Separation mechanism of the primary Si phase from the hypereutectic Al–Si alloy using a rotating magnetic field during solidification

Separation mechanism of the primary Si phase from the hypereutectic Al–Si alloy using a rotating magnetic field during solidification

Available online at www.sciencedirect.com ScienceDirect Acta Materialia 72 (2014) 57–66 www.elsevier.com/locate/actamat Separation mechanism of the ...

4MB Sizes 0 Downloads 8 Views

Available online at www.sciencedirect.com

ScienceDirect Acta Materialia 72 (2014) 57–66 www.elsevier.com/locate/actamat

Separation mechanism of the primary Si phase from the hypereutectic Al–Si alloy using a rotating magnetic field during solidification J.C. Jie a, Q.C. Zou a, J.L. Sun a, Y.P. Lu a, T.M. Wang a,⇑, T.J. Li b,⇑ a

School of Material Science and Engineering, Dalian University of Technology, Dalian 116024, Liaoning, People’s Republic of China Laboratory of Special Processing of Raw Materials, Dalian University of Technology, Dalian 116024, People’s Republic of China

b

Received 18 January 2014; received in revised form 13 March 2014; accepted 14 March 2014

Abstract Understanding solidification behavior under an intense flow field is important in controlling the microstructure and macrostructure of alloys in industry. In the present study, we show that using a rotating magnetic field (RMF) during solidification of hypereutectic Al–Si alloy can efficiently congregate the primary Si phase to the inner wall of the crucible and form a Si-rich layer with 65–698 wt.% Si content. The Al–Si melt flow under an RMF and the temperature field of the liquid metal are the two dominant conditions for the segregation of the primary Si phase. The intense melt flow, i.e., secondary flow and Taylor–Go¨rtler vortices, carries the bulk liquid with higher Si content to promote the growth of the primary Si phase formed close to the inner wall of the crucible where the temperature is low, finally resulting in the remarkable segregation of the primary Si phase. This work has demonstrated that a forced intense melt flow combined with proper cooling conditions can greatly change the solidification structure of alloys, which is beneficial to microstructure control. Ó 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Rotating magnetic field; Al–Si alloy; Separation; Solidification; Microstructure

1. Introduction It has been recognized for many years that fluid flow within a melt can have profound effects on the solidification structure. Whenever fluid flow is artificially applied during the casting of alloys, some changes in the microstructure can be observed [1–6]. One observation, for instance, is a transition from equiaxed to globular microstructure, dendrite growth or grain refinement, or macrosegregation [4,5]. Research under microgravity conditions in the last decades has shed light on the importance of fluid flow, and it became clear that even in the best experimental ⇑ Corresponding authors. Tel./fax: +86 411 8470 8940.

E-mail addresses: [email protected] (T.M. Wang), [email protected] cn (T.J. Li).

set-ups, residual flows can change the microstructure appreciably [7]. One of the common flows in a solidifying, electrically conducting melt can be generated by applying time-varying magnetic fields. The most prominent realizations of these are induction coils [8] in furnaces and rotating [9] or traveling [10] magnetic fields. Such fields are a powerful tool to provide a wide variety of flow patterns through which the solidifying microstructure can be tailored in situ. Considerable progress has been made in the past 10 years, in particular with rotating magnetic field (RMF)-driven fluid flow, by employing careful model experiments of low-melting-point alloys and adapted numerical simulations [3,4,11,12]. A standard case is the axisymmetric arrangement of a cylindrical liquid metal column exposed to an RMF [13,14].

http://dx.doi.org/10.1016/j.actamat.2014.03.031 1359-6454/Ó 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

58

J.C. Jie et al. / Acta Materialia 72 (2014) 57–66

As is well known, inhomogeneous solute distribution is accompanied by a gradual change in composition and is an important and well-known phenomenon during solute-rich alloy solidification, which includes macrosegregation and microsegregation. This may occur on various scales in the ingot; microsegregations are linked to composition variations on the scale of the lowest solid structure, whereas macrosegregations concern the scale of the product [15– 17]. Macrosegregation often results in the deterioration of the material properties. However, it has a great use in some cases, such as the preparation of materials with high purity. The hypereutectic and hypoeutectic Al–Si alloys are extensively used in the automotive and aerospace industries due to their low density, good castability, low thermal expansion and higher mechanical properties [8,18]. The effect of fluid flow on typical hypoeutectic Al–Si-based cast alloys is well understood, since in the last decade intensive research on that subject was performed, including experimental investigation and simulation methods [19–27]. For hypereutectic Al–Si alloys, because the primary Si phase is first formed and can move with the liquid metal, it is considered that a forced flow field may have a remarkable influence on the segregation behavior of Si solute. Hence, in the present study, we introduced an intense melt flow induced by RMF to greatly alter the solidification structure of hypereutectic Al–30Si alloy. The mechanism of segregation behavior of the primary Si phase during solidification under RMF was investigated by a series of experiments. 2. Experimental A hypereutectic Al–Si alloy with 30 wt.% Si was prepared by conventional casting from commercial Al (99.7 wt.%) and metallurgical Si (99.7 wt.%). First, the Al–Si alloy was melted in an electrical resistance furnace, and then poured into a preheated (840 °C) cylindrical graphite crucible with an inner diameter of 60 mm. The crucible was placed under an RMF with a frequency of 50 Hz, which was induced by a three-phase, three-pole magnetic generator. The magnetic flux densities in the present study were 12, 17 and 25 mT, respectively. After switching on the RMF, the Al–Si melt was stirred vigorously by the electromagnetic field. A large amount of the primary Si phase congregated close to the inner wall of the crucible during solidification. According to the experimental requirement, the thermal insulation material was placed at different positions of the graphite crucible to prevent the Al–Si melt from fast cooling and investigate the segregation behavior under variant cooling conditions. Fig. 1 is a schematic illustration of the experimental setup. As an example, asbestos was placed at the top of the crucible. To measure the liquid temperature, the heads of thermocouples 1, 2 and 3 with a distance interval of 15 mm were fixed 1 mm into the liquid metal close to the sidewall, and thermocouple 4, which measured the central liquid, was placed 3 mm above the bottom surface of the

Fig. 1. Schematic illustration of Al–30Si alloy solidified under an RMF, in which the thermal insulation material is placed at the top of the crucible. Thermocouples 1, 2, 3 and 4 were used to measure the temperatures of liquid metal.

crucible. All the thermocouples were connected to the temperature recording system. The prepared samples were cut to expose the vertical and cross-sections, and the microstructures were observed by an MEF-4A optical microscope. 3. Results Fig. 2 shows the vertical sections of Al–30Si alloy solidified under an RMF with different magnetic flux densities. It should be noted that the thermal insulation material was placed on the top position of the crucible during the solidification process, as shown in Fig. 1. Thus, the heat is released from the bottom and sidewall. For the alloy solidified without RMF, similar to the previous work [28], the primary Si phase tends to uniformly distribute in the alloy (Fig. 2a). Because the Si content is high and the size of Si is large, the primary Si phase can drop and form holes in the sample during the grinding process. The length of the primary Si phase is 2–6 mm due to the slow cooling rate. However, when solidified under RMF with different magnetic flux densities, a large amount of the primary Si phase is separated from the Al–30Si alloy and distributes in the periphery of the sample. In addition, the separation effect of 25 mT (Fig. 2d) is better than those of 12 mT (Fig. 2b) and 17 mT (Fig. 2c). Furthermore, the primary Si phase is still formed in the center of the alloy. In order to further exhibit the segregation of the primary Si phase, the half cross-sections of Al–30Si alloy solidified under RMF with different magnetic flux densities were examined and the results are shown in Fig. 3. It should be noted that the cross-sections were cut from the position that is 20 mm above the bottom surface of the ingots. For the alloy solidified without RMF, the primary Si phase tends to uniformly distribute in this section of alloy (Fig. 3a). The primary Si-rich layers are formed when the Al–30Si alloy solidified under RMF with different magnetic

J.C. Jie et al. / Acta Materialia 72 (2014) 57–66

59

Fig. 2. Vertical sections of Al–30Si alloy solidified under an RMF with different magnetic flux densities: (a) 0 mT, (b) 12 mT, (c) 17 mT and (d) 25 mT.

Fig. 3. Cross-sections of Al–30Si alloy solidified under an RMF with different magnetic flux densities: (a) 0 mT, (b) 12 mT, (c) 17 mT and (d) 25 mT.

flux densities. Under a 12 mT RMF, the Si-rich layer has a smaller thickness (Fig. 3b) than 17 mT (Fig. 3c) and 25 mT

(Fig. 3d) RMFs. In addition, the Si content of the Si-rich layer is measured to be in the range of 65–69 wt.%.

60

J.C. Jie et al. / Acta Materialia 72 (2014) 57–66

The microstructures at different positions in Figs. 3c and 2c are presented in Fig. 4. It can be seen from Fig. 4a (position 1) that the coarse primary Si phase congregates together. A small amount of Al-rich structures distributes in the inter-region of the primary Si phase. Fig. 4b (position 2) presents the interface between the Si-rich region and Al-rich region, where the size of the primary Si phase is much smaller than those formed close to the inner sidewalls. Fig. 4c (position 3) reveals that the inside structure of the ingot is mainly the Al–Si eutectic structure, and little primary Si phase can be found in the middle of the ingot. In addition, some primary Si particles can be found in the center part of the sample (position 4 in Fig. 2c) solidified under an RMF of 17 mT, as shown in Fig. 4d. 4. Discussion 4.1. Lorenz force and flow field under RMF The applied RMF is of low frequency and low induction. The low-frequency case is justified due to the relative pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 frequency <1 leading to a skin depth ð 2=xrl0 Þ of 37 mm, which exceeds the radius of the mold we use (R = 30 mm), where l0 is the magnetic permeability, r is the electrical conductivity and x is the angular frequency of the RMF. The low-induction condition implies that the angular velocity induced by the RMF does not change the magnetic field applied. Once the low-frequency and low-conduction conditions are fulfilled, the Lorentz force of the melt under the RMF

has an azimuthal component, and the meridional component of the Lorentz force can be neglected due to its minor effect on the fluid flow in comparison to the azimuthal component [29]. The magnetic force FL can be written as a function of the position only and has the following shape [30]: 1 F L ¼ r  x  B2  r  f ðz; rÞ 2

ð1Þ

where B is the magnetic flux density, r is the radius and the shape function f(z, r) reflects the influence of the finite length of the cylinder. In the cross-section, the Lorentz force has a tangential direction, and thus the liquid metal mainly exhibits a primary motion in the form of a swirl due to the large tangential Lorentz force caused by RMF. The associated centrifugal force is balanced by a radial pressure gradient, and this pressure gradient is also imposed throughout the boundary layer on the base. Of course, the swirl in this boundary layer is diminished through viscous drag, and so there is a local imbalance between the radial pressure force and centrifugal force. The result is a radial inflow along the bottom, with the fluid eventually drifting up and out of the boundary layer, which has the form of two toroidal vortices [31]. The secondary flow results in a radial inward flow along the solidification front on the bottom position of crucible, which transports the solute atoms towards the axis of the ingot [32,13]. Important non-dimensional parameters are the Hartmann and the Reynolds numbers, given by pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Ha ¼ BR0 r=ð2qtÞ and Rew ¼ xR20 =t, respectively. The

Fig. 4. Microstructures of different positions: (a) position 1 in Fig. 3c, (b) position 2 in Fig. 3c, (c) position 3 in Fig. 3c and (d) position 4 in the frame of Fig. 2c.

J.C. Jie et al. / Acta Materialia 72 (2014) 57–66

61

magnetic Taylor number representing a measure of the magnetic force can be derived as follows: Ta ¼ Ha2 Rew ¼

rxB2 R40 2lm

ð2Þ

The magnetic field is characterized by the magnetic flux density B and the frequency x, whereas q, l and m stand for the density, dynamic viscosity and kinematic viscosity, respectively. R0 is the inner radius of crucible. For weak magnetic fields, the flow field is laminar when the Ta has a magnitude of 105 according to the value of height/radius [30,33]. However, in the present study, the Taylor number is estimated to be 5.3  108 for the magnetic flux density of 25 mT, which is larger than those of the present reported results. Increasing the magnetic field leads to the flow instabilities, inducing Taylor–Go¨rtler vortices that move along the sidewalls of the cylinder and dissipate inside the Bo¨dewadt layer [30,33,34]. Such vortex structures are very efficient with respect to melt mixing [14]. The higher the Ta, the more Taylor–Go¨rtler vortices will be introduced; the flow is fully turbulent in the present study. However, due to the complex heat release and evolvement of the moving solid Si phase with lower electrical conductivity, the calculation and simulation of flow field are very difficult. Based on the direct numerical simulation and experimental results, the flow characteristics can be determined with the low Ta number. Nikrityuk et al. have studied the three distinct fluid flow phases [13]. During the first two phases (initial adjustment and inertial phase), acceleration of the liquid takes place in close similarity to the isothermal spin-up. The third phase is characterized by a braking of the fluid flow due to the progressive solidification increasing the aspect ratio of the liquid (diameter/ height) and decreasing the forcing. Although the simulation and calculation are difficult with increasing Ta number, the basic characteristics are the same for different Ta numbers. The differences between them are mainly the flow intensity and turbulence, such as the appearance of the Taylor–Go¨rtler vortex. However, the structure formation is just related to the basic characteristics of the flow field of alloys solidified under RMF. 4.2. Separation mechanism under RMF 4.2.1. Temperature distribution in the melt To clarify the segregation mechanism, the Al–30Si alloy solidified under an RMF of 25 mT was chosen as a typical example. Temperature distribution plays an important role in the formation of the Si-rich layer. Due to the application of thermal insulation material at the top of the crucible (see Fig. 1), there are mainly two temperature gradients in the liquid metal, the inner and outside metal together with the top and bottom parts. For example, Fig. 5 presents the cooling curves of Al–30Si alloy at different positions close to the inner crucible wall, together with the central part of liquid metal. It can be seen that a difference of

Fig. 5. Cooling curves of Al–30Si alloy at different positions under an RMF of 25 mT.

40 °C exists between positions 1 and 3 close to the sidewall, and the cooling rate is 17 K min1. For the case of the central liquid metal, the cooling rate is calculated to be 14 K min1 measured by thermocouple 4. Thus, the cooling rate of the central liquid metal is lower than those of the positions near the sidewall, which is important to the segregation of primary Si phase during the solidification process. The primary Si phase is first precipitated from the Al– 30Si melt during solidification, and the Si content in the remaining liquid decreases with temperature, which is based on the Al–Si equilibrium phase diagram. Without the RMF, the Si contents at different positions of the liquid alloy remain 30 wt.%, if neglecting the tiny compositional fluctuation and thermal convection. For the case with the forced convection caused by the RMF, whether the Si content in the flowing liquid metal changes during solidification is important to reveal the segregation mechanism. The formed Si phase floats in the liquid metal and moves with the rotating liquid. The solid–liquid interface is clear, and it is easy to pour out the moving liquid or semisolid from the crucible and leave the solidified Al–Si alloy. Before pouring out the liquid metal, the temperature was measured by a K-type thermocouple that was immersed below the central surface of the liquid metal. The pouredout liquid or semisolid was then heated to the liquid state to make Si distribute homogeneously, and then poured into a water-cooled copper mold for fast solidification. Thus, the Si content of the poured liquid or semisolid is homogenous and can be measured. A comparison of the calculated Si content of the poured-out liquid metal according to the pouring temperature is based on Ref. [35] and the measured Si content can reveal whether the state of the rotating metal is liquid or semisolid. Fig. 6 presents the measured Si content of the pouredout liquid metal and calculated Si content at different pour temperatures. It is surprising to see that the Si content of the rotating liquid metal decreases along the liquidus

62

J.C. Jie et al. / Acta Materialia 72 (2014) 57–66

primary Si phase moves due to the push of rotating liquid metal, which may result in the segregation of the primary Si phase. 2. The secondary flow and vortices transfer the bulk melt with higher temperature to the inner wall of the crucible, which is beneficial to the growth of the already precipitated primary Si phase there.

Fig. 6. Measured Si content of the poured-out liquid metal and theoretically calculated Si content at different pouring temperatures.

temperature, indicating a segregation process accompanied by solidification. The corresponding solidified shells obtained at different temperatures are shown in Fig. 7. The Si-rich layer is thin at 750 °C (Fig. 7a), and becomes thickened gradually with reducing the temperature to 650 °C and 600 °C, as shown in Fig. 7b and c. Hence, the formation of the Si-rich layer is a gradual process with decreasing temperature. 4.2.2. Formation of Si-rich layer According to the above results, two mechanisms are assumed to be responsible for the formation of the Si-rich layer: 1. The formed primary Si phase in the bulk liquid during solidification is transferred by the secondary flow and vortices to the inner wall, and then captured by the advancing solid–liquid interface. The basis of this assumption takes into account the electrical conductivity difference of the primary Si phase and the Al–Si melt, which are 5  104 and 3.6  106 S m1, respectively [36,37]. Thus, the primary Si phase suffers a much smaller Lorentz force due to the smaller electrical conductivity compared to the liquid metal. According to Eq. (1), the Lorentz force of the primary Si phase can be neglected compared to the liquid metal. The formed

The primary difference between the two mechanisms mentioned above is that the secondary flow transfers the primary Si phase or carries the Si-rich liquid metal to the solid–liquid interface. In order to clarify this question, a special experiment was carried out. The Al–30Si alloy was first heated to 840 °C and then cooled to different temperatures including 780, 750 and 710 °C by switching off the power. In this process, the semisolid was stirred by a graphite stick to prevent the primary Si phase from forming a mesh-shape around the inner wall of crucible. The temperature difference between the core and periphery of the sample was controlled to less than 2 K, so that the Si phase or atoms can distribute homogeneously in the melt. When the required temperature was achieved, the alloy was carried into the RMF and then solidified. The flow of the alloy was still very intense because a large amount of liquid metal still existed in each sample and the primary Si phase can move with the liquid metal. The precipitated volume fractions of the primary Si phase are 50%, 30% and 10% at temperatures of 710, 750 and 780 °C, respectively. The images of the solidified samples are shown in Fig. 8. Position 1 highlighted by the arrow is the coarse primary Si phase that is already precipitated from the melt before being placed under the RMF, while position 2 is formed after being placed under the RMF. It can clearly be seen that the coarse primary Si phase that already precipitated from the melt before being placed under the RMF cannot be carried to the periphery of the sample, but exhibits a nearly homogeneous distribution. The Si-rich zone, which was precipitated after being placed in the RMF, is formed on the bottom part of sample at 780 °C (Fig. 8a) and 750 °C (Fig. 8b). However, no Si-rich layer is formed at 710 °C, and the primary Si phase distributes in the sample homogeneously (Fig. 8c). The above results suggest that the primary Si phase cannot be carried

Fig. 7. The solidified shells obtained at different pouring temperatures: (a) 700 °C, (b) 650 °C and (c) 600 °C.

J.C. Jie et al. / Acta Materialia 72 (2014) 57–66

63

Fig. 8. Images of Al–30Si alloy solidified under an RMF at temperatures of (a) 780 °C, (b) 750 °C and (c) 710 °C. Position 1 is the coarse primary Si phase precipitated from the melt before being placed under an RMF; position 2 is formed after it is placed under an RMF.

to the periphery of the sample by forced convection during the solidification process. Thus, the Si-rich liquid metal is carried to the periphery of the sample, which is responsible for the segregation of the primary Si phase. It can be concluded from the above experimental results that the formation of a Si layer in Al–30Si alloy solidified under an RMF is a complex coupling process of cooling conditions, flow field and crystal growth. Fig. 9 is a schematic illustration of the segregation mechanism of Al–30Si alloy solidified under an RMF. At different temperatures, the volume fractions of the precipitated primary Si phase remaining in the melt are shown in Fig. 9a. It can be seen that the volume fraction of the primary Si phase is variant at different temperatures in an equilibrium solidification process. Therefore, the volume fraction of the primary Si phase at different positions in the melt is nonuniform due to the time-dependent temperature difference. In the present study, the temperature gradient is very small and the Si distribution is homogeneous when the liquid metal is placed in an RMF at the beginning. The intense secondary flow and Taylor–Go¨rtler vortices occur soon after the liquid metal is placed under the RMF (Fig. 9b). With prolonged time, the obvious temperature difference at different positions in the melt appears (see Fig. 5), resulting in the inhomogeneous distribution of Si atoms in the melt and volume fraction of the primary Si phase. The primary Si phase is first formed in the region with lower temperature, i.e., the bottom and sidewall of the crucible, as shown in Fig. 9c. Thus, the position close to the bottom and sidewalls will be rich in the precipitated primary Si phase, while it is poor in the center of the alloy. Furthermore, the size of the primary Si phase increases with decreasing temperature. When the sizes of the primary Si phases increase to a certain extent, they will contact each other and exhibit a mesh-shape that can reduce the flow velocity of the liquid metal. In addition, the viscosity of the liquid metal near the crucible wall increases with decreasing temperature due to the increasing amount of primary Si phase. The above reasons can keep a nearly fixed enrichment of primary Si phase near the crucible wall

(Fig. 9c), which can resist the sweep of liquid metal. This is the fundamental condition to form the Si-rich layer. As already mentioned, the bulk liquid metal can be carried to the periphery of the sample by the secondary flow, and then returns to the central part along the Bo¨dewadt layer. In this process, the melt temperature will decrease when it is close to the inner wall, and will increase when it returns to the central part (Fig. 9d). Thus, on one hand, the Si content in the liquid metal will saturate once it is close to the inner wall of the crucible. On the other hand, the already formed primary Si phase will serve as a growth site. In addition, the Si-rich liquid metal can also directly nucleate in the vicinity of the inner wall of the crucible, where the temperature is low. Thus, the liquid metal continuously flows in the crucible. Flow close to the inner wall of the crucible can reduce the Si content of liquid metal, resulting in the growth of the already precipitated primary Si phase. Finally, the primary Si phase can be separated in the periphery of the sample, and the Al–Si eutectic structure is finally formed in the inner part of the alloy. In addition, the centrifugal force Fc plays an important role in the formation of the Si-rich layer, and can be expressed as follows: F c ¼ mrx2

ð3Þ

The centrifugal force vs. gravity can be expressed as: N¼

x2 r g

ð4Þ

In our study, the maximum ratio on its radial profile can be estimated to be 3.3. The large centrifugal force may fracture or distort the deposited primary Si phase in the Si-rich layer. Furthermore, the Al–Si melt inside the inter-region of the primary Si phase can be squeezed out through compression by centrifugal force and Ekman pumping. Thus, the content of primary Si phase in the layer can be extended up to 65–69 wt.%, which is far more than the normal composition. Due to the formation of the Si-rich layer around the periphery of the sample, the Al–Si melt adjacent to the

64

J.C. Jie et al. / Acta Materialia 72 (2014) 57–66

Fig. 9. Schematic illustration of segregation mechanism of Al–30Si alloy during solidification under an RMF. (a) The volume fractions of the precipitated primary Si phase and the remaining primary Si phase at different temperatures, (b) temperature and Si distribution are homogeneous when the liquid metal is placed under an RMF at the beginning, (c) the motionless Si-mesh formed in the inner crucible wall due to the cooling condition, (d) Si content in the bulk liquid decreases, caused by the secondary flow and (e) the braking of the primary flow and stopping of the segregation process. The circle with arrows represents the secondary flow and Taylor–Go¨rtler vortices cased by the RMF. The melt temperature and volume fraction of the primary Si phase along the dotted line are depicted in each illustration.

layer is in a S-poor state, which is extended to the eutectic composition. With decreasing temperature, the Al–Si alloy will finally be solidified with the eutectic content. When the alloy is further solidified, a braking of the primary flow appears when approximately one-third of the Al–Si liquid melt is solidified, due to the progressive solidification increasing the aspect ratio of the liquid (diameter/height) and decreasing the forcing [13]. Hence, in the last stage of solidification, the segregation process stops (see Fig. 9e), and there are still some coarse primary Si particles in the sample, as shown in Fig. 2b and c. For the hypoeutectic Al–Si alloy solidified under an RMF, the periphery of the sample will be enriched in the primary a-Al phase, while the Si content inside the sample can be extended to the eutectic composition [20,22,38,39]. The advection of Si-rich liquid due to the secondary flow is responsible for the development of a liquid channel on the axis of rotation inside the mushy zone. By contrast, for the hypereutectic Al–30Si alloy, the primary Si phase is first precipitated adjacent to the inner crucible wall. The Si-rich liquid metal is transferred to the inner wall of the crucible by secondary flow and Taylor–Go¨rtler

vortices, which is beneficial to the growth of the primary Si phase and the formation of a Si-rich layer. Meanwhile, the rejected Al solute is transferred to the bulk liquid, which results in the formation of the Al–Si eutectic in the central part of sample. 4.2.3. Different microstructures under different conditions According to the analysis above, the Si distribution can be controlled by the cooling condition and flow field under an RMF. The cooling condition is the key condition, and the flow field of liquid metal under an RMF is the necessary condition. Controlling the flow field intensity and positions of heat transfer, the Si-rich layer can be obtained on the different positions of Al–30Si alloy. The corresponding results are shown in Fig. 10. When the central and bottom parts were heat-preserved, the Si phase could be enriched in the upper part solidified under RMF with a magnetic flux density of 12 mT, as shown in Fig. 10a. When the top and bottom parts were heat-preserved, the Si-rich layer was formed at the central part of the sample when solidified under an RMF of 12 mT (see Fig. 10b). When the top and central parts were heat preserved, the

J.C. Jie et al. / Acta Materialia 72 (2014) 57–66

65

Fig. 10. Different distributions of primary Si phase under different conditions: (a) top part, 12 mT, (b) central part, 12 mT and (c) bottom part, 17 mT.

Si phase could be segregated at the bottom part solidified under an RMF of 17 mT, as shown in Fig. 10c. Thus, the distribution of the primary Si phase can be controlled by altering the cooling condition and flow field induced by the RMF. For the industrial application of hypereutectic Al–Si alloy, the proper cooling rate and flow field are very important to control the distribution of the primary Si phase. If the cooling rate is high enough, the primary Si phase can distribute homogeneously due to the limited time for the diffusion of the Si atoms. This situation comes into existence whatever the flow field is. If the cooling rate is very low, the macrosegregation will appear at a proper temperature gradient, and the extent will increase with increasing liquid flow velocity. 5. Conclusions The present study investigated the segregation mechanism of primary Si phase during solidification of hypereutectic Al–Si alloy under the RMF. By controlling the cooling condition and flow field induced by RMF, the Si-rich layer with a high Si content was achieved. The segregation mechanism contains several steps as follows: 1. During solidification process, the primary Si phase is first nucleated on the crucible wall or formed close to the wall where the temperature is lower. A primary Si mesh appears when the size of primary Si phase increases to an extent. This is the fundamental condition to form the Si-rich layer. 2. The intense melt flow, i.e., secondary flow and Taylor– Go¨rtler vortices, brings about the bulk liquid with high Si content to the inner wall region where the temperature is low. The precipitated primary Si phase there incorporates the Si atoms from the liquid metal and grows continuously. At the same time, the Si content and equilibrium temperature of the bulk liquid metal reduce. With prolonged time, the Si content of the bulk

liquid metal can be reduced to the eutectic point, and the Si-rich layer is finally formed. 3. The additional pressure on the inner wall caused by the centrifugal force compresses the Si-rich layer and further squeezes out the Al-rich melt, resulting in the high Si volume fraction. In summary, the segregation of primary Si phase is a couple interaction process of heat transfer, flow of liquid metal and crystal growth under RMF.

Acknowledgements The authors gratefully acknowledge the support of the Natural Science Foundation of China (Nos. 51134013, 51271042, 51104029), the China Postdoctoral Science Foundation (No. 2013M530913), the Fundamental Research Funds for the Central Universities of China and National Natural Science Foundation (No. 51274054). In addition, the authors are grateful for the useful discussions with Dr. S.J. Wang and Ms. H.W. Wang. References [1] Kurs W, Fisher DJ. Fundamentals of solidification. Aedermannsdorf: Trans Tech Publications; 1990. [2] Haverkort JW. Metall Mater Trans B 2010;41B:1240. [3] Gerbeth G, Eckert K, Odenbach S. Eur Phys J Spec Top 2013;8:1. [4] Eckert S, Nikrityuk PA, Willers B, Raebiger D, Shevchenko N, Neumann-Heyme H, Travnikov V, Odenbach S, Voigt A, Eckert K. Eur Phys J Spec Top 2013;137:123. [5] Asta M, Beckermann C, Karma A, Kurz W, Napolitano R, Plapp M, et al. Acta Mater 2009;57:941. [6] Boettinger WJ, Coriell SR, Greer AL, Karma A, Kurz W, Rappaz M, et al. Acta Mater 2000;48:43. [7] Zimmermann G, Weiss A. Microgr Sci Technol 2005;16:143. [8] Totten GE, MacKenzie DS. Handbook of aluminum: physical metallurgy and processes, vol. 1. New York: Marcel Dekker Inc; 2003. [9] Griffiths WD, McCartney DG. Mater Sci Eng A 1996;216:47. [10] Medina M, Terrail YDU, Durand F, Fautrelle Y. Metall Mater Trans B 2004;35B:743.

66

J.C. Jie et al. / Acta Materialia 72 (2014) 57–66

[11] Stiller J, Koal K, Nagel WE, Pal J, Cramer A. Eur Phys J Spec Top 2013;122:111. [12] Cramer A, Eckert S, Gerbeth G. Eur Phys J Spec Top 2013;41:25. [13] Nikrityuk PA, Eckert K, Grundmann R. Int J Heat Mass Transfer 2006;49:1501. [14] Eckert K, Gerbeth G, Eckert S, Nikrityuk PA, Ra D. Metall Mater Trans B 2007;38:977. [15] Nadella R, Eskin DG, Du Q, Katgerman L. Prog Mater Sci 2008;53:421. [16] Beckermann C. Int Mater Rev 2002;57:243. [17] Flemings MC. ISIJ Int 2000;40:833. [18] Davis JR. ASM handbook: properties and selection: nonferrous alloys and special-purpose materials, vol. 02. Materials Park, OH: ASM International; 1990. [19] Steinbach S, Ratke L. Mater Sci Eng A 2005;413:200. [20] Zimmermann G, Weiss A, Mbaya Z. Mater Sci Eng A 2005;413–414:236. [21] Steinbach S, Ratke L. Int J Cast Metal Res 2007;20:140. [22] Steinbach S, Ratke L. Metall Mater Trans A 2007;38A:1388. [23] Willers B, Eckert S, Nikrityuk PA, Raebiger D, Dong J, Eckert K, et al. Metall Mater Trans B 2008;39:304.

[24] Metan V, Eigenfeld K, Raebiger D, Leonhardt M, Eckert S. J Alloys Compd 2009;487:163. [25] Noeppel A, Ciobanas A, Wang XD, Zaidat K, Mangelinck N, Budenkova O, et al. Metall Mater Trans B 2010;41:193. [26] Metan V, Eigenfeld K. Eur Phys J – Spec Top 2013;220:139. [27] Roosz A, Kovacs J, Ronafoldi A, Kovacs A. Mater Sci Forum 2013;752:57. [28] Nishi Y, Kang Y, Morita K. Mater Trans 2010;51:1227. [29] Priede J, Gelfgat YM. Magnetohydrodynamics 1996;32:249. [30] Grants I, Gerbeth G. J Fluid Mech 2001;431:407. [31] Davidson PA. J Fluid Mech 1992;245:669. [32] Priede J, Gerbeth G. J Cryst Growth 2005;285:261. [33] Grants I, Gerbeth G. J Fluid Mech 2002;463:229. [34] Richardson BAT. J Fluid Mech 1974;63:593. [35] Murray JL, McAIister AJ. Bull Alloy Phase Diagr 1984;5:74. [36] Kadkhodabeigi M, Safarian J, Tveit H, Tangstad M, Johansen ST. Trans Nonferr Met Soc China 2012;22:2813. [37] Dong J, Eckert K. Metall Mater Trans B 2008;39B:304. [38] Hainke M, Friedrich J, Uller GM. J Mater Sci 2011;9:2011. [39] Steinbach S, Ratke L. Metall Mater Trans A 2007;38:1388.