TiFe Precipitation Behavior and its Effect on Strengthening in Solution Heat-Treated Ti-5 Al3 . 5 Fe During Isothermal Aging

We investigated the TiFe precipitation behavior of solution heat-treated Ti-5Al-3.5Fe during isothermal aging, quantified the effect of precipitation on strengthening by evaluating the hardness, and compared it to the effect of Ti3Al precipitation in Ti-6Al-4V. TiFe precipitates formed both at grain boundaries (GBs) and within the grain matrices. Phase transformation from the β to α phase also occurred during isothermal aging; this transformation generated lamellar interphase boundaries between the transformed α phase and remaining β phase in prior β grains. These interphase boundaries enabled the formation of in-grain TiFe precipitates by acting as a nucleation site. GB precipitation did not require prior β → α phase transformation to generate nucleation sites (i.e., interphase boundaries), so TiFe precipitation could occur immediately upon isothermal aging. Thus, GB precipitation proceeded more quickly than in-grain precipitation; as a result, precipitates were larger and more spherical at the GBs than in grains. The strengthening behavior exhibited by TiFe precipitation differed obviously from that caused by Ti3Al precipitation in Ti-6Al-4V because of its differing precipitation kinetics and related microstructural evolution.

Aging heat treatment in Ti alloys can increase their strengths by generating precipitates in their microstructures [6].A typical precipitate of Al-containing Ti alloys is Ti 3 Al, whose precipitation behavior has been extensively investigated using Ti-6Al-4V [11][12][13].Ti 3 Al precipitates form during aging heat treatment at temperatures of <550 • C [12,13], and their formation causes a significant increase in the material strength [12].Precipitation also occurs in the Ti-Al-Fe system at 530 • C, but the most common precipitate is TiFe, not Ti 3 Al [14].TiFe is made by the β-phase stabilizing element Fe, whereas Ti 3 Al is made by the α-phase stabilizing element Al; therefore, the precipitation characteristics of TiFe and the resultant strengthening in Ti-Al-Fe are expected to differ from those of Ti 3 Al precipitation in Ti-6Al-4V.Understanding precipitation behavior is essential for the design of of appropriate aging heat-treatment conditions.However, TiFe precipitation behavior in Ti-Al-Fe alloys remains poorly understood.
In this work, we investigated the TiFe precipitation behavior of a Ti-Al-Fe system during isothermal aging and evaluated the material hardness to quantify the effect of precipitation on strengthening.We found that the strengthening behavior of TiFe precipitation in the Ti-Al-Fe system differed significantly from that of Ti3Al precipitation in Ti-6Al-4V.The difference may arise from differing precipitation kinetics and related microstructural evolution.

Materials and Methods
A Ti-5Al-3.5Fe alloy ingot produced by induction skull melting was used in this work.To break up solidification structures, the ingot was forged at 1100 °C in the β-phase field and then hotrolled at 930 °C in the (α+β)-phase field; the processing temperatures were set with reference to the equilibrium phase diagram of the alloy (Figure 1).The rolled material was solution heat-treated by water-quenching from 950 °C in the (α+β)-phase field; the resulting material was defined as the initial material.To investigate TiFe precipitation evolution during aging, the initial material was heat-treated at 500 °C, at which temperature TiFe can form (Figure 1).The microstructures were examined by scanning electron microscopy (SEM, model: JSM-7001F, JEOL, Tokyo, Japan) at an acceleration voltage of 20 kV and by transmission electron microscopy (TEM, model: JEM 2100, JEOL, Tokyo, Japan) at an acceleration voltage of 200 kV.For SEM measurement, the sample surface was mechanically polished and chemically etched using a solution of 96 mL H2O + 2 mL HNO3 + 2 mL HF.TEM samples were prepared using a twin-jet electro-polisher (model: Tenupol-5, STRUERS, Cleveland, OH, USA) in a solution of 100 mL HClO4 + 900 mL CH3OH.Crystal structures of different phases were determined by X-ray diffraction (XRD) with Cu Kα radiation (model: Max-2500VL, Rigaku, The woodlands, TX, USA).Hardness was evaluated using a micro-Vickers tester (model: FM-700, Future-Tech, Kanagawa, Japan).

Results
The initial material that had been water-quenched (i.e., solution heat-treated) from the (α+β)phase region showed that equiaxed β (βR) grains were retained in the microstructure (Figure 2a).In The microstructures were examined by scanning electron microscopy (SEM, model: JSM-7001F, JEOL, Tokyo, Japan) at an acceleration voltage of 20 kV and by transmission electron microscopy (TEM, model: JEM 2100, JEOL, Tokyo, Japan) at an acceleration voltage of 200 kV.For SEM measurement, the sample surface was mechanically polished and chemically etched using a solution of 96 mL H 2 O + 2 mL HNO 3 + 2 mL HF.TEM samples were prepared using a twin-jet electro-polisher (model: Tenupol-5, STRUERS, Cleveland, OH, USA) in a solution of 100 mL HClO 4 + 900 mL CH 3 OH.Crystal structures of different phases were determined by X-ray diffraction (XRD) with Cu Kα radiation (model: Max-2500VL, Rigaku, The woodlands, TX, USA).Hardness was evaluated using a micro-Vickers tester (model: FM-700, Future-Tech, Kanagawa, Japan).

Results
The initial material that had been water-quenched (i.e., solution heat-treated) from the (α+β)-phase region showed that equiaxed β (β R ) grains were retained in the microstructure (Figure 2a).In α+β Ti alloys, thin α-phase lath structures are generally formed inside β-phase grains during water-quenching from high temperatures [15][16][17][18][19].The Fe in the Ti-5Al-3.5Fealloy has 2.5 times the β-phase stabilization ability of V in Ti-6Al-4V [8].Thus, the Ti-5Al-3.5Fealloy underwent almost no β → α phase transformation during solution heat treatment and the initial material shows equiaxed β R grains.We also observed primary α-phase (α p ) grains because the initial material had been cooled from the (α+β)-phase region.α+β Ti alloys, thin α-phase lath structures are generally formed inside β-phase grains during waterquenching from high temperatures [15][16][17][18][19].The Fe in the Ti-5Al-3.5Fealloy has 2.5 times the βphase stabilization ability of V in Ti-6Al-4V [8].Thus, the Ti-5Al-3.5Fealloy underwent almost no β  α phase transformation during solution heat treatment and the initial material shows equiaxed βR grains.We also observed primary α-phase (αp) grains because the initial material had been cooled from the (α+β)-phase region.The microstructure was significantly changed by isothermal aging.In the material aged for 10 h (Figure 2b), an α-phase lath structure was developed inside the βR grains; this change indicates the β  α phase transformation.As the aging duration was increased to 100 h, the α-phase laths became thickened (Figure 2c).In contrast, the αp-phase grains showed no changes during heat treatment (Figure 2b,c).The β  α phase transformation is also verified by XRD patterns (Figure 3).The initial material showed very strong peaks of the β phase, which confirm that its microstructure was mainly the βR phase; weak peaks of the α phase were due to the presence of αp grains.In the materials aged for both 10 and 100 h, however, the peaks of the β phase almost completely disappeared and new peaks of the α phase developed.This observation confirms that aging heat treatment induced β  α phase transformation.A slight shift in the XRD peaks was also observed as the aging proceeded (Figure 3).This phenomenon was because a significant lattice distortion in the initial material due to rapid cooling (i.e., water quenching) was relieved by β  α phase transformation and TiFe precipitation that occurred during aging.The microstructure was significantly changed by isothermal aging.In the material aged for 10 h (Figure 2b), an α-phase lath structure was developed inside the β R grains; this change indicates the β → α phase transformation.As the aging duration was increased to 100 h, the α-phase laths became thickened (Figure 2c).In contrast, the α p -phase grains showed no changes during heat treatment (Figure 2b,c).The β → α phase transformation is also verified by XRD patterns (Figure 3).The initial material showed very strong peaks of the β phase, which confirm that its microstructure was mainly the β R phase; weak peaks of the α phase were due to the presence of α p grains.In the materials aged for both 10 and 100 h, however, the peaks of the β phase almost completely disappeared and new peaks of the α phase developed.This observation confirms that aging heat treatment induced β → α phase transformation.A slight shift in the XRD peaks was also observed as the aging proceeded (Figure 3).This phenomenon was because a significant lattice distortion in the initial material due to rapid cooling (i.e., water quenching) was relieved by β → α phase transformation and TiFe precipitation that occurred during aging.α+β Ti alloys, thin α-phase lath structures are generally formed inside β-phase grains during waterquenching from high temperatures [15][16][17][18][19].The Fe in the Ti-5Al-3.5Fealloy has 2.5 times the βphase stabilization ability of V in Ti-6Al-4V [8].Thus, the Ti-5Al-3.5Fealloy underwent almost no β  α phase transformation during solution heat treatment and the initial material shows equiaxed βR grains.We also observed primary α-phase (αp) grains because the initial material had been cooled from the (α+β)-phase region.The microstructure was significantly changed by isothermal aging.In the material aged for 10 h (Figure 2b), an α-phase lath structure was developed inside the βR grains; this change indicates the β  α phase transformation.As the aging duration was increased to 100 h, the α-phase laths became thickened (Figure 2c).In contrast, the αp-phase grains showed no changes during heat treatment (Figure 2b,c).The β  α phase transformation is also verified by XRD patterns (Figure 3).The initial material showed very strong peaks of the β phase, which confirm that its microstructure was mainly the βR phase; weak peaks of the α phase were due to the presence of αp grains.In the materials aged for both 10 and 100 h, however, the peaks of the β phase almost completely disappeared and new peaks of the α phase developed.This observation confirms that aging heat treatment induced β  α phase transformation.A slight shift in the XRD peaks was also observed as the aging proceeded (Figure 3).This phenomenon was because a significant lattice distortion in the initial material due to rapid cooling (i.e., water quenching) was relieved by β  α phase transformation and TiFe precipitation that occurred during aging.The XRD patterns of the aged material also showed TiFe peaks (Figure 3), which indicates TiFe precipitate formation in the microstructure.TiFe precipitates were not clearly visible in the SEM image (Figure 2) because they are very fine (<100 nm [14]).TiFe precipitates were examined in detail using TEM.In the material aged for 10 h (Figure 4a), numerous and very fine (15-90 nm) TiFe precipitates were formed throughout the microstructure, both at grain boundaries (GBs) and within grains (Figure 4b).GB precipitates (Figure 4c) and in-grain precipitates (Figure 4d) both had a body-centered cubic (BCC) crystal structure, which is characteristic of TiFe [20].The GB precipitates were relatively spherical, but most in-grain precipitates were rectangular cuboids.The matrix showed strong hexagonal close-packed diffraction patterns (Figure 4e), which is consistent with the mostly α-phase matrix.Isothermal aging for 100 h increased the size of the spherical GB precipitates substantially, whereas in-grain precipitates increased moderately in size and became increasingly spherical (Figure 5a).The number of the precipitates was considerably decreased from 140/µm 2 in the material aged for 10 h (Figure 4a) to 37/µm 2 in the material aged for 100 h (Figure 5a).The matrix was almost completely transformed to the α phase (Figure 5b).The XRD patterns of the aged material also showed TiFe peaks (Figure 3), which indicates TiFe precipitate formation in the microstructure.TiFe precipitates were not clearly visible in the SEM image (Figure 2) because they are very fine (<100 nm [14]).TiFe precipitates were examined in detail using TEM.In the material aged for 10 h (Figure 4a), numerous and very fine (15-90 nm) TiFe precipitates were formed throughout the microstructure, both at grain boundaries (GBs) and within grains (Figure 4b).GB precipitates (Figure 4c) and in-grain precipitates (Figure 4d) both had a bodycentered cubic (BCC) crystal structure, which is characteristic of TiFe [20].The GB precipitates were relatively spherical, but most in-grain precipitates were rectangular cuboids.The matrix showed strong hexagonal close-packed diffraction patterns (Figure 4e), which is consistent with the mostly α-phase matrix.Isothermal aging for 100 h increased the size of the spherical GB precipitates substantially, whereas in-grain precipitates increased moderately in size and became increasingly spherical (Figure 5a).The number of the precipitates was considerably decreased from 140/μm 2 in the material aged for 10 h (Figure 4a) to 37/μm 2 in the material aged for 100 h (Figure 5a).The matrix was almost completely transformed to the α phase (Figure 5b).The XRD patterns of the aged material also showed TiFe peaks (Figure 3), which indicates TiFe precipitate formation in the microstructure.TiFe precipitates were not clearly visible in the SEM image (Figure 2) because they are very fine (<100 nm [14]).TiFe precipitates were examined in detail using TEM.In the material aged for 10 h (Figure 4a), numerous and very fine (15-90 nm) TiFe precipitates were formed throughout the microstructure, both at grain boundaries (GBs) and within grains (Figure 4b).GB precipitates (Figure 4c) and in-grain precipitates (Figure 4d) both had a bodycentered cubic (BCC) crystal structure, which is characteristic of TiFe [20].The GB precipitates were relatively spherical, but most in-grain precipitates were rectangular cuboids.The matrix showed strong hexagonal close-packed diffraction patterns (Figure 4e), which is consistent with the mostly α-phase matrix.Isothermal aging for 100 h increased the size of the spherical GB precipitates substantially, whereas in-grain precipitates increased moderately in size and became increasingly spherical (Figure 5a).The number of the precipitates was considerably decreased from 140/μm 2 in the material aged for 10 h (Figure 4a) to 37/μm 2 in the material aged for 100 h (Figure 5a).The matrix was almost completely transformed to the α phase (Figure 5b).TiFe precipitation clearly affected the Vickers hardness H V of the alloy (Figure 6).The initial material had H V ~450; this was increased to ~510 with 10 h aging and rapidly decreased for >10 h aging.The material showed H V ~400 after 100 h aging.This nonlinear trend in precipitation strengthening clearly differs from that of Ti-6Al-4V (Figure 6, [21]).TiFe precipitation clearly affected the Vickers hardness HV of the alloy (Figure 6).The initial material had HV ~ 450; this was increased to ~510 with 10 h aging and rapidly decreased for >10 h aging.The material showed HV ~ 400 after 100 h aging.This nonlinear trend in precipitation strengthening clearly differs from that of Ti-6Al-4V (Figure 6, [21]).

Discussion
Precipitation generally occurs at microstructural defects such as GBs and interphase boundaries because they act as nucleation sites for precipitation [22,23].However, our results showed that TiFe precipitates formed not only at GBs, but also inside the grain matrix in which no preferential nucleation sites had been observed.XRD (Figure 3) and TEM (Figures 4 and 5) results showed that the βR phase in the initial material was transformed to α phase during aging.The transformation yields a lamellar structure comprising transformed α phase and remaining β phase inside prior β grains [24,25]; the boundaries between these phases are well-known nucleation sites for precipitation [22].Accordingly, we expect that the β  α phase transformation occurred before in-grain precipitation, generating such a lamellar structure, and that the boundaries enabled the formation of in-grain TiFe precipitates by acting as a nucleation site.This expectation is supported by the observation that in-grain TiFe precipitates formed as if they were aligned at lamellar interphase boundaries (Figure 4).However, lamellar interphase boundaries were uncommon in the material aged for 10 h (Figure 4e) and almost completely disappeared after aging for 100 h (Figure 5b).This microstructural evolution may occur because, during TiFe precipitation, the β-stabilizing Fe atoms are expelled from the remaining β phase and move to the precipitates.Thus, the stability of the remaining β phase decreased as precipitation proceeded, and as a consequence, additional β  α phase transformation occurred, so the remaining β phase was converted to the α phase and the interphase boundaries were eliminated.
The morphology of in-grain TiFe precipitates changed as precipitation proceeded (Figures 4a  and 5a).In-grain TiFe precipitates were mostly cuboid (Figure 4a,b) in the material aged for 10 h; this shape implies high coherency between the precipitates and matrix [26].Maintaining a high coherency between different phases can minimize the free energy of the entire system by lowering

Discussion
Precipitation generally occurs at microstructural defects such as GBs and interphase boundaries because they act as nucleation sites for precipitation [22,23].However, our results showed that TiFe precipitates formed not only at GBs, but also inside the grain matrix in which no preferential nucleation sites had been observed.XRD (Figure 3) and TEM (Figures 4 and 5) results showed that the β R phase in the initial material was transformed to α phase during aging.The transformation yields a lamellar structure comprising transformed α phase and remaining β phase inside prior β grains [24,25]; the boundaries between these phases are well-known nucleation sites for precipitation [22].Accordingly, we expect that the β → α phase transformation occurred before in-grain precipitation, generating such a lamellar structure, and that the boundaries enabled the formation of in-grain TiFe precipitates by acting as a nucleation site.This expectation is supported by the observation that in-grain TiFe precipitates formed as if they were aligned at lamellar interphase boundaries (Figure 4).However, lamellar interphase boundaries were uncommon in the material aged for 10 h (Figure 4e) and almost completely disappeared after aging for 100 h (Figure 5b).This microstructural evolution may occur because, during TiFe precipitation, the β-stabilizing Fe atoms are expelled from the remaining β phase and move to the precipitates.Thus, the stability of the remaining β phase decreased as precipitation proceeded, and as a consequence, additional β → α phase transformation occurred, so the remaining β phase was converted to the α phase and the interphase boundaries were eliminated.
The morphology of in-grain TiFe precipitates changed as precipitation proceeded (Figures 4a  and 5a).In-grain TiFe precipitates were mostly cuboid (Figure 4a,b) in the material aged for 10 h; this shape implies high coherency between the precipitates and matrix [26].Maintaining a high coherency between different phases can minimize the free energy of the entire system by lowering the interphase boundary energy [26].In this case, precipitates tend to grow in a shape similar to their crystal structure [27].The in-grain precipitates may have formed cuboid shapes because they preferentially grew in a particular crystallographic direction (<110> in the BCC structure [28]).
Metals 2018, 8, 875 6 of 8 However, as precipitates grew further, the misfit strain required to maintain the coherency was increased, so eventually, the coherency failed and precipitates formed spherical shapes that reduced the misfit strain [26]; thus, large in-grain precipitates were spherical after 100 h aging (Figure 5a).
During aging, GB precipitation proceeded more quickly than in-grain precipitation.After 10 h aging (Figure 4b), GB precipitates were more spherical and larger than in-grain precipitates.In the material aged for 100 h, the GB precipitates had grown more significantly than the in-grain precipitates (Figure 5a).Unlike in-grain precipitation, GB precipitation did not require prior β → α phase transformation to create nucleation sites (i.e., interphase boundaries); therefore, GB precipitation could occur immediately upon isothermal aging.GBs permitted faster Fe diffusion for TiFe precipitation than the lamellar interphase boundaries did [29,30]; this difference further accelerated GB precipitation.
In the Ti-5Al-3.5Fealloy tested, the Vickers hardness H V increased significantly for the first 10 h of aging at 500 • C, compared to the hardness of Ti-6Al-4V (Figure 6).This result indicates that precipitation strengthening occurred and was greater in Ti-5Al-3.5Fethan in Ti-6Al-4V for 10 h aging.The number of individual precipitates is an important factor in determining the degree of precipitation strengthening [31,32].During aging at 500 • C for 10 h, numerous TiFe precipitates (140/µm 2 ) formed in the Ti-5Al-3.5Fealloy (Figure 4a), whereas a limited number (~16/µm 2 ) of Ti 3 Al precipitates formed in Ti-6Al-4V during aging at 575 • C for 10 h [21]; this quantity (16/µm 2 ) was estimated based on an interpolation using three results: 14/µm 2 at 2 h, 38/µm 2 at 144 h, and 73/µm 2 at 576 h [21].Thus, H V increased more rapidly in Ti-5Al-3.5Fethan in Ti-6Al-4V.More numerous TiFe precipitates may have formed because, in the Ti matrix, Fe atoms have diffusivities ~33 times higher than Al atoms [33].The fast Fe diffusion facilitated Fe clustering and thus accelerated the kinetics of TiFe precipitation, permitting the nucleation of numerous TiFe precipitates during relatively brief aging.
However, the number of TiFe precipitates considerably decreased in Ti-5Al-3.5Feafter 100 h aging (Figure 5a).This phenomenon occurred because the precipitation proceeded mainly by the coalescence of preformed precipitates rather than by the nucleation of new precipitates, as a thermodynamically driven process to lower the total free energy of the system by reducing the number of interphase boundaries [26].Thus, the present alloy showed a rapid reduction in hardness at aging durations of >10 h (Figure 6).In addition, considering that interphase boundaries increase the strength of α+β Ti alloys [34], the disappearance of the remaining β phase after 100 h aging may have contributed to the decreased hardness.In contrast, the number of Ti 3 Al precipitates in Ti-6Al-4V increased even during aging heat treatment over 100 h [21], possibly because the relatively slow kinetics of Ti 3 Al precipitation extended the period during which nucleation-dominant precipitation occurred, compared to the case of TiFe precipitation.Moreover, in Ti-6Al-4V, the remaining β phase is stable during aging [21] because its major precipitate (Ti 3 Al) is made not by the β-phase stabilizing element V, but by the α-phase stabilizing element Al; therefore, strengthening by lamellar interphase boundaries can occur.The results indicate that Ti-5Al-3.5Fehas an optimal aging time to maximize precipitation strengthening.This possibility should be evaluated in other Ti-Al-Fe alloys.

Conclusions
This work investigated TiFe precipitation behavior in solution heat-treated Ti-5Al-3.5Feduring isothermal aging at 500 • C, and compared this alloy's hardness to that of Ti-6Al-4V to quantify the effect of TiFe precipitation on strengthening.TiFe precipitates were formed at GBs and within the grain matrix.During the aging of Ti-5Al-3.5Fe,the β → α phase transformation generated a lamellar structure composed of transformed α phase and remaining β phase in prior β grains.These interphase boundaries enabled the formation of in-grain TiFe precipitates by acting as a nucleation site.GB precipitation did not require prior β → α phase transformation to create nucleation sites and thus could occur immediately; therefore, GB precipitation proceeded more quickly than in-grain precipitation.As a result, GB precipitates were larger and more spherical than in-grain precipitates.The alloy hardness increased by TiFe precipitation strengthening with ≤10 h aging, but rapidly decreased with longer aging.This precipitation strengthening behavior differed from that

Figure 1 .
Figure 1.Equilibrium phase information of Ti-5Al-3.5Fealloy used in this work as a function of temperature, showing possible phases and their fraction; this result was obtained by JMatPro-v8 calculation with Ti database.

Figure 1 .
Figure 1.Equilibrium phase information of Ti-5Al-3.5Fealloy used in this work as a function of temperature, showing possible phases and their fraction; this result was obtained by JMatPro-v8 calculation with Ti database.

Figure 3 .
Figure 3. XRD patterns of initial Ti-5Al-3.5Fealloy and alloy after aging at 500 °C for 10 or 100 h.

Figure 3 .
Figure 3. XRD patterns of initial Ti-5Al-3.5Fealloy and alloy after aging at 500 °C for 10 or 100 h.

Figure 3 .
Figure 3. XRD patterns of initial Ti-5Al-3.5Fealloy and alloy after aging at 500 • C for 10 or 100 h.

Figure 4 .
Figure 4. TEM analysis results of Ti-5Al-3.5Fealloy after aging for 10 h.(a) Bright-field image, (b) magnification of the area A indicated in (a), and (c-e) diffraction patterns of three important phases observed in (b).

Figure 4 . 8
Figure 4. TEM analysis results of Ti-5Al-3.5Fealloy after aging for 10 h.(a) Bright-field image, (b) magnification of the area A indicated in (a), and (c-e) diffraction patterns of three important phases observed in (b).

Figure 4 .
Figure 4. TEM analysis results of Ti-5Al-3.5Fealloy after aging for 10 h.(a) Bright-field image, (b) magnification of the area A indicated in (a), and (c-e) diffraction patterns of three important phases observed in (b).

Figure 5 .
Figure 5. TEM analysis results of Ti-5Al-3.5Fealloy after aging for 100 h.(a) Bright-field image and (b) corresponding diffraction patterns.

Figure 5 .
Figure 5. TEM analysis results of Ti-5Al-3.5Fealloy after aging for 100 h.(a) Bright-field image and (b) corresponding diffraction patterns.

Figure 6 .
Figure 6.Variation in average hardness with aging time; the value at aging time = 0 h indicates the hardness for the initial material.Error bars indicate the standard deviations.A result of Ti-6Al-4V [21] is presented for comparison.

Figure 6 .
Figure 6.Variation in average hardness with aging time; the value at aging time = 0 h indicates the hardness for the initial material.Error bars indicate the standard deviations.A result of Ti-6Al-4V [21] is presented for comparison. Metals 2018, 8, x; doi: FOR PEER REVIEW www.mdpi.com/journal/metals Metals 2018, 8, x; doi: FOR PEER REVIEW www.mdpi.com/journal/metals