Aluminium Loss During Ti-Al-X Alloy Smelting Using the VIM Technology

1


Introduction
Rapid advancements in technology, observed over the recent years, have mostly resulted from improvements in production technologies of various types of metallic, ceramic and composite materials with unique properties.Such materials are e.g.metal alloys based on intermetallic Fe-Al phases, nickel-based superalloys or titanium alloys.The last mentioned, due to their small specific densities, high corrosion resistance or hightemperature strength, are applied in e.g.aerospace, defence and automotive industries.Among titanium-based materials, the most commonly used types are Ti-Al-X alloys.However, their production technologies are associated with many difficulties due to three factors: a high smelting temperature, high reactivity of titanium with the crucible materials and a potential for evaporation of the alloy components with higher vapour pressures than that of the metal matrix during the smelting process [1][2][3][4][5][6][7][8][9][10][11].
To avoid the problem of titanium reactivity with crucible materials, cold crucible furnaces are increasingly being used.In this type of furnaces, crucible is usually made of water-cooled copper segments.The electro-dynamic interaction between the currents induced in the crucible and the charge causes the liquid metal to be pushed away from the walls of the crucible, and the charge only contacts its base..The problem of metal bath component loss during production of a specific alloy is particularly important when even a small change of their contents leads to altered alloy properties.In the present paper, aluminium evaporation from the Ti-Al-Nb, Ti-Al-V and Ti-Al alloys during smelting in a crucible vacuum induction melting furnace has been discussed.

Research methodology
The research experiments were performed on multicomponent alloys; their composition is presented in Table 1.The melting system applied in the experiments was a SecoWarwick VIM-20 vacuum induction melting furnace; its image is presented in Fig. 1.
Fig. 1.The Seco-Warwick VIM-20 vacuum induction melting furnace: 1 -the furnace chamber, 2 -the crucible, 3 -the induction coil, 4 -the door to close the chamber tightly, 5 -the measurement system, 6 -the ingot mould, 7 -the ingot mould heater This furnace type selection resulted from its being a device widely used for titanium alloy smelting.At the beginning of each experiment, an alloy sample (about 1000 g) was placed in a graphite crucible located inside the induction coil of the furnace.When the furnace chamber was closed, a precisely specified vacuum was generated using diffusive and Roots pumps.After stabilisation of the chamber pressure, the crucible was heated up to the set temperature.During each experiment, metal specimens were collected and analysed for the aluminium content.The experiments were performed at 10 to 1000 Pa for 1973 K and 2023 K.

Composition of the gaseous phase over liquid Ti-Al-X alloys
To determine the equilibrium composition of the gaseous phase over the investigated alloys, the thermodynamic database HSC Chemistry ver.6.1 was used [12].Estimated values of the equilibrium vapour pressure for Ti, Al, V and Nb over pure components are listed in Table 2, and the vapour pressures of these metals over the analysed alloys -in Table 3.They were determined based on the thermodynamic data regarding activity coefficients of the individual components of the analysed liquid titanium alloys included in the papers by Semiatin, Song, Zhu and Belyanchikov [13][14][15][16].Fig. 2 shows ratios of the titanium vapour pressures to the aluminium values over these alloys.The data in Fig. 2 demonstrate that for the range of 1923 K to 2173 K, the ratio is 0.01 to 0.05.This means the aluminium equilibrium pressures significantly exceed those for titanium and, therefore, there is a potential for intense evaporation of this component of the investigated alloys during smelting, from the thermodynamic point of view.

Experimental results and discussion
Overall results of all experimental melting processes for the Ti-Al-Nb, Ti-Al-V and Ti-Al alloys are summarised in Table 4.In addition to the experimental parameters, it contains the values of final aluminium content in the alloy, relative loss of this metal from the alloy and the aluminium evaporation flux.The Table 4 data show that the determined aluminium loss for all tested alloys increased as the operating pressure of the melting system decreased, and ranged from 4 % to 25 %.By analogy, the vacuum increase (1000 to 10 Pa) caused higher aluminium evaporation flux values.
While analysing the process of aluminium evaporation from liquid titanium alloys during their smelting in the vacuum induction melting furnace, three factors that may affect the rate of this process should be mentioned: the operating pressure in the melting system, the metal bath surface area and its stirring rate.There are four pressure ranges to alternately affect the evaporation process discussed [17][18][19][20][21].The first range is related to pressures that cause the metal evaporation rate to reach its maximum which remains stable as the pressure further decreases.So-called free evaporation is observed at that time during which the evaporating metal atoms or molecules that are leaving the metal bath surface do not collide with other gaseous molecules so their velocity is the same as it was while leaving this surface.The other pressure range concerns the values for which the process rate is practically controlled only by mass transfer processes in the liquid phase.When the pressure in the system is raised above this value, a change of the stage that determines the analysed process is observed.This refers to pressures of up to several hundred Pa.Within this pressure range, the evaporation rate is determined by mass transfer processes in both the gaseous and liquid phases.For higher pressures, diffusion control is present and the process rate is determined exclusively by the mass transfer in the gaseous phase.Such a process control is usually seen for pressures above 1000 Pa.
The value of evaporation flux of the liquid alloy component is directly proportional to the evaporation surface area i.e. the metal bath surface in the analysed case which, for the induction furnace, increases with higher power values.The higher power results in a considerably larger bath surface area due to meniscus formation which is the effect of electromagnetic field influencing the liquid metal.It is illustrated in Fig. 3  Simulations related to the effects of the operating electrical parameters of the induction furnace showed that increasing frequencies of the induction coil-supplying current were associated with higher mean bath stirring rates within the whole liquid metal volume as well as with higher near-surface rates [22].In addition, these values depend on the position of the crucible versus the induction coil.While changing the crucible location inside the induction coil, we alter the electromagnetic field distribution and, thus, the bath stirring rate, which is observed at the metal surface in particular.Greater distances of the lower bath surface from the lower induction coil edge (the crucible is not symmetrically placed against the induction coil) result in smaller stirring rates of the metal [23].Increased stirring rates may cause intensified evaporation during melting processes performed at the vacuum of less than several Pa.
The paper [24] presents results of a mathematical simulation allowing determination of the liquid titanium surface area and the mean bath stirring rate for the system used in the experiments discussed in this paper.To determine the velocity field, a model of coupled electromagnetic and hydrodynamic fields of liquid metal was applied.It should be noted that an essential problem that impedes modelling of this process is the fact that the electromagnetic field affects the hydrodynamic field of liquid metal bath, while changing its surface shape.Due to the shape changes that cause electromagnetic field distribution alterations, it is necessary to calculate the electromagnetic and hydrodynamic fields at each simulation step [25,26] or with a frequency sufficient for maintaining the current field distribution by means of the mathematical extrapolation [27,28].The numerically determined shape of the liquid titanium surface for the following model parameters is presented in Fig. 4: current intensity 738 A, current frequency 5 kHz, titanium resistivity (metal) 1.7e -6 Ω⋅m, graphite resistivity (crucible) 10 Ω⋅m and copper resistivity (induction coil) 0.018 Ω⋅m.The liquid titanium velocity field for the same melting system is presented in Fig. 5.The results of simulations performed for both the meniscus size and the stirring rate of liquid titanium are summarised in Table 5.The experiment-determined relationships between the aluminium evaporation flux and the operating pressure in the melting system for the investigated Ti-Al-Nb and Ti-Al-V alloys are shown in Figs. 8 and 9. Figure 10 presents determined conventional pressure ranges for both intense and minimum aluminium evaporation from the investigated alloys.These curves were plotted based on relationships proposed in the paper [11] where the authors demonstrated that for multi-component titanium alloys containing up to 23 % wt.Al, the value of allowable pressure pall could be determined as follows: p all = 5.5⋅p Al (1) while the critical pressure: p cri = 0.55⋅p Al (2) where: p Al -equilibrium pressure of Al over the alloy.
The range above the line pall corresponds to the pressure at which aluminium evaporation is negligible while the range below the line p cri corresponds to the pressures at which significant aluminium loss during Ti-Al melting processes in vacuum induction melting furnaces is observed.Fig. 10.Effects of melting temperature on the pall and pcri parameters for the investigated Ti-Al-X alloys

Fig. 2 .
Fig. 2. Effects of temperature on the p Ti / p Al value

Fig. 3 .
Fig. 3. Images of the surface of aluminium melted using various furnace operating power values: a) 11 kW; b) 37 kW

Fig. 5 .
Fig. 5. Liquid titanium velocity field [24] Table 5. Results of the simulations for both the velocity field and the surface area of liquid titanium melted in the VIM-20 furnace at 2173 K Maximum EM force 52.9 kN Mean EM force 11.7 kN Maximum titanium volumetric velocity 0.18 m/s Mean titanium volumetric velocity 0.06 m/s Titanium free surface area (meniscus) 0.0068 m 2 Surface area of the crucible cross-section 0.0060 m 2 Maximum titanium surface velocity 0.185 m/s Mean titanium surface velocity 0.135 m/s Figs. 6 and 7 present the experiment-determined relationships between relative aluminum loss from the Ti-Al-Nb, Ti-Al-V alloys and the operating pressure in the melting system.The parameter value for all the alloys ranged from 4 % to 25 %.

Fig. 6 .Fig. 7 .
Fig. 6.Effects of pressure on relative titanium loss from the Ti-Al-Nb alloy

Fig. 8 .Fig. 9 .
Fig. 8. Effects of the operating pressure in the induction furnace on the aluminium evaporation flux from the Ti-Al-Nb alloy

Table 1 .
Chemical composition of the investigated alloy

Table 2 .
Determined o i p values for titanium, aluminum, vanadium and niobium Alloy component

Table 3 .
Determined vapour pressure values for titanium, aluminum, vanadium and niobium over the liquid alloys Ti-Al-V, Ti-Al-Nb and Ti-Al

Table 4 .
Results of vacuum melting experiments for the Ti-Al-Nb, Ti-Al-V and Ti-Al alloys