Analysis of the development of abnormal grains structures during Beta annealing of Ti-64 wrought products

The β -annealing of Titanium-6Al-4V (Ti64) wrought aerospace components can lead to the development of abnormal grain structures (AGS) that jeopardise material performance. Therefore, developing an in-depth understanding into the origins of AGS will help in the design of processing routes that can avoid the conditions that lead to their development. This research demonstrate the ap-plication of novel concurrent in-situ heating and electron back scatter diﬀraction (EBSD) techniques to help elucidate possible mechanisms for the development of AGS. It was found that primary- α ( α p ) may play a key role, acting as a second phase particle, in pinning the β -phase grain boundaries during recrystallisation. The strengthening of a large area cube component texture macrozone, consisting of predominantly low angle grain boundaries, also helps in the development of AGS.


Introduction
The microstructure of the alloy titanium 6 aluminium 4 vanadium (Ti-64) can be tailored through thermomechanical processing, enabling its use in different aerospace applications. For example, beta annealing is used to produce a microstructure with high crack growth resistance which is desired in many structural aerospace applications. The β-annealing of Ti-64 components is carefully controlled to produce the required microstructure parameters. Despite this, in some instances, a transformed-β microstructure consisting of a coarse bimodal (and therefore abnormal) grain size distribution, referred to as an "abnormal grain structure" (AGS), appears in forgings with the abnormally large grains developing at the mid-section. This microstructure is unfit for purpose and therefore rejected. The development of AGSs is sometimes referred to as arising from abnormal grain growth [1] however, more recent research has demonstrated that the phenomena may be due to recrystallisation instead [2] and therefore "abnormal grain growth" is neither suitable nor descriptive of the phenomenon at play.
In this paper we present the results of a set of novel in situ experiments designed to elucidate the conditions and mechanisms that lead to AGSs. This has involved the use of fast electron backscatter diffraction (EBSD) in a scanning electron microscope (SEM) equipped with a high temperature heating stage to study in situ the microstructure and texture evolution in a representative β-annealed forging.

Material studied
A Ti-64 forged rectangular section from a starting 250 mm 2 diameter billet known to contain AGSs was supplied by Airbus. The forging was sliced in half along the draw direction (DD) and one half subjected to an industry standard β-anneal at T β + 30 • C for 1 hour and air cooled, whilst the other half was retained in the as-forged condition. The specimen directions are defined as follows: DD = draw direction, FD = principle forging direction (maximum strain) and TD = transverse forging direction (minimum strain). 1

Forged material -AGSβ-annealed microstructure
An area of the β-annealed specimen that contained AGS was sectioned from the parent block into two specimens, one prepared for optical microscopy and the other prepared for EBSD microscopy by grinding to 4000 grit followed by 20 minutes of polishing, using a mixture of OPS/H 2 O 2 with a 4:1 ratio. Etching of the optical specimen was conducted using Kroll's reagent. The DD/FD plane of the abnormal region was mapped by EBSD, over a 25 x 4 mm area using a voltage of 20kV, step size of 10 µm and a working distance of 20 mm. The corresponding "as-forged" section (non heat treated) that married up to the location of the β-annealed section that exhibited AGS, was then selected for use in the in situ experiment.
The EBSD data was analysed using the MTEX software package [3]. Reconstruction of the prior β-grain texture was achieved using software developed by Davies et al. [5], [6] at the University of Sheffield.

2.3
In situ EBSD mapping during β annealing A 2 mm 3 cube of the as forged material was removed from the centre of the section, with respect to the forging directions. The DD/TD face was then prepared for EBSD analysis using the same grinding and polishing regime mentioned. The specimen was then cemented onto the centre of a heating chip using carbon based cement, and the cement cured for 2 hours at 25 • C, followed by a further 2 hours at 93 • C and a final 2 hours at 260 • C to ensure good electrical conductivity. A similar sized specimen of pure zirconium was also prepared and mounted to the heating chip along side the Ti64 sample to act as an "oxygen-getter", in order to reduce the amount of latent oxygen in the microscope chamber that could absorb into the Ti64 sample during heating and increase the β-transus temperature (T β ).
In situ, the specimen was pre-heated to 300 • C to ensure efficient heating and contact of the thermocouple with the heating stage. After the pre-heat test, the system was left under vacuum at ambient temperature for approximately 16 hours to help remove latent oxygen. The sample was then heated to progressively higher temperatures, within the α + β phase regime, and iso-thermally held in order to carry out EBSD mapping. The EBSD scans were conducted at 20 kV, 70 • tilt and at a working distance of 20 mm. The scan size was approximately 1 mm 2 with a step size of 1 µm for scans below 970 • C and 3 µm for scans above 970 • C.

Optical microscopy and EBSD mapping β-annealed forging -AGS
Optical microscopy of the β-annealed forged material that contains a typical AGS region is shown in figure 1. The micrograph depicts the transformed-β microstructure with homogeneous grains at the FD surfaces. At the mid-section of the sample, the unambiguous presence of prior β-grain boundaries become difficult to observe. However, within this region, a large prior-β island grain of approximately 6.5 mm in length can be seen (red arrow). This type of microstructure, with a bimodal grainsize distribution, is typical of AGS. A black box indicates the representative area of interest that was mapped by EBSD ( figure   2). Initial EBSD analysis of the as-received β-annealed forging, given in figure 1, revealed the presence of very large abnormal prior β-grains of about 0.5 cm in size in the centre of the forging, that transform on cooling to a typical bimodal β-transformed microstructure with thick primary-α (α p ) colonies at their grain boundaries and fine α-lamella plates both within the prior β-grains and in the surrounding matrix.
Reconstruction of the β-phase from the orientations of the α-lamella colonies revealed that the ab-

As forged microstructure
A back-scatter SEM image of the centre of the forged starting material is shown in figure 3 depicting a bimodal microstructure with regions of α p and transformed-β, giving secondary α (α s ) with some retained β.

In situ β-annealing
The region studied during the in situ heating experiment was obtained from the centre of the as-forged block which was selected to marry-up to the location in the β-annealed section that exhibited AGS. The starting textures of the α and β phases are shown in figure 4 (using EBSD maps) with corresponding pole figures obtained at 800 • C. It can be seen that this region has a relatively strong α-texture with a The progression of the phase transformation tracked by in situ heating between 800 • C and 977 • C is shown in figure 5. At 800 • C, the α-phase had a volume fraction of about 90%, existing as both α p and α s . However, it was difficult to distinguish between primary and secondary alpha using only the EBSD map. Although the majority of the map area has the cube orientation, a few non-cube oriented β grains can also be seen, some of which appear to be associated to the α-phase texture band highlighted in figure   4 (red arrow). At 940 • C the α-phase volume fraction reduced to 16%. Only large surviving α-grains (approximately 50 µm) can be seen in both the phase map and α-phase IPF map that correspond to α p . The phase map in figure 5b and the IPF map in figure 5h show that the orientations of the newly transformed-β are those of the β grains that were present at 800 • C (figure 5g). Therefore, whilst the dominant texture is the cube texture, as before, there are also some highly misorientated grains that have grown from the pre-existing misorientated β-phase nuclei that were seen at 800 • C.
At 977 • C, the α-phase reduced to a volume fraction of only 2% (figure 5c), which was somewhat unevenly distributed. The highly misorientated β-grains from within the α-phase texture band were, however, no longer present, having been replaced by newly recrystallised grains that share a similar orientation to that of the cube component.
At 999 • C, the α → β phase transformation is complete, as seen in figure 6. The entire mapped surface now contains only β-phase subgrains whose orientations lie within 10 • of the ideal cube texture component. The (100) pole figure demonstrates an extremely sharp and strong cube texture (max = 45 m.r.d.) whilst the misorientation histogram of figure 6b indicates all the subgrain boundary misorientations were lower than 10 • , with over 60 % lower than 5 • .

Discussion
Initial investigation into the AGSs in a β-annealed forged section by EBSD microscopy and reconstruction of the super-transus β-phase texture, unambiguously identified the presence of large abnormal grains with high angle grain boundary misorientations with respect to a surrounding matrix, composed of cube-aligned subgrains. They are therefore similar in nature to the coarse grains found in rolled plate [2] and probably grow in a similar manner, via recrystallisation. However, unlike in rolled plate, the number of grains that grow successfully was much lower in the forging, leading to only a few, but very large abnormal grains.
This implies that conditions for this form of recrystallisation to occur happen only rarely in the forging studied here.
By annealing a specimen in-situ, that exhibited a prior β-phase consisting of a strong cube orientation, has demonstrated that the final texture that developed during transition through the α + β phase field grows directly from the pre-existing β-phase orientations. This highlights the importance of controlling the β-phase texture during the prior deformation steps.
The small area studied showed that there were several non-cube oriented β-grains in the as forged material that could have become new coarse grains after β-annealing. However, the experiment also showed that when the size of these misorientated grains is similar to that of the cube oriented subgrains, the subgrains grow to consume the misorientated grains. This is consistent with a thermodynamic viewpoint, since high-angle grain boundaries have higher energy and therefore such grains will disappear on annealing, everything else being equal. The situation is further complicated here however, since the remnant α p acts to pin grain boundaries.
The dissolution of α-phase occurs in a two-stage process. Initially, the finer lamellar α s transforms to β regrowing from the retained β orientation, with respect to the Burgers relation, leaving α p to act as second phase pinning particles in a classical 'Zener' sense. Figure 4c demonstrates this effect during the recrystallisation process, as regions pinned by less α p particles exhibit β-phase subgrains of larger size than those β-subgrains that are surrounded by α p at their boundaries. As a grain of high boundary mobility would need to reach a critical size with respect to its neighbours in order to enter a stage of discontinuous growth [7], as seen in AGSs, this may be a candidate mechanism that is capable of enabling the development of such microstructures in Ti-64. Further work is therefore needed to establish the dissolution kinetics of α p with respect to the α-phase texture bands that develop during α + β phase deformation.
Texture banding in the α-phase (figure 5d) may also play a direct role in the development of misorientated β-grains that become embedded within the cube texture component. By comparing figures 5d and e, there is evidence of a relationship in the location of α and β-phase texture bands that suggests some form of mechanism during concurrent deformation of the two phases is at play that produces highly misorientated grains within the subgrain matrix. 7

Conclusions
Large area EBSD mapping of typical AGS and in situ heating with concurrent EBSD mapping of ti64 forgings has shown: • Large abnormal β-grains, of high boundary mobility, develop through uninhibited growth into a matrix of low angle subgrains of a predominantly cube texture.
• Heterogeneous dissolution of α p may play an important role in determining the recrystallisation behaviour of the β-phase during heating through the α + β-phase window.
• Regions of texture banding within the α-phase could provide the misorientated β-grains that act as nuclei for highly misorientated abnormal β-grain growth.
• It is still unclear what the exact conditions that are needed for a misorientated grain to grow in a such strongly texture region. We are planning to carry out computational simulations of this mechanism to study the competing effects of grain boundary pinning, misorientation and sub-grain size.