Characterization of the high temperature deformation behavior of two intermetallic TiAl–Mo alloys

Characterization of the high temperature deformation behavior of two intermetallic TiAl–Mo alloys

Materials Science & Engineering A 648 (2015) 208–216 Contents lists available at ScienceDirect Materials Science & Engineering A journal homepage: w...

5MB Sizes 0 Downloads 1 Views

Materials Science & Engineering A 648 (2015) 208–216

Contents lists available at ScienceDirect

Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea

Characterization of the high temperature deformation behavior of two intermetallic TiAl–Mo alloys Flora Godor a,n, Robert Werner a, Janny Lindemann b, Helmut Clemens a, Svea Mayer a a b

Department of Physical Metallurgy and Materials Testing, Montanuniversität Leoben, Roseggerstr. 12, A-8700 Leoben, Austria GfE Fremat GmbH, Lessingstr. 41, D-09599 Freiberg, Germany

art ic l e i nf o

a b s t r a c t

Article history: Received 15 June 2015 Received in revised form 14 September 2015 Accepted 15 September 2015 Available online 18 September 2015

Intermetallic titanium aluminides are promising candidates for high-temperature components in aero and automotive applications. To enable good processing characteristics with an optimized final microstructure, the hot-working parameters and the fraction of the β/βo-TiAl phase at deformation temperature are of particular interest. Therefore, the high-temperature deformation behavior of two γ-TiAl based alloys with the nominal compositions Ti–41Al–3Mo–0.5Si–0.1B and Ti–45Al–3Mo–0.5Si–0.1B, in at%, was studied. At room temperature both alloys contain the ordered phases γ-TiAl, βo-TiAl and small amounts of α2-Ti3Al. In order to investigate dynamic restoration during thermomechanical processing, s isothermal compression tests were conducted on a Gleeble 3500 simulator and corresponding flow curves were measured. The tests were carried out at temperatures from 1150 °C to 1300 °C, applying strain rates ranging from 0.005 s  1 to 0.5 s  1, up to a true strain of 0.9. The deformed microstructural states of the multiphase alloys, particularly the dynamically recrystallized grain sizes, were characterized by means of scanning electron microscopy and electron back-scatter diffraction. To compare the microstructure right before and after deformation heat treatments were additionally performed at the same temperatures as the compression tests were carried out. The experimentally determined flow stress data were described with two different constitutive models (Sellars–McTegart model, Hensel–Spittel model). The experimentally determined dynamically recrystallized grain sizes of the hot-deformed microstructures were linked with the Zener–Hollomon parameter calculated from the simulation through a power law. & 2015 Elsevier B.V. All rights reserved.

Keywords: Intermetallics Constitutive equation Dynamic recrystallization Deformation behavior Heat treatment Microstructure

1. Introduction Structural materials should have low densities and withstand high stresses and temperatures to fulfill the requirements of the automotive and aircraft engine industry. In future, fuel consumption, harmful emissions, like the greenhouse gases (CO2, NOx), as well as noise levels must be reduced [1]. Therefore, in the last two decades great effort was undertaken to develop innovative intermetallic γ-TiAl based alloys. These alloys possess the potential to replace other materials, for example the heavier nickel base alloys or steels, in application for turbocharger turbine wheels in automotive engines or for turbine blades in the low pressure turbine of aircraft engines. Intermetallic γ-TiAl based alloys have low densities (about 4 g cm  3), good corrosion and oxidation resistance, a high melting point and outstanding mechanical properties, such as a high specific strength and Young's modulus [2]. Their special n

Corresponding author. Fax: þ43 3842 402 4202. E-mail address: [email protected] (F. Godor).

http://dx.doi.org/10.1016/j.msea.2015.09.077 0921-5093/& 2015 Elsevier B.V. All rights reserved.

thermo-physical properties are due to their mixed metallic and covalent bonding characteristics and the ordered phases occurring at low temperatures. However, this is also the reason for the brittleness and the difficult workability of this alloy class [3]. For a long time, manufacturing of TiAl alloys with conventional forming processes has posed a challenge for technical applications. It is therefore necessary to study the material's behavior at high temperature deformation to design an economic manufacturing process with optimized forming parameters. Many research groups have investigated the deformation behavior of γ-TiAl based alloys consisting of three phases α/α2 (α: A3 structure, α2: D019 structure), β/βo (β: A2 structure, βo: B2 structure) and γ (L10 structure) or consisting of two phases α/α2 and γ as reported in [4–14] amongst others. Up to now, in the literature only limited information can be found about the hot-deformation behavior of alloys containing high amounts of β/βo-phase. In the present work the microstructural evolution of two different TiAl–Mo alloys with high contents of β/βo-phase was investigated after hot-deformation. At low temperatures the alloys consist of the ordered phases

F. Godor et al. / Materials Science & Engineering A 648 (2015) 208–216

209

Fig. 1. SEM images (BSE mode) showing the cast/HIP microstructure: (a) Ti–41Al–3Mo–0.5Si–0.1B alloy (termed: Ti–41Al–3Mo) and (b) Ti–45Al–3Mo–0.5Si–0.1B alloy (termed: Ti–45Al–3Mo).

βo-TiAl, γ-TiAl and negligible amounts of α2-Ti3Al. At higher temperatures, however, βo-TiAl and α2-Ti3Al disorder to β-Ti(Al) and α-Ti(Al), respectively. In order to study dynamic recovery and recrystallization during thermomechanical processing, isothermal compression tests were s carried out on a Gleeble 3500 simulator. During deformation flow curves (true stress–true strain curves) were measured. The tests were conducted at temperatures from 1150 °C to 1300 °C, applying strain rates ranging from 0.005 s  1 to 0.5 s  1, up to a true strain of 0.9. The deformed microstructure of the alloys, particularly the dynamically recrystallized grain sizes, were analyzed using scanning electron microscopy (SEM) and electron back-scatter diffraction (EBSD). The experimentally obtained flow stress data were mathematically described with two different constitutive models (Sellars–McTegart, Hensel–Spittel). The experimentally determined grain sizes of the deformed microstructures were linked with the Zener–Hollomon parameter calculated from the simulation through a power law. To analyse the initial microstructure before the onset of hotdeformation both alloys were subjected to heat treatments. The temperatures were chosen according to the applied hot-deformation temperatures (1150–1300 °C). A short-term annealing for 5 min was conducted to adjust the microstructure, which is present prior to deformation. To compare these microstructural states with a state closer to equilibrium, annealing treatments for 60 min were performed additionally.

2. Materials and methods 2.1. Experimental Two γ-TiAl based alloys with the nominal compositions Ti– 41Al–3Mo–0.5Si–0.1B (termed: Ti–41Al–3Mo) and Ti–45Al–3Mo– 0.5Si–0.1B (termed: Ti–45Al–3Mo) in at% were investigated. In general, Mo is added to TiAl alloys to improve their high temperature performance. It is a strong β-stabilising element [11] and impedes diffusion processes by increasing the activation energy of diffusion and thus decreasing the climb rate of dislocations. This enhances the creep and thermal stability and enables applications at higher temperatures. On the other side the kinetics of phase transformations as well as recrystallization processes are slowed down. However, Mo tends to have a negative impact on the oxidation resistance [15–17]. As Si has a low solubility in the alloys, silicides are formed, which improve the creep strength and stabilize the microstructure at elevated temperatures [3]. Furthermore, Si shifts the α/α2-phase field to higher temperatures and increases the γ-solvus temperature [18]. As B forms borides during solidification, it is used for grain refinement [19]. Ingots were produced by means of vacuum arc remelting and

centrifugal casting in permanent molds at GfE Metalle und Materialien GmbH, Germany. In order to close residual porosity, the ingots were additionally hot-isostatically pressed (HIPed) at a temperature of 1210 °C and a pressure of 200 MPa for 4 h followed by furnace cooling with a cooling rate of o 8 K min  1. For the heat treatments rectangular block specimens (20  20  10 mm3) and for the compression tests cylindrical specimens with 10 mm in diameter and 15 mm in length were machined from the HIPed ingots. s The compression tests were conducted on a Gleeble 3500 simulator at 1150 °C, 1200 °C, 1250 °C, and 1300 °C with constant strain rates of 0.005 s  1, 0.05 s  1, and 0.5 s  1, up to a true strain of 0.9. The specimens were resistance heated with a heating rate of 5 K s  1 and soaked for 5 min, before deformation was started in vacuum with a pressure of 10  4 mbar. During hot-compression flow curves were measured by converting load-displacement data into true stress–true strain data. After reaching the final degree of deformation, the heating was switched off and the specimens were cooled to room temperature. Both alloys were subjected to annealing treatments (5 min and 60 min) to characterize the microstructure at test temperature prior to deformation. The specimens were therefore heated in the high-temperature chamber furnace RHF 1600, annealed for 5 min or 60 min respectively and subsequently water quenched. The annealing temperatures correspond with those of the compression tests (1150 °C, 1200 °C, 1250 °C and 1300 °C). The deformed specimens were cut parallel to the compression axis. Metallographic preparation was performed according to [20] in order to show the initial, deformed and annealed microstructures. The microstructures were analysed by means of SEM employing a Zeiss EVO 50. In Fig. 1 the microstructures in the asreceived condition (cast/HIP) are presented. In the back-scattered electron (BSE) mode the Mo-rich β-phase exhibits the brightest contrast, whereas the γ-phase appears as the darkest. To determine the phase transition temperatures between 1000 °C and 1450 °C, differential scanning calorimetry (DSC) measurements were performed [21]. The phase fractions of the initial and annealed specimens were examined by X-ray diffraction (XRD) on a Bruker AXS D8 Advance Diffractometer in Bragg– Brentano arrangement using Cu-Kα-radiation. For the Rietveld analysis the software package Topas of Bruker AXS was used. In order to identify the microstructural mechanisms during the forming process and to measure grain sizes, especially the dynamically recrystallized grain size, EBSD measurements were conducted. The EBSD of the FIB VERSA 3D microscope equipped with an EDAX Hikari XP EBSD camera and the software TSL OIM Analysis 7 was used for the characterization of the deformed microstructures. For these measurements an area of 90  90 mm2 was scanned with a step size of 0.3 mm.

210

F. Godor et al. / Materials Science & Engineering A 648 (2015) 208–216

2.2. Simulation

be linked with the experimentally determined recrystallized grain sizes dRxx using the following equation:

In the second part of this work the experimentally obtained flow curves were described with two different constitutive equations. To this end, the Sellars–McTegart model (ST-model) [22] was compared with the Hensel–Spittel model (HS-model) [23]. The STmodel is based on the following hyperbolic-sine Arrhenius-type relationship:

dRxx = k⋅Z −m,

Z = Z (ϵ, ϵ̇, T ) = ϵ̇·e

Q (ϵ) RT

wherein k and m are constants and m is characteristic for the respective material.

3. Results and discussion

= f (σ (ϵ, ϵ̇, T ))

= A(ϵ)· sinh(α(ϵ)·σ(ϵ, ϵ̇, T ))n(ϵ) ,

(1)

where Z is the Zener–Hollomon parameter acting as a temperature-compensated strain rate, ε̇ is the strain rate, Q(ε) is the activation energy of deformation, R is the universal gas constant, T is the deformation temperature in [K] and σ (ε, ε,̇ T ) is the true stress. The parameters A(ε), α(ε), n(ε) and Q(ε) are dependent on the material and are assumed to be constant for each individual strain rate [5,22]. In the HS-model a material-specific initial value A0 is multiplied by the factors Kϑ , Kε and Kε̇, expressing the influence of the forming parameters (temperature, strain and strain rate).

σ = σHS = σHS(ϵ, ϵ,̇ ϑ) = A 0 ⋅Kϑ⋅Kϵ⋅ Kϵ ̇ =A 0 ⋅[exp(m1⋅ϑ)]⋅[ϵm2⋅exp(m4 /ϵ)⋅(1 + ϵ)(m5 ⋅ϑ)⋅exp(m7⋅ϵ)] ⋅[(ϵ̇/ϵ0̇ )(m3 + m8 ⋅ϑ)]

(3)

(2)

In Eq. (2) ε is the true strain, ϑ is the deformation temperature in [°C], A0 and the exponents m1–m8 are fitting parameters describing the effect of deformation conditions on the flow stress σ and ε̇0 is a normalization parameter for the strain rate with the value 1 s  1 [23–25]. All parameters were calculated with Wolfram Mathematica 9.0 by using a multi-nonlinear regression analysis. The Zener–Hollomon parameter Z, calculated from Eq. (1), can

3.1. Heat treatments The heat treatments prove, that the investigated materials reach a state close to thermodynamic equilibrium after annealing for 60 min. After annealing for 5 min, which represents the microstructure prior to deformation, the microstructure is in a state of non-equilibrium. As can be seen in Fig. 2, the differences between the phase fractions after annealing for 5 min and 60 min are larger at higher temperatures because of enhanced diffusion processes. Furthermore, with increasing temperature the γ-phase fraction decreases until the γ-solvus temperature is reached, whereas the α/α2-phase fraction increases. The Ti–41Al–3Mo alloy consists of the βo- and γ-phase until 1200 °C, but for the Ti–45Al– 3Mo alloy the two-phase state remains preserved until 1250 °C. The SEM images of the annealed microstructures, as shown in Fig. 3, confirm these findings. The DSC measurements indicate that the γ-solvus temperature of the Ti–41Al–3Mo alloy is 1245 °C, whereas in the Ti–45Al–3Mo alloy it is 1305 °C. In the first alloy the ordering reaction βo2β occurs at 1200 °C, in the second one at 1265 °C [21]. 3.2. Deformation behavior In Fig. 4 the results of the compression tests and of the

Fig. 2. Phase fractions of the Ti–41Al–3Mo alloy (a,b) and of the Ti–45Al–3Mo alloy (c,d) after annealing for 5 min (a,c) and for 60 min (b,d). After the respective heat treatment the specimens were water quenched. The data were measured by means of XRD and evaluated by Rietveld analysis [18].

F. Godor et al. / Materials Science & Engineering A 648 (2015) 208–216

211

Fig. 3. SEM images taken in BSE mode of the Ti–41Al–3Mo alloy (a,b) and of the Ti–45Al–3Mo alloy (c,d) after annealing at 1200 °C for 5 min (a,c) and for 60 min (b,d). After the heat treatments the specimens were water quenched. While the images (a–c) show a two-phase (βo, γ) microstructure, it is evident that in (d) three phases (βo, γ, α2) coexist, which is in agreement with Fig. 2.

simulation are exemplarily presented for hot-deformation at 1200 °C and different strain rates. The measured flow curves (solid lines) can be divided into three parts. At the beginning the true stress increases because of work-hardening. This implies that the dislocation density increases until the peak value is reached. When a critical degree of deformation is exceeded, dynamic restoration processes set in causing a decrease of the true stress. Dynamic recrystallization does not necessarily start at the peak stress, but already at lower strain values as reported in references [26–28]. In

the third stage the true stress drops to a steady state value, which indicates the balance between work hardening and dynamic recrystallization. The decrease of the true stress is characteristic, if dynamic recrystallization and not dynamic recovery dominates during deformation [22,29]. In Fig. 4 it is also demonstrated that the peak values of the Ti– 45Al–3Mo alloy are significantly higher, because at deformation temperature the alloy contains less βo-phase. The flow curves of both alloys show that the true stress is higher at lower

Fig. 4. Experimentally determined flow curves (solid lines) of the Ti–41Al–3Mo (a,c) and of the Ti–45Al–3Mo (b,d) at 1200 °C applying three different strain rates. The solid symbols show the flow curves calculated from the ST-model (a,b), whereas the open symbols show the flow curves obtained from the HS-model (c,d).

212

F. Godor et al. / Materials Science & Engineering A 648 (2015) 208–216

Fig. 5. SEM images taken in BSE mode of the Ti–41Al–3Mo alloy (a,b,c) and of the Ti–45Al–3Mo alloy (d,e,f) after deformation at 1200 °C with 0.005 s  1 (a,d), 0.05 s  1 (b,e) and 0.5 s  1 (c,f). In all micrographs the compression direction is vertical.

deformation temperatures and higher strain rates. Moreover, at higher strain rates there is less time for dislocation movement, on the other hand higher temperatures facilitate the dislocation movements by thermal activation [30]. According to [1,3,5,10] the flow softening after the peak stress can be ascribed to dynamic recrystallization. Thus, the measured decrease of the flow stress is higher for the Ti–45Al–3Mo alloy, which contains more γ-phase. This suggests that mainly the γ-phase recrystallizes during hotcompression. After compression SEM images in BSE mode were taken, focusing on the central part of the specimens. In Fig. 5 the microstructures after deformation at 1200 °C with different strain rates are exemplarily shown. In accordance with the heat treatment study (Fig. 2(a and c)) at 1200 °C the microstructures of both alloys contain two phases (βo, γ) and only an insignificant volume fraction of α2-Ti3Al. It is apparent that at lower strain rates the grains are more spherical, whereas after deformation at higher strain rates the grains are elongated perpendicular to the compression axis. In order to guarantee comparability only the results of the Gleeble tests conducted in the (βo þ γ)-phase field region were taken into consideration in the simulation part as well as for the EBSD measurements and for the identification of the deformation behavior. Furthermore, the fringe zone of the specimens were examined with SEM. At the investigated deformation temperatures and strain rates no cracks or round-type cavities were found, thus no flow instabilities occur. The dynamically recrystallized grain size dRxx was determined separately for each phase using grain color maps obtained from the EBSD scans. The software TSL OIM Analysis 7, which was used for the evaluation of the EBSD measurements, calculates grain diameters from the area assuming the grain is a circle. Thereby, the measured grain sizes are smaller for higher strain rates and lower deformation temperatures as reported in [1,3–5,10] as well. A closer analysis of the grain morphology and grain sizes proves that both phases (βo þ γ) participate in deformation, although for both alloys at 1150 °C, 1200 °C and for the Ti–45Al–3Mo alloy up to 1250 °C the βo-phase is ordered. Morris et al. [6] also found that in a Ti–44Al–2Mo alloy both, the βo- and the γ-phase, show deformability at elevated temperatures. By means of SEM and transmission electron microscopy (TEM) the authors could show

that the γ-TiAl phase was deformed by ordinary 1/2 o110] dislocations and mechanical twinning along 1/6 o112]. At the beginning of the deformation process only the γ-phase is deformed. When higher loads are applied or when a sufficient stress concentration is reached at the βo/γ-grain boundaries, the βo-phase starts to deform as well. Then ordinary o 1004 dislocations and o111 4 superdislocations on {110} planes are activated in the βophase by compression [6]. In both alloys before deformation the microstructure consists of large, irregularly shaped grains. The Ti–41Al–3Mo alloy contains approximately 45 m% γ and 55 m% βo at deformation temperatures (1150 °C, 1200 °C; see Fig. 2(a and c)). The βo-phase forms a continuous matrix, where elongated γ-grains are embedded. As the grain size maps indicate, the largest grains in both phases are about 25 mm. After the compression tests the average grain sizes are smaller, the largest βo-grain sizes are 12 mm, whereas the largest γ-grain sizes are 16 mm. The βo- and γ-grains rotate into a direction perpendicular to the forming direction and are divided into subgrains with different orientations. An overview of the orientation maps of the βo-phase in the Ti–41Al–3Mo after annealing for 5 min at 1200 °C and after compression at 1150 °C and 1200 °C with strain rates of 0.05 s  1 and 0.5 s  1 is given in Fig. 6. In the enlarged grain boundary maps (Fig. 6) high-angle grain boundaries (disorientation 415°) are marked with blue lines, low-angle grain boundaries (disorientation o5°) with red lines and grain boundaries with a disorientation of 5–15° with green lines. In the βo-phase the grains are elongated and surrounded by high-angle grain boundaries. Within these grains high-angle grain boundaries and also low-angle grain boundaries are detected, the latter forming subgrains. Therefore, it is likely that in the βo-phase continuous dynamic recrystallization is the main mechanism during hot-compression. This observation is supported by the fact, that within the βo-phase dynamic recovery processes are preferred due to its high stacking fault energy [31,32]. Continuous recrystallization is connected to extended recovery processes, which proceed after strong work-hardening or when the sliding of grain boundaries is impeded [28,33]. Then the whole microstructure is renewed by formation of high-angle and low-angle grain boundaries, which are homogeneously distributed over the deformed area. In contrast, during ordinary dynamic recrystallization the

F. Godor et al. / Materials Science & Engineering A 648 (2015) 208–216

213

Fig. 6. Orientation maps generated from the EBSD measurements of the βo-phase in the Ti–41Al–3Mo alloy after annealing for 5 min at 1200 °C (a) and after deformation at 1150 °C (b,c) and 1200 °C (d,e) with strain rates of 0.005 s  1 (b,d) and 0.05 s  1 (c,d).

microstructure is rebuilt by nucleation and growth including movement of high-angle grain boundaries [28,33]. In the γ-phase, beside a small fraction of elongated grains, also globular ones can be found. The grain boundary maps show that the globular grains are sharply surrounded by high-angle grain boundaries and that there are no subgrains inside. The elongated grains show a subgrain structure, which indicates the occurrence of dynamic recovery. Thus, it can be concluded that in the γ-phase of the Ti– 41Al–3Mo alloy both, dynamic recrystallization and dynamic recovery, contribute to the restoration of the microstructure, whereas dynamic recrystallization is the dominating mechanism. In the Ti–45Al–3Mo alloy the specimens, which were deformed at 1150 °C, 1200 °C and 1250 °C, were analysed by means of EBSD. In this temperature range the alloy contains approximately 75 m% γ and 25 m% βo. After annealing for 5 min the γ-phase forms a continuous matrix with embedded βo-grains. The largest βo-grains are measured as 12 mm, whereas the largest γ-grain size is about 32 mm. Before deformation only grains surrounded by high-angle grain boundaries were detected. After the Gleeble tests the alloy develops grain sizes up to 8 mm in both phases. As the alloy starts to deform the γ-matrix has to endure a large part of the stress during deformation. The influence of the forming parameters on the deformation behavior can also be expressed by the Zener– Hollomon parameter Z. With decreasing temperature and/or

increasing strain rate the parameter Z increases. An increase of Z includes a stronger driving force for recrystallization because of the higher dislocation density resulting in a lower grain size of the recrystallized grains. On the other hand the fraction of recrystallized grains is smaller when forming at low temperatures with high strain rates [4,34]. Within the γ-TiAl phase only globularly developed grains were found, meaning that dynamic recrystallization is the predominant restoration mechanism. However, the fraction of recrystallized γ-grains depends on the forming parameters and on the β/γ-ratio. From Fig. 4 it is evident that flow softening due to dynamic recrystallization is stronger in the Ti– 45Al–3Mo alloy than in the Ti–41Al–3Mo alloy, because it contains more γ-TiAl phase. It is reported by several research groups that in the γ-TiAl phase dynamic recrystallization is the preferred mechanism of restoration during hot-forming [3,4,10,35]. In the βophase of the Ti–45Al–3Mo alloy, however, mostly elongated grains with a subgrain structure were detected (comparable to the grain boundary maps shown in Fig. 6 for the Ti–41Al–3Mo alloy). Therefore, the dominating mechanism during compression is continuous recrystallization. The investigations proved that in both alloys within the βophase continuous recrystallization is the main mechanism during deformation, whereas the γ-phase tends to recrystallize dynamically. For further information and a clearer separation of the

214

F. Godor et al. / Materials Science & Engineering A 648 (2015) 208–216

Fig. 7. Relationship between the constitutive parameters of the ST-model and the true strain ε as derived for the Ti–45Al–3Mo alloy. The discrete data points were calculated for every true strain value recorded during the compression tests by using Eq. (1). With these data points an interpolation (dashed line) and a polynomial fit (solid line) was conducted.

deformation mechanisms additional phase-specific investigations, for example in-situ high-energy XRD experiments, e.g. see Ref. [8], would be required. 3.3. Constitutive modeling As described in Section 3.2 only the flow stress data obtained from the compression tests in the (βo þ γ)-phase field region were used for the constitutive modeling of the flow curves. Therefore, the calculations of the Ti–41Al–3Mo alloy were performed using a 3  2-matrix (3 strain rates, 2 deformation temperatures), however, for the Ti–45Al–3Mo alloy a 3  3-matrix was used (3 strain rates, 3 deformation temperatures). The modeling was conducted with a Wolfram Mathematica 9.0 program code developed by Werner et al. [5,21]. In Fig. 7 the constitutive parameters of the ST–model are shown for the Ti–45Al–3Mo alloy, which were obtained by solving Eq. (1) for each parameter separately. The discrete data points result from the arithmetic mean of the slopes (Fig. 7(a)–(c)) or the ordinate (Fig. 7(d)) of a linear data fit conducted for every true strain value (ε ¼0–0.9). Furthermore, the dependence of the respective constitutive parameter from the true strain values ε was calculated by interpolation and polynomial fit of the experimental data. By using the ST-model the measured true stress–true strain data can be well approached as shown in Fig. 4(a and b). The workhardening and also the flow softening are described accurately. However, the drawback is that this model can only be used to predict flow stresses within the experimentally determined strain range as depicted in Fig. 8. For example finite element analysis requires analytical flow curves that can be extrapolated to higher strain levels. If an extrapolation of the flow stress data is demanded it is recommended to use the HS–model instead. From Fig. 8 it is obvious that the HS–model provides a proper reproduction of the flow stress data within the experimental strain range. Additionally, it is suitable for extrapolation to higher strain levels. A further advantage is the shorter program code, because all required fitting parameters (A0 and m1–m8) can be calculated

Fig. 8. Flow curves of the Ti–45Al–3Mo alloy at a constant strain rate of 0.5 s  1 at different deformation temperatures. Comparison between the experimentally determined flow curves (solid lines), the calculated flow curves from the ST–model (triangular shaped symbols) and from the HS–model (circular shaped symbols).

Table 1 Parameter of the HS–model calculated from Eq. (2) for the Ti–41Al–3Mo alloy and the Ti–45Al–3Mo alloy. Parameter

Unit

Ti–41Al–3Mo

Ti–45Al–3Mo

A0 m1 m2 m3 m4 m5 m7 m8

[MPa] [°C  1] [Dimensionless] [Dimensionless] [Dimensionless] [°C  1] [Dimensionless] [°C  1]

2,303,990  0.00872333  0.437465 0.916574  0.0241143 0.00267539  2.0393  0.000578847

377,820  0.00622528  0.196918  1.24363  0.0192131  0.000253714  0.172619 0.00130777

according to Eq. (2) in a single step through a multi-nonlinear regression analysis. The corresponding parameters for the HS– model are given in Table 1. From the ST–model (see Eq. (1)) the activation energy of

F. Godor et al. / Materials Science & Engineering A 648 (2015) 208–216

215

Fig. 9. Relationship between the dynamically recrystallized γ-grain size dRxx and the Zener–Hollomon parameter Z for the Ti–41Al–3Mo and the Ti–45Al–3Mo alloy. The values for Z (ε,̇ T ) were calculated with the activation energies from the ST–model (see text). The plotted data points were obtained from measurements at 1150 °C and 1200 °C with 0.005 s  1, 0.05 s  1, 0.5 s  1. In case of alloy Ti–41Al–3Mo the experiment at 1150 °C and 0.5 s  1 was not successful, therefore this data point is not included in the diagram. Due to a lack of material the tests could not be repeated.

deformation Q(ε) can be calculated for every true strain value. At the final degree of deformation (ε ¼0.9) the activation energies in the Ti–41Al–3Mo alloy and Ti–45Al–3Mo alloy are 267 kJ mol  1 and 553 kJ mol  1, respectively. Other research groups also report activation energies for various TiAl alloys within 300 kJ mol  1 and 700 kJ mol  1 [4,36,37]. As demonstrated in Fig. 4, the flow stresses are higher in the Ti–45Al–3Mo alloy, which contains more γ-TiAl. The calculated activation energies are also higher within this alloy, proving that with increasing βo-content the flow stress and therefore the activation energy decreases. In order to guarantee good workability and optimized microstructures, the forming parameters and also the fraction of βo-TiAl must be chosen carefully. Using the activation energies the Zener–Hollomon parameter Z (ε,̇ T ) can be determined for every value of strain rate and temperature T (Eq. (1)). By employing Eq. (3) the experimentally determined recrystallized grain size dRxx can be linked with the Zener–Hollomon parameter Z, which is obtained from the ST–model. As the EBSD experiments proved that only the γ-phase recrystallizes during compression, the recrystallized grain sizes were evaluated phasespecifically for the γ-phase. In contrast, within the βo-phase the dominating mechanism during deformation is continuous dynamic recrystallization, which is a strong recovery process. The relationship between the dynamically recrystallized γ-grain size dRxx and the Zener–Hollomon parameter Z is shown in Fig. 9. Linear regression results in m-values of about  0.14 for both alloys. Kim [4] reported similar values for lamellar γ-TiAl alloys consisting of three phases (α/α2 þ β/βo þ γ).

4. Conclusions In the present study the high temperature deformation behavior of two Mo bearing γ-TiAl based alloys, Ti–41Al–3Mo–0.5Si– 0.1B and Ti–45Al–3Mo–0.5Si–0.1B, which contain high amounts of the ordered B2 βo-phase, was investigated. Isothermal compression tests were performed between 1150 °C and 1300 °C applying strain rates of 0.005 s  1, 0.05 s  1 and 0.5 s  1. During deformation true stress–true strain curves were generated. For the detailed analysis of the deformation mechanisms only the specimens, which were deformed within the (βo þ γ)-phase field region, have

been considered. The Ti–41Al–3Mo–0.5Si–0.1B alloy consists of two phases (βo-TiAl and γ-TiAl) until a temperature of 1200 °C is reached, whereas the Ti–45Al–3Mo–0.5Si–0.1B alloy shows a twophase microstructure up to 1250 °C. With increasing Al-content the γ-phase fraction in the alloy increases, whereas the βo-phase fraction decreases. The present fraction of βo-TiAl, however, has a great impact on the workability. An increase of the βo-phase reduces the maximum flow stress as well as the activation energy for deformation. In order to enable conventional forming processes, for example conventional forging, both forming parameters and fraction of βo-phase must be set properly. The measured flow curves and the EBSD measurements demonstrate the influence of the forming parameters. The grain sizes decrease with increasing strain rate and decreasing deformation temperature. Simultaneously, the flow stress increases. In the γ-phase mainly dynamic recrystallization occurs. The βophase tends to recover dynamically during deformation, due to its high stacking fault energy. As a consequence of the strong recovery effect, this phase undergoes a so-called continuous recrystallization. In the simulation part the Sellars–McTegart model and the Hensel–Spittel model were compared. For the experimentally measured strain range both models are suitable for predicting flow curves. However, only the HS-model can be used for extrapolating flow curves to higher strain levels and can thus provide input data for finite element simulations. Using the ST–model the Zener– Hollomon parameter Z( ε̇,T) can be calculated for each deformation temperature and strain rate. The experimental part and the simulation part could be combined with a power law, which links the dynamically recrystallized grain size with the calculated Zener–Hollomon parameter.

Acknowledgments The authors thank Gerhard Hawranek for expert technical assistance. This work was conducted within the framework of the German BMBF project O3X3530A.

216

F. Godor et al. / Materials Science & Engineering A 648 (2015) 208–216

References [1] H. Clemens, S. Mayer, Adv. Eng. Mater. 15 (2013) 191. [2] W. Wallgram, T. Schmölzer, L. Cha, G. Das, V. Güther, H. Clemens, Int. J. Mater. Res. 100 (2009) 1021. [3] F. Appel, J.D.H. Paul, M. Oehring, Gamma Titanium Aluminide Alloys: Science and Technology, Wiley-VHC, Weinheim, Deutschland, 2011. [4] H.Y. Kim, W.H. Sohn, S.H. Hong, Mater. Sci. Eng. A 251 (1998) 216. [5] R. Werner, E. Schwaighofer, M. Schloffer, H. Clemens, J. Lindemann, S. Mayer, Adv. Mater. Res. 922 (2014) 807. [6] M.A. Morris, Y.G. Li, Mater. Sci. Eng. A 197 (1995) 133. [7] F. Appel, R. Wagner, Mater. Sci. Eng. R Rep. 22 (1998) 187. [8] T. Schmölzer, K.D. Liss, C. Kirchlechner, S. Mayer, A. Stark, M. Peel, H. Clemens, Intermetallics 39 (2013) 25. [9] A. Bartels, C. Hartig, S. Willems, H. Uhlenhut, Mater. Sci. Eng. A 239–240 (1997) 14. [10] V. Seetharaman, S.L. Semiatin, Metall. Mater. Trans. A 27 (1996) 1987. [11] M. Takeyama, S. Kobayashi, Intermetallics 13 (2005) 993. [12] R. Kainuma, Y. Fujita, H. Mitsui, I. Ohnuma, K. Ishida, Intermetallics 8 (2000) 855. [13] T. Tetsui, K. Shindo, S. Kobayashi, M. Takeyama, Scr. Mater. 47 (2002) 399. [14] T. Tetsui, K. Shindo, S. Kaji, S. Kobayashi, M. Takeyama, Intermetallics 13 (2005) 971–978. [15] Y. Shida, H. Anada, Mater. Trans. 35 (1994) 623. [16] D. Pilone, F. Felli, A. Brotzu, Intermetallics 43 (2013) 131. [17] R. Pflumm, A. Donchev, S. Mayer, H. Clemens, M. Schütze, Intermetallics 53 (2014) 45. [18] T. Schmölzer, S. Mayer, E. Schwaighofer, T. Lippmann, P. Staron, A. Stark, H. Clemens, The Effect of Si on Phase Fractions and Transition Temperatures of a TiAl–Mo Alloy, Annual Report, Desy, Hamburg, Deutschland, 2010. [19] U. Hecht, V. Witusiewicz, A. Drevermann, J. Zollinger, Intermetallics 16 (2008) 969. [20] M. Schloffer, T. Schmölzer, S. Mayer, E. Schwaighofer, G. Hawranek,

[21] [22] [23]

[24] [25] [26] [27] [28] [29] [30] [31] [32] [33]

[34] [35] [36] [37]

F. Schimansky, F. Pyczak, H. Clemens, M. Schloffer, T. Schmoelzer, S. Mayer, E. Schwaighofer, Prakt. Metallogr. 48 (2011) 594. F. Godor, Charakterisierung Des Warmumformverhaltens Zweier Legierungen Im System TiAlMo (Diploma thesis), Montanuniversität Leoben, Austria, 2014. C.M. Sellars, W.J. McTegart, Acta Metall. 14 (1966) 1136. A. Hensel, T. Spittel, Kraft- Und Arbeitsbedarf Bildsamer Formgeburgsverfahren, VEB Deutscher Verlag für Grundstoffindustrie, Leipzig, Deutschland, 1978. M. El Mehtedi, F. Musharavati, S. Spigarelli, Mater. Des. 54 (2014) 869. N.S. Viii, S.B. Heidelberg, Part 2: Non-Ferrous Alloys – Light Metals, Springer Berlin Heidelberg, Berlin, Heidelberg, 2011. M. Luton, C. Sellars, Acta Metall. 17 (1969) 1033. T. Sakai, J.J. Jonas, Acta Metall. 32 (1984) 189. T. Al-Samman, G. Gottstein, Mater. Sci. Eng. A 490 (2008) 411. L. Cheng, X. Xue, B. Tang, H. Kou, J. Li, Intermetallics 49 (2014) 23. E. Macherauch, H.W. Zoch, Praktikum in Werkstoffkunde, Vieweg þTeubner, Wiesbaden, Deutschland, 2011. B. Liu, Y. Liu, Y.P. Li, W. Zhang, A. Chiba, Intermetallics 19 (2011) 1184. C. Leyens, M. Peters, Titanium and Titanium Alloys, Wiley-VHC, Weinheim, Deutschland, 2003. R.D. Doherty, D.A. Hughes, F.J. Humphreys, J.J. Jonas, D. Juul Jensen, M. E. Kassner, W.E. King, T.R. McNelley, H.J. McQueen, A.D. Rollett, Mater. Sci. Eng. A 238 (1997) 219. H.Z. Niu, Y.Y. Chen, S.L. Xiao, F.T. Kong, C.J. Zhang, Intermetallics 19 (2011) 1767. M. Nobuki, K. Hashimoto, J. Takahashi, T. Tsujimoto, Mater. Trans. 31 (1990) 814. Z.J. Pu, K.H. Wu, J. Shi, D. Zou, Mater. Sci. Eng. A 192-193 (1995) 780. M. Nobuki, T. Tsujimoto, ISIJ Int. 31 (1991) 931.

True Detective 9.1 | Großhandel & Sonderposten | The End Of The Fucking World