- Email: [email protected]
Effect of cooling rate on the solidification of Cu58Co42
Materials Science and Engineering A 449–451 (2007) 644–648
Effect of cooling rate on the solidification of Cu58Co42 S. Curiotto a,b,c,∗ , N.H. Pryds a , E. Johnson a,b , L. Battezzati c a
Materials Research Department, AFM 228, Risø National Laboratory, Frederiksborgvej 399, DK-4000 Roskilde, Denmark b Niels Bohr Institute, Nanoscience Centre, University of Copenhagen, Copenhagen, Denmark c Dipartimento di Chimica IFM, Centro di Eccellenza NIS, Universit` a di Torino, Via P. Giuria 9, 10125 Torino, Italy
Received 21 August 2005; received in revised form 14 December 2005; accepted 24 February 2006
Abstract This paper deals with Co42 Cu58 (at.%) alloy processed to obtain a cooling rate gradient during solidification. The Co–Cu system displays a metastable miscibility gap in the liquid state. Undercooling of the melt beyond a certain limit results in liquid separation and formation of droplets of the minority phase in a matrix of the dominant phase. The material was melted and cast in a wedge shaped copper mould to obtain a gradient of cooling rates. The cooling history of the alloy during the solidification was monitored by insertion of thermocouples at different positions in the mould. During cooling, the melt separated into droplets of Co-rich and Cu-rich phases. The Co-rich droplets are mainly coagulated along the major axis of the samples and an explanation of this segregation mechanism is proposed. The distribution of the size of the Co-rich particles was measured in various areas along the sample and correlated to the cooling rate. © 2006 Elsevier B.V. All rights reserved. Keywords: Mould casting; Copper alloys; Metastable monotectic; Undercooling; Monotectic
1. Introduction Under conventional solidification conditions, alloys with a miscibility gap in the liquid state display a clear separation of the liquid in two layers, with the denser phase at the bottom. In some alloy systems the miscibility gap is metastable and the liquid–liquid separation can be achieved only by undercooling the liquid; this is the case for some Cu-based alloys. In the Cu–Co system the existence of the metastable liquid–liquid separation was reported for the first time by Nakagawa [1], who measured the change of magnetic susceptibility as a function of demixing and examined the microstructure of quenched specimens. Later, Elder et al. [2] determined experimentally the position of the Cu–Co miscibility gap by processing a range of samples with different nominal composition using electromagnetic levitation and studying the composition by electron microprobe analysis. They also carried out a prelimi-
∗ Corresponding author at: Niels Bohr Institute, Nanoscience Centre, University of Copenhagen, Copenhagen, Denmark. Tel.: +45 46775853; fax: +45 46775758. E-mail address: [email protected] (S. Curiotto).
0921-5093/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2006.02.375
nary work to model the phase diagram considering both stable and metastable equilibria. The effect of high cooling rates on the microstructure of Cu–Co alloys was studied by Munitz and Abbaschian [3–5]. They used scanning electron microscopy to examine samples processed by electron beam surface melting and related their microstructure to the undercooling. Recently, several authors investigated the extension of the miscibility gap, by differential scanning calorimetry or by directly measuring the temperature of the samples during the cooling from the liquid [6–9] while they observed the microstructure of the specimens after the solidification. Accurate studies of the microstructures formed during the solidification were carried out by Robinson et al. [10] and Cao et al. [8]. They also examined the relationships between undercooling and radius of the Co-rich droplets, finding that the maximum diameter of Co-rich spheres increased with the undercooling. Recently, Lu et al. [11] performed rapid solidification experiments by drop tube processing and studied the size distribution of Cu-rich particles formed in the matrix during the solidification. In this work the dependence of microstructure on cooling rate in a Cu58 Co42 alloy is studied using a wedge shaped copper mould and then characterising the microstructure with an optical and scanning electron microscope.
S. Curiotto et al. / Materials Science and Engineering A 449–451 (2007) 644–648
2. Experimental Co–Cu alloys have been synthesized from highly pure Cu (99.999%) and Co (99.9%). The composition Cu58 Co42 (at.%), corresponding to the maximum of the miscibility gap [6,12], has been chosen to facilitate the liquid–liquid separation. The components have been pre-alloyed with a total mass of approximately 400 g in an arc-melting furnace, evacuating and purging the chamber several times with high purity Ar and using lumps of Zr as getters. The master-alloy has then been cast into a copper wedge mould to obtain a gradient of cooling rates. The casting has been carried out in a closed chamber evacuated several times and re-filled with inert gas (Ar). The alloy has been induction melted in a graphite crucible and poured into a wedge shaped copper mould closed at the bottom, 8 cm high, 2.4 cm thick and 4 cm wide. The temperature–time curve during the solidification of the alloy has been recorded with Pt–PtRh thermocouples covered with a ceramic compound of high thermal conductivity and high electric resistance. The thermocouples have been inserted in the center of the mould at three different positions along the main axis: 7 mm (A), 42 mm (B) and 62 mm (C), respectively from the wedge tip (see insert ‘a’ of Fig. 1). The samples have been sectioned vertically along the major axis, polished and examined without etching by optical and scanning electron microscopy (SEM). X-ray diffraction (XRD) of different areas of the samples has been performed to determine the structure of the phases using a wide-angle goniometer in the Bragg–Brentano configuration with Co K␣ radiation. 3. Results and discussion A typical temperature–time curve recorded from thermocouple at position (B) is shown in the insert ‘b’ of Fig. 1. The maximum cooling rate has been determined from the slope of the curves. An empirical relationship between the cooling rate . (T in K/s) and the half thickness of the wedge (z in mm), i.e.
Fig. 1. (a) Schematic illustration of the wedge mould. (b) Temperature as a function of time for thermocouple (B). (c) Cooling rate (T˙ ) vs. half thickness of the wedge (z). The full squares are the experimentally determined points and the line is the fit of Eq. (1) to the data points.
645
the distance from the walls to the central axis of the mould, has been used to fit the data [13,14]: T˙ = kzn
(1)
where k and n are constants dependent on the material and on the geometry of the mould. Eq. (1) can be linearized and the linear fit is shown in Fig. 1c. We have found k = 2042 and n = −1.8, respectively. After solidification, the examination of the solidified microstructure revealed that the wedge shaped ingot contains two phases: (1) Co-rich particles mostly found in the middle of the section, along the major axis, and (2) Cu-rich matrix, which fills the space around the particles. The Co-rich particles are approximately spheroidal, characteristic shape often observed when a liquid is demixed. The effect of cooling rate on the particle size has been studied by measuring areas of the Co-rich droplets by image analysis of micrographs taken by optical microscopy along the major axis of the wedge. Fig. 2a shows the distribution of the total area occupied by a specific class of particles at different locations along the wedge. The measured particles sizes have been related to the local cooling rate calculated via Eq. (1) and superimposed in Fig. 2a. The local microstructure at two different positions along the wedge is also shown in Fig. 2b and c. From these micrographs, the differences in the droplets size can be clearly seen. The general tendency as seen from this figure is that the size of the Co-rich particles increases with the distance from the tip of the specimen, i.e. as the cooling rate is decreased. At low cooling rates, the demixed Co-rich droplets coalesced more as a result of remaining longer in the liquid state. Because of the shape of the mould, a gradient of cooling rates exists between the tip and the top of the wedge resulting in a large variation of the microstructure. Close to the tip (z = 0.5 mm) the cooling rate calculated using Eq. (1) is about 7 × 103 K/s; the droplets size is relatively small, there are no Co-rich regions with area larger than 8000 m2 , few smaller than 10 m2 and most of them are between 10 and 300 m2 . Going further from the tip, the number of small spheres decreases while the number of large particles increases. Beyond 2 cm from the tip the particles are very large, often around one millimetre in diameter. Fig. 3 shows a SEM micrograph of an upper part of the sample. The dark regions are Co-rich, while the bright are Cu-rich as confirmed by energy dispersive X-ray spectroscopy. Fine Corich droplets surround large particles of the same phase. These large spherical particles have a dark rim at the interface with the Cu-rich matrix and white fine Cu-rich spherical spheres in the core. This microstructure can be explained from the fact that at the top of the wedge, although the cooling rate is not high, the sample experienced high undercooling and remained liquid for a long time before nucleation took place. When the undercooled melt reaches the miscibility gap it separates in two liquids, one Cu-rich, the other Co-rich, as predicted by the phase diagram calculated according to [12] and shown in Fig. 4. The liquid–liquid separation is a continuous process, so the first Co-rich droplets formed contain much Cu, but continuing with demixing, i.e. lowering the temperature, those formed afterwards become richer in
646
S. Curiotto et al. / Materials Science and Engineering A 449–451 (2007) 644–648
Fig. 2. (a) Distribution of Co-rich particles in the Cu-rich matrix for different positions along the wedge. The distribution is given as percentage of area occupied by Co-rich particles for discrete dimensional classes. For each measured region the corresponding cooling rate (right y-axis) with error bar is also given. (b) and (c) Optical microscope images taken at the tip and at a position 2 cm away from the tip, respectively. In these micrographs the bright area is Cu-rich and the dark areas are Co-rich.
cobalt as the system follows the line of the miscibility gap. The coagulation is a continuous process as well, so the first droplets form the core of the large Co-rich particles. These inner liquid regions are too far from the boundary with the Cu-rich liquid to adjust the composition by means of diffusion through the interface. Therefore, as the temperature decreases, in order to follow the equilibrium, demixing must take place inside the Co-rich core of the large particles with formation of Cu-rich droplets. In Fig. 3 the small dark particles are apparently moving towards the nearest large Co-rich drop. The particles move toward the region with lower interfacial energy because of Marangoni motion. In fact, if there is a temperature gradient in the liquid, an interfacial tension gradient also exists, so the particles of the minor liquid phase move towards the warmer region as also found for Cu–Fe alloys [15]. This explains the reason why most of the Co is located in the middle of the wedge that was warmer than the walls. Besides, the demixing is an exothermic process [6] and
the heat is dissipated faster in the Cu-rich phase than in the Corich (Cu has a heat conductivity four times that of the Co [16]). Therefore, locally, the Co-rich regions might be at slightly higher temperature with respect to the Cu-matrix. Hence, as can be seen in Fig. 3, there is coalescence of the small Co-rich droplets towards the bigger ones. The same mechanism of coagulation can explain the egg-type core microstructures observed in Co-Cu alloys after high temperature differential scanning calorimetry [6–8,17]. Fig. 5 represents a section of a large Co-rich drop exactly in the middle of the wedge, with Cu-rich particles elongated perpendicularly to the major axis of the sample. The Cu-rich droplets are now the minor phase inside the Co-rich particle and so they move towards the warmer, inner zone, in the opposite direction of the extraction of heat. This explains also the elongated shape, perpendicular to the major axis of the wedge. The movement is towards the interior, as indicated by the higher
S. Curiotto et al. / Materials Science and Engineering A 449–451 (2007) 644–648
Fig. 3. SEM image obtained by backscattered electrons of a region at the top of the wedge. The Co-rich phase appears dark, the Cu-rich phase bright. Co-rich drop that is demixed internally and bordered by a dark rim. The large drop is surrounded by smaller droplets.
Fig. 4. Cu–Co phase diagram as calculated in [12]. The dashed line is the metastable miscibility gap. The thick line is the cooling path of the undercooled Cu58 Co42 alloy.
647
Fig. 6. SEM low magnification image obtained by backscattered electrons of a region at the top of the wedge. The Co-rich phase appears dark, the Cu-rich phase bright. It shows large Co-rich particles with Cu-rich drops inside and dendrites starting from the edges.
density of Cu-rich particles in the middle of the zone. They can be present because trapped during the coagulation of the Corich droplets or because demixed from the Co-rich liquid. Their large dimensions are due to the collision and coagulation of little droplets during the motion. Fig. 6 is taken at lower magnification in comparison with Fig. 3. It shows very large Co-rich regions with Cu-rich particles inside and dendrites starting from the edges. The shape of the small droplets is expected because a round form allows reduction of surface energy. The larger particles would need more time to accommodate their shape: before rearranging it they solidified and are not spherical. Besides, after the demixing and the solidification of the Co-rich particles, the Cu-rich liquid is still rich in Co, but it must reach the composition of the stable liquidus curve of the phase diagram. Therefore the excess of Co solidifies as dendrites starting from the previously solidified Co-rich large particles. The solidification then finishes with the peritectic transformation of the Cu-rich liquid. X-ray diffraction investigations show the presence of two fcc (face centred cubic) phases both at the tip and at the top of the sample, one richer in Co and the other in Cu. The -Co phase has not been observed. 4. Summary
Fig. 5. Optical microscope image from a region at the central axis of the wedge, the bright regions are Cu-rich and the dark are Co-rich. Cu-rich particles inside a large Co-rich drop; they solidified while moving towards the middle of the sample and elongated perpendicularly to the major axis of the wedge.
A Co42 Cu58 alloy has been solidified in a wedge shaped Cu mould to obtain a cooling rate gradient along the sample. The cooling rates have been measured during the casting by Pt–PtRh thermocouples and the relationship between the cooling rate and the half thickness of the wedge has been found. The maximum cooling rate, reached at the tip of the alloy, was about 7 × 103 K/s. The microstructure has been analysed by optical and scanning electron microscopy. Due to the relatively high cooling rate, the alloy experienced undercooling and liquid–liquid metastable phase separation with the formation of Co-rich droplets in a Cu-rich matrix. Because of Marangoni
648
S. Curiotto et al. / Materials Science and Engineering A 449–451 (2007) 644–648
motion, the particles move toward the warmer zones, in the opposite direction of the extraction of heat. The distribution of Co-rich particles in different zones of the wedge has been measured and related to the cooling rate. The Co-rich droplets had more time to coalesce and therefore become larger in the areas cooled more slowly. The structure of both the Cu-rich and the Co-rich phases observed is fcc; the -Co phase has never been found. Acknowledgment The work has been supported by the European Space Agency within the project “CoolCop” (ESA-MAP AO 99-010). References [1] Y. Nakagawa, Acta Metall. 6 (1958) 704–711. [2] S.P. Elder, A. Munitz, G.J. Abbaschian, Mater. Sci. Forum 50 (1989) 137–150. [3] A. Munitz, R. Abbaschian, J. Mater. Sci. 26 (1991) 6458–6466. [4] A. Munitz, R. Abbaschian, J. Mater. Sci. 33 (1998) 3639–3649.
[5] A. Munitz, R. Abbaschian, Metall. Mater. Trans. A 27 (1996) 4049–4059. [6] C.D. Cao, G.P. Gorler, D.M. Herlach, B. Wei, Mater. Sci. Eng. A 325 (2002) 503–510. [7] C.D. Cao, T. Letzig, G.P. Gorler, D.M. Herlach, J. Alloys Compd. 325 (2001) 113–117. [8] C.D. Cao, D.M. Herlach, M. Kolbe, G.P. Gorler, B. Wei, Scripta Mater. 48 (2003) 5–9. [9] I. Yamauchi, N. Ueno, M. Shimaoka, I. Ohnaka, J. Mater. Sci. 33 (1998) 371–378. [10] M.B. Robinson, D. Li, T.J. Rathz, G. Williams, J. Mater. Sci. 34 (1999) 3747–3753. [11] X.Y. Lu, C.D. Cao, M. Kolbe, B. Wei, D.M. Herlach, Mater. Sci. Eng. A 375–377 (2004) 1101–1104. [12] M. Palumbo, S. Curiotto, L. Battezzati, CALPHAD-Computer coupling of phase diagrams and thermochemistry 30 (2006) 171–178. [13] N.H. Pryds, X. Huang, Metall. Mater. Trans. A 31 (2000) 3155–3166. [14] N.H. Pryds, A.S. Pedersen, Metall. Mater. Trans. A 33 (2002) 3755–3761. [15] C.P. Wang, X.J. Liu, I. Ohnuma, R. Kainuma, K. Ishida, Science 297 (2002) 990–993. [16] R.D. Lide, Handbook of Chemistry and Physics, CRC Press, 2004, pp. 12–219. [17] S. Curiotto, N. Pryds, E. Johnson, L. Battezzati, Metallurgical and Materials Transactions A 37 (2006) 2361–2368.
Effect of cooling rate on the solidification of Cu58Co42 S. Curiotto a,b,c,∗ , N.H. Pryds a , E. Johnson a,b , L. Battezzati c a
Materials Research Department, AFM 228, Risø National Laboratory, Frederiksborgvej 399, DK-4000 Roskilde, Denmark b Niels Bohr Institute, Nanoscience Centre, University of Copenhagen, Copenhagen, Denmark c Dipartimento di Chimica IFM, Centro di Eccellenza NIS, Universit` a di Torino, Via P. Giuria 9, 10125 Torino, Italy
Received 21 August 2005; received in revised form 14 December 2005; accepted 24 February 2006
Abstract This paper deals with Co42 Cu58 (at.%) alloy processed to obtain a cooling rate gradient during solidification. The Co–Cu system displays a metastable miscibility gap in the liquid state. Undercooling of the melt beyond a certain limit results in liquid separation and formation of droplets of the minority phase in a matrix of the dominant phase. The material was melted and cast in a wedge shaped copper mould to obtain a gradient of cooling rates. The cooling history of the alloy during the solidification was monitored by insertion of thermocouples at different positions in the mould. During cooling, the melt separated into droplets of Co-rich and Cu-rich phases. The Co-rich droplets are mainly coagulated along the major axis of the samples and an explanation of this segregation mechanism is proposed. The distribution of the size of the Co-rich particles was measured in various areas along the sample and correlated to the cooling rate. © 2006 Elsevier B.V. All rights reserved. Keywords: Mould casting; Copper alloys; Metastable monotectic; Undercooling; Monotectic
1. Introduction Under conventional solidification conditions, alloys with a miscibility gap in the liquid state display a clear separation of the liquid in two layers, with the denser phase at the bottom. In some alloy systems the miscibility gap is metastable and the liquid–liquid separation can be achieved only by undercooling the liquid; this is the case for some Cu-based alloys. In the Cu–Co system the existence of the metastable liquid–liquid separation was reported for the first time by Nakagawa [1], who measured the change of magnetic susceptibility as a function of demixing and examined the microstructure of quenched specimens. Later, Elder et al. [2] determined experimentally the position of the Cu–Co miscibility gap by processing a range of samples with different nominal composition using electromagnetic levitation and studying the composition by electron microprobe analysis. They also carried out a prelimi-
∗ Corresponding author at: Niels Bohr Institute, Nanoscience Centre, University of Copenhagen, Copenhagen, Denmark. Tel.: +45 46775853; fax: +45 46775758. E-mail address: [email protected] (S. Curiotto).
0921-5093/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2006.02.375
nary work to model the phase diagram considering both stable and metastable equilibria. The effect of high cooling rates on the microstructure of Cu–Co alloys was studied by Munitz and Abbaschian [3–5]. They used scanning electron microscopy to examine samples processed by electron beam surface melting and related their microstructure to the undercooling. Recently, several authors investigated the extension of the miscibility gap, by differential scanning calorimetry or by directly measuring the temperature of the samples during the cooling from the liquid [6–9] while they observed the microstructure of the specimens after the solidification. Accurate studies of the microstructures formed during the solidification were carried out by Robinson et al. [10] and Cao et al. [8]. They also examined the relationships between undercooling and radius of the Co-rich droplets, finding that the maximum diameter of Co-rich spheres increased with the undercooling. Recently, Lu et al. [11] performed rapid solidification experiments by drop tube processing and studied the size distribution of Cu-rich particles formed in the matrix during the solidification. In this work the dependence of microstructure on cooling rate in a Cu58 Co42 alloy is studied using a wedge shaped copper mould and then characterising the microstructure with an optical and scanning electron microscope.
S. Curiotto et al. / Materials Science and Engineering A 449–451 (2007) 644–648
2. Experimental Co–Cu alloys have been synthesized from highly pure Cu (99.999%) and Co (99.9%). The composition Cu58 Co42 (at.%), corresponding to the maximum of the miscibility gap [6,12], has been chosen to facilitate the liquid–liquid separation. The components have been pre-alloyed with a total mass of approximately 400 g in an arc-melting furnace, evacuating and purging the chamber several times with high purity Ar and using lumps of Zr as getters. The master-alloy has then been cast into a copper wedge mould to obtain a gradient of cooling rates. The casting has been carried out in a closed chamber evacuated several times and re-filled with inert gas (Ar). The alloy has been induction melted in a graphite crucible and poured into a wedge shaped copper mould closed at the bottom, 8 cm high, 2.4 cm thick and 4 cm wide. The temperature–time curve during the solidification of the alloy has been recorded with Pt–PtRh thermocouples covered with a ceramic compound of high thermal conductivity and high electric resistance. The thermocouples have been inserted in the center of the mould at three different positions along the main axis: 7 mm (A), 42 mm (B) and 62 mm (C), respectively from the wedge tip (see insert ‘a’ of Fig. 1). The samples have been sectioned vertically along the major axis, polished and examined without etching by optical and scanning electron microscopy (SEM). X-ray diffraction (XRD) of different areas of the samples has been performed to determine the structure of the phases using a wide-angle goniometer in the Bragg–Brentano configuration with Co K␣ radiation. 3. Results and discussion A typical temperature–time curve recorded from thermocouple at position (B) is shown in the insert ‘b’ of Fig. 1. The maximum cooling rate has been determined from the slope of the curves. An empirical relationship between the cooling rate . (T in K/s) and the half thickness of the wedge (z in mm), i.e.
Fig. 1. (a) Schematic illustration of the wedge mould. (b) Temperature as a function of time for thermocouple (B). (c) Cooling rate (T˙ ) vs. half thickness of the wedge (z). The full squares are the experimentally determined points and the line is the fit of Eq. (1) to the data points.
645
the distance from the walls to the central axis of the mould, has been used to fit the data [13,14]: T˙ = kzn
(1)
where k and n are constants dependent on the material and on the geometry of the mould. Eq. (1) can be linearized and the linear fit is shown in Fig. 1c. We have found k = 2042 and n = −1.8, respectively. After solidification, the examination of the solidified microstructure revealed that the wedge shaped ingot contains two phases: (1) Co-rich particles mostly found in the middle of the section, along the major axis, and (2) Cu-rich matrix, which fills the space around the particles. The Co-rich particles are approximately spheroidal, characteristic shape often observed when a liquid is demixed. The effect of cooling rate on the particle size has been studied by measuring areas of the Co-rich droplets by image analysis of micrographs taken by optical microscopy along the major axis of the wedge. Fig. 2a shows the distribution of the total area occupied by a specific class of particles at different locations along the wedge. The measured particles sizes have been related to the local cooling rate calculated via Eq. (1) and superimposed in Fig. 2a. The local microstructure at two different positions along the wedge is also shown in Fig. 2b and c. From these micrographs, the differences in the droplets size can be clearly seen. The general tendency as seen from this figure is that the size of the Co-rich particles increases with the distance from the tip of the specimen, i.e. as the cooling rate is decreased. At low cooling rates, the demixed Co-rich droplets coalesced more as a result of remaining longer in the liquid state. Because of the shape of the mould, a gradient of cooling rates exists between the tip and the top of the wedge resulting in a large variation of the microstructure. Close to the tip (z = 0.5 mm) the cooling rate calculated using Eq. (1) is about 7 × 103 K/s; the droplets size is relatively small, there are no Co-rich regions with area larger than 8000 m2 , few smaller than 10 m2 and most of them are between 10 and 300 m2 . Going further from the tip, the number of small spheres decreases while the number of large particles increases. Beyond 2 cm from the tip the particles are very large, often around one millimetre in diameter. Fig. 3 shows a SEM micrograph of an upper part of the sample. The dark regions are Co-rich, while the bright are Cu-rich as confirmed by energy dispersive X-ray spectroscopy. Fine Corich droplets surround large particles of the same phase. These large spherical particles have a dark rim at the interface with the Cu-rich matrix and white fine Cu-rich spherical spheres in the core. This microstructure can be explained from the fact that at the top of the wedge, although the cooling rate is not high, the sample experienced high undercooling and remained liquid for a long time before nucleation took place. When the undercooled melt reaches the miscibility gap it separates in two liquids, one Cu-rich, the other Co-rich, as predicted by the phase diagram calculated according to [12] and shown in Fig. 4. The liquid–liquid separation is a continuous process, so the first Co-rich droplets formed contain much Cu, but continuing with demixing, i.e. lowering the temperature, those formed afterwards become richer in
646
S. Curiotto et al. / Materials Science and Engineering A 449–451 (2007) 644–648
Fig. 2. (a) Distribution of Co-rich particles in the Cu-rich matrix for different positions along the wedge. The distribution is given as percentage of area occupied by Co-rich particles for discrete dimensional classes. For each measured region the corresponding cooling rate (right y-axis) with error bar is also given. (b) and (c) Optical microscope images taken at the tip and at a position 2 cm away from the tip, respectively. In these micrographs the bright area is Cu-rich and the dark areas are Co-rich.
cobalt as the system follows the line of the miscibility gap. The coagulation is a continuous process as well, so the first droplets form the core of the large Co-rich particles. These inner liquid regions are too far from the boundary with the Cu-rich liquid to adjust the composition by means of diffusion through the interface. Therefore, as the temperature decreases, in order to follow the equilibrium, demixing must take place inside the Co-rich core of the large particles with formation of Cu-rich droplets. In Fig. 3 the small dark particles are apparently moving towards the nearest large Co-rich drop. The particles move toward the region with lower interfacial energy because of Marangoni motion. In fact, if there is a temperature gradient in the liquid, an interfacial tension gradient also exists, so the particles of the minor liquid phase move towards the warmer region as also found for Cu–Fe alloys [15]. This explains the reason why most of the Co is located in the middle of the wedge that was warmer than the walls. Besides, the demixing is an exothermic process [6] and
the heat is dissipated faster in the Cu-rich phase than in the Corich (Cu has a heat conductivity four times that of the Co [16]). Therefore, locally, the Co-rich regions might be at slightly higher temperature with respect to the Cu-matrix. Hence, as can be seen in Fig. 3, there is coalescence of the small Co-rich droplets towards the bigger ones. The same mechanism of coagulation can explain the egg-type core microstructures observed in Co-Cu alloys after high temperature differential scanning calorimetry [6–8,17]. Fig. 5 represents a section of a large Co-rich drop exactly in the middle of the wedge, with Cu-rich particles elongated perpendicularly to the major axis of the sample. The Cu-rich droplets are now the minor phase inside the Co-rich particle and so they move towards the warmer, inner zone, in the opposite direction of the extraction of heat. This explains also the elongated shape, perpendicular to the major axis of the wedge. The movement is towards the interior, as indicated by the higher
S. Curiotto et al. / Materials Science and Engineering A 449–451 (2007) 644–648
Fig. 3. SEM image obtained by backscattered electrons of a region at the top of the wedge. The Co-rich phase appears dark, the Cu-rich phase bright. Co-rich drop that is demixed internally and bordered by a dark rim. The large drop is surrounded by smaller droplets.
Fig. 4. Cu–Co phase diagram as calculated in [12]. The dashed line is the metastable miscibility gap. The thick line is the cooling path of the undercooled Cu58 Co42 alloy.
647
Fig. 6. SEM low magnification image obtained by backscattered electrons of a region at the top of the wedge. The Co-rich phase appears dark, the Cu-rich phase bright. It shows large Co-rich particles with Cu-rich drops inside and dendrites starting from the edges.
density of Cu-rich particles in the middle of the zone. They can be present because trapped during the coagulation of the Corich droplets or because demixed from the Co-rich liquid. Their large dimensions are due to the collision and coagulation of little droplets during the motion. Fig. 6 is taken at lower magnification in comparison with Fig. 3. It shows very large Co-rich regions with Cu-rich particles inside and dendrites starting from the edges. The shape of the small droplets is expected because a round form allows reduction of surface energy. The larger particles would need more time to accommodate their shape: before rearranging it they solidified and are not spherical. Besides, after the demixing and the solidification of the Co-rich particles, the Cu-rich liquid is still rich in Co, but it must reach the composition of the stable liquidus curve of the phase diagram. Therefore the excess of Co solidifies as dendrites starting from the previously solidified Co-rich large particles. The solidification then finishes with the peritectic transformation of the Cu-rich liquid. X-ray diffraction investigations show the presence of two fcc (face centred cubic) phases both at the tip and at the top of the sample, one richer in Co and the other in Cu. The -Co phase has not been observed. 4. Summary
Fig. 5. Optical microscope image from a region at the central axis of the wedge, the bright regions are Cu-rich and the dark are Co-rich. Cu-rich particles inside a large Co-rich drop; they solidified while moving towards the middle of the sample and elongated perpendicularly to the major axis of the wedge.
A Co42 Cu58 alloy has been solidified in a wedge shaped Cu mould to obtain a cooling rate gradient along the sample. The cooling rates have been measured during the casting by Pt–PtRh thermocouples and the relationship between the cooling rate and the half thickness of the wedge has been found. The maximum cooling rate, reached at the tip of the alloy, was about 7 × 103 K/s. The microstructure has been analysed by optical and scanning electron microscopy. Due to the relatively high cooling rate, the alloy experienced undercooling and liquid–liquid metastable phase separation with the formation of Co-rich droplets in a Cu-rich matrix. Because of Marangoni
648
S. Curiotto et al. / Materials Science and Engineering A 449–451 (2007) 644–648
motion, the particles move toward the warmer zones, in the opposite direction of the extraction of heat. The distribution of Co-rich particles in different zones of the wedge has been measured and related to the cooling rate. The Co-rich droplets had more time to coalesce and therefore become larger in the areas cooled more slowly. The structure of both the Cu-rich and the Co-rich phases observed is fcc; the -Co phase has never been found. Acknowledgment The work has been supported by the European Space Agency within the project “CoolCop” (ESA-MAP AO 99-010). References [1] Y. Nakagawa, Acta Metall. 6 (1958) 704–711. [2] S.P. Elder, A. Munitz, G.J. Abbaschian, Mater. Sci. Forum 50 (1989) 137–150. [3] A. Munitz, R. Abbaschian, J. Mater. Sci. 26 (1991) 6458–6466. [4] A. Munitz, R. Abbaschian, J. Mater. Sci. 33 (1998) 3639–3649.
[5] A. Munitz, R. Abbaschian, Metall. Mater. Trans. A 27 (1996) 4049–4059. [6] C.D. Cao, G.P. Gorler, D.M. Herlach, B. Wei, Mater. Sci. Eng. A 325 (2002) 503–510. [7] C.D. Cao, T. Letzig, G.P. Gorler, D.M. Herlach, J. Alloys Compd. 325 (2001) 113–117. [8] C.D. Cao, D.M. Herlach, M. Kolbe, G.P. Gorler, B. Wei, Scripta Mater. 48 (2003) 5–9. [9] I. Yamauchi, N. Ueno, M. Shimaoka, I. Ohnaka, J. Mater. Sci. 33 (1998) 371–378. [10] M.B. Robinson, D. Li, T.J. Rathz, G. Williams, J. Mater. Sci. 34 (1999) 3747–3753. [11] X.Y. Lu, C.D. Cao, M. Kolbe, B. Wei, D.M. Herlach, Mater. Sci. Eng. A 375–377 (2004) 1101–1104. [12] M. Palumbo, S. Curiotto, L. Battezzati, CALPHAD-Computer coupling of phase diagrams and thermochemistry 30 (2006) 171–178. [13] N.H. Pryds, X. Huang, Metall. Mater. Trans. A 31 (2000) 3155–3166. [14] N.H. Pryds, A.S. Pedersen, Metall. Mater. Trans. A 33 (2002) 3755–3761. [15] C.P. Wang, X.J. Liu, I. Ohnuma, R. Kainuma, K. Ishida, Science 297 (2002) 990–993. [16] R.D. Lide, Handbook of Chemistry and Physics, CRC Press, 2004, pp. 12–219. [17] S. Curiotto, N. Pryds, E. Johnson, L. Battezzati, Metallurgical and Materials Transactions A 37 (2006) 2361–2368.