Effects of Zr Addition on Precipitation and Stability of Fe₂Nb Laves and Ni₃Nb Phases in Novel Fe–Cr–Ni–Nb Austenitic Heat-resistant Steels

Abstract

Effects of Zr addition on the microstructure formation of two intermetallic phases of Fe₂Nb Laves (TCP) and Ni₃Nb (GCP) in a carbon-free Fe-20Cr-35Ni-2.5Nb (at.%) novel austenitic heat-resistant steel at elevated temperatures were examined from the viewpoints of thermodynamics and kinetics. From a thermodynamic perspective, the maximum solubility of Zr in the matrix phase is about 0.1 at.% under homogenization treated states at around 1473 K. The dissolved Zr in the matrix eventually reduces the solubility limit of Nb, thereby playing an important role in promoting the formation of grain-boundary Fe₂Nb phase as well as that of thermodynamically stable Ni₃Nb-δ phase within grain interiors after long-term aging at 1073 K. The Zr dissolved in the matrix after the homogenization treatment is found to enrich in δ phase formed during the aging and raises the temperature limit of formation of δ phase by about 50 K. From a kinetic perspective, because of the increase in Nb supersaturation, the dissolved Zr not only enhances the precipitation of Laves phase at grain boundaries but also promotes the homogeneous nucleation of metastable Ni₃Nb-γ″ phase within grain interiors from the beginning of aging and retards the phase transformation from γ″ to δ. These microstructure changes by Zr can be interpreted in terms of the phase equilibria and relative phase stability among different phases. The present results reveal that even a small amount of Zr in solution, not segregation to grain boundaries, has tremendous effects on the precipitation behavior and morphology of intermetallic phases.

1. Introduction

To improve the energy efficiency of a coal-fired power plant and reduce CO₂ emission it is necessary to raise the steam temperature in the boilers.^1,2,3,4) In an advanced ultra-supercritical (A-USC) thermal power plant for the next generation operated above 973 K,⁵⁾ a 10⁵ h creep-rupture strength of 100 MPa or higher^3,4) is required for the boiler tubes/pipes and turbine rotor components. Many Ni-based superalloys strengthened by Ni₃Al-γ′ phase^3,4,6,7,8) have been developed as candidate for the A-USC power plant components. On the other hand, the steels strengthened by transition metal carbides do not fulfill the requirement.^3,4,9,10) However, Takeyama et al. have recently proposed a new concept to develop novel carbon-free austenitic heat-resistant steels strengthened by two intermetallic phases of topologically close-packed (TCP) Fe₂Nb Laves and geometrically close-packed (GCP) Ni₃Nb,^{11,12,13,14,15,16,17,18,19,20,21,22,23)} based on thermodynamics and precipitation kinetics aspects. They found that the allowable temperature of 10⁵ h creep-rupture strength under 100 MPa in the novel steels can exceed 973 K and the temperature is almost comparable to that of a candidate nickel-base alloy Nimonic 263.^24,25) The superior creep-rupture strength of the novel steels is mainly due to the precipitation of TCP Fe₂Nb Laves phase at grain boundaries (GBs) named as the “Grain-boundary Precipitation Strengthening (GBPS)” mechanism,^{11,12,15,16,17,18,19,20,21,22,23,26)} together with the precipitation of metastable Ni₃Nb-γ″ and subsequent stable Ni₃Nb-δ phase within grain interiors.¹⁸⁾

The addition of Zr is reported to enhance the creep-rupture strength of these novel steels.¹³⁾ In general, Zr is widely accepted as a grain-boundary strengthener in heat resistant alloys^27,28,29,30) due to segregation to grain boundaries because of its large atomic size with respect to the host atom.³¹⁾ However, in our previous study, the small amount of Zr additions to one of the novel steels based on Fe-20Cr-30Ni-2Nb (at.%) was found to enhance the grain-boundary precipitation of the Fe₂Nb Laves phase.¹⁸⁾ Thus, the strengthening of the steel by the Zr addition could be associated with the increase in the precipitation of GB Laves phase based on the GBPS mechanism. In addition, we also revealed that Zr was enriched in the metastable Ni₃Nb-γ″ phase precipitated within the grain interiors and retarded the transformation to the stable Ni₃Nb-δ phase.¹⁸⁾ These results suggest that the solid solution of Zr atoms in the matrix somehow affects the precipitation behavior of the intermetallic phases, although the grain-boundary segregation of Zr cannot be ruled out.

In this study, thus, several carbon-free novel austenitic steels containing Zr up to 0.5 at.% were prepared, and the effects of Zr addition on the precipitation and stability of TCP Fe₂Nb Laves and GCP Ni₃Nb phases were examined from thermodynamics and kinetics viewpoints, for the sake of further development of the novel steels.

2. Experimental Procedure

Three steels with a base composition of Fe-20Cr-35Ni-2.5Nb (all compositions are at.% unless otherwise stated) containing a different amount of Zr up to 0.5 at.% were used in this study. The nominal compositions and the heat treatments employed to these steels are summarized in Table 1. These steels are hereafter referred to as base (no Zr addition) and Zr content. The 30 g button ingots of each composition were prepared by arc melting under argon atmosphere using the metals of 99.9% purity. During melting process each ingot was turned over several times to eliminate un-melted metals and to homogenize the mixture. The weight loss of each ingot after the melting is less than 0.1 g. The ingots were first homogenized to eliminate the microstructure inhomogeneity occurring during solidification and then cold-rolled to about 30% in height reduction. The cold-rolled ingots were then homogenized again to adjust the grain size to about 150 to 200 μm, followed by water quench. Subsequently, the homogenized specimens were aged at given temperatures up to 3600 h.

Table 1. Nominal compositions and heat-treatment conditions of the steels studied.

Steel	Nominal composition/at.%					Homogenization treatment	Aging conditions
	Cr	Ni	Nb	Zr	Fe		973 K–1173 K 12 s–3600 h
Base	20	35	2.5	–	bal.	1523 K/1 h
0.1Zr	20	35	2.5	0.1	bal.	1463 K/2 h
0.3Zr	20	35	2.5	0.3	bal.	1443 K/4 h
0.5Zr	20	35	2.5	0.5	bal.

* Aging for up to 15 minutes was done by salt bath.

For microstructure observations, the heat-treated specimens were mechanically polished, followed by electro-polishing in a solution of phosphoric acid containing supersaturated chromic anhydride. The microstructures were examined using a field emission scanning electron microscope (FE-SEM) with an acceleration voltage of 10 to 15 kV with a working distance of 10 mm. The compositions of the phases present in the heat-treated specimens were analyzed using a field-emission electron probe microanalyzer (FE-EPMA) equipped with wave-length dispersive spectrometer (WDS) under 15 kV and 20 nA. The compositions were determined by point analysis at 10 different positions for each phase and the areas analyzed were larger than 1 μm to eliminate the broadening effect. The data with the total mass of 100 ± 2 wt.% were averaged with an accuracy of ± 0.01 wt.%. The elemental mappings were also performed for thin or small phases to qualitatively analyze the compositions with a step size of 0.06 μm and dwell time of 10 ms.

The grain-boundary area fraction of precipitates (ρ), which is defined as the grain-boundary area decorated with precipitates with respect to the total grain boundary area, was quantified using the following Eq. (1):   

$$\rho \text{(\%)}=\frac{\sum l_i}{L}\times \text{100}$$(1)

where the numerator is the sum of the measured grain-boundary lengths l_i covered by the number of precipitates (i) and L is the total length of analyzed grain boundaries (about 1 mm), using an appropriate magnification of several backscattered electron images (BEIs). The hardness of grain interiors was measured by Micro-Vickers hardness machine with a load of 9.8 N and time for 10 s. The indentation size was smaller than the grain size, indicating that the values are the grain-interior hardness values.

3. Results and Discussion

3.1. Effect of Zr on the Long-term Aging Microstructures

Figure 1 shows the secondary electron images (SEIs) for the change in microstructures with the increase in Zr content of the steels aged at 1073 K for the longest aging time of 3600 h, closest to the equilibrium state. In the base steel (Fig. 1(a)), as previously reported,²⁶⁾ the grain boundaries are decorated mostly by the spheroidal Fe₂Nb (C14) Laves phase, together with a small amount of Ni₃Nb-δ (D0_a) phase, and the area fraction ρ is about 60%. Within the grain interiors, two kinds of precipitates with a short ellipsoidal shape and a long needle-like shape are uniformly distributed, and they are respectively identified as the Laves and δ phases. In the Zr added steels (Figs. 1(b) to 1(d)), the grain boundaries are obviously covered with globular shaped precipitates with a size of 1 to 2 μm, and the ρ values become higher than that of the base steel, regardless of the Zr content. However, some large particles with about 5 μm in size are often observed at the grain boundaries with increasing Zr content. Within grain interiors, the addition of 0.1 at.% Zr increases the precipitation density of the thicker and elongated precipitates forming the Widmanstatten morphology (Fig. 1(b)). The precipitation morphologies in the Zr added steels are almost the same, regardless of the Zr content. The measured ρ and H_v values are plotted as a function of the Zr content, as shown in Fig. 2. The ρ and H_v values increase from 60% to 74% and from 2.0 GPa to 2.3 GPa, respectively, by adding 0.1% Zr, and the both values remain almost unchanged with a further increase in Zr content.

Fig. 1.

Secondary electron images (SEIs) showing microstructures of (a) base, (b) 0.1Zr, (c) 0.3Zr, and (d) 0.5Zr aged at 1073 K for 3600 h.

Fig. 2.

Change in the area fraction covered by the Fe₂Nb ρ and the hardness H_v within grain interiors with Zr content in the steels aged at 1073 K for 3600 h.

Figure 3 shows the elemental mapping corresponding to the microstructure of the 0.1Zr shown in Fig. 1(b). At grain boundaries, most of the precipitates are depleted in Fe, Cr and Ni, and enriched in Nb with respect to the γ (fcc) matrix phase, indicating that they are the Laves phase, based on our phase diagram study.¹⁴⁾ However, the precipitate formed at the triple junction is enriched in Ni and Nb and further depleted in Fe and Cr, indicating the Ni₃Nb-δ phase. It is important to note that Zr is enriched in the region where the δ phase is formed. Within grain interiors, as indicated by arrows the needle-shaped precipitates having a straight habit plane with three directions, forming the Widmanstatten morphology in each grain, are identified as the δ phase where Ni and Nb are enriched and Fe and Cr are depleted. Note that Zr is enriched in the δ phase, in good agreement with the previous report.¹⁸⁾ The elongated thicker precipitates formed on and along the δ phase are identified as the Laves phase where Fe, Cr and Ni are depleted but Nb is enriched. There is no indication of Zr enrichment in the Laves phase.

Fig. 3.

Elemental mappings of the 0.1Zr aged at 1073 K for 3600 h.

From a thermodynamic viewpoint, the present results clearly demonstrate that the addition of only 0.1 at.% Zr to the base steel is effective in enhancing the formation of Fe₂Nb Laves phase at the grain boundaries and Ni₃Nb-δ phase within grain interiors at 1073 K. The Zr addition seems to enhance the formation of grain-boundary (GB) δ phase as well, as shown in Fig. 3. However, we confirmed that the amount of GB δ phase is still small, so that the increase in the ρ shown in Fig. 2 is mainly due to the formation of the Laves phase. In addition, an important finding is that Zr is prone to enrich in the Ni₃Nb-δ phase.

In order to understand the reason why such a small amount of Zr is effective in increasing the formation of Fe₂Nb phase at the grain boundaries as well as Ni₃Nb-δ phase within the grain interiors, the compositions of γ matrix phase in the specimens after the homogenization treatment (HT) and those in the same specimens aged at 1073 K shown in Fig. 1, together with some precipitated phases, were analyzed, and the results are summarized in Table 2. From these datasets, the concentrations of Nb and Zr dissolved in the matrix are plotted as a function of the Zr content in the steels in Fig. 4. In the HT state, the solubility of Nb in the matrix phase is nearly the same level of 2.3 at.% to 2.4 at.%, regardless of the Zr contents in the steels studied. At 1073 K, however, the solubility of Nb in the matrix phase in the base steel is 0.8 at.%, and it decreases down to about 0.6 at.%, regardless of the Zr content (Fig. 4(a)). Thus, the Nb supersaturation, ΔX_Nb, at 1073 K is significantly increased by the Zr additions. The solubility of Zr in the matrix phase in the HT state is about a level of 0.07 at.% to 0.08 at.%, regardless of the Zr content in the steels studied, and it is reduced down to about 0.02 at.% at 1073 K (Fig. 4(b)). Note that the large round-shaped particles observed in both 0.3Zr and 0.5Zr steels (Figs. 1(c), 1(d)) are found to contain about 13 at.% Zr and 8 at.% Nb (Table 2), indicating that more than 0.1 at.% addition of Zr to the base steel induces the formation of a highly Zr enriched phase, presumably Ni₅Zr (C15_b) phase, as a primary solidification phase. Thus, the solubility limit of Zr in the matrix at the HT temperature regions is about 0.08 at.%, and it is about 0.02 at.% to 0.03 at.% at 1073 K.

Table 2. The analyzed compositions of the phases present in steels aged at 1073 K for 3600 h.

Steel	Heat treatment	Phase present	Compositions/at.%
			Cr	Ni	Nb	Zr	Fe
Base	HT	γ	20.3	34.5	2.30	–	bal.
	1073 K/3600 h	γ	21.2	34.0	0.80	–	bal.
		ε	13.5	24.5	25.7	–	bal.
		δ	3.40	67.1	21.0	–	bal.
0.1Zr	HT	γ	21.2	34.5	2.42	0.07	bal.
	1073 K/3600 h	γ	22.2	33.7	0.63	0.02	bal.
		ε	14.2	24.1	25.6	0.01	bal.
		δ	*	*	*	*	*
0.3Zr	HT	γ	21.4	34.1	2.38	0.08	bal.
	1073 K/3600 h	γ	22.4	33.5	0.63	0.02	bal.
		ε	*	*	*	*	*
		δ	*	*	*	*	*
		Primary	2.97	64.5	8.21	12.5	bal.
0.5Zr	HT	γ	21.7	33.8	2.26	0.08	bal.
	1073 K/3600 h	γ	22.6	33.4	0.61	0.03	bal.
		ε	14.0	24.7	25.0	0.05	bal.
		δ	*	*	*	*	*
		Primary	2.76	65.2	7.77	13.0	bal.

*  The size of the precipitates is too small to analyze.

Fig. 4.

Solubility of (a) Nb and (b) Zr dissolved in the γ matrix at the homogenization treated state (HT) and at 1073 K as a function of the Zr content in the steels.

In general, Zr is believed to segregate to grain boundaries,²⁹⁾ because of the larger atomic radius (1.602 Å), 15% larger than those of the major matrix constituent elements of Fe (1.274 Å), Ni (1.246 Å) and Cr (1.282 Å) atoms under the CN (coordination number) of 12.^31,32) Therefore, a small amount of Zr of about 0.1 wt.% is often added to heat resistant materials as a grain boundary strengthener.²⁹⁾ However, the present result clearly demonstrates that most of the Zr atoms are existing in solution in the γ matrix, rather than segregating to the grain boundaries. Thus, it is evident that the increase in the ΔX_Nb by the Zr addition gives rise to the increase in the ρ. The increase in the ΔX_Nb also plays an essential role in increasing the precipitation density of Ni₃Nb-δ phase within the grain interiors since Nb is a main constituent element of the δ phase. However, the Zr supersaturation, ΔX_Zr, also contributes in enhancing the formation of the δ phase at 1073 K, as shown in Fig. 4(b), since Zr is enriched in the δ phase.

The decrease in the solubility of Nb in the γ matrix phase is related to the Zr in solution. We tentatively calculated the lattice parameters a of the matrix phase in the base and 0.1Zr steels using the analyzed compositions of the elements in Table 2, based on the hard sphere model ($\sqrt{2}a=4\overline{r}$) together with Vegard’s law, where $\overline{r}$ is an average atomic size of the matrix phase with fcc structure using the atomic radii of the constituent elements of Fe, Ni, Cr, Nb (1.468 Å) and Zr with CN12 as mentioned before,³²⁾ given by the following equation:   

$$\overline{r}={\displaystyle \sum }_i^n{x}_i{r}_i$$

where x_i is the mole fraction of i atom, and r_i is the atomic radii of i atom. The calculated lattice parameters of the steels with and without Zr are exactly the same value of 3.586 Å, indicating that the lattice expansion of the matrix phase by solution of 0.8 at.% Nb in the base steel corresponds to that by the solution of 0.63 at.% Nb and 0.02 at.% Zr together in the 0.1Zr steel. The calculated result clearly suggests a certain threshold value for the lattice expansion limit in solution by the solute elements in the matrix. Furthermore, the higher the solubility of Zr, the lower the solubility of Nb, which is responsible for the increase in ΔX_Nb, and thereby enhancing the formation of Fe₂Nb and Ni₃Nb-δ phases at 1073 K.

The increase in ΔX_Nb by the Zr addition would raise the temperature limit of the thermodynamically stable δ phase. Figure 5 shows an elemental mapping of the microstructures of the base and 0.1Zr steels aged at 1123 K for 1460 h. This aging time is far beyond the aging time of 3600 h at 1073 K, according to Larson-Millar conversion. In both steels, the grain-boundary precipitates are the Laves phase where Ni is depleted and Nb is enriched, and the ρ is higher in the Zr added steel. Within the grain interiors, in the base steels all of the precipitates with an elongated shape of 2 μm in length is the Laves phase where Ni is depleted and Nb is enriched (Fig. 5(a)). In Zr added steel, most of the precipitates with the same shape are also the Laves phase, and the precipitation density becomes higher than that in the base steel. Besides the laves phase, some needle-like precipitates indicated by yellow arrows are identified as Ni₃Nb-δ where Ni, Nb and Zr enriched (Fig. 5(b)). The result of the increase in upper temperature limit of the formation of thermodynamically stable Ni₃Nb-δ phase by Zr addition indicates that Zr addition stabilized the δ phase. It is also noteworthy that long needle-like precipitates pointed by red arrows in Zr added steels (Fig. 5(b)) are the Laves phase, not the δ phase, and the precipitation morphology of the Laves phase by Zr additions obviously changes to Widmanstatten-type at 1123 K.

Fig. 5.

Elemental mappings of (a) base and (b) 0.1Zr aged at 1123 K for 1460 h.

3.2. Effect of Zr Addition on the Microstructure Formation

In the previous section, regardless of the Zr content of the steels, the 0.1 at.% Zr dissolved in the matrix plays an important role in enhancing the formation of Fe₂Nb at the grain boundaries and that of Ni₃Nb-δ phase within the grain interiors, from the thermodynamics viewpoint. This section focuses on the 0.1Zr steel and how the Zr in solution affects the precipitation behavior of both intermetallic phases, from the kinetics viewpoint.

Figure 6 shows a typical example of the change in the precipitation morphology with Zr addition at different aging time at 1073 K in the base and 0.1Zr steels. In the base steel after 1 h aging, the elongated Fe₂Nb Laves phase with an average length of 1 μm is non-uniformly precipitated at the grain boundaries, and no precipitates are observed within the grain interiors (Fig. 6(a)). After 24 h aging, the GB Laves phase becomes larger and the ρ is still less than 50%. Within the grain interiors, three different shapes of inhomogeneously distributed precipitates are observed; very fine dotted particles (< 500 nm) identified as metastable Ni₃Nb-γ″ (D0₂₂) phase,²⁶⁾ needle-shaped precipitates (1 to 2 μm) identified as stable Ni₃Nb-δ (D0_a) phase transformed from the γ″ phase, and ellipsoidal particles (1 μm) identified as Fe₂Nb (Fig. 6(b)). In contrast, in the 0.1Zr steel, the finer GB Laves phase particles are homogeneously distributed, and the ρ apparently increases even after 1 h. Within the grain interiors, the very fine γ″ precipitates are already formed and homogeneously distributed even after 1 h aging (Fig. 6(c)). After further aging of 24 h, the uniform distribution of the GB Laves phase particles are kept and the ρ is much higher than that in the base steel. Within the grain interiors, the γ″ phase grows to some extent and the transformed product of needle-shaped δ phase is developed. In addition, thick elongated precipitates identified as Fe₂Nb are also formed (Fig. 6(d)). Then, the quantitative analyses of the ρ and H_v in the base and 0.1Zr steels aged for various aging times were preformed, and the values are plotted as a function of aging time in Fig. 7. In both steels, the ρ values progressively increase with increasing the aging time and reach the maximum values shown in Fig. 2 after 1200 h aging. The value of ρ in the Zr added steel becomes higher by approximately 14% from the beginning of the precipitation to the entire aging to 3600 h (Fig. 7(a)). In the case of H_v within the grain interiors (Fig. 7(b)), both steels show nearly the same values of 1.5 GPa in the HT state. In the base steel, the H_v starts increasing after aging for a few hours and reaches a peak of 2.2 GPa at about 100 h, followed by a gradual decrease to 2.0 GPa to 3600 h. In the 0.1Zr steel, the H_v starts increasing after aging for a few minutes by an order of magnitude faster than that in the base steel and rapidly increases to a peak of 2.8 GPa at around 10 h, followed by a slight decrease to 2.3 GPa with aging to 3600 hs. It is worth mentioning that the H_v value doubles after the few hours of aging by adding only 0.1 at.% Zr to the base steel.

Fig. 6.

SEIs showing the difference in microstructure of (a, b) base and (c, d) 0.1Zr aged for 1 h (a, c) and 24 h (b, d) at 1073 K.

Fig. 7.

Change in (a) ρ and (b) H_v within the grain interiors with aging time in the base and Zr added steels at 1073 K.

The same microstructure analyses were done at 1123 K, and the change in ρ and H_v with aging time is shown in Fig. 8. Both steels show similar trends of ρ with aging time, although the values of the Zr added steel become higher by approximately 10% than that in the base steel during the entire aging. The ρ values after aging for 1200 h at 1123 K is higher than those at 1073 K for 3600 h (Fig. 8(a)). The H_v values in the both steels also show the similar trend (Fig. 8(b)), although the increased value of the peak hardness by the Zr addition becomes smaller than that obtained at 1073 K (Fig. 7(b)). The corresponding microstructures before and at the peak hardness for both steels are shown in Fig. 9. Before the peak of 1 h aging (Figs. 9(a), 9(c)), the base steel shows only the GB Laves phase sparsely precipitated and almost no precipitates within the grain interiors, whereas the Zr added steel shows the GB Laves phase covered most of the grain boundaries and very fine γ″ phase particles precipitated within the grain interiors either homogeneously or with aligned morphology depending on the grains. At the peak of 24 h aging (Figs. 9(b), 9(d)), although the morphology of the GB Laves phase is nearly the same, the precipitation morphology within the grain interiors is distinctly different between the two steels. The base steel shows a sparse distribution of ellipsoidal particles identified as the Laves phase, whereas the Zr added steel exhibits the Widmanstatten morphology consisting of thick and elongated precipitates with 5 μm in length. The elongated precipitates were identified as the Laves phase where Fe is enriched, although some amount of thin needle-shaped δ phase with about 1 μm in length exists. Note that the precipitation density of the Laves phase in the Zr added steel is higher than that in the base steel, and this is responsible for the increase in the H_v values by the Zr addition (Fig. 8(b)).

Fig. 8.

Change in (a) ρ and (b) H_v within grain interiors with aging time in the base and Zr added steels at 1123 K.

Fig. 9.

Microstructure change with aging time at 1123 K for (a, c) 1 h and (b, d) 24 h, nearly the peak hardness in Fig. 8(b), of the base (a, b) and the 0.1Zr (c, d).

Based on these results, effects of Zr addition on the precipitation kinetics of TCP Laves phase and metastable and stable GCP Ni₃Nb phases are depicted as time-temperature precipitation (TTP) diagram together with the TTP in the base steel^26,33) in Fig. 10, where the ε_GB and ε_GI are the Laves phase precipitated at the grain boundaries and within the grain interiors, respectively. There are several points to be noted on Zr effects on the precipitation kinetics. Firstly, the Zr addition accelerates the formation of GB Laves phase (P_s (ε_GB)) by nearly an order of magnitude. Secondly, the Zr addition raises the nose temperature of the C curve of the formation of γ″ phase within the grain interiors (P_s (γ″)) by about 50 K and also accelerates the formation of γ″ phase by an order of magnitude. Thirdly, the Zr addition also raises the nose temperatures of the C curves of the start (T_S) and finish (T_f) of the γ″ to δ phase transformation by about 50 K and enhances the phase transformation kinetics above the nose temperature but slows it down below the temperature. Lastly, the Zr addition retards the formation of the Laves phase within the grain interiors (P_s (ε_GI)), because of the consumption of the ΔX_Nb by the formation of γ″ and δ phases.

Fig. 10.

Effect of the 0.1 at.% Zr addition on the precipitation kinetics, based on the previously reported time-temperature-precipitation (TTP) diagram of the base.^26,33)

3.3. Zr Effect on the Formation of TCP and GCP Phases

In section 3.1, the addition of 0.1 at.% Zr to the base steel of Fe-20Cr-35Ni-2.5Nb revealed to bring on the following three distinct changes from the thermodynamic viewpoint: decreasing the Nb solubility in the γ matrix (Fig. 4), increasing the ρ of the GB Fe₂Nb Laves phase (Fig. 2) and stabilizing the Ni₃Nb-δ phase by nearly 50 K up to about 1073 K (Fig. 5). These changes can be interpreted by the relative phase stability change among γ, Fe₂Nb-ε (TCP) and Ni₃Nb-δ (GCP) phases.¹⁴⁾ Figure 11 shows a schematic illustration of the Gibbs energy/composition diagram of the three phases at a given temperature. In the case of the steel without Zr, the terminal compositions of the three phases in equilibrium each other, so called “the three-phase tie triangle” can be obtained from the common tangent plane among the free energy curvilinear surfaces of the three phases, depicted as a solid line at the bottom of Fig. 11. Since Zr is identified to enrich in the δ phase, the free energy of the δ phase should be decreased. The curvature of the free energy curvilinear surface of the γ phase is larger than those of the intermetallic phases, so that the decrease in the energy of the δ phase makes the contact point (solid circle) of γ phase on the common tangent plane further away toward Ni-(Fe+Cr) line, thereby changing the three-phase tie triangle as shown by the dotted line in Fig. 11, resulting in the decrease in solubility of Nb in the matrix phase. This is responsible for the increase in the ρ of GB Laves phase. The change in the three-phase tie triangle also makes it possible to change the phase equilibria from the γ+ε two-phase region to the γ+ε+δ three-phase region in a certain alloy composition range as shown in the solid red circle.

Fig. 11.

Schematic illustration showing the change in the phase equilibria among γ, Fe₂Nb-ε (TCP) and Ni₃Nb-δ (GCP) phases by the addition of Zr, based on a thermodynamic viewpoint. The solid line is for the base steel system and the broken line for the Zr added system. (Online version in color.)

In section 3.2, from the kinetic viewpoint, the addition of 0.1 at.% Zr to the base steel was shown to enhance the formation of metastable Ni₃Nb-γ″ (GCP) phase within the grain interiors in short-term aging (Figs. 6, 7, 8, 9), prior to the formation of stable Ni₃Nb-δ phase after long-term aging. Formation of this metastable γ″ phase is commonly observed in Ni-base alloys containing Nb, such as IN718, although the mechanism of formation has not been uncovered yet. The enhancement of the formation of γ″ phase in this study is associated with the decrease in Nb solubility in the matrix, that is, the increase in Nb supersaturation, ΔX_Nb. Figure 12 shows a schematic illustration showing the γ+ε+δ three-phase tie triangles of the steels without and with Zr at a given temperature, depicted based on the thermodynamic consideration of Zr effects shown in Fig. 11, where the terminal composition of the metastable γ″ phase is considered the same as that of the δ phase. This assumption is reasonable since the transformation (γ″→δ) is the second order transformation and the Zr is already enriched in the γ″ phase.¹⁸⁾ Now let us consider the decomposition process of the supersaturated γ phase (γ^ss) at the point O during aging at the given temperature, where the point O in the three-phase regions is a fixed alloy composition in both steels after the homogenization treatment. In both steels, the Laves phase forms first mainly at the grain boundaries, followed by the γ″ formation within the grain interiors (Fig. 10). Suppose that the ΔX_Nb is maintained in the regions away from the grain boundaries, the formation of metastable γ″ phase would occur along the following decomposition process:   

$${\gamma }^{\text{ss}}\to {\gamma }^{\text{neq}1}+\gamma ″,$$

where the γ^neq1 is a non-equilibrium γ phase with a composition at the point of e and f which is the intersection of the line ① and the extrapolation line of the γ/γ+δ phase boundary of the base steel and that of the steel with Zr, respectively, as shown in Fig. 12. This decomposition is highly possible because the D0₂₂ structure of the γ″ phase is one of the derivatives of fcc structure with similar lattice parameters,³⁴⁾ reducing the thermal barriers energy for the nucleation, which makes fine coherent homogeneous nucleation possible, as shown in Figs. 6(c) and 9(c). The driving force for the nucleation of γ″ phase is associated with the Nb supersaturation ($\mathrm{\Delta }{X}_{\text{Nb}}^{\text{m}}$), that is, the difference in Nb concentration between the points O (γ^ss) and e (γ^neq) for the base steel and that between the points O and f for the Zr added steel. As shown in Fig, 12, the $\mathrm{\Delta }{X}_{\text{Nb}}^{\text{m}}$ in the steel with Zr is higher than that in the base steel, resulting in the formation of a higher amount of γ″ phase, and this is in good agreement with the present results.

Fig. 12.

Schematic illustration showing the decomposition processes of the supersaturated γ phase for the formation of Fe₂Nb Laves (TCP) and Ni₃Nb (GCP) phases, based on thermodynamic and kinetic viewpoints. (Online version in color.)

As aging proceeds, the following two reactions take place:   

$$\gamma ″\to \delta$$

  

$${\gamma }^{\text{neq}1}\to \epsilon +{\gamma }^{\text{neq}2}$$

The former is the phase transformation process to the stable Ni₃Nb-δ phase. On the other hand, the latter is the further decomposition process of γ^neq1 to form the Laves (ε) phase, where the γ^neq2 is another non-equilibrium γ phase with a composition at the point h and j which is the intersection of the γ/γ+ε phase boundary and the tie-line ② for the base steel and the tie-line ③ for the steels with Zr, respectively, as shown in Fig. 12. This decomposition process results in the formation of the Laves phase within the grain interiors, and the volume fraction of the Laves phase is obviously higher in the steel with Zr than that in the base steel, which is in good agreement with the present experimental results (Figs. 5 and 9).

It is interesting to discuss the change in the precipitation morphology of the Fe₂Nb Laves phase within the grain interiors from a random-distribution type consisting of an ellipsoidal shape to the Widmanstatten-type consisting of a needle-like shape (Figs. 5 and 9(b), 9(d)). This is associated with the transformation process of GCP and TCP phases as mentioned above. Figure 13 schematically illustrates how the precipitation morphology of the Fe₂Nb Laves phase becomes Widmanstatten-type in terms of the precipitation sequence of the precipitates within a grain interior with aging, particularly the case where the <111>_γ orientation of the grain is normal to the observation surface (paper plane). In the beginning of aging, metastable γ″ phase homogeneously precipitates (Fig. 13(a)), and because of the difference in the precipitation habit plane against γ matrix phase between the γ″ phase ((001)_γ″//{001}_γ, <010]_γ″// <100>_γ) and δ phase ((010)_δ //{111}_γ, [100]_δ // <011>_γ), the γ″ phase transforms to the δ phase with aligning along the four {111} planes of the γ phase, as shown in Figs. 6(c) and 9(c). Once the transformation is completed, the precipitation morphology of the δ phase becomes the Widmanstatten-type (Figs. 13(a), 13(b)). This is because the δ phase precipitates with a platelet shape having the above crystallographic orientation relationship, so that in most cases the δ phase is visible along three different line directions intersecting the four {111}_γ planes on the observation surface for each grain, except the grain with <100>_γ orientation perpendicular to the observation surface where only two orthogonal intersection line directions exist. Since the δ phase can dissolve less Fe and Cr atoms (Table 2), the formation of δ phase results in the enrichment of these atoms locally around the δ phase, enhancing the nucleation of Laves phase at the δ/γ interfaces (Fig. 13(b)). In case that the precipitation density of the γ″ phase is small, the length of the δ phase is limited so that the further transformation from δ to ε results in a random distribution morphology of the Laves phase. On the other hand, if the precipitation density is high like the case in the present steel with Zr, the transformed morphology of the δ phase becomes Widmanstatten-type consisting of long platelet shape, so that the further transformation from the δ phase to ε phase gives rise to the unique Widmanstatten-type morphology of the Laves phase. The morphology change in the Laves phase thus suggests that the γ″→δ phase transformation preferentially takes place prior to the precipitation of the Laves phase within the grain interiors in the composition and temperature ranges where both δ and ε phases form within the grain interiors during aging.

Fig. 13.

Schematic illustration showing how the precipitation morphology of the Fe₂Nb Laves phase becomes Widmansttaten-type, in terms of the transformation sequence of the precipitates within the grain interiors during aging: (a) Ni₃Nb-γ″, (b) Ni₃Nb-δ, (c) Fe₂Nb-ε phase. The lines represent the intersection of four {111}_γ planes on the observation surface in a grain with <111>_γ orientation perpendicular to the observation surface.

4. Conclusions

The effects of Zr addition on the precipitation behavior and morphology of the Fe₂Nb Laves (TCP) phase and Ni₃Nb (GCP) phases in a carbon-free Fe-20Cr-35Ni-2.5Nb novel austenitic heat-resistant steel have been investigated from viewpoints of thermodynamics and kinetics, and the following conclusions can be drawn:

(1) The maximum solubility of Zr in the fcc matrix in the novel steel is about 0.1 at.% and 0.02 at.% at around 1473 K and 1073 K, respectively. The further addition of Zr induces a Zr enriched primary solidification phase.

(2) The Zr dissolved in the matrix decreases the Nb concentration in the fcc matrix phase in equilibrium with the Laves and δ phases, thereby increasing the Nb supersaturation for the formation of the intermetallic phases.

(3) The Zr dissolved in the matrix at 1473 K is mostly enriched in the Ni₃Nb-δ phase formed by aging at 1073 K, and the temperature limit of thermodynamically stable δ phase increases to 1123 K, nearly 50 K higher that in the steel without Zr.

(4) At the grain boundaries, the Zr dissolved in the matrix is effective in promoting the precipitation of grain-boundary Fe₂Nb Laves phase from the biginning of the aging, and the grain-boundary area fraction covered by the Laves phase in the steel with 0.1 at.% Zr becomes higher by at least 10% than that in the steel without Zr at 1073 K and 1123 K.

(5) Within the grain interiors, the Zr dissolved in the matrix enhances the homogeneous nucleation of metastable Ni₃Nb-γ″ phase in the early stage of aging at 1073 K and 1123 K, and the formation of transformed Ni₃Nb-δ phase from the γ″ phase is eventually enhanced as the aging proceeds. The further aging promotes the formation of ε phase at the γ/δ interfaces, and this is more obvious at 1123 K.

(6) The formation sequence from γ″→δ→ε phase within the grain interiors drastically changes the precipitation morphology of the Laves phase from a random distribution type to Widmanstatten-type by Zr addition.

(7) From these results, it is Zr in solution, not segregation to the grain boundaries, that causes the change in precipitation microstructures of TCP and GCP phases at grain boundaries and within the grain interiors, even though the amount of dissolved Zr in the matrix is very small. These findings on the Zr effects would make great contributions to construct the microstructure design principle to further improve the high-temperature strength of the novel steels.

Acknowledgements

This study was supported by the Grant-in-Aid (JY240118) on the Advanced Low Carbon Technology Research and Development Program (ALCA), Japan Science and Technology Agency (JST).

References

• 1)   F.  Masuyama: ISIJ Int., 41 (2001), 612. https://doi.org/10.2355/isijinternational.41.612
• 2)   R.  Viswanathan,  J. F.  Henry,  J.  Tanzosh,  G.  Stanko,  J.  Shingledecker,  B.  Vitalis and  R.  Purgert: J. Mater. Eng. Perform., 14 (2005), 281. https://doi.org/10.1361/10599490524039
• 3)   R.  Viswanathan,  J.  Shingledecker,  J.  Hawk and  S.  Goodstein: Proc. 2nd ECCC Creep Conf. on Creep & Fracture in High Temperature Components, (Zurich), ed. by I. A. Shibli and S. R. Holdsworth, DEStech Publications, Inc., Lancaster, PA, (2009), 31.
• 4)   J.  Shingledecker,  R.  Purgert and  P.  Rawls: Proc. 7th Int. Conf. on Advances in Materials Technology for Fossil Power Plants, (Waikoloa), ed. by D. Gandy and J. Shingledecker, ASM International, Materials Park, OH, (2013), 41.
• 5)   M.  Fukuda,  E.  Saito,  H.  Semba,  J.  Iwasaki,  S.  Izumi,  S.  Takano,  T.  Takahashi and  Y.  Sumiyoshi: Proc. 7th Int. Conf. on Advances in Materials Technology for Fossil Power Plants, (Waikoloa), ed. by D. Gandy and J. Shingledecker, ASM International, Materials Park, OH, (2013), 24.
• 6)   M.  Render,  M. L.  Santella,  X.  Chen,  P. F.  Tortorelli and  V.  Cedro III: Metall. Mater. Trans. A, 52 (2021), 2601. https://doi.org/10.1007/s11661-021-06253-1
• 7)   C.  Crozet,  A.  Devaux and  D.  Béchet: MATEC Web Conf., 14 (2014), 17006. https://doi.org/10.1051/matecconf/20141417006
• 8)   H.  Semba,  H.  Okada,  T.  Hamaguchi,  S.  Kurihara and  K.  Tanaka: Proc. Joint EPRI - 123HiMAT Int. Conf. on Advances in High-Temperature Materials, (Nagasaki), Vol. 60, ed. by J. Shingledecker and M. Takeyama, ASM International, Materials Park, OH, (2019), 513.
• 9)   A.  Iseda,  H.  Okada,  H.  Semba and  M.  Igarashi: Energy Mater., 2 (2007), 199. https://doi.org/10.1179/174892408X382860
• 10)   A. F.  Padilha and  P. R.  Rios: ISIJ Int., 42 (2002), 325. https://doi.org/10.2355/isijinternational.42.325
• 11)   M.  Takeyama: Mater. Sci. Forum, 539–543 (2007), 3012.
• 12)   M.  Takeyama: 194th and 195th Nishiyama Memorial Lecture, ISIJ, Tokyo, (2008), 1 (in Japanese).
• 13)   Y.  Yamamoto,  M.  Takeyama,  Z. P.  Lu,  C. T.  Liu,  N. D.  Evans,  P. J.  Maziasz and  M. P.  Brady: Intermetallics, 16 (2008), 453. https://doi.org/10.1016/j.intermet.2007.12.005
• 14)   Y.  Hasebe,  K.  Hashimoto and  M.  Takeyama: J. Jpn. Inst. Met., 75 (2011), 265 (in Japanese). https://doi.org/10.2320/jinstmet.75.265
• 15)   I.  Tarigan,  N.  Takata and  M.  Takeyama: JSPS 123rd Committee Report on Heat-resistant Materials and Alloys, Vol. 52, JSPS, Tokyo, (2011), 63.
• 16)   I.  Tarigan,  N.  Takata and  M.  Takeyama: Proc. 12th Int. Conf. on Creep and Fracture of Engineering Materials and Structures (JIMIS 11), (Kyoto), ed. by K. Maruyama et al., The Japan Institute of Metals, Sendai, (2012), 55.
• 17)   Y.  Misosaku,  I.  Tarigan,  T.  Kimura,  N.  Takata,  M.  Ueda,  T.  Maruyama and  M.  Takeyama: Tetsu-to-Hagané, 100 (2014), 1158 (in Japanese). https://doi.org/10.2355/tetsutohagane.100.1158
• 18)   N.  Kanno,  K.  Yoshimura,  N.  Takata,  I.  Tarigan and  M.  Takeyama: Mater. Sci. Eng. A, 662 (2016), 551. https://doi.org/10.1016/j.msea.2016.03.038
• 19)   M.  Takeyama: Bull. Iron Steel Inst. Jpn., 23 (2018), 414 (in Japanese).
• 20)   M.  Takeyama: 237th and 238th Nishiyama Memorial Lecture, ISIJ, Tokyo, (2019), 53 (in Japanese).
• 21)   M.  Takeyama: Bull. Iron Steel Inst. Jpn., 24 (2019), 505 (in Japanese).
• 22)   M.  Takeyama: Kinzoku, 90 (2020), 550 (in Japanese).
• 23)   M.  Takeyama: Materia Jpn., 60 (2021), 281 (in Japanese). https://doi.org/10.2320/materia.60.281
• 24)   J. C.  Zhao,  V.  Ravikumar and  A. M.  Beltran: Metall. Mater. Trans. A, 32 (2001), 1271. https://doi.org/10.1007/s11661-001-0217-4
• 25)   M.  Detrois,  P. D.  Jablonski and  J. A.  Hawk: Mater. Sci. Eng. A, 799 (2021), 140337. https://doi.org/10.1016/j.msea.2020.140337
• 26)   F.  Gao: Ph.D. thesis, Tokyo Institute of Technology, (2016).
• 27)   R. F.  Decker and  J. W.  Freeman: The mechanism of beneficial effects of boron and zirconium on creep-rupture properties of a complex heat-resistant alloy, University of Michigan Library, https://deepblue.lib.umich.edu/handle/2027.42/4525, (accessed 2006-02-03).
• 28)   J. H.  Schneibel,  C. L.  White and  M. H.  Yoo: Metall. Trans. A, 16 (1985), 651. 10.1007/BF02814239
• 29)   J. R.  Davis: ASM Specialty Handbook: Heat-Resistant Materials, ASM International, Materials Park, OH, (1997), 231.
• 30)   Y.  Tsai,  S.  Wang,  H.  Bor and  Y.  Hsu: Mater. Sci. Eng. A, 607 (2014), 294. https://doi.org/10.1016/j.msea.2014.03.136
• 31)   W.  Hume-Rothery,  R.  Smallman and  C. W.  Haworth: The Structure of Metals and Alloys, Institute of Metals, London, (1969), 100.
• 32)   E. T.  Teatum,  K. A.  Gschneidner, Jr. and  J. T.  Waber: Compilation of Calculated Data Useful in Predicting Metallurgical Behavior of the Elements in Binary Alloy Systems, Los Alamos Scientific Laboratory, Los Alamos, NM, (1968), 11. https://doi.org/10.2172/4789465
• 33)   H.  Li,  F.  Gao,  S.  Kobayashi and  M.  Takeyama: JSPS 123rd Committee Report on Heat-resistant Materials and Alloys, Vol. 56, JSPS, Tokyo, (2016), 121.
• 34)   F.  He,  D.  Chen,  B.  Han,  Q.  Wu,  Z.  Wang,  S.  Wei,  D.  Wei,  J.  Wang,  C. T.  Liu and  J.  Kai: Acta Mater., 167 (2019), 275. https://doi.org/10.1016/j.actamat.2019.01.048