Diffusive phase transformation in a Cu Zn alloy under rapid heating by electropulsing

Transcription

1 PHILOSOPHICAL MAGAZINE LETTERS, MAY 2004 VOL. 84, NO. 5, Diffusive phase transformation in a Cu Zn alloy under rapid heating by electropulsing Yizhou Z. Zhouy Lehrstuhl Werkstoffkunde und Technologie der Metalle (WTM), Universita t Erlangen-Nu rnberg, Martensstr. 5, Erlangen, Germany Wei Zhang, Jingdong Guo and Guanhu He Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang , PR China [Received in final form 22 December 2003 and accepted 9 January 2004 ] Abstract Electropulsing with a damped oscillating waveform was employed to heat cycle a Cu Zn alloy containing a and b 0 phases. It was found that significant long-range diffusion and then a diffusive phase transformation a þ b 0!b can occur in the alloy during the course of heating, even when the heating rate during the electropulsing treatment is very high (10 6 Ks 1 ). The phenomenon is different from the common cases of rapid heating and rapid cooling. It implies that electropulsing can dramatically enhance diffusion in the alloy. } 1. Introduction In previous work (Zhou et al. 2002a, b, 2003), we reported that the microstructure of metallic materials could be dramatically refined by the rapid heating cooling cycles of an electropulsing treatment. It was proposed that the mechanism of refinement was related to a solid-state phase transformation following nucleation and growth mechanisms. This implied that the phase transformation during heating was not a non-diffusive phase transformation (a martensitic or a massive phase transformation). We know that solid-state phase transformations can generally be classified into two kinds: diffusive and non-diffusive. Under conditions of rapid heating and/or rapid cooling, since there is no significant long-range diffusion on a short time scale, a diffusive phase transformation cannot be expected to take place. The heating rate of high-current-density electropulsing can reach Ks 1. Thus, we need to know whether or not a diffusive phase transformation can take place within such a short time, although it is well known that the flow of electric current can give rise to mass transport, a process known as electromigration (Huntington and Grone 1961, Blech 1976, Ho and Kwok 1989). In the present work, a diffusive phase transformation in a Cu Zn alloy was found during the rapid heating cycle of an electropulsing treatment. The phenomenon cannot be explained by classical electromigration theory. yauthor for correspondence. Yizhou.Zhou@ww.uni-erlangen.de. Philosophical Magazine Letters ISSN print/issn online # 2004 Taylor & Francis Ltd DOI: /

2 342 Y. Z. Zhou et al. } 2. Experimental details The nominal chemical composition of the Cu Zn alloy used in the present investigation was 59.4 wt% Cu 40.6 wt% Zn. The alloy was annealed at 873 K for 30 min and then cooled in air. After the thermal treatment, the alloy had two phases at room temperature: the a phase and the b 0 phase. Flat samples were cut from the annealed alloy and were divided into three groups: H samples, that were not to undergo any further treatment; HE samples, samples to be treated by electropulsing; HL samples, samples to be treated by a pulsed laser. Electropulsing was performed under ambient conditions, using the discharge of capacitor banks. Each sample was treated twice, the parameters of the electropulsing being chosen to be the same in all cases. The interval between the two electropulsing treatments was long enough to ensure that the sample cooled to room temperature before the second treatment. In order to obtain a high cooling rate in the effective middle part of sample after electropulsing, the size of the two ends was made much larger than that of the middle part so that current density was much smaller there. The middle parts of these HE samples were 3 mm long, 3 mm wide and 1.5 mm thick. The ends of each sample were put into Cu electrodes during electropulsing treatment, so that their temperature rise was very small and they could be regarded as at room temperature. A high cooling rate could thereby be obtained in the effective part of the sample. The waveform of the electropulsing was shown to be a damped oscillating wave using a Rogowski coil and a TDS3012 Tektronix digital storage oscilloscope (figure 1). The pulse duration was about 800 ms and the period of oscillation t p ¼ 120 ms. The maximum current density was j m ¼ 18.0 ka mm 2. With these parameters, the temperature of the sample by Joule heating was measured to be about 1110 K by a K-type thermocouple (diameter, 80 mm) soldered to the sample at its middlemost part. This temperature is very close to the melting point (1175 K) of the alloy. HL samples were 40 mm long, 20 mm wide and 3 mm thick. They were subjected to pulsed treatment from a Nd-doped yttrium aluminium garnet laser device. The pulse duration was about 1000 ms. The surface temperature of the illuminated part of each sample could be controlled by adjustment of the voltage applied to the laser device and the distance between the sample and a focusing lens. It was controlled to be very close to the melting point of the alloy. As for the electropulsing experiments, Current density (ka/mm 2 ) Time (µs) Figure 1. Waveform of electropulsing.

3 Diffusive phase transformation in a Cu Zn alloy 343 the HL samples were treated twice, the experimental conditions being exactly the same. The interval between the two laser treatments was long enough to allow cooling to room temperature before the second treatment. A JSM-6301F JEOL field emission scanning electron microscope and a JEM JEOL transmission electron microscope were employed to observe the microstructure of the samples. The samples for scanning electron microscopy examination were polished and etched. The chemical composition of microareas in the samples was determined with an energy-dispersive X-ray spectroscope in the scanning electron microscope. The characteristics of the untreated H and the electropulsed HE samples were examined by X-ray diffractometry using Cu ka radiation. } 3. Experimental results From microstructure observations of the samples, we found that the electropulsing and laser treatments, although both rapid, differ in their effects. Figure 2 shows scanning electron micrographs of the H, HE and HL samples. In Figure 2 (a), the raised microstructure is the a phase and the cupped microstructure is the b 0 phase. The average grain size of the a-phase microstructure is about 30 mm. The composition of the a phase was found to be 62.8 wt% Cu 37.2 wt% Zn, while that of the b 0 -phase was 55.9 wt% Cu 44.1 wt% Zn by energy-despersive spectroscopy (EDS) analysis. In Figure 2 (b), the microstructure of the a phase can hardly be seen; that is the sample consists of almost a single b 0 phase. EDS analysis showed that the composition of the microstructure in the figure is uniform, and of composition 59.6 wt% Cu 40.4 wt% Zn, which is almost the same as the nominal composition of the alloy. In figure 2 (c), although the temperature of the surface during laser treatment was close to that during electropulsing, one sees that the a and b 0 phases still remain after this treatment. EDS analysis showed that the composition of the a phase is 62.5 wt% Cu 37.5 wt% Zn, while that of the b 0 phase is 56.1 wt% Cu 43.9 wt% Zn, the values almost the same as those for the untreated H samples. The results of the EDS analysis mean that diffusion between the a phase and the b 0 phase is strong during electropulsing treatment and that the average displacement of the atoms is no less than 10 mm within the pulse duration of 800 ms. In combination with X-ray diffraction (XRD) analysis and transmission electron microscopy (TEM) observations, it can be confirmed that the microstructure of the electropulsed samples is almost a single b 0 phase. Figure 3 shows XRD patterns of the untreated H samples and the electropulsed HE samples. The diffraction peaks of the a phase can hardly be seen in the HE sample. Figure 4 (a) shows a bright-field TEM image taken from the HE sample. One can see that there are three grains in the figure. Selectedarea electron diffraction patterns of the three grains were found to be exactly the same; one of these is shown in figure 4 (b). Of course, in order to obtain the diffraction pattern shown in figure 4 (b), it was necessary to rotate the different grains by different amounts. Analysis of the diffraction pattern showed that the microstructure shown in figure 4 (a) is the b 0 phase. It must be pointed out that the a phase in HE samples could not be found by TEM observations. However, many coarse-grained a-phase microstructures could be observed on the surface of HL samples by TEM. According to these experimental results, almost all the a phase transforms to the b 0 phase after double electropulsing treatment. However, the same does not occur after a double pulsed laser treatment.

4 344 Y. Z. Zhou et al. (a) β α (b) β (c) β α Figure 2. 10µm Scanning electron micrographs of samples: (a) without any treatment; (b) treated twice by electropulsing; (c) treated twice by pulsed laser excitation. } 4. Discussion and conclusions Figure 5 shows a local Cu Zn binary alloy phase diagram (Massalski et al. 1990), in which the nominal composition of the alloy used in this work is shown as a straight line A; the compositions of the a and b 0 phases in the untreated H samples are indicated as straight lines B and C respectively. According to this phase diagram

5 Diffusive phase transformation in a Cu Zn alloy 345 α(111) β (110) Intensity α(200) β (200) α(220) β (211) a b Figure θ (deg) XRD patterns of samples: (a) without any treatment; (b) treated twice by electropulsing. (a) 0.5µm (b) 112 [110], CsCl-type structure Spots of second diffraction Figure 4. (a) A bright-field TEM image; (b) the corresponding selected-area electron diffraction pattern taken from a sample treated twice by electropulsing. and knowledge of phase transformation of the alloy (Brooks 1982), the course of the microstructure changes in the alloy during heating cooling cycles is as follows. During the heating cycle, the a and the b 0 phases transform to the b phase by longrange atom drift, when the temperature is higher than the phase transformation

6 346 Y. Z. Zhou et al B A C Temperature C C C β α C L C Weight Percent Zinc Weight percent zinc Figure 5. Local Cu Zn binary alloy phase diagram. β temperature of þ 0! (about 1003 K). During cooling, the a phase will precipitate from the b phase by long-range diffusion of atoms, if the cooling rate is not too high. However, the high-temperature b phase will be quenched to the b 0 phase, and the a phase will be prevented from precipitating from the b phase in the case of rapid cooling, because there is no significant long-range diffusion during the cooling cycle. It must be emphasized that the phase transformation þ 0! is a diffusive phase transformation. Namely, the phase transformation cannot take place without long-range diffusion. In the present work, the heating rates during electropulsing and laser treatment could be defined as the ratio of the maximum temperature to the pulse duration. According to the above values, the heating rate during electropulsing is almost the same as that during the pulsed laser treatment, both being about Ks 1. However, the cooling rate during the electropulsing experiment was far less than that during the laser experiment, because the cooling rate of the electropulsed HE samples was controlled by their larger ends, while that of the laser-treated HL samples resulted directly from their matrices. In combination with the above analysis concerning the phase transformation of the alloy, it can be concluded that the single b 0 phase microstructure should also be formed in the illuminated part after laser treatment, if the a and the b 0 phases can transform to the b phase during the heating cycle of the laser treatment. However, the experimental results showed that the a and the b 0 phases still remain in the illuminated part after laser treatment. The reason for the difference between the microstructures in the two treatments must be that there is significant long-range diffusion during the heating cycle of the electropulsing treatment, while there is no significant long-range diffusion during the heating cycle of the laser treatment. Rapid long-range migration of atoms within such a short time is unusual. In an electropulsing system, a flux of atoms can be driven by a composition gradient, an electric current and a temperature gradient. Based on the experimental

7 Diffusive phase transformation in a Cu Zn alloy 347 results of the laser treatment, it is clear that the difference between the atomic densities of the a and b 0 phases provides an insufficient composition gradient to allow long-range migration of atoms. Classical electromigration theory cannot explain the phenomenon of rapid diffusion either. The average atom drift velocity V ie resulting from an electric current is given by Huntington and Grone (1961), Blech (1976), Ho and Kwok (1989) V ie ¼ J ie c i ¼ D i kt ez j ¼ D 0 kt ez j exp Q i, ð1þ kt where J ie is the flux of atoms, c i is the atomic density, D i the diffusion coefficient, k Boltzmann s constant, T the temperature, e the electronic charge, Z the effective charge, j the current density, the resistivity, D 0 the diffusion pre-exponential factor and Q i the activation energy. For the investigated alloy, Cu is the solvent and Zn the solute, with the drift velocity of Zn atoms being higher than that of Cu atoms. Therefore, an estimate of the drift velocity is based on the behaviour of the Zn atoms. According to equation (1) with the values D 0 ¼ m 2 s 1 (Guy and Hren 1974), Q i ¼ J mol 1 (Guy and Hren 1974), Z ¼ 23.3 (Ho and Kwok 1989) and j ¼ j m ¼ 18.0 ka mm 2, it can be found that V ie is of the order of 10 2 mms 1. It is clear therefore that the drift velocity is very low and the atomic displacement cannot reach 10 mm by this process during the electropulsing treatment. In addition, the electropulsing is an alternating treatment and the current, and hence atom drift direction, will undergo reversals. Since the properties of the a and b 0 phases are different, their temperatures due to Joule heating can be different during the electropulsing treatment. Therefore, an atom flow resulting from a temperature gradient (thermotransport) could occur. However, thermotransport is generally a second-order effect compared with electromigration (Hummel 1994). Furthermore, any temperature difference between the phases will disappear in very short time in this work. Therefore, the phenomenon in the present work is unlikely to be explained in terms of a temperature gradient. According to classical electromigration theory, the diffusion coefficient cannot be changed by an electric current, which can be inferred from the definition of D i in equation (1); D i is the diffusion coefficient in a current-free system. However, an unchanged diffusion coefficient cannot explain the phenomenon in this work. Therefore, we think that the main reason for the rapid atom drift in the present work arises from a diffusion coefficient that is dramatically enhanced by the electropulsing waveform. Possibilities for enhancing the diffusion coefficient by electropulsing include (a) an increase in the pre-exponential factor D 0, which could result from heating on an atomic scale, increasing the atom vibration frequency and also the activation entropy, since we know D 0 ¼ fa 2 exp (S/k) (Bocquet et al. 1983), where f is a constant for a certain system, a is the lattice constant, is the atom vibration frequency and S is the change of activation entropy, and

8 348 Diffusive phase transformation in a Cu Zn alloy (b) a reduction in the activation energy Q i, which could result from a reduction in the strength of opposing atom motion and an increase in defect density. However, these mechanisms are only tentative suggestions. Similar experimental results were found in other recent work, in which it was also suggested that a possible explanation lay in the diffusion coefficient (Bertolino et al. 2002). In summary, a diffusive phase transformation, a þ b 0!b, takes place in a Cu Zn alloy containing a and b 0 phases during rapid heating by electropulsing treatment. In common cases of rapid heating, a diffusive transformation is not found. For example, no equivalent transformation occurs when the heating is carried out by laser pulse excitation. The diffusive transformation cannot be explained by classical electromigration theory and is possibly associated with a dramatic enhancement of the diffusion coefficient in the investigated system as a consequence of the electric field used in the electropulsing treatment. ACKNOWLEDGEMENTS Financial support by The National Natural Science Foundation of China (grants and ) and The National Major Basic Research Development Program Item of China (grant G ) is acknowledged. The authors thank Professor Lu Ke and Wang Yuanming for many helpful discussions. REFERENCES BERTOLINO, N., GARAY, J., ANSELMI-TAMBURINI, U., and MUNIR, Z. A., 2002, Phil. Mag. B, 82, 969. BLECH, I. A., 1976, J. appl. Phys., 47, BOCQUET, J. L., BREBEC, G., and LIMOGE, Y., 1983, Physical Metallurgy, third edition, edited by R. W. Cahn and P. Haasen (Amsterdam: North-Holland), chapter 8, p BROOKS, C. R., 1982, Heat Treatment, Structure and Properties of Nonferrous Alloys, (Materials Park, Ohio: American Society for Metals), pp GUY, A. G., and HREN, J. J., 1974, Elements of Physical Metallurgy, Third edition (Reading, Massachusetts: Addison-Wesley), p HO, P. S., and KWOK, T., 1989, Rep. Prog. Phys., 52, 301. HUMMEL, R. E., 1994, Int. Mater. Rev., 39, 97. HUNTINGTON, H. B., and GRONE, A. R., 1961, J. Phys. Chem. Solids, 20, 76. MASSALSKI, T. B., OKAMOTO, H., SUBRAMANIAN, P. R., and KACPRZAK, L., 1990, Binary Alloy Phase Diagrams, second edition (Materials Park, Ohio: American Society for Metals), p ZHOU, Y. Z., ZHANG, W., SUI, M. L., LI, D. X., HE, G. H., and GUO, J. D., 2002a, J. Mater. Res., 17, 921. ZHOU, Y. Z., ZHANG, W., WANG, B. Q., and GUO, J. D., 2003, J. Mater. Res., 18, ZHOU, Y. Z., ZHANG, W., WANG, B. Q., HE, G. H., and GUO, J. D., 2002b, J. Mater. Res., 17, 2105.