1 Controlled Residual Surface Contamination of γtial, Induction Melted in Ceramic Crucibles J. Barbosa, A. Caetano Monteiro, Universidade do Minho, Guimarães, Portugal C. Silva Ribeiro, FEUP, Porto, Portugal Abstract This paper describes the melting procedure of gamma TiAl, using a multi-layered refractory crucible based on stabilized zirconia. Hardness and residual element contamination will be presented and discussed for samples let solidified and cooled inside the crucible in order to simulate the worst cooling conditions, regarding metal-mould reaction (samples subjected to the longest cooling time experimentally available). Results will be compared with samples melted and solidified in calcia and single stabilized zirconia crucibles. Some microhardness, residual contamination profiles and surface finish of samples poured into different refractory investment casting moulds will also be presented and discussed. 1. Introduction On the last decades, titanium and titanium alloys have been under intensive research and development, both for critical applications (aerospace, aeronautic, automotive and chemical industries) and for non-critical ones (sports equipment, photography, arquitecture, food industry). As a result of such effort, a large number of titanium alloys with combination of excellent tensile properties, creep strength and corrosion resistance, associated to a superior strength-to-weight ratio, are now available. Nevertheless, the use of titanium alloys is still reduced - the world consumption in 1998 was around 60,000 tons 1,2 mainly due to the extremely high price of ready to use parts. The actual production for consumer products is expected to increase on the next years, but that will surely depend on an effective cost reduction, either by the development of technological factors (new alloys and production techniques) or by a decrease in the prices of raw materials and on production costs. The major part of titanium and titanium alloys components is produced by plastic formation (forging, rolling, extrusion) and machining, and just a small quantity is produced by casting. However, the increase in the use of casting processes to produce near net shape titanium parts, is believed to be one of the most effective ways to decrease the actual production costs. 1 2. Casting of Titanium Alloys The production of high quality titanium castings is a difficult and expensive task. The main reasons for that are: the high melting point; the extremely high reactivity of titanium alloys against a large amount of elements (solid, liquids and gaseous) at high temperatures, with particular emphasis to oxygen. For this reason, traditional casting techniques and materials cannot be used, both for the melting and the moulding operations, and melting and pouring have to be performed under vacuum
2 or inert gas. Usually, arc-melting equipments with non-consumable electrodes and water-cooled copper crucibles are used to obtain billets of the desired composition, which are remelted several times until a quite homogeneous composition is achieved. To produce cast parts, consumable electrode arc melting (VAR) or induction skull melting (ISM) are used to melt previously produced billets and scrap, and the molten metal poured into ceramic moulds made by the investment casting process. In order to avoid metal-mould reaction, the traditional moulding materials cannot be used, because they can lead to a severe contamination of the castings. There are some good references to some moulding materials, but until now any mould material was found to be absolutely inert against titanium. This casting procedure presents some drawbacks: high melting cost due to high energy consumption and multi-melting operations; difficulty to control the superheating; alloy heterogeneity; high installation costs; castings contamination. The use of a ceramic crucible to melt titanium alloys by the induction melting process might be a possible solution, both to decrease melting costs (due to energy costs saving) and to eliminate some casting defects, such as misrun (due to poor superheating), which are the two most important drawbacks of the concurrent melting processes. However, the high reactivity of titanium alloys may impair the quality of the castings made using this process, due to reactions between the melt and the ceramic materials, present both in the crucible and in the mould. These reactions may contaminate the castings, due to the absorption of some residual elements, and chemical heterogeneities, inclusions and structural variations are also frequent. During the melting and solidification/cooling stages, metal-crucible and metal-mould reactions are possible to occur, with a greater probability to the first one, as the metal is at the liquid state and diffusion rates are usually higher. A molten metal-crucible reaction usually produces a uniform increase of the residual element content in the metal, due to the stirring effect of the induction field. During solidification, contamination mainly affects the outside region of the casting, from which results a higher content of contaminant elements near the interface between the casting and the mould wall, continually decreasing to a sort of plateau in which the level of contaminants is held constant (beyond a certain distance from the interface). Refractory materials used to produce foundry moulds are usually single or complex oxides and when a titanium alloy is poured into such moulds, oxygen absorption is usually present, as the mould oxides are less stable than those resulting from the reaction of the melt components with oxygen. This element has a marked tendency to form an interstitial solid solution with titanium. In practice, a harder superficial layer with higher oxygen content can frequently be found in titanium castings, known as "alpha-case". 3 Although there are some evident benefits from the use of refractory crucible induction melting when compared to alternative techniques (ISM or VAR) lower melting costs, lower investment costs, higher superheating, easier operation procedures this application is still depending on the development of suitable ceramic crucibles, because until now no single material was found to be absolute inert against titanium and titanium alloys.
3 3. Experimental Technique During the last 3 years, the authors have done research work on this field. Different kinds of crucibles and ceramic moulds have been evaluated, by melting and pouring TiAl, and results are published elsewhere. 4,5,6 Among tested ceramic crucibles ZrO 2 stabilized with different materials, CaO, graphite, and multi-layer ZrO 2 Y 2 O 3 crucibles only CaO and multi-layer ZrO 2 /Y 2 O 3 crucibles revealed to be thermodynamically stable enough to melt TiAl. However, some materials such CaO have high higroscopicity and poor thermal shock resistance, which decrease the possibility to use them on a regular basis. For this reason, multi-layer crucibles where found to be the best. The main objective of this paper is to summarize the results obtained by allowing samples to solidify inside the different crucibles, and describe the results obtained by the use of different moulding materials on the characteristics of TiAl castings, produced both by the investment casting and graphite permanent mould processes. Evaluation of castings was performed considering: microhardness and contaminant concentration profiles; correlation between oxygen concentration and microhardness profiles with the alphacase extension; surface roughness of the castings. The entire processing (melting, pouring and cooling) was performed inside a sealed chamber, where a controlled atmosphere of commercial pure argon was maintained. The chamber was open only after the sample reached room temperature. The melting procedure was the same as used in previous work and published elsewhere. 4,5,6 Melting stock weights 100 g, made of titanium rod grade II and commercial pure 99,8% aluminium. Samples solidified inside the crucibles were 40 mm in diameter and 22 mm long. Cast specimens were cylinders 20 mm in diameter and 85 mm long, and were poured around 1600ºC. Moulds were made using the shell investment casting process, but graphite permanent shells were also used. Production of investment casting shells was performed using traditional dip coatingstuccoing techniques, and permanent graphite shells were produced by machining. Shells were centrifugally poured, and for that operation they were placed inside a cylindrical positioning box and sustained inside by filling the remaining space inside the box with sodium silicate/co 2 moulding sand. Ceramic moulds (shells) have been produced using three different slurry-stucco systems, based on different binders and refractories (table 1). Table 1: Materials used to produce ceramic shells System nature Layers Slurry Stucco Nr. of layers Binder Refractory Silica Face coat and backup Colloidal SiO 2 Fused SiO 2 SiO 2 6 Zirconia Face coat and backup Colloidal ZrO 2 ZrO 2 ZrO 2 6 Multi-material 1st layer Colloidal Y 2 O 3 Y 2 O 3-1 2nd layer Colloidal Y 2 O 3 Y 2 O 3 ZrO 2 1 Backup layers Colloidal ZrO 2 ZrO 2 ZrO 2 6
4 Shells have been dried and dewaxed by thermal-shock and fired at 1100 ºC for 4 hours (ZrO 2 based and multi-material shells) or 1000 ºC for 2 hours (SiO 2 based shells). Mould temperature during pouring was around 100 ºC. 4. Experimental results Samples for characterization were collected from the middle section of the castings, by cutting the cylinders at half their height, and prepared using traditional metallographic techniques: surfaces were prepared by mechanical polishing and etched with Kroll solution (2%HF, 4%HNO 3, 94%H 2 O). Phase and chemical composition identification were performed by quantitative EDS analysis with standards of pure Ti, Al and Zr, using a JEOL JSM 35C scanning electron microscope. The carbon content of samples was measured by microprobe analysis, using a Cameca XS50 equipment. A secondary ion mass spectrometer (SIMS) was used to determine the oxygen contamination of the samples. Microhardness was evaluated on a Shimadzu hardness tester, using a 50g load, for 15 seconds. Surface roughness (Ra) was determined on a Perthometer S5P. 4.1 Samples solidified inside the crucibles The as cast microstructure of all samples presents two microconstituents: one with a strong lamellar dendritic pattern with two phases (α 2 + γ), and a interdendritic γ phase (figure 1). The dendritic constituent is present in higher quantity in all the samples, and the volume fraction of interdendritic γ increases from the outside of samples to the inside of them, following the decrease of cooling rate 4. Although, in samples obtained in multi-layer crucibles, the volume fraction of interdendritic γ is smaller than that found in samples previously obtained in single stabilized ZrO 2 crucibles. This is due to the absence of residual Zr in the γ phase, which was found to stabilize that phase. 5 α 2 +γ γ Fig. 1 - Microstructure of a casting melted and cooled in a calcia stabilized zirconia crucible In previous work, it was reported that stabilized ZrO 2 crucibles were unsuitable to melt TiAl, as the castings appeared to be contaminated with Zr and O, and revealed a significant extension of alphacase. 4,5,6 Following to that work, new ZrO 2 crucibles were tested. To decrease interaction with the molten metal, base crucibles were inside coated with Y 2 O 3 (which is a material thermodynamically
5 more stable than ZrO 2 ), in order to avoid direct contact between ZrO 2 and the metal. Samples melted and cooled in such multi-layer crucibles didn t show any contamination with Zr or any other residual element, what confirms their high stability. Microhardness value is in practice constant throughout the sample (around HV), and alpha-case was not detected. Microhardness value is similar to that found by authors in samples processed in CaO crucibles, in previous work 4,5,6, and it is in accordance with other references for TiAl, 7 what might indicate the absence of other kind of non-evaluated residual elements, as, for instance, oxygen. The results of microhardness testing and zirconium contamination are presented in figure 2. ZrO 2 /Y 2 O 3 Crucible ZrO 2 /CaO Crucible 600 10 500 6 550 500 450 350 8 6 4 2 Zr (%at) 450 350 5 4 3 2 1 Zr (%at) 250 0 250 0 CaO Crucible Multi-layer ZrO 2 /Y 2 O 3 Crucible 330 350 310 290 250 270 HV (α 2 + g) HV (γ) % Residual element (α 2 + γ) % Residual element (γ) Fig. 2 Microhardness and Zr profiles in samples obtained in stabilized ZrO 2 and ZrO 2 based multi-layer crucibles 4.2 Samples cast in ceramic shells and graphite permanent moulds Microstructure of samples poured into ceramic shells is quite similar to microstructure of samples solidified inside the crucibles, although, the volume fraction of γ phase is in this case much smaller (figure 3a). In samples cast in permanent graphite moulds, structure is fully lamellar α 2 + γ all over the samples, showing a very small volume fraction of isolated γ phase in the centre of the cross section of castings (figure 3b). However, its dimension didn t allow microhardness measurement and chemical composition evaluation.
6 The difference in microstructures is due to different cooling rates of both samples, with the fully lamellar structure corresponding to a much higher cooling rate, as also referred by other authors. 8 The presence of a small amount of isolated γ phase on the fully lamellar samples is due to the lower cooling rate of the inside region of the samples, when compared with the cooling rate of the outside regions. This effect is much more evident when the mould material presents a very high thermal conductivity (as graphite, for instance), and for that reason differences in the volume fraction of γ phase from the inside to the outside of castings is almost inexistent in samples cast in ceramic shells. a) b) Fig. 3 - Microstructure of castings poured into a) ceramic shells and b) permanent graphite moulds Microhardness and residual element concentration profiles of all poured samples are shown in figure 4. In samples cast in SiO 2 and ZrO 2 moulds there is evidence of a microhardness gradient, from the surface to the inside of the castings, which leads to an alpha-case extension of 250 and 50 µm, respectively. The development of that gradient is similar to those that were found in samples that were allowed to cool and solidify inside the crucibles: near the surface there is a rapid decrease of microhardness with the increase in distance from the sample / refractory interface. The inner part of the samples has a near constant hardness value, as shown in figure 4, which suggests that there is a constant level of contamination resulting either from the period when the metal was in the molten state or already present in the raw material itself. The outside profile suggests that it resulted from the metal-ceramic reaction during the cooling stage, in the solid state. It is considered that hardness increase is an outcome of contamination with residual elements. According to figure 4 microhardness and residual metallic elements segregation profiles both for SiO 2 moulds (Si) and ZrO 2 moulds (Zr) do not match, however oxygen distribution and microhardness profiles of lamellar microconstituent are quite similar (figures 4 and 5). This suggests that in samples cast in SiO 2 and ZrO 2 moulds, the presence of oxygen is the main responsible for the hardness increase near the surface. Samples cast in graphite moulds present an alpha-case extension of 150 µm, which is believed to result from surface contamination of the castings with carbon, as this element easily forms interstitial solid solutions with titanium. According to figure 4, microhardness and carbon concentration profiles have a similar development, and both assume an almost constant value for distances from the casting surface greater than 150 µm, which suggests their dependence. Samples cast in multi layer ZrO 2 /Y 2 O 3 moulds don t show any microhardness variation with increasing
7 distances from the casting surface (figure 4). Besides, no residual elements were found (Zr or Y). This confirms the high stability of Y 2 O 3, as suggested by its low free energy of formation, and the good results we had before, using this material as inside layer of ZrO 2 crucibles. 380 360 340 320 280 260 240 Alpha-case - 50 µm ZrO 2 mould 0,8 0,6 0,4 0,2 Zr (at. %) 450 430 410 390 370 350 330 310 290 270 Alpha-case - 250 µm SiO 2 mould 0,9 0,7 0,5 0,3 0,1 Si (at. %) Graphite mould Multi-layer ZrO 2 /Y 2 O 3 mould 380 360 340 320 280 260 240 1 Alpha-case - 150 mm 2 1,5 1 0,5 0 C (at. %) 380 360 340 320 280 260 240 HV (α 2 + g) HV (γ) % Residual element (α 2 + γ) % Residual element (γ) Fig. 4 - Microhardness and residual elements concentration profiles of samples cast in graphite, ZrO 2, SiO 2 and multi-layer ZrO 2 /Y 2 O 3 moulds. ZrO2 mould SiO2 mould 100000 Oxygen (counts) 10000 1000 75 µm 0 200 600 800 230 µm Depth from surface (mm) Fig. 5 - Oxygen distribution profiles in samples cast in ZrO 2 and SiO 2 moulds
8 The surface finishing of castings was evaluated by measuring its roughness (table 2). The best results were achieved on samples cast in graphite moulds (around 1 µm). Surface finishing of castings obtained in ZrO 2 and multi layer ZrO 2 /Y 2 O 3 ceramic shells was also excellent, but castings produced in SiO 2 shells revealed adherent moulding material, and roughness measurement was quite difficult. However on those areas where no adherent material was present, it was possible to find an average surface roughness around 4 µm, which was the highest value of all cast samples. Table 2 - Surface roughness (Ra) of cast samples. Mould material Graphite ZrO 2 ZrO 2 /Y 2 O 3 SiO 2 Surface roughness (Ra) 1 µm 2,1 µm 1,6 µm 4 µm Conclusions 1. A multi-layer ZrO 2 /Y 2 O 3 crucible was produced and used to induction melt TiAl alloys. These crucibles perform much better than traditional ZrO 2 crucibles. It was not noticed significant metal contamination with residual elements, and crucibles present a good thermal shock resistance. 2. Castings were produced by the investment casting process, using different moulding materials and by permanent graphite mould. In castings produced in multi-layer ZrO 2 /Y 2 O 3 shells, alpha-case was not detected, and microhardness was constant all over the samples. Those castings revealed an excellent surface roughness less than 2 µm. 3. Microhardness value of the lamellar constituent is a good indicator to identify a casting contaminated with oxygen; the alpha-case extension corresponds to the diffusion depth of oxygen in the casting. 4. This casting method allows the production of sound TiAl castings, and might be a valuable alternative process to produce titanium aluminide castings for non-critical applications. References (1) J.L. Martin: Proceedings of the 14 th Annual Conference and Exhibition of the International Titanium Association, 1998, 1. (2) D. Fischer: Proceedings of the 14 th Annual Conference and Exhibition of the International Titanium Association, 1998, 252. (3) T.S. Piwonka: Proceedings of the 42nd Annual Technical Meeting of the Investment Casting Institute, 1994. (4) J. Barbosa: International Journal of Cast Metals Research, 12 (2000), 293. (5) J. Barbosa: Key Engineering Materials, 188 (2000), 45. (6) J. Barbosa: Materiais 2001 1 st International Materials Symposium, 2001. (7) Materials Properties Handbook Titanium Alloys, Section VI Titanium Aluminides, ASM, June 1994. (8) M. Takeyama: Structural Intermetallics, 1993, 167.