients yield information on the concurrent behavior of reversible and irreversible trapping sites. Notice from these polarization curves that hydrogen transport in the SA condition is relatively fast when compared with either the AR or the A800 conditions. Apparently, in the SA condition, steady state is reached earlier than in any other heat conditions due to the lack of gb M23C6 carbides which are expected to act as effective irreversible hydrogen trapping sites [16]. The resultant steady state permeation fluxes (Jss), as well as effective hydrogen diffusivities (Deff) and hydrogen surface concentrations (Co) were estimated by considering the widely employed
J.M. Zagal et al. / Corrosion Science 50 (2008) 3371–3377
3373
Fig. 2. Electrochemical hydrogen permeation cell assembly.
Fig. 3. SEM micrographs of Inconel alloy 690 in (a) as received condition, (b) solid solution, (c) solid solution followed by 800 anneal and (d) Ti-rich particles and twins in the solid solution condition.
time-lag method described in detail somewhere else [17]. Table 4 gives the estimated permeability Jss, Deff and Co values. Notice from this table that the steady state hydrogen permeation and the corresponding diffusivity and surface concentration exhibit the highest values for the SA condition. In contrast, in the AR condition the
magnitudes of these parameters were the lowest, with intermediate permeability values found in the A800 condition. In particular, in the SA condition the exhibited Deff is 15-fold higher than in the AR condition. Except for the extent of gb coverage which was reduced due to grain growth during heat treating,
3374
J.M. Zagal et al. / Corrosion Science 50 (2008) 3371–3377
Fig. 4. (a) Fractured surface of an Inconel membrane and (b) EDX peaks from a second coarse phase showing Ti as the main element.
the ‘‘background” density of other trapping sites in this alloy is expected to be similar for the three heat conditions. Hence, the reduction in Deff for the AR and 800A can be attributed to the interaction of hydrogen with the M23C6 gb carbides. Apparently, gb carbide coverage is highly effective for irreversible hydrogen trapping. Moreover, a discrete gb carbide coverage (AR) rather than a semicontinuous one (A800) seems to be increasingly effective for initial hydrogen trapping purposes (see Table 4). Neglecting the effect of other trapping sites in the Inconel matrix such as dislocations as they are very similar for the three heat conditions, the dominant irreversible trapping effect is that of gb M23C6 carbides. From the work of Symons et al. [16] it was found that the hydrogen binding energies with M23C6 carbides in Inconel alloy 690 is 37 kJ/mol which is almost twice the one found at either gbs (12–20 kJ/mol [19,20]) or dislocations (15.4 kJ/mol [5]). In addition, Ti-precipitates are expected to be preferential irreversible hydrogen traps as their potential binding energies would easily exceed those for M23C6 carbides [21]. However, there are no reports on the hydrogen binding energies for these phases. In addition, even though the fraction of Ti-phases is relatively small (0.007, see Table 3), they can also play a major role on hydrogen trapping in this alloy. However, the Ti phases are found in the three heat conditions and can be considered as a ‘‘background trap-
ping” feature in this Inconel alloy. Hence, it can be safely assumed that any additional irreversible trapping effects have to be related to the interaction of hydrogen with the M23C6 precipitates. Fig. 6b shows second polarization hydrogen transients for the permeability curves exhibited by the tested permeation membranes. Second hydrogen transients meaning a subsequent hydrogen permeation through the membrane after the charging current was turned off following the first polarization (i.e., after the hydrogen flux returned back to zero following which the current was turned back on). Notice in this case that the irreversible traps which got filled up in the first polarization are expected to retain their hydrogen in subsequent polarizations. Thus, they are not expected to contribute to subsequent permeation kinetics. From the relative positions of the second polarization hydrogen transients it is evident that hydrogen permeation is still relatively fast in the SA condition, even though the Deff is reduced by almost three-fold (see Table 4). In addition, the effective hydrogen diffusivity drastically decreases in the A800 condition (approximately by one order of magnitude). In contrast, in the alloy in the AR condition there is a significant increase in Deff (almost six-fold). Although, these results are not clearly understood, it is likely that in the SA and A800 conditions, once the irreversible traps become saturated with hydrogen, the role of the gbs as fast hydrogen
3375
J.M. Zagal et al. / Corrosion Science 50 (2008) 3371–3377
Fig. 5. (a) Fractured surface of an Inconel membrane and elemental mapping of a rectangular Ti-containing phase showing that other than Ti (d), it does not contain significant amounts of (b) C or (c) Cr.
Table 3 Quantitative metallographic determinations as a function of the heat condition Inconel 690
Average grain size (lm)
Carbide fraction
Ti-precipitate fraction
Mean carbide size (lm)
Mean Ti precipitate size (lm)
As-received Solution annealed 800 anneal
29.03 58 69.2
0.0187 0.01 0.037
0.0072 0.0071 0.0073
1.01 – –
3.4 3.37 3.63
transport paths is effectively reduced. In particular, it is expected that in the A800 condition once the semi-continuous gb carbide film interfaces get saturated with hydrogen, its subsequent permeation is drastically affected. Apparently, hydrogen transport across the membrane thickness is severely hindered due to the interaction of diffusible hydrogen with the semi-continuously hydrogen saturated gb carbide interfaces. In addition, the dendrite and globular interfacial gb carbide morphologies are highly irregular and thus provide relatively large areas for hydrogen trapping. The aforementioned effect is not found in the AR condition, probably as a result of the discrete nature of the gb carbide precipitation, as well as to the increasing amount of gb area. Under these conditions, it is apparent that after saturation of the irreversible trapping sites, the interaction of high energy trapping sites with diffusible hydrogen is not appreciable, thus accounting for the exhibited increase in Deff values. In as much as gbs are low energy irreversible hydrogen traps (Eb = 12–20 kJ/mol [19,20]), it is likely that upon hydrogen exposure (first polarization transients), as they become saturated with
hydrogen, the overall hydrogen diffusivity along the gbs is enhanced. This effect is particularly notorious in the SA condition where most of the gbs are free from carbide particles. However, once the gbs become saturated with hydrogen, they no longer seem to act as effective high diffusivity hydrogen paths. In turn, this can explain the results for the exhibited Deff (see Table 4) obtained from second polarization curves. In general, hydrogen trapping can be described by the McNabb and Foster equation [22]
Do =D ¼ 1 þ ½k=p ðN t Þ
ð1Þ
where Nt is the trap density, Do is the diffusivity in the pure metal (no traps), D is the apparent diffusivity, k and p are the trapping and ¼ kCð1 nÞ pn, where n is release rate parameters defined by dn dt the fraction of traps occupied at time t and C is the concentration of diffusible hydrogen and thus capable of undergoing a trapping reaction. From the above equation when k ffi p, there is no irreversible trapping and D ffi Do. This condition is expected to be attained dur-
3376
J.M. Zagal et al. / Corrosion Science 50 (2008) 3371–3377
1 0.9 0.8 0.7
J/Jss
0.6 0.5 First Polarization_as received First Polarization_Solution-Annealed First Polarization_800-Anneal
0.4 0.3 0.2 0.1 0 0
1000
2000
3000
4000
5000
6000
Time (seconds) 1 0.9 0.8
From the work of Pound et al. [23], the k parameter in Eq. (1) can be directly related to the density of irreversible and the size of the trap defect. Apparently, in this heat condition both, the trap density and mean trap size are relatively large. In particular, the carbide size factor is relatively large due to the dendrite/globular morphologies (p k) and can thus account for the exhibited Deff values. In this case, the interaction between diffusible hydrogen and the hydrogen saturated irreversible gb trapping sites is unavoidable. In the AR condition, irreversible hydrogen trapping as described by the k parameter is not as severe as in the A800 condition and it is consistent with the experimental Deff values (see Table 4). From the outcome of this work is is clear that the most detrimental condition in terms of retaining hydrogen along the gbs, is in the A800 condition. In this case, once irreversible trapping sites consisting of semi-continuous gb carbide interfaces get saturated with atomic hydrogen they effectively hinder diffusible hydrogen transport across the membrane. Yet, the most susceptible microstructures for IGSCC in PWR environments expected in Inconel alloys correspond to the solid solution condition [8–15]. In this condition, hydrogen trapping at the gbs is not as significant as when gb carbides are present. Thus, it is likely that hydrogen in combination with local plasticity events are needed to reach a critical condition which will trigger crack initiation/propagation at the gbs. 4. Conclusions
0.7
J/Jss
0.6 0.5 2nd Polarization_as received 2nd Polarization_Solution-Annealed 2nd Polarization_800-Anneal
0.4 0.3 0.2 0.1 0 0
1000
2000
3000
4000
5000
6000
Time (seconds) Fig. 6. Normalized output of hydrogen flux versus time. (a) First polarization curves and (b) second polarization curves for the Inconel alloy in the AR, SA and 800A conditions.
Table 4 Experimental hydrogen permeability data as a function of the heat condition Inconel 690
Jss (mol/s cm2)
Deff (tlag) (cm2/s)
Co (tlag) (mol/cm3)
First transient As received Solution annealed 800 anneal
2.9 10 6 13.9 10 5 1.67 10 7
1.15 10 7 5.34 10 6 1.93 10 6
0.63 3.14 0.0048
Second transient As received Solution annealed 800 anneal
1.54 10 7 9.75 10 5 2.63 10 6
6.9 10 7 2.5 10 6 2.19 10 7
0.0124 4.67 0.876
ing second polarization permeation curves as most irreversible trapping sites become saturated with hydrogen. However, it does not seem to hold in the A800 condition where most of the gbs are covered with semi-continuous M23C6 carbide films.
1. Initial hydrogen permeation testing in Inconel 690 indicates that the presence of second phases such as M23C6 leads to a reduction in the effective hydrogen diffusivity when compared with the solid solution condition. 2. Second polarization hydrogen transient curves indicate that once the second phases become saturated with hydrogen, the exhibited Deff is reduced by almost an order of magnitude in the A800 heat condition. Apparently, a continuous coverage with hydrogen saturated gb carbide interfaces is highly effective in hindering hydrogen permeation. 3. In the SA condition it is found that the hydrogen permeation rates found from first and second polarization curves exhibited a slight decrease in Deff from 5.34 10 6–2.5 10 6 cm2/s, respectively. Apparently, during first polarization transients, gbs act as fast diffusion paths for hydrogen. However, once the gbs get saturated with hydrogen, subsequent hydrogen permeation (second transients) along these paths is no longer as effective. 4. In the AR condition, a discrete gb coverage with M23C6 carbides is found to be highly effective in reducing Deff when the alloy is first exposed to hydrogen. However, once the irreversible traps become saturated with hydrogen, hydrogen permeability exhibits significant improvements, probably as a result of the lack of appreciable interaction between diffusible hydrogen and irreversible trapping sites. References [1] J.-D. Mithieux, F. Vaillant, D. Buisine, Y. Brechet, F. Louchet, in: International Conference on Corrosion–Deformation Interactions (CDI), Nice France, Eurocorr ’96, vols. IV/IX, 1996, pp. 2-1 2-4. [2] I. Lenartova, M. Habashi, V. Vodarek, J. Galland, M. Tvrdy, L. Hyspecka, in: International Conference on Corrosion–Deformation Interactions (CDI), Nice France, Eurocorr ’96, vols. IV/IX, 1996, pp. 2-1–2-4. [3] P. Yu, H-C. Yao, Corrosion 46 (1990) 391–402. [4] J.J. Kai, G.P. Yu, C.H. Tsai, M.N. Liu, S.C. Yao, Metall. Trans. A 20 (1989) 2057– 2067. [5] S. Smuk, H. Hanninen, Y. Jagodzinski, O. Tarasenko, P. Aaltonen, in: F.P. Ford, S M. Bruemmer, G.S. Was (Eds.), Nineth International Symposium on Environmental Degradation of Materials in Nuclear Power Systems – Water Reactors, TMS, 1999, pp. 59 67.
J.M. Zagal et al. / Corrosion Science 50 (2008) 3371–3377 [6] H. Kawamura, H. Hirano, S. Shirai, H. Takamatsu, T. Matsunga, K. Yamaoka, K. Oshiden, H. Takiguchi, in: F.P. Ford, S.M. Bruemmer, G.S. Was (Eds.), Nineth International Symposium on Environmental Degradation of Materials in Nuclear Power Systems – Water Reactors, TMS, 1999, pp. 601 608. [7] J.B. Ferguson, H.F. Lopez, Metall. Mater. Trans. A 37 (2006) 2471–2479. [8] G. Sui, J.M. Titchmarsh, G.B. Hey, J. Congleton, Corros. Sci. 39 (1997) 565–587. [9] N. Totuska, E. Lunarska, G. Grangnolino, Z. Szklarska-Smialowska, Corrosion 43 (1987) 505–514. [10] G.P. Airey, Corrosion 35 (1979) 129–135. [11] P.G. Shewmon, Y. Shen, Metall. Trans. A. 22 (1991) 1857–1864. [12] C. Shen, P.G. Shewmon, Metall. Trans. A 21 (1990) 1261–1271. [13] T. Magnin, N. Renaudot, F. Foct, Mater. Trans., JIM 41 (2000) 210–218. [14] S. Matshumima, Y. Shimizu, Transactions of the Japan Institute of Metals 24 (1983) 149–153.
3377
[15] H.F. Lopez, A. Mehboob, J.L. albarran, L. Martinez, in: N.R. Moody, A.W. Thompson, R.E. Ricker, G.W. Was, R.H. Jones (Eds.), Conference Proceedings, Hydrogen Effects on Materials Behavior and Corrosion Deformation Interactions, TMS, Moran Wyoming, 2002, pp. 989–998. [16] D. Symons, G. Young, J. Scully, Metall. Mater. Trans. A 32A (2001) 369– 377. [17] M.A.V. Devanathan, Z.O.J. Stachursky, Proc. Roy. Soc. Lond. Ser. A 270 (1962) 90–102. [18] V. Venkatesh, H.J. Rack, Mater. Sci. Technol. 15 (1999) 408–412. [19] D.H. Lassila, H.K. Birnbaum, Acta Metall. Mater. 36 (1988) 2821–2825. [20] J. Yao, S.A. Meguid, J.R. Cahoon, Metall. Trans. A 24 (1993) 105–112. [21] G.M. Pressouyre, I.M. Bernstein, Metall. Trans. A 9 (1978) 1571–1579. [22] A. McNabb, P.K. Foster, Trans. TMS-AIME 227 (1963) 618–627. [23] B.G. Pound, R.M. Sharp, G.A. Wright, Acta Metall. 35 (1987) 263–270.