Cu interface under uniform and gradient high magnetic fields

Materials Science and Engineering A 495 (2008) 244–248

Diffusion layer growth at Zn/Cu interface under uniform and gradient high magnetic fields Dong-gang Li a , Qiang Wang a,∗ , Guo-jian Li a , Xiao Lv a , Keiji Nakajima a,b , Ji-cheng He a a

Key Laboratory of Electromagnetic Processing of Materials (Ministry of Education), Northeastern University, Shenyang, China b Division of Applied Process Metallurgy, Department of Materials Science and Engineering, Royal Institute of Technology (KTH), SE-100 44 Stockholm, Sweden Received 21 March 2007; received in revised form 24 September 2007; accepted 10 October 2007

Abstract As a common phenomenon occurring in many material processes, diffusion may induce significant changes in composition and microstructure near the interface. In the present study, liquid/solid (Zn/Cu) interface diffusion experiments in high magnetic fields (up to 12 T) were conducted and the thickness changes of diffusion layer under different magnetic field conditions were examined. It was found that there were no noticeable effects of high magnetic fields on the formation of intermetallic phases at the interface. However, the magnetic flux density exerted a non-linear influence on the diffusion layer thickness. This phenomenon should be attributed to the effect of magnetic fields suppressing natural convection and inducing thermo-electromagnetic convection. In addition, the diffusion of Zn into Cu could be retarded by a magnetic field gradient. These results indicate that both the strength and the gradient of high magnetic fields can be used to control the diffusion behavior. © 2008 Elsevier B.V. All rights reserved. Keywords: Diffusion; Liquid/solid; Interface; Convection; High magnetic fields

1. Introduction Diffusion, which often occurs in the vicinity of the interface of a binary alloy system in many material processes, usually results in great changes of alloy composition and microstructure. Additionally, some diffusion-related behaviors may lead to the formation of some layer-pattern compounds at the interface, either useful or harmful for material properties. Therefore, the control of diffusion process is one of the effective measures to improve material properties. With the recent development of high magnetic field technology, some researchers paid more attention to the diffusion control by high magnetic fields [1,2]. Many experiments indicated that high magnetic fields can change the movement of active atoms and vacancies by an intensive Lorentz force or magnetic force. Khine et al. [3] reported that the convection could be significantly suppressed by the Lorentz force during self-diffusion experiments of liquid germa-

∗

Corresponding author. Tel.: +86 24 83681726; fax: +86 24 83681758. E-mail address: [email protected] (Q. Wang).

0921-5093/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2007.10.096

nium in the case of 3 T magnetic field. Miyake et al. [4] discussed that under a static magnetic field, the obtained diffusion coefficient in liquid metal was of the same order of magnitude with the data measured in microgravity environment, in which there is not almost any convection. Nakamichi et al. [5] found that the diffusion of carbon in ␥-iron was retarded by application of a 6 T uniform magnetic field, but it was enhanced by a negative magnetic field gradient, which means that carbon atoms move towards the direction with a higher magnetic field strength. However, Nakajima et al. [6] reported that a magnetic field had no obvious effect on the diffusion of nickel in titanium. There were some ambiguity about the effects of high magnetic fields on diffusion, so it is necessary to have a deep insight into the mechanism how uniform and gradient high magnetic fields play roles in diffusion process for fundamental as well as practical reasons. Earlier work on diffusion under high magnetic fields was related to a solid/solid diffusion couple. But in industrial application, interdiffusion in a liquid/solid system is the key issue for solidification process, hot-dip coating process, diffusion welding, etc. Since it is easier for a high magnetic field to control the movement of atoms in liquid phase than in solid phase, a liquid

D.-g. Li et al. / Materials Science and Engineering A 495 (2008) 244–248

245

Fig. 1. (a) Schematic view of the superconducting magnetic field heat treatment system, (b) distribution of the magnetic field along the bore axis and (c) the temperature profile of the heating treatment process.

Zn/solid Cu diffusion couple is used to study the diffusion layer growth at the interface under different magnetic field conditions in the current paper. 2. Experimental procedure Experimental materials were composed of 99.99 wt.% pure copper cold rolled rods and 99.99 wt.% pure zinc granules. Every diffusion couple was produced by filling 5.5 g of zinc granules into a cylindrical copper crucible with an inner diameter of 10 mm. The inner surface of the Cu crucibles was mechanically polished (by a velocity-controlled flat-drill coated with a polishing cloth) to form a clean and smooth interface between Cu and Zn. The zinc granules were cleaned with HCl 3 vol.% before the experiments in order to limit the layer of zinc oxides. The experimental apparatus is based on a superconducting magnet (JMTD-12 T100, JASTEC, Japan) with a bore of 100 mm diameter, in which a resistance furnace (i.d. 33 mm) was installed for melting and solidifying the specimens (Fig. 1(a)). An axial magnetic field with a maximum magnetic flux density (B) of 12 T at the centre of the bore and the maximum value of B dB/dz of ±564 T2 /m at ±105 mm from the centre of the bore was applied upward along the cylindrical crucible axis. Fig. 1(b) shows the

distribution of the magnetic field along the bore axis. The influence of the uniform magnetic field was examined with magnetic flux density of 4.5, 8, 8.8, 10.5 and 12 T, respectively. The magnetic field gradients [B(dB/dz) = ±166 T2 m−1 with B = 8.8 T] were applied by placing the samples apart from the uniform magnetic field region for ±45 mm. The magnetic field was applied during all the heating treatment process. Temperature was controlled by means of a programming thermometer with an R-type thermocouple (the precision of ±0.1 K). The samples were heated up to 723 K (the melting point of pure zinc is 692.5 K) and kept at this temperature for 30 min, then they were cooled down to room temperature (Fig. 1(c)) under a protective atmosphere of high purity argon with a flow rate of 40 ml/min. Thereafter, the 24 samples (3 samples used for each experimental condition) were cut lengthwise with a linear cutting machine, mechanically polished and buff-finished using SiC suspension with a diameter of 3 ␮m to a mirror surface. The microstructure of the diffusion layers at the bottom of the crucible was observed with LEICA metallographic microscope. Furthermore, the average thickness of the diffusion layers was measured carefully (at least nine regions in each sample) using optical microscopy. The concentration distribution of Cu and Zn across the layers along the diffusion direction

Fig. 2. Micrograph of the intermediate layers (a) without a magnetic field and (b) with a magnetic field of 8.8 T. The bright-colored islands on the top of the micrographs are also ␧ phase.

246

D.-g. Li et al. / Materials Science and Engineering A 495 (2008) 244–248

Fig. 3. Concentration profiles of Zn along the direction normal to the interface with different magnetic field strength. The applied step length of electron probe line scanning is 10 ␮m.

was investigated by EPM-810Q electron probe microanalysis (EPMA). 3. Results and discussion The formation and growth of diffusion layers at the Cu/Zn interface can be interpreted as follows: firstly, atoms from the liquid metal penetrated into solid metal Cu through grain-boundary paths to reach the solubility limit. Then, as a consequence of diffusional mixing, a copper zinc layer appeared. At the same time, the active copper atoms diffused into the liquid metal region. Finally, due to phase transformation and recrystallization, the diffusion layers composed of columnar crystals were formed. Fig. 2(a) and (b) are the optical micrographs showing the microstructure for Cu/Zn couples annealed at 723 K without and with magnetic field, respectively. The arrow at the right side of Fig. 2b indicates the direction of the applied magnetic field. Whatever the magnetic field strength and gradient, the interface is always composed of two intermetallic compounds ␥ (Cu5 Zn8 ) and ␧ (CuZn5 ) according to EPMA (Fig. 3). A third intermetallic compound ␤ should be formed according to phase diagram analysis [7]. Apart from one ␧ layer was with irregular edges, which located adjacent to the zinc matrix, the other ␥ layer was found to be flat. But the ␤ layer is so thin that it is shown as a bright golden line located between the ␥ layer and Cu side in the optical micrographs. The width of ␤ phase can be described as: w␤ = x␤/␥ − xCu/␤ , where x␤/␥ is the migration distance of the ␤/␥ interface and xCu/␤ is the migration distance of the Cu/␤ interface. The formation of ␤ layer, which was not evident with the experimental results (EPMA), could be a consequence of the small difference of migration distance between these two interfaces. As a result, in these experiments the phase composition of diffusion layers has not been affected by high magnetic fields. 3.1. Diffusion under uniform high magnetic fields As shown in Fig. 3, with the modification of magnetic field intensity, the thickness of the diffusion layers along the EPMA line scanning changes markedly. This means that the thickness of the total and individual diffusion layers may change with the

Fig. 4. Diffusion layers thickness in copper–zinc couples annealed under different magnetic field strength.

variation of magnetic field conditions. Fig. 4 indicates the nonmonotonic relationship between the magnetic field intensity and the diffusion layer thickness. Here, the thickness of ␤ intermetallic phase was not measured because it is negligible comparing with that of ␥ and ␧ phase. It is found that the curve of thickness exhibits a decrease with a peak centred on a field of about 8 T when B changes from 0 to 10.5 T. The total thickness of diffusion layers in the sample annealed at a magnetic field of 8.8 T is almost equal to that of the ordinarily (B = 0) annealed sample. However, above 10.5 T, the thickness of the diffusion layers increases slightly with increasing the magnetic field strength. This phenomenon can be attributed to two effects of uniform high magnetic field, suppressing natural convection and inducing thermo-electromagnetic convection [8]. During the penetration process of the liquid Zn into the solid Cu, the convection of the liquid metal (located near the interface) exerts a direct influence on diffusion behavior, because buoyant convection and thermocapillary convection will affect mass and heat transport processes on which the diffusion depends. Stronger convection currents can increase the concentration of solute at the solid/liquid interface. Thereby, the diffusion flux of solute is increased by convection according to Fick’s first law: J = −D(∂C/∂X) (where J is the diffusion flux, D the diffusion coefficient and ∂C/∂X is the concentration gradient). On the one hand, a uniform magnetic field can retard the growth of diffusion layers indirectly by suppressing the natural convection. Many researchers have applied a magnetic field to suppress the convection in liquid metals and semiconductors due to their large electrical conductivities [9,10]. The retarding effect of magnetic field on natural convection is responsible for the decrease of the thickness of diffusion layers with the magnetic field strength increasing. On the other hand, at a given magnetic field strength, a new type of convection named thermoelectromagnetic convection (TEMC) [11] would appear. Since a temperature gradient at the solid/liquid interface always results in thermoelectric currents in liquid metals (Seebeck effect), the interaction of the magnetic field with the thermoelectric currents then products a thermo-electromagnetic flow. However, convection due to TEMC is also damped by the high magnetic field

D.-g. Li et al. / Materials Science and Engineering A 495 (2008) 244–248

Fig. 5. Distribution of the area fractional variation of ␧ phase in the Zn matrix under magnetic fields of (a) 0 T, (b) 8.8 T and (c) 12 T. The bright dendritic in the micrographs are ␧ phase and the black point on the top of every micrograph represent the area fraction of ␧ phase within an experimental error. The area fraction of ␧ phase was measured on the cross-section observation of the samples by LEICA metallographic analysis software.

[12]. The magnitude of TEMC therefore goes through a maximum as a function of the magnetic field. The appearance and disappearance of TEMC corresponds to the experimental results that an increase in thickness of diffusion layers with B = 8 T and a significant decrease in it with B = 10.5 T occurred. As to the slight enhancement in the growth of the diffusion layers, when the magnitude of magnetic field strength is higher than 10.5 T, an assumption could be made that the high magnetic field can apply heavy intensity magnetization energy on the atomic scale, and hence can increase the transition frequency of atoms, Γ , as mentioned in reference [13]. Due to the fact that the increment of Γ by magnetic field could result in the enhancement of diffusion coefficient D, one can expect that the thickness of diffusion layers will increase when B is high enough. In addition to thickness measurement, the area fractional variation of ␧ phase in Zn matrix for every sample was measured by LEICA metallographic analysis software. The microstructure in Zn matrix indicates that the area fractional variation of ␧ phase as shown in Fig. 5 agrees well with the change in thickness of diffusion layers under the same magnetic field strength. It can be considered that the higher fraction of ␧ phase is corresponding to the larger thickness in diffusion layers. These results reveal that the diffusion of Cu atoms in Zn is significantly affected by the magnetic fields as well. 3.2. Diffusion under high gradients of magnetic field For comparison, the thickness of diffusion layers at a magnetic field of 8.8 T with different magnetic field gradients is shown in Fig. 6. It is found that the thickness is decreased about 20 ␮m in a magnetic field gradient of 166 T2 m−1 whether the magnetic field gradient is negative (−) or positive (+). To clarify the origin of the observed magnetic field gradient effect on diffusion, we should consider the influence of magnetic force on the atoms transition [14] and on the convection [15] in the

247

Fig. 6. The thickness of diffusion layers as a function of the magnetic field gradients (B = 8.8 T).

liquid metal Zn (diamagnetic material). Furthermore, the magnetic free energy gradient that is involved by a magnetic field gradient may be possible to affect the vacancy formation energy [5]. About the effect of magnetic field gradient on interdiffusion between Cu and Zn at solid/liquid interface should be examined deeply in the future. 4. Conclusions High magnetic fields were applied to study the growth of diffusion layers at the interface zones of Zn/Cu diffusion couples. The results reveal that both the strength and the gradient of high magnetic fields can control the diffusion behavior effectively. (i) Whatever the magnetic field strength and gradient, the product at the interface always contains two intermetallic compounds: ␥ and ␧. (ii) The diffusion layers thickness exhibits a non-monotonic change with the magnetic field strength because of the two effects of the uniform magnetic field on convection. (iii) A negative or positive field gradient can retard zinc diffusion in copper resulting in the decrease of the thickness of diffusion layers. Acknowledgements This work was supported by the National Natural Science Foundation of China (Grant No. 50374027), the Program for New Century Excellent Talents in University, PR China (Grant No. NCET-06-0289) and the 111 project (Grant No. B07015). References [1] M. Kasuga, T. Takano, S. Akiyama, K. Hiroshima, K. Yano, K. Kishio, J. Cryst. Growth 275 (2005) e1545–e1550. [2] A.V. Pokoev, D.I. Stepanov, I.S. Trofimov, V.F. Mazanko, Phys. Status Solidi A 137 (1993) K1–K3.

248

D.-g. Li et al. / Materials Science and Engineering A 495 (2008) 244–248

[3] Y.Y. Khine, R.M. Banish, J.I.D. Alexander, J. Cryst. Growth 250 (2003) 274–278. [4] T. Miyake, Y. Inatomi, K. Kuribayashi, Jpn. J. Appl. Phys. Part 2 (Lett.) 41 (2002) L811–L813. [5] S. Nakamichi, S. Tsurekawa, Y. Morizono, T. Watanabe, M. Nishida, A. Chiba, J. Mater. Sci. 40 (2005) 3191–3198. [6] H. Nakajima, S. Maekawa, Y. Aoki, M. Koiwa, Trans. Jpn. Inst. Met. 26 (1985) 1–6. [7] T.B. Massalski, J.L. Murray, L.H. Bennett, H. Baker (Eds.), Binary Alloy Phase Diagrams, vol. 1, American Society for Metals, Ohio, USA, 1986, p. 981. [8] Q. Wang, D.G. Li, K. Wang, Z.Y. Wang, J.C. He, Scripta Mater. 56 (2007) 485–488.

[9] V. Botton, P. Lehmann, R. Bolcato, R. Moreau, Energy Convers. Manage. 43 (2002) 409–416. [10] Y. Kishida, K. Takeda, Int. J. Appl. Electromag. Mater. 2 (1992) 291–299. [11] R. Moreau, O. Laskar, M. Tanaka, D. Camel, Mater. Sci. Eng. A173 (1993) 93–100. [12] A. Croll, F.R. Szofran, P. Dold, K.W. Benz, S.L. Lehoczky, J. Cryst. Growth 183 (1998) 554–563. [13] J. Zhao, P. Yang, F. Zhu, C.Q. Cheng, Scripta Mater. 54 (2006) 1077–1080. [14] J.H. Huang, Diffusion of Metals and Alloys, Mechanical Industry Press, Beijing, China, 2001, p. 26. [15] T. Tagawa, R. Shigemitsu, H. Ozoe, Int. J. Heat Mass Transfer 45 (2002) 267–277.