Evolution of Globular Microstructure and Rheological Properties of Stellite™ 21 Alloy after Heating to Semisolid State

Semisolid processing of metal alloys usually consists of two stages: (1) the initial preparation of the feedstock material, usually in the form of a billet, with a near-spheroidal microstructure in the solidus-liquidus range and (2) forming the final near-net-shape product by thixoforging or thixocasting process. A number of methods have been proposed for the preparation of non-dendritic feedstock materials (Ref 14, 19, 20), with the magneto-hydro-dynamic stirring (MHD) method being the most commonly used commercial route for light metal alloys, and sprayforming, although rather expensive clearly applicable for some high value alloys for which other methods do not work. Furthermore, two other methods were developed based on thermo-mechanical treatment of alloys, directly applicable to wrought metal alloys. One is the strain-induced melt-activated (SIMA) method suitable for industrial applications because it is relatively inexpensive and offers an easy way to obtain the non-dendritic feedstock by just heating the highly strained feedstock to the semisolid state. The feedstock is hot-worked material, e.g., rolled, forged or extruded in the solid state above recrystallization temperature and then cooled; such feedstock develops fine globular microstructure after being heated up to the semisolid state. In this study, we used a combination of sprayforming and SIMA methods. It is worth noting that a similar method to the SIMA, the recrystallization and partial melting (RAP) route is another possible way to develop suitable microstructures (Ref 21-23), where the starting material is warm as opposed to hot worked (Fig. 3).

The first stage of the experimental work involved a high-temperature DTA analysis of the Stellite™ alloy being studied; this analysis was conducted using STA JUPITER 449 analyzer produced by Netzsch. The DTA heating curve and the liquid fraction distribution as a function of temperature for Stellite™ 21 alloy are shown in Fig. 2. The measurements were taken at a heating rate of 5° per minute. It should be indicated that applied heating rate is relatively low in order to obtain conditions close to equilibrium state. In real processes, kinetics of melting process depends on the heating rate which is rather higher than during DTA analysis.

Samples were cut from sprayformed and hot-extruded Stellite™ 21 alloy billets. Their microstructure is shown in Fig. 4, to demonstrate the state of the extruded material. The samples were etched in the mixture of 15 ml of HCI, 10 ml of acetic acid and 10 ml of HNO₃ for 2 min, but this process did not allow the structure resulting from the extrusion process to be clearly observed. Micrographs did not reveal any grain boundaries, but irregular interdendritic and intergranular carbides could be seen, visible as light globules, with an average size of one or two microns; the black parts indicate intergranular lamellar eutectic and the gray the actual grains. The average alloy composition includes (in weight percentage): 61.36%-Co, 27.10%-Cr, 5.69%-Mo, 1.65%-Fe, 2.93%-Ni, 0.59%-Si, 0.25%-C, 0.25%-W, 0.18%-Mn. The microstructural morphology was confirmed on the basis of chemical analysis of selected areas using energy-dispersive spectroscopy (EDS); using FEI E-SEM XL30 with EDS detector EDAX Genesis 4000. These areas are shown in Fig. 5 and the chemical analysis in Table 1. It is easy to recognize that the alloy microstructure of Stellite™ 21 consists of Cr-rich Co solid solution and two types of Cr-rich carbides. The EDS results together with x-ray pattern (Fig. 6) show that the Co solid solution is a mixture of γ-fcc and ε-hcp phases, and two types of carbides Cr₇C₃ and Cr₂₃C₆. The Cr₇C₃ carbide is eutectic with the Co solid solution, while the Cr₂₃C₆ carbide is a precipitate with fishbone shape.

Samples of approximately 7 × 7 × 10 mm dimensions were suspended on Kanthal wire in a resistance tube furnace and heated to a temperature of approximately 1350 °C. To obtain the expected amount of liquid fraction in the semisolid state, the soaking temperatures were taken from the DTA analysis, with temperatures of 1320, 1340, 1360 and 1375 °C corresponding, respectively, to 1%, 5%, 20% and 40% liquid fractions, in accordance with Fig. 2. The samples were soaked at temperature for 10 min and then rapidly quenched in a water tank placed below the furnace. During the heating of the samples, a protective atmosphere of argon gas was used to reduce oxidation. After quenching, the samples were examined by optical and scanning electron microscopy in combination with energy-dispersive spectroscopy. The samples for SEM investigations were polished using diamond powder, and no etching was applied.

The microstructures taken at various temperatures, which correspond to different amount of the liquid phase in the semisolid state, are shown in Fig. 7. Temperature variations from 1320 to 1375 °C approximately caused variations of liquid fraction from 1 to 40%. The resulting microstructures appear to be composed of regular globular particles embedded in the eutectic formed from the liquid phase, and these microstructures appear suitable for thixoforming. The microstructures were obtained using optical microscopy.

The composition of the material was investigated for both the spheroidal solid particles and their surroundings using the standard EDAX ZAF quantification method, and the results are summarized in Table 2. The areas where composition was analyzed are shown in Fig. 8. They total 100%; however, only elements with a content of over 0.2 wt.% are shown. In both cases, the beam scanned an area of 20 μm × 20 μm, and depth of scanning is assessed as equaling some 1.6 μm, and therefore the measurement gave an average chemical composition in both the globular solid and the low melting phase. No oxygen content was observed inside the globular solid and inside the low melting phase, as determined by means of EDS. These results confirm that oxidation can be avoided during thixoforming of high melting point alloys through appropriate protective gas atmospheres, such as Ar or N₂ (Ref 9). The average composition of the alloy grains after application of the SIMA method is almost the same as grains from as-received material. But the lamellar eutectic is richer in cobalt, mainly due to the higher amount of γ-fcc phase, which occurs under equilibrium conditions. The weight percent of chromium and carbon in area of lamellar eutectic corresponds to stoichiometric Cr₇C₃ carbide. In the sprayformed as-received alloy, the solidification process proceeds under non-equilibrium conditions, and the carbon content does not correspond to the phase diagram.

XRD patterns of Stellite ™ 21 before and after heating have been shown in Fig. 6 and 9. Figure 6 shows that the as-received alloy includes both γ(fcc) and ε(hcp) phases; the γ-Co phase being mainly 〈111〉 and 〈100〉 textured and the ε-Co phase mainly 〈0001〉, 〈1010〉 and 〈1011〉 textured. The presence of ε-Co phase results from the strain-induced martensitic transformation caused by the extrusion process that the investigated alloy was subjected to. After the heating tests (Fig. 9), the γ-Co phase is mainly observed with fcc structure. The signal of ε-Co phase is very low, which indicates that the strain-induced phase transformation under mechanical stresses proceeded to a small extend. This transformation can occur probably under rapid quenching of samples after aging at high temperatures.

Although the globular microstructure of the material can be examined by optical microscopy, SEM offers higher resolution images, allowing the resolution of individual carbide particles to be achieved, something that cannot be obtained with optical micrographs. Furthermore, additional information on the composition of phases in multicomponent alloys, for which phase diagrams are not easily available, may also be obtained when using EDS. The control of composition is especially critical where samples are subjected to external forces or non-uniform heating.

The rheological analysis was carried out using the high-temperature viscometer FRS 1600, designed by the Anton Paar company, with a furnace that allows measurements in the temperature range 400-1500 °C. Rheological analysis of Stellite™ 21 alloy in the semisolid state requires work around the 1300-1450 °C temperature range, and the viscosity measurements were performed using Searle’s method (Ref 24), carried out using a rotational viscometer with a stationary cup (outer cylinder). In this method, the rod is rotated and the cup is stationary, and the cylinders are concentric, i.e., axi-symmetric with the rotation axis of the inner cylinder. The outer and inner cylinders have 28 and 15 mm in diameter, respectively. They were made from alumina (Al₂O₃). In order to prevent wall slippage on the rotating rod and the crucible, their lateral surfaces have profiled, serrated shape. During experiment, the furnace chamber was blown by the argon gas in order to prevent samples oxidation. Additionally inside furnace chamber, the graphite element was placed to bond rest of oxygen. Before the measurement, the sample was melted to liquid state to place the rotating rod (inner cylinder) inside the crucible (outer cylinder). Next the sample was cooled to assumed temperature, and the viscosity measurement was taken. All the time during temperature changes, the sample was slowly sheared with the rate equals 5 s⁻¹.

As a first step, measurements were concerned with the determination of the relationship between the alloy viscosity and the temperature (Fig. 10). As a basic rule, decrease in temperature causes increase in registered viscosity, especially in the case of metal alloys in the semisolid state. The highest viscosity increase appears for temperatures below 1365 °C. These experiments confirmed results of DTA analysis and allowed to identify the appropriate temperature range for thixocasting of Stellite™ 21 alloy approximately between 1370 and 1390 °C.

In the second step of experimental work an analysis of Stellite™ 21 viscosity in the semi solid state versus the shear rate was carried out. The shear rate ramp was applied in order to control its changes in the range from 0.1 to 20 s⁻¹ during measurements. In the range of higher values of the shear rate, the viscosity changes are much less, then in the range of lower values of the shear rate. Moreover, the measurement of viscosity for higher shear rate could follow in the wake of turbulent flow inside the crucible. Figure 11 shows the viscosity curves versus shear rate at the above mentioned temperatures in the semiliquid state for the alloy tested. Before each viscosity measurement, during change in the temperature, the samples were sheared with a rate of about 5 s⁻¹. The temperature changed with rate of 1° per minute. The shape of the curves describing the relationship between the viscosity and the shear rate indicates the shear thinning behavior of the semisolid Stellite™ slurry, i.e., its viscosity decreases with an increasing shear rate.

Numerical simulations, using for example the CDF (Computational Fluid Dynamics) method require usually the viscosity models in the form of mathematical equations. One of them is Carreau-Yasuda equation, used very often in the software developed for computer simulations of the casting processes. It should be mentioned that the Carreau-Yasuda equation takes into consideration the shear thinning phenomenon, which occurs in the liquid metals, but this model does not allow the thixotropy phenomenon, which appears in semi-solid alloys, to be described.where \(\dot{\gamma }\) is strain rate, η ₀ zero strain rate viscosity, η _∞ infinite strain rate viscosity, λ phase shift, n power law coefficient and α Yasuda coefficient.

The obtained parameter values (shown in Table 3) of Eq 1 can be directly applied in the numerical simulations. The approximation was carried out using the gradient optimization method for minimization of differences between measured and calculated viscosity values. The correctness of the approximation of the viscosity curves could be observed in Fig. 12. The values of Carreau-Yasuda parameters versus temperature are also shown in Fig. 13.