15th World Conference on Titanium (Ti-2023) THE NATURE, EVOLUTION AND EFFECT OF THE OMEGA PHASE IN Ti-15Mo (wt.%) Nicholas Jones1, Nicole Church1, Christian Talbot1, Joe Bennett1, Lewis Owen2 & Howard Stone1 1 Department of Materials Science and Metallurgy, University of Cambridge, 27 Charles Babbage Road, Cambridge, CB3 0FS, UK 2 Department of Material Science and Engineering, The University of Sheffield, Mappin Street, Sheffield, S1 3JD, UK. Metastable β titanium alloys are highly attractive materials for a number of different industrial sectors due to the broad range of properties that can be obtained. Typically, processing these alloys requires the retention of the β phase in a metastable state, which is commonly achieved by rapid cooling from a high temperature solution heat treatment. However, this processing route also tends to form the ω phase, which can have a pronounced influence on the behaviour of the material. Here, through the use of in situ synchrotron diffraction, the evolution and influence of the ω phase has been studied. The results indicate a change in the mechanical properties of the alloy does coincide with the formation of isothermal omega but relates to both the fraction of the omega present and the evolution of the β phase. Keywords: Ti-Mo; Omega; Mechanical Properties; Synchrotron Radiation 1. Introduction Metastable β alloys offer the broadest range of properties, from low modulus alloys utilised in biomedical devices to high strength alloys used in the landing gear assemblies of the largest airframes. In many cases, their ability to retain the β phase in a metastable form through rapid cooling from high temperatures is exploited to achieve the final desired properties. Yet, such processing also leads to the occurrence of the ω phase, which is widely regarded as detrimental to alloy performance. Throughout the literature, two forms of ω are generally reported; 1) a compositionally indistinct athermal form (ωath) that develops on cooling; and, 2) a compositionally distinct isothermal form that evolves through diffusional processes during low temperature (200 - 450˚C) heat treatments [1]. Despite the importance of this class of alloys, there remains significant debate around the nature and influence of the ω phase, along with the mechanisms that lead to its formation [2-5]. The presence of ω is often purported to result in an embrittled material. However, this statement is imprecise as alloys that clearly contain ωath following rapid cooling from the β phase field often exhibit extensive plastic regions [6, 7] or undergo diffusionless shear transformations [8, 9]. In reality, it is the ωiso that affects mechanical properties. Yet even when solely considering reports relating to the ωiso phase, there are substantial discrepancies between different studies. For example, Mantri et al. [6] report that the change from ωath to ωiso in a Ti-12.5Mo (wt.%) following a 48 h heat treatment at 475˚C caused the dominant deformation mechanism of the β phase to change from twinning to slip. This was rationalised in terms of a substantial increase in the stability of the β phase as it became enriched in Mo. Similarly, Lai et al. [10] also reported a change in mechanical properties in Ti-12Mo (wt.%) following heat treatment. However, they observed a complete loss of ductility following only 10 minutes at 400˚C. In contrast, very short heat treatments of 1 minute at temperatures below 250˚C have been shown to dramatically increase the yield strength of Ti-12Mo (wt.%), whilst retaining an ability to undergo > 50% plastic strain [7]. However, ductility was substantially reduced when heat treating at higher temperatures. Whilst all of these studies attribute the change in mechanical behaviour to Mo enrichment of the β phase, there remains a lack of clarity over the precise underlying mechanism. In addition, it is notable that there are marked variations in key parameters between the studies, such as heat treatment temperatures and time. Such variations, and those found in other related studies, would likely lead to significant differences in the size and volume fraction of ωiso precipitates. The consequence of this on the macroscopic alloy properties still requires full rationalisation. As such, further studies are needed to elucidate the influence of, and mechanisms by which, the ω phase affects the behaviour of metastable β alloys. Currently, the transition from the β + ωath microstructure, often present following rapid cooling from a solution heat treatment, to a β + ωiso microstructure is not well understood. However, experimental investigation of this evolution is challenging as it requires in situ characterisation techniques. Resistivity measurements acquired over a broad range of temperatures for Ti-15Mo (wt.%) suggested that during initial heating (≲ 200˚C) ωath reverted to β, before ωiso formation took place between ~ 200 and ~ 400˚C [11, 12]. More recent complementary in situ neutron diffraction experiments have confirmed these trends and allowed a more quantitative assessment of phase evolution [13]. However, despite these data, clarity is still required in 15th World Conference on Titanium (Ti-2023) relation to how the ωiso nucleates and when it impacts the mechanical behaviour of the material. As such, this work seeks to clarify the nature and evolution of the ω phase present in the metastable β alloy Ti-15Mo (wt.%) using in situ techniques, and to link these results to its influence on properties. 2. Experimental Methods A bar of Ti-15Mo (wt.%), with a diameter of 8 mm and ~150 mm long, was provided for this work by Timet UK ltd. The actual composition of material is reported below in Table 1 [14]. A 7 mm slice was taken across the bar’s axis and cold rolled to a ~95% reduction. Small tensile specimens, with a gauge width of 0.5 mm, and 5 mm diameter discs were removed from the rolled strip by electrical discharge machining. All specimens were sealed in evacuated quartz tubes and heat treated at 900˚C for 300 s followed by air cooling. In situ heating and cooling experiments were performed on the I12 beamline at Diamond Light Source using a Linkam TST350 and a Linkam 1500V, which heated and cooled at 25˚C min-1. A 0.5 × 0.5 mm monochromatic incident beam, with an energy of ~ 80 keV, was used to illuminate the samples in a transmission Debye-Scherrer configuration, with diffraction data collected by a CdTe 2M Pilatus area detector located ~ 730 mm from the sample. The 2D diffraction data were integrated using the DAWN software [15, 16] and sequential fitting of individual diffraction peaks achieved with a Gaussian function in Wavemetrics Igor Pro. Microstructural characterisation was performed in a Zeiss GeminiSEM 300. Electron backscattered diffraction (EBSD) data were obtained using an Oxford Instruments Symmetry detector with an accelerating voltage of 25 kV and an 120 µm aperture. Processing of these data was performed using the Channel 5 Tango software. Ex-situ mechanical property evaluation was achieved through tensile testing performed on an Instron 3367B load frame, with a 30 kN load cell and a 12.5 mm Epsilon contact extensometer. Table 1: Measured compositions of the studied material (wt%) Ti Mo Fe O N C H 84.26 15.56 0.01 0.13 0.01 0.01 0.02 3. Results and Discussion 3.1. Initial condition Following heat treatment at 900˚C the initial microstructure of the material predominantly comprised equiaxed β grains, Figure 1, which were all within the range 20 – 80 µm. Inside these grains, a number of dark contrast parallel linear features could be seen. Further analysis using EBSD indicated that these features corresponded to thin regions of β phase that were misoriented to the rest of the grain. The misorientation measured across the interface was consistent with these features being <1 1 3>{3 3 2} twins. Complementary high energy synchrotron diffraction data collected from the material in the same initial condition are shown in Figure 2. A number of sharp high intensity peaks were present in the diffraction pattern, corresponding to the bcc β phase. A non-linear least squares regression analysis of these peaks identified the β lattice parameter to be 3.27 Å. Interspersing the β peaks were some much lower intensity reflections, which had substantially broader profiles. The most prominent example of one of these features can be seen at ~ 7.3˚2θ. When analysed, the positions of these peaks were consistent with the ω phase, which is well known to form in this alloy [5, 14]. Given that the material had been rapidly cooled from a heat treatment above the β transus, it was likely that this was the athermal form of the ω phase and, therefore, compositionally indistinct from parent β phase, as has been shown previously in reference [5]. A non-linear least squares regression analysis of the diffraction data gave lattice parameters of a = 4.86 Å and c = 2.88 Å for the ωath phase. Figure 1: Band contrast image of the initial condition, with EBSD orientation data overlaid (IPF-Z colouring). 15th World Conference on Titanium (Ti-2023) Figure 2: Synchrotron X-ray diffraction pattern of the initial material condition following 300 s at 900˚C. 3.2. Cooling and heating experiments To study the influence of temperature on the constituent phases of the alloy, a tensile specimen was mounted in a Linkam TST350 stage and illuminated by a synchrotron X-ray beam. The specimen was then subjected to a thermal cycle between -150˚C and 350˚C whilst diffraction patterns were collected at ~ 1 s intervals. Throughout the thermal cycle, the overall form of the diffraction patterns did not substantially vary outside of what would be expected for changes in temperature. Critically, at no point were any additional reflections observed in the diffraction data, indicating that no further phases formed during the thermal cycle. This is particularly relevant as the αʺ martensite phase has often been reported in this [17, 18] and other metastable β alloy systems [19-21] at or below room temperature. Whilst there were no gross changes in the diffraction data during the thermal cycle, there were some more subtle changes relating to the intensity of the β and ω phase reflections. To study these changes in more detail, the evolution of the ω peaks was tracked using a sequential single peak fitting routine. The output of this analysis for the most prominent (1 1 2)ω reflection is shown in Figure 3. During the cooling segment of the cycle, corresponding to the blue data points in Figure 3, the intensity of the ω reflections continuously increased in a near linear manner. This would suggest that the fraction of the ω phase present in the material was increasing inversely with temperature. Such behaviour would be consistent with the formation of stable ω, which forms through a phonon softening mode once a critical temperature, commonly referred to as ωs, has been reached. Figure 3: Evolution of the (1 1 2)ω peak area during thermal cycling between -150 and 350 ˚C. Upon heating from -150 ˚C to room temperature the intensity of the ω reflection immediately began to decrease, as shown by the red data points in Figure 3. The evolution of this peak followed the reverse trajectory of that observed on cooling, suggesting that temperature alone was the key driving factor. These observations are highly comparable with those of De Fontaine et al. [2], although in the current data there was no evidence of any hysteretic behaviour. The absence of any hysteresis indicated that this transformation must occur by an athermal mechanism, such as a phonon mediated process, rather than one involving longer range diffusion. When heating from room temperature to 300˚C, similar behaviour to that seen below 0˚C was observed. The ω peak intensity continued to reduce with temperature, indicating a reduction in the fraction of this phase within the microstructure and, correspondingly, a reduction in phase stability. It should be noted that despite the continual decrease in intensity, there was clear evidence for the presence of the ω phase throughout the heating segment. As such, this indicates that the stability limit of the ω phase in this alloy must be above 350˚C, consistent with neutron scattering data that suggest a solvus temperature of ~ 550˚C [13]. This observation is in stark contrast to other binary metastable β alloys, such as Ti-24Nb (at.%), where there is no evidence of the ωath phase above ~ 80˚C [20]. Interestingly, when heated above 330˚C, the intensity of the ω reflections began to show a positive trend with temperature, that is an increase in intensity as the temperature rose. This observation suggested that the fraction of ω in the microstructure increased rapidly at these temperatures, which would seem logical based on other similar data considering the diffusionally driven formation of the isothermal ω phase (ωiso) [6, 10, 13]. 15th World Conference on Titanium (Ti-2023) 3.3. Isothermal heat treatments The formation of ωiso in metastable β titanium alloys is well known but limited data exists that describes its evolution. To obtain a better understanding of how the ωiso develops as a function of time, a 5 mm diameter disc of the same heat treated material as studied in section 3.2 was heated to temperatures between 200 and 325˚C and exposed for 7 h (25.2 ks) in a Linkam 1500 V stage. As with the data presented in section 3.2, this experiment was performed within the synchrotron and diffraction data were acquired at regular intervals, ~37 s, throughout the thermal exposure. Example diffraction patterns corresponding to the start and the end of the thermal exposure at 300˚C are shown in Figure 4. From these data, it was evident that during this exposure the intensity of the ω peaks had substantially increased, consistent with previous reports in the literature [13]. This increase was most prominent for the (1 1 2)ω reflection, located at ~ 7.3 2θ but the same trend can be seen in all other ω peak locations. Similar behaviour was observed for all other exposure temperatures. Crucially, in none of the datasets was there any evidence of additional diffraction peaks forming during the exposure. This indicated that other phases, such as the equilibrium α phase, had not formed during these heat treatments. Figure 4: Synchrotron X-ray diffraction patterns corresponding to the material at the start (0 s) and end (25.2 ks) of the 300˚C thermal exposure. To elucidate the development of the ωiso phase during these exposures, the evolution of the (1 1 2)ω peak intensity was tracked through a sequential peak fitting routine. These data are plotted in Figure 5 and highlight a number of interesting points. In all cases, during heating to the exposure temperature the intensity of the ω reflection decreased. This is consistent with the data presented in Figure 3 and further supports the idea that the stability of the ωath present following the 900˚C heat treatment decreased as the material was heated. Similarly, the magnitude of the decrease in (1 1 2)ω intensity was greater when heated to higher temperatures. As noted in section 3.2, a distinct change in behaviour was observed above a critical temperature with the intensity of the ω reflections beginning to increase. However, within the dataset the evolution of the (1 1 2)ω intensity was noticeably different depending on the exposure temperature. At the lowest exposure temperature of 200˚C, the evolution of ω was very slow, with the peak intensity steadily increasing for the entire exposure time. Notably, even after a 7 h exposure, the peak intensity was only just equal to that originally present at room temperature. This suggests that the fraction of ωiso in the material following 7 h at 200˚C was the same or less than the fraction of ωath in the initial air cooled condition. Figure 5: Evolution of the (1 1 2)ω peak area during heating to and a 7 h (25.2 ks) exposure at 200, 250, 300 and 325˚C. At 250˚C the evolution of the ωiso phase was more pronounced than when the material was aged at 200˚C. The peak intensity rose sharply within the first 20 minutes before the increase in intensity became more gradual for the remainder of the isothermal exposure. At this temperature the peak area returned to the same intensity as it had been in the initial air cooled condition within 40 minutes and had developed a greater volume fraction by the end of the 7h exposure. The evolution of the ω phase at 300 and 325˚C were similar, with an extremely rapid initial increase in peak area that became far more gradual as the exposure progressed. This type of evolution is in line with previous observations that the ωiso phase can form very quickly during the initial stages of heat treatment. In this case, the peak intensity had returned to the initial room temperature level within 8 minutes at 300˚C and within 5 minutes at 325˚C. Furthermore, at both temperatures ~ 70% of the 15th World Conference on Titanium (Ti-2023) final ωiso fraction formed within the first hour of the exposure. The isothermal evolution of a phase is classically described using the Johnson-Mehl-Avrami-Kolmogorov expression, given below in equation 1. In this expression Vf is the volume fraction of transformed material, k is the reaction constant, t is the time over which the transformation has occurred and n is the Avrami constant. Critically, the value of the Avrami constant can provide insight into the nature of the transformation. 𝑉𝑓 = 1 − 𝑒𝑥𝑝(−𝑘𝑡𝑛) (1) The material studied in the present work was taken from rolled strip and despite the fact that it had recrystallised during heat treatment, the grain assembly was still not a good approximation of a crystallographic powder. As such, the use of Rietveld refinement to ascertain phase volume fractions would not be appropriate. Consequently, a modified version of equation 1 was used, where the degree of transformation is used instead of volume fraction; 𝑉𝑓 = 𝑓𝜔(𝑡) 𝑓𝜔 𝑚𝑎𝑥 (2) where fω(t) is the fraction of ω present at time t and fωmax is the maximum fraction that forms (taken to be the maximum fitted peak area) [22, 23]. In the current analysis, phase fractions were obtained from the normalised intensity of the (1 1 2)ω reflection, with t = 0 taken to be the time at which the sample first reached the ageing temperature following heating. For the exposures between 200 and 300˚C, a single low magnitude Avrami constant, between 0.3 & 0.5, was found to reasonably describe the ω phase evolution data in each case. In contrast, the data corresponding to exposure at 325˚C showed two distinct regions with the first being very short and having a steep gradient (n ~ 1.9), followed by a far longer region with a much shallower gradient (n ~ 0.45). Avrami constants with magnitudes ~ 0.5 have previously been linked to diffusional growth processes [22], which would be consistent with ωiso formation. Studies characterising the formation of α in other metastable β alloys have also observed a two stage process, where n ~ 1.2 in the first stage before dropping to a value < 1 in the following stage [22, 24]. In these prior studies, this initial region has been linked with the nucleation of the developing phase. However, it is not clear why this feature would only be seen at 325˚C in the present work. This aspect of the data requires further investigation and careful comparison to the behaviour of other metastable β binary alloys. 3.4. Mechanical Properties As highlighted in the introduction, many studies have reported a severe embrittling of metastable β alloys once ωiso forms. To assess the influence that the formation of ωiso had on the mechanical properties of the present material, a number of tensile tests were performed on specimens in different heat treated conditions. A selection of these data are presented in Figure 6. Figure 6: Tensile stress-strain data for Ti-15Mo following a number of different heat treatments. In all conditions, the material exhibited linear elastic behaviour before yielding and plastically deforming with essentially no work hardening. As might be expected, the solution heat treated material had the lowest yield stress (~ 900 MPa) and the greatest ductility. The formation of ωiso clearly stiffened the material, whilst also increasing the yield stress and reducing alloy ductility. Interestingly, and in contrast to some reports in the literature, the reduction in ductility was not instantaneous and, as can be seen from Figure 6, appeared to depend not only on the fraction of ωiso in the material but likely also its morphology and size. To obtain a more direct assessment of the influence of the ω phase, the mechanical properties of the alloy was also investigated through a series of in situ elastic loading cycles to a peak stress of 600 MPa. Each loading cycle was performed following a heat treatment at 300˚C, which progressively increased the cumulative exposure duration the material had spent at temperature. Plane specific diffraction elastic properties were obtained for both the β and ω phases by tracking the response of individual peaks during the loading segments of the cycle. 15th World Conference on Titanium (Ti-2023) In the solution heat treated condition the β planes had diffraction elastic constants in the range ~ 73 – 86 GPa, whilst the ω phase was found to be much stiffer with diffraction elastic constants between 145 and 150 GPa. Isothermal ageing led to a progressive increase of the β peak diffraction elastic constants, consistent with the trends shown from the ex situ mechanical testing results shown in Figure 6. In contrast, the diffraction elastic constants for the ω phase remained consistent throughout. These results suggest that the macroscopic stiffening of the material is a consequence of both an increase in the fraction of the ωiso phase and the associated increase in the β phase modulus as it becomes progressively enriched in Mo. 4. Conclusions In this paper, the evolution of the ω phase and its effect on the mechanical properties of a Ti-15Mo (wt.%) alloy have been studied. Through the use of in situ synchrotron diffraction studies a number of new observations have been made. In the initial solution heat treated condition, the material consists of equiaxed grains of metastable β, which themselves contain the ω phase. When cooling and heating from room temperature to cryogenic temperatures the volume fraction of ω changes in an identical manner, indicating that this must be the compositionally indistinct athermal form. Upon heating, the volume fraction of the ω phase continued to decrease until ~ 300˚C, above which it began to increase, consistent with the formation of the compositionally distinct isothermal variant. Similar evolution was noted during longer thermal exposures, where the rate of ωiso formation and the final fraction was heavily dependent on the exposure temperature. As with previous reports, the formation of the ωiso phase was found to stiffen and strengthen the material but at the expense of ductility. However, unlike prior research, this effect occurred more slowly and appeared to be as a consequence of both an increase in ω fraction and an associated change in the properties of the β phase as it became enriched in Mo. 5. Acknowledgements The authors would like to acknowledge Diamond Light Source for the provision of beam time (MG30411, MG31965 & MG33585) and S. Michalik and L. D. Connor for their assistance. 6. References 1. S.K. Sikka, et al. Progress in Materials Science 27 (1982) 245 - 310 2. D. De Fontaine, et al. Acta Metallurgica 19 (1971) 1153 - 1162 3. A. Devaraj, et al. Acta Materialia 60 (2012) 596- 609 4. S. Nag, et al. Physical Review Letters 106 (2011) 245701 5. J.M. Bennett, et al. Scripta Materialia 107 (2015) 79-82 6. S.A. Mantri, et al. Scripta Materialia 130 (2017) 69-73 7. F. Sun, et al. Materials Research Letters 5 (2017) 547-553 8. N.G. Jones, et al. in Proceedings of the 13th World Conference on Titanium (2016) 899-904 9. R.J. Talling, et al. Acta Materialia 57 (2009) 1188 - 1198 10. M.J. Lai, et al. Scripta Materialia 193 (2021) 38- 42 11. P. Zhanal, et al. Acta Physica Polonica A 128 (2015) 779-782 12. P. Zhanal, et al. Journal Of Materials Science 53 (2017) 837-845 13. P. Zhanal, et al. Materials … 12 (2019) 3570 14. J.M. Bennett, et al. Materials Characterization 142 (2018) 523-530 15. J. Filik, et al. Journal of Applied Crystallography 50 (2017) 959-966 16. M. Basham, et al. Journal of Synchrotron Radiation 22 (2015) 853-858 17. C.H. Wang, et al. Journal of Alloys and Compounds 720 (2017) 488-496 18. F. Sun, et al. Acta Materialia 61 (2013) 6406- 6417 19. E.M. Hildyard, et al. Acta Materialia 237 (2022) 118161 20. E.M. Hildyard, et al. Acta Materialia 199 (2020) 129-140 21. N.L. Church, et al. Shape Memory and Superelasticity 7 (2021) 166-178 22. S. Malinov, et al. Journal of Alloys and Compounds 348 (2003) 110-118 23. M. Naveen, et al. Journal of Alloys and Compounds 616 (2014) 607-613 24. S. Bein, et al. Journal De Physique IV 6 (1996) 99-107 1. Introduction 2. Experimental 3. Results and disussion 4. Conclusions 5. Acknowledgements 6. References 1. Introduction 2. Methods 3. Results 3.1. The effect of temperature on Ms 3.2. Microstructural changes during thermal cycling 3.3. Using thermal cycling to mitigate against functional fatigue 4. Discussion 5. Conclusion 6. Acknowledgements 7. Data Accessability 8. References 1. E.M. Hildyard, L.D. Connor, L.R. Owen, D. Rugg, N. Martin, H.J. Stone, N.G. Jones, Acta Mater 199 (2020) 129–140. 2. N. Church, L. Connor, N. Jones, Scripta Materialia 222 (2023) 115035.