- Email: [email protected]
Dislocation density of pure copper processed by accumulative roll bonding and equal-channel angular pressing
Materials Characterization 104 (2015) 101–106
Contents lists available at ScienceDirect
Materials Characterization journal homepage: www.elsevier.com/locate/matchar
Dislocation density of pure copper processed by accumulative roll bonding and equal-channel angular pressing Yoji Miyajima a,⁎, Satoshi Okubo a, Hiroki Abe a,1, Hiroki Okumura a,2, Toshiyuki Fujii b, Susumu Onaka a, Masaharu Kato a a b
Department of Materials Science and Engineering, Graduate School of Science and Engineering, Tokyo Institute of Technology, 4259-J2-63, Nagatsuta-cho, Midori-ku, Yokohama 226-8502, Japan Department of Metallurgy and Ceramics Science, Graduate School of Engineering, Tokyo Institute of Technology, 2-12-1-S8-7, Ookayama, Meguro-ku, Tokyo 152-8552, Japan
a r t i c l e
i n f o
Article history: Received 5 February 2015 Received in revised form 7 April 2015 Accepted 10 April 2015 Available online 11 April 2015
a b s t r a c t The dislocation density of pure copper fabricated by two severe plastic deformation (SPD) processes, i.e., accumulative roll bonding and equal-channel angular pressing, was evaluated using scanning transmission electron microscopy/transmission electron microscopy observations. The dislocation density drastically increased from ~1013 m−2 to about 5 × 1014 m−2, and then saturated, for both SPD processes. © 2015 Elsevier Inc. All rights reserved.
Keywords: Ultrafine grain Dislocation density Dislocation structure Grain size STEM TEM Cu
1. Introduction Ultrafine grained (UFG) metals fabricated by severe plastic deformation (SPD) processes having the grain size d less than 1 μm have been widely investigated because of their high strength [1]. Such high strength can be understood qualitatively by both the high density of grain boundaries and high dislocation density ρ via grain refinement strengthening and dislocation strengthening, respectively. Although many papers report grain size d, the number of papers reporting the dislocation density ρ is limited [2–5]. This is because the experimental procedure for evaluating ρ is relatively complicated compared with that for evaluating d. Among several methods to evaluate ρ, such as, X-ray diffraction line profile analysis (XLPA) [6,7], differential scanning calorimetry (DSC) [6], electrical resistivity measurements [3], hardness measurements [8], and transmission electron microscopy/scanning electron transmission electron microscopy (TEM/STEM) [2,4], the TEM/STEM method is the only one by which dislocations are observed directly. The dislocation density obtained by the TEM/STEM method is often said to be smaller than the
⁎ Corresponding author. E-mail address: [email protected] (Y. Miyajima). 1 Current affiliation: Denso Corporation, Kariya 448-8661, Japan. 2 Current affiliation: Mitsubishi Materials Corporation, Tokyo 100-8117, Japan.
http://dx.doi.org/10.1016/j.matchar.2015.04.009 1044-5803/© 2015 Elsevier Inc. All rights reserved.
real density: some dislocations may disappear from the surface of the thin foil specimen due to the image force, or, some dislocations satisfy the invisible condition for the diffracted electron beam. Nevertheless, we believe that the advantage of the direct observation compensates for the possibility of underestimation of ρ. As far as the authors know, there is only one systematic TEM/STEM study on the change in ρ in SPD processed aluminium, i.e., 2N–Al (purity of 99%) [4], whereas, no systematic investigation has been reported for copper apart from a few reports of relatively-highly deformed copper by the SPD process [2,9,10]. Therefore, the aim of this study is to systematically evaluate ρ of SPD processed pure copper using TEM/STEM.
2. Materials and methods Pure Cu sheets and Cu rods having the purity of 99.99% (4N–Cu) annealed at 873 K for 2 h (7.2 ks) were subjected to an accumulative roll bonding (ARB) and an equal-channel angular pressing (ECAP) processes, respectively [1,11,12]. Both SPD processes were performed at room temperature. The initial grain size of 4N–Cu is about 35 μm. The ARB process consists of four steps; cutting a metal sheet into two pieces, cleaning with acetone and subsequent wire brushing on one side of each sheet, stacking the two pieces of sheet as the two brushed surfaces stack, and roll bonding with the rolling reduction of 50%. Fig. 1(a) shows the schematic diagram of ARB and its sample
102
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106
Fig. 1. Schematic illustration of (a) accumulative roll bonding process and (b) equal-channel angular pressing process. The sample coordinates were also displayed.
coordinates. Rolling, transverse and normal directions are denoted as RD, TD and ND, respectively. Here, TD is perpendicular to both ND and RD. In this study, the rolled sheet was immediately water quenched in order to avoid the effect of process heat during the roll bonding. The surface of the rolls used for the ARB process was lubricated by machine oil, and the ARB process was applied up to eight cycles. Hereafter, the ARB N cycle sample is denoted as ARB Nc, and N is the ARB cycle number. The detail of the ARB can be found elsewhere [2–4,11,12]. The initial sizes of the sheets were 1 mm, 90 mm, and 400 mm for the thickness, the width, and the length, respectively. The ECAP process was performed using a metal die having a channel path bent 90°. The cross sectional shape of the path is circular, and extrusion direction (ED), TD and ND are defined as shown in Fig. 1(b). The sizes of the cylindrical samples were 10 mm and 60 mm for the diameter and the length, respectively. The cylindrical samples were inserted into the path via route Bc, in other words, the cylindrical samples were rotated by 90° around the height direction at each pass. The channel of the metal die and the cylindrical samples was lubricated by molybdenum disulphide. The ECAP process was applied up to 12 passes. Hereafter, ECAP n passes sample is denoted as ECAP np. Both the ARB and the ECAP processed 4N–Cu were observed by STEM and TEM including high-voltage electron microscopy (HVEM). JOEL JEM-2100F (STEM mode) with the acceleration voltage of 200 kV with Gatan bright field (BF) detector was used for STEM, whereas, Hitachi H1250 with the acceleration voltage of 1000 kV and JOEL JEM2100F (TEM mode) with the acceleration voltage of 200 kV were used for TEM. All the TEM and STEM specimens were cut from both the ARB and the ECAP processed samples using an electrical wire discharge machine, Brother HS-300, as the observations were performed from TD. The surface of the specimens was mechanically polished using SiC paper until the thickness of the specimen becomes about 150 μm. The specimens were also electrolytically polished using Struers TenuPole 3 with the applied voltage at 7 V in the mixture of distilled water, phosphoric acid, and ethanol for 7:2:1 in volume at 273 K until the thickness becomes around 20 μm. Subsequently, the specimens were electrolytically polished with the applied voltage at 7 V in the mixture of nitric acid and methanol for 1:3 in volume at 223 K for perforation. Apart from annealed 4N–Cu and ECAP 1p, seven grains satisfying diffraction conditions for visible dislocations in the grains were observed by STEM in order to evaluate dislocation density ρ of the specimen with Ham's intersection method [13]. Whereas, ρ of the annealed 4N– Cu and ECAP 1p were measured from only one grain since the grain
size was too large to measure several grains. The procedure for Ham's intersection method is schematically shown in Fig. 2. First, a mesh is drawn on a TEM image of a grain satisfying the diffraction condition, and the total length L of the mesh is measured. Secondly, the number of intersections m between the mesh and dislocations are counted. Finally, Eq. (1) is used to evaluate ρ. ρ¼
2m Lt
ð1Þ
Here, the thickness of the specimen t was evaluated from thickness fringes which appear at a high-angle grain boundary (HAGB) under two-beam condition or systematic reflection [4]. It is known that the thickness fringes appear periodically with discrete depth, the so-called extinction distance ξg [14], and therefore, the specimen thickness can be evaluated once the number of thickness fringes at HAGB is counted. Table 1 shows some relationships between the reflection and ξg for Cu, which is copied from Table A.4.2 in reference [14]. It is pointed out that the values of ξg in the literature are for 100 kV electrons, but, the acceleration voltage of TEM used in this study is 200 kV. Thus, the values of
Fig. 2. Schematic illustration of Ham's method. The thick lines represent grain boundary, mesh consists of thin lines, small circles indicates the intersections, and bent lines are dislocations.
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106 Table 1 relationships between the reflection and extinction distance for Cu taken from reference [14]. Reflection Extinction distance [nm]
100 [kV] 200 [kV]
111
200
220
311
222
24.2 30.7
28.1 35.6
41.6 52.7
50.5 64.0
53.5 67.8
ξg for 200 kV electrons are also displayed in Table 1, which was evaluated by multiplication of the correlation factor (1.268) taken from Table A.4.1 in reference [14] and the values of ξg for 100 kV. 3. Results and discussions Fig. 3 shows HVEM photographs of ARB processed 4N–Cu. Grain thickness decreases and grains are more elongated along RD with increasing N. Fig. 4 shows TEM images of ECAP processed 4N–Cu. Rather equiaxed grains can be observed. However, some grains are elongated along the ECAP last shear plane which is 45° to both ND and ED. It
103
should be noted that the HVEM photograph of 4N–Cu ARB processed by 6c and 8c have already been displayed in the literature in order to discuss the mechanical properties [2]. The mean grain separation along ND was measured from Fig. 3 as a function of the ARB cycle number N using the line interception method and the results are shown in Fig. 5. The data points of ARB 6c and ARB 8c were taken from the literature. It can be seen that grain thickness of the ARB processed 4N–Cu decreases as N increases and almost saturates at about 130 nm after ARB 4c. Similarly, the mean grain separation was measured from Fig. 4 as a function of the ECAP pass number n, as shown in Fig. 6. The grain size of ECAP processed 4N–Cu reaches to be almost constant with the value of about 250 nm at ECAP 4p. Although the TEM observations of dislocations have been performed for decades [14], some basics are stated as follows. In both HVEM and TEM images, some grains seem to have dislocations as shown in Figs. 3 and 4. Such grains satisfy the appropriate diffraction conditions for visualising the dislocations. The other grains seem to have no dislocations, which only means that they do not satisfy the appropriate diffraction conditions for dislocations. In other words, it is impossible to judge from only one BF image whether grains have high dislocation
Fig. 3. High voltage transmission electron microscopy photograph of pure copper accumulative roll bonding processed (a) 1 cycle, (b) 2 cycle, (c) 3 cycle, (d) 4 cycle, (e) 6 cycle, and (f) 8 cycle. It should be noted that a part of the image of 6 cycle and 8 cycle have already been displayed in [2].
104
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106
Fig. 4. TEM images of pure copper equal-channel angular pressing processed (a) 4 pass, (b) 8 pass, and (c) 12 pass.
density or not. Furthermore, the thickness of the observed area is also important, since, the apparent density of dislocation (total length per unit area) can be different depending on the thickness of the specimen at the region even if the dislocation density is the same. Figs. 7 and 8 show the examples of STEM images used for the evaluation of ρ for the ARB and the ECAP processed 4N–Cu, respectively. It is possible to see the elongated grains with dislocations in Fig. 7 and the relatively equiaxed grains with dislocations in Fig. 8. Figs. 9 and 10 show the dislocation density ρ of the ARB and the ECAP processed 4N–Cu versus N and n, respectively. Some of the data points (ARB 6c, ARB 8c, ECAP 4p, and ECAP 8p) were taken from the literature [2,9]. As can be seen from these figures, ρ immediately increases to around 5 × 1014 m− 2 at ARB 1c and ECAP 1p from the value of ~ 1013 m−2 for annealed 4N–Cu (ARB 0c or ECAP 0p). Furthermore, it is noteworthy that the values of ρ for SPD processed 4N–Cu are almost independent of N and n. Moreover, the value 5 × 1014 m−2 is the same regardless of the SPD processes, either ARB or ECAP. Interestingly, the rapid increase of ρ at the first ARB cycle and the subsequent saturation of ρ with increasing N have also been observed using XRD and DSC [6].
The microstructural evolution of plastically deformed pure fcc metals have already been reported by Hansen, which is so-called “grain subdivision” [15]. First, dislocations are introduced into coarse grains at the initial stage of plastic deformation, and then, dislocations start forming dislocation structures such as dislocation cells at some points. With introducing more dislocations by further plastic deformation, dislocation cells become low-angle grain boundaries (LAGBs), and finally, HAGBs are formed resulting from the increase of misorientation angle due to the further plastic deformation. However, there has been no experimental evidence on the evolution of ρ during SPD using TEM/STEM observations. From the present study for 4N–Cu and the previous one for 2N–Al [4], the change in ρ during SPD can be summarised as follows. (1) ρ increases with increasing the effective strain and d changes resulting from the change of the grain shape according to the geometrical constraints of SPD. (2) ρ reaches the maximum value and dislocations start to form dislocation structure such as cell structure. (3) The formed cell walls become LAGBs and the average misorientation continuously increases with increasing the effective strain. Thus, d decreases with increasing the effective strain, whereas the value of ρ stays nearly
Fig. 5. Accumulative roll bonding cycle number dependence of mean grain separation along normal direction for pure copper. It should be noted that the data point of 6 cycle and 8 cycle are taken from [2].
Fig. 6. Equal-channel angular pressing pass number dependence of mean grain separation for pure copper.
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106
105
Fig. 7. Scanning transmission electron microscopy images of pure copper accumulative roll bonding processed by (a) 1 cycle, (b) 2 cycle, (c) 3 cycle, (d) 4 cycle, (e) 5 cycle, (f) 6 cycle, (g) 7 cycle, and (h) 8 cycle. It should be noted that images (f) and (h) are taken from [2].
Fig. 8. Scanning transmission electron microscopy images of pure copper equal channel angular pressing processed by (a) 1 pass, (b) 4 pass, (c) 8 pass, and (d) 12 pass. It should be noted that images of (b) and (c) are taken from [9].
106
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106
SPD process to reach a saturation value in ρ due to the low mobility of dislocations. 4. Conclusions Dislocation density of pure copper after severe plastic deformation was evaluated using transmission electron microscopy/scanning transmission electron microscopy observations. Both accumulative roll bonding and equal-channel angular pressing processes give the similar trend that dislocation density drastically increased from ~1013 m−2 to around 5 × 1014 m−2 immediately after the first process, and, the values of dislocation density remain unchanged after the first process. Such trend is thought to be typical of pure metals with medium to high stacking fault energy which are subjected to accumulative roll bonding and equal-channel angular pressing. Fig. 9. Accumulative roll bonding cycle number dependence of dislocation density of pure copper. It should be noted that values of 6 cycle and 8 cycle are used in the literature [2].
Acknowledgments This study was financially supported by a Grant-in-aid for Scientific Research on Innovative Area “Bulk Nanostructured Metals” No. 22102006 through the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. The HVEM observation was performed at the HVEM room of Tokyo Institute of Technology, which was supported by the Nanotechnology Support Project of MEXT, Japan. The authors are most grateful to Professor Eiichi Sato, Japan Aerospace Exploration Agency (JAXA), for the use of rolling mill in his laboratories. References
Fig. 10. Equal-channel angular pressing pass number dependence of dislocation density of pure copper. It should be noted that values of 6 pass and 8 pass are used in the literature [9].
constant. (4) Some LAGBs become HAGBs with increasing effective strain and the fraction of HAGBs increases, but, the value of ρ is again constant. It should be pointed out that d here is the grain size measured using all boundaries observed by TEM/STEM, and therefore, the boundaries contain dislocation walls, LAGB, and HAGB. For ARB processed 2N–Al, the saturation value for ρ of about 1 × 1014 m−2 has been reported [4]. In this case, however, the starting material was not fully recrystallized, and therefore, the initial value of ρ was already around 1 × 1014 m−2. Nevertheless, the saturation value of ρ under SPD seems to be affected by some physical parameters, such as melting point and stacking fault energy (SFE), in addition to the purity of the metals since the value is obviously the result of the balancing of introduced dislocations and recovery, i.e., about 5 × 1014 m− 2 for 4N–Cu and about 1 × 1014 m− 2 for 2N–Al. Although there is no systematic study on 4N–Al, it is expected that the saturation value of ρ is around 1 × 1013 m−2, smaller than that for 2N–Al, judging from the reported value for 4N–Al ARB processed 6c by Kamikawa [5]. The present study implies that the saturation value of dislocation density is independent of specific SPD processes even for other fcc materials with medium to high SFE such as Cu, Ni, and Al, and that the saturation is reached immediately after the first SPD process. However, fcc metals with low SFE may require more than one pass/cycles of an
[1] A. Azushima, R. Kopp, A. Korhonen, D.Y. Yang, F. Micari, G.D. Lahoti, P. Groche, J. Yanagimoto, N. Tsuji, A. Rosochowski, A. Yanagida, Severe plastic deformation (SPD) processes for metals, CIRP Ann. Manuf. Technol. 57 (2008) 716–735. [2] Y. Miyajima, H. Abe, T. Fujii, S. Onaka, M. Kato, Effects of Si on mechanical properties and microstructure evolution in ultrafine-grained Cu–Si alloys processed by accumulative roll bonding, Acta Mater. 61 (2013) 1537–1544. [3] Y. Miyajima, S.Y. Komatsu, M. Mitsuhara, S. Hata, H. Nakashima, N. Tsuji, Change in electrical resistivity of commercial purity aluminium severely plastic deformed, Philos. Mag. 90 (2010) 4475–4488. [4] Y. Miyajima, M. Mitsuhara, S. Hata, H. Nakashima, N. Tsuji, Quantification of internal dislocation density using scanning transmission electron microscopy in ultrafine grained pure aluminium fabricated by severe plastic deformation, Mater. Sci. Eng. A 528 (2010) 776–779. [5] N. Kamikawa, X. Huang, N. Tsuji, N. Hansen, Strengthening mechanisms in nanostructured high-purity aluminium deformed to high strain and annealed, Acta Mater. 57 (2009) 4198–4208. [6] Y. Miyajima, T. Ueda, H. Adachi, T. Fujii, S. Onaka, M. Kato, Dislocation density of FCC metals processed by ARB, 6th International Conference on Nanomaterials by Severe Plastic Deformation, IOP Publishing, Metz 2014, p. 012138. [7] J. Gubicza, G. Ribarik, G.R. Goren-Muginstein, A.R. Rosen, T. Ungar, The density and the character of dislocations in cubic and hexagonal polycrystals determined by Xray diffraction, Mater. Sci. Eng. A 309 (2001) 60–63. [8] S. Graca, R. Colaco, R.A. Carvalho, R. Vilar, Determination of dislocation density from hardness measurements in metals, Mater. Lett. 62 (2008) 3812–3814. [9] S. Okubo, H. Abe, Y. Miyajima, T. Fujii, S. Onaka, M. Kato, Effects of temperature and strain rate on plastic deformation of ultrafine-grained copper prepared by equalchannel angular pressing, Mater. Trans. 55 (2014) 1525–1530. [10] X. Huang, N. Hansen, N. Tsuji, Hardening by annealing and softening by deformation in nanostructured metals, Science 312 (2006) 249–251. [11] Y. Saito, H. Utsunomiya, N. Tsuji, T. Sakai, Novel ultra-high straining process for bulk materials — development of the accumulative roll-bonding (ARB) process, Acta Mater. 47 (1999) 579–583. [12] Y. Saito, N. Tsuji, H. Utsunomiya, T. Sakai, R.G. Hong, Ultra-fine grained bulk aluminum produced by accumulative roll-bonding (ARB) process, Scr. Mater. 39 (1998) 1221–1227. [13] R.K. Ham, Philos. Mag. 6 (1961) 1183–1184. [14] P. Hirsch, A. Howie, R.B. Nicholson, D.W. Pashley, M.J. Whelan, Electron Microscopy of Thin Crystals, Robert E. Krieger Publishing, Florida, 1977. [15] N. Hansen, New discoveries in deformed metals, Metall. Mater. Trans. A 32 (2001) 2917–2935.
Contents lists available at ScienceDirect
Materials Characterization journal homepage: www.elsevier.com/locate/matchar
Dislocation density of pure copper processed by accumulative roll bonding and equal-channel angular pressing Yoji Miyajima a,⁎, Satoshi Okubo a, Hiroki Abe a,1, Hiroki Okumura a,2, Toshiyuki Fujii b, Susumu Onaka a, Masaharu Kato a a b
Department of Materials Science and Engineering, Graduate School of Science and Engineering, Tokyo Institute of Technology, 4259-J2-63, Nagatsuta-cho, Midori-ku, Yokohama 226-8502, Japan Department of Metallurgy and Ceramics Science, Graduate School of Engineering, Tokyo Institute of Technology, 2-12-1-S8-7, Ookayama, Meguro-ku, Tokyo 152-8552, Japan
a r t i c l e
i n f o
Article history: Received 5 February 2015 Received in revised form 7 April 2015 Accepted 10 April 2015 Available online 11 April 2015
a b s t r a c t The dislocation density of pure copper fabricated by two severe plastic deformation (SPD) processes, i.e., accumulative roll bonding and equal-channel angular pressing, was evaluated using scanning transmission electron microscopy/transmission electron microscopy observations. The dislocation density drastically increased from ~1013 m−2 to about 5 × 1014 m−2, and then saturated, for both SPD processes. © 2015 Elsevier Inc. All rights reserved.
Keywords: Ultrafine grain Dislocation density Dislocation structure Grain size STEM TEM Cu
1. Introduction Ultrafine grained (UFG) metals fabricated by severe plastic deformation (SPD) processes having the grain size d less than 1 μm have been widely investigated because of their high strength [1]. Such high strength can be understood qualitatively by both the high density of grain boundaries and high dislocation density ρ via grain refinement strengthening and dislocation strengthening, respectively. Although many papers report grain size d, the number of papers reporting the dislocation density ρ is limited [2–5]. This is because the experimental procedure for evaluating ρ is relatively complicated compared with that for evaluating d. Among several methods to evaluate ρ, such as, X-ray diffraction line profile analysis (XLPA) [6,7], differential scanning calorimetry (DSC) [6], electrical resistivity measurements [3], hardness measurements [8], and transmission electron microscopy/scanning electron transmission electron microscopy (TEM/STEM) [2,4], the TEM/STEM method is the only one by which dislocations are observed directly. The dislocation density obtained by the TEM/STEM method is often said to be smaller than the
⁎ Corresponding author. E-mail address: [email protected] (Y. Miyajima). 1 Current affiliation: Denso Corporation, Kariya 448-8661, Japan. 2 Current affiliation: Mitsubishi Materials Corporation, Tokyo 100-8117, Japan.
http://dx.doi.org/10.1016/j.matchar.2015.04.009 1044-5803/© 2015 Elsevier Inc. All rights reserved.
real density: some dislocations may disappear from the surface of the thin foil specimen due to the image force, or, some dislocations satisfy the invisible condition for the diffracted electron beam. Nevertheless, we believe that the advantage of the direct observation compensates for the possibility of underestimation of ρ. As far as the authors know, there is only one systematic TEM/STEM study on the change in ρ in SPD processed aluminium, i.e., 2N–Al (purity of 99%) [4], whereas, no systematic investigation has been reported for copper apart from a few reports of relatively-highly deformed copper by the SPD process [2,9,10]. Therefore, the aim of this study is to systematically evaluate ρ of SPD processed pure copper using TEM/STEM.
2. Materials and methods Pure Cu sheets and Cu rods having the purity of 99.99% (4N–Cu) annealed at 873 K for 2 h (7.2 ks) were subjected to an accumulative roll bonding (ARB) and an equal-channel angular pressing (ECAP) processes, respectively [1,11,12]. Both SPD processes were performed at room temperature. The initial grain size of 4N–Cu is about 35 μm. The ARB process consists of four steps; cutting a metal sheet into two pieces, cleaning with acetone and subsequent wire brushing on one side of each sheet, stacking the two pieces of sheet as the two brushed surfaces stack, and roll bonding with the rolling reduction of 50%. Fig. 1(a) shows the schematic diagram of ARB and its sample
102
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106
Fig. 1. Schematic illustration of (a) accumulative roll bonding process and (b) equal-channel angular pressing process. The sample coordinates were also displayed.
coordinates. Rolling, transverse and normal directions are denoted as RD, TD and ND, respectively. Here, TD is perpendicular to both ND and RD. In this study, the rolled sheet was immediately water quenched in order to avoid the effect of process heat during the roll bonding. The surface of the rolls used for the ARB process was lubricated by machine oil, and the ARB process was applied up to eight cycles. Hereafter, the ARB N cycle sample is denoted as ARB Nc, and N is the ARB cycle number. The detail of the ARB can be found elsewhere [2–4,11,12]. The initial sizes of the sheets were 1 mm, 90 mm, and 400 mm for the thickness, the width, and the length, respectively. The ECAP process was performed using a metal die having a channel path bent 90°. The cross sectional shape of the path is circular, and extrusion direction (ED), TD and ND are defined as shown in Fig. 1(b). The sizes of the cylindrical samples were 10 mm and 60 mm for the diameter and the length, respectively. The cylindrical samples were inserted into the path via route Bc, in other words, the cylindrical samples were rotated by 90° around the height direction at each pass. The channel of the metal die and the cylindrical samples was lubricated by molybdenum disulphide. The ECAP process was applied up to 12 passes. Hereafter, ECAP n passes sample is denoted as ECAP np. Both the ARB and the ECAP processed 4N–Cu were observed by STEM and TEM including high-voltage electron microscopy (HVEM). JOEL JEM-2100F (STEM mode) with the acceleration voltage of 200 kV with Gatan bright field (BF) detector was used for STEM, whereas, Hitachi H1250 with the acceleration voltage of 1000 kV and JOEL JEM2100F (TEM mode) with the acceleration voltage of 200 kV were used for TEM. All the TEM and STEM specimens were cut from both the ARB and the ECAP processed samples using an electrical wire discharge machine, Brother HS-300, as the observations were performed from TD. The surface of the specimens was mechanically polished using SiC paper until the thickness of the specimen becomes about 150 μm. The specimens were also electrolytically polished using Struers TenuPole 3 with the applied voltage at 7 V in the mixture of distilled water, phosphoric acid, and ethanol for 7:2:1 in volume at 273 K until the thickness becomes around 20 μm. Subsequently, the specimens were electrolytically polished with the applied voltage at 7 V in the mixture of nitric acid and methanol for 1:3 in volume at 223 K for perforation. Apart from annealed 4N–Cu and ECAP 1p, seven grains satisfying diffraction conditions for visible dislocations in the grains were observed by STEM in order to evaluate dislocation density ρ of the specimen with Ham's intersection method [13]. Whereas, ρ of the annealed 4N– Cu and ECAP 1p were measured from only one grain since the grain
size was too large to measure several grains. The procedure for Ham's intersection method is schematically shown in Fig. 2. First, a mesh is drawn on a TEM image of a grain satisfying the diffraction condition, and the total length L of the mesh is measured. Secondly, the number of intersections m between the mesh and dislocations are counted. Finally, Eq. (1) is used to evaluate ρ. ρ¼
2m Lt
ð1Þ
Here, the thickness of the specimen t was evaluated from thickness fringes which appear at a high-angle grain boundary (HAGB) under two-beam condition or systematic reflection [4]. It is known that the thickness fringes appear periodically with discrete depth, the so-called extinction distance ξg [14], and therefore, the specimen thickness can be evaluated once the number of thickness fringes at HAGB is counted. Table 1 shows some relationships between the reflection and ξg for Cu, which is copied from Table A.4.2 in reference [14]. It is pointed out that the values of ξg in the literature are for 100 kV electrons, but, the acceleration voltage of TEM used in this study is 200 kV. Thus, the values of
Fig. 2. Schematic illustration of Ham's method. The thick lines represent grain boundary, mesh consists of thin lines, small circles indicates the intersections, and bent lines are dislocations.
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106 Table 1 relationships between the reflection and extinction distance for Cu taken from reference [14]. Reflection Extinction distance [nm]
100 [kV] 200 [kV]
111
200
220
311
222
24.2 30.7
28.1 35.6
41.6 52.7
50.5 64.0
53.5 67.8
ξg for 200 kV electrons are also displayed in Table 1, which was evaluated by multiplication of the correlation factor (1.268) taken from Table A.4.1 in reference [14] and the values of ξg for 100 kV. 3. Results and discussions Fig. 3 shows HVEM photographs of ARB processed 4N–Cu. Grain thickness decreases and grains are more elongated along RD with increasing N. Fig. 4 shows TEM images of ECAP processed 4N–Cu. Rather equiaxed grains can be observed. However, some grains are elongated along the ECAP last shear plane which is 45° to both ND and ED. It
103
should be noted that the HVEM photograph of 4N–Cu ARB processed by 6c and 8c have already been displayed in the literature in order to discuss the mechanical properties [2]. The mean grain separation along ND was measured from Fig. 3 as a function of the ARB cycle number N using the line interception method and the results are shown in Fig. 5. The data points of ARB 6c and ARB 8c were taken from the literature. It can be seen that grain thickness of the ARB processed 4N–Cu decreases as N increases and almost saturates at about 130 nm after ARB 4c. Similarly, the mean grain separation was measured from Fig. 4 as a function of the ECAP pass number n, as shown in Fig. 6. The grain size of ECAP processed 4N–Cu reaches to be almost constant with the value of about 250 nm at ECAP 4p. Although the TEM observations of dislocations have been performed for decades [14], some basics are stated as follows. In both HVEM and TEM images, some grains seem to have dislocations as shown in Figs. 3 and 4. Such grains satisfy the appropriate diffraction conditions for visualising the dislocations. The other grains seem to have no dislocations, which only means that they do not satisfy the appropriate diffraction conditions for dislocations. In other words, it is impossible to judge from only one BF image whether grains have high dislocation
Fig. 3. High voltage transmission electron microscopy photograph of pure copper accumulative roll bonding processed (a) 1 cycle, (b) 2 cycle, (c) 3 cycle, (d) 4 cycle, (e) 6 cycle, and (f) 8 cycle. It should be noted that a part of the image of 6 cycle and 8 cycle have already been displayed in [2].
104
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106
Fig. 4. TEM images of pure copper equal-channel angular pressing processed (a) 4 pass, (b) 8 pass, and (c) 12 pass.
density or not. Furthermore, the thickness of the observed area is also important, since, the apparent density of dislocation (total length per unit area) can be different depending on the thickness of the specimen at the region even if the dislocation density is the same. Figs. 7 and 8 show the examples of STEM images used for the evaluation of ρ for the ARB and the ECAP processed 4N–Cu, respectively. It is possible to see the elongated grains with dislocations in Fig. 7 and the relatively equiaxed grains with dislocations in Fig. 8. Figs. 9 and 10 show the dislocation density ρ of the ARB and the ECAP processed 4N–Cu versus N and n, respectively. Some of the data points (ARB 6c, ARB 8c, ECAP 4p, and ECAP 8p) were taken from the literature [2,9]. As can be seen from these figures, ρ immediately increases to around 5 × 1014 m− 2 at ARB 1c and ECAP 1p from the value of ~ 1013 m−2 for annealed 4N–Cu (ARB 0c or ECAP 0p). Furthermore, it is noteworthy that the values of ρ for SPD processed 4N–Cu are almost independent of N and n. Moreover, the value 5 × 1014 m−2 is the same regardless of the SPD processes, either ARB or ECAP. Interestingly, the rapid increase of ρ at the first ARB cycle and the subsequent saturation of ρ with increasing N have also been observed using XRD and DSC [6].
The microstructural evolution of plastically deformed pure fcc metals have already been reported by Hansen, which is so-called “grain subdivision” [15]. First, dislocations are introduced into coarse grains at the initial stage of plastic deformation, and then, dislocations start forming dislocation structures such as dislocation cells at some points. With introducing more dislocations by further plastic deformation, dislocation cells become low-angle grain boundaries (LAGBs), and finally, HAGBs are formed resulting from the increase of misorientation angle due to the further plastic deformation. However, there has been no experimental evidence on the evolution of ρ during SPD using TEM/STEM observations. From the present study for 4N–Cu and the previous one for 2N–Al [4], the change in ρ during SPD can be summarised as follows. (1) ρ increases with increasing the effective strain and d changes resulting from the change of the grain shape according to the geometrical constraints of SPD. (2) ρ reaches the maximum value and dislocations start to form dislocation structure such as cell structure. (3) The formed cell walls become LAGBs and the average misorientation continuously increases with increasing the effective strain. Thus, d decreases with increasing the effective strain, whereas the value of ρ stays nearly
Fig. 5. Accumulative roll bonding cycle number dependence of mean grain separation along normal direction for pure copper. It should be noted that the data point of 6 cycle and 8 cycle are taken from [2].
Fig. 6. Equal-channel angular pressing pass number dependence of mean grain separation for pure copper.
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106
105
Fig. 7. Scanning transmission electron microscopy images of pure copper accumulative roll bonding processed by (a) 1 cycle, (b) 2 cycle, (c) 3 cycle, (d) 4 cycle, (e) 5 cycle, (f) 6 cycle, (g) 7 cycle, and (h) 8 cycle. It should be noted that images (f) and (h) are taken from [2].
Fig. 8. Scanning transmission electron microscopy images of pure copper equal channel angular pressing processed by (a) 1 pass, (b) 4 pass, (c) 8 pass, and (d) 12 pass. It should be noted that images of (b) and (c) are taken from [9].
106
Y. Miyajima et al. / Materials Characterization 104 (2015) 101–106
SPD process to reach a saturation value in ρ due to the low mobility of dislocations. 4. Conclusions Dislocation density of pure copper after severe plastic deformation was evaluated using transmission electron microscopy/scanning transmission electron microscopy observations. Both accumulative roll bonding and equal-channel angular pressing processes give the similar trend that dislocation density drastically increased from ~1013 m−2 to around 5 × 1014 m−2 immediately after the first process, and, the values of dislocation density remain unchanged after the first process. Such trend is thought to be typical of pure metals with medium to high stacking fault energy which are subjected to accumulative roll bonding and equal-channel angular pressing. Fig. 9. Accumulative roll bonding cycle number dependence of dislocation density of pure copper. It should be noted that values of 6 cycle and 8 cycle are used in the literature [2].
Acknowledgments This study was financially supported by a Grant-in-aid for Scientific Research on Innovative Area “Bulk Nanostructured Metals” No. 22102006 through the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. The HVEM observation was performed at the HVEM room of Tokyo Institute of Technology, which was supported by the Nanotechnology Support Project of MEXT, Japan. The authors are most grateful to Professor Eiichi Sato, Japan Aerospace Exploration Agency (JAXA), for the use of rolling mill in his laboratories. References
Fig. 10. Equal-channel angular pressing pass number dependence of dislocation density of pure copper. It should be noted that values of 6 pass and 8 pass are used in the literature [9].
constant. (4) Some LAGBs become HAGBs with increasing effective strain and the fraction of HAGBs increases, but, the value of ρ is again constant. It should be pointed out that d here is the grain size measured using all boundaries observed by TEM/STEM, and therefore, the boundaries contain dislocation walls, LAGB, and HAGB. For ARB processed 2N–Al, the saturation value for ρ of about 1 × 1014 m−2 has been reported [4]. In this case, however, the starting material was not fully recrystallized, and therefore, the initial value of ρ was already around 1 × 1014 m−2. Nevertheless, the saturation value of ρ under SPD seems to be affected by some physical parameters, such as melting point and stacking fault energy (SFE), in addition to the purity of the metals since the value is obviously the result of the balancing of introduced dislocations and recovery, i.e., about 5 × 1014 m− 2 for 4N–Cu and about 1 × 1014 m− 2 for 2N–Al. Although there is no systematic study on 4N–Al, it is expected that the saturation value of ρ is around 1 × 1013 m−2, smaller than that for 2N–Al, judging from the reported value for 4N–Al ARB processed 6c by Kamikawa [5]. The present study implies that the saturation value of dislocation density is independent of specific SPD processes even for other fcc materials with medium to high SFE such as Cu, Ni, and Al, and that the saturation is reached immediately after the first SPD process. However, fcc metals with low SFE may require more than one pass/cycles of an
[1] A. Azushima, R. Kopp, A. Korhonen, D.Y. Yang, F. Micari, G.D. Lahoti, P. Groche, J. Yanagimoto, N. Tsuji, A. Rosochowski, A. Yanagida, Severe plastic deformation (SPD) processes for metals, CIRP Ann. Manuf. Technol. 57 (2008) 716–735. [2] Y. Miyajima, H. Abe, T. Fujii, S. Onaka, M. Kato, Effects of Si on mechanical properties and microstructure evolution in ultrafine-grained Cu–Si alloys processed by accumulative roll bonding, Acta Mater. 61 (2013) 1537–1544. [3] Y. Miyajima, S.Y. Komatsu, M. Mitsuhara, S. Hata, H. Nakashima, N. Tsuji, Change in electrical resistivity of commercial purity aluminium severely plastic deformed, Philos. Mag. 90 (2010) 4475–4488. [4] Y. Miyajima, M. Mitsuhara, S. Hata, H. Nakashima, N. Tsuji, Quantification of internal dislocation density using scanning transmission electron microscopy in ultrafine grained pure aluminium fabricated by severe plastic deformation, Mater. Sci. Eng. A 528 (2010) 776–779. [5] N. Kamikawa, X. Huang, N. Tsuji, N. Hansen, Strengthening mechanisms in nanostructured high-purity aluminium deformed to high strain and annealed, Acta Mater. 57 (2009) 4198–4208. [6] Y. Miyajima, T. Ueda, H. Adachi, T. Fujii, S. Onaka, M. Kato, Dislocation density of FCC metals processed by ARB, 6th International Conference on Nanomaterials by Severe Plastic Deformation, IOP Publishing, Metz 2014, p. 012138. [7] J. Gubicza, G. Ribarik, G.R. Goren-Muginstein, A.R. Rosen, T. Ungar, The density and the character of dislocations in cubic and hexagonal polycrystals determined by Xray diffraction, Mater. Sci. Eng. A 309 (2001) 60–63. [8] S. Graca, R. Colaco, R.A. Carvalho, R. Vilar, Determination of dislocation density from hardness measurements in metals, Mater. Lett. 62 (2008) 3812–3814. [9] S. Okubo, H. Abe, Y. Miyajima, T. Fujii, S. Onaka, M. Kato, Effects of temperature and strain rate on plastic deformation of ultrafine-grained copper prepared by equalchannel angular pressing, Mater. Trans. 55 (2014) 1525–1530. [10] X. Huang, N. Hansen, N. Tsuji, Hardening by annealing and softening by deformation in nanostructured metals, Science 312 (2006) 249–251. [11] Y. Saito, H. Utsunomiya, N. Tsuji, T. Sakai, Novel ultra-high straining process for bulk materials — development of the accumulative roll-bonding (ARB) process, Acta Mater. 47 (1999) 579–583. [12] Y. Saito, N. Tsuji, H. Utsunomiya, T. Sakai, R.G. Hong, Ultra-fine grained bulk aluminum produced by accumulative roll-bonding (ARB) process, Scr. Mater. 39 (1998) 1221–1227. [13] R.K. Ham, Philos. Mag. 6 (1961) 1183–1184. [14] P. Hirsch, A. Howie, R.B. Nicholson, D.W. Pashley, M.J. Whelan, Electron Microscopy of Thin Crystals, Robert E. Krieger Publishing, Florida, 1977. [15] N. Hansen, New discoveries in deformed metals, Metall. Mater. Trans. A 32 (2001) 2917–2935.