Distribution Behavior of B and P during Al-Si Melt Directional Solidification with Open-Ended Crucible

Effect of directional solidification

Figure 4 shows the vertical cross-section of the ingots and the macrostructure of the marked areas. The mass fractions of the primary Si flakes in five different parts from the bottom to the top of each ingot are shown in Figure 5.

Figure 4:

Photographs of vertical cross section of solidified Al-Si ingots. (a) Al-20Si; (b) Al-30Si; (c) Al-40Si; and their corresponding optical microstructures (d), (e) and (f) of the marked area.

Figure 5:

Mass fraction of primary silicon flakes in different parts of ingots.

It can be seen in Figure 4 that the primary Si phases in Al-20Si (wt %) alloy are mainly in grainy shape, but the primary Si phases in Al-30Si and Al-40Si alloys are plate-like. The needle-like primary Si phases at the bottom and the edge of the ingots indicate that the nucleation commences from these areas. In Figure 5, the content of the primary Si flakes increases with increase of Si content in these three samples and decreases from the bottom to the top in each ingot, which was specifically discussed in our early publication [22]. In addition, the amount of the impurities B and P in the primary Si flakes increases from the bottom to the top as shown in Figure 6. All the above indicates the impurities removal mechanism of directional solidification: Firstly, the primary Si flakes forms at the bottom end that has lower temperature, and then the impurities are rejected to the eutectic Al-Si melt when the primary Si flakes grow. Finally, the directional solidification drives the impurities to the top of the ingots.

Figure 6:

Distribution of B and P in five different parts from bottom to top of ingots. (a) B in primary Si flakes; (b) B in eutectic Si powders; (c) P in primary Si flakes; and (d) P in eutectic Si powders.

There are interfaces between the parts with dense and sparse primary Si phase, which are marked by dotted lines in Figure 4. The same phenomenon is shown in Figure 5, in which the mass fraction of the primary Si flakes drops at the interfaces. The part with dense primary Si phases approximately accounts for 60 % of the ingots in Al-20Si and Al-30Si and 80 % in Al-40Si. Furthermore, the distribution of the primary Si flakes in Figure 5 may be described by directional solidification theory [23]. At the beginning of the solidification, the primary Si flakes form and grow at the bottom of the ingot. When the solidification continues, the Si concentration ahead of the solid/liquid interface decreases, forming a concentration gradient. Then the fraction of the primary Si phases in the solidified part decreases until the composition of melt reaches the eutectic point. Finally, the concentration of the alloy remains at the eutectic point till the end of the solidification. However, the fraction of the primary Si phases at the top of the ingots does not reach zero in Figure 5. Many primary Si grains can be found in the top area of the ingots as shown in Figure 4. This is caused by the influence of the impurities, oxides and protective Ar atmosphere.

Distribution of impurities elements

To obtain distribution of B and P in the Al-Si alloys, the concentration profiles of B and P in the primary Si flakes and the eutectic Si powders along the ingot growth direction are constructed from the data of ICP-OES analyses, which is shown in Figure 6.

The concentration of B in the MG-Si raw materials is 25 ppmw. After the Al-Si solvent refining, the B contents in the primary Si flakes were generally less than 10 ppmw, and the content of B in Al-30Si is the lowest in these three samples. This might be related to the TiB₂ particles formed during the solidification process, which needs further study. However, for the eutectic Si powder, B contents were below 10 ppmw only in Al-20Si, all others were about 20 ppmw, which have only little purification effect. In addition, the content of B in the primary Si flakes and the eutectic Si powders fluctuates without obvious trend, which indicates that the directional solidification does not strongly affect the distribution of B.

On the other hand, the concentration of P gradually increases from the bottom to the top of the samples, which reflects the effect of the directional solidification. And it is obvious that the content of P increases gradually when Si content decreases from 40~20 % in Al-Si alloys. The concentrations of P are 43 and 26 ppmw in the MG-Si and the Al of the raw materials, respectively. However, it reaches above 43 ppmw in Al-20Si. This is due to the formation of P containing particles and trapping of these particles by the growing primary Si phases at the low solidification temperatures, which is suggested by Li [21] and Ban [24].

The contents of B and P in the primary Si flakes and the eutectic Si powders are less than that in raw materials MG-Si. It can be concluded that B and P are segregated to the Al-Si melt during the forming and growing of the primary Si flakes, and P gradually segregates to the top of ingots during directional solidification. In addition, the contents of B and P have different trends with both Si content in the Al-Si alloys and directional solidification. This could be caused by the following mechanism. The raw materials MG-Si contain some metal element Ti. AlP and TiB₂ particles can form before the primary Si formation during the refining process [24, 25]. The AlP particles can act as nucleation sites for Si phase growth [24], but the TiB₂ particles cannot [25]. So, the alloys with lower solidification temperature contain higher amount of P than B.

Segregation coefficient

An apparent segregation coefficient, k_a, is used to characterize the segregation behavior during the directional solidification process. Assuming that there is a complete mixing of elements in the Al-Si melt and no diffusion in the solidified primary Si flakes, which meets the basic requirement of Gulliver-Scheil model. The content of B and P in the solidified Si can be described by Scheil equation:

where C_s, C₀ and f_s are the impurity content in the primary Si flakes, the initial impurity composition in the melt and the fraction of solid, respectively. Considering the law of mass conservation that the total amount of solute in the solid must be equal, the formula is obtained by integration of eq. (1)

then,

where Cs‾ and f_s are the average composition of the impurity element in the primary Si and the fraction of the primary Si in the alloy which can be obtained from Al-Si phase diagram, respectively. In this work, the primary Si flakes mainly gather at the bottom 3/5 of the ingots as shown in Figure 4, so Cs‾ could be calculated from the first three data points of impurities in Figure 6.

The apparent segregation coefficient of B and P (k_aB and k_aP) can be obtained from eq. (3), the results are shown in Figure 7. It can be seen that the apparent segregation coefficients of B and P decrease when the content of Si increase in the Al-Si alloys. On the other hand, the solidification temperature of the alloys rises with increase of Si content in the Al-Si phase diagram (Figure 2(b)). So it can be concluded that apparent segregation coefficients increase with decrease of solidification temperature of the alloys in this work.

Figure 7:

Segregation coefficient of B and P.

Yoshikawa [8, 9] reported that the segregation coefficient of B and P can be expressed by the following equations:

where k_B and k_P are the segregation coefficient of B and P, respectively, T is the absolute temperature. For the solidification discussed before is a non-isothermal process, so k_a is actually an average of segregation coefficient kˉ of the solidification temperature range, which can be expressed as:

where T₁ is the liquidus temperature, T₀ is the eutectic temperature, and k(T) is the k_B or k_P function of temperature, which is shown in eqs (4) and (5). Bring eqs (4) and (5) into eq (6), the equilibrium segregation coefficient of B and P in Al-Si alloys could be worked out. The comparison with the apparent segregation coefficients of B and P is shown in Figure 7.

The apparent segregation coefficients of B and P are much larger than the equilibrium coefficients and the apparent segregation coefficients of B and P in Al-20Si are even greater than 1. The reasons of the huge numerical difference are as follows. Firstly, the solidification in this work is non-equilibrium and the raw material used in this work is MG-Si and commercial Al. Secondly, the raw material contains high concentration of metal elements such as Fe, Ti and Ca, which would combine to form specific compounds, such as AlP [24, 26] and TiB₂ [25, 27]. Some of these intermetallic compound particles can form before Si phase growth and act as nucleation sites for Si phase growth. Finally, the solidification rate in this work is too fast to let the growing primary Si phases push the compound particles out from them. The compound particles can be trapped easily in this kind of growth condition [28].

In addition, the trends of the apparent segregation coefficients and the equilibrium ones are contradictory. This indicate that kinetic factors have great influence on the apparent segregation coefficients of B and P during directional solidification of the Al-Si alloys, particularly more prominent at the low solidification temperature side. It can be explained by mechanisms of diffusion and convection during the solidification. The schematic drawing of electromagnetic stirring is shown in Figure 8. The viscosity (μ) of the Al-Si melt and the diffusion coefficient (D) of B and P correlate with temperature (T), as shown in eq. (7) and (8).

Figure 8:

Schematic drawing of electromagnetic stirring mechanism.

where T is the absolute temperature, μ₀ and D₀ are coefficients, E_μ and E_D are the activation energies and R is the universal gas constant. When the temperature increases, the viscosity of the Al-Si melt decreases and the diffusion coefficient of B and P increases. Thus the amount of convection and diffusion in the melt would be raised. In the hypereutectic Al-Si alloys, when the Si content increases, the liquidus temperature rises. In the meantime, the heating power and therefore the amount of electromagnetic stirring increase, so the amount of convection and diffusion increase with increasing Si content. As a result, the impurity elements B and P can be segregated out of Si effectively when Si content of the Al-Si alloys increases.

Li [20, 21] calculated the apparent segregation coefficient of B and P in Al-30Si alloys at the cooling rate of about 1 mK/s. It is found that the apparent segregation coefficient of P increases with decrease in average solidification temperature, which agrees with the trend in this work. Ban [16] calculated the apparent segregation coefficient of B and P in Al-30Si alloys at the cooling rates of 8.33 and 25 mK/s, respectively. The apparent segregation coefficient of B and P in Al-30Si at different cooling rates are summarized and shown in Figure 9. The apparent segregation coefficients of B and P increase by logarithmic relationship with increase of cooling rate. This also indicates that kinetics has great influence on the apparent segregation coefficients of B and P during silicon purification by the Al-Si solvent refining method. In the Al-Si alloys, B and P need a long time to diffuse from the silicon phase surface to the melt during the formation and growth of the primary Si flakes. Thus the slower cooling rate allows more impurities segregate from the growing primary silicon flake surface to the remaining Al-Si melt.

Figure 9:

Effect of cooling rate on apparent segregation coefficient.