Further Research on Hardenability of Ti-1023 β Alloy Leading to a New DOE

As part of a treatise on the hardenability of Ti-1023, a summary about the nucleation mechanisms of primary and secondary alpha, the viability of two-step aging, and existence of precipitation-free zones is provided. Assumptions are given on the nucleation mechanism of primary alpha particles and the effect of higher beta stabiliser content on stress-induced martensitic transformation. It is reminded that further research into two-step aging would be beneficial. A proposed design of the experiment deals with the existence of martensite in stressed samples, so that the martensitic transformation is minimized by an increased content of beta stabilisers, volume fraction of primary alpha, and beta grain size. Finally, a microstructure with stress-induced omega particles as precursors of secondary alpha particles may result in a better trade-off of mechanical properties; it is suggested that this be explored.


Introduction
Ti-1023 is a beta tanium alloy designed to provide be�er mechanical proper es than Ti64 and was patented by Timet in 1974 [1]. This alloy is currently available in three modifica ons AMS 4984, 4986, 4987 and finds an applica on in aircra� structural parts such as landing gear, actuators, flap tracks, rotor heads, fi ngs and fasteners [2] due to material proper es such as its excellent combina on of strength, duc lity, toughness, machinability, weight, superior fa gue proper es, corrosion resistance, deep hardenability with reasonable high-temperature strength and creep stability [3][4] [5][6] [7][8] [9]. Due to a wide range of chemical composi on, the molybdenum-aluminium equivalent [Mo-Al] eq can acquire values app. 5.3-10.4 [10] [2], and Ti-1023 is thus posi oned predominantly as a near beta alloy. The beta transus for a varying composi on is not known but some reported values are between 788-805°C [11] [4][12] [13]. A varying composi on influencing mainly the β-transus temperature, microstructure and material proper es has been causing confusion among researchers since its incep on due to possible forma on of martensite upon quenching to room temperature, especially when quenching thick sec ons [4] Ti-1023 is beta forged to increase fracture toughness and finish alpha-beta forged to keep a sufficient level of fracture toughness and to improve duc lity [3]. The alloy composi on and thermomechanical processing -such as forging history and heat treatments -have an impact on the mechanical proper es primarily by a modifica on of the volume frac on, morphology and distribu on of α phase [20].
The resul ng microstructure of samples quenched from above or below the beta transus and strengthened by α-aging has very good fa gue cycle proper es and crack growth resistance [21] [6]. Hence the major interest lies in tensile strength, elonga on and fracture toughness and their trade-off. Depending on thermomechanical treatment, the ul mate yield strength varies ca. 260-Hence, it is presumed that Ti-1023 with the [Mo-Al] eq =5.3-10.4 can go through sympathe c nuclea on at a higher range, mixed at middle and possibly interface instabili es at a lower range. The impact of various mechanisms of forma on of α p lamellae on mechanical proper es in general is not clear.
The mechanism of nuclea on of the secondary alpha α s in the absence of pre-exis ng isothermal omega ω i phase was documented on many occasions such as in step-quenching from the solu on treatment temperature [24], rapid hea ng [25] or up-quenching [26] or mixed hea ng [27] to aging temperature and suppression of ω i forma on as a consequence of a higher content of aluminium or oxygen [28]. Terlinde et al. [22] described a secondary alpha α s of size 3-8x0.1 micron that were created in an autocataly c manner in the case of a high hea ng rate (not specified) to an aging temperature >400°C in Ti-1023 alloy not forming martensite upon beta quenching. For the purpose of crea on of an isothermal transforma on diagram for Ti-1023, Toran and Biederman [14] experimented with step quenching from above the beta transus temperature and confirmed that above 550°C, the acicular secondary alpha α s formed upon annealing from beta phase. Below 550°C, the acicular martensite formed first. However, neither the composi on of this alloy nor any results of mechanical tes ng were available. Most likely the annealed microstructure of samples formed by step-quenching from above the beta transus and samples previously quenched to room temperature and rapidly heated or up-quenched to the same temperature is very similar despite the fact that the linear defects forming the athermal omega ω a embryos transform to ω a in the la�er case [29][25] [26].
The loca on of nuclea on of secondary alpha α s par cles assisted directly or indirectly by isothermal omega ω i par cles has received a lot of a�en on recently, especially in the case of samples beta quenched to room temperature. Nuclea on of secondary alpha α s par cles has been presumed at the β-ω i interfaces for both high and low misfit alloys [30][31] [32][33] [34], in ω i par cles [35], near β-ω i interface [36] and instead of ω i par cles [27]. During aging, Ti-1023 behaves as an alloy with a high linear misfit. The isothermal omega ω i par cles adopt a cubical shape to minimize the total elas c strain. The la ce misfit increases un l exceeding a cri cal value causing forma on of interfacial disloca ons where preferen ally the secondary alpha α s nucleates. Hence, secondary alpha α s par cles nucleate at the β-ω i interfaces [31] [37]. Beta solu on treated and water quenched Ti-1023 samples exhibited three different forms of isothermal omega ω i par cles such as nondescript, ellipsoidal and cubical depending on aging temperature and me [4]. Hadjadj et al. [38] experimented with alpha-beta solu on treated water quenched Ti-1023 alloy. The atomic probe measurements of the directly aged specimens at a temperature of 300°C showed isothermal omega ω i zones depleted in all alloying elements and a�er a longer annealing me also secondary alpha α s zones with a higher content of aluminium. The authors presumed that secondary alpha α s precipitated preferen ally at the β-ω i interface. Three different types of nuclea on mechanism of secondary alpha α s phase during aging were observed such as uniform at temperature <450°C, sympathe c at >400°C and preferen al along β-β grain boundaries at >650°C. The achievable strength was the highest in the uniform nuclea on scheme [4]. At low hea ng rates, it was observed that forma on of isothermal omega ω i par cles was being replaced by fine secondary alpha α s at aging temperatures >400°C in Ti-1023 alloy not forming martensite upon beta quenching [22].
A two-step aging process led to an improved combina on of strength and duc lity due to pre-exis ng ω i par cles which directed the distribu on and morphology of α s par cles [39] [40]. Ellipsoidal isothermal omega ω i par cles formed during a short anneal of a beta solu on treated and water quenched Ti-1023 sample at 400°C. The second anneal at 300°C for 45min resulted in cubical shape ω i par cles [4]. A two-step aged Ti-1023 beta quenched sample to room temperature had slightly larger secondary alpha α s par cles and gave very good fracture toughness and strain [7]. However, more sophis cated research into mechanical proper es is not available.
Quazi et al. [41] observed precipitate free zones, presumably due to grain boundaries ac ng as sinks for oxygen in samples aged at increased aging temperature in high-oxygen-enriched Ti-Nb alloys. Mantri et al. [42] and Sakamoto et al. [43] observed an improved duc lity in samples forming precipitate free zones. Precipitate free zones are not typical for Ti-1023, however Duerig et al. [4] described a thick layer of grain boundary alpha forming at high aging temperature 680°C a�er 2000min in an alloy with an oxygen content 0.15 wt pct. Precipitate free zones of a slightly smaller width are apparent from the SEM pictures.
The omega phase has been sa sfactorily stress-induced previously by cold rolling or compression at room temperature [44] [45] [46] [47]. Similar experimental work rela ng to Ti-1023 has not been located probably due to the stress-induced martensi c transforma on typical for this alloy [48][4] and the previously men oned existence of martensite upon beta quenching. The stress-induced martensi c transforma on is however avoidable by both increasing of a volume frac on of primary alpha α p [4] and increasing the beta grain size [49] [50]. Presumably, with a higher beta stabiliser content, the effect of stress-induced martensi c transforma on reduces as well. It is further not clear how the structure of the samples changes upon applica on of various strains and strain rates during an upse ng process at room temperature.

Further work with Ti-1023
Barriobero-Vila et al. [19] conducted probably the most sophis cated experimental work with differen al scanning calorimetry (DSC) and high-energy X-ray diffrac on (HEXRD)describing the phase transforma ons in beta quenched Ti-1023 when hea ng from room temperature to 600°C under various hea ng rates. The [Mo-Al] eq was low at 6.36 and orthorhombic martensite α'' was found in a retained β phase matrix upon quenching. Apart from orthorhombic martensite, athermal omega ω a was also found. Depending on a hea ng rate the orthorhombic martensite decomposed to beta phase, or also directly to omega phase. The authors suggested a diffusion-controlled mechanism of growth of isothermal omega ω i from beta at low hea ng rates and forma on of isothermal omega ω i from orthorhombic martensite at the highest hea ng rate. An extremely high volume frac on of isothermal omega ω i about 70-80% of the en re volume was found at temperature ca. 250-400°C for a high hea ng rate 50°C/min. Considering the fact, secondary alpha α s likely precipitates preferen ally at β-ω i interface [38], it might be beneficial to explore the effect of hea ng rate and the volume frac on of isothermal omega ω i for precipita on of secondary alpha α s .

Design of experiment
The primary objec ve of this paper was to provide background research to develop a new route to hardening of cylindrical parts of Ti-1023. Two extremes of composi on would be used at two levels of [Mo-Al] eq . The cylindrical workpieces would be heattreated at four levels, two above and two below beta transus. For both types of material this allows a determina on of a suitable beta grain size and a content of primary alpha α p to avoid either type or martensi c transforma on. The cylindrical samples shall be further upset at room temperature at two levels such as 5 and 15% at two levels of strain rate.

Summary
The author believes that given experimental work will lead to a more efficient and effec ve process achieving superior mechanical proper es of small cylindrical parts. In addi on, several research sugges ons have been given that could be clarified as part of planned experimental work.