Microstructural design and mechanical properties of a cast and heat-treated intermetallic multi-phase γ-TiAl based alloy

Intermetallics 44 (2014) 128e140

Contents lists available at ScienceDirect

Intermetallics journal homepage: www.elsevier.com/locate/intermet

Microstructural design and mechanical properties of a cast and heattreated intermetallic multi-phase g-TiAl based alloy Emanuel Schwaighofer a, *, Helmut Clemens a, Svea Mayer a, Janny Lindemann b, c, Joachim Klose c, Wilfried Smarsly d, Volker Güther e a

Department of Physical Metallurgy and Materials Testing, Montanuniversität Leoben, Roseggerstr. 12, A-8700 Leoben, Austria Chair of Physical Metallurgy and Materials Technology, Brandenburg University of Technology, Konrad-Wachsmann-Allee 17, D-03046 Cottbus, Germany GfE Fremat GmbH, Lessingstr. 41, D-09599 Freiberg, Germany d MTU Aero Engines GmbH, Dachauer Str. 665, D-80995 Munich, Germany e GfE Metalle und Materialien GmbH, Höfener Str. 45, D-90431 Nuremberg, Germany b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 21 May 2013 Received in revised form 12 September 2013 Accepted 16 September 2013 Available online 10 October 2013

Advanced intermetallic multi-phase g-TiAl based alloys, such as TNM alloys with a nominal composition of Tie43.5Ale4Nbe1Moe0.1B (in at.%), are potential candidates to replace heavy Ni-base superalloys in the next generation of aircraft and automotive combustion engines. Aimed components are turbine blades and turbocharger turbine wheels. Concerning the cost factor arising during processing, which e additionally to material costs e significantly influences the final price of the desired components, new processing solutions regarding low-cost and highly reliable production processes are needed. This fundamental study targets the replacement of hot-working, i.e. forging, for the production of turbine blades. But without forging no grain refinement takes place by means of a recrystallization process because of the lack of stored lattice defects. Therefore, new heat treatment concepts have to be considered for obtaining final microstructures with balanced mechanical properties in respect to sufficient tensile ductility at room temperature as well as high creep strength at elevated temperatures. This work deals with the adjustment of microstructures in a cast and heat-treated TNM alloy solely by exploiting effects of phase transformations and chemical driving forces due to phase imbalances between different heat treatment steps and compares the mechanical properties to those obtained for forged and heat-treated material. Ó 2013 Elsevier Ltd. All rights reserved.

Keywords: A. Titanium aluminides, based on TiAl B. Mechanical properties at ambient temperature B. Mechanical properties at high temperatures B. Phase transformation C. Heat treatment D. Microstructure

1. Introduction Recent publications underline the enormous potential and huge demand for novel b-solidifying multi-phase g-TiAl based alloys for structural components like turbine blades and turbocharger wheels in modern high-performance combustion engines [1e7], such as the b-stabilized TNM alloy (TNM ¼ TiAleNbeMo) with a nominal composition of Tie43.5Ale4Nbe1Moe0.1B (in at.%) [8e10]. The main advantages of this alloy are a low density of about 4.1 g/cm3, a high specific elastic modulus of about 28 GPa/(g/cm3) and a high specific tensile yield strength of about 140 MPa/(g/cm3) at 750  C. This materials class also exhibits excellent oxidation resistance at service temperatures up to 800  C due to the high content of Al and Nb [11,12]. Detailed information regarding the TNM alloying concept can be found in papers of Clemens et al. [8,9,13,14]. In

* Corresponding author. Tel.: þ43 3842 4024204; fax: þ43 3842 4024202. E-mail address: [email protected] (E. Schwaighofer). 0966-9795/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.intermet.2013.09.010

general, the mechanical properties of TiAl alloys are determined by their chemical composition. However, for a fixed chemical composition balanced mechanical properties can be set only by an adjustment of appropriate microstructural features due to multiple heat treatments. In this context, the quantitative influence of various microstructural constituents on mechanical properties and, thus, the identification of critical processing parameters influencing the appearance of the final microstructure, are an essential knowledge for an application-oriented process and microstructure design. In current TiAl technology, as-cast and hot-isostatically pressed (cast/HIP) material is the starting condition for thermomechanical processing to adjust specific microstructures by means of forging and/or ensuing heat treatments. In general, forging of g-TiAl based alloys is conducted in a temperature range from 0.7 to 0.9$Tm, where Tm specifies the melting temperature in K, and requires a careful choice of processing parameters like temperature and strain rate [3,4,10]. After forging, which for TNM alloys can be either done under isothermal or near-conventional conditions, the material has

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

to be heat-treated via a multi-step heat treatment to ensure balanced mechanical properties at ambient as well as elevated temperatures. The major advantage of forging and subsequent heat treatments is the adjustment of small a2/g-colonies, which lead to an improved ductility of the final microstructure [15e17]. Recently, Wallgram et al. [18] have shown that grain refinement can be achieved by a combination of forging and post-forging recrystallization heat treatment, where phase transformations in interaction with recrystallization are taking place. This process leads to a homogeneous distribution of all phases within a fine-grained microstructure. Adversely, hot-forming procedures might imply the formation of undesired deformation textures, which lead to a more or less pronounced anisotropy of the mechanical properties. A forging texture, for example, cannot be fully removed by means of subsequent heat treatments. In addition, segregation effects arising from the solidification process cannot be completely compensated by hot-working and subsequent heat treatments. In the last decade, process development and microstructural evolution of b-solidifying g-TiAl based alloys were studied by several research groups for improving the mechanical properties. Mainly, the adjustment of desired microstructures was conducted by hot-forming operations and subsequent heat treatments starting from a cast/HIP ingot or powder metallurgically produced prematerial [9,19e24]. In these investigations, forming procedures, such as hot-extrusion and hot-forging, are playing a central role in respect to the evolution of the final microstructure. Thereby an increased dislocation density can help to refine coarse microstructural constituents by means of recrystallization processes. Throughout the employment of subsequent multi-step heat treatments, significant modifications of the microstructure can be set toward balanced mechanical properties of the final heat-treated material as reported in Ref. [14]. Furthermore, also experiments to study the influence of microstructural evolution starting from the cast/HIP condition were performed by means of single high temperature heat treatments with subsequent slow cooling [25]. Imaev et al. [26] showed an approach of grain refinement starting from an as-cast b-solidifying TiAl alloy via heat treatments in the single b-phase field followed by oil-quenching and subsequent isothermal annealing. Yang et al. [27] investigated the effect of discontinuous coarsening on grain refinement of cast/HIP peritectic TiAl alloys throughout the use of a cyclic heat treatment. However, in these studies the focus was put on grain refinement and not on the resulting mechanical properties and their optimization, i.e. high creep strength and sufficient ductility at room temperature (RT). In the framework of the present study, a number of multi-step heat treatments were applied to a cast/HIP TNM alloy in order to achieve similar mechanical properties compared to forged and heat-treated material [14]. Thereby, heat treatments within the single b-phase field are playing a key role for obtaining chemical and microstructural homogeneity and, thus, to reach acceptable mechanical properties without a prior hot-forming operation. Additionally, the development of a cyclic heat treatment for obtaining microstructures from the cast/HIP condition with improved RT-ductility is a further step to address higher ductility requirements [28]. The characterization of final heat-treated TNM alloys on the nano-scale regarding lamellar spacing within the

129

a2/g-colonies and the existence of u-phase domains with B82 structure within the bo-phase is reported in Refs. [14,29e31]. Based on these results, it is assumed that the observed “nanofeatures” are similar in our heat-treated microstructures, therefore TEM studies were neglected. Nevertheless, the present paper tries to identify the critical features for the design of microstructures by means of cost-effective and robust heat treatments starting from the cast/HIP condition in order to adjust balanced mechanical properties without hot-forming. 2. Material and experimental The TNM material of the present investigation was produced by GfE Metalle und Materialien GmbH, Nuremberg, Germany, via the so called “advanced beta process”. By this technique a powder metallurgically compacted electrode is vacuum arc remelted (VAR) for two times to achieve adequate chemical and structural homogeneity [32]. Detailed information on the applied melting technique can be found in Ref. [33]. Subsequently, the chemically homogenized electrode is melted in a VAR skull melter and cast by a centrifugal casting process [10]. The produced ingots had a diameter of 60 mm and a raw length of 297 mm. Due to the fast cooling rate during centrifugal casting a fine-grained cast microstructure is obtained and the deviation of the Al content along the ingot axis is smaller than  0.1 at.% [34]. The as-cast condition shows a small fraction of residual micropores, which are eliminated by the ensuing HIP step. Thereby, the material is heated up to 1200  C, held for 4 h at 200 MPa and subsequently cooled via furnace cooling (<8 K/min) to RT. The chemical composition of the TNM material is shown in Table 1. As analytical methods X-ray fluorescence spectroscopy (XRF), inductively coupled plasmaoptical emission spectroscopy (ICP-AES) and carrier gas hot extraction analysis were used (for elemental assignment see Table 1). Evidently, the deviation from the nominal TNM alloy composition is with Tie43.67Ale4.08Nbe1.02Moe0.1B rather small. The phase transition temperatures and prevailing phase fractions at different annealing temperatures were primary characterized by means of in situ synchrotron heating experiments using the high energy X-ray diffraction (HEXRD) setup of the HZG beamline HARWI II at DESY in Hamburg, Germany. Detailed information regarding the experimental setup and evaluation of raw data can be found in Refs. [30,35]. Complementary, the phase transition temperatures were verified by means of differential scanning calorimetry (DSC) measurements with a heating rate of 20 K/min from RT to 1450  C using a high-temperature DSC 404F3 from Netzsch, Germany. A sample size of 3 mm in diameter and approximately 1 mm in height, which corresponds to a sample mass of about 40  3 mg, was used. The crucibles consisted of Al2O3. An Y2O3 suspension was used to improve the thermal contact between the samples and the crucible. The measurements were conducted under Ar-atmosphere with a gas flow of 50 ml/min. The temperature measurement of the DSC setup was calibrated by means of enthalpy standards. With both methods, HEXRD and DSC experiments, a temperature accuracy of about 5  C can be obtained. Furthermore, thermodynamic equilibrium calculations based on the CALPHAD

Table 1 Chemical composition of the investigated TNM alloy. The concentrations of Ti, Al, Nb, and Mo were obtained by means of X-ray fluorescence spectroscopy (XRF), whereas the amounts of B, Si, Fe, Cu, Cr, Ni, and C were determined by means of inductively coupled plasma-optical emission spectroscopy (ICP-OES). Carrier gas hot extraction analysis was conducted to quantify the impurity elements O, N and H.

m.% at.%

Ti

Al

Nb

Mo

B

Si

Fe

Cu

Cr

Ni

C

O

H

N

bal. bal.

28.75 43.67

9.24 4.08

2.38 1.02

0.025 0.095

0.009 0.013

0.049 0.036

0.013 0.008

0.008 0.006

0.011 0.008

0.004 0.014

0.042 0.108

0.001 0.041

0.001 0.003

130

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

method were performed with the software package Thermocalc using a commercial available database for TiAl [36] in order to obtain additional information on phase transition temperatures and phase fractions [30]. The material for mechanical testing was heat-treated in an industrial furnace under atmospheric conditions. Cylindrical blanks with a diameter of 10 mm and a length of 80 mm were spark eroded from the cast/HIP ingot. The blanks were then heat-treated at temperatures within the range of 850  C to 1425  C. For further investigation in the SEM, the cut surfaces of selected heat-treated blanks were ground and additionally polished by means of vibration polishing in order to prevent surface cracks, especially in the gphase, as well as to reduce residual stresses for subsequent X-ray diffractometry (XRD) and macro-hardness testing. The SEM investigations were performed with an FE-REM MIRA from Tescan, Czech Republic, equipped with a field emission gun. The images were recorded in back-scattered electron (BSE) mode. For a detailed description of metallographic preparation and quantitative analysis of TNM alloys the reader is referred to [30,37]. Hardness measurements according to Vickers were conducted with a Vickers hardness testing machine from Frank, Germany. Within this study, the HV10 hardness values are given as arithmetic mean values from at least 3 single hardness measurements. The tolerance range of the mean hardness values correlates to 5 HV10. Static tensile tests were performed with a universal tensile testing machine Inspect 100 kN from Hegewald & Peschke, Germany, in the temperature range from RT to 800  C with a strain rate of 8$105 s1. The tensile test specimens had a diameter of 4 mm and a length of 20 mm within the gauge volume. The surface of the specimens was polished electrolytically in order to rule out surface effects induced during manufacturing. The determined tensile properties are specified as the arithmetic mean values of at least two valid tensile tests per heat treatment and testing condition in order to achieve minimum statistics. Short-term tensile creep tests under constant load were conducted on final heat-treated samples using Mark II TC 30 and TC 50 creep testing machines from Denison Mayes Group, Great Britain. The creep tests were performed at a constant temperature of 750  2  C applying an initial stress of 150  2 MPa. The creep specimens had a gauge length of 30 mm and a diameter of 6 mm. The creep deformation was measured with two simultaneously mounted extensometer bars with inductive displacement

transducers. The final creep curve was derived as the arithmetic mean value from two individual short-term creep measurements.

3. Results and discussion 3.1. TNM system TNM alloys are novel multi-phase g-TiAl based alloys which solidify via the b-phase as shown in the phase diagram TieAle 4Nbe1Moe0.1B displayed in Fig. 1a [28,38]. The stabilization of the b-phase is achieved by addition of Nb and Mo, whereby Mo has an approximately four times stronger b-stabilizing effect than Nb [39]. During solidification TNM alloys undergo a transformation pathway simplified expressed as L / L þ b / b / b þ a / a þ b þ g / a þ b þ bo þ g / a þ bo þ g / a þ a2 þ bo þ g / a2 þ bo þ g (see Fig. 1a). In this transformation pathway, the precipitation sequence of Ti-borides during solidification is neglected. In the as-cast TNM microstructure Ti-borides are situated at the grain boundaries of the former body-centered cubic (bcc) b-Ti grains [40]. During solidification the Ti-borides act as efficient grain refiners by restricting the formation length of hexagonal closedpacked (hcp) a-Ti grains and enhancing heterogeneous nucleation of the a-phase during b to a transformation regarding the Burgers orientation relationship {110}bjj(0001)a and <111>bjj < 1120 > a [40e42]. The experimentally verified quasi-binary section through the TNM alloying system as shown in Fig. 1a predicts no occurrence of a single a-phase field region for a chemical composition of Tie 43.67Ale4Nbe1Moe0.1B. The phase fraction diagram depicted in Fig. 1b for the actual chemical composition of Tie43.67Ale4.08Nbe 1.02Moe0.1B confirms this prediction. The slightly higher contents of the b-stabilizers yield to a higher b-stabilization. Thereby, the occurrence of a single a-phase field region is not desired due to the danger of excessive grain growth during heat treatments [37]. Further cooling leads to a diffusion-controlled precipitation of gTiAl (L10-structure) from a-phase according the Blackburn orientation relationship (0001)ajj{111}g and < 1120 > ajj<101]g, forming a/g-colonies [3,43]. Also the formation of globular or lensshaped g-precipitates within the b-phase can be observed, whereby the morphological feature depends on the prevailing cooling rate. At RT a and b are present as ordered phases a2-Ti3Al and bo-TiAl with D019- and B2-structure, respectively [30,44]. After

Fig. 1. a) Experimental quasi-binary section through the TNM alloying system according to [28,38]. The Al-content of the investigated alloy is indicated as a vertical line. b) Course of phase fractions with temperature for the investigated alloy composition Tie43.67Ale4.08Nbe1.02Moe0.1B.

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

131

Table 2 Phase transition temperatures of the investigated TNM alloy with a composition of Tie43.67Ale4.08Nbe1.02Moe0.1B. The transition temperatures (bold) were experimentally determined by means of in situ HEXRD and DSC measurements with applied heating rates (HR) of 2 K/min und 20 K/min, respectively. Phase transition temperatures [ C]

Experimental Phase transition temperatures from HEXRD and DSC

HEXRD (HR 2 K$min1)

DSC (HR 20 K$min1)

Teut Ta2/a Tbo/b Tgsolv Tb

1160 1160e1175 1175e1205 1255 1405

1160 11601175 11751205 1255 Out of range

1165 1165e1190 1190e1220 1260 1415

cooling to RT no indication of the formation of u-phase or their derivatives was found. The phase fraction evolution within the temperature regime from 1100  C to 1350  C was determined by in situ HEXRD heating experiments on cast/HIP material with a heating rate of 2 K/min close to thermodynamic equilibrium (Fig. 1b). Thereby, the finegrained and nearly texture-free microstructure after b-solidification [13,25,40] can be easily investigated by means of X-ray diffraction methods, i.e. HEXRD as well as conventional lab-scale XRD, due to sufficient grain statistics and acceptable chemical homogeneity [35]. However, low diffusivity at low temperatures, i.e. retarded kinetics, and excessive grain growth at high temperatures close to the single b-phase field region are limiting the reasonable use of these methods. The phase transition temperatures from in situ HEXRD measurements were confirmed by DSC experiments, which were conducted up to 1450  C (see Table 2). Thereby, the applied DSC heating rate of 20 K/min yields to 5e10  C higher phase transition temperatures when compared to extrapolated equilibrium values for 0 K/min, as obtained in pre-studies [45,46]. The course of the phase fractions above 1350  C was estimated by means of thermodynamic calculations employing the software Thermocalc in order to gain a qualitative trend of phase evolution during the solidification sequence (Fig. 1b). At temperatures below 1100  C, the prevailing phase fractions were determined by isothermal heat treatments at 850  C and 900  C for 6 h and subsequent lab-scale XRD measurements combined with quantitative phase analysis by means of Rietveld refinement [47]. Here, the starting microstructure has a significant effect on the prevailing phase fractions after the heat treatments because of the low diffusivity within the low-temperature regime. 3.2. Initial cast/HIP microstructure The cast/HIP condition is the starting microstructure for the ensuing heat treatments. The microstructure after HIPing is shown

in Fig. 2a and it is termed a so-called “nearly lamellar g” (NL g) microstructure. The white arrow shows a coarse-grained a2/g-colony embedded in an otherwise fine-grained matrix. Typical dimensions of such coarse elongated a2/g-colonies are about 150 mm in length and 30 mm in width. The microstructural features are better resolved in Fig. 2b. The thickness of the laths within the a2/gcolonies is within the mm-range and thus rather large. During HIPing a decomposition reaction within the a2/g-colonies according to a / b þ g takes place, which leads to an increase in thickness of the g-lamellae as well as to the formation of secondary precipitates of the bo-phase (bo,sec) within the a2/g-colonies. The decomposition reaction results from a strong phase imbalance of the as-cast starting condition. In the cast/HIP microstructure the a2/gcolonies are surrounded by ordered bo-phase and globular g-grains (Fig. 2b). Within the bo-phase also u-domains can be found. Whereas the u-domains are easily detected by XRD, they cannot be resolved in the SEM due to their small size (<10 nm). 3.3. Heat treatment study Fig. 3 illustrates the applied heat treatments. The temperatures of the heat treatments are indicated in absolute values as well as relative to the respective phase transition temperatures, e.g. eutectoid temperature Teut, g-solvus temperature Tgsolv or b-transus temperature Tb. Each single heat treatment step is indicated by temperature ( C), holding time (hours or minutes) and cooling or heating condition (air cooling AC, furnace cooling FC or furnace heating FH). The corresponding cooling or heating rates are given in the insert in Fig. 3. The solution heat treatments were varied from a temperature below Tgsolv up to a temperature within the single bphase field region. The aim of this heat treatment step is to supersaturate the a-phase upon cooling, which is required to obtain a fine lamellar spacing within the a2/g-colonies by the subsequent annealing near service temperature, the so-called stabilization or precipitation treatment. Therefore, the blanks were annealed at

Fig. 2. Reference microstructure determined by SEM in BSE mode: a) As-cast and hot-isostatically pressed (cast/HIP) starting condition. The white arrow indicates a coarse a2/gcolony. Coarse a2/g-colonies have a typical dimension of 150 mm in length and 30 mm in width. b) Microstructural details of the cast/HIP microstructure.

132

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

Fig. 3. Schematic overview of the applied heat treatment study starting from the cast/HIP microstructure. The first heat treatment step was varied from temperatures below the gsolvus temperature (Tgsolv) up to temperatures within the single b-phase field region. Subsequently, for HT #1 to HT #4 a stabilization treatment at 850  C for 6 h followed by furnace cooling was conducted. In HT #5 the stabilization treatment was performed at 900  C. In HT #6 an intermediate heat treatment step was carried out at 1255  C in order to adjust an almost fully lamellar microstructure. The cyclic heat treatment after the b-homogenization in HT #7 was primarily conducted to transform residual g-phase to globular ggrains.

850  C or 900  C for 6 h followed by furnace cooling. Thereby, fine precipitated g-lamellae within the a2/g-colonies lead to a cellular reaction (CR) starting from the a2/g-colony boundaries according to (a2 þ g)lamellar / (a2 þ bo þ g)cellular. The thickness of the g-laths within the a2/g-colonies is dominated by the temperature of the stabilization treatment, whereas the annealing time plays a minor role, and increases with increasing temperature [14]. Additionally, small lens-shaped g-grains (glens) can be found within the bophase. The slow furnace cooling after the stabilization treatment leads to the formation of u-phase below w825  C [31,48] and a reduction of internal stresses [14,28]. The solution heat treatment in HT #1 at a temperature of 35  C below Tgsolv yields to a reduction of b-phase and an adjustment of a certain fraction of g-phase (see evolution of phase fractions in Fig. 1b). The microstructure after HT #1 is shown in Fig. 4a. The corresponding microstructural features together with HV10 are listed in Table 3. A description of the determination of microstructural constituents is given in the appendix. The microstructure consists of 72 vol.% a2/g-colonies, 11 vol.% bo-phase situated on colony boundaries and 17 vol.% globular g-grains, which are located within the former bo-phase. The a2/g-colonies have a typical dimension of 150 mm length and 30 mm width, which is in the same range as obtained for the cast/HIP condition (see Fig. 2a,b). During HT #1 no significant growth of the a-grains during the first heat treatment step has occurred. The grain size aspect ratio of the formed a2/g-colonies, which corresponds to the ratio of length to width of the elongated a2/g-colonies, remains constant at a value of 5 when compared to the cast/HIP microstructure. Due to the reduction of b/bo-phase during HT #1 the seam of bo-phase at the colony boundaries is in the range of 5e15 mm and, therefore, smaller than the cast/HIP starting condition. The reduction of bophase also leads to a disappearance of the secondary bo precipitates within the a2/g-colonies when compared to the cast/HIP condition. After HT #1 the coarse g-lamellae within the a2/g-colonies could not be completely removed, but the fraction of coarse g-lamellae was reduced from 59 vol.% to 34 vol.%. For both comparative conditions the thickness of the coarse g-lamellae is in the range of 1e3 mm. Due to a slight supersaturation of the a2-phase after the first solution heat treatment step in HT #1 a volume fraction of about 15 vol.% fine lamellar g-phase is present. In comparison, the

content of fine lamellar g-phase within the cast/HIP material is at zero. A holding time of 3 h at 1220  C in HT #1 also leads to a small increase of the content of globular g-grains as well as to an increase in their median diameter from 6.5 to 7.5 mm when compared to the cast/HIP starting condition. However, no spherodization of the coarse g-lamellae has taken place, which indicates only a small difference of the interfacial energies between the a- and g-phase [3]. Therefore, the microstructural features after HT #1 are similar to the cast/HIP condition, with the only exception of a higher fraction of fine g-lamellae within the a2/g-colonies. The similarity of the microstructures also leads to comparable Vickers hardness values for the cast/HIP condition and HT #1 (w355 HV10, see Table 3). The solution annealing in HT #2 at Tgsolv, which corresponds to the minimum fraction of the b-phase (see Fig. 1b), further decreases the amount of b- and g-phase. After HT #2, the volume fraction of the a2/g-colonies increased to 94% (see Fig. 4b and compare to Fig. 2a). Due to growth of the a-grains the size of the a2/g-colonies increased up to 200 mm in length and 50 mm in width, decreasing the grain size aspect ratio to 4. Additionally, the microstructure consists of 2 vol.% bo-phase situated at the colony boundaries and 2 vol.% of residual globular g-grains. The seam thickness of the bophase is further reduced to 1e10 mm due to the small volume fraction of b-phase present at this annealing temperature (see Fig. 4b and compare to Fig. 2b). The median diameter of the globular g-grains is decreased to 5.2 mm, which is also an effect of the reduction of the g-phase. After HT #2 an amount of 22 vol.% of coarse g-laths within the a2/g-colonies is remaining. The thickness of the g-laths is comparable to those after cast/HIP and HT #1. The higher supersaturation of the a2-phase after solution annealing and AC has led to an increase of the fraction of fine g-laths within the a2/g-colonies to 50 vol.%. As a consequence the Vickers hardness increased to 383 HV10. During the solution annealing of HT #3 at a temperature 45  C above Tgsolv the g-phase dissolves completely and only disordered a- and b-phase are existing during annealing. At this temperature, the b-phase primarily controls the grain growth of the a-phase. At RT the fraction of the ordered bo-phase is higher than after HT #2. The portion of b-phase at annealing temperature, which, for example, can be measured by means of in situ HEXRD

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

measurements, does not exactly relate to the fraction of bo-phase present at RT after AC which can be attributed to the fast retransformation kinetics of b / a. This type of microstructure termed “nearly lamellar b” (NL b) microstructure, which means that no globular g-grains are present. HT #3 leads to an adjustment of 94 vol.% a2/g-colonies (see Fig. 4c), whereby the size of the a2/gcolonies is smaller than in HT #2 due to a better stabilization of grain growth by means of a high fraction of b-phase at the annealing temperature of 1300  C (see Table 3). The a2/g-colonies are 150 mm long and 40 mm wide, yielding to a grain size aspect ratio of 4. Approximately 4 vol.% bo-phase is situated at the a2/gcolony boundaries and 2 vol.% are assigned to cellular reaction and secondary precipitates of the bo-phase within the a2/g-colonies. It is assumed that the entire g-phase is precipitated as fine g-lamellae within the a2/g-colonies, neglecting the amount of lens-shaped ggrains within the bo-phase. The higher b-phase fraction present

133

during the first heat treatment step leads to an enrichment of bophase at the a2/g-colony boundaries with a thickness ranging from 5 to 30 mm. The hardness of this type of microstructure is increased up to 427 HV10. The solution annealing within the single b-phase field region in HT #4 yields upon AC to an almost complete transformation of b / a, which can be used for a refinement of coarse-grained a2/g-colonies [49]. Therefore, holding time, sample geometry and cooling conditions are the critical parameters for the adjustment of a homogeneously refined microstructure. Additionally, the heat treatment within the single b-phase region leads to a chemical homogenization due to fast diffusion processes, whereby segregations from the casting process can be compensated. A further advantage is the large temperature window of heat treatments within the single b-phase field region and the small variation of Tb with chemical composition within the TNM composition range [39,50]. The refined and texture-

Fig. 4. Microstructures after heat treatments according to Fig. 3 aee) Simple two-step heat treatments and feg) multi-step heat treatments. The corresponding heat treatment number is shown in the upper right corner. h) Forged and heat-treated material (forged þ HT) [14]. The inserts show microstructural details in higher magnification.

w50 w25

w25

0 15 50 74 73 76 64 4 55 1e3 1e3 1e3 e e e e 1e3 1e3 59 34 22 e e e e 19 1 i i i e e e e h h 6.5 7.5 5.2 e e e e 10.2 5.2 15 17 2 e e e e 34 10 3 0 2 2 1 1 2 0 3 5e20 5e15 1e10 5e30 1e5 1e5 1e5 5e20 5e20

g g

CR ¼ cellular reaction, SP ¼ secondary bo-precipitates, i ¼ inhomogeneous, h ¼ homogeneous.

11 11 2 4 2 2 1 11 2 5 5 4 4 7 7 5 7 1 30 30 50 40 15 15 25 15 25 150 150 200 150 100 100 125 100 25 71 72 94 94 97 97 97 55 85 74 66 74 74 73 76 64 57 66 14 11 4 6 3 3 3 11 5 12 23 22 20 24 21 33 32 29

g g g b b b

NL NL NL NL NL NL FL NL NL Cast/HIP HT#1 HT#2 HT#3 HT#4 HT#5 HT#6 HT#7 Forged & HT

Fraction Fraction Median Spatial Fraction Thickness Fraction Thickness [vol.%] [vol.%] diameter distribution [vol.%] [vol.%] [nm] [mm] [mm] Thickness [mm] Fraction Length Width Grain size Fraction [mm] aspect [vol.%] [mm] [vol.%] ratio [%]

g bo a2

CR þ SP Globular Seam at colony boundaries inclusive u-phase

bo-phase a2/g-colonies

Heat treatment

Type of Phase fractions via XRD microstructure measurements [vol.%]

Morphological characteristics of different microstructures

g-phase

Coarse lamellar

Fine lamellar

351 362 383 427 451 446 427 355 e

       

5 5 5 5 5 5 5 5

Vickers hardness HV10

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140 Table 3 Microstructural parameters of the cast/HIP starting material, the cast and heat-treated variants as well as the forged and heat-treated reference condition [14]. The table indicates the type of microstructure and compares the volume fractions of the prevailing phases as well as the morphological characteristics for each microstructural condition. The thickness of the fine g-lamellae within the a2/g-colonies was derived from TEM investigations conducted in Refs. [29,30]. The Vickers hardness can be interpreted as an overall property of each microstructural condition. Details on the determination of microstructural constituents can be found in the appendix.

134

free microstructure after HT #4 is shown in Fig. 4d. The final microstructure after the stabilization treatment consists of 97 vol.% a2/gcolonies and 2 vol.% bo-phase at the colony boundaries. The residual content of 1 vol.% belongs to secondary precipitates of the bo-phase within the a2/g-colonies. No cellular reaction is observed (see insert in Fig. 4d). By means of HT #4 the coarse-grained a2/g-colonies of the cast/HIP condition are refined to a length of 100 mm and a width of 15 mm, increasing the grain size aspect ratio to 7. According to HT #3 all of the g-phase is present as fine g-lamellae within the a2/g-colonies, again neglecting the small fraction of g-phase within the bophase. Compared to HT #3, the Vickers hardness could be increased to 451 HV10 by means of HT #4 which is a consequence of the high content of fine a2/g-colonies exhibiting a narrow lamellar spacing. In HT #5 the stabilization treatment was conducted at 900  C. The microstructure after HT #5 shows the same microstructural features as that of HT #4 (see Fig. 4e and d). The difference is a higher average thickness of the g-laths of about 50 nm within the a2/g-colonies and the onset of cellular reaction (see insert in Fig. 4e), both caused by the higher annealing temperature of 900  C during the stabilization treatment. The g-lath thickness were derived from TEM investigations as reported in Refs. [29,30]. The hardness value of 446 HV10 is in the range of the hardness obtained for HT #4. In HT #6 an intermediate heat treatment step compared to HT #4 was conducted at 1255  C, which is in the range of the b-phase minimum (Fig. 1b), in order to reduce the content of the b-phase and to adjust an almost fully lamellar (FL) microstructure with a very low residual bo-phase fraction (Fig. 4f). As a less advantageous effect the heat treatment near the b-phase minimum leads to grain growth of the afore refined microstructure. After HT #6 the a2/gcolonies have a typical length of 125 mm and a width of 25 mm, leading to a grain size aspect ratio of 5. The microstructure consists of 97 vol.% a2/g-colonies and only a small fraction of 1 vol.% of bophase and about 2 vol.% of cellular reaction and secondary bo-precipitates. Despite the high volume fraction of a2/g-colonies, only a volume fraction of 64% g-phase was determined which exists within the colonies. In comparison to the microstructure after HT #2, where the solution and stabilization treatments are nearly identical to that of HT #6, the g-phase content is about 74 vol.%. Therefore, it can be assumed that the stabilization treatment conducted at 850  C did not attain thermodynamic equilibrium after annealing for 6 h. The hardness of the microstructure after HT #6 amounts 427 HV10. In HT #7 a cyclic heat treatment was conducted after b-homogenization in order to obtain globular g-grains within the finegrained microstructure. The ensuing two-step heat treatment has the same purpose as in case of HT #1, whereby HT #7 should lead to even more fine-grained a2/g-colonies showing a significantly higher content of homogeneously distributed globular g-grains. After this heat treatment the microstructure consists of 55 vol.% a2/g-colonies, 11 vol.% bo-phase and 34 vol.% globular g-grains (Fig. 4g). The refined a2/g-colonies have a length of 100 mm and a width of 15 mm, corresponding to those obtained for HT #4 and HT #5 (see Table 3). The relatively high content of 11 vol.% bo-phase corresponds to the bophase fraction after HT #1. The thickness of the bo-phase seam around the a2/g-colonies is in the range of 5e20 mm. The high content of 34 vol.% globular g-phase leads to a larger median diameter of 10.2 mm of the g-grains but also to a significant fraction of undesired coarse g-laths within the a2/g-colonies. Only 4 vol.% gphase can be assumed to be precipitated as fine g-lamellae within the a2/g-colonies. As a consequence of the high volume fraction of globular g-phase the Vickers hardness decreases to 355 HV10, which is in the range obtained for the cast/HIP condition and material subjected to HT #1. The forged and heat-treated (forged þ HT) reference material [14] as depicted in Fig. 4h represents the desired microstructure to

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

be achieved by heat treatments conducted on cast/HIP material. It consists of 85 vol.% a2/g-colonies, 3 vol.% bo-phase and 10 vol.% globular g-grains. The remaining amount of 2 vol.% corresponds to cellular reaction and secondary bo-precipitates. The a2/g-colonies have a diameter in the range of 25 mm and exhibit a globular shape, which means that the grain size aspect ratio is about 1. In comparison the cast and heat-treated microstructures are significantly larger and exhibit grain size aspect ratios in the range of 4e7, which means that without a forging step the a2/g-colonies are always elongated (see Table 3). The forged and heat-treated microstructure was also subjected to a stabilization treatment at 850  C. The globular g-grains exhibit a median diameter of 5.2 mm and are homogeneously distributed within the microstructure. The fraction of coarse g-lamellae is with 1 vol.% low, thus 50 vol.% of fine g-laths can be found within the a2/g-colonies. 3.4. Tensile tests as function of temperature Fig. 5 shows the temperature-dependent mechanical properties of the cast/HIP and heat-treated samples gained from tensile tests conducted at RT, 300  C, 700  C and 800  C. The Rp0.2 tensile yield strength as a function of temperature is shown in Fig. 5a. In order to achieve a high yield strength the microstructure must consist of a large volume fraction of homogeneously distributed fine a2/g-colonies with a lamellar spacing in the range of 25e50 nm, see HT #4 and HT #5 in Fig. 5a. There, the microstructures after b-homogenization and ensuing stabilization treatment (HT #4 and HT #5) exhibit with about 950 MPa the highest Rp0.2 values of all investigated cast and heat-treated microstructural conditions at RT. The yield strength decreases with increasing size of the a2/g-colonies (compare HT #4 with HT #3 and HT #6). Depending on their size, location and orientation within the specimen volume they can act as

135

critical defects [51e53], which lead to enhanced brittleness and early fracture during tensile testing. In Fig. 5a also invalid Rp0.2 yield strength data are indicated for HT #3 and HT #6 (marked with a cross). Thereby, at RT and 300  C the plastic deformation of the specimen until fracture was smaller than 0.2%, thus no valid Rp0.2 value can be determined. That means that during quasi-static loading the fracture strength RB was reached before Rp0.2. Therefore, it is obvious that the microstructures with coarse-grained a2/gcolonies are considered to be less tolerant against damage due to their limited plastic deformation behavior at temperatures below the brittle-to-ductile transition temperature (BDTT). For detailed information concerning the plasticity of g-TiAl based alloys the reader is referred to [17,54], where the role of dislocations and mechanical twinning on the deformation behavior is comprehensively summarized. The Rp0.2 yield strength of the NL b-type microstructures after HT #3, HT #4 and HT #5, as well as after HT #6 are in the range of the forged and heat-treated material (see solid line in Fig. 5a). At ambient temperature the Rp0.2 yield strength of the NL g microstructure after HT #2 is lower than that of the forged and heat-treated material, whereas the Rp0.2 values are comparable at elevated temperatures. The Rp0.2 values of the NL g microstructures of the cast/HIP starting condition as well as after HT #1 and HT #7 are similar, but significantly lower when compared to the forged and heat-treated condition. Within the NL g-type microstructures the g-phase is the softest phase due to its comparably easy to activate glide and twinning systems, e.g. see Refs. [17,54], therefore plasticity is enhanced at the presence of spatially expanded g-phase regions, i.e. coarse g-lamellae within the a2/g-colonies and/or globular g-grains. As a consequence, a higher fraction of expanded g-phase features leads to a significant decrease of the Rp0.2 yield strength (compare HT #1 and HT #7 with HT #2 in Fig. 5a). Overall, the Rp0.2 yield strength values decrease for the b-homogenized

Fig. 5. Mechanical properties gained from uniaxial tensile tests as a function of temperature ranging from RT to 800  C. The assignment of the various heat treatments is shown in the inset of a). The solid lines mark the yield strength and total fracture elongation reported for forged and heat-treated TNM material [14]. a) Rp0.2 tensile yield strength; x: invalid test, i.e. plastic fracture elongation smaller than 0.2%, b) Rm ultimate tensile strength, c) At total fracture elongation (see also Table 4), and d) E elastic modulus.

136

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

Table 4 Total fracture elongation At of the cast and heat-treated microstructures according to Fig. 5c. At [%]

Cast/HIP

HT#1

HT#2

HT#3

HT#4

HT#5

HT#6

HT#7

25  C 300  C 700  C 800  C

0.61 1.46 1.73 21.23

0.85 1.21 1.88 14.39

0.68 0.78 1.06 1.89

0.54 0.70 0.96 1.24

0.81 0.95 1.06 1.85

0.83 0.96 1.28 1.75

0.80 0.55 1.11 1.88

1.16 1.23 1.99 6.78

condition after HT #4 from 950 MPa at RT to 570 MPa at 800  C and from 680 MPa at RT to 450 MPa at 800  C for the condition after HT #7 showing a globular g-phase fraction of 34 vol.%. The microstructures after HT #4 and HT #7 exhibit a difference in Rp0.2 of w300 MPa at RT and w150 MPa at 800  C. The course of the ultimate tensile strength Rm within the investigated temperature regime from 25  C to 800  C is shown in Fig. 5b. At RT the Rm values are only slightly higher than the corresponding Rp0.2 values of each specific microstructural condition (compare Fig. 5a with Fig. 5b). At 300  C the Rm/Rp0.2 ratio increases which is primarily caused by enhanced dislocation mobility with increasing temperature [54e56]. After HT #2, HT #3 and HT #6, where grain growth during the heat treatment leads to coarsegrained a2/g-colonies, the Rm values are significantly reduced due to early fracture caused by these critical defects. Therefore, the microstructures after HT #3 and HT #6 are not suitable for applications, where ductility is playing a dominant design role. The microstructural homogenization after HT #4, HT #5 and HT #7, where no grain growth occurred subsequent to the b-homogenization treatment, shows no early fracture and only a small scatter of the tensile test data was observed due to reduced defect size and the prevailing homogeneous microstructure. As a common characteristic of g-TiAl based alloys, a brittle-toductile transition behavior is also obtained for TNM alloys. As shown in Fig. 5c (see also Table 4) this is true for the cast/HIP condition, HT #1 and HT #7 where the fracture elongation at 800  C is considerably increased when compared to the lower temperatures. For the remaining microstructural variants the maximum test temperature of 800  C was obviously below BDTT. At RT the microstructures after HT #1 and HT #7, which both exhibit spatial expanded regions of g-phase, show a pronounced fracture elongation. The cast/HIP microstructure has a reduced plastic deformability, presumably due to the presence of coarse-grained a2/gcolonies. In contrast to the deformation behavior of the NL b-type microstructures, where plastic deformability is rather small and expected to be localized within the g-laths of the a2/g-colonies, the

globular g-grains of the NL g-type microstructures have a stronger effect on increasing the plastic deformability (compare the fracture elongation obtained for HT #1 and HT #7 at RT). Above 700  C the total fracture elongation increases significantly, especially for the NL g-type microstructures which exhibit a higher content of spatially expanded g-phase regions (see cast/HIP, HT #1 and HT #7 in Table 4). The increase in fracture elongation of the NL b-type microstructures (HT #3 to HT #6) is rather small compared to the NL g microstructures. The observed phase transformations, such as bo / u, will also influence the overall deformation behavior as reported in Ref. [31]. In comparison, the forged and heat-treated material exhibits a pronounced plastic deformation behavior within the whole investigated temperature range due to the presence of fine equiaxed a2/g-colonies and a volume fraction of 10% of ductile globular g-grains. The temperature dependence of the elastic moduli E of the different heat-treated conditions is shown in Fig. 5d. Within this study no certain correlation of E to the specific heat treatment was found. Therefore, it can be assumed that the mean value of the elastic modulus exhibits a maximum inaccuracy of about 5 GPa. E decreases significantly with increasing temperature. At RT E has a mean value of 158 GPa and decreases to 107 GPa at 800  C. 3.5. Correlation of temperature-dependent Rp0.2 yield strength with RT-hardness The average temperature-dependent Rp0.2 yield strength values of the cast/HIP and heat-treated material conditions from Fig. 5a were sorted by their corresponding values of Vickers hardness HV10 at RT (see Table 3), which is a characteristic measure of each specific microstructural condition. The resulting 3D plot is shown in Fig. 6a. Thereby, the set of Rp0.2 values was fitted by means of an adapted relation based on Kocks model [57]. The fitting procedure was applied by means of the program Origin 8.5.1 from OriginLab Corporation, USA, after the LevenbergeMarquardt algorithm. In the 2 present case, the adjusted coefficient of determination R is 0.9763

Fig. 6. a) Rp0.2 yield strength profiles as function of temperature for the cast/HIP and heat-treated microstructures sorted by their corresponding Vickers hardness at RT. b) Contour plot of the fitted surface in a).

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

Fig. 7. Short-term creep properties of the heat-treated TNM microstructures compared to the cast/HIP starting material. The creep tests were performed at 750  C and 150 MPa. The highest (cast/HIP starting material) and lowest (HT #6) creep rate are indicated.

(see insert in Fig. 6a), which corresponds to good fitting quality. Thereby, the plot shows a maximum deviation DRp0.2 of approximately  40 MPa in small localized regions, whereby the rest of the fit matches accurately with the measured values. The contour plot of Fig. 6a is shown in Fig. 6b and corresponds to a 2D projection of the fitted surface. Employing the results to forged TNM material with NL b microstructure as reported in Ref. [7], which exhibits a hardness of 410  5 HV10, Rp0.2 values of 835 MPa at RT and 550 MPa at 800  C would be expected according to Fig. 6b. In tensile tests, Rp0.2 values of 848 MPa at RT and 545 MPa at 800  C were measured, which exemplarily underline the good agreement of the applied surface fitting to RT Vickers hardness. 3.6. Short-term creep tests Tensile creep tests were conducted in order to evaluate the creep resistance of the different microstructures shown in Fig. 4. The obtained creep curves are depicted in Fig. 7. All NL g-type microstructures with a significant volume fraction of globular ggrains, such as the microstructures of the forged and heat-treated condition [58] as well as after HT #7, show creep properties in the range of the cast/HIP starting condition. Obviously, the small size of the fine-grained and equiaxed a2/g-colonies of the forged and heat-treated material compensate the effect of the low content of 10 vol.% globular g-grains when compared to the microstructure adjusted with HT #7. After HT #7 the a2/g-colonies are coarser and show an elongated feature, whereas the amount of globular ggrains is with 34 vol.% significantly higher. Geometrical limitations, i.e. a fine lamellar spacing within the a2/g-colonies, lead to a significant improvement of creep resistance due to restricted dislocation mobility within the small g-laths in the a2/g-colonies as

137

reported in Refs. [17,59,60]. The NL g-type microstructure after HT #1, where the geometrically expanded g-phase regions are only prevailing in the form of coarse g-laths within coarse-grained a2/gcolonies, shows a significant lower creep deformation after 200 h. In contrast to the NL g-type microstructures, the NL b-type microstructures exhibit a better creep performance, whereby the primary creep region is prolonged and shifted to smaller strains. The microstructures with fine a2/g-colonies after HT #4 and HT #5 exhibit a lower creep resistance compared to the microstructures with coarse-grained a2/g-colonies after HT #2, HT #3 and HT #6. The higher annealing temperature of 900  C during the stabilization treatment in HT #5 leads to an increased average lamellar spacing within the a2/g-colonies and, therefore, to an increase in creep deformation when compared to the microstructure after HT #4. The heat treatment variants leading to coarse-grained a2/gcolonies with fine lamellar spacing exhibit the highest creep resistance of all investigated microstructures. Here, residual expanded g-phase regions, i.e. a small content of coarse g-laths within the a2/g-colonies as well as residual globular g-grains after HT #2, and/or expanded regions of bo-phase at the a2/g-colony boundaries after HT #3 lead to slightly higher creep deformation after 200 h when compared to the near fully lamellar microstructure after HT #6. The appearance of bo-phase reduces the creep resistance due to enhanced diffusivity within the ordered bcc lattice as well as through deformation accommodation by formation of u-phase [9,61]. The most creep resistant microstructure exists after HT #6. It consists of 97 vol.% coarse-grained a2/g-colonies with a lamellar spacing in the range of 25 nm. Obviously, fully lamellar microstructures exhibit the best creep performance of all microstructure variants within the TNM alloy system. The influence of other microstructural features e such as u-phase and lensshaped g-grains within the bo-phase as well as secondary bo-precipitates within the a2/g-colonies e are assumed to be constant for all prevailing microstructural conditions. After 200 h all investigated microstructures exhibit a significant smaller creep strain than 1%, whereby the creep deformation compared to the cast/HIP condition can be improved by a factor of 2e3. The creep rates after 200 h vary from 7∙109 s1 for the cast/HIP condition to 2∙109 s1 for the microstructure after HT #6. 3.7. Assessment of the microstructural constituents In the following, the influence of the major microstructural constituents on the mechanical properties of TNM alloys are discussed in order to assess the potential of the different microstructures adjusted within the current heat treatment study. A summary of the evaluation of the investigated microstructural constituents can be found in Table 5. During heat treatments around Tgsolv the microstructural homogeneity is strongly influenced by the presence of phases (b and g) which retard grain growth. The yield strength at RT strongly increases with decreasing size and increasing portion of fine a2/g-colonies within the microstructure. In this context, compare the b-homogenized

Table 5 Assessment of microstructural constituents influencing the mechanical properties of TNM alloys, which were heat-treated subsequent to the cast/HIP process. Note that the comparison is only valid for a random orientation of the a2/g-colonies. Microstructural parameters

Suppression of grain growth

Yield strength at RT

Ductility at RT

Creep resistance

Small size of a2/g-colonies Fine a2/g- lamellar spacing Large fraction of a2/g-colonies bo(b)-phase at colony boundaries Globular g-grains

‒ (x) ‒ DD DD

DD D DD D ‒‒

(x) (x) ‒ (x) DD

‒ D DD ‒ ‒‒

DD (‒ ‒ ‒) strong positive (negative), D (L) positive (negative), (x) no influence or correlation found within this study.

138

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

microstructure after HT #4 with the microstructure after HT #7. Expanded bo(u)-phase situated at the boundaries of the a2/g-colonies is the hardest of the three major phases existing within the TNM system as reported in Refs. [30,31] and, therefore, shows a strengthening effect at RT. In contrast, a high volume fraction of globular g-grains leads to a significant decrease of the yield strength at RT. In contrast to bo, the g-phase is the softest phase in the TNM alloy [30] and, thus, increases the plastic deformability. Furthermore, the ductility at RT seems not to be dependent from the grain size aspect ratio of the a2/g-colonies, whereas the small size of the a2/g-colonies of the forged and heat-treated condition increases the ductility significantly. Here, the forged and heattreated material has shown the highest fracture elongation at RT [14]. A small width of the g-laths within the a2/g-colonies decreases ductility due to limited dislocation mobility [17,60]. However, the creep resistance of the cast and heat-treated conditions can be increased by a high fraction of a2/g-colonies with a fine lamellar spacing (see HT #6 in Fig. 7), whereby a small a2/g-colony size leads to a decrease in creep strength. Furthermore, high volume contents of globular g-grains lead to a significant reduction of creep resistance due to enhanced dislocation mobility. Therefore, the choice of the final microstructure depends significantly on the design criteria of the respective field of application.

Community’s Seventh Framework Programme (FP7/2007e2013) under grant agreement no. 226716. Many thanks are given to Dr. Arno Bartels for fruitful discussions.

Appendix. Determination of microstructural constituents In the following, the determination of the microstructural features from a combined evaluation of XRD phase fractions and quantitative analysis of SEM micrographs is described. In the first step of the evaluation, the RT phase fractions fa2;XRD , fbo;XRD and fgXRD of the a2-, bo- and g-phases, respectively, have to be determined by means of Rietveld analysis [47]. The Rietveld refinement, however, only provides information about the fractions of the prevailing phases, but usually gives no evidence about their morphological appearance. Thereby, the refined phase fractions have to be converted from m.% to vol.% by using the density of the refined elementary cell in order to allow a comparison with the results derived from the quantitative analysis of the SEM micrographs, which are given in vol.%. In the second step, depending on the type of microstructure the content of globular g-grains fgglob has to be obtained from SEM images by means of quantitative analysis. From that, the total fraction of g-lamellae fglam can be calculated according to

4. Conclusions

fglam The mechanical properties of a cast/HIP TNM alloy with a nominal composition of Tie43.5Ale4Nbe1Moe0.1B can strongly be altered by subsequent heat treatments. Coarse-grained a2/g-colonies, which arise during hot-isostatic pressing, are present within the final microstructure after “conventional” solution heat treatments below the b-transus temperature Tb. Therefore, early fracture can occur during tensile tests conducted at RT. Accordingly, a microstructural homogenization heat treatment is needed to prevent early failure and, therefore, to increase the damage tolerance of the material. A new heat treatment approach by means of a simple short-term annealing within the single b-phase field region combined with a subsequent stabilization treatment near service temperatures demonstrate the major advantages of heat treatments starting from the cast/HIP condition: lower annealing time, better chemical and microstructural homogeneity as well as no statistical significant texture. After b-homogenization and a subsequent stabilization treatment (i.e. HT #4) higher tensile yield strength and higher creep resistance can be adjusted when compared to the forged and heat-treated material (NL g), however, the ductility at RT is reduced. The fracture elongation at RT of cast/HIP material can be optimized by employing a cyclic heat treatment conducted subsequently to the b-homogenization treatment (HT #7). Thereby, the obtained ductility is still in the range of previously published forged and heat-treated TNM material [14], whereby the tensile yield strength is reduced to lower values. The creep properties are comparable to the forged and heat-treated material. Generally, cast and heat-treated TNM microstructures possess the potential to concur with forged and heat-treated TNM material, especially when high strength and creep resistance are required and the demand on RTfracture elongation is moderate.

¼ fgXRD  fgglob :

(A.1)

For microstructures without the presence of globular g-grains the g-phase fraction from the Rietveld refinement represents the total fraction contained within the a2/g-colonies. In the present study it is necessary to separate fglam in a fine-lamellar fraction fglam;fine and a coarse-lamellar fraction fglam;coarse . Thereby, fglam;coarse has to be determined from SEM micrographs, whereby this determination can involve errors due to the available resolution of the SEM. Accordingly, fglam;fine can be calculated by means of

fglam;fine

  ¼ 100  fgglob þ fglam;coarse :

(A.2)

After that, the volume fraction of a2/g-colonies fa2 =g can be derived from the summation of a2-phase fraction and lamellar gphase content

fa2 =g

¼ fa2;XRD þ fglam :

(A.3)

In the next step, the content of bo-phase fbo;SEM has to be determined from SEM micrographs. Thereby, lens-shaped g-grains and u-domains within the bo-phase are added to the bo-phase content, but the resulting inaccuracy is rather small. In comparison to the bo-phase fraction fbo;XRD determined from XRD measurements, the bo-phase content obtained by evaluation of SEM micrographs has to be the same or smaller due to the better resolution of the XRD measurements with regard to the very fine secondary bo-precipitates within the a2/g-colonies:

fbo;XRD  fbo;SEM

(A.4)

The residual fraction fres can be calculated according to

Acknowledgement

  fres ¼ 100  fa2 =g þ fbo;SEM þ fgglob : ¼ fbo;sec þ fCR  0;

The major part of this work was carried out within the framework of the BMBF project O3X3530A, Germany. The support of the DESY management, User Office and HZG beamline staff for performing in situ HEXRD phase evolution experiments at HARWI II, W2 beamline at DESY, Germany, is gratefully acknowledged. The research activities also received funding from the European

fres includes the sum of secondary bo-precipitate content fbo;sec and the fraction of cellular reaction fCR. As a necessary condition the residual content has to be greater than 0 vol.%, whereby fres strongly depends on the accuracy of how fa2 =g, fbo;SEM and fgglob can be determined. In this study, fres can only be considered as a qualitative measure for fbo;sec and fCR.

(A.5)

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

Additionally, the microstructural constituents have to be quantified by means of their characteristic dimensions. Characteristic dimensions of a2/g-colonies are their length and width as well as their grain size aspect ratio. Here, the obtained maximum values of the prevailing dimensions of the a2/g-colonies are stated only. In case of cast and heat-treated TNM alloys, where the a2/g-colonies are elongated, it is not possible to determine the true size distribution of a2/g-colonies from SEM micrographs. The thickness of bophase seam at the a2/g-colony boundaries is another important factor. Globular g-grains can be described by their median diameter, their spatial distribution within the microstructure e either inhomogeneously situated at colony boundaries or homogeneously distributed within the microstructure e and their grain size distribution, which can be measured by means of quantitative methods from the SEM micrographs. The g-lamellae within the a2/g-colonies are characterized by their thickness.

References [1] Tetsui T, Shindo K, Kobayashi S, Takeyama M. A newly developed hot worked TiAl alloy for blades and structural components. Scripta Materialia 2002;47: 399e403. [2] Appel F, Oehring M, Wagner R. Novel design concepts for gamma-base titanium aluminide alloys. Intermetallics 2000;8:1283e312. [3] Appel F, Oehring M. g-Titanium aluminide alloys: alloy design and properties. In: Titanium and titanium alloys. Weinheim: Wiley-VCH; 2003. p. 89e152. [4] Kestler H, Clemens H. Production, processing and applications of g-TiAl based alloys. In: Titanium and titanium alloys: fundamentals and applications. Weinheim: Wiley-VCH; 2003. p. 351e92. [5] Kim YW, Morris D, Yang R, Leyens C. Structural aluminides for elevated temperature applications. Warrendale: TMS; 2008. [6] Clemens H, Smarsly W. Light-weight intermetallic titanium aluminides e status of research and development. Advanced Materials Research 2011;278: 551e6. [7] Gaitzenauer A, Müller M, Clemens H, Voigt P, Hempel R, Mayer S. Eigenschaftsoptimiertes warmumformen einer intermetallischen TitanaluminidLegierung. BHM Berg- Und Hüttenmännische Monatshefte 2012;157:319e 22. [8] Chladil H, Clemens H, Otto A, Güther V, Kremmer S, Bartels A, Gerling R. Charakterisierung einer b-erstarrenden g-TiAl-Basislegierung. BHM 2006;151: 356e61. [9] Clemens H, Wallgram W, Kremmer S, Güther V, Otto A, Bartels A. Design of novel b-solidifying TiAl alloys with adjustable b/B2-phase fraction and excellent hot-workability. Advanced Engineering Materials 2008;10:707e13. [10] Clemens H, Mayer S. Design, processing, microstructure, properties, and applications of advanced intermetallic TiAl alloys. Advanced Engineering Materials 2013;15:191e215. [11] Yoshihara M, Miura K. Effects of Nb addition on oxidation behavior of TiAl. Intermetallics 1995;3:357e63. [12] Lin JP, Zhao LL, Li GY, Zhang LQ, Song XP, Ye F, Chen GL. Effect of Nb on oxidation behavior of high Nb containing TiAl alloys. Intermetallics 2011;19: 131e6. [13] Clemens H, Chladil HF, Wallgram W, Zickler GA, Gerling R, Liss KD, Kremmer S, Güther V, Smarsly W. In and ex situ investigations of the b-phase in a Nb and Mo containing g-TiAl based alloy. Intermetallics 2008;16:827e33. [14] Wallgram W, Schmoelzer T, Cha L, Das G, Güther V, Clemens H. Technology and mechanical properties of advanced g-TiAl based alloys. International Journal of Materials Research 2009;100:1021e30. [15] Bartels A, Koeppe C, Mecking H. Microstructure and properties of Ti-48Al-2Cr after thermomechanical treatment. Materials Science and Engineering: A. 1995;192-193:226e32. [16] Appel F, Kestler H, Clemens H. Forming. In: Westbrook JH, Fleischer RL, editors. Intermetallic compounds e principles and practice. Chichester, UK: John Wiley & Sons, Ltd; 2002. p. 617e42. [17] Appel F, Paul JDH, Oehring M. Gamma titanium aluminide alloys: science and technology. Weinheim: Wiley-VCH; 2011. [18] Wallgram W, Clemens H, Schloffer M. Method for producing a component and components of a titanium-aluminum base alloy, patent application number 20110277891, 2011. [19] Wu X. Review of alloy and process development of TiAl alloys. Intermetallics 2006;14:1114e22. [20] Appel F, Oehring M, Paul JDH, Klinkenberg C, Carneiro T. Physical aspects of hot-working gamma-based titanium aluminides. Intermetallics 2004;12: 791e802. [21] Appel F, Paul JDH, Oehring M, Clemens H, Fischer F-D. Physical metallurgy of high Nb-containing TiAl alloys. Zeitschrift Fuer Metallkunde/Materials Research and Advanced Techniques 2004;95:585e91.

139

[22] Takeyama M, Kobayashi S. Physical metallurgy for wrought gamma titanium aluminides: microstructure control through phase transformations. Intermetallics 2005;13:993e9. [23] Imayev VM, Imayev RM, Oleneva TI, Khismatullin TG. Microstructure and mechanical properties of the intermetallic alloy Tie45Ale6(Nb, Mo)e0.2B. The Physics of Metals and Metallography 2008;106:641e8. [24] Imayev V, Oleneva T, Imayev R, Christ H-J, Fecht H-J. Microstructure and mechanical properties of low and heavy alloyed g-TiAl þ a2-Ti3Al based alloys subjected to different treatments. Intermetallics 2012;26:91e7. [25] Imayev RM, Imayev VM, Khismatullin TG, Oehring M, Appel F. New approaches to designing alloys based on g-TiAl þ a2-Ti3Al phases. The Physics of Metals and Metallography 2006;102:105e13. [26] Imayev VM, Imayev RM, Khismatullin TG. Mechanical properties of the cast intermetallic alloy Tie43Ale7(Nb, Mo)e0.2B (at %) after heat treatment. The Physics of Metals and Metallography 2008;105:484e90. [27] Yang J, Wang JN, Wang Y, Xia Q. Refining grain size of a TiAl alloy by cyclic heat treatment through discontinuous coarsening. Intermetallics 2003;11: 971e4. [28] Schwaighofer E, Schloffer M, Schmoelzer T, Mayer S, Lindemann J, Guether V, Klose J, Clemens H. Influence of heat treatments on the microstructure of a multi-phase titanium aluminide alloy. Practical Metallography 2012;49:124e 37. [29] Cha L, Clemens H, Dehm G. Microstructure evolution and mechanical properties of an intermetallic Tie43.5Ale4Nbe1Moe0.1B alloy after ageing below the eutectoid temperature. International Journal of Materials Research (Formerly Z Metallkd) 2011;102:703e8. [30] Schloffer M, Iqbal F, Gabrisch H, Schwaighofer E, Schimansky FP, Mayer S, Stark A, Lippmann T, Göken M, Pyczak F, Clemens H. Microstructure development and hardness of a powder metallurgical multi phase g-TiAl based alloy. Intermetallics 2012;22:231e40. [31] Schloffer M, Rashkova B, Schoeberl T, Schwaighofer E, Zhang Z, Clemens H, Mayer S. Evolution of uo-phase in a b-stabilized multi-phase TiAl alloy and its effect on hardness. 2013 Submitted to Acta Materialia. [32] Güther V, Joos R, Clemens H. Microstructure and defects in g-TiAl based vacuum arc remelted ingot materials. In: Hemker KJ, editor. 3rd Int. Symp. on structural intermetallics. USA: Jackson Hole WY; 2001. [33] Achtermann M, Fürwitt W, Guether V, Nicolai H.-P. Method for producing a gTiAl base alloy. Patent EP2010/064306, 2011. [34] Achtermann M, Guether V, Klose J, Nicolai H-P. Production of g-TiAl based feed stock materials for subsequent investment casting and forging operations. In: Presentation at 4th international workshop on titanium aluminides, GfE metalle und materialien GmbH, Nuremberg, Germany 2011. [35] Schmoelzer T, Liss K-D, Staron P, Mayer S, Clemens H. The contribution of high-energy X-rays and neutrons to characterization and development of intermetallic titanium aluminides. Advanced Engineering Materials 2011;13: 685e99. [36] Saunders N. Phase equilibria in multi-component g-TiAl based alloys. In: Kim YW, Dimiduk DM, Loretto MH, editors. Gamma titanium aluminides. Warrendale, PA: TMS; 1999. p. 183e8. [37] Schloffer M, Schmoelzer T, Mayer S, Schwaighofer E, Hawranek G, Schimansky FP, Pyczak F, Clemens H. The characterisation of a powder metallurgically manufactured TNMTM titanium aluminide alloy using complimentary quantitative methods. Practical Metallography 2011;48:594e604. [38] Clemens H, Boeck B, Wallgram W, Schmoelzer T, Droessler LM, Zickler GA, Leitner H, Otto A. Experimental studies and thermodynamic simulations of phase transformations in Ti-(41-45)Al-4Nb-1Mo-0.1B alloys. In: Materials Res. Soc. Symp. Proc. Warrendale: MRS; 2008. p. 115e20. [39] Schloffer M, Themessl A, Schwaighofer E, Clemens H, Heutling F, Helm D, Achtermann M, Mayer S. Phase transitions and phase equilibria in the TiAlNb-Mo system. In: Presentation at 4th international workshop on titanium aluminides, GfE metalle und materialien GmbH, Nuremberg, Germany 2011. [40] Hecht U, Witusiewicz V, Drevermann A, Zollinger J. Grain refinement by low boron additions in niobium-rich TiAl-based alloys. Intermetallics 2008;16: 969e78. [41] Burgers WG. On the process of transition of the cubic-body-centered modification into the hexagonal-close-packed modification of zirconium. Physica 1934;1:561e86. [42] Oehring M, Stark A, Paul JDH, Lippmann T, Pyczak F. Microstructural refinement of boron-containing b-solidifying g-titanium aluminide alloys through heat treatments in the b phase field. Intermetallics 2013;32:12e20. [43] Blackburn MJ, editor. The science, technology and application of titanium. Oxford: Pergamon Press Ltd.; 1970. p. 633e43. [44] Watson IJ. In situ characterization of a Nb and Mo containing g-TiAl based alloy using neutron diffraction and high-temperature microscopy. Advanced Engineering Materials 2009;11:932e7. [45] Schwaighofer E. Influence of heat treatments on the microstructure and mechanical properties of cast and hot-isostatically pressed TNM alloys. Diploma thesis. Montanuniversität Leoben; 2010. [46] Lindemann J. Unpublished data, 2011. [47] McCusker LB, Von Dreele RB, Cox DE, Louër D, Scardi P. Rietveld refinement guidelines. Journal of Applied Crystallography 1999;32:36e50. [48] Stark A, Oehring M, Pyczak F, Schreyer A. In situ observation of various phase transformation paths in Nb-rich TiAl alloys during quenching with different rates. Advanced Engineering Materials 2011;13:700e4. [49] Schwaighofer E, Clemens H, Mayer S. Unpublished data 2009e2012.

140

E. Schwaighofer et al. / Intermetallics 44 (2014) 128e140

[50] Güther V. TNM data sheet Nr. 1, GfE Metalle Und Materialien GmbH; 2010. Nuremberg, Germany. [51] Koeppe C, Bartels A, Seeger J, Mecking H. General aspects of the thermomechanical treatment of two-phase intermetallic TiAl compounds. Metallurgical Transactions A 1993;24:1795e806. [52] Koeppe C, Bartels A, Clemens H, Schretter P, Glatz W. Optimizing the properties of TiAl sheet material for application in heat protection shields or propulsion systems. Materials Science and Engineering: A. 1995;201: 182e93. [53] Viswanathan GB, Vasudevan VK. Processing, microstructure and tensile properties of a Ti-48 AT.% Al alloy. Scripta Metallurgica et Materialia 1995;32: 1705e11. [54] Appel F, Wagner R. Microstructure and deformation of two-phase g-titanium aluminides. Materials Science and Engineering: R: Reports 1998;22:187e268. [55] Lin D, Wang Y, Law CC. Thermal activation processes of tensile deformation in g-TiAl alloy. Materials Science and Engineering: A. 1997;239e240:369e77.

[56] Appel F, Lorenz U, Oehring M, Sparka U, Wagner R. Thermally activated deformation mechanisms in micro-alloyed two-phase titanium amminide alloys. Materials Science and Engineering: A. 1997;233:1e14. [57] Follansbee PS, Kocks UF. A constitutive description of the deformation of copper based on the use of the mechanical threshold stress as an internal state variable. Acta Metallurgica 1988;36:81e93. [58] Schloffer M. Gefüge und Eigenschaften der intermetallischen TNM-Legierung. PhD thesis. Montanuniversität Leoben; 2013. [59] Chatterjee A, Mecking H, Arzt E, Clemens H. Creep behavior of g-TiAl sheet material with differently spaced fully lamellar microstructures. Materials Science and Engineering: A. 2002;329e331:840e6. [60] Dehm G, Motz C, Scheu C, Clemens H, Mayrhofer P, Mitterer C. Mechanical size-effects in miniaturized and bulk materials. Advanced Engineering Materials 2006;8:1033e45. [61] Wang J, Nieh T. Creep of a beta phase-containing TiAl alloy. Intermetallics 2000;8:737e48.