Precipitation Behavior of ω_o Phase in Ti-37.5Al-12.5Nb Alloy

Abstract

Mutual transformation between α₂ and ω_o phases has been an interesting topic in recent years. In this study, martensitic α₂ was obtained by air-cooling from 1250 °C in Ti-37.5Al-12.5Nb (at%) alloy while four ω_o variants formed in the β_o phase matrix during the cooling process. Nonetheless, only one ω_o variant was observed at the periphery of the α₂ plates in the β_o phase and the orientation relationship between these two phases was [$0001$] α₂//[$1\overline{2}10$] ω_o; ($11\overline{2}0$) α₂//(0002) ω_o. Thin γ plates precipitated within the α₂ phase and were thought to be related to the appearance of ω_o phase. The redistribution of the compositions during the phase transformations was studied by energy dispersive X-ray spectroscopy analysis. The corresponding mechanisms of the phase transformations mentioned above are discussed.

1. Introduction

High Nb-containing TiAl (Nb-TiAl) alloys have been considered as potential materials for high-temperature applications due to their low density, high strength, good oxidation resistance, and creep properties [1,2,3]. Recently, Stark et al. showed that the amount of ω_o phase increased with the content of Nb in high Nb-TiAl alloy [4,5]. Meanwhile, Nb is a β phase stabilizer that extends the β phase field and facilitates the ordered ω (ω_o) phase transitions in the β_o phase in high Nb-TiAl alloys [6,7,8,9]. Numerous studies have reported the ω_o phase transformations in high Nb-TiAl alloys, indicating that the ω_o phase is stable at 700–900 °C [10,11]. However, these studies have mainly focused on the transition process between the β_o and ω_o phases [6,7,8,9,10,11,12,13] or the α₂ to β_o phase [14,15,16]. Recently, some reports concentrated on the precipitation of the ω_o phase in α₂ laths during aging and studied the relationship between the α₂ and ω_o phases [10,17,18,19,20]. To summarize, there are two different thoughts regarding the precipitation of ω_o from α₂ phase. First, Huang et al. reported the perpendicular decomposition of coarse α₂ laths in Ti-44Al-8Nb-B alloy and suggested that the α₂ to β_o(ω) transformation occurred after exposing at 700 °C in air for up to 10,000 h, indicating the occurrence of α₂→β_o→ω_o transformation [17]. Similar cases of α→β→ω transformation in titanium alloys had been reported by Vohra et al. [18] and Gupta et al. [19]. Second, Bystrzanowski et al. observed that the applied stress could enhance the ω_o precipitation and suggested that the ω_o precipitation was directly transformed from the α₂ phase [20]. Furthermore, Song et al. observed the direct α₂ to ω_o phase transformation in Ti-45Al-9Nb alloy after aging at 900 °C [10]. Although these studies discussed the transformation process and orientation relationships (ORs) between the ω_o and α₂ phases, few reports focused on the nucleation sites of the ω_o phase and the preferential ORs between the ω_o and α₂ phases. Moreover, the nucleation behavior of ω_o particles associated with α₂ phase in β_o phase has scarcely been reported. In this work, the precipitation of ω_o phase in Ti-37.5-12.5Nb alloy was examined. The preferential OR between the ω_o and α₂ phases was evaluated. The corresponding mechanisms were also discussed.

2. Materials and Methods

An ingot of the Ti-37.5Al-12.5Nb (at%) used in this study was prepared using induction levitation melting. The ingot was flipped and remelted three times to ensure compositional homogeneity.

Table 1. Chemical composition of the as-cast material.

Elements	Ti	Al (at%)	Nb (at%)	O (wt%)	N (wt%)
Composition	Bal.	37.0	13.0	0.018	0.0086

lists the chemical compositions measured via wet chemical analysis. Specimens with sizes of 10 × 10 × 10 mm were cut from the center of the ingot by electric-discharge machining. The specimens were heat treated at 1250 °C for 2 h followed by air cooling with a cooling rate of approximately 20 K/s. The microstructures after heat treatments were examined using a Zeiss Supra 55 scanning electron microscope (SEM) in back-scattered electron (BSE) mode. Thin foils used for transmission electron microscopy (TEM) observation were prepared by twin-jet electro-polishing in a solution of 65 vol% methanol, 30 vol% butanol, and 5 vol% perchloric acid at 30 V and −30 °C. TEM analysis was conducted on a Tecnai G² F30 field emission transmission electron microscope operating at 300 kV. The compositions were obtained by energy dispersive X-ray spectroscopy (EDS) on TEM. Each parameter was an average value of more than five results measured at different locations.

3. Results

The actual composition of the alloy is Ti-37.0Al-13.0 Nb, as obtained by chemical analyses, which is close to the nominal composition. Figure 1a shows the BSE image of the Ti-37.5Al-12.5Nb alloy after air-cooling. The microstructure is composed of α₂ plates and β_o matrix which can be identified by TEM as in Figure 1b and Figure 2b. The well-defined dark lines (arrowed in Figure 1a) observed in α₂ plates are believed to be the midribs of the martensite, which is similar to the result for the α₂ phase form from the β phase by iced-brine quenching in Ti-44Al-4Nb-4Hf-0.1Si [21]. It is difficult to distinguish whether the ω_o phase exists in the β_o region or not from the SEM image. Thus, the precipitation behavior of the ω_o phase can be studied by using TEM. Figure 1b shows the bright-field TEM image of the air-cooled sample. Some particles with sizes of tens of nanometers distribute uniformly in the β_o region. The corresponding selected area diffraction (SAD) pattern of the β_o region is shown in Figure 1c, indicating that β_o phase can readily transform to ω_o phase during air cooling in this alloy.

Commonly, the observed ω_o phase in high Nb-TiAl alloys can form from the “ω-collapse” in β_o phase [6], i.e., the {111} β_o layers “collapse” and the “-A-B-A-B-A-B-A-” stacking sequence in β_o phase transforms into “-A-B/A-B-A/B-A-”. There are four equivalent {111} β_o layers so that four possible ω_o variants can form in one β_o grain. Considering the ORs between these four ω_o variants and β_o phase, the {$01\overline{1}0$} ω_o diffraction spots of two ω_o variants (denoted as “ω_o1” and “ω_o2” in Figure 1c) can be observed at 1/3{112} β_o under the zone axes: <110> β_o//<$2\overline{11}0$> ω_o1, ω_o2. However, the <$01\overline{1}2$> zone axes of “ω_o3” and “ω_o4” are parallel with <110> β_o thus the diffraction patterns of these zone axes are overlapped completely. As a result, the intensities of the superposition spots of the ω_o and β_o phases are significantly increased in Figure 1c.

Figure 2a shows the bright-field image of the α₂ plates. The circled area in Figure 2a demonstrates an almost precipitate-free region except for some particles that precipitate at the boundary of the α₂ phase. The corresponding SAD pattern of this area is shown in Figure 2b. Only one {$01\overline{1}0$} ω_o spot exists at 1/3{112} β_o under the same beam direction as Figure 1c, which indicates that only one ω_o variant exists at the α₂/β_o boundary. The OR between these phases is obtained as:

$$\left[110\right] {\mathsf{\beta }}_{\mathrm{o}}//[0001]{\mathsf{\alpha }}_2//[1\overline{2}10] {\mathsf{\omega }}_{\mathrm{o}}; (1\overline{1}1) {\mathsf{\beta }}_{\mathrm{o}}//(11\overline{2}0) {\mathsf{\alpha }}_2//\left(0002\right){\mathsf{\omega }}_{\mathrm{o}}$$

Figure 2c is the corresponding dark field image taken by using the diffraction spot of the ω_o variant, as circled in the SAD pattern in Figure 2b. It is indicated that the precipitates at the α₂/β_o boundary are of one kind of ω_o variants. The compositions of the different phases (denoted in Figure 2a) were obtained by EDS equipped on TEM in

Table 2. Energy dispersive X-ray spectroscopy (EDS) results of different phases.

Regions	Composition (at %)
	Ti	Al	Nb
βo martix	49.9 ± 0.8	38.5 ± 0.5	11.6 ± 0.8
ωo	49.5 ± 0.7	36.7 ± 0.5	13.8 ± 0.8
α2 plate	51.1 ± 0.8	38.1 ± 0.5	10.8 ± 0.8

. The results show that the ω_o precipitates at the periphery of the α₂ phase are more concentrated in Nb than β_o-martix and α₂ phase. This case can be interpreted in that Nb is a stabilizing element of the ω_o phase rather than of the β_o and α₂ phases [22,23,24]. Thus, ω_o precipitations at the α₂/β_o boundary are more concentrated in Nb than β_o and α₂ phases because of the expulsion of Nb in the β_o and α₂ phases during the cooling process.

Figure 3 shows the High-resolution TEM (HRTEM) image of the α₂/β_o interface obtained under [0001] α₂ direction. The fast Fourier transformation (FFT) images of the ω_o and β_o areas are shown in Figure 3b,c respectively. It is demonstrated that ω_o phases sized about a few tens of nanometers nucleate at the α₂ boundary in Figure 3a. This indicates that the preferential ω_o phase can nucleate at the boundary of α₂ phase and keep a certain OR with α₂ phase.

Further studies on the interior of the α₂ phase reveal that there are a few stacking faults in it. Figure 4a shows some fine-scale planar defects within the α₂ phase, indicating certain phase transformations occur, as also suggested by the distortions at the boundary of the α₂ plate. Figure 4b is the HRTEM image of the interface between the α₂ and β_o phase. The β_o region and an ω_o precipitate nucleated at the α₂ boundary are observed (the corresponding FFT images are shown in Figure 4c,d). It is worth pointing out that although the beam direction is [$11\overline{2}0$] α₂ and the ω_o precipitate is under [0001] ω_o direction, the OR between these two phases is as same as those obtained in Figure 2 and Figure 3. The magnified image of Figure 4b is shown in Figure 4e, the atomic stacking sequence of the α₂ phase changes from “-ABAB-” to “-ABCABC-” (FCC-stacking) due to Shockley partial dislocations moving on alternate basal plane (0001) α₂ planes. Repeating this mechanism every two basal planes of the hexagonal matrix leads to the crystal structure change, thus, the stacking faults can act as the nucleus of the γ phase [25,26]. Moreover, it has also been reported that γ phase precipitated in this alloy after annealing at 700 °C for 26 days [6]. However, the γ phase observed in [6] consists of γ grains precipitated directly from the matrix and not thin γ laths transformed from the α₂ phase. Despite the different morphologies, these facts suggest that the γ phase is an equilibrium phase.

4. Discussion

4.1. Single ω_o Variant Nucleated at the α₂ Boundary

Transformation matrices are useful for calculating the ORs between the precipitation and matrix phase, which has been widely used in calculating the habit-planes and misorientations [27,28,29,30]. As described above, four possible ω_o variants exist in the β_o phase. The ORs between the ω_o variants and β_o phase and between the ω_o and α₂ phases are calculated by using the transformation matrices. The transformation matrices B for β_o to ω_o phases are shown in

Table 3. The transformation matrices B for four ω_o variants from the β_o phase.

Variants	Orientation Relationship	Transformation Matrices B for βo to ωo
B1	(111) βo//(0001) ωo; [$1\overline{1}0$] βo//[$2\overline{11}0$] ωo	$\begin{array}{ccc}\sqrt{2}/2& -\sqrt{2}/2& 0\\ \sqrt{6}/6& \sqrt{6}/6& -\sqrt{6}/3\\ \sqrt{3}/3& \sqrt{3}/3& \sqrt{3}/3\end{array}$
B2	($\overline{1}11$) βo//(0001) ωo; [$110$] βo//[$2\overline{11}0$] ωo	$\begin{array}{ccc}\sqrt{2}/2& \sqrt{2}/2& 0\\ -\sqrt{6}/6& \sqrt{6}/6& -\sqrt{6}/3\\ -\sqrt{3}/3& \sqrt{3}/3& \sqrt{3}/3\end{array}$
B3	($1\overline{1}1$) βo//(0001) ωo; [$110$] βo//[$2\overline{11}0$] ωo	$\begin{array}{ccc}\sqrt{2}/2& \sqrt{2}/2& 0\\ -\sqrt{6}/6& \sqrt{6}/6& \sqrt{6}/3\\ \sqrt{3}/3& -\sqrt{3}/3& \sqrt{3}/3\end{array}$
B4	($11\overline{1}$) βo//(0001) ωo; [$1\overline{1}0$] βo//[$2\overline{11}0$] ωo	$\begin{array}{ccc}\sqrt{2}/2& -\sqrt{2}/2& 0\\ -\sqrt{6}/6& -\sqrt{6}/6& -\sqrt{6}/3\\ \sqrt{3}/3& \sqrt{3}/3& -\sqrt{3}/3\end{array}$

(see Appendix A.1 for details). According to the Burgers OR between the α₂ and β_o phases: {110} β_o//($0001$) α₂; <$1\overline{1}1$> β_o//<$11\overline{2}0$> α₂, six β_o variants can form from the α₂ phase and the transformation matrices C for these variants are shown in

Table 4. The transformation matrices C for six β_o variants from the α₂ phase.

Variants	Orientation Relationship	Transformation Matrices C for α2 to βo
C1	(110) βo//(0001) α2; [$1\overline{1}1$] βo//[$2\overline{11}0$] α2	$\begin{array}{ccc}1/\sqrt{3}& 1/\sqrt{6}& 1/\sqrt{2}\\ -1/\sqrt{3}& -1/\sqrt{6}& 1/\sqrt{2}\\ 1/\sqrt{3}& -2/\sqrt{6}& 0\end{array}\times L$
C2	(110) βo//(0001) α2; [$1\overline{11}$] βo//[$2\overline{11}0$] α2	$\begin{array}{ccc}1/\sqrt{3}& -1/\sqrt{6}& 1/\sqrt{2}\\ -1/\sqrt{3}& 1/\sqrt{6}& 1/\sqrt{2}\\ -1/\sqrt{3}& -2/\sqrt{6}& 0\end{array}\times L$
C3	(110) βo//(0001) α2; [$1\overline{1}1$] βo//[$\overline{1}2\overline{1}0$] α2	$\begin{array}{ccc}1/\sqrt{3}& 1/\sqrt{6}& 1/\sqrt{2}\\ -1/\sqrt{3}& -1/\sqrt{6}& 1/\sqrt{2}\\ 1/\sqrt{3}& -2/\sqrt{6}& 0\end{array} \times L\times \begin{array}{ccc}-1& 1& 0\\ -1& 0& 0\\ 0& 0& 1\end{array}$
C4	(110) βo//(0001) α2; [$1\overline{11}$] βo//[$\overline{1}2\overline{1}0$] α2	$\begin{array}{ccc}1/\sqrt{3}& -1/\sqrt{6}& 1/\sqrt{2}\\ -1/\sqrt{3}& 1/\sqrt{6}& 1/\sqrt{2}\\ -1/\sqrt{3}& -2/\sqrt{6}& 0\end{array} \times L\times \begin{array}{ccc}-1& 1& 0\\ -1& 0& 0\\ 0& 0& 1\end{array}$
C5	(110) βo//(0001) α2; [$1\overline{1}1$] βo//[$\overline{11}20$] α2	$\begin{array}{ccc}1/\sqrt{3}& 1/\sqrt{6}& 1/\sqrt{2}\\ -1/\sqrt{3}& -1/\sqrt{6}& 1/\sqrt{2}\\ 1/\sqrt{3}& -2/\sqrt{6}& 0\end{array} \times L\times \begin{array}{ccc}0& -1& 0\\ 1& -1& 0\\ 0& 0& 1\end{array}$
C6	(110) βo//(0001) α2; [$1\overline{11}$] βo//[$\overline{11}20$] α2	$\begin{array}{ccc}1/\sqrt{3}& -1/\sqrt{6}& 1/\sqrt{2}\\ -1/\sqrt{3}& 1/\sqrt{6}& 1/\sqrt{2}\\ -1/\sqrt{3}& -2/\sqrt{6}& 0\end{array} \times L\times \begin{array}{ccc}0& -1& 0\\ 1& -1& 0\\ 0& 0& 1\end{array}$

(see Appendix A.2 for details). The hypothesis is that the ORs between the α₂ and ω_o phases can be transferred by the relationships between α₂ to β_o and β_o to ω_o. The OR between the α₂ and ω_o phases (matrices T) can be readily obtained by

T_α2→ωo = B × C

(1)

L is the transformation matrix from the crystallographic coordinate system to the orthogonal coordinate system (see Appendix A). By using Equation (1), the matrices T for α₂ to ω_o phases are obtained, deriving the arbitrary parallel crystallographic directions between the α₂ and ω_o phases. According to these results, only four ω_o variants have good lattice matching and two ORs between the α₂ and ω_o phases are obtained (see Appendix A.2 for details):

$$<2\overline{11}0>{\mathsf{\alpha }}_2//<0001>{\mathsf{\omega }}_{\mathrm{o}}; \left\{0002\right\}{\mathsf{\alpha }}_2//\left\{2\overline{11}0\right\} {\mathsf{\omega }}_{\mathrm{o}}\mathrm{ORI}$$

$$<2\overline{11}0>{\mathsf{\alpha }}_2//<2\overline{2}01>{\mathsf{\omega }}_{\mathrm{o}}; \left\{0002\right\}{\mathsf{\alpha }}_2//\left\{01\overline{1}2\right\} {\mathsf{\omega }}_{\mathrm{o}}\mathrm{ORII}$$

ORII can be also expressed as <$\overline{1}010$>α₂//<$2\overline{4}23$>ω_o; {$0002$} α₂//{$01\overline{1}2$} ω_o by using a superimposed stereographic projection. Thus, both ORs calculated from the transformation matrices are the same as the results obtained by edge-to-edge matching calculation in Ti-45Al-9Nb alloy [10]. According to the results, the selection of OR between the α₂ and ω_o phases is essentially based on which ω_o variant has good lattice matching with α₂ phase. It was reported that the misfit between the α₂ and ω_o phases had a minimum value if ORI was formed [10]. Thus, it is believed that only one ω_o variant nucleated at the α₂ boundary.

Moreover, according to the EDS results in Table 2, the composition of the ω_o particle nucleated at the boundary of the α₂ phase is more concentrated in Nb than the β_o matrix. It is suggested that the ω_o phase primarily nucleates at the boundary of the α₂ phase and enriches in Nb during growth. Thus, the untransformed area of the periphery of the α₂ phase is depleted in Nb so that the precipitate-free regions are observed.

4.2. Thin γ Plates Precipitated within the α₂ Phase

Because of the misfit matching between the α₂ and ω_o phases, distortions at the interface are expected. The interplanar spacings of the (0002) α₂ and ($11\overline{2}0$) ω_o planes measured by HRTEM and SAD software are 0.233 nm and 0.229 nm, respectively. That is a 1.7% mismatch in the interplanar spacing when ORI is formed between the α₂ and ω_o phases. It means that α₂ phase may have an extra (0002) α₂ plane after successive stacking of 59 pairs of ($11\overline{2}0$) ω_o and (0002) α₂ planes. The interplanar spacing of (111) γ is 0.232 nm is smaller than that of the (0002) α₂ but larger than ($11\overline{2}0$) ω_o. Thus, the fine γ plates may relieve the distortion at the interface of the ω_o and α₂ phases (Figure 4e). It has been reported that the precipitation of γ in α₂ is simply a HCP→FCC structure change which can be brought about if a/6 <$10\overline{1}0$> type Shockley partials move on alternate basal plane (0001) α₂ planes [31,32]. As mentioned above, an extra (0002) α₂ plane exists after successive stacking of 59 pairs of ($11\overline{2}0$) ω_o and (0002) α₂ planes. This case may cause the distorted-region separated along the α₂ and ω_o interface. As a consequence, the sliding of the partial dislocations in the distorted-region can produce separated fine γ plates at certain intervals. Moreover, the interplanar spacing of 59 (0002) α₂ planes is approximate 13.7 nm, which is consistent with the average spacing of the γ laths, which is approximate 12 nm as obtained from Figure 4a.

5. Conclusions

In this work, the precipitation of ω_o phase in Ti-37.5Al-12.5Nb alloy was examined mainly by TEM. The ORs between different phases were calculated. The main results are summarized as follows:

• Only one ω_o variant preferentially nucleates at the α₂ boundaries. This is because the minimum misfit exists at the α₂/ω_o interface if the OR between these two phases is: <$2\overline{11}0$> α₂//<0001> ω_o; {$0002$} α₂//{$2\overline{11}0$} ω_o.
• Precipitate-free regions are observed at the α₂ boundaries. EDS results indicate that the ω_o precipitates are more concentrated in Nb than β_o-matrix. The preferred nucleation of the ω_o variant causes solute depletion surrounding the α₂ plates, which inhibits the nucleation and growth of new ω_o precipitates in the un-precipitated regions.
• Thin γ plates precipitate within the α₂ phase. These fine γ plates can relieve the distortion caused by the mismatch at the α₂/ω_o interface.