Effect of far-field flow on a columnar crystal in the convective undercooled melt*
Ji Xiao-Jiana), Chen Ming-Wena), Xu Xiao-Huaa), Wang Zi-Dongb)
School of Mathematics and Physics, University of Science and Technology Beijing, Beijing 100083, China
School of Materials Science and Engineering, University of Science and Technology Beijing, Beijing 100083, China

Corresponding author. E-mail: chenmw@ustb.edu.cn

Project supported by the Overseas Distinguished Scholar Program by the Ministry of Education of China (Grant No. MS2010BJKJ005), the National Natural Science Foundation of China (Grant No. 10972030), and the Science and Technology Support Project of Jiangxi, China (Grant No. 20112BBE50006).

Abstract

The growth behavior of a columnar crystal in the convective undercooled melt affected by the far-field uniform flow is studied and the asymptotic solution for the interface evolution of the columnar crystal is derived by means of the asymptotic expansion method. The results obtained reveal that the far-field flow induces a significant change of the temperature around the columnar crystal and the convective flow caused by the far-field flow accelerates the growth velocity of the interface of the growing columnar crystal in the upstream direction and inhibits its growth velocity in the downstream direction. Our results are similar to the experimental data and numerical simulations.

Keyword: 64.70.D–; 68.08.–p; 81.30.Fb; columnar crystal; far-field uniform flow; interface morphology; crystal growth
1. Introduction

The growth behavior and pattern formation of the columnar crystal during the solidification of the undercooled melt is an important issue in the materials science and condensed matter physics and has been extensively investigated during the past years.[14] Coriell and Parker[5] analyzed the morphological stability of the columnar crystal undergoing diffusion-controlled growth in consideration of isotropic interfacial tension and revealed that the growth of the crystal is unstable when the radius is greater than and stable when the radius is less than the critical radius. The analogous mechanism is also found by Mullins and Sekerka.[6] Chen et al.[7] studied the microstructure of different positions of the columnar crystal experimentally and analyzed the process of the formation and growth of the columnar crystal in the continuous directional solidification. It is shown that the growth of the columnar crystal is affected by both the columnar crystal merge mechanism and the elimination rule. Buchholz and Engler[8] found that the deformation of the solute layer around the dendrites tips caused by the forced convection makes the columnar dendrites tips incline in the upstream direction.[9, 10] In the solidification, the crystal growth will be affected by many factors, such as undercooling, [11] interface kinetics, [12] anisotropic surface tension, [13, 14] flow, [1519] etc. These experiments and numerical simulations deepen the understanding of the crystal growth. However, since the nonlinear system of the crystal growth with flow has no exact analytical solutions available generally, theoretical studies on the crystal growth with flow are relatively few. Chen et al.[20, 21] studied the effect of the far-field flow on the interface morphology of the spherical crystal. In this paper, the effect of the convective flow caused by the far-field flow on the interface evolution of the columnar crystal is discussed by using the asymptotic expansion method, the results obtained show the interface morphology of the columnar crystal affected by the far-field flow.

2. Theoretical model

Consider the growth of an infinite columnar crystal with an initial interface radius in the convective undercooled melt. We assume that the melt is an incompressible Newtonian fluid. In the undercooled melt, far from the columnar crystal the temperature is , the undercooling is , where is the solidification equilibrium temperature of the pure substance. The columnar crystal is influenced by the far-field flow Ū , which points to the cross section of the columnar crystal in the rightward direction. Let Ū denote the convective velocity in the liquid phase, denote the pressure, and denote the temperatures respectively for the solid and liquid phases, denote the interface of the columnar crystal, in which represents time.

We rescale the temperature scale by Δ T, velocity scale by the characteristic velocity of the interface V, length scale by , time scale by , pressure scale by ρ LV2, where ρ L is the density of the melt, then we introduce the following dimensionless physical quantities:

where , kL is the heat conduction coefficient of the liquid phase, Δ H is the latent heat of fusion per unit volume. We use coordinate (r, θ ) to describe the cross section perpendicular to symmetric axis of the columnar crystal affected by the far-field flow as shown in Fig. 1.

Fig. 1. A columnar crystal in the undercooled melt affected by the far-field uniform flow.

The governing equations for the convective velocity U contain the continuity equation and Navier– Stokes equation as

where , υ is the kinematical viscosity, is the gradient operator, 2 is the Laplacian operator.

The convective velocity U in the far field satisfies: as r→ ∞ ,

where U is a dimensionless velocity, i is the unit vector in the positive direction of x axis.

The heat transfer processes for the temperature fields are governed by thermal conduction equations

where

where cp is the specific heat, κ S and κ L are, respectively, the thermal diffusivities of the solid and liquid phases.

At the interface r = R(θ , t), the temperature satisfies the following Gibbs– Thomson condition and energy conservation condition:

where

and K is the local mean curvature at the interface, UI is the growth velocity of the interface, n is the unit exterior vector normal to the crystal– melt interface, γ is the isotropic surface energy, μ is the interfacial kinetics coefficient, kS is the heat conduction coefficient of the solid phase.

The temperature for the liquid phase TL in the far field satisfies: as r→ ∞ ,

The initial condition of the interface is

We consider the case of small undercooling or large latent heat, thus the parameter ε is very small, then the columnar crystal growth system (2)– (10) becomes a singularly perturbed boundary problem.

3. Asymptotic solution

For the case of the small parameter, we seek the asymptotic expansion solutions for the growth system (2)– (10) as ε → 0,

Substituting Eq. (11) into Eqs. (2)– (10) and expanding the interface curvature into

we equate the coefficients of like powers of ε and derive each order approximation in Eq. (11).

If ∂ /∂ t = 0 is assumed, we obtain the leading order terms for the flow field and temperature fields satisfying

At the interface r = R0 (t), the interface conditions satisfy

The leading order term of the temperature field for the liquid phase TL0 in the far field satisfies: as r→ ∞ ,

The initial condition is

Approximatively, we replace (U0 · )U0 in Eq. (14) by (U · )U0, where U = (U cosθ , − U sinθ ). Then the solution U0 of Eq. (14) is decomposed into

where φ and χ satisfy respectively

where Y is the solution of the Helmholtz equation 2Yb2Y = 0, b is subject to Ω /U = 1/2b. The solution of Eq. (23) is

where B0 and B1 are constants to be determined. By solving the Helmholtz equation, χ is expressed as

where a = 1.7811, C0 and C1 are constants to be determined. Substituting Eqs. (25) and (26) into Eq. (22), we derive the leading order approximation solutions for the flow field U0 = (u0, v0) and P0 can be expressed as

With Eqs. (15), (16), and the interface condition (18), we obtain the leading order approximation solutions for the temperature fields written as

where a0 (t) and bS0 (t) are functions to be determined. It is seen that as r→ ∞ , equation (28) does not satisfy the far-field condition (20). Like Coriell and Parker, [5] we take a special treatment for the far field such that R satisfies

then the leading order approximation solutions for the temperature fields are expressed as

where Aλ = ln R − ln R0 and R0 follows the ordinary differential equation

With the initial condition (21), R0 is determined by the implicit function

If ∂ /∂ t = O(1) and ε → 0 are assumed, we obtain the first-order terms for the temperature fields satisfying

At the interface r = R0(t), the interface conditions satisfy

where

The initial condition is

Considering the function constitution of inhomogeneous term (U0 · )TL0, we seek the solutions of the following form for Eqs. (35)– (41):

Substituting Eqs. (42)– (44) into interface conditions (37)– (39), we obtain

where g1(t) satisfies the ordinary differential equation

With the initial condition (41), the solution of g1 (t) is expressed as

Then, the first-order approximation solutions for the temperature fields and R1 are, respectively, expressed as

Therefore, we obtain the first-order asymptotic solutions of Eq. (11). The corresponding growth velocity of the interface is

With the asymptotic solution, we analyze the effect of the convective flow on the interface morphology of the columnar crystal. Figure 2 shows the temperature variation in the liquid phase along different directions. It is found that the temperature which is not affected by the far-field flow increases more remarkably than that affected by the far-field flow as time increases. Figure 3 shows the isothermal diagram of the liquid phase. The isotherm in the upstream direction where the flow comes is denser than that in the downstream direction, and the temperature variation in the upstream direction is larger than that in the downstream direction (as shown in Fig. 4). Figure 5 shows that the radius of the interface affected by the flow increases more remarkably than that not affected by the flow as time increases, it implies that the far-field flow promotes the growth velocity of the interface (as shown in Fig. 6).

Fig. 2. The temperature variation of the liquid phase versus time t at r = 15, where k = 1, C = 0.3, M = 0.09, Γ = 0.01, ε = 0.2, Aλ = 3.3, U = 0.5; θ = 2π /3, 3π /4, π (from top to bottom for cases affected by U).

Fig. 3. The isothermal diagram of the liquid phase at t = 120, where k = 1, C = 0.3, M = 0.09, Γ = 0.01, ε = 0.2, U = 0.5, Aλ = 3.3.

Fig. 4. The temperature variation of the liquid phase versus r at t = 120, where k = 1, C = 0.3, M = 0.09, Γ = 0.01, ε = 0.2, Aλ = 3.3, U = 0.5; θ = 2π /3, 3π /4, π (from top to bottom for cases affected by U).

Fig. 5. The variation of the radius of the interface versus time t, where k = 1, C = 0.3, M = 0.09, Γ = 0.01, ε = 0.2, Aλ = 3.3, U = 0.5; θ = π , 3π /4, 2π /3 (from top to bottom for cases affected by U).

Fig. 6. The variation of the growth velocity of the interface versus time t, where k = 1, C = 0.3, M = 0.09, Γ = 0.01, ε = 0.2, Aλ = 3.3, U = 0.5; θ = π , 3π /4, 2π /3 (from top to bottom for cases affected by U).

In the paper, we introduce the dimensionless physical quantities to simplify the equations and interface conditions, if we return to the dimensional physical quantities, we can compare qualitatively with a real system. For example, the solidification equilibrium temperature of Fe particles is TM = 1728 K, the undercooling of Fe particles during the solidification is Δ T = 373 K, the latent heat of Fe particles per unit volume is Δ H = 2.404 × 109J!· m− 3, the specific heat is cp = 477.3 J· kg− 1· K− 1, the density is ρ L = 7874 kg· m− 3, , we obtain the variations of the radius and the growth velocity of the interface versus time, respectively. From Figs. 7– 8, we see that the variations of the radius and the growth velocity of the interface in a real system are similar to that in the dimensionless system.

Fig. 7. The variation of the radius of the interface versus time t, where θ = π , 3π /4, 2π /3 (from top to bottom for cases affected by U).

Fig. 8. The variation of the growth velocity of the interface versus time t, where θ = π , 3π /4, 2π /3 (from top to bottom for cases affected by U).

It is shown from the above analysis that the convective flow significantly influences the interface evolution. When 0 < θ < π /2, if R0 > Γ , then dR0/dt > 0, and ∂ R/∂ t < dR0/dt; if R0 < Γ , then dR0/dt < 0, and ∂ R/∂ t > dR0/dt. When π /2 < θ < π , if R0 > Γ , then dR0/dt > 0, ∂ R/∂ t > 0, and ∂ R/∂ t > dR0/dt; if R0 < Γ , then dR0/dt < 0, ∂ R/∂ t < 0, and ∂ R/∂ t < dR0/dt. It means that the convective flow accelerates the growth velocity of the interface of the growing columnar crystal in the upstream direction and inhibits its growth velocity in the downstream direction; whereas the convective flow enhances the decay velocity of the interface of the decaying columnar crystal in the upstream direction and inhibits its decay velocity in the downstream direction. Figures 9– 11 show the evolution of a columnar crystal at different times (or equivalently, for different R0) affected by the far-field flow. The theoretical results are similar to the experimental data and numerical simulations.[8]

Fig. 9. The effect of the far-field flow on the evolution of a columnar crystal, where R0 = 7, U = 0.5, k = 1, C = 0.3, M = 0.09, Γ = 0.01, ε = 0.2, Aλ = 3.3.

Fig. 10. The effect of the far-field flow on the evolution of a columnar crystal, where R0 = 13, U = 0.5, k = 1, C = 0.3, M = 0.09, Γ = 0.01, ε = 0.2, Aλ = 3.3.

Fig. 11. The effect of the far-field flow on the evolution of a columnar crystal, where R0 = 16, U = 0.5, k = 1, C = 0.3, M = 0.09, Γ = 0.01, ε = 0.2, Aλ = 3.3.

4. Conclusion

We investigate the growth behavior of a columnar crystal in the convective undercooled melt affected by the far-field flow and derive the approximation solution for the interface evolution by using the asymptotic expansion method. The results reveal that the far-field flow induces a significant change of the temperature around the columnar crystal and the temperature variation in the upstream direction is larger than that in the downstream direction. The convective flow accelerates the growth velocity of the interface of the growing columnar crystal in the upstream direction and inhibits its growth velocity in the downstream direction. The theoretical results are similar to the experimental data and numerical simulations.

Reference
1 Mei X Y, Ma W H, Wei K X, Yang B and Dai Y N 2009 Acta Scientiarum Naturalium Un. 48 89 [Cited within:1]
2 Yan N, Dai F P, Wang W L and Wei B B 2011 Chin. Phys. Lett. 28 078101 DOI:10.1088/0256-307X/28/7/078101 [Cited within:1] [JCR: 0.811] [CJCR: 0.4541]
3 Chen C, Chen Z, Zhang J and Yang T 2012 Acta Phys. Sin. 61 108103(in Chinese) [Cited within:1] [JCR: 1.016] [CJCR: 1.691]
4 Du L F, Zhang R, Xing H, Zhang L M, Zhang Y and Liu L 2013 Acta Phys. Sin. 62 106401(in Chinese) DOI:10.7498/aps.62.106401 [Cited within:1] [JCR: 1.016] [CJCR: 1.691]
5 Coriell S R and Parker R L 1965 J. Appl. Phys. 36 632 DOI:10.1063/1.1714041 [Cited within:2] [JCR: 0.71]
6 Mullins W W and Sekerka R F 1963 J. Appl. Phys. 34 323 DOI:10.1063/1.1702607 [Cited within:1] [JCR: 0.71]
7 Chen Y J, Chen Q, Wang Z D, Hu H Q, Liu Y M and Lian Y D 2004 J. Tsinghua Sci. Technol. 44 1464(in Chinese) [Cited within:1] [CJCR: 0.517]
8 Buchholz A and Engler S 1996 Comput. Mater. Sci. 7 221 DOI:10.1016/S0927-0256(96)00084-5 [Cited within:2] [JCR: 1.878]
9 Lv H, Tang S L and Zhou W P 2011 Chin. Phys. Lett. 28 084708 DOI:10.1088/0256-307X/28/8/084708 [Cited within:1] [JCR: 0.811] [CJCR: 0.4541]
10 Cao B, Lin X, Wang M and Huang W D 2012 Chin. Phys. B 21 086401 DOI:10.1088/1674-1056/21/8/086401 [Cited within:1] [JCR: 1.148] [CJCR: 1.2429]
11 Chen C C, Lin H K and Lan C W 2014 J. Cryst. Growth 385 121 DOI:10.1016/j.jcrysgro.2013.03.053 [Cited within:1] [JCR: 1.552]
12 Chen M W, Ni F, Wang Y L, Wang Z D and Xie J X 2011 Acta Phys. Sin. 60 068103(in Chinese) [Cited within:1] [JCR: 1.016] [CJCR: 1.691]
13 Chen M W, Lan M, Yuan L, Wang Y Y, Wang Z D and Xu J J 2009 Chin. Phys. B 18 1691 DOI:10.1088/1674-1056/18/4/069 [Cited within:1] [JCR: 1.148] [CJCR: 1.2429]
14 Zhang Y P, Lin X, Wei L, Peng D J, Wang M and Huang W D 2013 Acta Phys. Sin. 62 178105(in Chinese) DOI:10.7498/aps.62.178105 [Cited within:1] [JCR: 1.016] [CJCR: 1.691]
15 Galenko P K, Funke O, Wang J and Herlach D M 2004 Mater. Sci. Eng. A 375–377 488 DOI:10.1016/j.msea.2003.10.021 [Cited within:1]
16 Cao B, Lin X, Wang M and Huang W D 2012 Chin. Phys. B 21 056401 DOI:10.1088/1674-1056/21/5/056401 [Cited within:1] [JCR: 1.148] [CJCR: 1.2429]
17 Zhang W and Chen C 2012 Chin. Phys. Lett. 29 084701 DOI:10.1088/0256-307X/29/8/084701 [Cited within:1] [JCR: 0.811] [CJCR: 0.4541]
18 Chen M W, Wang B and Wang Z D 2013 Chin. Phys. B 22 116805 DOI:10.1088/1674-1056/22/11/116805 [Cited within:1] [JCR: 1.148] [CJCR: 1.2429]
19 Bhattacharyya K and Pop L 2014 Chin. Phys. B 23 024701 DOI:10.1088/1674-1056/23/2/024701 [Cited within:1] [JCR: 1.148] [CJCR: 1.2429]
20 Chen M W, Wang Z D, Xie J X and Xu J J 2008 Acta Mech. Sin. 24 681 DOI:10.1007/s10409-008-0169-7 [Cited within:1] [CJCR: 0.558]
21 Chen M W and Wang Z D 2013 Chin. Phys. B 22 098104 DOI:10.1088/1674-1056/22/9/098104 [Cited within:1] [JCR: 1.148] [CJCR: 1.2429]