Phase Transformation Behavior of a β-Solidifying γ-TiAl-Based Alloy from Different Phase Regions with Various Cooling Methods

Abstract

The phase transformation behavior of Ti-42Al-5Mn (at.%) alloy from different phase regions with various cooling rates was investigated based on electron probe micro analyzer-backscattered electrons (EPMA-BSE). It is shown that β→α₂′ takes place when this alloy is cooled at a high rate, such as water quenching (WQ), oil cooling (OC), from β single phase. With the decreasing cooling rate to air cooling (AC), β→α₂′ is restrained and β→γ is promoted by forming γ platelets. The room-temperature microstructure is β_o + α₂ when alloy cooled (WQ and OC) from (β + α) dual-phase. However, under AC, β→γ occurs and γ platelets form. It should be noted that α₂→γ happens when this alloy cooled from 1180 °C (>T_eut) by OC and AC, forming an incomplete lamellae (α₂/γ) structure in the α₂ phase. However, when the alloy cooled from 1100 °C (<T_eut), α₂/γ→β_o,sec occurs and complete lamellae generates in α₂ phase.

1. Introduction

In general, Ti and Ti alloys are quite attractive for aerospace, medical, and industrial applications [1,2,3,4], of which gamma-titanium aluminide (γ-TiAl) has been considered as one of the topmost candidates for high-temperature structural material due to its low density and high specific modulus and strength [5,6,7,8,9]. However, the intricate processing behavior, particularly the poor ductility and the low hot workability appears to be the main obstacle for a wide industrial application of γ-TiAl alloys [9]. In 1997, β-solidifying γ-TiAl alloys (via β-phase solidification) were proposed by Naka [10], also called beta solidified gammalloys (BSG) by Kim in 2018 [11], arousing increasing attention due to their excellent workability and the appropriate solidification process without any peritectic segregation [12]. These BSG alloys usually have one or more β stabilizing elements, such as Mo, Cr, Mn, V, Nb and a lower Al concentration (42–44%) when compared to conventional gammalloys (CG) [11], which finally results in their solidification pathway going through the single β-phase region.

As for the CG alloys, typically four types of microstructures can be obtained by heat treatment around the gamma dissolution temperature (T_γsolv), namely fully lamellar (FL), nearly lamellar (NL), duplex (DP) and near gamma (NG) [13,14]. The solidification procedure from the liquid to the stable solid state for the CG alloys can be given as L→L + β→(β + α) or (L + α)→α→α + γ→α₂ + γ. In the case of BSG alloys, since they contain a sufficient amount of β-stabilizing elements, the solidification pathway appears to be more complicated, and the typical structures of CG alloys are also no longer applicable. Clemens et al. [15,16] defined three types of structures for Ti-43.5Al-4Nb-1Mo-0.1B (TNM) alloy (BSG), including NL + β₁, NL + β₂, and NL + γ. The three microstructure consist of α₂/γ + β_o + γ, but there are significant differences in their morphology. From the aspect of the γ phase morphology, the γ phase in the first two kinds of structure is platelet, and the γ phase in the latter structure is block. Compared with NL + β₁ and NL + β₂, the difference between the two microstructures is that there are very few β_o,sec and γ phases around the colony boundary for NL + β₁, while there are plenty of β_o,sec and γ phases for NL + β₂.

Ti-42Al-5Mn, developed by Tetsui et al. in 2002 [17], is a typical BSG alloy, and it has the advantage of low cost and great workability, and seems to be a promising material for automotive industrial applications. Since then, Tetsui et al. reported some studies on this alloy from 2002 to 2013 [18,19,20], but their investigations were still limited to smelting and forging. Our team has also conducted more in-depth research on Ti-42Al-5Mn in recent years. Our previous research confirmed its solidification pathway from the liquid to room temperature as the following: liquid→liquid + β→β→β + α→β + α + γ→β_o + α₂ + γ→β_o + γ + α₂/γ→β_o + γ + α₂/γ + β_o,sec, with the T_β = 1311 °C, T_γsolv = 1231 °C, T_{β→βo/α→α2} = 1168 °C, T_eut = 1132 °C and T_{α2/γ→βo,sec} ≈ 1120 °C [21]. On this basis, the continuous cooling transformation curve (CCT) of the forged alloy was examined and a CCT diagram of the alloy was established [22]. It has been found that five kinds of phase transformation exist during continuous cooling from 1300 °C (forging temperature) with the cooling rates of 0.1 °C/s, 0.5 °C/s, 2 °C/s, 10 °C/s, 50 °C/s, and 200 °C/s, including β→α, β→γ, β→β_o/α→α₂, α₂→α₂/γ and α₂/γ→β_o,sec. Despite this, the study of the alloy is still at an initial stage; the corresponding microstructure evolution and phase transformation are not well understood. In order to exert the advantages of the alloy, it is overwhelmingly indispensable to grasp their phase transformation behavior.

In the present study, the phase transformation behavior of as-casted Ti-42Al-5Mn alloy from different phase regions (β, α + β, α + β + γ) with various cooling rate was further investigated. The work aimed to clarify the phase transition characteristics of Ti-42Al-5Mn in order to provide fundamental data, knowledge and information (DKI) for BSG alloy development.

2. Experimental Methods

Ti-42Al-5Mn (designated 42Al-5Mn) (at.%) as-cast ingot, with the dimension of Φ120 mm × 400 mm, was prepared by the Institute of Metal Research (IMR) (Shenyang, China) improved vacuum induction melting (VIM) and vacuum arc remelting (VAR) using titanium sponge (99.9 wt.%), pure aluminum (99.9 wt.%), and purified manganese (99 wt.%).

The samples with size of Φ8 mm × 8 mm were cut along the edge of the foot section of the as-cast ingot. Subsequently, each sample was put into a separate sealing tube filled with argon atmosphere to avoid oxidation, and the sealing tube was then placed in the muffle furnace. In order to ensure the consistency of each experiment, the samples were first heated to 1340 °C to ensure the α₂/γ lamellae and γ grain could be completely dissolved into the β phase. After holding for 30 min, samples were cooled by water quenching (WQ), oil cooling (OC), and air cooling (AC), and the decreasing ordering in cooling speed was WQ, OC, AC [23]. The other specimens were then furnace cooled (FC, valued about 5 °C·min⁻¹ [24]) to the target temperatures of 1260 °C (α + β), 1180 °C (α + β + γ, close to T_eut), and 1100 °C (α + β + γ). After holding for 30 min, these samples were also cooled with the WQ, OC, and AC methods, as illustrated in Figure 1. The microstructures of the specimens were characterized by electron probe micro analyzer (EPMA, JXA-8230, Northeastern University, Shenyang, China) with back scattered electron (BSE) and wave dispersive spectroscopy (WDS) mode.

3. Results

3.1. Microstructure Cooled from 1340 °C

Figure 2 shows the microstructure characteristics of the samples subjected to 1340 °C with different cooling methods. In this paper, β_o (ordered body-centered cubic (bcc) B2 structure) and α₂ (ordered hexagonal D0₁₉ structure) are used to describe the room-temperature phase composition. The reason is that the disorder–order phase transformation shows the least sensitivity to the cooling rate in γ-TiAl alloy, namely a high cooling rate could not suppress the nucleation for the transformation [25].

Based on our previous research [21], there should be only pure β phase at 1340 °C since the Ti-42Al-5Mn is in the β single-phase region. However, from Figure 2a, some needlelike martensitic structures, namely α′₂ [26] was observed in the above WQ microstructures. As described in reference. [22], the martensitic phase is considered to be the decomposition of β phase by β→α₂′. Moreover, upon quenching from the single β-phase field region a martensitic transformation was reported in “high-β” bearing γ-TiAl based alloys, such as TiAl-Mo [26], TiAl-V [27] or TiAl-Nb-Hf [28].

When cooled with the OC method (Figure 2b), the martensitic phase also existed. This means that the β→α₂′ cannot be restrained even though the cooling rate decreases to OC. It can also be seen that the size of α′₂ is obviously coarsened due to the relatively lower cooling rate than WQ. Besides, in this situation, some fine γ (ordered face-centered tetragonal L1₀ structure) phase named γ-platelets (γ_p) [29] are formed within the β_o phase, which is the decomposition of the β phase by β→γ.

With the cooling rate decreasing to AC (Figure 2c), it can be found that the needlelike martensitic structure disappeared, which suggests that the β→α₂′ is completely restrained at this cooling rate. The β_o phase is mainly distributed in parallel strips which are quite different from the above samples. The α₂ phase, defined as the supersaturated α₂-grain [30], was found between the parallel lines. It should be mentioned that the nucleated γ phase from β_o was also very fine, which is similar to the OC sample.

3.2. Microstructure Cooled from 1260 °C

Figure 3 presents the microstructure characteristics of the samples subjected to 1260 °C with different cooling methods. As can be seen in Figure 3a, in comparison with Figure 2a, plenty of supersaturated α₂-grains are precipitated from the single β phase when the sample was cooled from 1340 °C to 1260 °C by furnace cooling. Both the WQ and OC samples consisted of β_o and α₂, and no martensitic α₂′ phase was detected. This indicates that the β→α₂′ transformation was not only affected by the cooling rate but also influenced by the temperature before cooling.

When it cooled with the AC method (Figure 3c), the microstructure consisted of β_o, α₂ and γ phases. Lots of fine γ_p were nucleated from the β phase, and the γ phase grew in a staggered manner in the β phase, not parallel, as in Figure 2c.

3.3. Microstructure Cooled from 1180 °C

Figure 4 shows the microstructure characteristics of the samples subjected to 1180°C with different cooling methods. As can be seen in Figure 4a, in comparison with Figure 3a, plenty of globular γ grains named γ_g [29] with a median size of 15 μm were nucleated from the β phase. In the α₂ region, a few dispersive γ laths named γ_l also formed.

With decreasing cooling rates to OC and AC, the sizes of the γ_g and β_o phases were all intensively increased to a median size of 40 μm. Moreover, the amount of γ_l phase in the supersaturated α₂-grains increased as well. The incomplete α₂/γ lamellae called (α + γ) still formed for these three samples, which might be due to the 1180 °C being relatively higher than the T_eut (~1140 °C) [21].

3.4. Microstructure Cooled from 1100 °C

Figure 5 shows the microstructure characteristics of the samples subjected to 1100 °C with different cooling methods. As can be seen in Figure 5a, in comparison with Figure 4a, the complete α₂/γ lamellae formed. It should be mentioned that some fine β_o,sec phases were also nucleated in the lamellae due to the 1100 °C in the temperature range of the α₂/γ→β_o,sec transformation. When the cooling rates decreased to OC and AC (Figure 5b,c), the amount of β_o,sec phases increased. The lamellar spacing and γ grain size also significantly increased with the decreasing cooling rate.

4. Discussion

As stated above, the phase transformation behavior of the Ti-42Al-5Mn alloy from different phase regions with various cooling rate was obtained, as shown in

Table 1. Phase transformation behavior of Ti-42Al-5Mn from different phase regions with various cooling rates.

Temperature/°C	Phase Region [21]	Phase Transformation
		WQ	OC	AC
1340	β	β→α′2	β→α′2	β→α, β→γ
1260	β + α	/	/	β→α, β→γ
1180	β + α + γ	α→α + γ	α→α + γ	α→α + γ
1100	β + α + γ	α→α2/γ,α/γ→βo,sec	α→α2/γ,α/γ→βo,sec	α→α2/γ, α/γ→βo,sec

. It can be seen that the influence of the temperature and cooling rate on the phase transformation and microstructure of this alloy at room temperature was very significant. For the same phase region, with the decrease of the cooling rate, the microstructure of the sample coarsened, and the types of phase transitions increased. This was mainly due to the greater time for the lower cooling rate, which favors more phase transitions.

It should be mentioned that the γ morphology at room temperature was also closely related to the temperature and cooling rates.

Table 2. Corresponding relationship between temperature, cooling rate and γ’ morphology.

Temperature/°C	1340			1260			1180			1100
Cooling method	WQ	OC	AC	WQ	OC	AC	WQ	OC	AC	WQ	OC	AC
γg	×	×	×	×	×	×	√	√	√	√	√	√
γp	×	×	√	×	×	√	×	×	×	×	×	×
γl	×	×	×	×	×	×	√	√	√	√	√	√

Note: “×” is not found; “√” is found.

shows the corresponding relationship between temperature, cooling rate and γ’ morphology. As can be seen from this table, three types of γ’ morphology can be identified as stated in Section 3, including γ grain or globular (γ_g), γ platelet (γ_p), and γ lath (γ_l), which are illustrated in Figure 6. The γ_g with block morphology, was directly nucleated from the β phase. It was found that when the holding temperature was below T_γsolv, the γ_g could be obtained under the present cooling methods, such as WQ, OC and AC. In the case of γ_p, it was also directly formed in the β phase. However, for this structure, the sample only cooled by AC could be obtained when the holding temperature was above T_γsolv. These fine γ_p phases are usually dispersed in the β region. The third one is γ_l, which was precipitated from the α₂/γ lamellae, and its formation condition was similar to γ_g, i.e., the γ_l phase was almost accompanied by the formation of the γ_g phase. It was ascertained that the treating temperature below T_γsolv can promote the formation of γ_g and γ_l, and inhibit the appearance of γ_p. Whereas, the temperature above T_γsolv can restrain the nucleation of γ_g and γ_l, and when combined with a relative low cooling rate, for example, the AC method, it will promote the appearance of γ_p.

Although the γ_g and γ_l are all nucleated from the β phase, the size was different between these two phases. It can be seen from Figure 2, Figure 3, Figure 4 and Figure 5, that the size of γ_g was significantly larger than that of γ_p, which might be the result of their different precipitation conditions. In this study, the fine γ_p is often formed when cooled from a temperature that does not contain the γ phase region, such as the β or (β + α) phase region. In this situation, the γ_p is precipitated from the β phase during the continuous cooling with a relatively slow cooling rate. Such a phase would have a smaller size compared to the γ_g. However, once the temperature is lowered to the phase region including the γ phase, such as the (β + α + γ) phase region, the pre-precipitated γ phase furnace cooling from 1340 °C to the (β + α + γ) phase region would be further coarsened during the following cooling to form the γ_g. Moreover, it can be also noted that the size of γ_g would be obviously coarsened with a further decrease of the cooling rate. Hence, it is suggested that γ_g and γ_l can be clarified into the same phase, but the grain size between them is different due to their different precipitation behavior.

In fact, the different sizes of γ at the α₂/γ colony boundary have a significant effect on the properties of the γ-TiAl alloy. For the globular γ phase (γ_g), Mayer et al. [15] suggested that the γ_g at the α₂/γ colony boundary is a ductile phase itself due to its high slip lines, and it will help improve the ductility of the γ-TiAl alloy. In addition, Schwaighofer et al. [30] proposed that the presence of the γ phases at the colony boundary can intensively retard grain growth, but a high volume fraction of globular γ-grains would lead to a significant decrease of the yield strength at room temperature (RT) and creep strength at elevated temperature. Recently, in our research [31], we concluded that the γ phase at the α₂/γ colony boundary benefit in terms of improving the ductility of Ti-42Al-5Mn alloy depends on its grain size. This means that the γ phase can only significantly improve the plasticity of the alloy if it has a suitable size, i.e., too large or too small will have more or less adverse effects on the ductility of γ-TiAl alloy plasticity. Hence, it is necessary to effectively control the grain size of the γ phase to ensure that the alloy has good comprehensive performance in the regulation of the alloy microstructure and mechanical properties.

In particular, the fully lamellar microstructure usually exhibits a superior combination of strength, creep resistance, fracture toughness and ductility [32], i.e., the γ phase at α₂/γ colony boundary might be detrimental to the creep strength, then various heat treatments were applied to reduce the volume fraction of γ phase at the α₂/γ colony boundary. For instance, for Ti-44.5Al-6.25Nb-0.8Mo-0.1B (BSG), Bolz et al. [33] found that when treating at 1270 °C, which is just slight below the T_γsolv (1280 °C), and finally in combination with a same annealing system (800 °C/6 h/FC), the volume fraction of the globular γ-grain sharply reduced from 61.4% to 12.9% with the increased cooling rate from FC to OC, while that of the lamellar α₂/γ colonies increased intensively from 23.5% to 84.8%. Moreover, they also clarified that the volume fraction of the globular γ-grain can decrease to nearly zero when treating at 1300 °C holding for 1 h and cooling with the AC method, in combination with the 800 °C/6 h/FC annealing system. This means that treating at a higher temperature (normally above T_γsolv) and combining with a higher cooling rate can help to suppress the globular γ-grain, which inversely promotes γ_l formation under 800 °C/6 h/FC annealing system. Based on the present research, for Ti-42Al-5Mn, it was also seen that the γ’ morphology can gradually transition from globular to platelet, and both γ_g and γ_p can disappear completely with an increased treating temperature above T_γsolv combined with a high cooling rate (WQ, OC). Due to the absence of annealing treatment, for the sample accompanied with no γ_g and γ_p phases, large amounts of supersaturated α₂ phase would in turn form.

For the binary TiAl alloys (CG), it should be clarified that the globular γ-grain can emerge when the treating temperature is in the (α + γ) region. With temperature increases to above T_γsolv, the full lamellar microstructure without any globular γ-grains can be formed. This suggests that the γ phase at the α₂/γ colony boundary can be completely restrained if treating with a relatively high temperature above T_γsolv both for the CG and BSG alloy.

In addition, it should be mentioned that the parent phase of the γ phase α₂/γ colony boundary is not consistent in different γ-TiAl alloys. For the CG alloys, it has been claimed that this γ phase would form in the α-phase when the cooling rate is sufficiently high to suppress the formation of lamellar (α₂/γ) microstructure [34,35]. For BSG alloys, however, this γ phase would only nucleate directly from the β phase [36]. In general, the transformation paths of γ either from β or from α are closely related to the solidification pathway of γ-TiAl alloy. For the CG alloys, the solidification procedure is: L→L + β→β + α or L + α→α→α + γ→α₂ + γ. At low temperatures, the β phase will completely transform into the α phase. In this situation, the γ phase can only emerge from α phase. In contrast, for the BSG alloys the solidification pathway is: L→L + β→β→β + α→β + α + γ→β_o + α₂ + γ. The (β + α) phase will turn to (β + α + γ) with decreasing temperature. Based on the present results, the γ phase nucleates directly from the β phase while the α phase transforms into α₂/γ lamella with a γ lath inside. The phase transformation from α₂-Ti₃Al(D0₁₉) of γ-TiAl(L1₀) not only includes the obvious change of the atom stacking order but also the re-adjustment of the localized chemical composition by long-distance diffusion [37]. This phase transformation is controlled by thermal activation and the reaction rate is relatively slow. In fact, Bolz et al. [33] have concluded that a higher driving force is apparently necessary for α→α₂ + γ (γ_l) than that of β→γ (γ_g and γ_p). On the contrary, both β_o-TiAl(B₂) [38] and γ-TiAl (L1₀) belong to the cubic structure and have the structure correlation shown in Figure 7. Moreover, the local chemical atom composition of β_o is much closer to the γ phase. Hence, it can be seen that the γ phase is more likely to be precipitated from the β phase with the bcc structure than the α phase with the hcp structure.

5. Conclusions

The phase transformation behavior of Ti-42Al-5Mn alloy from different phase regions with various cooling rates was studied. The basic conclusions are the following:

• The β→α₂′ takes place when this alloy is cooled at a high rate (WQ and OC) from β single phase. With the decreasing cooling rate to AC, β→α₂′ is restrained and the β→γ is promoted by the formation of γ platelets.
• The room-temperature microstructure is β_o+α₂ when this alloy is cooled by WQ and OC from (β + α) the dual-phase. However, under AC, the β→γ takes place and γ platelets form.
• The α₂→γ occurs when this alloy is cooled from the (β + α + γ) temperature, slightly higher than T_eut (1132 °C), by WQ, OC and AC, forming incomplete lamellae (α₂/γ) structures in the α₂ phase. In this situation, plenty of globular γ grains rather than γ platelets are nucleated from the β phase, and the size of γ grains increased intensively with the decreasing cooling rate.
• When the alloy cooled from 1100 °C (<T_eut), the α₂/γ→β_o,sec occurs and complete lamellae are generated in the α₂ phase.