- Email: [email protected]
Three dimensional atom probe and first-principles studies on spinodal decomposition of Cr in a Co-alloyed maraging stainless steel
Scripta Materialia 121 (2016) 37–41
Contents lists available at ScienceDirect
Scripta Materialia journal homepage: www.elsevier.com/locate/scriptamat
Regular Article
Three dimensional atom probe and first-principles studies on spinodal decomposition of Cr in a Co-alloyed maraging stainless steel Jialong Tian a,c, Wei Wang a,⁎, Lichang Yin a,b,⁎, Wei Yan a, Yiyin Shan a, Ke Yang a a b c
Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China Shenyang National Laboratory for Materials Science, 72 Wenhua Road, Shenyang 110016, China School of Materials and Metallurgy, Northeastern University, Shenyang 110004, China
a r t i c l e
i n f o
Article history: Received 5 January 2016 Received in revised form 21 April 2016 Accepted 22 April 2016 Available online xxxx Keywords: Maraging stainless steels Spinodal decomposition First-principles calculation Three- dimensional atom probe
a b s t r a c t The effect of Co addition on spinodal decomposition of Cr in a new maraging stainless steel was investigated by three dimensional atom probe (3DAP). The concentration profile of Cr and analysis by maximum likelihood method indicated an increase of spinodal decomposition amplitude of Cr in the Co-alloyed maraging stainless steel. The first-principles calculations showed that the increased Fe-Fe ferromagnetic interaction caused by Co addition might facilitate the formation of Cr-rich clusters in bcc Fe. © 2016 Elsevier Ltd. All rights reserved.
The addition of Co in maraging stainless steels could be traced back to 1960's when Pyromet X-12 came on the scene [1]. It is generally believed that Co addition could lower the solid solubility of Ti or Mo in martensite matrix of maraging stainless steels, thus producing an increased volume fraction of Mo or Ti containing precipitates to increase the strength [2,3]. However, it was also proposed that Co would inhibit the dislocation recovery [4] and decrease the sizes of precipitates and martensitic blocks, thus producing a higher secondary hardening [5,6]. Whatever the strengthening mechanism is, it is no doubt that Co addition can lead to an increase in strength of maraging stainless steels. However, it is still unclear how Co addition influences other properties of maraging stainless steel, e.g., corrosion resistant. Recently, the authors have developed a new Co-alloyed maraging stainless steel, which shows a higher strength compared with that without Co [7]. Specifically, when the Co content is 13 wt.%, the tensile strength of the Co-alloyed steel reaches 1928 MPa. Unfortunately, the corrosion resistance of this steel shows an opposite tendency. For example, the Co-alloyed steel with the highest tensile strength showed the worst corrosion resistance. According to previous researches, the poor corrosion resistance of stainless steels always resulted from the Cr-depletion region occurrence [8–11]. It can be concluded that Co should promote the occurrence of Cr-depletion region, which should be come from the spinodal decomposition that will be discussed later. Within the miscibility gap of the FeCr phase diagram, α-α' phase separation could be accomplished by two
⁎ Corresponding authors. E-mail addresses: [email protected] (W. Wang), [email protected] (L. Yin).
http://dx.doi.org/10.1016/j.scriptamat.2016.04.033 1359-6462/© 2016 Elsevier Ltd. All rights reserved.
different paths namely nucleation and growth (NG) and spinodal decomposition (SD) [12]. Since the phase separation found in this work has been demonstrated to be a spinodal decomposition, the phase separation in other literatures is also named spinodal decomposition although this terminology may not be fully accurate for particular conditions. For the spinodal decomposition in Fe-Cr alloys, it has firstly become a subject of intensive investigations due to the so-called “475 °C embrittlement” resulted from spinodal decomposition [13–16]. In order to improve the comprehensive properties including corrosion resistance, the Fe-Cr alloys have been alloyed with Mo, Ni, Co, etc. Thus, much work has been performed to discuss the effect of alloying elements on spinodal decomposition [13,14,17–19]. Also, because of its potential application as a construction material for nuclear reactors, the effect of ion irradiation on spinodal decomposition [20] as well as Cr segregation [21,22] was also investigated. In conclusion, much work has been occupied on the description of experiment results and the relevant negative effect, however, the underlying theoretical evidence was seldom investigated [23]. Thus, exploring the functional mechanism of alloying elements (Co in this study) on spinodal decomposition in Fe-Cr alloy is worthwhile. In this work, the 3DAP technology associated with a maximum likelihood method has been employed to investigate the effect of Co on the spinodal decomposition of Cr in the new Co-alloyed maraging stainless steels. The main reason for the promoted spinodal decomposition of Cr should be due to the Co addition, which was studied based on the firstprinciples calculations. The chemical compositions of the experimental steels with 0, 5 and 13 wt.% Co are presented in Table 1. Except for Co, the contents of other
38
J. Tian et al. / Scripta Materialia 121 (2016) 37–41
Table 1 Chemical compositions of the experimental steels. wt.%
C
Cr
Ni
Mo
Ti
Al
S
Fe
0 wt.% Co alloy 5 wt.% Co alloy 13 wt.% Co alloy
0.002
12.31
5.42
Co 0.02
5.08
0.41
0.05 0.003
P
0.004
76.70
0.003
12.06
5.13
5.05
5.13
0.38
0.05 0.005
0.003
72.49
0.005
12.33
4.55
13.10
5.59
0.41
0.09 0.004
0.002
63.92
alloying elements are the same except for acceptable fluctuation. The experimental steels were firstly melted in a 25 kg vacuum induction furnace and then remelted by vacuum arc melting. The remelted ingot was homogenized at 1523 K and then forged into 25 mm square bars. Specimens for tests and observation were subjected to a solution treatment at 1323 K for 1 h followed by a cryogenic treatment in liquid nitrogen (77 K) for 8 h. Finally, the specimens were aged at 773 K for 0.5, 3, 12, 20, 40 and 100 h, respectively, followed by air cooling. In order to prepare the tips for the 3DAP investigations, square rods of 0.5 mm × 0.5 mm × 15 mm were cut from the aged bulks, and then were etched to sharp needles by a standard two-step electro-polishing technique [24]. 3DAP experiment was performed in a LEAP™ 3000HR atom probe at a base temperature of 60 K with a pulse fraction ratio of 0.2 and a pulse repetition rate of 200 kHz in ultrahigh vacuum of 10−8 Pa. IVAS™ 3.6.2 software was used to perform the atom reconstruction and the data analysis [25]. To quantify the spinodal decomposition evolution as a function of aging time, a maximum likelihood method [26] was applied based on the concentration profile of
Cr. The spinodal decomposition amplitude was finally calculated by computed iteration. Fig. 1 presents 3DAP elemental maps of Cr atoms within analyzed volumes of 30 × 30 × 60 nm3 in all steels aged at 773 K for different times. It can be seen that Cr atoms distributed homogeneously in all steels after cryogenic treatment (Fig. 1a,c,e). However, heterogeneous distribution of Cr atoms can be observed with increasing aging time except for the 0 wt.% Co specimen. Fig. 1a and b show the 3DAP elemental maps of Cr atoms in the 0 wt.% Co specimen, and it can be seen from the two figures that the Cr atoms always distributed randomly even after aging for 100 h. As for the 5 wt.% Co specimen, the decomposition microstructure is observed in the specimen aged for 100 h, as shown in Fig. 1d. However, when the Co content is increased to 13 wt.%, clustering of Cr atoms occurred after aging for 0.5 h, as seen in Fig. 1f. In particular, when the aging time is prolonged to 3 h and 100 h, the decomposition microstructure consisting of Cr-rich zones and Cr-depleted zones can be easily identified, as shown in Fig. 1g and h. The 3D reconstruction provides a good representation of the space distribution of the Cr atoms, however, the results can't unequivocally demonstrate the expected trends of spinodal decomposition. As an illustration, concentration profiles and related concentration frequency distributions of Cr atoms in the three aged steels were calculated. Then, a maximum likelihood method was used to calculate the spinodal decomposition amplitude of Cr atoms based on the Cr concentration profiles. Fig. 2a and b show the corrosion current density and spinodal decomposition amplitude calculated by maximum likelihood method as a function of aging time in the three steels, respectively. It can be seen from Fig. 2 that the corrosion current density and spinodal
Fig. 1. 3DAP chromium atoms mapping of three alloy specimens aged at 773 K for different times. All the analyzed volumes are with the dimensions of 30 × 30 × 60 nm3. (a), (c) and (e) are specimens after cryogenic treatment in 0 wt.% Co, 5 wt.% Co and 13 wt.% Co alloys; (b) and (d) are specimens after aging for 100 h in 0 wt.% Co, 5 wt.% Co alloys; (f), (g) and (h) are specimens after aging for 0.5 h, 3 h and 100 h respectively in 13 wt.% Co alloy.
J. Tian et al. / Scripta Materialia 121 (2016) 37–41
Fig. 2. Variations trends of (a) corrosion current density and (b) spinodal decomposition amplitude as a function of aging time for three alloy specimens aged at 773 K.
decomposition amplitude show approximately similar increasing trend with aging time. These results also demonstrate that spinodal decomposition played a key role on corrosion resistance of the maraging stainless steels. It can be seen from Fig. 2b that the decomposition kinetics of the 5 wt.% Co specimen is much slower than that of the 13 wt.% Co specimen. Decomposition amplitude of 13 wt.% Co specimen increased with aging time and nearly approached a plateau when the aging time reached 20 h. However, no plateau was observed in 0 and 5 wt.% Co specimens even aging time has reached 100 h. It was also observed that the visual spinodal decomposition morphology corresponds to the decomposition amplitude of 0.06–0.08. The addition of alloying elements in Fe-Cr alloys could influence the spinodal decomposition process. Systematic investigations have been performed to reveal the effect of Co on spinodal decomposition in FeCr-Co permanent alloys. Consistent conclusion has also been reached that Co extends the spinodal decomposition region (composition range and temperature range) [13,17,18]. The effect of alloying elements such as Ni, Mn and Cu on spinodal decomposition has also been studied [14,19]. It is concluded that both Ni and Mn enhance the kinetics of spinodal decomposition, in comparison, the effect of Cu is quite weak. The main difference in the three steels is the content of Co element, indicating that Co might be a stimulating effect on the spinodal decomposition of Cr during the aging process. It should be noted that, different from permanent alloy with higher Cr content, maraging stainless steel used in this study has a much lower Cr content (~ 13 wt.%) that is far away from the critical spinodal decomposition concentration (20 wt.%). By now, no one has reported the observation of spinodal decomposition occurred at early aging stage in low-Cr stainless steel.
39
According to the previous studies [27–29], spinodal decomposition of Cr could only be observed after long term aging in low-Cr stainless steels. Thus, Co has a strong enhancing effect on spinodal decomposition. Then, what is the functional mechanism of Co in the low-Cr stainless steel? Aiming to understand the effect of Co on the spinodal decomposition of Cr, the first-principles calculations have been performed to investigate the electronic structure and stability of Fe-Cr-Co alloys with different Cr contents. All calculations were carried out within the framework of density functional theory implemented in the Vienna Ab Initio Simulation Package [30,31], using the generalized gradient approximation of Perdew-Burke-Ernzerhof [32]. The electron-ion interactions were described by using the frozen-core project or augmented wave approach [33,34]. The energy cut-off for plane wave was set to be 300 eV. The Gamma-centered Monkhorst-Pack grids with a 6 × 6 × 6 k-point sampling were used for all calculations. All atoms are allowed to fully relax during geometry optimization with the convergence criteria of 1 × 10−6 eV in total energy. Considering the ferromagnetic and antiferromagnetic characteristics of Fe (Co) and Cr atoms, respectively, spin polarized calculations were performed for all cases. In order to obtain a systematic insight into the development of Cr-rich clusters, i.e., spinodal decomposition, twelve periodic models based on a 3 × 3 × 3 bcc supercell (including 54 atoms) were constructed to represent Fe-Cr-Co alloys with different Cr contents, while with fixed contents of 0 and 5.6 at.% for Co. As examples, four models of the Fe-Cr-Co alloys with 16.7 at.% Cr are shown in Fig. 3, demonstrating four cases with different Cr distributions, dispersed and clustered without (Models A and B) or with (Models C and D) Co addition. According to the 3DAP elemental maps of Co atoms in Fig. 5 of the Supplementary material, it can be seen clearly that Co atoms showed a dispersive distribution in all the specimens at different heat treated conditions. The nearest neighbor distance analysis (NNA) has also demonstrated the random distribution of Co atoms. Thus, for the model with Co addition, Co atoms are added into the supercell with completely dispersive distribution. The arabic numerals presented in these models are used to denote the Fe-Cr-Co alloys with different Cr contents, and for example, the model with arabic numerals 1, 2, and 3 shown in Fig. 3a (c) represents a Fe-Cr-Co alloy with 5.6 at.% Cr and without (with) Co addition. The purposes of these four models used in the calculations and the detailed information of chemical composition of each model are presented in Table 2.Two to six and nine Cr atoms are mixed into the bcc Fe, respectively, to model Fe-Cr-Co alloys with different Cr contents. For each case with a specified Cr content, two configurations are considered: (i) Cr atoms take a clustered distribution with the nearest neighboring distance; (ii) Cr atoms take a dispersed distribution with remote Cr-Cr distance. The cluster-formation energy of Cr in the Fe-Cr-Co alloy was calculated based on the equation: ΔE = EC-ED, where EC and ED are the total energies of Fe-Cr-Co alloy with clustered and dispersed Cr. Therefore, negative (positive) value of ΔE indicates that the Cr atoms
Fig. 3. Atomic positions of Cr, Co and Fe in (a) Model A; (b) Model B; (c) Model C; (d) Model D. Crx, x = 1–9 denotes Cr atom will replace Fe atom at the numbered sites. Models A and C are used to evaluate the configuration with remote Cr-Cr positions. Models B and D are used to evaluate the first nearest neighbors of Cr-Cr interactions.
40
J. Tian et al. / Scripta Materialia 121 (2016) 37–41
Table 2 The purpose of the models used for the calculation and the detailed chemical compositions in each models. Fe(54-m-n)Co(m)Cr(n), m = 0.3 and n = 2.3,5,9 denote the atom compositions in every system. Crx, x = 1–9 denotes the atom labeled by numbers shown in Fig 3. Purpose
Model Systems
Preference of Cr to be dispersed or clustered with different Cr concentrations.
A, B
Preference of Cr to be dispersed or clustered with Co addition.
C, D
Fe52Cr2(Cr1Cr5); Fe51Cr3(Cr1Cr3Cr5); Fe49Cr5(Cr1–Cr5); Fe45Cr9(Cr1–Cr9); Fe49Co3Cr2(Cr1Cr5); Fe48Co3Cr3(Cr1Cr3Cr5); Fe46Co3Cr5(Cr1–Cr5); Fe42Co3Cr9(Cr1–Cr9);
prefer to be clustered (dispersed). The detail calculation results about the cluster-formation energy in each model are given in Table 3 of the Supplementary material. As can been seen in Table 3 of the Supplementary material, the value of ΔE is positive for the Fe-Cr-Co alloys when the number of Cr atoms is less than or equal to 4, then turns to be negative when the number of Cr atoms increases to 5 and more, implying that the Cr atoms prefer to be dispersed (clustered) at low (high) concentration, which can be well explained by the negative (positive) formation energy of Fe-Cr alloy with low (high) Cr content [35]. In order to identify the specific transition concentration of Cr distribution from being dispersed to being clustered in the Fe-Cr-Co alloys, the cluster-formation energies are plotted as a function of the Cr concentration in Fig. 4. Clearly shown in Fig. 4, the calculated ΔE decreases linearly with increasing the concentration of Cr for both Fe-Cr and FeCr-Co alloys. Moreover, the concentration of Cr at the transition point of ΔE from positive to negative is estimated to be 8.8 at.% for Fe-Cr alloy, which decreases to 7.3 at.% for Fe-Cr-Co alloy. This means that the addition of Co promotes the clustering tendency of Cr atoms and facilitates the formation Cr-rich clusters, lowering the transition concentration from dispersed to clustered Cr in Fe-Cr alloy. This result is well agreed with the miscibility gap of α-phase in Fe-Cr-Co alloy [36], which demonstrated that addition of Co extended the difference in concentration between Fe-rich phase and Cr-rich phase. Magnetic effects have been demonstrated to be the origin of segregation, precipitation and phase separation in Fe-Cr alloys [37–41]. Thus, we have further calculated the magnetic moments of the Fe-Cr-
Fig. 4. The cluster-formation energy (green line) and the difference of averaged magnetic moments of Fe between two Cr distributions (red lines) as a function of Cr concentration with and without Co addition. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Co alloys, in order to explore the stimulating mechanism of Co addition for the formation of Cr-rich clusters observed in the present experiment. The detailed calculation results including the magnetic moments and total energies for the Fe-Cr-Co alloys with different Cr contents are presented in Table 3 of the Supplemenatry material. It can be seen from Table 3 of the Supplementary material that for the central atom of Cr cluster, the value of μ Cr is negative when two or three Cr atoms are mixed into the bcc Fe, then changes to positive when five and nine Cr atoms are involved into the bcc Fe, due to anti-ferromagnetic characteristic of Cr and only really central Cr existed for Cr5 and Cr9 clusters. Which can also explain why μ Cr is very close to zero but remains to be negative (see Table 3 of the Supplementary material for details) for the cases of four and six Cr atoms involved in the Fe-Cr (and Fe-Cr-Co) alloys, due to no real central Cr existed for Cr4 and Cr6 clusters considered in this work. This complex situation of the magnetic moments of Cr in different clusters mainly comes from the magnetic frustrations demonstrated in previous work [35]. Interestingly, it was found that the average magnetic moment of Fe atoms (μ Fe ) in Fe-Cr alloys increased due to Co addition for both dispersed and clustered distributions of Cr atoms, as presented in Table 3 of the Supplementary material. This finding also agrees with the magnetic moment enhancement for Fe atoms of Fe-Co clusters or alloys due to alloying with Co [42,43]. In order to understand the relationship between the magnetic interactions and the stability for different space distribution (clustered or dispersed) of Cr atoms in Fe-Cr and Fe-Cr-Co alloys considered in this work, we have plotted the difference of averaged magnetic moments ( Δμ Fe ) of Fe between two Cr distributions, clustered and dispersed, as a function of Cr concentration in Fig. 4. As we can see in Fig. 4 and Table 3 of the Supplementary material, Δμ Fe increases linearly from − 0.033 (− 0.021) to 0.043 (0.067) μ B with increasing the Cr concentration from 3.7 at.% to 16.7 at.% for Fe-Cr (Fe-Cr-Co) alloys. Easy to understand that, the positive (negative) Δμ Fe means stronger (weaker) ferromagnetic interaction, resulting in higher (lower) stability of corresponding alloys with clustered Cr. Clearly, an increment of Δμ Fe (about 0.02 μ B ) due to Co addition was observed for Fe-Cr-Co alloys compared with Fe-Cr alloys. The transition concentration of Cr from negative to positive Δμ Fe decreases from 9.7 at.% for Fe-Cr to 7.6 at.% for Fe-Cr-Co alloys, qualitatively agreed with the stability results of Fe-Cr-Co alloys. One should note that the increased Δμ Fe (~ 0.02 μ B ) due to Co addition is about 1% of the magnetic moment of each Fe atom (2.20 μ B for ferromagnetic ground state of bulk bcc Fe). Considering that the calculated ferromagnetic interaction energy of pure bcc Fe is 0.47 eV per atom, which is the total energy difference between non-magnetic and ferromagnetic state of bulk bcc Fe and has approximately linear relationship with Δμ Fe according to the classic Heisenberg model of ferromagnetic interactions. Therefore, 1% change of μ Fe should result in an energy change of 4.7 meV for each Fe atom, totally 235 meV for the supercell with about 50 Fe atoms used in our DFT calculation. This estimation of the energy change (235 meV) is quantitatively in the same order of the cluster-formation energy differences between Fe-Co-Cr and Fe-Cr alloys, ranging from 100 to 280 meV for most Cr concentrations as shown in Fig. 4. Thus, it is reasonably to conclude that the increased Fe-Fe ferromagnetic interaction resulted from Co addition increases the stability of Fe-Cr-Co alloys with clustered Cr, thus facilitating the formation of Cr-rich clusters. In summary, spinodal decomposition of Cr atoms at early aging stage was firstly observed in a low-Cr maraging stainless steel. Spinodal decomposition amplitude as a function of aging time was calculated by maximum likelihood method associated with 3DAP results. It can be concluded that Co addition promotes the spinodal decomposition of Cr atoms during the aging process. The functional mechanism of Co addition was also studied by the first-principles calculations. The calculated results indicate that the increased Fe-Fe magnetic interaction resulted from Co addition aggravates the tendency to form Cr-rich clusters in bcc Fe.
J. Tian et al. / Scripta Materialia 121 (2016) 37–41
Acknowledgments This work was supported by a funding from the Natural Science Foundation of China (No. 51201160 and 51472249). We acknowledge Dr. Teng Yang for fruitful discussions on the magnetic interaction. The theoretical calculations were performed on TianHe-1 (A) at National Supercomputer Center in Tianjin. Appendix A. Supplementary data
[15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25]
Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scriptamat.2016.04.033.
[26] [27]
References [1] R.F. Decker, S. Floreen, Maraging Steel: Recent Developments and Applications, in:Maraging Steel – The First 30 Years, The Minerals, Metal & Materials Society, Huntington, 1988. [2] W. Sha, A. Cerezo, G.D.W. Smith, Metall. Mater. Trans. A 24 (1993) 1251. [3] Y. He, K. Yang, W.S. Qu, F.Y. Kong, G.Y. Su, Mater. Lett. 56 (2002) 763. [4] G.R. Speich, D.S. Dabowski, L.F. Porter, Metall. Trans. 4 (1973) 303. [5] A.S. Murthy, J.E. Medvedeva, D. Isheim, S.L. Lekakh, V.L. Richards, D.C. Van Aken, Scr. Mater. 66 (2012) 943. [6] D. Coutsourdais, Mem. Rev. De Metallurgie. 58 (1961) 503. [7] Y.C. Li, W. Yan, J.D. Cotton, G.J. Ryan, Y.F. Shen, W. Wang, Y.Y. Shan, K. Yang, Mater. Des. 82 (2015) 56. [8] M. Schwind, J. Kållqvist, J.O. Nilsson, J. Ågren, H.O. Andrén, Acta Mater. 48 (2000) 2473. [9] M.P. Ryan, D.E. Williams, R.J. Chater, B.M. Hutton, D.S. McPhail, Nature 415 (2002) 770. [10] D.E. Williams, Y.Y. Zhu, J. Electrochem. Soc. 147 (2000) 1763. [11] E.G. Webb, R.C. Alkire, J. Electrochem. Soc. 149 (2002) B272. [12] D. Chandra, L.H. Schwartz, Met. Trans. 2 (1971) 511. [13] S. Jin, S. Mahajan, D. Brasen, Metall. Trans. A. 11A (1980) 69. [14] P. Hedström, Y.F. Hu, J. Zhou, S. Wessman, M. Thuvabder, J. Odqvist, Mater. Sci. Eng. A 574 (2013) 123.
[28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43]
41
K.F. Ha, H.M. Zhang, K.L. Jing, Metall. Trans. A. 20A (1989) 2563. K.H. Park, J.C. Lasalle, L.H. Schwartz, Acta Metall. 34 (1986) 1853. H. Kaneko, M. Homma, K. Nakamura, IEEE Trans. Magn. 13 (1977) 1325. F. Zhu, P. Haasen, R. Wagner, Acta Metall. 34 (1986) 457. J.E. Brown, G.D.W. Smith, Surf. Sci. 246 (1991) 285. M.M. Li, M.K. Miller, W.Y. Chen, J. Nucl. Mater. 462 (2015) 214. V. Kuksenko, C. Pareige, P. Pareige, J. Nucl. Mater. 425 (2012) 125. A. Bhattacharya, E. Meslin, J. Henry, C. Pareige, B. Décamps, C. Genevois, D. Brimbal, A. Barbu, Acta Mater. 78 (2014) 394. V.I. Razumovskiy, A.V. Ruban, P.A. Korzhavyi, Phys. Rev. B 84 (2011) 024106. M.K. Miller, A. Cerezo, M. Hetherington, G. Smith, Atom Probe Field Ion Microscopy, Oxford University Press, Oxford, 1996. B. Gault, D. Haley, F.D. Geuser, M. Moody, E. Marquis, D. Larson, B. Geiser, Ultramicroscopy 111 (2011) 448. T.J. Godfrey, M.G. Hetherington, J.M. Sassen, G.D.W. Smith, J. Phys. 49 (1988) C6–421. D.H. Ping, M. Ohnuma, Y. Hirakawa, Y. Kadoya, K. Hono, Mater. Sci. Eng. A 394 (2005) 285. M.K. Miller, K.F. Russell, Appl. Surf. Sci. 94 (95) (1996) 398. K. Stiller, M. Hattestrand, F. Danoix, Acta Mater. 46 (1998) 6063. G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) RC558. G. Kresse, J. Furthmuller, Phys. Rev. B 54 (1996) 11169. J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865. P.E. Blöchl, Phys. Rev. B 50 (1994) 17953. G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758. T.P.C. Klaver, R. Drautz, M.W. Finnis, Phys. Rev. B 74 (2006) 094435. H. Kaneko, M. Homma, K. Nakamura, IEEE T. Magn. 13 (1977) 1325. P.A. Korzhavyi, A.V. Ruban, J. Odqvist, J.O. Nilsson, B. Johansson, Phys. Rev. B 79 (2009) 054202. A.V. Ruban, P.A. Korzhavyi, B. Johansson, Phys. Rev. B 77 (2008) 094436. O.I. Gorbatov, P.A. Korzhavyi, A.V. Ruban, B. Johansson, Y.N. Gornostyrev, J. Nucl. Mater. 419 (2011) 248. A.V. Ruban, V.I. Razumovskiy, Phys. Rev. B 86 (2008) 174111. D.N. Manh, M.Y. Lavrentiev, S.L. Dudarev, J. Compute. Aid. Des. 14 (2007) 159. G. Mpourmpakis, G.E. Froudakis, A.N. Andriotis, M. Menon, Phys. Rev. B 72 (2005) 104417. A. Diaz-Ortiz, R. Drautz, M. Fahnle, H. Dosch, J.M. Sanchez, Phys. Rev. B 73 (2006) 224208.
Contents lists available at ScienceDirect
Scripta Materialia journal homepage: www.elsevier.com/locate/scriptamat
Regular Article
Three dimensional atom probe and first-principles studies on spinodal decomposition of Cr in a Co-alloyed maraging stainless steel Jialong Tian a,c, Wei Wang a,⁎, Lichang Yin a,b,⁎, Wei Yan a, Yiyin Shan a, Ke Yang a a b c
Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China Shenyang National Laboratory for Materials Science, 72 Wenhua Road, Shenyang 110016, China School of Materials and Metallurgy, Northeastern University, Shenyang 110004, China
a r t i c l e
i n f o
Article history: Received 5 January 2016 Received in revised form 21 April 2016 Accepted 22 April 2016 Available online xxxx Keywords: Maraging stainless steels Spinodal decomposition First-principles calculation Three- dimensional atom probe
a b s t r a c t The effect of Co addition on spinodal decomposition of Cr in a new maraging stainless steel was investigated by three dimensional atom probe (3DAP). The concentration profile of Cr and analysis by maximum likelihood method indicated an increase of spinodal decomposition amplitude of Cr in the Co-alloyed maraging stainless steel. The first-principles calculations showed that the increased Fe-Fe ferromagnetic interaction caused by Co addition might facilitate the formation of Cr-rich clusters in bcc Fe. © 2016 Elsevier Ltd. All rights reserved.
The addition of Co in maraging stainless steels could be traced back to 1960's when Pyromet X-12 came on the scene [1]. It is generally believed that Co addition could lower the solid solubility of Ti or Mo in martensite matrix of maraging stainless steels, thus producing an increased volume fraction of Mo or Ti containing precipitates to increase the strength [2,3]. However, it was also proposed that Co would inhibit the dislocation recovery [4] and decrease the sizes of precipitates and martensitic blocks, thus producing a higher secondary hardening [5,6]. Whatever the strengthening mechanism is, it is no doubt that Co addition can lead to an increase in strength of maraging stainless steels. However, it is still unclear how Co addition influences other properties of maraging stainless steel, e.g., corrosion resistant. Recently, the authors have developed a new Co-alloyed maraging stainless steel, which shows a higher strength compared with that without Co [7]. Specifically, when the Co content is 13 wt.%, the tensile strength of the Co-alloyed steel reaches 1928 MPa. Unfortunately, the corrosion resistance of this steel shows an opposite tendency. For example, the Co-alloyed steel with the highest tensile strength showed the worst corrosion resistance. According to previous researches, the poor corrosion resistance of stainless steels always resulted from the Cr-depletion region occurrence [8–11]. It can be concluded that Co should promote the occurrence of Cr-depletion region, which should be come from the spinodal decomposition that will be discussed later. Within the miscibility gap of the FeCr phase diagram, α-α' phase separation could be accomplished by two
⁎ Corresponding authors. E-mail addresses: [email protected] (W. Wang), [email protected] (L. Yin).
http://dx.doi.org/10.1016/j.scriptamat.2016.04.033 1359-6462/© 2016 Elsevier Ltd. All rights reserved.
different paths namely nucleation and growth (NG) and spinodal decomposition (SD) [12]. Since the phase separation found in this work has been demonstrated to be a spinodal decomposition, the phase separation in other literatures is also named spinodal decomposition although this terminology may not be fully accurate for particular conditions. For the spinodal decomposition in Fe-Cr alloys, it has firstly become a subject of intensive investigations due to the so-called “475 °C embrittlement” resulted from spinodal decomposition [13–16]. In order to improve the comprehensive properties including corrosion resistance, the Fe-Cr alloys have been alloyed with Mo, Ni, Co, etc. Thus, much work has been performed to discuss the effect of alloying elements on spinodal decomposition [13,14,17–19]. Also, because of its potential application as a construction material for nuclear reactors, the effect of ion irradiation on spinodal decomposition [20] as well as Cr segregation [21,22] was also investigated. In conclusion, much work has been occupied on the description of experiment results and the relevant negative effect, however, the underlying theoretical evidence was seldom investigated [23]. Thus, exploring the functional mechanism of alloying elements (Co in this study) on spinodal decomposition in Fe-Cr alloy is worthwhile. In this work, the 3DAP technology associated with a maximum likelihood method has been employed to investigate the effect of Co on the spinodal decomposition of Cr in the new Co-alloyed maraging stainless steels. The main reason for the promoted spinodal decomposition of Cr should be due to the Co addition, which was studied based on the firstprinciples calculations. The chemical compositions of the experimental steels with 0, 5 and 13 wt.% Co are presented in Table 1. Except for Co, the contents of other
38
J. Tian et al. / Scripta Materialia 121 (2016) 37–41
Table 1 Chemical compositions of the experimental steels. wt.%
C
Cr
Ni
Mo
Ti
Al
S
Fe
0 wt.% Co alloy 5 wt.% Co alloy 13 wt.% Co alloy
0.002
12.31
5.42
Co 0.02
5.08
0.41
0.05 0.003
P
0.004
76.70
0.003
12.06
5.13
5.05
5.13
0.38
0.05 0.005
0.003
72.49
0.005
12.33
4.55
13.10
5.59
0.41
0.09 0.004
0.002
63.92
alloying elements are the same except for acceptable fluctuation. The experimental steels were firstly melted in a 25 kg vacuum induction furnace and then remelted by vacuum arc melting. The remelted ingot was homogenized at 1523 K and then forged into 25 mm square bars. Specimens for tests and observation were subjected to a solution treatment at 1323 K for 1 h followed by a cryogenic treatment in liquid nitrogen (77 K) for 8 h. Finally, the specimens were aged at 773 K for 0.5, 3, 12, 20, 40 and 100 h, respectively, followed by air cooling. In order to prepare the tips for the 3DAP investigations, square rods of 0.5 mm × 0.5 mm × 15 mm were cut from the aged bulks, and then were etched to sharp needles by a standard two-step electro-polishing technique [24]. 3DAP experiment was performed in a LEAP™ 3000HR atom probe at a base temperature of 60 K with a pulse fraction ratio of 0.2 and a pulse repetition rate of 200 kHz in ultrahigh vacuum of 10−8 Pa. IVAS™ 3.6.2 software was used to perform the atom reconstruction and the data analysis [25]. To quantify the spinodal decomposition evolution as a function of aging time, a maximum likelihood method [26] was applied based on the concentration profile of
Cr. The spinodal decomposition amplitude was finally calculated by computed iteration. Fig. 1 presents 3DAP elemental maps of Cr atoms within analyzed volumes of 30 × 30 × 60 nm3 in all steels aged at 773 K for different times. It can be seen that Cr atoms distributed homogeneously in all steels after cryogenic treatment (Fig. 1a,c,e). However, heterogeneous distribution of Cr atoms can be observed with increasing aging time except for the 0 wt.% Co specimen. Fig. 1a and b show the 3DAP elemental maps of Cr atoms in the 0 wt.% Co specimen, and it can be seen from the two figures that the Cr atoms always distributed randomly even after aging for 100 h. As for the 5 wt.% Co specimen, the decomposition microstructure is observed in the specimen aged for 100 h, as shown in Fig. 1d. However, when the Co content is increased to 13 wt.%, clustering of Cr atoms occurred after aging for 0.5 h, as seen in Fig. 1f. In particular, when the aging time is prolonged to 3 h and 100 h, the decomposition microstructure consisting of Cr-rich zones and Cr-depleted zones can be easily identified, as shown in Fig. 1g and h. The 3D reconstruction provides a good representation of the space distribution of the Cr atoms, however, the results can't unequivocally demonstrate the expected trends of spinodal decomposition. As an illustration, concentration profiles and related concentration frequency distributions of Cr atoms in the three aged steels were calculated. Then, a maximum likelihood method was used to calculate the spinodal decomposition amplitude of Cr atoms based on the Cr concentration profiles. Fig. 2a and b show the corrosion current density and spinodal decomposition amplitude calculated by maximum likelihood method as a function of aging time in the three steels, respectively. It can be seen from Fig. 2 that the corrosion current density and spinodal
Fig. 1. 3DAP chromium atoms mapping of three alloy specimens aged at 773 K for different times. All the analyzed volumes are with the dimensions of 30 × 30 × 60 nm3. (a), (c) and (e) are specimens after cryogenic treatment in 0 wt.% Co, 5 wt.% Co and 13 wt.% Co alloys; (b) and (d) are specimens after aging for 100 h in 0 wt.% Co, 5 wt.% Co alloys; (f), (g) and (h) are specimens after aging for 0.5 h, 3 h and 100 h respectively in 13 wt.% Co alloy.
J. Tian et al. / Scripta Materialia 121 (2016) 37–41
Fig. 2. Variations trends of (a) corrosion current density and (b) spinodal decomposition amplitude as a function of aging time for three alloy specimens aged at 773 K.
decomposition amplitude show approximately similar increasing trend with aging time. These results also demonstrate that spinodal decomposition played a key role on corrosion resistance of the maraging stainless steels. It can be seen from Fig. 2b that the decomposition kinetics of the 5 wt.% Co specimen is much slower than that of the 13 wt.% Co specimen. Decomposition amplitude of 13 wt.% Co specimen increased with aging time and nearly approached a plateau when the aging time reached 20 h. However, no plateau was observed in 0 and 5 wt.% Co specimens even aging time has reached 100 h. It was also observed that the visual spinodal decomposition morphology corresponds to the decomposition amplitude of 0.06–0.08. The addition of alloying elements in Fe-Cr alloys could influence the spinodal decomposition process. Systematic investigations have been performed to reveal the effect of Co on spinodal decomposition in FeCr-Co permanent alloys. Consistent conclusion has also been reached that Co extends the spinodal decomposition region (composition range and temperature range) [13,17,18]. The effect of alloying elements such as Ni, Mn and Cu on spinodal decomposition has also been studied [14,19]. It is concluded that both Ni and Mn enhance the kinetics of spinodal decomposition, in comparison, the effect of Cu is quite weak. The main difference in the three steels is the content of Co element, indicating that Co might be a stimulating effect on the spinodal decomposition of Cr during the aging process. It should be noted that, different from permanent alloy with higher Cr content, maraging stainless steel used in this study has a much lower Cr content (~ 13 wt.%) that is far away from the critical spinodal decomposition concentration (20 wt.%). By now, no one has reported the observation of spinodal decomposition occurred at early aging stage in low-Cr stainless steel.
39
According to the previous studies [27–29], spinodal decomposition of Cr could only be observed after long term aging in low-Cr stainless steels. Thus, Co has a strong enhancing effect on spinodal decomposition. Then, what is the functional mechanism of Co in the low-Cr stainless steel? Aiming to understand the effect of Co on the spinodal decomposition of Cr, the first-principles calculations have been performed to investigate the electronic structure and stability of Fe-Cr-Co alloys with different Cr contents. All calculations were carried out within the framework of density functional theory implemented in the Vienna Ab Initio Simulation Package [30,31], using the generalized gradient approximation of Perdew-Burke-Ernzerhof [32]. The electron-ion interactions were described by using the frozen-core project or augmented wave approach [33,34]. The energy cut-off for plane wave was set to be 300 eV. The Gamma-centered Monkhorst-Pack grids with a 6 × 6 × 6 k-point sampling were used for all calculations. All atoms are allowed to fully relax during geometry optimization with the convergence criteria of 1 × 10−6 eV in total energy. Considering the ferromagnetic and antiferromagnetic characteristics of Fe (Co) and Cr atoms, respectively, spin polarized calculations were performed for all cases. In order to obtain a systematic insight into the development of Cr-rich clusters, i.e., spinodal decomposition, twelve periodic models based on a 3 × 3 × 3 bcc supercell (including 54 atoms) were constructed to represent Fe-Cr-Co alloys with different Cr contents, while with fixed contents of 0 and 5.6 at.% for Co. As examples, four models of the Fe-Cr-Co alloys with 16.7 at.% Cr are shown in Fig. 3, demonstrating four cases with different Cr distributions, dispersed and clustered without (Models A and B) or with (Models C and D) Co addition. According to the 3DAP elemental maps of Co atoms in Fig. 5 of the Supplementary material, it can be seen clearly that Co atoms showed a dispersive distribution in all the specimens at different heat treated conditions. The nearest neighbor distance analysis (NNA) has also demonstrated the random distribution of Co atoms. Thus, for the model with Co addition, Co atoms are added into the supercell with completely dispersive distribution. The arabic numerals presented in these models are used to denote the Fe-Cr-Co alloys with different Cr contents, and for example, the model with arabic numerals 1, 2, and 3 shown in Fig. 3a (c) represents a Fe-Cr-Co alloy with 5.6 at.% Cr and without (with) Co addition. The purposes of these four models used in the calculations and the detailed information of chemical composition of each model are presented in Table 2.Two to six and nine Cr atoms are mixed into the bcc Fe, respectively, to model Fe-Cr-Co alloys with different Cr contents. For each case with a specified Cr content, two configurations are considered: (i) Cr atoms take a clustered distribution with the nearest neighboring distance; (ii) Cr atoms take a dispersed distribution with remote Cr-Cr distance. The cluster-formation energy of Cr in the Fe-Cr-Co alloy was calculated based on the equation: ΔE = EC-ED, where EC and ED are the total energies of Fe-Cr-Co alloy with clustered and dispersed Cr. Therefore, negative (positive) value of ΔE indicates that the Cr atoms
Fig. 3. Atomic positions of Cr, Co and Fe in (a) Model A; (b) Model B; (c) Model C; (d) Model D. Crx, x = 1–9 denotes Cr atom will replace Fe atom at the numbered sites. Models A and C are used to evaluate the configuration with remote Cr-Cr positions. Models B and D are used to evaluate the first nearest neighbors of Cr-Cr interactions.
40
J. Tian et al. / Scripta Materialia 121 (2016) 37–41
Table 2 The purpose of the models used for the calculation and the detailed chemical compositions in each models. Fe(54-m-n)Co(m)Cr(n), m = 0.3 and n = 2.3,5,9 denote the atom compositions in every system. Crx, x = 1–9 denotes the atom labeled by numbers shown in Fig 3. Purpose
Model Systems
Preference of Cr to be dispersed or clustered with different Cr concentrations.
A, B
Preference of Cr to be dispersed or clustered with Co addition.
C, D
Fe52Cr2(Cr1Cr5); Fe51Cr3(Cr1Cr3Cr5); Fe49Cr5(Cr1–Cr5); Fe45Cr9(Cr1–Cr9); Fe49Co3Cr2(Cr1Cr5); Fe48Co3Cr3(Cr1Cr3Cr5); Fe46Co3Cr5(Cr1–Cr5); Fe42Co3Cr9(Cr1–Cr9);
prefer to be clustered (dispersed). The detail calculation results about the cluster-formation energy in each model are given in Table 3 of the Supplementary material. As can been seen in Table 3 of the Supplementary material, the value of ΔE is positive for the Fe-Cr-Co alloys when the number of Cr atoms is less than or equal to 4, then turns to be negative when the number of Cr atoms increases to 5 and more, implying that the Cr atoms prefer to be dispersed (clustered) at low (high) concentration, which can be well explained by the negative (positive) formation energy of Fe-Cr alloy with low (high) Cr content [35]. In order to identify the specific transition concentration of Cr distribution from being dispersed to being clustered in the Fe-Cr-Co alloys, the cluster-formation energies are plotted as a function of the Cr concentration in Fig. 4. Clearly shown in Fig. 4, the calculated ΔE decreases linearly with increasing the concentration of Cr for both Fe-Cr and FeCr-Co alloys. Moreover, the concentration of Cr at the transition point of ΔE from positive to negative is estimated to be 8.8 at.% for Fe-Cr alloy, which decreases to 7.3 at.% for Fe-Cr-Co alloy. This means that the addition of Co promotes the clustering tendency of Cr atoms and facilitates the formation Cr-rich clusters, lowering the transition concentration from dispersed to clustered Cr in Fe-Cr alloy. This result is well agreed with the miscibility gap of α-phase in Fe-Cr-Co alloy [36], which demonstrated that addition of Co extended the difference in concentration between Fe-rich phase and Cr-rich phase. Magnetic effects have been demonstrated to be the origin of segregation, precipitation and phase separation in Fe-Cr alloys [37–41]. Thus, we have further calculated the magnetic moments of the Fe-Cr-
Fig. 4. The cluster-formation energy (green line) and the difference of averaged magnetic moments of Fe between two Cr distributions (red lines) as a function of Cr concentration with and without Co addition. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Co alloys, in order to explore the stimulating mechanism of Co addition for the formation of Cr-rich clusters observed in the present experiment. The detailed calculation results including the magnetic moments and total energies for the Fe-Cr-Co alloys with different Cr contents are presented in Table 3 of the Supplemenatry material. It can be seen from Table 3 of the Supplementary material that for the central atom of Cr cluster, the value of μ Cr is negative when two or three Cr atoms are mixed into the bcc Fe, then changes to positive when five and nine Cr atoms are involved into the bcc Fe, due to anti-ferromagnetic characteristic of Cr and only really central Cr existed for Cr5 and Cr9 clusters. Which can also explain why μ Cr is very close to zero but remains to be negative (see Table 3 of the Supplementary material for details) for the cases of four and six Cr atoms involved in the Fe-Cr (and Fe-Cr-Co) alloys, due to no real central Cr existed for Cr4 and Cr6 clusters considered in this work. This complex situation of the magnetic moments of Cr in different clusters mainly comes from the magnetic frustrations demonstrated in previous work [35]. Interestingly, it was found that the average magnetic moment of Fe atoms (μ Fe ) in Fe-Cr alloys increased due to Co addition for both dispersed and clustered distributions of Cr atoms, as presented in Table 3 of the Supplementary material. This finding also agrees with the magnetic moment enhancement for Fe atoms of Fe-Co clusters or alloys due to alloying with Co [42,43]. In order to understand the relationship between the magnetic interactions and the stability for different space distribution (clustered or dispersed) of Cr atoms in Fe-Cr and Fe-Cr-Co alloys considered in this work, we have plotted the difference of averaged magnetic moments ( Δμ Fe ) of Fe between two Cr distributions, clustered and dispersed, as a function of Cr concentration in Fig. 4. As we can see in Fig. 4 and Table 3 of the Supplementary material, Δμ Fe increases linearly from − 0.033 (− 0.021) to 0.043 (0.067) μ B with increasing the Cr concentration from 3.7 at.% to 16.7 at.% for Fe-Cr (Fe-Cr-Co) alloys. Easy to understand that, the positive (negative) Δμ Fe means stronger (weaker) ferromagnetic interaction, resulting in higher (lower) stability of corresponding alloys with clustered Cr. Clearly, an increment of Δμ Fe (about 0.02 μ B ) due to Co addition was observed for Fe-Cr-Co alloys compared with Fe-Cr alloys. The transition concentration of Cr from negative to positive Δμ Fe decreases from 9.7 at.% for Fe-Cr to 7.6 at.% for Fe-Cr-Co alloys, qualitatively agreed with the stability results of Fe-Cr-Co alloys. One should note that the increased Δμ Fe (~ 0.02 μ B ) due to Co addition is about 1% of the magnetic moment of each Fe atom (2.20 μ B for ferromagnetic ground state of bulk bcc Fe). Considering that the calculated ferromagnetic interaction energy of pure bcc Fe is 0.47 eV per atom, which is the total energy difference between non-magnetic and ferromagnetic state of bulk bcc Fe and has approximately linear relationship with Δμ Fe according to the classic Heisenberg model of ferromagnetic interactions. Therefore, 1% change of μ Fe should result in an energy change of 4.7 meV for each Fe atom, totally 235 meV for the supercell with about 50 Fe atoms used in our DFT calculation. This estimation of the energy change (235 meV) is quantitatively in the same order of the cluster-formation energy differences between Fe-Co-Cr and Fe-Cr alloys, ranging from 100 to 280 meV for most Cr concentrations as shown in Fig. 4. Thus, it is reasonably to conclude that the increased Fe-Fe ferromagnetic interaction resulted from Co addition increases the stability of Fe-Cr-Co alloys with clustered Cr, thus facilitating the formation of Cr-rich clusters. In summary, spinodal decomposition of Cr atoms at early aging stage was firstly observed in a low-Cr maraging stainless steel. Spinodal decomposition amplitude as a function of aging time was calculated by maximum likelihood method associated with 3DAP results. It can be concluded that Co addition promotes the spinodal decomposition of Cr atoms during the aging process. The functional mechanism of Co addition was also studied by the first-principles calculations. The calculated results indicate that the increased Fe-Fe magnetic interaction resulted from Co addition aggravates the tendency to form Cr-rich clusters in bcc Fe.
J. Tian et al. / Scripta Materialia 121 (2016) 37–41
Acknowledgments This work was supported by a funding from the Natural Science Foundation of China (No. 51201160 and 51472249). We acknowledge Dr. Teng Yang for fruitful discussions on the magnetic interaction. The theoretical calculations were performed on TianHe-1 (A) at National Supercomputer Center in Tianjin. Appendix A. Supplementary data
[15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25]
Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scriptamat.2016.04.033.
[26] [27]
References [1] R.F. Decker, S. Floreen, Maraging Steel: Recent Developments and Applications, in:Maraging Steel – The First 30 Years, The Minerals, Metal & Materials Society, Huntington, 1988. [2] W. Sha, A. Cerezo, G.D.W. Smith, Metall. Mater. Trans. A 24 (1993) 1251. [3] Y. He, K. Yang, W.S. Qu, F.Y. Kong, G.Y. Su, Mater. Lett. 56 (2002) 763. [4] G.R. Speich, D.S. Dabowski, L.F. Porter, Metall. Trans. 4 (1973) 303. [5] A.S. Murthy, J.E. Medvedeva, D. Isheim, S.L. Lekakh, V.L. Richards, D.C. Van Aken, Scr. Mater. 66 (2012) 943. [6] D. Coutsourdais, Mem. Rev. De Metallurgie. 58 (1961) 503. [7] Y.C. Li, W. Yan, J.D. Cotton, G.J. Ryan, Y.F. Shen, W. Wang, Y.Y. Shan, K. Yang, Mater. Des. 82 (2015) 56. [8] M. Schwind, J. Kållqvist, J.O. Nilsson, J. Ågren, H.O. Andrén, Acta Mater. 48 (2000) 2473. [9] M.P. Ryan, D.E. Williams, R.J. Chater, B.M. Hutton, D.S. McPhail, Nature 415 (2002) 770. [10] D.E. Williams, Y.Y. Zhu, J. Electrochem. Soc. 147 (2000) 1763. [11] E.G. Webb, R.C. Alkire, J. Electrochem. Soc. 149 (2002) B272. [12] D. Chandra, L.H. Schwartz, Met. Trans. 2 (1971) 511. [13] S. Jin, S. Mahajan, D. Brasen, Metall. Trans. A. 11A (1980) 69. [14] P. Hedström, Y.F. Hu, J. Zhou, S. Wessman, M. Thuvabder, J. Odqvist, Mater. Sci. Eng. A 574 (2013) 123.
[28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43]
41
K.F. Ha, H.M. Zhang, K.L. Jing, Metall. Trans. A. 20A (1989) 2563. K.H. Park, J.C. Lasalle, L.H. Schwartz, Acta Metall. 34 (1986) 1853. H. Kaneko, M. Homma, K. Nakamura, IEEE Trans. Magn. 13 (1977) 1325. F. Zhu, P. Haasen, R. Wagner, Acta Metall. 34 (1986) 457. J.E. Brown, G.D.W. Smith, Surf. Sci. 246 (1991) 285. M.M. Li, M.K. Miller, W.Y. Chen, J. Nucl. Mater. 462 (2015) 214. V. Kuksenko, C. Pareige, P. Pareige, J. Nucl. Mater. 425 (2012) 125. A. Bhattacharya, E. Meslin, J. Henry, C. Pareige, B. Décamps, C. Genevois, D. Brimbal, A. Barbu, Acta Mater. 78 (2014) 394. V.I. Razumovskiy, A.V. Ruban, P.A. Korzhavyi, Phys. Rev. B 84 (2011) 024106. M.K. Miller, A. Cerezo, M. Hetherington, G. Smith, Atom Probe Field Ion Microscopy, Oxford University Press, Oxford, 1996. B. Gault, D. Haley, F.D. Geuser, M. Moody, E. Marquis, D. Larson, B. Geiser, Ultramicroscopy 111 (2011) 448. T.J. Godfrey, M.G. Hetherington, J.M. Sassen, G.D.W. Smith, J. Phys. 49 (1988) C6–421. D.H. Ping, M. Ohnuma, Y. Hirakawa, Y. Kadoya, K. Hono, Mater. Sci. Eng. A 394 (2005) 285. M.K. Miller, K.F. Russell, Appl. Surf. Sci. 94 (95) (1996) 398. K. Stiller, M. Hattestrand, F. Danoix, Acta Mater. 46 (1998) 6063. G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) RC558. G. Kresse, J. Furthmuller, Phys. Rev. B 54 (1996) 11169. J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865. P.E. Blöchl, Phys. Rev. B 50 (1994) 17953. G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758. T.P.C. Klaver, R. Drautz, M.W. Finnis, Phys. Rev. B 74 (2006) 094435. H. Kaneko, M. Homma, K. Nakamura, IEEE T. Magn. 13 (1977) 1325. P.A. Korzhavyi, A.V. Ruban, J. Odqvist, J.O. Nilsson, B. Johansson, Phys. Rev. B 79 (2009) 054202. A.V. Ruban, P.A. Korzhavyi, B. Johansson, Phys. Rev. B 77 (2008) 094436. O.I. Gorbatov, P.A. Korzhavyi, A.V. Ruban, B. Johansson, Y.N. Gornostyrev, J. Nucl. Mater. 419 (2011) 248. A.V. Ruban, V.I. Razumovskiy, Phys. Rev. B 86 (2008) 174111. D.N. Manh, M.Y. Lavrentiev, S.L. Dudarev, J. Compute. Aid. Des. 14 (2007) 159. G. Mpourmpakis, G.E. Froudakis, A.N. Andriotis, M. Menon, Phys. Rev. B 72 (2005) 104417. A. Diaz-Ortiz, R. Drautz, M. Fahnle, H. Dosch, J.M. Sanchez, Phys. Rev. B 73 (2006) 224208.