Abstract
In this study, hot workability of Ti–35V–15Cr–0.3Si–0.1C alloy is investigated in the temperature range of 900–1150°C and strain rate range of 0.01–10 s− 1 using processing maps. In the maps, the stability domain displays the feature of dynamic recrystallisation and dynamic recovery. The fraction of recrystallisation in this alloy is much larger than that in other β-Ti alloys, which can be attributed to the effect of titanium carbides by a ‘particle stimulated nucleation’ mechanism. Higher temperature and lower strain rate are helpful for breaking down the carbides and improving the workability. However, deforming >1100°C would result in grain coarsening. In instability domain, the occurrence of flow localisation, shear bands and cracking is discussed. The optimised processing window is in 1050–1100°C /0.01–0.1 s− 1.
Introduction
Titanium and titanium alloys are excellent candidates for aerospace applications owing to their high strength/weight ratio and excellent corrosion resistance. 1 However, titanium and titanium alloys have a fatal weakness of their combustion sensitivity, which has been well documented.2–3 Thin sections of conventional titanium alloys will burn in air when ignited by a localised heat source such as friction heating due to a heavy rub. Developing the burn resistant titanium alloys will extend their application to more components of aero engines, thus displacing steels and nickel alloys of almost twice the density. 4 For this reason, in the last decades, more and more attention has been paid to the burn resistant titanium alloys in different countries, especially to the Ti–V–Cr series burn resistant titanium alloys.5–6 It's worth mentioning that Pratt and Whitney 5 developed a burn resistant alloy based on Ti–35V–15Cr in 1985, designated as alloy C, which has high burn resistance up to 600°C. Alloy C has been successfully used in F119 engine that powers the F22 jet fighter. 7
In the present research, Ti–35V–15Cr–0.3Si–0.1C is a β stabilised Ti–V–Cr series burn resistant titanium alloy. The addition of carbon element brings significant change to the β titanium alloys. It has been found that appropriate carbon addition improves the thermal stability and ductility of burn resistant β titanium alloys because of the formation of TiCx, which acts as an oxygen getter [forming Ti(CO) and removing the oxygen from the β matrix]. 8 Sun and Lavernia made a comparison between the creep behaviour of the Ti–35V–15Cr and Ti–35V–15Cr–0.2C alloys, concluding that adding C to the burn resistant titanium alloys leads to greater creep resistance, which is attributed to the second particle strengthening of carbide particles by inhibiting the dislocation motion in the matrix. 9 In industrial practice, however, as a unique β titanium alloy, Ti–35V–15Cr–0.3Si–0.1C is very difficult to process due to severe surface or edge cracking. The poor workability of Ti–V–Cr series alloys is also reported in the study on alloy C (Ti–35V–15Cr) 10 and Ti40 (Ti–25V–15Cr–0.3Si). 11 Although several technical methods such as canned forging in Ti–25V–15Cr–0.3Si 11 or plasma sprayed coating in alloy C 10 have been reported to solve this problem successfully, no reports on hot workability of Ti–35V–15Cr–0.3Si–0.1C alloy have been published. So, investigation on the workability of this alloy is needed in order to optimise the processing parameters and obtain the desired microstructure.
Experimental
The alloy used in the present work with the nominal chemical composition of Ti–35V–15Cr–0.3Si–0.1C (wt-) was provided by Western Superconducting Technologies Co., Ltd. Ti–35V–15Cr–0.3Si–0.1C is a highly stabilised β titanium alloy since it has the highest molybdenum equivalency (47.5 wt-) of all commercial β titanium alloys according to the ‘moly equivalent’ equation in Ref. 12, which is used to indicate the β stability of titanium alloys. The optical microstructure of the received material is shown in Fig. 1. As can be seen, its microstructure consists of coarse β grains and some titanium carbides in β matrix, showing a quite different microstructure with traditional two-phase titanium alloys. The as received material was applied a small deformation (∼20) along the radial direction of the ingot before it was provided to us.
Microstructure of as received Ti–35V–15Cr–0.3Si–0.1C
In order to ensure consistency and eliminate the effect of drawing, all the specimens for compression test were cut in a homogeneous macrosection after close examination. The specimens were than machined into cylinder 10 mm in diameter and 15 mm in height according to the standard method for hot compression test. A series of isothermal compression tests were conducted on a computer controlled Gleeble-1500 thermal simulator in the deformation temperature range of 900–1150°C with 50°C intervals, strain rate of 0.01, 0.1, 1 and 10 s− 1 and the height reduction in 10–70 with an interval of 20. A special graphite based lubricant was coated on both ends of the specimen, and a 0.1 mm tantalum foil was placed between die surface and specimen to prevent cementation and minimise friction. Samples were heated to the corresponding testing temperatures with the rate of 10°C s− 1 and kept for 5 min before compression so as to obtain a uniform deformation temperature. The specimens were cooled in the air after hot compression. The true stress–strain data were recorded automatically in the compression process. Deformed specimens were then sectioned parallel to the compression axis and prepared for metallographic examination using standard procedures. Metallographic specimens after grinding, polishing and etching were observed with Olympus/PMG3 optical microscope and JSM-6390 scanning electron microscope (SEM). Some of the specimens were also examined by JEM-200CX transmission electron microscope (TEM). Transmission electron microscopy foils were prepared in a twin jet electropolishing devise. The foils were of low quality because the coarse carbides can easily fall off. Other methods were also tried, but proved not much helpful.
Results and discussions
Flow stress analyses
The flow stress curves of Ti–35V–15Cr–0.3Si–0.1C in the deformation temperature range of 900–1150°C for different strain rates (0.01–10 s− 1) are shown in Fig. 2. Like most other titanium alloys, the flow stress increased significantly with decreasing temperature and increasing strain rate, indicating the sensitivity of flow stress to the variations of deformation temperature and strain rate. Especially, Ti–35V–15Cr–0.3Si–0.1C alloy exhibits higher stress level compared with TC21 (α+β), 13 Ti60 (near α) 14 and Ti40 (β) 15 at the same deformation conditions, which can be attributed to the extremely high alloying additions and their effect of solution strengthening on the β titanium matrix. In addition, continuous flow softening is observed at almost all temperatures and strain rates used in the present work, and the softening tendency is greater at lower temperatures and higher strain rates. In compression of coarse grain CP–Ti (single alpha phase), curves show an opposite trend, which is a continuous flow hardening behaviour. 16 The flow softening behaviour in β alloys is believed to be associated with temperature rise, dynamic recrystallisation, spheroidisation or flow instability duo to localisation, cracking, etc.17,18 In many titanium alloys, according to the study in the authors’ group, temperature rise plays an important role in flow softening during compression, especially at higher strain rate. However, it is not the only cause of softening; microstructural change also leads to softening. In this study, it can be found in the following section that microstructural change including dynamic recovery (DRV), dynamic recrystallisation, the breaking down of carbides and flow localisation occur in different deforming conditions. In addition, it is interesting to note from the curves that at strain rates >0.1 s− 1, there is a distinct peak in the flow stress in the early stages of deformation followed by steady state at higher strains, showing a discontinuous yielding behaviour. Such discontinuous yielding behaviour is observed in many titanium alloys and has been attributed to the generation of mobile dislocations from the grain boundaries.19,20 Another noticeable feature in Fig. 2 is that the curves at high strain rate of 10 s− 1 show very obvious oscillation, which are suggestive of instabilities, such as serious flow localisation or cracking. 15 To conclude, the shape of stress–strain curves indicates some features that help with identifying the possible mechanisms of hot deformation. However, detailed microstructural examinations are still required in order to confirm the exact mechanisms.
Stress–strain curves of Ti–35V–15Cr–0.3Si–0.1C in different conditions
Processing maps
The approach of processing maps, used to model the constitutive behaviour of the material, was developed by Y. V. R. K. Prasad on the basis of the principles of dynamic materials model.21,22 The total input power on the workpiece is dissipated by two means: (i) a temperature rise and (ii) a microstructural change. The efficiency of power dissipation through microstructural change processes is given by
. The variation of the efficiency η with temperature and strain rate constitutes a power dissipation map exhibiting different domains, which may be directly related with specific microstructural mechanisms.
A continuum instability criterion based on the extremum principles of irreversible thermodynamics as applied to large plastic flow is used in Prasad's model to identify the regimes of flow instabilities.
21
Flow instabilities may occur when the following equation
23
is satisfied
By applying the stress–strain data to compute the parameter m, η and ξ, two processing maps obtained for Ti–35V–15Cr–0.3Si–0.1C at the strain of 0.4 and 0.7 are plotted using MATLAB software, as shown in Fig. 3a and b respectively. The contour numbers in the figures are the efficiency of power dissipation (η), and the shaded regions enclosed by heavy line represent the instability regions, in which the parameter ξ gets negative values. Overall, the efficiency of power dissipation takes higher values at the lower part of the maps, and presents lower values at the upper part, indicating that when material deformed at a lower strain rate, more input power is dissipated by microstructural transformation, while at higher strain rate, most of the input energy is dissipated by other means. Apparently, the former way is more desirable. Specifically, as can be seen from the Fig. 3a, power dissipation efficiency reaches peak value of 72 at 1060°C /0.01 s− 1, and in Fig. 3b, peak efficiency is ∼64 at 1050°C/0.01 s− 1.
Processing maps obtained for Ti–35V–15Cr–0.3Si–0.1C at the strain of a 0.4 and b 0.7
The instability region occurs in the region of high strain rates (the shaded regions in the maps). With increasing strain from 0.4 to 0.7, the area of instability region increases, indicating that the processing window becomes narrow when increasing strain, and this is consistent with the industrial practice. Moreover, it is noticeable that in both maps, the processing window is relatively wider at higher deformation temperatures, which can be explained by the better workability of material at higher temperatures than that at lower temperatures.
In order to investigate the underlying deformation mechanisms and verify the reliability of process parameters predicted by processing map, the evidence of deformation in these domains are identified and validated through microstructure observations in the following sections.
Stability domain
As shown in Fig. 3, stability regions in both maps are at the low strain rate area. Peak power dissipation efficiencies in both maps occur at strain rate of 0.01 s− 1 and temperatures ∼1050°C, indicating a favourable workability in this region. For more details, as can be seen from the maps, domains #1 in Fig. 3a and #2 in Fig. 3b in the ranges of 925–1100°C and 0.01–0.03 s− 1 are drawn by a red line to represent the highest power dissipation efficiency area. The power dissipation efficiency values in domains #1 and #2 are higher than ∼46. To determinate the deformation mechanisms in two domains of peak efficiency, the microstructure of the specimen deformed at 950°C, 1050°C, 1150°C/0.01 s− 1 is examined, as shown in Fig. 4a–c respectively. The microstructure exhibits a characteristic of dynamic recrystallisation (DRX). It is noted that recrystallised grains prefer to occur in the vicinity of TiCx. In addition, both the fraction of the recrystalllised β grain and the grain size increase as deformation temperature increased. The corresponding recrystalllised grain size is ∼7 μm (Fig. 4a), ∼15 μm (Fig. 4a) and ∼40 μm (Fig. 4c) at temperatures of 950, 1050 and 1150°C and strain rate of 0.01 s− 1 respectively, indicating that deforming at temperature >1100°C would result in quick grain coarsening. In previous reports, it has been widely recognised that high power dissipation efficiency is associated with DRX.21,24,25 However, less DRX can be observed in the middle strain rate area of the stability domain. Transmission electron microscopy observation (Fig. 5) finds dislocation walls and subgrains in the microstructure, suggesting that (DRV) occurs widely in this area.
a 950°C; b 1050°C; c1150°CTypical microstructures at different deforming temperatures of Ti–35V–15Cr–0.3Si–0.1C
Transmission electron microscopy microstructure at 1050°C/1 s− 1 of Ti–35V–15Cr–0.3Si–0.1C
Nevertheless, it seems the large fraction of DRX in the present β alloy is still an abnormal phenomenon. Previous studies suggest that DRV is the controlling mechanism of microstructural evolution during hot deformation in β alloys, 26 although DRX of β phase has also been reported. 15 A continuous DRX mechanism has been found responsible for extensive grain refinement and good workability in β titanium alloys. 27 The conventional discontinuous DRX can only be found at original β grain boundaries. 28 As in the case of Ti40 (a similar β alloy: Ti–25V–15Cr–0.2Si), 15 DRV is so fast that strain energy cannot be accumulated to a critical level for nucleation of DRX in many conditions. ‘necklace’ recrystallised fine grains can only be found along the original grain boundaries, and the fraction of DRX is therefore much less than that in the present alloy.
The key for investigating the pervasive DRX in the present alloy is on the understanding of the role of titanium carbides on DRX. Adding C into Ti–V–Cr alloy leads to a special two-phase microstructure that is unlike the conventional two-phase (α+β) alloys. Rather than α phase in conventional titanium alloys, the second phase in this alloy is titanium carbides, much harder than the α phase in α+β titanium alloys. In addition, the eutectic carbides can be broken down into particles by hot working. In metallic alloys, recrystallisation is frequently connected with second phase particles, as such particles have a strong effect on the recrystallisation kinetics and microstructure through a so called ‘particle stimulated nucleation (PSN)’ mechanism. 29 During the deformation of a particle containing alloy, the enforced strain gradient in the vicinity of a non-deformable particle creates a region of high dislocation density and large orientation gradient (particle deformation zone), which is an ideal site for the development of a recrystallisation nucleus. 30
In Fig. 4a, optical microscopy pictures of Ti–35V–15Cr–0.3Si–0.1C clearly reveal that new recrystallised grains prefer to nucleate near the carbides. Further works by the SEM and TEM (transmission electron microscopy) method confirm the relation between carbides and recrystallisation. An SEM picture (Fig. 6a) shows that new recrystallised fine grains with a diameter of ∼8 μm form in the vicinity of discontinuous carbides. Figure 6b exhibits on a more microlevel that a spindle shaped new grain, which is 1 μm wide and 4 μm long, originates at the carbides. Therefore, it can be concluded that PSN mechanism is responsible for the pervasive discontinuous recrystallisation within β grains in the present alloy and the existence of carbides in β matrix of this alloy accelerates the process of recrystallisation during hot deformation.
a graph (SEM) shows that new recrystallized fine grains form in vicinity of discontinuous carbides; b TEM graph shows that spindle shaped recrystallized grain formed at carbidesCarbides versus recrystallization
It is of great significance to understand the effects of titanium carbides on recrystallisation. In the present study, the material is a highly stabilised β titanium alloy, in which no obvious phase transformation occurs during deformation and subsequent cooling. So recrystallisation might be the only method to produce a completely new grain structure with a modified grain size, shape, and texture that will further improve the workability. By the controlling of material's processing, it is possible to control the distribution of second phase particles in the microstructure relatively accurately. Namely, if carbides can be broken down to small size and distribute more uniformly, it is possible to get a fine homogeneous recrystallised microstructure. Thus, it provides us an approach to use second particles as a method of controlling the grain size and microstructures during thermomechanical processing.
Instability domain
A large instability domain is predicted in the region of high strain rates, as shown in the shaded area of the maps, indicating a poor workability of the present alloy. Instability domain has a low level of power dissipation efficiency, which is < 40. As can be seen from the processing maps, instability domain expands as strain increases or deformation temperature decreases. The instability mechanism is found to associate with the occurrence of flow localisation, adiabatic shear band and surface cracking.
Figure 7 shows the deformed specimen of Ti–35V–15Cr–0.3Si–0.1C alloy after deforming at 900°C and with a strain rate of 10 s− 1. It can be seen that a surface crack occurs at an angle of 45° to the compressive axis (along the maximum shear stress plane), indicating this is a shearing deformation mechanism. From the microstructure, an adiabatic shear band with a width of ∼500 μm can be clearly seen in the middle area. Figure 8 shows the microstructure of specimen deformed at 1000°C/1 s− 1, exhibiting a feature of flow localisation. Strain is localised at the centre zone of the cross-section of the specimen with a width of < 1 mm. Flow localisation has been believed to associate with local temperature rise at higher strain rates (higher than 1 s− 1). During deformation, lots of deformation heat generate and dissipate. However, at higher strain rates, the heat has no time to dissipate due to short deforming time (at 10 s− 1, the total deforming time for a 15 mm tall specimen is < 0.1 s) and low thermal conductivity of titanium alloy, so local material gets softened immediately and flow localisation occurs. Generally, the formation of adiabatic shear band or severe flow localisation can result in flow stress softening dramatically (see the strain–stress curves) and thus lead to low efficiency of power dissipation.
Deformed specimen and its microstructure after deforming at 900°C and with strain rate of 10 s− 1
Microstructure of specimen deformed at 1000°C/1 s− 1
Another feature of the instability domain is surface cracking. Besides the intrinsic factors of the material (low ductility), the photographs of the cross-section show that the cracks occur at the end of shear band or flow localisation band, suggesting that local large strain gives rise to surface cracking. Further examination of the deformed samples reveals that cracking is relevant to the coarse titanium carbides. As shown in Fig. 9a, the distribution of the cracks on the surface is consistent with the carbide distribution. Metallographic examination on the cross-section of the samples further proves that the crack is originated at the carbides (Fig. 9b). In fact, most industrial alloys contain a second phase. If the second phase is hard and brittle, and in worst case distributed in the shape of continuous course network, it will deteriorate hot workability and toughness of the material. A common example is the eutectic carbide in high speed steel. 31 During working, both matrix metal and the second phase deform under the shearing force. However, the strain is inhomogeneous between matrix and second phase because of their different properties. An inner stress field therefore develops and increases with strain increases until a crack originates.
a graph (SEM) on sample surface shows that cracks distribute along long lamellas carbides; b OM graph on cross-section of sample shows crack is beginning at carbidesCracking versus carbides
Processing windows
For industrial productions like cogging and forging, the determination of processing window is a key technical issue since desired microstructure could be achieved only if the deforming conditions are carefully controlled. Therefore, two levels of issues should be concerned. First of all, the deformed billet or forging should not have defects (like deep cracks) that will affect the integrity of the finished part. Next is the microstructural aspect, like the deforming uniformity, grain refining, etc. For the present alloy, as mentioned above, strain rate plays a most significant role in cracking and deformation non-uniformity during processing. In detail, with the increase in strain rate, deformation non-uniformity increases. When samples are deformed at strain rate >1 s− 1, flow localisation occurs and deformation is localised in the middle area of samples. In some cases (low temperature or large reduction), severe localised deformation might lead to surface cracking. Deforming at low strain rate is not only good for uniform deformation and reducing cracking, but also good for microstructures. From the perspective of facilitating DRX, it can be concluded that lower strain rate, higher temperature (low Z) and larger reduction (high strain) are favourable, as they provide optimum kinetic conditions for DRX. However, it should be noted that too high temperatures would result in grain coarsening quickly, especially >1100°C. A processing mechanism map can thus be schematically plotted, shown in Fig. 10. It can be concluded that the present alloy is difficult to process because of its narrow processing window. Instability domain must be avoided when processing since flow localisation or cracking might occur. In stability zone, Ti–35V–15Cr–0.3Si–0.1C gets relatively uniform deformation and DRX takes place in this zone. This is very helpful to refine microstructure and improve mechanical properties of the material.
Processing mechanism map of Ti–35V–15Cr–0.3Si–0.1C under investigated conditions
Based on the above analysis, the applicable processing window should be carefully controlled in the range of 1050–1100°C/0.01–0.1 s− 1 in order to get a desirable microstructure of fine grain with small uniformly distributed carbides. Canned forging is thus strongly recommended in order to increase forging efficiency and reduce cracking. In practice, a Ti–35V–15Cr–0.3Si–0.1C slab has been successfully forged using the optimised parameters. Figure 11 shows the finished forging slab and its fine microstructure with dispersed carbide particles and fully DRX.
Finished forging slab of Ti–35V–15Cr–0.3Si–0.1C and its fine microstructure with dispersed carbide particles and fully DRX
Conclusions
Compression of Ti–35V–15Cr–0.3Si–0.1C exhibits higher stress level than conventional titanium alloys, and the flow stress is sensitive to the variation of deformation temperature and strain rate. Deformation in the stability domain is beneficial for DRX and DRV, while in the instability domain, flow localisation, shear bands or surface cracking would occur. Higher temperature and lower strain rate are helpful for breaking down the coarse carbides and promoting recrystallisation. However, deforming at temperatures >1100°C would result in grain coarsening quickly. The pervasive recrystallisation of the β phase in Ti–35V–15Cr–0.3Si–0.1C is attributed to the effect of TiCx and its ‘PSN’ mechanism. Particle deformation zone acts as prior nucleation sites during deformation. Breaking down the coarse TiCx dendrites into particles is in favour of the homogeneous nucleation of the recrystallised grains, which can in turn improve the workability. The processing window should be carefully controlled in the range of 1050–1100°C/0.01–0.1 s− 1.
Footnotes
Acknowledgements
This work was supported by the National Natural Science Foundation of China with grant no. 51075333.