Cu Segregation around Metastable Phase in Al-Mg-Si Alloy with Cu

Transcription

1 Materials Transactions, Vol. 48, No. 5 (2007) pp. 967 to 974 Special Issue on New Developments and Analysis for Fabrication of Functional Nanostructures #2007 The Japan Institute of Metals Cu Segregation around Metastable Phase in Al-Mg-Si Alloy with Cu Kenji Matsuda 1; *, Daisuke Teguri 2, Tatsuo Sato 3, Yasuhiro Uetani 4 and Susumu Ikeno 1 1 Graduate School of Science and Engineering for Research, University of Toyama, Toyama , Japan 2 Graduate School, Toyama University, Toyama , Japan 3 Graduate School of Science and Engineering, Tokyo Institute of Technology, Tokyo , Japan 4 Toyama Prefectural University, Toyama , Japan Cu distribution around Q 0 -phase in an Al-1.0 mass% Mg 2 Si-0.5 mass% Cu alloy has been investigated by the analytical transmission electron microscope (TEM), energy-filtering TEM (EFTEM), and high angular annular dark field scanning TEM (HAADF-STEM) techniques in order to determine the effect of Cu on the precipitation of this alloy. Cu-segregation around rod-shaped precipitates in the sample aged at 523 K for 24 ks was confirmed by energy dispersive X-ray spectroscopy, elemental maps, and HAADF-STEM images. Cu-segregation was also detected from small precipitates, which show a hexagonal network with a lattice constant of 1.04 nm in their HRTEM images. Coarse rods show homogeneous distribution of Cu, and chemical composition is similar to that of coarse Q 0 - or Q-phase. Small precipitates less than 10 nm were also confirmed in as-quenched samples, and those precipitates seem to form during quenching. [doi: /matertrans ] (Received November 1, 2006; Accepted December 28, 2006; Published April 25, 2007) Keywords: aluminum-magnesium-silicon alloy, copper addition, metastable phase, energy-filtering transmission electron microscopy 1. Introduction Al-Mg-Si alloys have attracted attention as materials for vehicles, by virtue of their light weight leading to fuel efficiency, and many recent studies have focused on improving their mechanical properties; specifically, strength and elongation. The precipitation sequence of this alloy is classically known to be as follows: 1) ssss! GP zones (rich in Mg and Si atoms) on f100g Al planes! ordered zones of 0 phase! equilibrium phase The addition of Cu has been known to improve ductility of an alloy, and GP zones and metastable phase make very important contributions to strength of this alloy system. 2) According to the quaternary alloy phase diagram,, Q, and phases exist in the range of 6000 series alloy. 3) Several recent studies report on Q 0 and Q phase in Al-Mg-Si alloys containing Cu. 2,4 8) The Q 0 phase is the metastable phase of the Q-phase, and has a hexagonal lattice and lattice constants of about a ¼ 1:04 nm and c ¼ 0:405 nm. Perovic et al. 4) proposed precipitating phases at peak strength conditions aged at 453 K in AA6111 and AA6016 Al alloys, where 00 phase exists in AA6016 alloy, whereas the two phases 00 and Q coexist in AA6111 alloy. Caylon and Buffat 5) discussed transformation from 0 -phase to B 0 -phase in Al-Mg-Si alloy and compared this transformation with that from QP, which is the precursor of the Q-phase, to Q-phase in Al-Cu-Mg-Si alloy. Miao and Laughlin 6) also concluded that in 6022 Al alloys, the precipitation of Q-phase is dependent on Cu concentration. In a 6022 alloy, which contains a small amount of Cu, the precipitation sequence consists of (sss)! GP zones! needlelike 00! rodlike 0 + lathshaped Q 0! + Si. In contrast, as Cu content increases, the precipitation sequence becomes (sss)! GP zones! needlelike 00! lath-shaped Q 0! Q + Si. In our study by transmission electron microscope (TEM), several kinds of metastable phases have been found from 6000 series alloys, and the crystal lattice of the TYPE-C precipitate is quite similar to that of the Q 0 -phase. 7 9) Moreover, Q 0 phase existing in the aged sample is surrounded by Cu segregation, and Cu concentration inside of Q 0 is lower than that at the interface between Q 0 and the matrix. 8) This means that Q 0 is not the only metastable precipitate in this alloy. In this study, Cu distribution around Q 0 -phase has been investigated by the analytical TEM, energy-filtering TEM (EFTEM), and high angular annular dark field scanning TEM (HAADF-STEM) techniques in order to determine the effect of Cu on the precipitation of this alloy. 2. Experimental An Al-1.0 mass% Mg 2 Si-0.5 mass% Cu alloy (0.5Cu alloy) was used for the present study. An Al-1.0 mass% Mg 2 Si-0.4 mass% Si alloy (excess Si alloy) was also prepared. The obtained ingots were homogenized at 723 K for 4 days, and were hot- and cold-rolled to sheets of 0.2 mm thickness. Sheets were solution heat-treated at 848 K for 3.6 ks and quenched in chilled water at 277 K. An aging treatment was performed at 523 K. Samples to be analyzed by transmission electron microscopy (TEM) were prepared by electro-polishing. The TEM (EM-002B, Topcon Co. Ltd., Tokyo, Japan) and EFTEM (JEOL-4010T, JEOL Co. Ltd., Tokyo Japan) were operated at 200 kv and 400 kv, respectively. Elemental maps by EFTEM were obtained by the 3 windows method. Generally, higher loss electron energies were used in this study, in view that an elemental map imaged using lower loss electron energies close to 100 ev is known to contain information of strain contrast. Pre-edge 1, pre-edge 2, and post edge of Cu-L, Mg-K and Si-K used are shown in Table 1. A field emission-type STEM (FE-STEM, JEOL- 2010F, JEOL Co. Ltd., Tokyo Japan) was also used to obtain an HAADF image. *Corresponding author, matsuda@eng.u-toyama.ac.jp

2 968 K. Matsuda, D. Teguri, T. Sato, Y. Uetani and S. Ikeno Fig. 1 Comparison of chemical composition between an elemental map and EDS for the same precipitate in Fig. 1(b). (a) Cu-L map, (b) estimated chemical compositions from elemental maps, (c) a zero-loss image, and (d) EDS data obtained for a precipitate of (c). Table 1 3. Results Parameters used for elemental maps (ev). Energy offset Slit width Pre-edge 1 Pre-edge 2 Post-edge Cu-L Mg-K Si-K A Cu map of Fig. 1(a) was obtained by EFTEM from the typical Q 0 -phase (Fig. 1(c)) and corresponded to the interface between the precipitate and the matrix. Mg and Si maps have also been obtained from the same precipitate and showed uniform distribution rather than segregation, although their images were omitted. The following equation has been applied to quantitative analysis of the elemental map in the present study: 10) N A ¼ I A 1 ð1þ I T A where N A is the number of atoms of element A per unit area, I A is intensity in elemental map of the element A per unit time, I T is intensity of a zero-loss image per unit time, and A is a partial cross-section for ionization of element A depending on a scattering angle, width of energy slit (E), and energy of primary electron (E O ). As A is given by EL/P software of our energy-filter, N A is calculated. To obtain the chemical composition for a unit volume, the number of atoms A per unit volume, n A was simply assumed as follows: n A ¼ N A =t ¼ I A 1 ð2þ I T t A where t is sample thickness. Sample thickness was also obtained from zero-loss peak, and its accuracy was about 5% as obtained from comparison with the CBD method. From this calculation, the precipitate includes Cu at a concentration less than 2 atom/nm 3, and the difference in Cu content between the matrix and the interface was 3.5 atom/nm 3,as shown in Fig. 1(b). When the fcc lattice of Al is assumed, Al atoms exist at 60.2 atom/nm 3. Thus, Cu concentration is about 5.8 at.% at the interface and 3.3 at.% inside the precipitate. This Cu-segregation was also confirmed from detailed EDS analysis using a 2 nm electron probe. Cu content is higher at points 2 and 4 in Fig. 1(d) than at points 1, 3 and 5. We also note that the ratio of Mg to Si is almost 1.0. This means that chemical composition inside this phase is actually similar to that of the type-c precipitate. An HAADF image was also obtained from this precipitate. The intensity of an HAADF image is proportional to the

3 Cu Segregation around Metastable Phase in Al-Mg-Si Alloy with Cu 969 Fig. 2 Comparison between EDS data and a HAADF-STEM image for the same precipitate. (a) a bright field STEM image, (b) EDS data obtained for a precipitate of (a), (c) a HAADF-STEM image of (a), and (d) an intensity profile of a rectangular region of (c). square of atomic number; therefore, brighter regions contain heavier atoms. 11) As Cu is the heaviest atom among Mg, Si, Al, and Cu, a brighter region in an HAADF image obtained for the precipitate contains a dominant amount of Cu. Figure 2(a) shows the result of a line analysis by EDS using a 0.2 nm electron probe. The intensity profile of an HAADF image shown in Fig. 2(d) includes 7 peaks; this corresponds to layers of precipitate marked 1 7 in Fig. 2(c). Layers 1 and 7 are actually at interfaces between this precipitate and the matrix, and peaks 1 and 7 are higher than the remaining peaks. This means that the corresponding regions contain the heaviest element. This result is in good agreement with the EFTEM result of Fig. 1 and supports its accuracy. Segregation of Cu has been also investigated from another approach. We expect Cu-maps parallel to h100i directions of the matrix to be obtained, because this precipitate is rodshaped and is parallel to h100i directions of the matrix and covered by Cu-segregation as illustrated in Fig. 3. In Fig. 4(b), which was obtained from the same area as Fig. 4(a), bright contrasts correspond to the rod-shaped precipitates that are parallel to the [100] and directions of the matrix in Fig. 4(a) marked by the arrows. Figures 4(c) 10 [1210] p [001] m [0001] p m Cu-segregation Type-C Fig. 3 A schematic illustration of Cu-distribution that surrounds the rodshaped precipitate.

4 970 K. Matsuda, D. Teguri, T. Sato, Y. Uetani and S. Ikeno (a) Zero-loss (b) Cu-map [001] (c) Zero-loss 30nm (d) Cu-map [001] [310] 30nm Fig. 4 EFTEM images at low magnification. (a) a zero-loss image parallel to [100] direction of the matrix, (b) Cu-L map corresponding to (a), (c) a zero-loss image parallel to ½130Š direction of the matrix, and (d) Cu-L map corresponding to (c). and (d) show the results obtained from the ½130Š direction of the matrix. As with the result of Fig. 4(a) and (b), bright contrasts in this Cu map also correspond to the rod-shaped precipitates. Figure 5(a) shows a zero-loss image in the sample overaged at 573 K for 60 ks. Coarse precipitates are observed in it. Fig. 5(b) is an enlarged image of the precipitate marked by the white arrow in (a), and its HRTEM image of Fig. 5(c) also shows a hexagonal network with a lattice constant of 1.04 nm. The intensity of the Cu map shows quite homogeneous distribution, as with Mg and Si. The number of atoms was obtained from the region marked by a white rectangular frame in Fig. 5(d) and is shown in Fig. 6(a). As shown in Fig. 6(b), EDS analysis was also obtained from 7 points on this coarse precipitate. No Cu-segregation was detected in Fig. 6(c). If Cu atoms are released during growth from the precipitate including higher Cu content, both Q 0 -phase, which contains Cu homogeneously, and type-c precipitate, with Cu-segregation, would be observed in aged samples. However, no Q 0 -phase is observed, which shows homogeneous Cu distribution in its Cu-map as in Fig. 5(d), especially in the early aging period. Figure 7 shows the Cu map obtained from a small precipitate in a sample aged at 523 K for 3.6 ks. This precipitate is small, but already shows a hexagonal network of 1.04 nm. Quantitative analysis of Cu shows that Cusegregation has already arisen. Some precipitates, which do not show a hexagonal network, did not show clear Cusegregation, although their images are omitted in this study. They are similar to the transition phases reported as randomtype precipitates. 12) 4. Discussion Electron diffraction patterns obtained from the Type-C and Q 0 -phase were compared in order to determine the difference in their crystal lattices. Table 2 shows the result of analysis of their diffraction patterns. The crystal lattice of the Q 0 -phase is

5 Cu Segregation around Metastable Phase in Al-Mg-Si Alloy with Cu 971 (a) (b) (c) 10 40nm (d) Cu-map 10nm 1.04nm 0.203nm 60 (e) Mg-map [100] (f) Si-map Fig. 5 EFTEM images of a large precipitate in the sample over-aged at 573 K for 60 ks. (a) a zero-loss image in the sample over-aged at 573 K for 60 ks. (b) an enlarged image of the precipitate marked by a white arrow in (a), (c) its HRTEM image. (d) Cu-L, (e) Mg-K, and (f) Si-K maps obtained for (c).

6 972 K. Matsuda, D. Teguri, T. Sato, Y. Uetani and S. Ikeno Fig. 6 (a) chemical compositions estimated from elemental maps in Figs. 5(d) (f), and (c) EDS analysis obtained from 7 points marked in (b). larger than that of the Type-C precipitate. From our previous work, the ratio of Mg to Si of the Type-C precipitate is about 1.0. The atomic ratio of elements in the Q 0 -phase was reported, and they are Mg : Si : Cu ¼ 8:7:2 13) or 8:6: 2. 14) The ratios Mg to (Si þ Cu) calculated from these are also about 1.0. When 2 Si atoms in the Type-C precipitate are replaced with 2 Cu atoms, the crystal lattice will be changed, because of the difference in atomic radius between Cu and Si. The changing lattice parameter is calculated by only changing of the volume of each lattice by the replacement of atoms. These results show close agreement with the result from electron diffraction patterns. Four possibilities for precipitation of Q 0 or Q-phase exist; that is, (1) transformation from the type-c precipitate, (2) transformation from the other transition phase, (3) nucleation on transition phases, and (4) direct nucleation in the matrix as a small Q 0 or Q-phase. Case (1) has been shown in the present work. The transition phase in case (2) means the random type, parallelogram-type precipitates, and the 0 -phase we reported. 15) These transition phases also exist in this sample, and include Cu, although their images are omitted. In our previous report, a complicated morphology of the cross-section of precipitate was found in this alloy, and both hexagonal networks of 1.04 nm for Q 0 -phase and 0.71 nm for 0 -phase in its HRTEM image have been recognized. 7) Q 0 - or Q-phase has been suggested to transform from the 0 -phase or nucleate on the interface between the 0 -phase and the matrix, and this corresponds to cases (2) and (3). The 4th possibility has been proposed by the TEM study shown in Fig. 8 obtained from the asquenched sample. There are small precipitates of less than 10 nm; their number per unit area was about 750 mm 2, and their cross sections are elongated. They were formed during quenching and their shape is quite different from the G.P. zone in the balanced alloy we proposed previously. 12) This is also different from the GP zone in Al-Cu alloy and the GPB zone in Al-Cu-Mg alloy. 16,17) These probably act as heterogeneous nucleation sites for the Q 0 phase or type-c precipitate. Their crystallography will be discussed elsewhere. A report has been published about quenching rate sensitivity for age-hardening of 6000 series Al alloys with Cu. 18) This precipitate may cause this behavior during quenching, because the added Cu for an Al-1.0 mass% Mg 2 Si alloy can be obtained as mass% at 848 K,

7 Cu Segregation around Metastable Phase in Al-Mg-Si Alloy with Cu 973 Fig. 7 (a) a zero-loss image obtained from a small precipitate in a sample aged at 523 K for 3.6 ks, (b) an enlarged image of (a), (c) its Cu- L map, and (d) the intensity profile obtained from a rectangular region in (c). Table 2 Comparison of the differences in lattice parameters between the Type-C and the Q 0 -phase. Type-C Q 0 -phase difference From actual SADPs a-axis nm nm nm c-axis Not detected Not detected From calculation a-axis nm nm nm c-axis nm nm nm which is the same temperature of the present solution heat treatment. 19) According to Wolverton s result by a first-principles total energy method, the fcc-based formation enthalpies with special quasi-random structures, H fcc SQS, for Cu-Mg and Mg- Si have been calculated as 2:6 and 16:4. 20) As the value of Cu-Mg is smaller than that of Mg-Si, formation of Cu-Mg clusters as shown in Fig. 8 is probably easy during quenching and/or at the early stage during aging. The Mg-Si or Mg-Si (-Cu) clusters can be formed at the same time; of course, Mg atoms are also consumed for clustering. Consequently, Si atoms can remain in the matrix at much higher concentration than in the matrix of the balanced alloy without Cu. Thus, the matrix includes Si in excess, and type- C precipitate can be formed in the matrix. The type-c precipitate has a tendency of heterogeneous nucleation that it can be formed on the dislocation or some interfaces between the precipitate and the matrix; 9) finally, the type-c precipitate nucleates on the Cu-Mg clusters as in Fig. 8. As is well known, dislocation loops can generally be observed in asquenched Al-Cu-Mg alloys. 21) However, no dislocation loops are observed in the as-quenched sample of this alloy. In addition, S 0 - or S-phase was not detected in this sample. Cu- Mg clusters probably cannot survive in this alloy during longterm aging, because of their stability. After dissolving, extra Cu atoms are attracted to the interface between the type-c and the matrix, because of their lattice misfits and thermal stability of the type-c precipitate. The Q 0 -phase probably nucleates heterogeneously on the Cu-Mg clusters as shown in Fig. 8, as the Q 0 -phase shows a similar tendency to the type-c precipitate according to the report of Yassar et al. 22) Both the type-c precipitate covered by Cu and the Q 0 -phase nucleated directly can grow to larger Q-phase after over aging. 5. Conclusions Distribution of copper around rod-shaped precipitates has been investigated by the analytical TEM, EFTEM, and

8 974 K. Matsuda, D. Teguri, T. Sato, Y. Uetani and S. Ikeno (a) [100] Acknowledgments The authors thank Dr. E. Okunishi of JEOL Ltd., for his support of HAADF-STEM work. This study was partially supported by the Research Project of the Venture Business Laboratory in University of Toyama, Japan. REFERENCES (b) 0.405nm [100] (c) 0.405nm 50nm [100] Fig. 8 (a) a bright field TEM image of the as-quenched sample. HRTEM images of 2 different dark dots in (a). (b) thin and long, and (c) thick and short ones. HAADF-STEM techniques in order to determine the effect of Cu on the precipitation in an Al-1.0 mass% Mg 2 Si-0.5 mass% Cu alloy. Cu-segregation around rod-shaped precipitates in the sample aged at 523 K for 24 ks was confirmed by EDS, elemental maps, and HAADF images. According to EFTEM observation along different directions of the matrix, an interface between precipitate and the matrix was covered by Cu. Cu-segregation was also detected from small precipitates, which show a hexagonal network with a lattice constant of 1.04 nm in their HRTEM images. Coarse rods show homogeneous distribution of Cu, and chemical composition is similar to that of Q 0 - or Q-phase. Small precipitates less than 10 nm were also confirmed in asquenched samples, and those precipitates seem to form during quenching. 1) H. K. Hardy and T. J. Heal: Report on Precipitation, Vol. 5, chapter 4, Progress in Metal Physics, (Pergamon press, London, 1954) pp ) D. J. Cahkrabarti and D. E. Laughlin: Progress Mater. Sci. 49 (2002) ) D. L. W. Collins: J. Inst. Metals 86 ( ) ) A. Perovic, D. D. Perovic, G. C. Weatherly and D. J. Lloyd: Scripta Mater. 41 (1999) ) C. Cayron and P. A. Buffat: Acta Mater. 48 (2000) ) W. F. Miao and D. E. Laughlin: Met. Mater. Trans. A 31A (2000) ) K. Matsuda, Y. Uetani, T. Sato and S. Ikeno: Met. Mater. Trans. A 32A (2001) ) K. Matsuda, D. Teguri, Y. Uetani, T. Sato and S. Ikeno: Scripta Mater. 47 (2002) ) K. Matsuda, Y. Sakaguchi, Y. Miyata, Y. Uetani, T. Sato, A. Kamio and S. Ikeno: J. Mater. Sci. 35 (2001) ) R. F. Egerton: Electron Energy-Loss Spectroscopy in the Electron Microscope, 2nd ed., Chapter 4, (Plenum press, New York, 1996) pp ) S. J. Pennycook and D. E. Jesson: Ultramicroscopy 37 (1991) ) K. Matsuda, H. Gamada, K. Fujii, Y. Uetani, T. Sato, A. Kamio and S. Ikeno: Met. Mater. Trans. A 29A (1998) ) L. Arnberg and B. Aurivillius: Acta Chem. Scand. A 34 (1980) ) C. Cayron, L. Sagalowicz, O. Beffort and P. A. Buffet: Philo. Mag. A 79 (1999) ) K. Matsuda, H. Gamada, K. Fujii, T. Yoshida, Y. Uetani, T. Sato, A. Kamio and S. Ikeno: J. Japan Inst. Metals 62 (1998) ) T. Sato and T. Takahashi: Trans. JIM 24 (1983) ) V. Radmilovic, G. Thomas, G. J. Shiflet and E. A. Starke, Jr.: Scripta Met. 23 (1989) ) I. R. Harris and P. C. Varley: J. Inst. Metals 82 ( ) ) H. Abe, S. Komatsu, K. Ikeda and T. Sakurai: J. Japan Inst. Light Metals 52 (2002) ) C. Wolverton: Acta Mater. 49 (2001) ) A. K. Mukhopadhyay, V. Singh, K. S. Prasad and C. R. Chakravorty: Acta Mater. 44 (1996) ) R. S. Yassar, D. P. Field and H. Weiland: Scripta Mater. 53 (2005)