Journal of Materials Science & Technology, 2020, 47(0): 152-161 DOI: 10.1016/j.jmst.2020.02.018

Research Article

Retarding the precipitation of η phase in Fe-Ni based alloy through grain boundary engineering

Honglei Hua,b, Mingjiu Zhao,a,*, Lijian Rong,a,*

aKey Laboratory of Nuclear Materials and Safety Assessment, Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110016, China

bSchool of Materials Science and Engineering, University of Science and Technology of China, 96 Jinzhai Road, Hefei 230026, China

Corresponding authors: * E-mail addresses:mjzhao@imr.ac.cn(M. Zhao),ljrong@imr.ac.cn(L. Rong).

Received: 2019-09-20   Accepted: 2019-12-3   Online: 2020-06-15

Abstract

It is important to inhibit the precipitation of η phases in precipitation strengthened Fe-Ni based alloys, as they will deteriorate not only the mechanical property but also the hydrogen resistance. The present investigation shows that grain boundary engineering (GBE) can retard the formation and growth of η phase in J75 alloy. After GBE treatment with 5% cold rolling followed by annealing at 1000 °C for 1 h, the fraction of special boundaries (SBs) increases from 38.4% in conventional alloy to 77.2% and the fraction of special triple junctions increases from 10% to 74%. During 800 °C aging treatment, quite amount of cellular η phases adjacent to random grain boundary (RGB) will be found in conventional alloy, and only a few small η phases have been observed in GBE treatment alloy subjected to the same aging treatment for long time. The reason for GBE in inhibiting precipitation of η phase can be attributed to not only introducing high fraction of SBs but also breaking the connectivity of RGB networks. As nucleation and growth of η phases on SBs are difficult due to their lower Ti concentration and diffusion rate, and the disruption of RGB networks reduces supply of Ti atoms to the η phases significantly, which impedes their growth at RGB.

Keywords: Fe-Ni based alloy ; η phase ; Precipitation behavior ; Random grain boundary connectivity ; Grain boundary engineering ; Boundary diffusion

PDF (6594KB) Metadata Metrics Related articles Export EndNote| Ris| Bibtex  Favorite

Cite this article

Honglei Hu, Mingjiu Zhao, Lijian Rong. Retarding the precipitation of η phase in Fe-Ni based alloy through grain boundary engineering. Journal of Materials Science & Technology[J], 2020, 47(0): 152-161 DOI:10.1016/j.jmst.2020.02.018

1. Introduction

In the last decades, with the improvement of hydrogen economy, the demand of structural materials with high strength and low hydrogen embrittlement sensitivity is incremental. Though, single-phase austenitic alloys have high resistance to hydrogen embrittlement, their yield strengths are typically below 350 MPa [1,2]. Thus, one kind of high-strength Fe-Ni based austenitic alloys (e.g. A286, J75 and so on) with a yield strength over 700 MPa have been successfully developed, which are mainly hardened by fine precipitates of an ordered, coherent L12 phase γʹ [Ni3 (Al, Ti)] formed during aging after solution heat treatment [[3], [4], [5]]. Owing to high yield strength, high ductility and good toughness, they have been wildly used as structural materials in the hydrogen environment [6,7].

However, the γʹ phase would evolve into hexagonal (DO24) η phase (Ni3Ti), especially aging at high temperature for a long time [8,9]. Also, creep or fatigue at high temperature would introduce η phase at grain boundary (GB) and extend to grain interior [10,11]. The formation of grain-boundary η phase has an important influence not only on the mechanical properties but also on hydrogen performance in Fe-Ni based austenitic alloys. On one hand, the formation of the η phases will consume the strengthening precipitate γʹ phases and deteriorate the hardness or strength of alloys [12,13]. On the other hand, the η phase will act as the nucleation site of the crack during deformation and is known to give rise to increase propensity of intergranular fracture [14,15]. In addition, previous investigations [16,17] also demonstrated that hydrogen was easy to accumulate at the incoherent interface between η phase and matrix, and thus resulted in an obvious hydrogen-induced loss of ductility and brittle fracture. Therefore, the η phase is considered as a deleterious and undesirable precipitate in Fe-Ni based alloys.

Many research efforts have been made to eliminate the precipitation of the η phase at GB. For example, precipitation of the η phase may be suppressed by reducing the ratio of Ti / Al [14,18,19]. However, lower ratio of Ti / Al will reduce not only the amount of the γʹ phases [20,21], but also their antiphase boundary energy. As a result, the strengthening effect of the γʹ phases would be weakened [22]. Moreover, our previous studies [17,23] found that addition of boron in the Fe-Ni based alloy could retard the precipitation of η phases at GBs, but increase the tendency to form liquation cracks in heat affected zone during welding because of the segregation of boron to the GB [24,25].

It is known that GB plays an important role in formation and growth of the η phases [14,15,26], as η phases always form at GBs and grow into the grain interior. Furthermore, scholars [27,28] found that η phases were inclined to precipitate at GBs with specific misorientation. However, it is well known that GBs may have different characters, such as low angle or high angle, and their influence on the formation and growth of η phases is not clear. Therefore, it is important to investigate the relationship between η phase precipitation behavior and GB character. Generally, boundary can be categorized according to the coincidence site lattice (CSL) model, whereby the boundary is assigned a ∑ value which is the reciprocal density of geometrically overlapped lattice points of adjoining grains [29,30]. The reported results demonstrated that low ∑ CSL boundary which is referred to as special boundary (SB) [31] does exhibit increased resistance to creep [32], fatigue [33], intergranular corrosion [34] and minor element segregation [35]. It has been found grain boundary engineering (GBE) can be applied to increase the fraction of low ∑ CSL boundaries [36,37]. GBE is the deliberate manipulation of grain boundary structure in order to improve material properties, and widely used in the low stacking-fault energy alloys which twin readily [36]. Typically, GBE treatment is referred to thermomechanical treatment which is composed of cold deformation followed annealing at different temperature [37,38]. Although, many studies have been carried out on the precipitation mechanism of η phase [12], and its relationships with heat treatment [39] and mechanical properties [17,23], there are few investigations which have been conducted on the positive role of GBE to inhibit η phase precipitation.

Thus, our research has two objectives: (1) to investigate the effect of the SBs and connectivity of random grain boundary (RGB) networks on the precipitation behavior of η phase, (2) to understand how GBE inhibits the formation and growth of the η phase.

2. Experimental

The experimental Fe-Ni based J75 alloy was produced by vacuum induction melting technique with the nominal chemical compositions as follows: 30Ni - 15Cr - 1.4Mo - 2.1Ti - 0.3Al - 0.25Si - 0.0015B - Fe bal (wt%). The ingot was homogenized at 1160 °C for 12 h, then rolled into 3 mm-thick plates at 1120 °C. All the plates were subjected to solution treatment at 980 °C for 1.5 h and subsequently water quenching (referred to as AM). Parts of plates were subjected to GBE treatment with 5% cold rolling then annealing at 1000 °C for 1 h followed by water quenching (referred to as GM). Then all the samples aged at 800 °C for 1 h and 12 h, respectively. In order to accelerate the precipitation of η phase and illustrate the role of GBE conveniently, the aging temperature of 800 °C is higher than the peak aging temperature of experimental alloy.

Microstructure was examined by using scanning electron microscope (SEM) with electron backscattered diffraction (EBSD) and transmission electron microscopy (TEM). The samples prior to EBSD examinations were prepared by conventional metallographic technique, followed by electro-polished in 10% perchloric acid to ethanol solution at 15 V for 30 s to remove residual surface stress generated during mechanical polishing. SEM samples were also slightly electrolytic etched in 10% oxalic acid at 0.1A for 4 min. SEM observations and EBSD analysis were all conducted on a ZEISS MERLIN Compact SEM at 20 kV, and Channel 5 software was used to judge the types of boundaries and calculate the proportion of SBs by length. The critical deviation in the SB was categorized according to the Palumbo-Aust criterion of Δθ = 15°∑-5/6 [40]. Thin foils for TEM analysis were prepared using double-jet electrochemical polishing in 10% perchloric acid ethanol solution at -30 °C. The observation was carried on JEM-2100.

Secondary ion mass spectrometry (SIMS) was used to identify the titanium and nickel distribution at GBs. A cuboidal sample with a dimension of 10 × 10 × 3 mm was machined, and mechanically and electrolytically polished to produce a smooth surface. The experiment was performed on a commercial TOF-SIMS 5 spectrometer (Iontof-Münster Germany) with the operation pressure of 10-9 mbar. The sample surface was sputtered by oxygen ion for 1000s to remove the contamination, which could achieve high sensitivity for nickel and titanium. Afterward, the surface analysis was performed on the central region of the sputter surface by analysis bismuth ion source.

3. Results

3.1. EBSD results

The microstructure, boundary characters and RGB networks of AM and GM obtained from EBSD are given in Fig. 1. The different colors in the Inverse Pole Figure (IPF) maps of AM and GM demonstrate different orientations of grains. The crystallographic orientations of grains in both alloys are nearly random (Fig. 1(a) and b). As shown in Fig. 1(c) and (d), SBs and RGBs marked by colored and black lines are both uniformly distributed. After GBE process, the length fraction of SBs increases form 38.4%-77.2%, while that of RGBs decreases from 61.6%-22.8%, correspondingly. The fractions of different SBs for both conditions have been summarized in Table 1. It is easy to find from Table 1, most of the SBs introduced by GBE are twin-related boundaries, and the ∑3 is the dominant one with a fraction of 66.1% in GM while 37.0% in AM. In addition, the fractions of ∑9 and ∑27 are also increased through interaction of boundaries (i.e., ∑3 + ∑3 = ∑9 or ∑3 + ∑9 = ∑27) [36].

Fig. 1.

Fig. 1.   Inverse pole figure (IPF) maps, GB maps and GB networks topology are shown in (a), (c) and (e) for AM, and (b), (d) and (f) for GM. (g) Fractions (in number) of triple junctions to access RGB networks connectivity in AM and GM. In (a), (b), (c) and (d), RGBs are in black while Σ3, Σ9, Σ27 and other Σ ≤ 29 SBs are colored red, green, blue and yellow respectively. In (e) and (f), RGBs and SBs are represented by black and gray lines.


Table 1   Boundary character distribution statistics of AM and GM.

TreatmentLength fraction / %
∑3∑9∑27∑ ≤ 29
AM37.00.60.0438.4
GM66.17.33.677.2

New window| CSV


Fig. 1(e) and (f) show RGBs and SBs on the boundary maps. It can be seen that the RGB networks have been fragmented after GBE treatment. The classification of triple junctions according to the numbers of RGB and SB (labelled as 3R-0S, 2R-1S, 1R-2S, 0R-3S) that they contain provides an indication of the fragmentation of RGB networks connectivity. Generally, a triple junction with more than two SBs is named special triple junction which would fragment the RGB networks effectively [41]. After GBE treatment, the proportion of 0R-3S triple junctions significantly increases from 4% to 62%, and that of the 3R-0S triple junctions decreases from 67% to 10%, just as illustrated in Fig. 1(g).

3.2. Microstructure observation

Fig. 2 is SEM micrographs of AM and GM aged at 800 °C for different time, showing the effect of GBE and aging time on the precipitation behavior of η phase. It can be seen that 1 h aging will introduce a small amount of η phases precipitated at GBs in AM, as indicated by the black arrows and local enlarged picture in Fig. 2(a). While, η phase can not be observed in GM (Fig. 2(b)). As the aging time increased to 12 h, a notable increase in the number and size of the η phases precipitated at GBs is observed in AM, and the η phases with a cellular morphology would grow into the adjacent grains (Fig. 2(c)). Nevertheless, only a small amount of η phases which are much smaller than that in AM can be found occasionally in GM (Fig. 2(d)). To further investigate the microstructure of η phase, TEM examination is carried out. Fig. 3(a) and (c) show lamellar η phases precipitated along GB and growing from GB to interior, separately. Electron diffraction patterns (Fig.3(b) and (d)) have further revealed that η phases aligned along {111} crystallographic plane of γ matrix and the orientation relationship between η phase and γ matrix is as follows: {0 0 1} η // {1 1 1} γ, <2 1 0> η // <1 1 0> γ, agreed with our previous results [23].

Fig. 2.

Fig. 2.   SEM micrographs of samples aged at 800 °C for (a) 1 h (c) 12 h of AM and (b) 1 h (d) 12 h of GM.


Fig. 3.

Fig. 3.   TEM micrographs of the η phase precipitated at GB in the AM aged at 800 °C for 12 h (a) and (c). (b) and (d) Diffraction pattern taken from (a) and (c).


3.3. Precipitation behavior of the η phase at different boundaries

Fig. 4 shows the precipitation behavior of the η phases at RGBs which is examined by SEM and EBSD. The boundary types are shown in the top right corner of the SEM pictures. After aging at 800 °C for 1 h, a small amount of η phases with a parallel rod-like morphology at RGBs are observed in AM, as indicated in Fig. 4(a), which has also been reported by many investigators in Fe-Ni based alloys [17,23,27]. However, η phase is not found in GM whether at RGB or not, as shown in Fig. 4(c). After aging at 800 °C for 12 h, quit amount of big cellular η phases are precipitated at RGBs in AM (Fig. 4(b)), while there are few at RGBs in GM (Fig. 4(d)). It is obvious that although η phase prefer to precipitate at RGB, precipitation of η phase at RGB in GM is more difficult compared with that in AM.

Fig. 4.

Fig. 4.   SEM and EBSD results of the η phases precipitated at RGBs in AM aged at 800 °C for (a) 1 h, (b) 12 h, and in GM aged at 800 °C for (c) 1 h, (d) 12 h. RGBs are in black while Σ3 and other Σ ≤ 29 SBs are colored red and yellow respectively.


In addition, no η phase is observed at ∑9 and ∑27 boundaries after aging at 800 °C for 1 h (Fig. 5(a)). Prolonging aging time to 12 h, quite fewer and smaller η phases can also be found at ∑9 and ∑27 boundaries in GM (Fig. 5(b)). More specifically, twin boundary which is considered as coherent ∑3 has higher resistance to the η phase precipitation, only discrete and tiny η phases can be occasionally found even after aging for 12 h in GM (Fig. 6(b)).

Fig. 5.

Fig. 5.   SEM and EBSD results of the η phases precipitated at different SBs in GM aged at 800 °C for (a) 1 h, (b) 12 h. RGBs are in black while Σ3, Σ9 and Σ27 are colored red, green and blue respectively.


Fig. 6.

Fig. 6.   SEM and EBSD results of the η phase precipitated at coherent ∑3 in GM aged at 800 °C for (a) 1 h, (b) 12 h. Coherent ∑3 is colored red.


To determine the effect of boundary character on η phase precipitation, we analyzed boundary types with η phases. Fig. 7 is the statistics of boundary fractions for different boundaries precipitated with the η phases in GM after aging at 800 °C for 12 h. After aging at 800 °C for 12 h, 80% of boundaries precipitated with η phases are RGBs. So it is clearly that η phases are difficult to form and grow at SBs while they are easy to precipitate at RGBs.

Fig. 7.

Fig. 7.   Statistics of fractions of different boundaries precipitated with η phase in GM after aging at 800 °C for 12 h.


4. Discussion

It is well known that there are two forms of η phases in Fe-Ni based alloys [27,42]: the discrete, parallel, rod-like morphology precipitates mainly at GBs and elongated plates with a Widmanstätten appearance within the grains. The latter morphology is predominant at aging temperature over 820 °C [12]. In this study, we mainly pay attention to the formation and growth process of the η phases at boundaries.

4.1. Characteristic of η phase precipitation

From the TEM results of Fig. 3, we can conclude that η phase precipitates along a {111} crystallographic plane of γ matrix no matter along or tilt at an angle to the boundary. It is noted that in present study the migrating of GB induced by η phases can change the plane index of GB, resulting the precipitating of η phases on new formed GB. As shown in Fig. 8(a), η phase growth will drive RGB migration and change its plane index. As a result, partial RGBs like A-A and B-B will parallel to {111} crystallographic plane, then η phases can precipitate along them. While other cellular η phases adjacent to the RGB would grow into the interior of grain with a tilt angle to the boundary and keep a parallel relationship each other as their interfaces with matrix are the same {111} γ plane. Due to the precipitation, the primary relatively smooth boundary will be replaced by new jagged one. The jagged boundary is concave between two η phases (indicated by arrow in Fig. 8(c)), as the growth of the η phases would consume the surrounding γʹ particles. The dissolution rate of γ′ in the middle of two η phases is lower compared with that in the area close to η phases, which can drag the migrating of GB. Unlike this, η phases precipitated at coherent ∑3 are discrete and would not drive boundary migration (Fig. 6(b)), in order to avoid the formation of new high energy RGB.

Fig. 8.

Fig. 8.   SEM and EBSD results of AM aged at 800 °C for 12 h. (a) SEM image. (b) Phase map consist of Fe and η phase. (c) Inverse pole figure (IPF) maps indicating the austenite around the η phase has the same orientation as the neighboring grain where growth starts. RGB are colored black.


In the meantime, the austenite which is between η phases and swept by the migrating boundary will change its orientation to the same as the grain which the cellular η phase growth starts (Fig. 8(c)). Here, we present detailed explanation as schematically shown in Fig. 9. The η phases grow from original boundary CD into grain A and the boundary will migrate towards grain A in a bowing form. In the same time, the orientation of austenite in region E which is originally a part of grain A will change into the same one as grain B. Thus, η phase will keep {0 0 1} η // {1 1 1} γ, <2 1 0> η // <1 1 0> γ orientation relationship with grain B and original boundary CD would disappear. This is advantage to reduce the formation free energy of η phase.

Fig. 9.

Fig. 9.   Schematic representation of the change in austenite orientation with the formation of η phase.


4.2. Effect of boundary type on the formation of η phase

As shown in Fig. 10(a), coarsening of the γʹ phases near the RGB is more than twice of it in the matrix after aging at 800 °C for 1 h. This is due to that atom diffusion rate along RGB is much higher than volume diffusion in matrix. Besides, the growth of γʹ phase near RGB also follows the Lifshitz - Slyozov - Wagner (LSW) theory based on the Ostwald ripening [[43], [44], [45], [46], [47]]. At the same time, coarsened γʹ phases will gradually align themselves along specific crystallographic planes of γ matrix as shown in the marked area in Fig. 10(a). According to Doherty [48], the driving force for this attraction is the removal of the elastically strained matrix lying between the two precipitates.

Fig. 10.

Fig. 10.   SEM microstructures near different grain boundaries in AM after aging at 800 °C for different time: (a) and (b) 1 h at RGB, (c) 1 h, (d) and (e) 12 h at coherent ∑3.


Prolong the aging time, Kusabiraki [45] pointed that the γʹ phase is in a state of hydrostatic compression. In such a condition, the strain energy might be relieved by the condensation of a layer of vacancies on the close-packed plane of the γʹ phase which would give rise to an extrinsic stacking fault [49]. This would induce the transformation of γʹ phase to η phase [50,51], as shown in Fig. 10(b).

On the contrary, the size of γʹ phase near coherent ∑3 boundary is similar to that in the matrix after aging for 1 h (Fig. 10(c)). Aligned γʹ phases and tiny η phases can be found at coherent ∑3 boundary and also in the interior of grain after aging for 12 h, as the black arrows indicated in Fig. 10(d) and (e). Apparently the coherent ∑3 shows high resistance to η phase precipitation.

It is easy to find η phase is inclined to form at RGB. Thus, there are quit amount of η phases observed in AM after long time aging treatment because of its high fraction of RGBs (Fig. 2(c)), while there are few in GM after subjected to the same aging treatment (Fig. 2(d)), as the fraction of RGBs dropped from 61.6% to 22.8%. It can be expected that SBs are favorable to restrain the formation of the η phases.

4.3. Effect of boundary character on η phase growth

Growth of the η phase is diffusion controlled process, which is mainly determined by the segregation of Ti at GBs and the Ti diffusion along GBs. As shown in Fig. 5, it should be noted that the size of the η phases at RGBs or SBs are different, and this is attributed to following reasons.

4.3.1. Effect of boundary type on titanium segregation

After solution treatment, Ti and Ni atoms prefer to segregate along boundaries, and the tendency of segregation to boundaries can be determined as follow [52]:

Cb≈C0exp(Eb/kT)/(1+C0exp(Eb/kT))

Where Cb is the concentration of segregation atoms at boundary, C0 is the concentration of solute atoms in the matrix, k is Boltzmann’s constant and T is absolute temperature, Eb is the binding energy of segregation element to the boundary which is influenced by boundary energy and segregation element. The higher GB energy will induce higher binding energy [53]. According to the Randle [31], the boundary energy of RGB, coherent ∑3 and incoherent ∑3 is 1.2, 0.01 and 0.1-0.6 J m-2, which would induce maximum Ti enrichment at RGB and minimal at coherent ∑3. By using SIMS, non-uniformly distributed Ni and Ti atoms are found (as shown in Fig. 11). It is reasonably to suppose that Ti inclines to segregate at the RGB but not at coherent ∑3 which is similar to the interior of grain after annealing at 980 °C for 1.5 h (as shown in Fig. 11(b)). This experimental result is consistent with the above theoretical analysis. Therefore, it can be predictable that compared with SB, high inclination of Ti segregation at RGB is beneficial for the growth of the η phase.

Fig. 11.

Fig. 11.   Secondary ion images show the distribution of (a) Ni and (b) Ti ions in alloy after annealing at 980 °C for 1.5 h.


4.3.2. Effect of boundary type on titanium atom diffusion

Besides segregation of Ti atoms at the boundary, their diffusion along boundary is also important to the growth of η phase.

Generally, the GB diffusion coefficient of Ti atom is four to six orders of magnitude larger than volume diffusion coefficient [54]. And, η phase growth will drive RGB migration. Therefore, boundary would provide a fast diffusion channel of Ti atoms to support growth of the η phase. However, boundary diffusion is very sensitive to its structure, and the diffusion velocity at RGB or SB is obviously different. It was demonstrated that the diffusion rate along ∑3 was typically 1-2 orders of magnitude lower than that along conventional RGB [55]. Similar results were also found for other SBs, such as ∑5 [56] and ∑13 [57] and so on [58]. Therefore, compared to SBs, number of Ti atoms can be transported quickly to the η phases along RGBs.

4.3.3. Effect of RGB connectivity on η phase growth

Compared with the fraction of RGBs, the connectivity of RGB networks has more important influence on the growth of η phase. When a η phase formed at RGB, its growth is primarily influenced by RGB connectivity. As shown in Fig. 12(a), almost all the big η phases in AM are found at RGBs which have a good connectivity with other RGBs (Fig. 12(b)). In contrast, only small η phases are precipitated at RGBs which are connected with SBs at two sides in GM, and the advancing growth of the η phases at RGBs is suppressed (Fig. 12(c) and (d)).

Fig. 12.

Fig. 12.   SEM maps of the η phase precipitated at (a) connected RGBs in AM, (c) unconnected RGBs in GM, after aging at 800 °C for 12 h. (b), (d) Grain boundary maps corresponding to (a) and (c), respectively. RGBs are in black while Σ3, Σ9 and Σ27 are colored red, green and blue respectively.


As shown in Fig. 4, the η phases precipitated at RGBs in AM are much larger than that in GM. Because ample Ti atoms are supplied by connective RGBs (Fig. 1(e)). After GBE treatment, few small η phases are found at RGBs after high temperature aging (Fig. 4(d)). The reason is that RGB connectivity is broken dramatically with the reduction of 3R-0S triple junctions from 67% in AM to 10% in GM (Fig. 1(f)and (g)). This microstructure optimization would be beneficial for suppressing the growth of η phase in GM as there are no enough fast diffusion paths for Ti atoms.

Based on the experimental observation and discussion above, it is proposed that the formation and growth of η phases at different boundaries are followed a sequence of events as schematically drawn in Fig. 13. After solution treatment, Ti atoms are prefer to segregate along RGBs rather than SBs both in AM and GM (Fig. 13(a) and (d)). During aging treatment, coarsening of the γʹ phases is easy to occur near parts of the RGBs than SBs, meanwhile the coarsened γʹ phases will gradually align themselves along specific crystallographic plane (Fig. 13(b) and (e), where other uniformly distributed γ′ phases in the interior of grain are omitted). Besides, transformation of γʹ to η will occur at RGB in AM only after short time aging (Fig. 13(b)), which can not be found in GM as indicated in Fig. 13(e). Prolonging the aging time, the further growth of η phases at RGB is affected by the connectivity of RGB networks. As shown in Fig. 13(c), once η phases form at a connected RGB during aging, more Ti atoms would be transported quickly from the adjacent connected RGBs to η nuclei to insure their growth (indicated by the blue arrows). As a result, large cellular η phases would form at RGB. So the high frequencies of 3R-0S and 2R-1S triple junctions in AM are beneficial for the growth of η phases. Instead, RGB is connected with other four SBs in GM, as shown in the Fig. 13(f). When η phases forming at the RGB, only few Ti atoms can be transported to there during aging as the Ti atoms transportation along SB is obviously slow. Therefore, growth of the η phases at RGBs in GM would be restrained. Thus the η phases at RGB in GM are much smaller than that found in AM. After long-term aging treatment, it is also worth noting that tiny and discrete η phases can be found both in the interior of grains or at coherent ∑3, as shown in the Fig. 13(c) and (f) which is induced by the volume diffusion of Ti in matrix.

Fig. 13.

Fig. 13.   Schematic illustration for formation and growth of η phase. Ti concentration at boundary after solution treatment in (a) AM and (d) GM. Formation of η phase at RGB after short time aging in (b) AM and (e) GM. Growth of η phase precipitated at (c) connected RGB in AM, and (f) unconnected RGB in GM after long time aging. RGBs and SBs are colored black and red, respectively.


5. Conclusions

Precipitation behavior of the η phase in the experimental alloy during aging at 800 °C was investigated. According to the observations presented in this study, it is easy to find that the formation and growth of the η phase could be restrained greatly through GBE. Some specific conclusions are summarized as follows.

(1) After GBE treatment, the fraction of SBs in Fe-Ni based alloy can be increased from 38.4%-77.2%, and the ∑3 is dominant one with an increase from 37.0%-66.1%, correspondingly. At the same time, the fraction of special triple junctions increases from 10% to 74% and the connectivity of RGB networks has been disrupted dramatically.

(2) In conventional treatment alloy, quite amount of cellular η phases at RGBs are found during 800 °C aging treatment, however, only a few small η phases have been observed in GBE treatment alloy subjected to the same aging treatment for long time.

(3) The reason for GBE in inhibiting precipitation of the η phase in the alloy can be attributed to not only introducing high fraction of SBs but also breaking the connectivity of RGB networks. Firstly, an increase in fraction of SBs is not in favor of precipitation for the η phases because of the lower concentration and diffusion rate of Ti atoms at SBs which are detrimental to the formation and growth of η phases. Secondly, the disruption of RGB networks reduces supply of Ti atoms to the η phases significantly, which impedes its growth at RGB.

Acknowledgments

This work was supported by the National Natural Science Foundation of China and China Academy of Engineering Physics [No. U1730140] and National Key Research and Development Program of China [Grant No. 2019YFB1505201].

Reference

Y.H. Fan, B. Zhang, J.Q. Wang, E.H. Han, W. Ke, J. Mater. Sci. Technol. 35 (2019) 2213-2219.

URL     [Cited within: 1]

A. Momeni, S.M. Abbasi, J. Mater. Sci. Technol. 27 (2011) 338-343.

URL     [Cited within: 1]

R. Gilles, D. Mukherji, H. Eckerlebe, L. Karge, P. Staron, P. Strunz, Th. Lippmann, J. Alloys. Compd. 612 (2014) 90-97.

URL     [Cited within: 1]

K.Y. Cheng, C.Y. Jo, T. Jin, Z.Q. Hu, J. Alloys. Compd. 509 (2011) 7078-7086.

URL     [Cited within: 1]

L.K. Singhal, J.W. Martin, Acta Metall. 16 (1968) 947-953.

URL     [Cited within: 1]

M.J. Zhao, Z.F. Guo, S.H. Chen, H. Liang, L.J. Rong, J. Mater. Sci. Technol. 30 (2014) 1155-1159.

URL     [Cited within: 1]

A.C. Wang, Y.Y. Li, C.G. Fan, K. Yang, D.F. Li, C.S. Shi, J. Mater. Sci. Technol. 11 (1995) 421-425.

URL     [Cited within: 1]

S. Asgari, Metall. Mater. Trans. A 37 (2006) 2051-2057.

URL     [Cited within: 1]

C.M. Tomasello, F.S. Pettit, N. Birks, J.L. Maloney, J.F. Radavich, Superalloys (1996) 145-151, 1996, Proc. Int. Symp., 8th.

[Cited within: 1]

N.D. Evans, P.J. Maziasz, R.W. Swindeman, G.D. Smith, Scr. Mater. 51 (2004) 503-507.

URL     [Cited within: 1]

B.S. Rho, S.W. Nam, Mater. Sci. Eng., A 291 (2000) 54-59.

URL     [Cited within: 1]

M. Seifollahi, S.H. Razavi, Sh. Kheirandish, S.M. Abbasi, J. Mater. Eng. Perform. 22 (2013) 3063-3069.

URL     [Cited within: 3]

W.J. Wan, G.W. Han, D. Bo, J. Iron Steel Res. Int. 17 (2010) 67-71.

[Cited within: 1]

Y.L. Xu, C.X. Yang, X.S. Xiao, X.L. Cao, G.Q. Jia, Z. Shen, Mater. Chem. Phys. 134 (2012) 706-715.

URL     [Cited within: 3]

Y.L. Xu, C.X. Yang, X.S. Xiao, X.L. Cao, G.Q. Jia, Z. Shen, Mater. Sci. Eng., A 530 (2011) 315-326.

[Cited within: 2]

J.A. Brooks, A.W. Thompson, Metall. Trans. A 24 (1993) 1983-1991.

URL     [Cited within: 1]

Z.F. Guo, H. Liang, M.J. Zhao, L.J. Rong, Mater. Sci. Eng., A 527 (2010) 6620-6625.

[Cited within: 4]

C.S. Wang, T.T. Wang, M.L. Tan, Y.A. Cuo, J.T. Guo, L.Z. Zhou, J. Mater. Sci. Technol. 31 (2015) 135-142.

URL     [Cited within: 1]

T.T. Wang, C.S. Wang, W. Sun, X.Z. Qin, J.T. Guo, L.Z. Zhou, Mater. Des. 62 (2014) 225-232.

URL     [Cited within: 1]

S.H. Chen, M.J. Zhao, L.J. Rong, Corros. Sci. 101 (2015) 75-83.

URL     [Cited within: 1]

S.H. Chen, M.J. Zhao, L.J. Rong, Mater. Sci. Eng., A 561 (2013) 7-12.

[Cited within: 1]

D. Raynor, J.M. Silcock, Methods Cell Sci. 4 (1970) 121-130.

[Cited within: 1]

M.J. Zhao, Z.F. Guo, H. Liang, L.J. Rong, Mater. Sci. Eng., A 527 (2010) 5844-5851.

URL     [Cited within: 4]

M.C. Chaturvedi, W. Chen, A. Saranchuk, N.L. Richards, Superalloys 718 (1997) 743-751, 625, 706 Var. Deriv., Proc. Int. Symp., 3rd.

[Cited within: 1]

W. Chen, M.C. Chaturvedi, N.L. Richards, Metall. Mater. Trans. A 32 (2001) 931-939.

URL     [Cited within: 1]

M. Savoie, C. Esnouf, L. Fournier, D. Delafosse, J. Nucl. Phys. Mater. Sci. Radiat. Appl. 360 (2007) 222-230.

[Cited within: 1]

X.Y. Li, J. Zhang, L.J. Rong, Y.Y. Li, Mater. Sci. Eng., A 488 (2008) 547-553.

[Cited within: 3]

B. Alabbad, S. Tin, Mater. Charact. 151 (2019) 53-63.

URL     [Cited within: 1]

S. Ranganathan, Acta Crystallogr. 21 (1966) 197-199.

DOI      URL     [Cited within: 1]

D.H. Warrington, P. Bufalini, Scr. Metall. Mater. 5 (1971) 771-776.

URL     [Cited within: 1]

V. Randle, P. Davies, B. Hulm, Philos. Mag. 79 (1999) 305-316.

URL     [Cited within: 2]

V. Thaveeprungsriporn, G.S. Was, Metall. Mater. Trans. A 28 (1997) 2101-2112.

URL     [Cited within: 1]

Y. Gao, M. Kumar, R.K. Nalla, R.O. Ritchie, Metall. Mater. Trans. A 36 (2005) 3325-3333.

DOI      URL     [Cited within: 1]

Y.L. Guo, E.H. Han, J.Q. Wang, J. Mater. Sci. Technol. 31 (2015) 403-412.

URL     [Cited within: 1]

M. Tomozawa, Y. Miyahara, K. Kako, Mater. Sci. Eng., A 578 (2013) 167-173.

[Cited within: 1]

V. Randle, Acta Mater. 52 (2004) 4067-4081.

URL     [Cited within: 3]

B.M. Guyot, N.L. Richards, Mater. Sci. Eng., A 395 (2005) 87-97.

URL     [Cited within: 2]

A.J. Schwartz, W.E. King, M. Kumar, Scr. Mater. 54 (2006) 963-968.

URL     [Cited within: 1]

W.J. Wan, G.W. Han, B. Deng, J. Iron Steel Res. Int. 17 (2010) 64-78.

[Cited within: 1]

G. Palumbo, K.T. Aust, Acta metall. mater. 38 (1990) 2343-2352.

URL     [Cited within: 1]

P. Fortier, W.A. Miller, K.T. Aust, Acta Mater. 45 (1997) 3459-3467.

URL     [Cited within: 1]

M.R. Jahangiri, H. Arabi, S.M.A. Boutorabi, Int. J. Miner., Metall. Mater. 20 (2013) 42-48.

URL     [Cited within: 1]

I.M. Lifshitz, V.V. Slyozov, J. Phys. Chem. Solids 19 (1961) 35-50.

URL     [Cited within: 1]

K. Kusabiraki, E. Amada, T. Ooka, ISIJ Int. 36 (1996) 208-214.

URL     [Cited within: 1]

K. Kusabiraki, Y. Takasawa, T. Ooka, ISIJ Int. 35 (1995) 542-547.

URL     [Cited within: 2]

H.B. Motejadded, M. Soltanieh, S. Rastegari, J. Mater. Sci. Technol. 28 (2012) 221-228.

URL     [Cited within: 1]

S. Tang, L.K. Ning, T.Z. Xin, Z. Zheng, J. Mater. Sci. Technol. 32 (2016) 172-176.

URL     [Cited within: 1]

R.D. Doherty, Met. Sci. 16 (1982) 1-14.

[Cited within: 1]

R.G. Baker, D.G. Brandon, J. Nutting, Philos. Mag. Abingdon (Abingdon) 4 (1959) 1339-1345.

URL     [Cited within: 1]

B.H. Kear, A.F. Giamei, J.M. Silcock, R.K. Ham, Scr. Metall. Mater. 2 (1968) 287-293.

URL     [Cited within: 1]

Y.U. Heo, M. Takeguchi, K. Furuya, H.C. Lee, Acta Mater. 57 (2009) 1176-1187.

URL     [Cited within: 1]

D. Mclean, Grain Boundaries in Metals, Clarendon Press, UK, 1957.

[Cited within: 1]

M. Yamaguchi, M. Shiga, H. Kaburaki, Science 307 (2005) 393-397.

DOI      URL     PMID      [Cited within: 1]

The sulfur-induced embrittlement of nickel has long been wrapped in mystery as to why and how sulfur weakens the grain boundaries of nickel and why a critical intergranular sulfur concentration is required. From first-principles calculations, we found that a large grain-boundary expansion is caused by a short-range overlap repulsion among densely segregated and neighboring sulfur atoms. This expansion results in a drastic grain-boundary decohesion that reduces the grain-boundary tensile strength by one order of magnitude. This decohesion may directly cause the embrittlement, because the critical sulfur concentration of this decohesion agrees well with experimental data on the embrittlement.

Y. Mishin, C. Herzig, Mater. Sci. Eng., A 260 (1999) 55-71.

URL     [Cited within: 1]

C. Minkwitz, C. Herzig, E. Rabkin, W. Gust, Acta Mater. 47 (1999) 1231-1239.

URL     [Cited within: 1]

Y. Mishin, Philos. Mag. 72 (1995) 1589-1607.

URL     [Cited within: 1]

Q. Ma, C.L. Liu, J.B. Adams, R.W. Balluffi, Acta Metall. Mater. 41 (1993) 143-151.

URL     [Cited within: 1]

A. Suzuki, Y. Mishin, Interface Sci. 11 (2003) 131-148.

URL     [Cited within: 1]

/