MICROSTUCTURE OF CAST TITANIUM ALLOYS

Transcription

1 MATERIALS FORUM VOLUME Edited by J.M. Cairney and S.P. Ringer Institute of Materials Engineering Australasia MICROSTUCTURE OF CAST TITANIUM ALLOYS M.J. Bermingham, S.D. McDonald, M.S. Dargusch, D.H. StJohn CAST Cooperative Research Centre, School of Engineering, The University of Queensland, Brisbane, Queensland, Australia ABSTRACT Due to the significant raw-material costs, the near net-shape manufacture of titanium alloy components is an attractive proposition. Processing through melting and solidification allows manipulation of the bulk composition and control of the microstructure. This research examines the effect of commonly used alloying additions on the as-cast microstructure of titanium alloys. In particular, the effects of incremental additions of aluminium and vanadium are used as a basis for explaining the as-cast microstructure of the most commonly used titanium-based engineering alloy, Ti6Al4V. 1. INTRODUCTION Titanium has emerged as a very attractive metal for numerous applications. It has the highest strength to density ratio in comparison to other widely used metals such as iron, nickel and aluminium based alloys. The tensile strength of titanium alloys is comparable to that of some stainless steels, as well as iron and cobalt based superalloys 1. Titanium has exceptional corrosion resistance in many environments, often exceeding the corrosion resistance of stainless steels. Although titanium has properties superior to many other materials its applications are limited primarily to its high cost. Traditionally titanium was confined to niche markets such as military and commercial aerospace. However emerging markets for titanium include chemical processing, biomedical, marine and consumer goods such as sporting equipment. Titanium demand is expected to double over the next decade 2. The high cost of titanium makes net shape manufacturing routes very attractive. Casting is a near net shape manufacturing route that offers significant cost advantages over forgings or fabricated structures. In addition, titanium components that are cast and hot isostatically pressed (HIP) exhibit mechanical properties comparable to forged counterparts. Like other cast metals, the final properties of titanium castings are highly dependant on the microstructure and therefore control of solidification is very important. In pure titanium, solidification occurs at 1668ºC with the formation of a body centered cubic crystal structure referred to as β-titanium. An allotrophic transformation occurs at 882ºC where the BCC structure transforms to hexagonal close-packed. The HCP structure is referred to as α-titanium, and the temperature above which the microstructure is 100% β-titanium is known as the β- transus. The addition of alloy elements can affect the β- transus temperature. Certain elements will raise the β- transus and thus are referred to as alpha stabilizers, and other elements will lower the β-transus, known as beta stabilizers. This work investigates the mechanisms of solidification and microstructural evolution in as-cast titanium alloys. The influences of aluminium and vanadium on the as-cast microstructure of commercial purity (CP) titanium are investigated. 2. EXPERIMENTAL The influence of aluminium on cast titanium was examined by making individual aluminium additions to commercial purity (CP) ASTM grade 2 titanium prior to melting, while the influence of vanadium was examined by remelting commercial Ti-6wt%Al- 4wt%V (ASTM grade 5) samples. One cylindrical sample of diameter 22 mm and height 17 mm was prepared for each melt. When required, high purity aluminium ( wt %) was added to the sample by drilling a hole into the side of the sample and placing the appropriate proportion of aluminium into the hole to give the desired bulk composition. This method of alloying is similar to that used by previous researchers 3. All samples were adjusted to equal weights of approximately 25.5 grams. Three samples of CP titanium, Ti-6Al and Ti-6Al-4V were cast with the averaged chemical results shown in Table 1. Table 1. Nominal and analysed composition of the experimental castings (three-cast averages). Average Analysed Composition Nominal Al V C O N Fe Ti Composition CP Ti < BAL Ti-6Al BAL Ti-6Al-4V BAL 84

2 Melting and casting of the 9 samples took place in random order using a tungsten arc melting furnace under a protective argon atmosphere (Autocast 230, Dentaurum ) with the melt cycle being controlled to achieve identical cycles between alloys. Casting took place into a copper mould of dimensions shown in Figure 1 that was maintained at room temperature. Figure 2. Cross section of CP Ti casting consisting predominantly of equiaxed prior-β grains. Figure 1. Schematic of the copper mould used for casting, and the casting unit. Temperature measurements were obtained using a Type B thermocouple. The thermocouple was inserted into a specially designed mould of the same dimensions in Figure 1, where the tip of the thermocouple was located in the approximate center of the casting. The thermocouple wire was shielded by a ceramic sheath, and only the very tip of the wires were exposed to the molten metal. This resulted in the fastest response time possible. The sampling speed was Hz. A variety of internal α-morphologies were present within the equiaxed prior-β grains. The average prior-β grain size in all samples was within the µm range. The unalloyed samples and the samples containing only aluminium all exhibited similar microstructures, however the addition of vanadium altered the microstructure. The major differences can be seen in Figure 3. All samples contained transformed β present in a variety of α-morphologies such as widmanstätten, After casting, the samples were sectioned longitudinally down the center and polished by firstly grinding to 1200 grit silicon carbide paper followed by final polishing in a solution containing 10% H 2 O 2 and a 90% colloidal silica solution. The samples were then etched (3ml HF, 30ml HNO 3, 67ml H 2 O 2 ) followed by examination using optical and scanning electron microscopy. 3. RESULTS The microstructure of all samples predominately consisted of equiaxed prior-β grains. A macrograph illustrating the equiaxed grains of a CP titanium casting can be seen in Figure 2. Figure 3. [A] Typical microstructue of CP Ti: grain boundary α (a), fine accicular α (b), widmanstätten α (c), serrated α (d). [B] Typical Ti6Al4V microstructure containing very fine acicular α (e), fine acicular α and β (f) and prior-β grain boundaries (g). 85

3 acicular, lamellar, serrated and plate-like α, some of which have been labeled in Figure 3 in accordance with the literature 4-6. As seen in Figure 3 (b), the addition of vanadium radically alters the observed microstructure. The microstructure now consists predominantly of very fine acicular α and some of this acicular α is delineated by intergranular β. The thermal data collected during solidification indicated the casting experienced very rapid cooling in the copper mould. Initial cooling rates were in excess of 150ºC per second over the first 1-2 seconds decreasing to approximately 30ºC per second during the β α transition. 4. DISCUSSION The complicated phase transitions that occur within titanium and its alloys make understanding solidification and microstructural formation more difficult than that of many other foundry alloys. Under equilibrium conditions, solidification occurs at 1668ºC with the formation of the high temperature BCC β- titanium phase. Upon further cooling, a phase change to HCP α-titanium occurs at 882ºC. The addition of alloying elements complicates this further, as certain elements may lower or raise the β α transition temperature. Furthermore, solidification of titanium castings rarely occurs under equilibrium conditions. Examination of the as-cast CP titanium microstructure yields little information about the solidification aspects of the high temperature β phase. Only prior-β grains remain, and within these grains exists a multitude of various α-morphologies. The prior-β grains remain visible because various α-morphologies nucleate from these boundaries, including grain boundary α. In addition, the α-phase does not cross prior-β grain boundaries. The addition of 6wt% aluminium to the CP titanium did not change the as-cast microstructure. The fact that the addition of aluminium did not induce microstructural change is not surprising after considering the binary Ti-Al phase diagram shown in Figure 4. The Ti-Al phase diagram shows that aluminium has a very high solid solubility in titanium, and at the low levels used in this research, no intermetallics or second phases form. Furthermore, aluminium is an alpha stabilizer, and hence no beta titanium is expected to remain in the as-cast microstructure. EDS analysis of the microstructure confirmed the presence of aluminium in solid solution in the alpha phase, and no indication of preferential aluminium rich α-morphologies could be determined. Figure 4. Binary phase diagrams for (a) Ti-Al and (b) Ti-V 7. The addition of 4wt% vanadium to the Ti-6Al did substantially impact the as-cast microstructure. However, once again there was no evidence of the solidification process involving the L β transformation. Prior-β grain boundaries were clearly visible as in the case of the CP and Ti-6Al microstructure, however the internal phases within the prior-β grains were altered dramatically. Instead of the structure described above the interior of the prior-β grains almost completely comprised of very fine acicular α + β. This α/β mixture is expected from the Ti-V phase diagram (Figure 4 (b)) as vanadium is a β stabilizer and allows α and β to coexist at room temperature. It remains unclear whether the β-ti in either CP Ti, Ti- 6Al or Ti-6Al-4V first nucleates and grows via a dendritic or cellular mechanism, as no evidence of dendrites can be seen in the as-cast microstructure. It is stated in the literature that dendritic growth morphologies are often obtained in Ti-6Al-4V castings however the β α +β transition eliminates evidence of prior dendrites 4. To examine this further, a supplementary experiment was performed in an attempt to observe the suspected dendritic morphology of the beta phase. A range of phase diagrams were investigated to select an insoluble element which would be rejected during initial solidification and be 86

4 unlikely to go back into solid solution. As shown in Figure 5 boron satisfied this criteria. The selection of boron was further supported by prior observations in the literature that boron-containing intermetallics are found scattered throughout as-cast structures when added to common titanium alloys 8, 9. Figure 5. Binary Ti-B phase diagram 7. One boron containing casting was made by adding boron in the form of an Al-4B master alloy to a CP Ti sample in the same manner as discussed for the aluminium-containing samples. All variables and casting parameters were maintained the same as those of the previous experiments. The final composition for this alloy was Ti-3.25wt%Al-0.135wt%B. It is evident that the addition of wt% boron significantly altered the as-cast microstructure. As suspected, the presence of an insoluble element delineated the β-dendrites making the structure clearly visible, as shown in Figure 6. Numerous clusters of rod shaped boron-containing particles were observed in the interdendritic regions. These rods were approximately 2-10 µm long and the presence of boron was confirmed by EDS. A phase which resembled fine widmanstätten α was present within the dendrites. As seen in Figure 6, the α-phase overlapped interdendritic regions (and the borides). This indicates that in three-dimensions the dendritic structure and the interdendritic phases offer little barrier to the growth of the α-phase. It is known that the α-plates grow during the β α transition however from the observations of this paper it remains unclear where these plates initially nucleate, although it is probable that they nucleate and grow from prior-β grain boundaries, as suggested in the literature 1. Unlike the other alloys investigated, the undulating dendritic interface made it difficult to clearly distinguish the prior-β grain boundaries in the boron containing sample. Figure 6. SEM (Secondary electron mode) images of Ti-3.25Al-0.135B showing (a) the dendritic microstructure delineated by intermetallics and (b) interdendritic boron-containing particles offer no substantial barrier to the growth of α-phases Although it is clear that the influence of boron in titanium promotes solidification via a dendritic mechanism it still remains unclear whether CP titanium, Ti-6Al or Ti-6Al-4V also solidify dendritically or by other mechanisms. A model which predicts whether a liquid alloy will solidify in a planar, cellular or dendritic fashion is shown in Equation G mc0 (1 k) Equation 1 VS kd Where G temperature gradient V s - growth velocity M - liquidus gradient C 0 - composition of the alloy k - partition coefficient D - diffusion coefficient in the liquid This equation shows that it is a combination of the solidification conditions (i.e. left hand side of Equation 1) and alloy properties (i.e. right hand side of Equation 1) that predict the growth morphology of the solidifying phase. When the term G/V s is minimized, the constitutional undercooling increases and there is an increased tendency for dendritic growth. C 0, m, k 87

5 and D are properties of the alloy, and hence these variables will vary with the addition of Al, V and B. The fact that the boron-containing casting had a visible dendritic morphology could be related to either a change in solidification structure or simply the delineation of dendrites via the presence of an insoluble phase. After consulting the phase diagrams in Figure 4 and Figure 5, it becomes apparent that in the case of the boron addition, the right side of Equation 1 is infinitely larger than the equivalent calculation for aluminium and vanadium additions used in this study. This is because the partitioning coefficient, k, for boron in titanium approaches 0, whereas those of aluminium and vanadium in titanium approach 1. This does not however discount the possibility of a dendritic growth morphology in the aluminium and vanadium containing castings because the G/V s term hasn t been considered. The obtained temperature measurement data indicate that the castings were subject to very high initial cooling rates, in the order of several hundred degrees celsius per second. The rapid cooling rate during solidification may increase the thermal undercooling and effectively reduce the G/V s ratio, promoting a solidification instability and hence a dendritic growth morphology. Furthermore, it is known that in pure metals it is possible to obtain dendritic growth morphologies if the undercooling is sufficient 11 and it is noted that trace impurities were present in all alloys used in this study. In light of this, it is proposed that all of our castings including CP Ti solidify via a dendritic mechanism and that the addition of boron simply allowed this structure to be visualised with ease. A schematic showing the proposed solidification and microstructural evolution of cast titanium alloys is presented in Figure CONCLUSION A series of titanium castings were produced and the effects of aluminium, vanadium and boron on the ascast microstructure investigated. No difference or variation in the microstructures could be determined between the commercial purity (CP) titanium and the Ti-6Al. However, the presence of vanadium altered the microstructure significantly from the CP and Ti-6Al samples. The addition of boron also changed the cast microstructure. When boron was present in low concentrations (0.135wt%), no prior-β grain boundaries could be observed and an extensive network of boron containing particles delineated β-ti dendrites. Despite being present above the β-transus temperature these intermetallics offered no barrier to the growth of the α- phase. A model was presented and it was proposed that all castings solidified via dendritic mechanisms. The absence of dendrites in the room temperature microstructures of CP, Ti6Al and Ti6Al4V was attributed to the high solubility of the alloy additions and the lack of any segregated phases to delineate the dendritic structure. Figure 7. Schematic of proposed microstructural evolution in titanium castings. In the presence of soluble elements a dendritic structure is possible but no evidence of this structure exists at room temperature. Insoluble elements allow prior β-dendrite arms to be visualised due to the presence of intermetallic precipitates. In both cases the morphology of the dendrites will depend on both the thermal conditions during solidification and the alloy composition. 88

6 References 1. M.J. Donachie: Titanium a technical guide, 2 nd ed. ASM International, Materials Park, OH, J. Simpson: Materials Technology, 2006, vol. 21(4), pp F.A. Crossley: SAMPE Journal, 1986, vol. 22(1), pp ASM Handbook: Titanium and titanium alloys, 1990, vol. 15, pp ASM Handbook: Titanium and titanium alloys, 1990, vol. 9, pp American Society for Metals: Metals Handbook. 8 th ed, 1972, vol ASM Handbook : Alloy phase diagrams, 1990, vol J. Zhu, et al.: Materials Science & Engineering, 2003, Vol. A339, pp S. Tamirisakandala, et al.: Scripta Materialia, 2005, vol. 43, pp M.C. Flemings: Solidification processing, McGraw-Hill, New York, W. Kurz, and D.J. Fisher: Fundamentals of solidification, 3 rd ed, Trans Tech Publications Ltd,