Microstructure and mechanical properties of titanium aluminide compositions containing Fe

Microstructure and mechanical properties of titanium aluminide compositions containing Fe

Materials Science & Engineering A 575 (2013) 152–159 Contents lists available at SciVerse ScienceDirect Materials Science & Engineering A journal ho...

2MB Sizes 0 Downloads 58 Views

Materials Science & Engineering A 575 (2013) 152–159

Contents lists available at SciVerse ScienceDirect

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

Microstructure and mechanical properties of titanium aluminide compositions containing Fe C.J. Bettles a,n, S. Tochon a, M.A. Gibson b, B.A. Welk c, H.L. Fraser c a

ARC Centre of Excellence for Design in Light Metals, Department of Materials Engineering, Monash University, Clayton 3800, Victoria, Australia CSIRO Light Metals Flagship, CSIRO Process Science and Engineering, Private Bag 33, Clayton South MDC, Clayton 3169, Victoria, Australia c Center for Accelerated Maturation of Materials, Department of Materials Science and Engineering, The Ohio State University, Columbus, OH, USA b

art ic l e i nf o

a b s t r a c t

Article history: Received 20 May 2012 Received in revised form 21 March 2013 Accepted 22 March 2013 Available online 3 April 2013

Six titanium aluminide alloys, based on the Ti–Al–Fe system, in the composition space defining the limits of the three-phase (γ+α2+β/B2) field have been prepared. The effects of heat treatment temperature (1000 1C and 1200 1C) and cooling rate (water quenching and air cooling) on the microstructure and room temperature mechanical behaviour (as determined by a shear punch test) are reported. It is shown that additions of Fe between 1.4 at% and 2.8 at% to the binary Ti–44Al alloys may improve room temperature ductility without sacrificing strength. The presence of the disordered bcc phase, rather than the ordered B2 phase, is required to achieve higher room temperature ductility. The preferred microstructure contains β-phase between 15% and 20%, together with a fraction of lamellar (γ+α2) colonies between 30% and 70%. The remainder of the microstructure is γ-phase. & 2013 Elsevier B.V. All rights reserved.

Keywords: Titanium alloys Intermetallics Electron microscopy Mechanical characterisation

1. Introduction Titanium aluminides have long been identified as promising high temperature materials. They possess an attractive range of mechanical and physical properties, but are compromised by poor ductility, at room temperature particularly. To alleviate this shortcoming, the preferred compositions are those with Ti:Al ratios which produce a duplex microstructure containing a combination of γ-phase grains and fine (α2+γ) lamellar colonies. The binary Al–Ti phase diagram shows that the (α2+γ) phase field exists over a wide range of compositions and, most interestingly, the solidification pathway may be through a β-phase field or peritectically through an α-phase field (L+β-α) [1,2]. There is a considerable body of research based on investigating the effects of small additions of ternary alloying elements on the room temperature ductility, and these have concentrated on modifying the lattices of the two primary α2 and γ phases [3,4]. Ternary and quaternary alloying additions have been employed to improve processability, and an alloy currently in application has a composition containing both chromium and niobium together with an aluminium content which ensures that solidification is through the peritectic reaction [5]. It is possible, therefore, with appropriate alloying additions and microstructural manipulation to achieve slight improvements in ductility whilst retaining the two-phase microstructure.

n

Corresponding author. Tel.: +61 3 99051966; fax: +61 3 99054940. E-mail address: [email protected] (C.J. Bettles)

0921-5093/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.msea.2013.03.058

An extension of this approach is to increase the amount of those alloying elements known to be β-stabilisers to move the composition from the two-phase into the three phase (β+α2+γ) region of the phase diagram, over specific temperature ranges. The main aim in this situation is to improve the elevated temperature formability of the titanium aluminides by being able to carry out the thermomechanical processing in the presence of the β-phase [6]. Clemens and co-workers [7,8] have developed a Ti51.9Al43Nb4Mo1B0.1 alloy, which displays a high volume fraction of β/B2-phase at elevated temperatures and consequently can be forged under near conventional processing conditions. In these cases, the β(B2)phase is not stable at room temperature so that a conventional duplex α2+γ microstructure exists after final processing. These elements may be either β-isomorphous, such as Mo, Zr and Nb [9,11] or β-eutectoid, such as Cr, Mn and W [12,13] in nature, or in combination [14]. Alternatively, the composition may be adjusted so that the three phases are retained at room temperature or that an entirely different region, two-phase (γ+β) for example, can exist [15,16]. It is possible to stabilise the β/B2 phase at room temperature by increasing the levels of alloying additions, but this can often be accompanied by the formation of ω-phase. Cheng and Loretto observed the formation of ω-phase in alloys containing either 8% Nb or 4% Nb and 4% Zr after heat treatment at 1350 1C and furnace cooling [17]. These same authors also observed that water quenching from the same temperature partially suppressed the formation of the ω-phase. Huang, in a grain refined version of the same alloy system that had been hot isostatically pressed, found that the B2-phase was present as a distribution of fine cells, and that the ω-phase could exist in two different forms, related to

C.J. Bettles et al. / Materials Science & Engineering A 575 (2013) 152–159

whether the phase was within the cells or the cell walls [18]. It was also proposed that the presence of the ω-phase contributed to the observed poor tensile ductility of the alloy. Iron additions have been less popular, primarily due to problems associated with segregation during solidification processing. However, it is possible that small additions of Fe can have a significant effect on the extent of the (α2+γ) phase field, and that the effect is composition sensitive. Inkson et al. found that small amounts of the ordered B2-phase were present in an alloy containing ∼45% Al, 1.6% Fe and 1.1% V after heat treatment at 1200 1C followed by furnace cooling [19]. Fe had preferentially partitioned to the B2-phase, which was found to have a Ti:Al ratio close to that of the α2 phase. Several authors have attempted to describe the ternary Al–Fe–Ti phase diagram, revealing in the 1000 1C isotherm a three-phase field for aluminium contents between approximately 35 at% and 45 at%, which is B2+TiAl+Ti3Al, where B2 is the ordered β-phase, and also the high solubility of Al in the Fe2Ti phase and the existence of the τ2 (cubic variant) and τ2* (tetragonal variant) phases [20–22]. Potentially, these intermetallic phases will have a deleterious effect on ductility. This work has been expanded and refined by both Kainuma [23,24] and Palm [25]. This region of the phase diagram is both complex and dynamic. At 1200 1C τ2 no longer exists and the β-Ti region has enlarged significantly. At 800 1C the Fe2Ti phase field has diminished and the τ2/τ2* phases exist in what is shown as a single field from 25–60 at% Al. The composition and structure information for the relevant phases are given in [25]. The microstructure of the γ-TiAl materials holds the key to their mechanical behaviour. In particular the plastic flow properties (including ductility and work hardening behaviour) are highly dependent on the fraction of γ-phase against lamellar colony fraction, size and composition of the γ-phase, and lamellar colony size and lamellar spacing [26–30]. In general terms, the duplex microstructures with primary γ-phase and lamellar colonies provide some tensile ductility at room temperature, whilst the fully lamellar structures tend to be more brittle. The yield strength increases as the fraction of lamellar structure increases, and in terms of a work hardening analysis (Jaoul–Crussard approach) Rao and Tangri show that two-phase alloys and single-phase γ alloys have multi-stage deformation mechanisms whereas single-phase α2 has a single stage deformation step [26]. These authors report that the work hardening coefficient, n, for the two-phase alloys, during Stage I deformation, is 0.18–0.19. Chan and Kim reported that the alloy Ti–47Al–2.6Nb–2.0(Cr+V) with a duplex microstructure has a ductility of 3.7% and an n value of 0.09 at 25 1C, whereas in a fully lamellar form the ductility is 0.88% and the n value is 0.162 [27]. Appel also observed that the work hardening characteristics of the two-phase and single-phase γ alloys were very similar, but confirmed that at the dislocation level the mechanisms are quite different, and this could have significant implications in terms of alloy development [28]. Chan has been particularly interested in the ductility and fracture toughness of brittle intermetallics, including the TiAl materials, and has presented theoretical analyses of crack instability behaviour [31,32]. The indication being that there are two means for improving fracture behaviour: intrinsic (related to properties inherent in the material) and extrinsic (related to structure); and that modifying the intrinsic behaviour is more appropriate for the aluminides. This is most readily achieved by alloying and/or ductile phase blunting of the crack tip. Whilst this is concerned with fracture toughness, the approach can also be used to consider improvements to ductility. Recently Chen et al. [33] found that the fracture toughness of a Ti– 45Al–2Nb–1.5V–1Mo–0.3Y alloy, with a microstructure consisting mainly of fine lamellar grains together with mixtures of γ and residual β phases along lamellar colony boundaries, displayed a higher KIC value up to 23.5 MPa m1/2 at room temperature, compared to a fully lamellar Ti–45Al–5Nb–0.3Y alloy.

153

Gorzel et al. have examined several alloys in the Ti–Al–Fe system, concentrating on microstructures arising after heat treatment at 1000 1C and a slow furnace cool [34]. Five of these alloys are in the region of interest to the present study, and the microstructures were multiphase containing B2, γ and τ2. The compressive response at room and elevated temperatures was determined. It is the room temperature properties that are of most relevance to this work, and it was reported that for alloys in a two-phase (γ+B2) region, yield strengths were highest, and increased as the fraction of B2 increased. Interestingly, a three phase microstructure, with equal fractions of γ and α2 and 20% B2, has the same yield strength as a (γ +B2) alloy with 22% B2. The overall conclusion was that (γ+τ2) and (B2+τ2) alloys have good combinations of modulus, compressive yield and hardness behaviour. In this study, it is the Ti–Al–Fe system that is of interest, particularly the composition space defining the limits of the threephase (γ+α2+β/B2) field. The effects of heat treatment temperature and cooling rate on the microstructure and room temperature mechanical behaviour (as determined by a shear punch test) of six alloys are reported.

2. Experimental details Six alloys were selected for investigation, spanning the (γ+α2), (α2+β/B2), (γ+α2+β/B2) and (γ+β/B2) phase fields. The positions on the partial ternary phase isotherms at 800 1C, 1000 1C and 1200 1C are shown in Fig. 1 and the chemical analyses for these alloys are given in Table 1. Note that the Ti, Al and Fe contents are in at%, and the oxygen content is given in wt%. The LENSTM deposition technique was used to prepare cylinders 13 mm diameter and 25 mm high. To produce a single composition, blended powder feedstock was used (source powder feedstock: gas atomised Ti50Al50 −140 mesh, gas atomised CP-Ti −140 mesh+325 mesh, both ex-Crucible Research, and 99.9% Fe plasma sprayed grade −100 mesh+325 mesh, ex-Atlantic Equipment Engineers). The blends were thoroughly mixed in a rotary mixer for 2 h prior to LENS™ processing. Two columns of the same composition were deposited simultaneously on the polished surface of a preprepared argon arc melted button (TEK Specialties Inc.), of nominal composition Ti56Al42Fe2, to minimise the thermal expansion mismatch with the base plate and facilitate the start-up of the deposition process. The alloys were heat treated for 6 h at both 1000 1C and 1200 1C under vacuum, and were either fast cooled through a water quench (WQ) or slow cooled (∼440 1C/min) in air (AC). Microstructures were observed by SEM and, in some cases, TEM. The SEM images were obtained using an FEI Phenom SEM, which has an accelerating voltage of 5 keV, operating in a backscattered electron mode. Phase identification was carried out using conventional (Cu Kα) X-ray diffraction (XRD). The differentiation between the disordered β and the ordered B2 phases was through X-ray diffraction, concentrating particularly on the (100) peak at 2θ ¼281 which only appears in the ordered patterns. The intensity of this peak is related to the composition, and can thus only be used as a general indication of ordering. The authors acknowledge that for a more in-depth analysis, electron diffraction would be a more appropriate technique. Phase fractions were estimated from the SEM micrographs, using the ImageJ software. The lamellar spacing was determined using a linear intercept method (ASTM E112). The values quoted demonstrate the range observed in the average spacing, taken over many colonies in different areas of the samples. The mechanical properties were determined using a shear punch test (SPT), which is suitable for very small samples [35]. This test is based on a blanking operation, and the stress state is predominantly one of shear. The results have been found to

154

C.J. Bettles et al. / Materials Science & Engineering A 575 (2013) 152–159

Table 2 Phases present and volume fractions (%) after heat treatment for each alloy composition. Alloy

1000 WQ

1000 AC

1200 WQ

1200 AC

1

α2 50 B2 50 –

α2 49 β/B2 49 τ2 2

α2 30 B2/β 65 g5

α2 36 B2 64 –

2

α2 49 B2 45 γ6 –

α2+63 β 30 γ4 τ2 3

α2 32 B2 68 – –

α2 36 B2 62 γ2 –

3

α2 50 γ 50 –

α2 49 γ 50 τ2 1

α2 50 γ 50 –

α2 64 γ 36 –

4

α2 47 γ 47 – τ2 6

α2 49 γ 49 τ2 2 –

α2 41 γ 41 β 18 –

α2 42 γ 52 β6 –

5

γ 82 α2 6 B2 12

γ 82 α2 12 τ2 6

γ 65 α2 18 β 17

γ 59 α2 25 β 16

6

γ 75 α2 15 τ2 10

γ 63 α2 17 τ2 20

γ 33 α2 34 τ2 33

γ 45 α2 44 β 11

3. Results and discussion The compositions were chosen to cross the three phase (α2+γ +B2/β) fields along lines of constant Fe content (A1, A2 and A5) and constant Al (A3, A4, A5 and A6) content. 3.1. Microstructure

Fig. 1. Portions of the (a) 800 1C, (b) 1000 1C and (c) 1200 1C isotherms of the Ti–Al–Fe phase diagram [25] showing the position of the six alloys prepared.

Table 1 Chemical analyses of the six alloys investigated (with Ti, Al and Fe in at% and oxygen in wt%). Alloy

Ti

Al

Fe

Oxygen (wt%)

A1 A2 A3 A4 A5 A6

65.70 62.06 56.43 55.40 53.46 52.34

32.05 35.44 42.80 43.19 43.76 43.51

2.25 2.50 0.77 1.41 2.78 4.15

0.2 0.24 0.19 0.25 0.18 0.19

correlate well with uniaxial tensile testing, and there are “rules of thumb” for converting the SPT results into equivalent tensile properties (UTS and ductility in particular) for many metals, but there has not been any such correlation for low ductility materials. The authors have carried out a direct comparison with tensile data from Forwood et al. [3] on compositions showing poor (Ti52Al48) and improved ductility ((Ti51.5Al47.5Zr1)98Cr2). It was found that tensile ductilities of 0.5% and 1.88% translated into εfracture under SPT of 0.12 and 0.18 respectively. Similarly, tensile strengths of 404 MPa and 662 MPa translated to sfracture of 355 and 417 MPa respectively. The results will be presented as SPT data, and the alloy comparisons made on that basis.

The phase identification was performed using a combination of back-scattered electron imaging (i.e. Z contrast) and X-ray diffraction (XRD), and the phases found to be present in each condition are listed in Table 2. Selected XRD traces are shown in Fig. 2, highlighting both the clear distinctions between the phases present and the difficulties associated with distinguishing between the ordered and disordered β-phases. The microstructures obtained from all compositions and the four heat treatments are described below and representative images highlighting the important features of the microstructures are shown in Figs. 3–7. 3.2. Alloy 1 Ti–32.05Al–2.25Fe This composition was expected to be single phase β at 1200 1C and (α2+B2) at 1000 1C. The microstructures after heat treatment at 1000 1C were very similar, irrespective of the cooling rate (WQ vs. AC) and the grain size was approximately 50 mm. Both microstructures contained the phases expected from the phase diagram —the equiaxed matrix β or B2 phase (after air cooling and water quenching respectively) and α2 along the grain boundaries and within the matrix grains—but in the air cooled structure a very small fraction of a Fe-rich phase was also observed. The volume fractions of the two phases were approximately equal. At 1200 1C, the microstructure shows both α2 and B2 phases. The equiaxed grain size is significantly larger at ∼100 mm and the grain boundary phase is much thicker (in the order of 10 mm) than after 1000 1C (Fig. 3). Overall, in both cases the amount of α2 has reduced, and that within the grains is blocky and up to 10 μm in size. The air cooled sample contains ∼36% α2 which is slightly more than the water quenched sample (35%). The matrix is the ordered B2-phase irrespective of cooling rate.

C.J. Bettles et al. / Materials Science & Engineering A 575 (2013) 152–159

These observations are mainly consistent with the phase diagram isotherms for the heat treatment temperatures. The isotherm at 800 1C shows a dramatic change, with the expansion of the τ/τ2 region, and Alloy 1 at equilibrium would be expected to fall within a two-phase field containing α2 and τ2 at that temperature. The Fe-containing phase is thus likely to be τ2, which has formed during slow cooling from 1000 1C.

B2 Ti3Al TiAl tau2 Beta A6 1200 AC

3.3. Alloy 2 Ti–35.44Al–2.50Fe

A2 1200 AC

A6 1000 AC 20

30

155

40

50

60

70

80

90

100

2 theta

Fig. 2. Selected XRD traces showing the positions of the peaks from the different phases observed in this alloy system.

α2

α2

B2

The 1200 1C isotherm has this alloy in the (α+β) two-phase field, close to the single β-phase boundary, whilst at 1000 1C it is within the target three-phase field. At 800 1C, the composition is still within a three-phase field, but this is now the (α2+γ+τ2) field. It was found that alloys of this composition heat treated at 1000 1C were of three phases β(B2)+α2+γ irrespective of cooling rate, with the γ-phase being in the minority (4–6% for 1000 1C AC and WQ). Representative microstructures are shown in Fig. 4. In the water quenched condition, the matrix phase is the β/B2, with the α2 present at grain boundaries and within grains. The γ has two morphologies—equiaxed globular primary phase and lenticular shaped secondary precipitates. The microstructure after quenching from 1000 1C retained the ordered B2 phase and is shown in Fig. 4. In contrast, air cooling has resulted in the presence of the disordered β-phase rather than the B2-phase and the additional formation of τ2, which appears to be located either within the α2-phase or at α2/β boundaries. The τ2 has a volume fraction of ∼2.5%. This is consistent with the observation for Alloy 1 and the phase diagram at 800 1C. The fraction of α2 is ∼63%, and this declines after water quenching to ∼49%. Treatment at 1200 1C has led to a two-phase microstructure and a significantly larger grain size. The α2-phase is found along the grain boundaries in a continuous fashion after both cooling rates, but within the grains, in the WQ material, it exists predominantly as large equiaxed primary α2 (total fraction 32%), whereas the air cooled material shows additional precipitation of α2 within the grains and a precipitate free zone adjacent to the grain boundary α2. The total fraction of α2 is ∼36%, of which approximately a third is located at the grain boundaries. 3.4. Alloy 3 Ti–42.8Al–0.77Fe

40 μm Fig. 3. BSE micrograph of Ti–32.05Al–2.25Fe (Alloy 1) after heat treatment at 12001 C WQ.

Alloy 3 displays the fully lamellar microstructure associated with the near-γ titanium aluminides (Fig. 5(a)). This is not the most desirable microstructure for “enhanced” ductility, that is generally the duplex structure with γ+(γ+α2)lam. Forwood et al. showed “ductile” microstructures which contained lamellar colonies plus faceted α2 and equiaxed γ [3]. Annealing at 1000 1C

α2

α2

α2 B2

τ2

B2

γ B2

15 μm

15 μm

80 μm

Fig. 4. BSE micrographs of Ti–35.4–Al2.5Fe (Alloy 2) after heat treatment at (a) 10001 C AC, (b) 10001 C WQ showing the large B2 grains (light), α2-phase (light grey) at both the grain boundaries and within the grains together with a smaller fraction of γ-phase (dark grey) and (c) 12001 C WQ.

156

C.J. Bettles et al. / Materials Science & Engineering A 575 (2013) 152–159

γ β α2+ γ+ β

α2+ γ

α2+ γ 20 μm

30 μm

Fig. 5. BSE micrographs of (a) Ti–42.8Al–0.77 Fe (Alloy 3) after heat treatment at 12001 C WQ, showing the fine colony structure with ∼10% intercolony (α2+γ) and (b) Ti–43.2Al–1.4Fe (Alloy 4) also after heat treatment at 12001 C WQ.

β

B2 γ

γ α2+ γ α2+ γ

20 μm

20 μm

Fig. 6. BSE micrographs of Ti–43.8Al–2.7Fe (Alloy 5) after heat treatment at (a) 10001 C WQ, and (b) 12001 C WQ.

τ2 γ α2

τ2

α2

γ

20 μm

5 μm

20 μm

Fig. 7. BSE micrographs of Ti–43.5Al–4.1Fe (Alloy 6) after heat treatment at (a) 1000 1C WQ (low magnification), (b) 1000 1C WQ (high magnification) and (c) 1200 1C WQ.

resulted in large colonies of very fine α2+γ lamellae (spacing 190– 250 nm), with a coarser more irregular lamellar structure at some intercolony boundaries. The water quenching appeared to result in a slightly coarser lamellar spacing, but this has not been verified statistically and could simply be an effect of orientation. The air cooled material also contains a very small amount (of the order of

1%) of τ2, at colony boundaries and within single colonies in regions where there is localised coarsening of the γ- and α2-phases. After 1200 1C annealing and slow cooling, the colonies are larger and the lamellar spacing definitely coarser than that after 1000 1C annealing. There are also two-phase regions between the lamellar colonies, containing blocky γ together with α2, which

C.J. Bettles et al. / Materials Science & Engineering A 575 (2013) 152–159

are not lamellar in morphology. The water quenched microstructure contains predominantly lamellar colonies between 20 mm and 40 μm across, which are smaller than those in the air cooled microstructure. However, whilst the spacing within a colony is uniform, there is considerable variation in the spacing from colony to colony, ranging from 300 nm to 500 nm. This microstructure produced the best strength and ductility for this composition, and is shown in Fig. 5(a). TEM confirmed that the lamellar structure displayed the well known orientation relationship between the γ- and α2-phases: o110 4γTo 11–204 α2 and {111}γT{0002}α2.

157

retain the two-phase microstructure on cooling from 1200 1C, with the XRD indicating that after air cooling the resulting microstructure contains ∼11% β-phase and equal amounts of blocky γ and α2. On water quenching the three phases present are γ+α2+τ2, in approximately equal fractions. There has been a general coarsening of the microstructure and a total loss of any lamellar morphology. It is clear from the above descriptions that several of the alloys do not contain the phases expected from the most recent isotherms published by Palm [25]. The most interesting observations are

3.5. Alloy 4 Ti–43.19Al–1.41Fe This composition was expected to be the desired three-phase structure after all heat treatments. However, the 1000 1C treatments resulted in the formation of τ2 after air cooling, rather than B2. τ2 was also present after water quenching, however this was in addition to the three expected phases. At the higher temperature, β remained after both slow and fast cooling, and was a significant fraction of the microstructure. In the material air cooled from 1200 1C, β was generally located alongside coarse γ and α2 in regions between the fine (γ+α2) lamellar colonies, and made up ∼6% of the total structure. In the water quenched material, which contains ∼70% of fine (∼500 nm spacing) lamellar colonies and is shown in Fig. 5(b), it is also the disordered β-phase that is present and, in addition to being between the lamellar colonies, it is also present within the individual colonies in bands of (β+γ). In total, the β-phase makes up ∼15–18% of the microstructure. TEM on the 1200 1C WQ material showed that the blocky primary γ-phase was made up of massive twins. However, the additional spots reported by Ducher to exist in some of the [110] diffraction patterns of the Fe-containing γ-phase were not observed in this condition [36]. 3.6. Alloy 5 Ti–43.8Al–2.78Fe Alloy 5 was selected to be in a two-phase γ+β/B2 region irrespective of the annealing temperature. This same alloy, if annealed at 800 1C, would also be in a two-phase region, but the phases would be γ and τ2. At this temperature the composition does lie, however, close to the three-phase (γ+α2+B2) boundary. Annealing at 1000 1C followed by slow cooling produces a microstructure dominated by blocky γ-phase together with α2+τ2 coexisting between the large γ-grains. Together the α2+τ2 constitute ∼18% of the structure and the ratio of τ2:α2 is 2:1. The water quench (Fig. 6(a)) produces (γ+α2+B2), existing as blocky γ-phase, (γ+α2) colonies (7% of the total) and (α2+B2) between γ-grains. The ratio of the B2:α2 in the intergranular phase is also 2:1 and the combined fraction of the two phases is ∼18%. Heat treating at 1200 1C clearly results in a three phase microstructure—γ+α2+β. The microstructure is loosely based on very coarse lamellar colonies (the width of an individual lamella is ∼1–2 mm), which contain pockets of β-phase, separated by broad regions containing all three phases. After air cooling, the β fraction is ∼16% and the colony fraction is ∼50%, and this has changed to ∼17% and ∼36% respectively after the faster cooling (Fig. 6(b)). 3.7. Alloy 6 Ti–43.51Al–4.15Fe The composition of Alloy 6 lies within the γ+B2 phase field at 1200 1C and inside the three-phase γ+B2+τ2 at 1000 1C, according to the phase diagrams. Heat treating at 1000 1C leads to a room temperature microstructure of γ+α2+τ2. B2 is not retained after either fast or slow cooling from this temperature. The slow cooled material is primarily γ-phase with (α2+τ2) between—the τ2 fraction is ∼20%. With quenching (Fig. 7(a)) the structure contains a mixture of γ and α2 together with τ2. It has not been possible to

(1) Alloy 1 is two-phase at 1200 1C, (2) Alloy 5 is three-phase at both 1200 1C and 1000 1C, and (3) Alloy 6 shows τ2 after quenching from 1200 1C. In an on-going work yet to be published, where compositionally graded samples with Fe contents up to 6 at% have been examined, we have observed that the levels of Fe that may substitute in the γ-TiAl lattice after treatment at 1200 1C exceed that appearing on the published isotherms, and we propose that the maximum Fe content is closer to 3.5 at% rather than 2.5 at%. There is also a similar increase in the maximum Fe in the α2 phase, and a significant increase in the Fe level at the “knee” of the (Ti)HT field that forms one corner of the three-phase (γ+α2+β) field. Overall, this would mean that the (γ+α2) phase field is increased and the (γ+α2+β) field is enlarged and shifted towards higher Fe levels. Under these circumstances Alloy 1 would be in the (γ+α2) region and Alloy 5 would be in a three-phase region at both 1200 1C and 1000 1C, and the observations presented here are consistent with this hypothesis. Clearly, further work is required to validate this assertion, and this is continuing. The presence of τ2 on slow cooling from treatments at 1000 1C is not entirely unexpected, given the presence of several τ2-containing phase fields at 900 1C. With the complexities around phase boundary shifts at temperatures between 900 1C and 1000 1C, arising from the formation of the large τ/τ2 phase field, it is difficult to construct continuous isopleths at either 44 at% Al or 2.5 at% Fe. At 1000 1C Alloy 6 should be within a τ2-containing field. However, at 1200 1C τ2 is unstable and does not appear on the equilibrium diagrams. At this stage, we cannot offer an explanation for the presence of τ2 after quenching from 1200 1C. However, the observation is not an isolated one, and the same behaviour has occurred in the compositionally graded samples mentioned earlier. Work is also continuing in this area.

4. Mechanical behaviour The shear punch test is useful for comparing the properties of materials when only small volumes are available. In this case it was deemed more appropriate to obtain repeated results using a small test piece rather than a single result from an alternative method. All results are averages of four tests, and the effective strain to failure and maximum effective stress for all alloys and heat treatment conditions are given in Tables 3 and 4, respectively. Included in these tables are the shear punch results for two compositions that Forwood et al. have previously investigated using conventional tensile testing [3]. Clearly, these two alloys have failure strains superior to any of the Fe-containing alloys. However, these compositions are higher in Al content than the alloys in this paper and they were also hot isostatically pressed prior to heat treatment. Morris et al. have reported tensile properties for aluminide alloys containing 44% Al and those heat treated at 1200 1C and WQ have a ductility of 0.2%, which would be comparable to the SPT fracture strains for the Fe-containing alloys [37]. The results for the Fe-

158

C.J. Bettles et al. / Materials Science & Engineering A 575 (2013) 152–159

Table 3 Strain at failure results from shear punch testing. Alloy

1000 1C WQ

1000 1C AC

1200 1C WQ

1200 1C AC

1 2 3 4 5 6 Ti52Al48 (Ti51.5Al47.5Zr1)98Cr2

0.0 0.0177 0.006 0.009 70.006 0.02 7 0.01 0.017 0.002 0.0067 0.004 – –

0.0 0.0 0.0137 0.01 0.0097 0.004 0.0087 0.004 0.0 – –

0.026 7 0.006 0.0147 0.004 0.0357 0.001 0.0477 0.008 0.041 70.004 0.0187 0.004 – –

0.0 0.00 0.0127 0.004 0.0147 0.001 0.03 7 0.005 0.0167 0.003 0.127 0.05 0.187 0.04

Table 4 Maximum stress (MPa) from shear punch testing. Alloy

1000 1C WQ

1000 1C AC

1200 1C WQ

1200 1C AC

1 2 3 4 5 6 Ti52Al48 (Ti51.5Al47.5Zr1)98Cr2

301 732 349 773 3477 22 3767 29 3417 22 269 7 43 – –

3567 59 1467 14 3677 17 412 7 5 281 739 1117 26 – –

560 7 19 492 7 15 408 7 3 446 74 481 7 10 382 7 26 – –

1587 8 1627 26 295 7 16 4127 10 4317 14 3777 25 3557 44 4167 16

600

Effective Stress

500 400 SPT 1200AC SPT 1000AC SPT 1200WQ SPT 1000WQ

300 200 Low Ductility

Improved Ductility

100 Zero Ductility 0 0

0.01

0.02

0.03

0.04

0.05

Effective Strain Fig. 8. Property space for the alloys (numbered as in Table 1) and heat treatment conditions, as defined by effective failure stress and effective failure strain.

containing alloys are presented graphically in Fig. 8 as a plot of effective fracture stress vs. effective fracture strain, and it can be seen that the alloys broadly fall into three behavioural envelopes based on ductility—εfracture ¼ 0, 0oεfracture o0.02, and εfracture 40.02 (designated as “improved”). The figure is plotted such that both composition (by alloy number) and heat treatment (by shape) can be distinguished. Interestingly, samples showing the “best” ductility are amongst those with the highest effective stress. It is clear that the fracture strain is not so much related to composition but to microstructure, in that it is possible to have a single composition (for example Alloy 1) displaying zero fracture strain under some heat treatment conditions (i.e. microstructure) and a value of 0.025 under another condition. This is not a new observation with several authors describing differences in ductility depending on whether the microstructure was fully lamellar or duplex in nature [10,38]. The plasticity can also be improved if the propensity for twinning in the γ-phase can be enhanced [39]. This enhancement can be achieved if the alloying additions are able to lower the stacking fault energy of the γ-phase. In addition, the lamellar spacing, λ, has

been shown to have a large effect—the yield strength/λ relationship is of the conventional Hall–Petch form for all λ4100 nm [40]. Therefore, as λ decreases, the yield strength increases and both ductility and work hardening coefficient decrease. The presence of the β/B2 phases can either be detrimental or beneficial to room temperature ductility, and this is often related to whether or not it is the ordered B2 or disordered β that is retained after annealing. Shi, in a Ti–Al–V alloy, showed that the ordered B2 had high strength and poor room temperature ductility compared to the disordered β form, and that as the fraction of B2 increased the yield strength increased and ductility decreased [41]. In terms of hardness, it was shown that B2 was harder than γ and β was softer. The region of the stress/ductility space defined as “improved” contains five samples, three of which have a similar structure. The heat treatment temperature was 1200 1C, in these three cases, and the compositions are ∼44% Al with 1.4–2.8% Fe. The structure is a mixture of lamellar colonies and interlamellar blocky β and γ phases (and perhaps a small fraction of α2). The lamellar spacing is coarse, in excess of 500 nm in one instance and 1 mm in the others. The fraction of the disordered β-phase is between 15% and 20%. The colony fraction lies between 35% and 70%, with the highest fraction being in the sample with the finer lamellar spacing. There does not seem to be any real correlation between ductility and the amount of γ-phase present, although in the higher Fe content alloy the heat treatment which gave the lower colony fraction and higher γ-phase fraction did have “improved” ductility. The best fracture strain (0.047) was in fact with a structure containing ∼10% γ-phase (Fig. 5(b)). If the fraction of the β-phase is reduced, and the lamellar colony fraction held constant, there is a decrease in ductility without any change in fracture stress. However, in this particular structure, the remainder is not solely an increased fraction of γ-phase but a mixture of blocky γ and α2, and it is likely that the increased α2 has contributed to the decline in ductility. Alloy 3, which contains ∼44% Al but a smaller amount (0.77) of Fe, can achieve a fracture strain (0.035) which places it in the “improved” property space without containing any β-phase. The microstructure is 90% lamellar, with the remainder being blocky γ and α2 (Fig. 5(a)). The most interesting aspect of this microstructure is that the average colony contains no more than 30 lamellae, and the selection of orientation is random so that adjacent colonies do not contain parallel lamellar orientations. The small colony size and increased ductility are consistent with the observations of Chan and Kim, who noted that in a TiAl alloy containing Nb+Cr+V “tensile ductility of the lamellar microstructure increases with decreasing colony size” [27]. The reproducibility of the data for this sample was very high, but only small changes to the microstructure caused a significant decline in ductility, with the remaining heat treatments of this composition producing fracture strains between 0.01 and 0.015. The final data point in this “improved” region has a strain of 0.026 and the highest fracture strength of any composition. The microstructure appears to be 70% β/B2 and 30% α2 and γ (with γ the minority phase). Other heat treatments for this composition (Alloy 1 Ti–

C.J. Bettles et al. / Materials Science & Engineering A 575 (2013) 152–159

32.05Al–2.25Fe), all clearly contained the ordered B2 phase and did not contain any γ-phase, and the ductility is zero. The XRD results for the 1200 1C water quenched condition (the “improved” data point) are inconclusive, with no clear peaks at the 2θ positions for the ordered structure. It is therefore possible that the specimen in this condition does indeed contain disordered β. A fracture strength of 560 MPa could therefore be somewhat unexpected, but the α2 fraction is present as a relatively continuous grain boundary phase, which may improve strength. The α2 within the grains is discrete and relatively large and should not make a significant contribution to strengthening. Again, somewhat small changes in microstructure can cause significant embrittlement, and this composition would not be considered to be a candidate for overall ductility improvement. Work hardening behaviour is not generally associated with brittle materials, but in those particular microstructures which displayed the greatest ductility it is possible to make an estimate of the work hardening coefficient from the shear punch data, nτ. The approach is described by Lucas, who also showed that for values of nτ o 0.2, there is a direct relationship between nτ and n (determined from conventional tensile testing) [42]. For those alloys with effective strain values in excess of 0.03, the work hardening coefficient was found to lie between 0.075 and 0.088, which is consistent with the observations for duplex microstructures of Chan and Kim [27]. Slow cooling from 1000 1C results in all alloys containing a measureable fraction of the τ2 phase, and generally this is an undesirable outcome. At higher Fe contents the τ2 coexists in a lamellar fashion with α2 and under these circumstances slight improvements in fracture strain have been observed. At faster cooling rates, τ2 was avoided and the B2-phase formed instead. It is interesting to note that, of the five results lying in the “improved” fracture strain region, four were achieved after a WQ from 1200 1C. There were no microstructural features common to all samples, but it does appear that on fast cooling the lamellar colonies formed in alloys A4 and A5 have a small lamellar spacing, which contributes to the higher fracture strain. 5. Conclusions The effect of additions of Fe on the microstructure and shear punch response of titanium aluminide compositions with aluminium contents between 32 and 44 at% was investigated. It was found that these additions resulted in ductilities lower than those achieved in current accepted compositions for γ titanium aluminides (which are, however, of somewhat higher Al content), and that the mechanical response was highly sensitive to the microstructure. The following observations can however be noted: 1. Additions of Fe to (γ+α2) titanium aluminide compositions containing ∼44 at% Al may show some ductility with concomitant improvements in strength. The best composition range for the Fe is between 1.4 and 2.8 at%. 2. The presence of disordered β-phase is desirable to achieve higher room temperature ductility in these Fe-containing alloys. The preferred microstructure contains between 15% and 20% β, together with a fraction of lamellar (γ+α2) colonies between 30% and 70%. The remainder of the microstructure is γ-phase. 3. Ordered B2-phase, τ2 and high fractions of blocky α2 may be detrimental to ductility.

159

Acknowledgements This work was funded, in part, under a collaborative agreement with the ARC Centre of Excellence for Design in Light Metals, CAST CRC and CSIRO Light Metals Flagship. The work was also supported, in part, by the Center for the Accelerated Maturation of Materials at The Ohio State University. The Australian Research Council is gratefully acknowledged for funding of the ARC Centre of Excellence for Design in Light Metals. The CAST CRC was established under, and is supported in part by, the Australian Government's Cooperative Research Centres (CRC) Scheme.

References [1] J.L. Murray, Phase Diagrams of Binary Titanium Alloys, ASM International, Metals Park, Ohio, 1987. [2] R.M Imayev, V.M. Imayev, M. Oehring, F. Appel, Intermetallics 15 (2007) 451–460. [3] C.T. Forwood, M.A. Gibson, P.R. Miller, C.J. Rossouw, A.J. Morton, in: M.V. Nathal, R. Darolia, C.T. Liu, P.L. Martin, D.B. Miracle, R. Wagner, M. Yamaguchi (Eds.), Structural Intermetallics, TMS, Warrendale PA, 1997 pp. 545–554. [4] T. Kawabata, H. Fukai, O. Izumi, Acta Mater. 46 (6) (1998) 2185–2194. [5] US 6,231,699, Heat treatment of gamma titanium aluminide alloys, May 15 2001. [6] J.S. Kim, Y.H. Lee, Y.-W. Kim, C.S. Lee, Mater. Sci. Forum 539–543 (2007) 1531–1536. [7] H. Clemens, W. Wallgram, S. Kremmer, V. Güther, A. Otto, A. Bartels, Adv. Eng. Mater. 10 (2008) 707–713. [8] H. Clemens, H.F. Chladil, W. Wallgram, G.A. Zickler, R. Gerling, K.-D. Liss, S. Kremmere, V. Güther, W. Smarsly, Intermetallics 16 (2008) 827–833. [9] T.T. Cheng, M.H. Loretto, Acta Mater. 46 (13) (1998) 4801–4819. [10] Y.G. Li, M.H. Loretto, Acta Metall. Mater. 42 (9) (1994) 2913–2919. [11] Y. Wang, Y. Liu, G.Y. Yang, H.Z. Li, B. Tang, Trans. Nonferrous Met. Soc. China 21 (9) (2011) 215–222. [12] W.J. Zhang, L. Francesconi, E. Evangelista, G.L. Chen, Scr. Mater. 37 (5) (1997) 627–633. [13] H. Zhu, D.Y. Seo, K. Maruyama, P. Au, Philos. Mag. Lett. 85 (7) (2005) 377–385. [14] D. Hu, A.B. Godfrey, M.H. Loretto, Intermetallics 6 (1998) 413–417. [15] D.S. Shih, Y.W. Kim, Sheet rolling and performance evaluation of beta gamma (β–γ) alloys, Ti-2007 Science and Technology, in: M. Niinomi, S. Akiyama, M. Ikeda, M. Hagiwara K. Maruyama(Eds.), Proceedings of the 11th World Conference on Titanium, Kyoto, 3–7 June 2007, pp. 1021–1024. [16] J.S. Kim, Y.H. Lee, Y.W. Kim, C.S. Lee, Mater. Sci. Forum 539–543 (2007) 1531–1536. [17] T.T. Cheng, M.H. Loretto, Acta Mater. 46913 (1998) 4801–4819. [18] Z.W. Huang, Acta Mater. 56 (2008) 1689–1700. [19] B.J. Inkson, C.B. Boothroyd, C.J. Humphreys, Acta Metall. Mater. 41910 (1993) 2867–2876. [20] M. Palm, G. Inden, N. Thomas, J. Phase Equil. 16 (3) (1995) 209–222. [21] L. Levin, A. Tokar, M. Talianker, E. Evangelista, Intermetallics 7 (1999) 1317–1322. [22] V. Raghavan, J. Phase Equil. 23 (4) (2002) 367–374. [23] R. Kainuma, Y. Fujita, H. Mitsui, I. Ohnuma, K. Ishida, Intermetallics 8 (2000) 855–867. [24] R. Kainuma, I. Ohnuma, K. Ishikawa, K. Ishida, Intermetallics 8 (2000) 869–875. [25] M. Palm, J. Lacaze, Intermetallics 14 (2006) 1291–1303. [26] P.Prasad Rao, K. Tangri, Mater. Sci. Eng. A 132 (1991) 49–59. [27] K.S. Chan, Y.-W. Kim, Metall. Trans A 23A (1992) 1663–1677. [28] F. Appel, U. Sparka, R. Wagner, Intermetallics 7 (1999) 325–334. [29] V. Seetharaman, S.L. Semiatin, Mater. Sci. Eng. A 299 (2001) 195–209. [30] G.B. Viswanathan, M.J. Mills, V.K. Vasudevan, Metall. Mater. Trans. A 34A (2003) 2113–2127. [31] K.S. Chan, Metall. Trans. A 24 (1993) 569583. [32] K.S. Chan, Metall. Mater. Trans. A 25A (1994) 299–308. [33] Y. Chen, H. Niu, F. Kong, S. Xiao, Intermetallics 19 (2011) 1405–1410. [34] A. Gorzel, M. Palm, G. Sauthoff, Z. Metallkd. 90 (1999) 64–70. [35] G.E. Lucas, J. Nucl. Mater. 117 (1983) 327. [36] R. Ducher, B. Viguier, J. Lacaze, Scr. Mater. 47 (2002) 307–313. [37] M.A. Morris, Y.G. Li, M. Leboeuf, Scr. Metall. Mater. 31 (4) (1994) 449–454. [38] V.M. Imaev, R.M. Imaev, T.I. Oleneva, T.G. Khismatullin, Phys. Met. Metallogr. 106 (6) (2008) 641–648. [39] F. Appel, M. Oehring, R. Wagner, Intermetallics 8 (2000) 1283–1312. [40] K. Maruyama, G. Suzuki, Hee Y. Kim, M. Suzuki, H. Sato, Mater. Sci. Eng. A 329–331 (2002) 190–195. [41] J.-D. Shi, Z. Pu, Z. Zhong, D. Zou, Scr. Metall. Mater. 27 (1992) 1331–1336. [42] G.E. Lucas, G.R. Odette, J.W. Sheckherd, ASTM STP 888 (1986) 112–140.