J. Cent. South Univ. (2016) 23: 508-514

DOI: 10.1007/s11771-016-3096-y

Microstructural evolution in a powder metallurgical Ti-7Mo alloy with continuous oxygen gradient

CHEN Zhi-xing(陈智星)¹, LIU Bin(刘彬)^{1, 2}, LIU Yong(刘咏)¹, ZENG Fan-pei(曾凡沛)², LU Jin-zhong(卢金忠)²

1. State Key Laboratory of Powder Metallurgy (Central South University), Changsha 410083, China;

2. Fujian Longxi Bearing (Group) Corp., Ltd., Zhangzhou 363000, China

 Central South University Press and Springer-Verlag Berlin Heidelberg 2016

Abstract: A titanium alloy containing continuous oxygen gradient was prepared by powder metallurgy (P/M) and the composition–property relationship was studied on a single sample. The alloy was sintered with layered powder of different oxygen contents via vacuum sintering and spark plasma sintering (SPS), respectively. After subsequent heat treatments, high-throughput characterizations of the microstructures and mechanical properties by localized measurements were conducted. The Ti-7% Mo (molar fraction) alloy with an oxygen content ranging from 1.3×10^-3 to 6.2×10^-5 (mass fraction) was obtained, and the effects of oxygen on the microstructural evolution and mechanical properties were studied. The results show that SPS is an effective way for fabricating fully dense Ti alloy with a compositional gradient. The average width of α′ phase coarsens with the increase of the content of oxygen. The content of α″ martensitic phase also increases with the content of oxygen. At oxygen contents of 3×10^-3 and 4×10^-3 (mass fraction), the Ti alloys present the lowest microhardness and the lowest elastic modulus, respectively. The results also indicate that the martensitic phases actually decrease the hardness of Ti-7Mo alloy, and oxygen effectively hardens the alloy by solid solution strengthening. Therefore, the high-throughput characterization on a microstructure with a gradient content of oxygen is an effective method for rapidly evaluating the composition–property relationship of titanium alloys.

Key words: titanium alloys; oxygen; martensitic transformation; high-throughput method; powder metallurgy (P/M); mechanical properties

1 Introduction

Titanium is an attractive material with a high specific strength and excellent corrosion resistance. Its applications have been expanded in many areas such as aerospace, marine, medical parts and automobiles [1-2]. The effects of oxygen on the mechanical properties and microstructures of titanium alloys are of considerable importance. WHITTAKER et al [3] and YAN et al [4] reviewed the dependence of mechanical properties on oxygen for Ti alloys fabricated via ingot metallurgy, powder metallurgy and additive manufacturing, and indicated that increasing the content of interstitial oxygen in titanium alloys results in an increase in yield strength, tensile strength and fatigue resistance at a given stress level, but has a detrimental effect on the ductility. It has been also reported that the addition of oxygen suppresses the stress-induced martensitic transformation [5-6], and the effect becomes stronger with increasing the content of oxygen. Whereas, it is also suggested that martensitic phases (e.g. α′, α″) can be induced or promoted by high concentrations of oxygen in Ti alloys [4]. However, the detailed mechanism and the compositional limit of oxygen for the microstructural evolution are still to be clarified. Therefore, understanding the effect of oxygen on the martensitic transformation is essential in designing and optimizing Ti alloys [7]. However, systematic work on the effects of oxygen on the microstructure and mechanical properties is still insufficient, because in order to obtain a comprehensive profile of such relationship, numerous alloys with different composites are needed.

Recently, in order to increase the efficiency of alloy design, some high throughput methods were developed. YOO et al [8] reported a ‘continuous phase diagramming’ technology. An epitaxial thin film of perovskite manganites with a composition gradient was deposited, and then continuous physical properties were mapped on the thin film. XIANG et al [9-11] suggested a similar method of ‘phase diagrams on a chip’. They deposited precursors of BaCo₃, Y₂O₃, Bi₂O₃, CaO, SrCo₃ and CuO onto the substrates of either MgO or LaAlO₃ one at a time, and obtained the compositional gradient through controlling the sintering process, then performed investigations of compositional analyses, phase identification and property characterizations on the thin film. ZHAO et al [12-16] improved the combinatorial ‘diffusion-multiple’ method, which includes preparing a sample with multiple alloy compositions in parallel, creating composition gradients by a long-term annealing, and then screening the structures and properties simultaneously. The above methods are suitable for studying the phase diagrams and evaluating the composition-structure-property relationships of condensed-matter systems in a time-efficient way. However, these methods are relatively complicated, and are not used in studying the compositional evolutions of interstitial elements, such as oxygen and nitrogen.

Therefore, the purpose of the present work is to evaluate the effects of oxygen on the microstructural evolution and clarify the relationship of oxygen content with microstructure, microhardness and the modulus of the titanium alloys by a newly-designed high-throughput experimental method.

2 Experimental methods

Three kinds of titanium powders with different oxygen contents were used in this work. The physical and chemical characteristics of the powders are shown in Fig. 1 and Table 1. Most of the powders are irregular in shape. Figure 2 shows the scheme used for the preparation of titanium alloys with continuous oxygen gradients. Layer A represents the powders with a nominal composition of Ti-7%Mo (molar fraction) prepared by mixing the titanium powder A and the molybdenum powder. Layers B and C have the same composition, but were prepared with titanium powders with different oxygen contents. The thickness of each layer was 1.5 mm. The three layers were die-pressed under a pressure of 500 MPa to a disk with a thickness of 4.5 mm and a diameter of 30 mm. Two sintering methods were used to consolidate the green compacts. One was the conventional vacuum sintering which was conducted at 1200 °C for 3 h in a vacuum of 5×10^-3 Pa. The other was the spark plasma sintering (SPS) which was performed at 1000 °C at a pressure of 70 MPa for 15 min, on a SPS system (HP D25/3 type, FCT Corporation, Germany) in Ar. The diameter of SPSed samples was 30 mm. After grinding and polishing, the SPSed samples were sealed in a quartz tube and an additional solution treatment was carried out at 1200 °C for 3 h to guarantee the diffusion of oxygen. Then, the solution treatment (ST) of the specimens at 900 °C for 30 min followed by water quenching was conducted. The specimens were sectioned vertically, ground, and polished by 0.05 μm colloidal silica. The microstructure was examined using scanning electron microscope (SEM) (Hitachi 6460, Hitachi Co. Ltd., Japan) and electron probe microanalysis (EPMA). X-ray diffraction (XRD) for phase analysis was conducted using a diffractometer (SIMENS D500, Bruker, Switzerland). The localized microhardness and the elastic modulus were evaluated by nano-indentation mapping technique on an ultra-nano-hardness tester (UNHT, Sweden).

Fig. 1 Representative morphologies of raw powders:

Table 1 Physical and chemical characteristics of raw elemental powders

Fig. 2 Scheme used for preparation of titanium samples with continuous oxygen gradient (CA-compression axis)

3 Results and discussion

3.1 Preparation of oxygen-gradient alloy through vacuum sintering

Figure 3 shows the microstructures of the Ti-7Mo sample prepared with vacuum sintering. The dark areas correspond to the residual pores. Three layers can be obviously observed in the microstructure from the top to the bottom. Layer A has a high porosity of about 7.2%, and the porosities for the Layer B and the layer C are 3.1% and 0.2%, respectively. High porosities in layer A and B result from the large particle sizes which cause the insufficient densification during pressing and sintering. Figure 4 shows the compositional mapping of oxygen in the sample. As the label on the right side of Fig. 4 shows, the different contrast represents different oxygen contents. Specifically, the selected brighter area of LayerA signifies that the oxygen atoms concentrate around the pores of sintered Ti alloy, and the content can reach as high as 7×10^-3 (mass fraction). Since the residual pores restrain the diffusion of the oxygen, and induce local concentration of oxygen, it is hard to obtain samples with continuous oxygen gradient through the conventional sintering.

Fig. 3 SEM image of Ti-7Mo alloy prepared by vacuum sintering at 1200 °C for 5 h (Black areas correspond to residual pores)

Fig. 4 Compositional mapping of oxygen in Ti-7Mo alloy prepared by vacuum sintering at 1200 °C for 5 h

3.2 Preparation of oxygen-gradient alloy through SPS

The SPS method was usually used to obtain fully dense powder metallurgical materials. Figure 5 shows the microstructure of the Ti-7Mo alloy prepared by SPS. The relative density is as high as 99.5 %, and nearly no pores can be observed. Unlike the alloy prepared by vacuum sintering, the three layers in the SPSed alloy are difficult to be identified because of high relative density. As the SPS was carried out at a relatively low temperature and short time, the diffusion of oxygen was insufficient. Therefore, a solution treatment was carried out. The temperature dependence of the diffusivity can be expressed by the diffusion coefficient D [17]:

                            (1)

where D⁰ is the pre-exponential factor, Q the diffusion energy, k the Boltzmann constant and T the absolute temperature. D has an exponent relation to the diffusion activation energy. Because the small size of the oxygen atom permits easy jumping between interstices, the activation energy of the diffusion of oxygen is much smaller than that of titanium, and the diffusion coefficient of oxygen is many orders of magnitude largerthan that for titanium. According to Ref. [1], the diffusion coefficient D of the oxygen in the β titanium alloys at 1200 °C is about 2×10^{-10 m2}/s, which is much higher than that of the self-diffusion of β titanium (8×10^{-13 m2}/s). The distance over which the diffusion occurs in a given time (t) can be approximately calculated by the following equation [18]:

                                   (2)

To get a continuous oxygen gradient, the optimum diffusion distance of the oxygen atoms in the present study should be close to the thickness of the layer (1.5 mm). For a diffusion coefficient of 2×10^{-10 m2}/s, the diffusion over a distance of 1.5 mm takes approximately 1.125×10⁴ s according to Eq. (2). Therefore, the time for the solution treatment at 1200 °C was set for 3 h. Figure 6 shows the compositional mapping of oxygen in the alloy after the solution treatment. An obvious oxygen gradient can be seen in the normal direction. From the top to the bottom, the measured oxygen content increases almost linearly from 1.3×10^-3 to 6.2×10^-3 (mass fraction) (Fig. 7). Therefore, it is suggested that the method combining SPS and long time solution treatment is effective in preparing the alloy with an oxygen gradient.

Fig. 5 SEM image of Ti-7Mo alloy SPSed at 1000 °C with a pressure of 70 MPa for 15 min

3.3 Characterization of microstructures

With the continuous oxygen gradient, it is convenient to disclose the variations of microstructuresand properties along with the oxygen content. Figure 8 shows the SEM images (in backscattered electron mode) of the solution-treated Ti-7Mo alloy, and Figs. 8 (a), (b) and (c) correspond to the top, middle and bottom of the sample, respectively. All the areas of the alloy exhibit martensitic structure since the solution treatment was conducted in the single β phase region. According to the XRD patterns shown in Fig. 9, the microstructure in Layer A with a low oxygen content shows fine, acicular martensitic structure of hexagonal α′ phase. With the oxygen content increasing to 4×10^-3 (mass fraction), the microstructure consists of α′ and α″ phases. The width of α′ phase is broadened, and a small amount of acicular orthorhombic α″ phases are observed. As for the area with an oxygen content of 6.2×10^-3 (mass fraction), there is also some acicular α″ phase, and the hexagonal α′ phase exhibits a lath-like morphology. The diffraction patterns are broadened marginally with the increase of the oxygen content. Through quantitative analyses, the average widths of α′ phase in Layer A, Layer B, Layer C are about 350 nm, 740 nm and 1.3 μm, respectively.

Fig. 6 Composition mapping of oxygen in Ti-7Mo alloy prepared by SPS and solution treatment

Fig. 7 Oxygen content as a function of distance to top of sample (Sample was prepared by SPS and solution treatment)

Fig. 8 SEM images (in backscattered electron mode) of solution treated Ti-7Mo sample ((a), (b) and (c) correspond to top, middle and bottom of sample, respectively)

Fig. 9 XRD patterns of solution treated Ti-7Mo sample (Layers A, B and C correspond to top, middle and bottom of sample, respectively)

Under rapid cooling condition, β phase may transform to α′ or α″ phase instead of α phase by martensitic transformations. The type of martensite is mainly dependent on the migration distance of the atoms in β grains during the period of martensitic transformation. When the migration distance is relatively longer, α′ phase is formed. On the contrary, as the migration distance shortens, α″ phase will be dominant in the microstructure of Ti alloys. The transition from hexagonal α′ phase to orthorhombic α″ phase occurs under a certain content of alloying element. The martensitic transformation takes place in the order of bcc → orthorhombic → hcp. The orthorhombic α″ phase is a transitional form in the process of α′ martensitic transformation. The alloying element plays a significant role in impeding the migration of atoms during lattice reconfiguration, and stops the martensitic transformation of α″. In the earlier study of Ti-Mo alloy system, the martensitic transformation was influenced significantly by the content of Mo. According to DAVIS et al [19], Ti-4Mo (mass fraction, %) alloy presented α′ martensite [19]. HO et al [20] reported the coexistence of α′ and α″ martensites in Ti-6Mo (mass fraction, %) alloy. When the Mo content was increased to 7.5%, the microstructure presented α″ phase and a small volume fraction of β phase [20].

Oxygen is an alpha-stabilizing element, contributing to the formation of α phase and destabilizing the bcc β phase. In the perspective of martensitic transformation, the precipitation of oxygen atom in Ti-Mo alloy increases the starting temperature of the martensitic transformation. This implies that martensitic phases (e.g. α′, α″) can be induced or promoted by high contents of oxygen in Ti alloys. Thus, the volume fraction and the average lath width of α′ martensite increase with the addition of oxygen content. In the present work, the increase of oxygen facilitates the formation of α″ martensitic phase in Ti-7Mo (mass fraction, %) alloy. It has been suggested that the presence of oxygen in the material distorts the lattice structure, causing crystalline microstrain [21]. TAHARA et al [7] suggested that the addition of oxygen induces lattice modulation, for instance, oxygen atoms act as random point defects, and generate local stress fields which induce a domain structure and suppress long-range α′ martensitic transformation. The migration distance of atom is limited by the resistance provided by oxygen atoms in the lattice reconfiguration.

It can be concluded that the interstitial oxygen promotes the growth of α′ phase, and increases the content of α″ phase as well.

3.4 Mechanical properties

Figure 10 shows the continuous change in the microhardness as a function of oxygen content. According to previous study, the hardness of α′martensite is merely higher than that of α solid solution, while the participation of α″phase decreases the strength and the hardness of Ti alloys dramatically. The orthorhombic α″structure is intermediate between bcc and hcp structures. Orthorhombic α″involves smaller strain than that required to produce hexagonal α′structure. The small strains cause the transformed α″martensitic structure to be less hardened [19]. The α′and α″phases are substitutional supersaturated solid solution, and contribute little to the strengthening of Ti alloy. In other words, martensitic strengthening might not be the main strengthening mechanism for Ti alloys. In principle,the additional oxygen involves in the formation of α″martensites lead to low hardness. Consequently, Layer A with relatively low oxygen content presents a hardness about HV 410, and Layer B, which consists of α′and α″martensite, presents the lowest hardness value of HV 379. This is the main reason that the hardness decreases with the content of oxygen from 1.3×10^-3 to 3×10^-3 (mass fraction).

Fig. 10 Microhardness as a function of oxygen content in Ti-7Mo alloy

With the oxygen content varying from 3×10^-3 to 6.2×10^-3 (mass fraction), the Vickers microhardness increases almost linearly from HV 379 to HV 443. This phenomenon is attributed to the solution strengthening of oxygen. The lattice distortion caused by oxygen results in the hardening (embrittlement) of the alloy.

For commercially pure titanium, the relationship between the hardness and the content of oxygen can be described as follows [22]:

                            (3)

where H₀ represents the hardness of the pure titanium, b is a constant and [O] is the oxygen content in the alloy. Therefore, the microhardness is proportional to the oxygen content. HEO et al [23] studied the effect of oxygen on the hardness of pure titanium obtained by electron beam melting, and found that with the increase of oxygen content from 5×10^-4 to 7.8×10^-3 (mass fraction), Vickers microhardness increases linearly from 110 kg/mm² to 340 kg/mm². The variations of the microhardness for Ti alloys which contain (3-6.2)×10^-3 (mass fraction) oxygen in the present study agree well with those proposed in Eq. (3) and Ref. [23]. However, the hardness in the present study is relatively higher than that in Ref. [23], which might result from the solution strengthening of element Mo.

Figure 11 shows the continuous change in the elastic modulus as a function of the oxygen content. The trend of elastic modulus is similar to that of the hardness as a function of the oxygen content. There is a decrease of the elastic modulus until around 4×10^-3 (mass fraction), reaching a minimum value, and then, there is a continuous growing trend with the oxygen content.

Fig. 11 Elastic modulus as a function of oxygen content in Ti-7Mo alloy

The modulus at a low oxygen area is almost the same to that reported by IVASISHIN et al [24] in a Ti-7Mo alloy with 100% martensite phase. This again indicates that the increased content of α″ phase decreases the elastic modulus. At oxygen content over 4×10^-3 (mass fraction), the modulus increases markedly, which is resulted from the strengthening effect of interstitial oxygen atoms. The effects of the oxygen content on the yield strength and the tensile strength have been studied in other works [3-4, 25-26]. It was suggested that the yield strength and the tensile strength increase linearly with increasing oxygen content, and the present results are well in agreement with the previous study.

In general, the high throughput characterizations of a single Ti alloy sample with a continuous oxygen gradient clearly indicate well the relationship of composition-microstructure-property.

4 Conclusions

A high-throughput method for studying composition–structure-property relationships based on P/M technology is proposed. Through conventional vacuum sintering, the oxygen gradient cannot be formed due to the high porosity in each layer, while a sample with continuous oxygen gradient is successfully prepared by the combination of SPS and solution treatments. A Ti-7Mo alloy with an oxygen contents ranging from (1.3-6.2)×10^-3 (mass fraction) was prepared, and the effect of oxygen on the martensitic transformation, microhardness and the elastic modulus was studied. It is found that the oxygen promotes the growth of α′ phase, and increases the α″ phase content. With the increase of oxygen content, the microhardness and the elastic modulus decrease at first, and then increase linearly by the solution strengthening of oxygen. At oxygen contents of 3×10^-3 and 4×10^-3 (mass fraction), the Ti alloys present the lowest microhardness and the lowest elastic modulus, respectively. Thus, it is an effective method for rapidly evaluating the oxygen effect on titanium alloys by using a single PM sample. With more measurements at micro-scale, other properties such as thermal conductivity, dielectric properties, and optical properties can also be evaluated.

References

[1] LUTJERING G, WILLAMS J C. Titanium [M]. 2nd ed. Berlin: Springer-Verlag, 2007: 1-12.

[2] LEYENS C, PETERS M. Titanium and titanium alloys: Fundamentals and applications [M]. Weinheim: Wiley-VCH, 2003: 333-423.

[3] WHITTAKER D. Powder processing, consolidation and metallurgy of titanium [J]. Powder Metall, 2012, 55(1): 1-11.

[4] YAN M, XU W, DARGUSCH M S, TANG H P, BRAND M, QIAN M. Review of effect of oxygen on room temperature ductility of titanium and titanium alloys [J]. Powder Metall, 2014, 57(4): 251-257.

[5] LIU Z, WELSCH G. Effect of oxygen and heat treatment on the mechanical properties of alpha and beta titanium alloys [J]. Metall Trans A, 1988, 19A: 527-542.

[6] TAHARA M, KANAYA T, KIM H Y, INAMURA T, HOSODA H, MIYAZAKI S. Heating-induced martensitic transformation and time-dependent shape memory behavior of Ti-Nb-O alloy [J]. Acta Mater, 2014, 80: 317-326.

[7] TAHARA M, KIM H Y, INAMURA T, HOSODA H, MIYAZAKI S. Lattice modulation and superelasticity in oxygen-added β-Ti alloys [J]. Acta Mater, 2011, 59: 6208-6218.

[8] YOO Y K, DUEWER F, YANG H, YI D, LI J W, XIANG X D. Room-temperature electronic phase transitions in the continuous phase diagrams of perovskite manganites [J]. Nature, 2000, 406: 704-708.

[9] XIANG X D, SUN X, BRICENO G, LOU Y, WANG K A, CHANG H. A combinatorial approach to materials discovery [J]. Science, 1995, 268: 1738-1740.

[10] XIANG X D. Combinatorial materials synthesis and screening: An integrated materials chip approach to discovery and optimization of functional materials [J]. Annu Rev Mater Sci, 1999, 29: 149-171.

[11] KOINUMA H, TAKEUCHI I. Combinatorial solid-state chemistry of inorganic materials [J]. Nat Mater, 2004, 3: 429-438.

[12] ZHAO J C. Combinatorial approaches as effective tools in the study of phase diagrams and composition-structure-property relationships [J]. Prog Mater Sci, 2006, 51: 557-631.

[13] ZHAO J C. A combinatorial approach for structural materials [J]. Adv Eng Mater, 2001, 3: 143-147.

[14] ZHAO J C. A combinatorial approach for efficient mapping of phase diagrams and properties [J]. J Mater Res, 2001, 16: 1565-1578.

[15] ZHAO J C, JACKSON M R, PELUSO L A, BREWER L N. A diffusion multiple approach for the accelerated design of structural materials [J]. MRS Bull, 2002, 27: 324-329.

[16] ZHAO J C. The diffusion-multiple approach to designing alloys [J]. Annu Rev Mater Res, 2005, 35: 51-73.

[17] MEHRER H. Diffusion in solids: Fundamentals, methods, materials, diffusion-controlled processes [M]. Berlin, Springer-Verlag, 2007: 313-316.

[18] JACKSON K A. Kinetic processes: Crystal growth, diffusion, and phase transitions in materials [M]. Weinheim, Wiley-VCH, 2004: 27-40.

[19] DAVIS R, FLOWER H M, WEST D R F. Martensitis transformations in Ti-Mo alloys [J]. J Mater Sci, 1979, 14: 712-722.

[20] HO W F, JU C P, CHERN LIN J H. Structure and properties of cast binary Ti-Mo alloys [J]. Biomaterials, 1999, 20: 2115-2122.

[21] VICENTE F B, CORREA D R N, DONATO T A G, ARANA-CHAVEZ V E, BUZALAF M A R, GRANDINI C R. The influence of small quantities of oxygen in the structure, microstructure, hardness, elasticity modulus and cytocompatibility of Ti-Zr alloys for dental applications [J]. Materials, 2014, 7: 542-553.

[22] LEYENS C, PETERS M. Titanium and titanium alloys: Fundamentals and applications [M]. Weinheim, Wiley-VCH, 2003: 211-212.

[23] HEO K H, MUNIRATHNAMA N R, LIMA J W, LEA M T, CHOI G S. Effect of oxygen and yttrium doping on the electrical resistivity and hardness of titanium metal obtained by electron beam melting [J]. Mater Chem Phys, 2008, 112: 923–927.

[24] IVASISHIN O M, FLOWER H M, LTJERING G. Titanium 99, Science and technology [C]// CRISM “Prometey”. St. Petersburg, Russia, 2000: 77-81.

[25] MIURA H. The influence of density and oxygen content on the mechanical properties of injection molded Ti–6Al–4V alloys [C]// Advances in Powder Metallurgy & Particulate Materials. Princeton, NJ, Metal Powder Industries Federation. 2010: 44-51.

[26] EBEL T, FERRI O M, LIMBERG W, OEHRING M, PYCZAK F, SCHIMANSKY F P. Metal injection moulding of titanium and titanium-aluminides [J]. Key Eng Mater, 2012, 520: 153-160.

(Edited by YANG Bing)

Foundation item: Project(2014CB6644002) supported by the National Basic Research Program of China; Project(2015CX004) supported by the Innovation-driven Plan in Central South University, China; Project(51301203) supported by the National Natural Science Foundation of China; Project(2014M551827) supported by the National  Science  Foundation  for Post-doctoral  Scientists  of  China; Project(2014GK3078) supported by the Science and Technology Planning of Hunan Province, China

Received date: 2015-11-12; Accepted date: 2016-01-27

Corresponding author: LIU Yong, Professor, PhD; Tel: +86-731-88836939; E-mail: yonliu@csu.edu.cn; LIU Bin, Associate Professor, PhD; Tel: +86-731-88836939; E-mail: binliu@csu.edu.cn