OF Tl-ALUMINIDE-BASED ALLOYS

J. W. Sears**, G. ltoh*,'and M. H. Loretto*

*IRC in Materials for High Performance Applications The University of Birmingham, Edgbaston, England B15 2TI,

**ManLabs, A Division of Alcan Corporation 21 Erie Street, Cambridge, Massachusetts, 02139

ABSTRACT

The aim of this programme is to spray form Ti Aluminides free from porosity and interstitial contaminatio.n. Initial work wa-s carried out by atomising IMI-318 (Ti-6-4) and Ti-aluminide alloys to assess the operation of the cold-wall induction bottom-pour (CWIBP) as a clean melting system. Ti-aluminide-based alloys have been sprayed onto flat and tubular substrates. The IRC plasma melting system at The University of Birmingham, Edgbaston, England, was used to provide ingot feed stock for the spray deposition and atomisation. The plasma system, used for ingots, was able to be converted for atomisation or spray deposition. Spray forming was facilitated by the use of a scanning or centrifugal atomiser. The operation of this process will be discussed and the effects of various operating parameters on the spray deposit product will be presented. The chemistry, microstructure, homogeneity and macrostructure of the as sprayed material has been examined.

INTRODUCTION

The Ti aluminides form a technologically important group of intermetallic compounds. TiAI (y) is one of these which exhibits great promise for commercial exploitation, being strong at elevated temperatures while also exhibiting superior oxidation resistance. However, the compound is brittle at ambient temperatures and this limits its application. Possible sources of this brittleness include compound purity, planarity of slip and the nucleation of voids at the points of intersection of deformation twins. Some attempts have been made to improve the room temperature ductility by alloying additions, the introduction of a dispersion of hard second phases, and the control of grain size. Consequently, there has been considerable interest in the application of powder (PM) and rapid solidification processing (RSP) to the preparation of these materials. Research has been focused on IM route by isothermal and the (PM) approach through powder consolidation via HIP, plasma spraying and mechanical alloying. Although, there has been considerable success in the IM and PM techniques, they have proved to be very expensive'. The spray forming of a deposit directly from the melt offers a more economical process to produce such difficult to form materials. As will be shown, deposits of Ti-aluminides can be produced directly from a stream of molten metal without the handling problems associated with the P M route.

Tilanium '92 Science and Technology Edited by F.H. Froes and I. Caplan The Minerals, Melals & Materials Society, 1993 987 It has been pointed out that there are as many as five different solid phases possible for an of composition Ti-52AI which has been undercooled -200 K below the equilibrium liquidus, these being ~(bee), a (hep), and the intermetallic compounds y (Llo), B2, and a2 (D019)2. It is known that when processing with the cold-wall induction bottom-pour (CWIBP), described below, that C and 02 can be introduced into the powders. Carbon is highly insoluble in TiAl, and the compound Ti2AIC may precipitate during consolidation and/or heat-treatment of the powder3. Part of the investigation, involving the powders and spray deposits, will be the examination of Ti-alloy particles before and after heat treatment to determine the extent of the C contamination. Oxygen affects the beta transus temperature of these alloys and adds to the material inherent brittleness.

COLD-WALL INDUCTION BOTTOM-POUR (CWIBP)

The CWIBP system includes a melt chamber capable of being pressurised to 170 kPa. This allows for the flexibility to ramp an overpressure in the melt chamber to account for the change in metalistatic head during the metal pour. A controlled overpressure allows for a constant metal flow rate which is critical in producing uniform spray deposits. The cold-wall crucible is a segmented, water-cooled, induction heated device. An induction-heated graphite nozzle at the base of this crucible constricts the molten metal into a stream for atomisation and spray forming.

Figure 1: a) CWIBP crucible installed in the IRC PACH furnace. b) Schematic of the CWIBP system.

Plasma-arc cold-hearth (PACH) melted ingots of IMl-318 (Ti-6-4), Ti-48AI, Ti-52Al, Ti- 48Al-2Nb-2Mn and Super-a2 (Ti-25Al-10Nb-3V-1Mo) were produced, as discussed before4, to provide billet material for subsequent thermal processing, powder production and consolidation and spray-formed deposits. Powders of IMl-318 (Ti-6-4), Ti-52AI and Super-a2 have been produced by CWIBP. The Ti-aluminide alloys, Ti-48AI, Ti-52AI, Ti-48Al-2Nb-2Mn and Super-a2 have been spray formed onto sheet and tubes. The spray system utilises the same CWIBP crucible that was used in powder production to provide a controlled, molten stream of metal to a scanning atomiser for the production of flat or drum preforms or to a centrifugal atomiser (CSD) for the production of rings. A photograph and schematic of the CWIPB device is shown in Figure l(a) and (b). The operation of this process has been discussed before5. The effects of various operating parameters on the spray deposit product will be presented. The chemistry, homogeneity and microstructure of the as sprayed material has been examined and compared with material made through both the PM and JM routes. The plasma-melted, as-cast and heat-treated microstructures of Ti aluminides have been described elsewhere6,7.

ATOMISATION

Alloys of IMI-318 (Ti-6-4), y-Ti (Ti-52Al) and Super a2-Ti were atomised. CWJBP served as the melt source to provide molten metal to the atomisation die for these alloys. Metal 988 temperature was measured using an IRCON, two-colour pyrometer. Superheat of - 30°C was obtained when melting IMI-318 (Ti-6-4) and ... 20°C for most of the Ti-aluminides. When the charge is completely melted, the nozzle is heated and molten metal begins to flow when the metal directly above the nozzle melts. Low powder yields, due to nozzle blow back and stream instability, were experienced during these early runs with lots of 3 to 6 kgs being produced from 25 kg charges (6 to 8 kgs retained in skull). Metal to gas flow rates of about 1:1 were planned. The details of the atomisation system has been described in detail before3. Shown in Table 1 is the data from the first atomisation runs. Mean particle sizes for these runs were higher than expected. Based on the data collected from these runs, it was found that the over pressure compensation was incorrect, resulting in higher than expected metal flow and therefore, low gas to metal flow rates resulting in larger powder particle size. Nozzle blockage by metal blow back was a main contributor to low yields. It is anticipated that, a change to an lower nozzle angle (<15°) will correct the blow back problem.

Table 1. Operating conditions from the initial atomisation runs.

RUN ALLOY CHARGE SKULL GAS/MET YIELD MEAN NOZZLE JET # WEIGHT WEIGHT RATIO• %*• SIZE® DIAMETER ANGLE l IMl-318 25 kgs 8.1 kgs 0.72 5.3 123 µm 5.0 mm 7.5° 2 IMl-318 24 kgs 7.9 kgs 0.41 2.0 133 µm 4.0 mm JOO 4 Ti-52AI 14 kgs 4.5 kgs 0.86 41.6 158 µm 5.0 mm 15° 5 Super-a2 27 kgs 7.2 kgs 0.62 9.3 144 µm 5.0 mm 15° •Gas I Met Ratio= kgs min-I Ar Gas+ kgs min-I Metal, ••Yield%= Total weight powder< 250 µm produced + Weight of metal poured through atomiser, @Mean size is at the 50% cumulative weight for all powders < 250 µm.

IMI-318

Table 2: Ti Alloy Powder 02 and C Levels from the Initial Atomisation Runs as a function of particle size.

RUN # ALLOY 45-90 µm 125-180 µm 45-90 µm 125-180 µm Oxv2en ppm Oxv2en ppm Carbon wt% Carbon wt% 1 IMl-318 2900 2600 0.08 0.06 2 IMl-318 4100 2500 0.08 0.06 4 Ti-52AI 1050 900 0.02 O.Dl 5 Super-a2 1850 1000 0.09 0.04

A TEM micrograph of IMl-318, ... 50 µm powder, Figure 2(b), shows a martensitic structure, common in rapidly solidified conventional Ti alloys. It was noted that the microstructures coarsen progressively from the surface to the centre of the particle. This observation agrees with the radiation cooling model presented by Sastry, et. aJ.8, further

989 discussion of this topic is outside the context of this paper and will be presented later. The microstructure was devoid of second phase particles (e.g., TiC present in Ti alloys contaminated by C) and similar to other reports on Ti with low levels of C9. This is to be expected, since the C concentration in the powder was less than the solubility limit in Ti.

2a 30µm 2b lµm Figure 2: a) SEM micrograph of IMI-318 powders produced by CWIBP gas atomisation. b) TEM micrograph showing martensitic microstructure of IMI-318 gas atomised powder 10.

Ti-52AI Powder The Ti-52Al powder, which was the best atomisation run (41.6 % yield), contained low C (0.01 to 0.02 wt.%) and 02 (900 to 1050 ppm). These results indicate very low interstitial pickup during atomisation (See Table 2), considering that the billet material was PACH double­ melted with blend levels at 600 ppm 02 and O.Olwt.% C.

5.0µm 3a 20.0 µm 3b Figure 3: a) SEM micrograph ofTi-52AI as-atomised powder. b) TEM micrograph ofTi-52AI as-atomised powderlO.

Typical microstructures of gas atomised powders of Ti-52AI are shown in figure 3(a) and (b). As can be seen, the microstructure is dendritic, with the dendrites being Ti3AI and the interdendritic regions being TiAI. These two phases have been identified by microdiffraction and X-ray studies''· When hot isostatically pressed (HIP'd) at 950°C for 4 hours under a pressure of 100 MPa, the dendritic microstructure transforms to equiaxed grains of TiAI. It is important to note that the microstructure consists of only one phases, namely TiAI, and that no carbides or nitrides, such as Ti2AI(C or N) or TiC, are present.

Super-alpha-2 powders The super-a2 powders had slightly higher levels of 02 and C than did the Ti-52AI powder, as shown in Table 2. It has been reported that 02 content has a significant effect on the beta­ transus of this alloy12, therefore, maintaining low 02 interstitial content is very important when processing this alloy. The powder produced in this run contained higher 02 levels than would 990 be desired. These powders were most likely contaminated due to a burnt o-ring that failed due to the nozzle blockage. 02 levels similar to the Ti-52Al or better should be attainable. A TEM micrograph of the super-a2 powder, Figure 4(a), shows APB's indicating an ordered structure. In Figure 4(b), a SEM micrograph of a 60 µm powder cross-section shows the distribution of primary a2 precipitates distributed through a fl matrix. These observations are similar to other work in progress13.

Figure 4: a) TEM micrograph of super-a2 as-atomised powder particle showing APB's, b) SEM of a cross-section of a 60 µm as-atomised super-a2 powder particle.

SPRAYFORMING

The sprayforming part of the system provided by Sprayforming Developments Limited (SDL), Swansea, U. K., is a novel three in one system. The three configurations of the system, shown in figure 5, allow the formation of Ti spray deposits in the form of thick sheet, tubular­ sprayed on outside of a drum and tubular-sprayed on the inside of a mould. The first two methods use a pneumatic scanner coupled with a gas atomiser to spray metal. Spray formed sheet is produced by moving a flat substrate under the spray in a direction perpendicular to the scanning motion. Tubular shapes are formed by rotating a drum under the spray with the rotational direction perpendicular to the scanning motion. Spraying inside of a mould is accomplished by centrifugal spray deposition (CSD). CSD consists of a rotating disk which sprays molten metal on the inside of a mould. The mould is programmed to move up and down as the spray deposit builds up on the inside surface.

Figure 5: Photographs of the spray forming systems, sheet and tubes are formed using a scanning atomiser and tubular shapes are formed from Centrifugal Spray Deposition (CSD).

Drum Preform The first spray-forming operations at the IRC were performed using Ti-aluminide ingot material (150 mm diameter), previously double-plasma melted, scanning onto a rotating drum preform. Objectives of these trials were to establish operating parameters. The data from Ti­ aluminide spray-forming experiments (1 to 3) are shown in Table 3. A more detailed description of the spray forming operation has been presented elsewhere3. An example of a deposit is shown 991 in figure 6(a), this preform was produced during run #1, scanning Ti-48AI onto the drum substrate, shown in Figure 5.

Table 3: Data from the first three spray-forming experiments using the 200 mm diameter drum substrate.

RUN ALLOY CHARGE SKULL GAS/MET AT/SC DRUM NOZZLE SUB. # WEIGHT WEIGHT RATIO• GAS .. SPEED DIA. PRE-HT 1 Ti-48Al 12 kl!S 3.5 kl!S 0.19 2.1 300 rpm 5.0 mm NONE 2 Ti-48-2-2 16.5 kgs 4.2 kgs 0.20 5.6 300 rpm 5.0 mm 200°c 3 Ti-52Al 16.5 kgs 11 kgs 0.30 6.2 100 rpm 4.0 mm 25o0 c •Gas I Met Ratio= kgs min-1 Ar Gas+ kgs min-1 Metal,•• AT/SC Gas= Atomiser Gas Flow (kgs min-1) + Scanner Gas Flow (kgs min-1)

Figure 6: a) Drum deposit of Ti-48AI, b) CSD deposit of Ti-52AI. Note the variation in size and surface texture.

The optical micrographs shown in figure 7, are examples of the microstructures obtained in Runs# 2. The deposition can be divided into three regions. The first is a splat layer where the first particles run together on the surface of the substrate (Figures 7(a)). Cooling is faster in this region, resulting in high porosity and evidence of many surviving particle boundaries. Region II begins as this layer builds-up, temperatures increase at the surface, allowing arriving particles to splatter and flow together before solidifying (Figures 7(b)). In region II the optimum spray conditions are obtained, which are characterised by a uniform microstructure and low porosity. The transition to region III (Figures 7(c)) begins when the metal flow rate becomes unsteady and breaks-up near the end of the run. Larger particles, at a wider size distribution, are sprayed during this time as the spray conditions deteriorate.

a 200 µm b 200 µm c 200 µm Figure 7: a) Splat-Substrate interface from Run #2 Ti-48-2-2 showing Region I, b) Run #2 - Region II, c) Run #2 - Region III.

992 Interstitial chemistries from the first two spray forming experiments (Table 4) were similar to those obtained from ingot material, as discussed above. 02 and C concentrations, in the spray-formed deposits, are only slightly higher than in the double-melted plasma ingot material.

Table 4: Interstitial chemistry Data from Spray-Forming Runs 1 and 2.

RUN ALLOY DEPOSIT OVERSPRAY OVERSPRAY DEPOSIT DEPOSIT # THICKNESS OXYGEN CARBON OXYGEN CARBON 1 Ti-48Al 4 mm 1500 ppm 600 oom 900 ppm 200 oom 2 Ti-48-2-2 6 mm 3400 ppm 1000 ppm 850 ppm 200 ppm

Centrifugal Spray Deposition (CSD) CSD is of particular interest in the spray forming of Ti alloys, since it can be operated under vacuum to minimise entrapped gas which would be detrimental for these alloys in rotating parts. The use of a centrifugal disk to atomise and distribute particles precludes the need for atomising and scanning gases. As of the present, the IRC system is not configured to operated under vacuum, but could be modified in the future. The experiments that have been run using CSD, indicate that higher deposition rates than with gas scanning are possible. Also, initial results showed that the porosities are much lower in CSD deposits than in the gas scanned material. The substrate-deposit interface produced by CSD is superior to the gas scanned devices, as shown in Figure 6(b). A super-02 alloy was sprayed by this method with the resultant microstructures shown in Figure S. There is minimal porosity in this deposit, and the grain structure is relatively coarse (compared to PM) but smaller than cast material and more uniform. The grains size is uniform from substrate surface to free surface (Compare Figure S(a) to S(b)). Note that the microstructure is devoid of porosity.

Sa 1.0mm Sb lOOµm Sc 1.0mm Figure S CSD spray formed Super-02. a) Micrograph of the deposit showing the substrate­ surface interface, b) Higher magnification from the same area as (a) showing fine precipitates of primary 02, c) The deposit free surface.

Plate Preform The flat preform offers a source of thick billet-like material. This configuration is also the easiest to operate. The initial experiment using the plate preform consisted of spraying Ti-4SAI alloy. The scanning conditions were such that a deposit of about 2 cm thick was built up on the 20 x 25 cm plate, with little oversprayed deposit. Metal flow rate for this run approached a singular value (6 kgs/min) due to improved overpressure controls. The plate was transversed under the spray three times at a speed of - 40 cm/sec. The microstructures shown in Figure 9 give a cross-sectional view from near the centre of the deposit. The deposit has three dense areas separated by bands of porosity. The porosity is present at the interface of the spray passes. One can infer from this that a singular pass or faster transverse speed may improve the deposit density. Mechanical test were being performed at the time of this writing.

993 Figure 9 a) Cross-section from the Ti-48Al sheet deposit, b) The substrate deposit interface, c) Porosity at the pass interface, d) Bulk microstructure showing the lamella y and a2.

CONCLUSIONS

It has been show that it is possible to spray form Ti-aluminides into various shapes without compromising the alloy's interstitial content. High density preforms have been able to be produced. Porosity will be a challenge to prevent in traditional gas-scan spray forming but with CSD, operating under a vacuum, a totally dense material may be achievable after thermo­ mechanical processing (i. e., HIP'ing). Grain size refinement has been shown to be much better than IM. Spray forming of Ti-aluminides presents an economical alternative to PM processing.

ACKNOWLEDGEMENTS

The author would like to thank the SERC for providing funds through the foundation grant with the IRC at The University of Birmingham, Edgbaston, England. Many thanks to the people at IMI for their contributions of time and materials. The assistance of the !RC staff, especially T. Perry, J. Boots, M. Glywn and A. Fellows made this project possible. Special thanks to T. Johnson and K. O'Rielly for their contributions.

1 Advanced Materials & Processes, ASM Publication, February 1992, p. 9. 2 J. L. Murray, "Binary Titanium Alloy Phase Diagram Compilation," Final Technical Report to the Office of Naval Research, Project No. NOOOl 4-85-F-0069, 1985. 3 M. J. Kaufman, D. G. Konitzer, R. D. Shull and H. L. Fraser, Scripta Metallurgica, Edited by Pergamon Press pie., Vol. 20, p. 103-108, (1986). 4 J. W. Sears, "Clean Melting Techniques for the Production of Titanium Alloy Powders and Spray Deposits", Proceedings: Advances in Powder Metallurgy - 1991, by APMI, Chicago, IL, Vol. 6, p. 335- 346. 5 J. W. Sears, "Titanium Spray Formed Structures'', Proceedings: 1992 PM World Congress, by APMI, San Francisco, CA, June 21- 26. 6 R. V. Ramanujan, I. P. Jones, J. M. Young and J. W. Sears. "The Influence of Heat Treatment on the Microstructure of Plasma-Melted g7based Titanium Aluminides," This conference. 7 C. Huang, T. A. Dean and Z. Chen, "The Influence of Thermomechanical Processing on the Microstructure and Properties of Plasma Melted Ti Aluminides, "This conference. 8 S. M. L. Sastry, T. C. Peng and B. London, "Rapidly Solidified Ti-Alloy Powders Produced by Plasma­ Arc-Melting I Centrifugal-Atomisation (PAMCA)," Advances in Powder Metallurgy -1989. Editors: T. G. Gasbarre and W. F. Jandeska, Jr., MPIF, Princeton, N.J., 1989, Vol. 3, p. 387-400. 9 J. W. Sears, et. al., "The Production of Rapidly Solidified Ti Alloy Powers"; Processing of Structural Metals by Raoid Solidification, Editors: F. H. Froes and S. J. Savage, ASM International, Metals Park, OH, 1987, p. 223-229. 10 T. T. Cheng, Work in progress at IRC, University of Birmingham, Edgbaston, England, 1992. 11 G. ltoh, Work in progress at IRC, University of Birmingham, Edgbaston, England, 1992. 12 A. Szaruga, M. Sajib, R. Omlor and H. A. Lipsitt, Scripta Metallurgica, Edited by Pergamon Press pie., Vol. 26, p. 787-790, (1992). 13 Z. Chen, Private conversation, University of Birmingham, Edgbaston, England, 1992. 994