E ﬀ ect of the High-Pressure Torsion (HPT) and Subsequent Isothermal Annealing on the Phase Transformation in Biomedical Ti15Mo Alloy

: Ti15Mo metastable beta Ti alloy was solution treated and subsequently deformed by high-pressure torsion (HPT). HPT-deformed and benchmark non-deformed solution-treated materials were annealed at 400 ◦ C and 500 ◦ C in order to investigate the e ﬀ ect of UFG microstructure on the α -phase precipitation. Phase evolution was examined using laboratory X-ray di ﬀ raction (XRD) and by high-energy synchrotron X-ray di ﬀ raction (HEXRD), which provided more accurate measurements. Microstructure was observed by scanning electron microscopy (SEM) and microhardness was measured for all conditions. HPT deformation was found to signiﬁcantly enhance the α phase precipitation due the introduction of lattice defects such as dislocations or grain boundaries, which act as preferential nucleation sites. Moreover, in HPT-deformed material, α precipitates are small and equiaxed, contrary to the α lamellae in the non-deformed material. ω phase formation is suppressed due to massive α precipitation and consequent element partitioning. Despite that, HPT-deformed material after ageing exhibits the high microhardness exceeding 450 HV.


Introduction
The interest in metastable β-Ti alloys has gradually increased due to their high specific strength, which make them ideal for applications in the aerospace industry [1]. Manufacturing of medical implants and devices is another high-added-value field and constitutes a prospective application of metastable β titanium alloys [2]. These alloys offer higher strength levels than commonly used α + β alloys due to controlled precipitation of tiny particles of α phase [3]. The strength of titanium and Ti-based alloys can be further improved by achieving an ultra-fine grained (UFG) structure via severe plastic deformation methods (SPD) [4][5][6][7]. Manufacturing of UFG metastable β-Ti alloys is of significant Metals 2019, 9,1194 2 of 14 interest as demonstrated by recent reports [8][9][10][11]. This study focuses on the effect of microstructure refinement by SPD on the precipitation of α phase upon subsequent thermal treatment.
The α phase forms in metastable β Ti alloys by a standard mechanism of nucleation and growth. The kinetics of α phase formation in a solution-treated material is generally given by chemical composition of the alloy and the temperature of ageing. However, α phase precipitation also strongly depends on the microstructure since α phase particles nucleate at preferential sites such as grain and subgrain boundaries and dislocations [12,13].
In some β-Ti alloy, nanosized ω phase particles can also act as preferential nucleation sites for precipitation of α phase, although the exact mechanism of ω formation has not been fully resolved yet [14][15][16][17][18]. It is therefore of significant interest to investigate α phase precipitation in the presence of high concentration of lattice defects (UFG material) in an alloy prone to the ω phase formation. Binary metastable Ti15Mo alloy used in this study consists, similarly to other β-Ti alloys with similar degree of β stabilization, of a mixture of β and ω ath (athermal ω) phases in the solution-treated (ST) condition [19,20]. This ω ath phase forms during quenching of the alloy from the temperatures above β transus by a displacive diffusionless mechanism [21]. However, it was also reported that ω phase can form as a result of high deformation, and is referred to as deformation induced ω [8,22]. The β → ω ath transformation is reversible up to a temperature of 110 • C [23]. Upon ageing at higher temperatures, the ω ath particles become stabilized by diffusion i.e., in Ti15Mo alloy by expelling Mo. This phase is referred to as ω iso (isothermal ω) [24]. The size of particles of ω iso phase is typically in the range of few nanometers up to 100 nm [19,24].
Solution-treated Ti15Mo alloy was processed by high-pressure torsion (HPT) [5] in order to achieve ultra-fine grained (UFG) microstructure with a high density of lattice defects. The equivalent von Mises strain achieved by HPT is heterogeneous and can be calculated according to Equation (1) [25]: where r represents the distance from the center of the sample, h is the thickness of the specimen, and N is the number of revolutions. The equivalent inserted von Mises strain after a single HPT rotation (N = 1, r = 10 mm and thickness h of 1 mm) ranges from 0 (the exact center) up to 35 = 3500% (periphery). Such extreme strain results in a dislocation density exceeding ρ = 5 × 10 14 m −2 and grain size in the range of hundreds of nanometers [11,26]. The objective of this study was to investigate the effect of SPD on the mechanisms and kinetics of α phase precipitation and to compare it with the precipitation in the non-deformed solution-treated material. Both laboratory X-ray diffraction (XRD) and high-energy XRD using synchrotron radiation (HEXRD) were employed for this experimental study.

Materials and Methods
Ti15Mo (wt%) alloy was supplied by Carpenter Co. (Richmond, VA, USA) in the form of a rod with the diameter of 10 mm. The as-received material was solution treated (ST) in an inert Ar atmosphere at the temperature of 810 • C for 4 h and subsequently water quenched. The cylindrical samples of the length of 5 mm were first cut from the rod and pressed with 6 GPa to achieve disk-shaped samples with the diameter of 20 mm and the thickness of approximately 1 mm. Disk samples were subsequently subjected to HPT deformation at room temperature at Ufa State Aviation Technical University (USATU), Russian Federation. Note that this pre-HPT deformation induced deformation of about 75% is significantly lower than the actual HPT deformation. The detailed description of the HPT method can be found elsewhere [5]. For this study, samples after N = 1 HPT rotation were prepared.
ST and HPT deformed samples (hereafter referred to as non-deformed and HPT-deformed, respectively) were aged at temperatures of 400 • C and 500 • C for 1, 4, and 16 h. The ageing was performed by immersing the samples to preheated salt-bath (i.e., with very high heating rate) without air access and subsequently water quenched. HPT samples for scanning electron microscopy (SEM) and microhardness study were prepared from the periphery part of the disks (>5 mm from the center) where the imposed strain is maximum. For XRD measurements, a quarter of a disk-shaped sample was used. All samples were mechanically grinded and polished by standard methods followed by a three-step vibratory polishing.
The microstructure of the specimens was observed using SEM Zeiss Auriga Compact Cross Beam (Jena, Germany) equipped with the energy dispersive spectroscopy (EDS) detector operated at 4 kV. K-line and L-line for Ti and Mo, respectively, were used for quantification of EDS data.
XRD measurement of the non-deformed Ti15Mo alloy was performed employing a Bruker D8 Advance powder X-ray diffractometer using Cu K α radiation (Bruker AXS GmbH, Karlsruhe, Germany), with a variable divergent slit and a Sol X detector. The width of the beam was 6 mm and the sample was rotated, allowing it to probe the whole specimen surface. XRD measurements of the HPT-deformed Ti15Mo alloy were carried out on Bruker D8 Discover powder X-ray diffractometer. Vertical Bragg-Brentano geometry (2.5 • Soller slits in both primary and secondary beam and 0.24 • divergence slit) with filtered Cu K α radiation was used. Beam size of 20 mm × 5-15 mm (depending on the θ angle) was used. Note that the beam size is comparable to the size of the HPT disk. Diffraction patterns in both cases were collected at room temperature in the 2θ range from 30 • to 130 • and were analyzed using LeBail approach in the program Jana2006 (Václav Petríček, Michal Dušek and Lukáš Palatinus, Institute of Physics Academy of Sciences, Prague, Czech Republic).
The HEXRD measurement was carried out at the P07-HEMS beamline of PETRA III (Deutsches Elektronen-Synchrotron, Hamburg, Germany) [27] using the energy of 100 keV (λ = 0.124 Å) in a transmission mode perpendicular to the HPT surface. Patterns of entire Debye-Scherrer rings were acquired ex-situ from the bulk of Ti15Mo samples at room temperature. A PerkinElmer XRD 1621 image plate detector was used. The samples were kept fixed during the acquisition and measured 5 mm from the center of the sample with an incident beam of slit size of 1 × 1 mm 2 . The acquired diffraction patterns were processed by Rietveld structural refinement using FullProf software (Juan Rodriguez Carvajal, ILL Grenoble, France). Azimuthal averaging over 360 • was performed first. HEXRD diffractograms after azimuthal averaging were treated by the March-Dollase approach in order to obtain at least rough estimates of volume fractions of individual phases. In fact, β phase peaks, which are the most intensive, could be fitted without the March-Dollase 'adjustment'. Nevertheless, diffractions of evolving α and ω phases had to be treated by the March-Dollase approach to achieve reasonable agreement with the measured data [28]. Even with the use of the March-Dollase approach the resulting R-factors of the fit accuracy range between 15-20 for α and ω phases, while R-factor for β phase is around 10.
The microhardness of the samples was measured using the Vickers method with the use of microhardness tester Qness Q10a (Golling, Austria). Note that all samples from HPT disks were cut from the area at least 5 mm distant from the center where microhardness of HPT N = 1 condition is saturated [8]. For each specimen 0.5 kg load and the dwell time of 10 s were applied. At least 20 indents were evaluated for each sample in order to get satisfactory statistical results.

Microstructure
The initial microstructure of the ST Ti15Mo alloy is shown in Figure 1. A coarse-grained (CG) structure consisting of grains of the average size of~50 µm is well visible due to the channeling contrast. In addition, brighter and darker and areas are also visible in the material due to the chemical contrast (marked by yellow arrows in Figure 1). These chemical inhomogeneities were investigated by EDS. Table 1 summarizes the results of the EDS point analysis. Note that results were obtained by standardless EDS in which the measured spectra are compared to the data collected from standards by the EDS manufacturer under different conditions (namely different beam conditions). Such data are therefore subjected to systematic error and as such they are not fully reliable in terms of exact quantitative Mo content determination. However, the relative difference in chemical composition between different areas is accurate and unambiguous. are therefore subjected to systematic error and as such they are not fully reliable in terms of exact quantitative Mo content determination. However, the relative difference in chemical composition between different areas is accurate and unambiguous.   The chemical inhomogeneities were also studied in the HPT-deformed sample. The SEM-BSE micrograph in Figure 3 clearly shows lighter and darker bands corresponding to the chemical composition differences, which were also confirmed by EDS. Darker areas with lower Mo were  are therefore subjected to systematic error and as such they are not fully reliable in terms of exact quantitative Mo content determination. However, the relative difference in chemical composition between different areas is accurate and unambiguous.  Table 1. Chemical composition of the darker and brighter bands (marked by yellow arrows in Figure  1) as determined by energy dispersive spectroscopy (EDS) point analysis.  The chemical inhomogeneities were also studied in the HPT-deformed sample. The SEM-BSE micrograph in Figure 3 clearly shows lighter and darker bands corresponding to the chemical composition differences, which were also confirmed by EDS. Darker areas with lower Mo were formed from curly bands in the non-deformed material ( Figure 1). In HPT deformed material, they are elongated in the direction of the deformation ( Figure 3).

The Evolution of Microstructure during Ageing
SEM-BSE micrographs in Figure 4 show the evolution of the microstructures of the nondeformed material after ageing at 400 °C and 500 °C for 1-16 h. The microstructure consists of coarsegrained β matrix. At least one triple-junction is shown in each image. After ageing at 400 °C for 1 and 4 h, only β matrix is observed. In the specimen aged for a longer time (400 °C/16 h), nanometer-sized precipitates are seen in the SEM-BSE micrograph (note the higher magnification of this micrograph). These small ellipsoidal particles are particles of ω phase, which are visible due to chemical partitioning-ω phase particles are slightly Mo depleted [29]. Ageing at the higher temperature of 500 °C resulted in a precipitation of continuous and coarse α phase along grain boundaries (hereafter referred to as grain boundary α or GB α), which is also Mo depleted and appears as a long dark particle along the former β/β boundary (indicated by a yellow arrow). In the vicinity of the β grain boundaries, α phase particles with a typical lamellar morphology precipitated. Small ellipsoidal particles in the grain interior belong to ω phase. The contrast of these particles increases with the increasing ageing time due to ongoing chemical partitioning. After ageing at 500 °C for 16 h, tiny ellipsoidal ω particles are clearly seen in grain interiors, GB α is visible along the former β/β grain boundaries and α lamellae span from the GB α to the grain interiors. In conclusion, ω particles with ellipsoidal morphology can be observed by SEM in the grain interior after ageing at 500 °C and the fraction of α phase particles with lamellar morphology increases with increasing time of ageing at 500 °C. The coexistence of all three β, α, and ω phases is observed. Figure 5 shows the microstructures of the HPT-deformed samples after ageing. Already after ageing at 400 °C/1 h, significant differences between the non-deformed and HPT-deformed specimens can be observed. In the non-deformed material, there is no evidence of α phase particles. On the other hand, small and equiaxed α particles already precipitated in HPT sample. In specimens aged at 400 °C for longer times of 4 h and 16 h, the volume fraction of the α phase increased and, simultaneously, α precipitates coarsened. Moreover, the precipitation is not homogeneous-some areas contain clearly more α phase particles. Ageing at 500 °C resulted in the formation of larger α particles, which are generally equiaxed, but not round-rather polygonal and sharp edged.  Figure 4 show the evolution of the microstructures of the non-deformed material after ageing at 400 • C and 500 • C for 1-16 h. The microstructure consists of coarse-grained β matrix. At least one triple-junction is shown in each image. After ageing at 400 • C for 1 and 4 h, only β matrix is observed. In the specimen aged for a longer time (400 • C/16 h), nanometer-sized precipitates are seen in the SEM-BSE micrograph (note the higher magnification of this micrograph). These small ellipsoidal particles are particles of ω phase, which are visible due to chemical partitioning-ω phase particles are slightly Mo depleted [29]. Ageing at the higher temperature of 500 • C resulted in a precipitation of continuous and coarse α phase along grain boundaries (hereafter referred to as grain boundary α or GB α), which is also Mo depleted and appears as a long dark particle along the former β/β boundary (indicated by a yellow arrow). In the vicinity of the β grain boundaries, α phase particles with a typical lamellar morphology precipitated. Small ellipsoidal particles in the grain interior belong to ω phase. The contrast of these particles increases with the increasing ageing time due to ongoing chemical partitioning. After ageing at 500 • C for 16 h, tiny ellipsoidal ω particles are clearly seen in grain interiors, GB α is visible along the former β/β grain boundaries and α lamellae span from the GB α to the grain interiors. In conclusion, ω particles with ellipsoidal morphology can be observed by SEM in the grain interior after ageing at 500 • C and the fraction of α phase particles with lamellar morphology increases with increasing time of ageing at 500 • C. The coexistence of all three β, α, and ω phases is observed. Figure 5 shows the microstructures of the HPT-deformed samples after ageing. Already after ageing at 400 • C/1 h, significant differences between the non-deformed and HPT-deformed specimens can be observed. In the non-deformed material, there is no evidence of α phase particles. On the other hand, small and equiaxed α particles already precipitated in HPT sample. In specimens aged at 400 • C for longer times of 4 h and 16 h, the volume fraction of the α phase increased and, simultaneously, α precipitates coarsened. Moreover, the precipitation is not homogeneous-some areas contain clearly more α phase particles. Ageing at 500 • C resulted in the formation of larger α particles, which are generally equiaxed, but not round-rather polygonal and sharp edged.

Evolution of Phase Composition during Ageing
The phase composition of the non-deformed and HPT-deformed samples before and after ageing is shown in laboratory XRD patterns in Figure 6 and Figure 7, respectively. Both the measured (thin black curves) and fitted (colored curves) XRD patterns are shown. The interplanar distance is displayed on the horizontal axis for the comparison to the HEXRD data, while the y-axis shows the intensity in a logarithmic scale, allowing one to distinguish small peaks. The most important peaks,

Evolution of Phase Composition during Ageing
The phase composition of the non-deformed and HPT-deformed samples before and after ageing is shown in laboratory XRD patterns in Figure 6 and Figure 7, respectively. Both the measured (thin black curves) and fitted (colored curves) XRD patterns are shown. The interplanar distance is displayed on the horizontal axis for the comparison to the HEXRD data, while the y-axis shows the intensity in a logarithmic scale, allowing one to distinguish small peaks. The most important peaks,

Evolution of Phase Composition during Ageing
The phase composition of the non-deformed and HPT-deformed samples before and after ageing is shown in laboratory XRD patterns in Figures 6 and 7, respectively. Both the measured (thin black curves) and fitted (colored curves) XRD patterns are shown. The interplanar distance is displayed on the horizontal axis for the comparison to the HEXRD data, while the y-axis shows the intensity in a logarithmic scale, allowing one to distinguish small peaks. The most important peaks, which best describe the evolution of emerging phases, are marked with arrows-full and open arrows for α and ω phase, respectively. A quantitative determination of phase content is not possible. The non-deformed specimen contains large grains with the size of hundreds of micrometers while HPT-deformed material is severely plastically deformed with high dislocation density and high internal stress resulted in the broadening of XRD peaks. Moreover, in both conditions, the grains have a preferred orientation as can be revealed from the relative intensity of the peaks. Therefore, laboratory XRD patterns could not be successfully fitted by any other method than the simple LeBail approach. However, several qualitative comparisons can be made.
The non-deformed specimen contains large grains with the size of hundreds of micrometers while HPT-deformed material is severely plastically deformed with high dislocation density and high internal stress resulted in the broadening of XRD peaks. Moreover, in both conditions, the grains have a preferred orientation as can be revealed from the relative intensity of the peaks. Therefore, laboratory XRD patterns could not be successfully fitted by any other method than the simple LeBail approach. However, several qualitative comparisons can be made. The non-deformed Ti15Mo alloy contains a mixture of β and ω phases. However, the identification of the ω phase content is difficult due to overlapping peaks of β and ω peaks. Nevertheless, ω peaks (11-22)ω and (11-21)ω can be observed at the inter planar distances dhkl ≈ 1.2 Å and dhkl ≈ 1.8 Å, respectively, as shown in Figure 6. Ageing of the non-deformed specimen at 400 °C/1 h (red curve in Figure 8) resulted in an increase of the intensity and narrowing of ω peaks (open arrows). In addition, small α phase peaks are also visible (full black arrows). XRD patterns of the nondeformed material aged at 400 °C for 1, 4, and 16 h are very similar. In the specimen aged at 500 °C for 1 h, the α phase is clearly present (full arrows in Figure 6). Moreover, its volume fraction increases with increasing ageing time (4-16 h), as also confirmed by SEM observations (cf. Figure 5); ω phase is still present in the specimen aged 500 °C even for the longest time of 16 h reported in [29]. The non-deformed Ti15Mo alloy contains a mixture of β and ω phases. However, the identification of the ω phase content is difficult due to overlapping peaks of β and ω peaks. Nevertheless, ω peaks (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22) ω and (11-21) ω can be observed at the inter planar distances d hkl ≈ 1.2 Å and d hkl ≈ 1.8 Å, respectively, as shown in Figure 6. Ageing of the non-deformed specimen at 400 • C/1 h (red curve in Figure 8) resulted in an increase of the intensity and narrowing of ω peaks (open arrows). In addition, small α phase peaks are also visible (full black arrows). XRD patterns of the non-deformed material aged at 400 • C for 1, 4, and 16 h are very similar. In the specimen aged at 500 • C for 1 h, the α phase is clearly present (full arrows in Figure 6). Moreover, its volume fraction increases with increasing ageing time (4-16 h), as also confirmed by SEM observations (cf. Figure 5); ω phase is still present in the specimen aged 500 • C even for the longest time of 16 h reported in [29]. The XRD pattern of the HPT deformed specimen exhibits significantly broadened peaks due to enhanced dislocation density and reduced crystallite size in this specimen. Moreover, the peaks are slightly shifted to different values of interplanar distances due to residual stresses in the deformed material-the direction of the shift depends on the type of the residual stress [30].
The ω phase content in the HPT sample seems to be inferior to that in the non-deformed sample; only a tiny peak can be resolved at the interplanar distance dhkl ≈ 1.2 Å. However, the most intensive peaks of the ω phase coincide with the peaks of the β phase.
In order to obtain more precise information about volume fraction of individual phases, HEXRD measurement was carried out on the HPT-deformed sample. In contrast to the laboratory XRD, HEXRD provides a better signal-to-noise ratio and the simultaneous measurement of the scattering signal in various directions due to the use of a 2D detector and subsequent azimuthal averaging. Consequently, a better resolution of small peaks, namely those of the ω phase, is achieved. Figure 8 shows the HEXRD pattern of the HPT specimen. Both the measured and fitted intensities as well as the difference curve between the fitted and measured intensity are shown in Figure 8. The results indicate that the HPT-deformed alloy is a two-phase material with volume fractions of the β and ω phase of 72% and 28%, respectively (the error of the volume fractions estimation is approximately ±5%).  In order to get more accurate results, the selected HPT-deformed specimen after ageing at 400 °C/1 h was examined using HEXRD. Both the measured and fitted intensity and the difference between the fitted and measured data are displayed in Figure 9. The XRD pattern of the HPT deformed specimen exhibits significantly broadened peaks due to enhanced dislocation density and reduced crystallite size in this specimen. Moreover, the peaks are slightly shifted to different values of interplanar distances due to residual stresses in the deformed material-the direction of the shift depends on the type of the residual stress [30].
The ω phase content in the HPT sample seems to be inferior to that in the non-deformed sample; only a tiny peak can be resolved at the interplanar distance d hkl ≈ 1.2 Å. However, the most intensive peaks of the ω phase coincide with the peaks of the β phase.
In order to obtain more precise information about volume fraction of individual phases, HEXRD measurement was carried out on the HPT-deformed sample. In contrast to the laboratory XRD, HEXRD provides a better signal-to-noise ratio and the simultaneous measurement of the scattering signal in various directions due to the use of a 2D detector and subsequent azimuthal averaging. Consequently, a better resolution of small peaks, namely those of the ω phase, is achieved. Figure 8 shows the HEXRD pattern of the HPT specimen. Both the measured and fitted intensities as well as the difference curve between the fitted and measured intensity are shown in Figure 8. The results indicate that the HPT-deformed alloy is a two-phase material with volume fractions of the β and ω phase of 72% and 28%, respectively (the error of the volume fractions estimation is approximately ±5%).
In order to get more accurate results, the selected HPT-deformed specimen after ageing at 400 • C/1 h was examined using HEXRD. Both the measured and fitted intensity and the difference between the fitted and measured data are displayed in Figure 9. In order to get more accurate results, the selected HPT-deformed specimen after ageing at 400 °C/1 h was examined using HEXRD. Both the measured and fitted intensity and the difference between the fitted and measured data are displayed in Figure 9. The volume fractions of individual phases in HPT-deformed sample before and after ageing at 400 °C/1 h are summarized in Table 2. The non-aged HPT-deformed material contains a high-volume fraction of the ω phase (28%). After ageing, the volume fraction decreases to approximately 9%. This is caused by enhanced volume fraction of the α phase, which reaches 23% in the aged condition. Table 2. Volume fraction of individual phases in HPT-deformed Ti15Mo alloy as determined from high energy synchrotron X-ray diffraction (HEXRD). The experimental errors are also shown.

Material
Volume Fraction of the β Phase Volume Fraction of the ω Phase Volume Fraction of the α Phase Ti15Mo HPT 72% ± 5% 28% ± 5% - The volume fractions of individual phases in HPT-deformed sample before and after ageing at 400 • C/1 h are summarized in Table 2. The non-aged HPT-deformed material contains a high-volume fraction of the ω phase (28%). After ageing, the volume fraction decreases to approximately 9%. This is caused by enhanced volume fraction of the α phase, which reaches 23% in the aged condition. Table 2. Volume fraction of individual phases in HPT-deformed Ti15Mo alloy as determined from high energy synchrotron X-ray diffraction (HEXRD). The experimental errors are also shown.

Material
Volume Fraction of the β Phase Volume Fraction of the ω Phase

Volume Fraction of the α Phase
Ti15Mo HPT 72% ± 5% 28% ± 5% -Ti15Mo HPT + ageing 400 • C/1 h 67% ± 5% 9% ± 3% 23% ± 4% Figure 10 shows the dependence of Vickers microhardness on ageing of the non-deformed and HPT-deformed samples. Ageing of both samples at 400 • C resulted in an abrupt increase of microhardness. In the non-deformed specimen, the microhardness increases with increasing ageing time at 400 • C. On the other hand, in the HPT-deformed sample aged at 400 • C for 1 h the microhardness reaches the maximum (500 HV). With increasing ageing time at 400 • C the microhardness continuously decreases. Specimens aged at 500 • C exhibit lower microhardness than specimens aged at 400 • C. The microhardness of HPT sample aged at 500 • C for 16 h even drops below the microhardness of the non-aged HPT specimen. Figure 10 shows the dependence of Vickers microhardness on ageing of the non-deformed and HPT-deformed samples. Ageing of both samples at 400 °C resulted in an abrupt increase of microhardness. In the non-deformed specimen, the microhardness increases with increasing ageing time at 400 °C. On the other hand, in the HPT-deformed sample aged at 400 °C for 1 h the microhardness reaches the maximum (500 HV). With increasing ageing time at 400 °C the microhardness continuously decreases. Specimens aged at 500 °C exhibit lower microhardness than specimens aged at 400 °C. The microhardness of HPT sample aged at 500 °C for 16 h even drops below the microhardness of the non-aged HPT specimen.

Enhanced α Phase Precipitation
A significant difference in the evolution of α phase in the non-deformed and the HPT-deformed conditions was observed. α phase is known to precipitate preferentially along the grain boundaries as so-called grain boundary α (GB α) [31]. Enhanced α phase precipitation was also found in predeformed materials due to the high dislocation density [32,33]. High concentration of defects in the HPT-deformed condition reduce the energy barrier for the nucleation of α phase. The growth of an α nuclei and its coarsening is controlled by the diffusion of Mo (β stabilizing element) in the β matrix [34]. It is well-known known that the pipe diffusion along dislocation cores as well as the diffusion rate along grain boundaries are several orders of magnitude higher than the bulk diffusion [35]. As a consequence, the growth of the α precipitates along grain boundaries is also accelerated. The enhanced precipitation of the α phase in severely deformed metastable β Ti alloys was reported in several studies [36,37]. In the coarse-grained material, α phase particles precipitate in the form of lamellae, because certain mutual orientations of neighboring α and β lattices are associated with the significantly lower interfacial energy and therefore, lamellar shape is optimal for the reduction of the total interfacial energy of a precipitate [38]. On the other hand, α particles in the HPT-deformed materials are equiaxed, but not round-detailed inspection of Figure 5 reveals that particles are rather Figure 10.
The evolution of the microhardness during ageing in the non-deformed and HPT-deformed samples.

Enhanced α Phase Precipitation
A significant difference in the evolution of α phase in the non-deformed and the HPT-deformed conditions was observed. α phase is known to precipitate preferentially along the grain boundaries as so-called grain boundary α (GB α) [31]. Enhanced α phase precipitation was also found in pre-deformed materials due to the high dislocation density [32,33]. High concentration of defects in the HPT-deformed condition reduce the energy barrier for the nucleation of α phase. The growth of an α nuclei and its coarsening is controlled by the diffusion of Mo (β stabilizing element) in the β matrix [34]. It is well-known known that the pipe diffusion along dislocation cores as well as the diffusion rate along grain boundaries are several orders of magnitude higher than the bulk diffusion [35]. As a consequence, the growth of the α precipitates along grain boundaries is also accelerated. The enhanced precipitation of the α phase in severely deformed metastable β Ti alloys was reported in several studies [36,37]. In the coarse-grained material, α phase particles precipitate in the form of lamellae, because certain mutual orientations of neighboring α and β lattices are associated with the significantly lower interfacial energy and therefore, lamellar shape is optimal for the reduction of the total interfacial energy of a precipitate [38]. On the other hand, α particles in the HPT-deformed materials are equiaxed, but not round-detailed inspection of Figure 5 reveals that particles are rather polygonal and sharp edged. It is assumed that α phase particles nucleate at triple junctions and all observed α particles are in fact GB α.
Ageing of HPT-deformed material (particularly at 400 • C/16 h) resulted in an inhomogeneous precipitation of α particles. Such inhomogeneity was reported to be caused by shear bands formed in the HPT deformed material [36,37,39,40]. However, we did not find any shear bands in the HPT material. On the other hand, we observed chemical inhomogeneities both in the non-deformed and in the HPT-deformed sample. In the latter case, the inhomogeneities are extended in the direction of HPT deformation (cf. Figure 3). As a consequence, the nucleation of the α phase particles may be therefore promoted in the areas depleted in Mo, even if the shape and the scale of precipitation inhomogeneities in Figure 5 cannot be directly compared to Mo-depleted regions in Figure 3 due to very different magnification (zone of observation).
In the specimen aged at 500 • C, ω phase was retained in the non-deformed material while it was completely absent in the HPT-deformed specimen. Enhanced precipitation of the α phase results in the rejection of the β stabilizing Mo to the surrounding β matrix causing a thermodynamic stabilization of the β matrix and suppression of the formation of the ω phase HPT deformed material [26]. Similar behavior, i.e., the preferred α phase precipitation over the formation of the ω phase, was observed in Ti-25Nb-2Mo-4Sn alloy deformed by cold-rolling [41].

Microhardness Evolution
Microhardness is significantly increased by HPT deformation as discussed in detail in [8] due to the microstructural refinement, the introduction of high dislocation density, and increased content of ω phase. Ageing of both non-deformed and HPT-deformed materials at 400 • C resulted in microhardness increase. Moreover, similar microhardness values were observed in both conditions. However, the similar increase of the microhardness can be attributed to different effects in both conditions.
The hardening of the non-deformed material aged at 400 • C is caused by the ω particles-the nano-sized ω particles are stabilized by diffusion, their size increases, and they act as much stronger obstacles for motion of dislocations. One may assume that a moving dislocation can pass through (cut) ω ath particles (known as Friedel effect [42]) as they are small and coherent. It is well known that the shear stress required for a dislocation to pass through a precipitate increases with its increasing size (within the Friedel's limit) and/or with increasing strength of the obstacle to dislocation motion [42,43]. Due to this and also because of the increasing volume fraction of ω phase, the hardness of the non-deformed material increases with increasing ageing time at 400 • C.
In HPT-deformed material, ageing at 400 • C already for 1 h results in the precipitation of tiny α phase particles, which are incoherent and cause significant Orowan strengthening. On the other hand, ω phase content is relatively low. The decreasing microhardness of the HPT material aged at 400 • C for longer times (4 and 16 h) may be related to the coarsening of the α + β microstructure. Both β matrix grains and α phase precipitates coarsen with increasing ageing time. The same process is even more pronounced during ageing at 500 • C. The microhardness was found to monotonically decrease with increasing ageing temperature and time. The maximum microhardness is therefore achieved in the HPT specimen aged at 400 • C. In a recent study [44], HPT deformation of Ti15Mo/TiB composite was performed at 400 • C and very high microhardness values (650 HV after N = 1 HPT revolution) were achieved.
The microhardness of the non-deformed material aged at 500 • C is inferior to that of the material aged at 400 • C for all ageing times. The relative decrease of microhardness of the non-deformed material with increasing temperature of ageing may be attributed to the decreasing volume fraction and increasing size of ω particles, whose size is well beyond the Friedel's limit.
Severe plastic deformation of the parent β phase and the introduction of high density of defects significantly accelerates the α phase precipitation. An important additional effect of this enhancement is the reduction of the content of ω phase and its disappearance at comparatively low ageing temperatures.

Conclusions
Metastable β titanium Ti15Mo alloy was prepared by HPT and subsequently aged at 400 • C and 500 • C. Phase transformations were observed by XRD and SEM. The following conclusions can be drawn from this study: Precipitation of the α phase in the HPT-deformed material is significantly enhanced by the high density of lattice defects such as dislocations or grain boundaries, which act as preferential nucleation sites for α phase precipitation. α phase particles in the non-deformed material precipitate in the form of lamellae, while in the HPT-deformed material, α precipitates are small, equiaxed, and polygonal in shape suggesting that they all formed as grain boundary α at the β grains triple junctions.
Two-phase α + β microstructure continuously coarsens with increasing temperature and time of ageing.
Deformation by HPT significantly increases the microhardness due to microstructural refinement, but also due to the formation of ω phase.
During annealing, the microhardness of the non-deformed material is governed mainly by the evolution of ω phase. In the HPT-deformed material, the main strengthening mechanism is the precipitation of fine α phase particles.

Conflicts of Interest:
The authors declare no conflict of interest.