- Email: [email protected]
Annealing behavior of cryogenically-rolled copper
Materials Science & Engineering A 585 (2013) 178–189
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Annealing behavior of cryogenically-rolled copper T. Konkova a, S. Mironov b,n, A. Korznikov a, M.M. Myshlyaev c, S.L. Semiatin d a
Institute for Metals Superplasticity Problems, Russian Academy of Science, 39 Khalturin Str., Ufa 450001, Russia Department of Materials Processing, Graduate School of Engineering, Tohoku University, 6-6-02 Aramaki-aza-Aoba, Sendai 980-8579, Japan Baikov Institute of Metallurgy and Material Science, Russian Academy of Science, 49 Lenin-av., Moscow 119991, Russia d Air Force Research Laboratory, Materials and Manufacturing Directorate, AFRL/RXCM, Wright-Patterson AFB, OH 45433-7817, USA b c
art ic l e i nf o
a b s t r a c t
Article history: Received 8 May 2013 Received in revised form 30 June 2013 Accepted 4 July 2013 Available online 31 July 2013
The static annealing behavior of cryogenically-rolled copper over a wide temperature range (50–950 1C) was established. At temperatures below 350 1C ( 0.5Tm), microstructure and texture evolution were interpreted in terms of discontinuous recrystallization. Grains having orientations close to (55;30/60;0), {236}〈385〉 (Brass-R), and {4;4;11}〈11;11;8〉 (Dillamore) were shown to recover rapidly and thus exhibited preferential growth during subsequent static recrystallization. At temperatures of 350 1C and higher, annealing behavior was dominated by abnormal grain growth. The abnormal character of this process was attributed to the relatively large spread in grain sizes produced during preceding recrystallization. & 2013 Elsevier B.V. All rights reserved.
Keywords: EBSD Nanostructured materials Thermomechanical processing Recrystallization Grain growth
1. Introduction There is significant commercial interest in the development of materials with ultrafine grain structures for structural applications. This interest is mainly driven by a substantial improvement in strength and ductility as well as a good balance of these properties. In addition, the control of the mechanical properties by processing may be an attractive alternative to expensive alloying. This could result in the use of fewer and simpler industrial alloys and would lead to economic benefits as well as improved recyclability. Techniques for the production of fine-grain alloys are thus of considerable commercial interest. Of particular importance are cost-effective methods that can be used to obtain large quantities of such materials. In this regard, an approach involving large deformation at cryogenic temperatures has recently attracted significant attention. It is believed that low temperatures may suppress dynamic recovery and stimulate mechanical twinning (e.g., [1, 2]) thereby enhancing the grain-refinement effect. This may decrease the level of strain necessary to achieve an ultra-fine microstructure and thus enable the application of conventional working processes such as rolling to produce such materials.
n
Corresponding author. Tel.: +81 22 795 7353; fax: +81 22 795 7352. E-mail addresses: [email protected] (T. Konkova), smironov@material. tohoku.ac.jp, [email protected] (S. Mironov), [email protected] (A. Korznikov), [email protected] (M.M. Myshlyaev), [email protected] (S.L. Semiatin). 0921-5093/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.msea.2013.07.042
To date, the majority of research in the field of cryogenic working has focused on aluminum and copper alloys [e.g., 1–7], most likely because of the superior ductility of these materials. It has been established that the key mechanism governing grainstructure evolution in both materials at cryogenic temperatures is the geometrical effect of strain per se. In other words, grains change their shape in proportion to the imposed strain, and noticeable grain subdivision and mechanical twinning are not observed [3,6]. By this means, a reasonably homogeneous grain structure, dominated by heavily elongated grains aligned with the direction of macroscopic material flow, is developed. Such grain structures typically contain a significant proportion of low-angle boundaries [3,6] and, in the case of copper, a high density of free dislocations [6]. The limited formation of deformation-induced boundaries during cryogenic deformation is believed to be partially associated with suppression of cross-slip at low temperatures [6]. This effect is also responsible for the strengthening of the {110}〈112〉 Brass texture in cryo-rolled materials [3,6]. Despite these characteristics, cryogenic deformation has been shown to be able to produce ultra-fine grain structures in both materials in some cases [3,6]. Not surprisingly, one important requirement to obtain and maintain a fine-grain microstructure is a high degree of microstructural stability within the expected range of use temperature and time. Hence, the annealing behavior of such materials is of interest. For cryo-deformed aluminum, this issue has been studied systematically by Zahid et al. [8]. For example, it was found that the fraction of low-angle boundaries increased progressively with
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
179
annealing temperature, leading eventually to discontinuous recrystallization. At intermediate temperatures, such behavior resulted in a bimodal structure comprising bands of coarse grains and fine subgrains. This unusual behavior has also been related to a strong texture in deformed material giving rise to the so-called “orientation pinning” effect [9] during subsequent grain growth. In contrast to aluminum, relatively little attention has been paid to the annealing behavior of cryo-deformed copper. It has been demonstrated that this material is often unstable at low and ambient temperatures; i.e., it is prone to discontinuous recrystallization or abnormal grain growth [6,10–13]. However, a comprehensive understanding of the effect of annealing temperature on microstructure and texture evolution of cryo-deformed copper is still lacking. The objective of the present work was to fill this gap in knowledge.
2. Material and experimental procedures The material used in the present work consisted of 99.9 wt pct. pure copper supplied as a hot-rolled bar. The as-received material was preconditioned by severe “abc” deformation [14] and then cryogenically rolled to 90 pct. overall thickness reduction (true strain¼ 2.3). The total thickness reduction was achieved using multiple passes of 10 pct. each. In order to provide cryogenicdeformation conditions, the rolling perform and work rolls were soaked in liquid nitrogen prior to each pass and held for 20 min. Immediately after the pass, the workpiece was re-inserted into liquid nitrogen. The total time of each pass (i.e., the exposure time of the specimen under ambient conditions) was only a few seconds. Heat transfer calculations revealed that the warming of the rolls and copper specimens prior to rolling due to free convection in air was small, resulting in temperature increases of the order only 1–4 1C. Additional details of the cryo-rolling process are described elsewhere [6]. To investigate the subsequent annealing behavior of the cryorolled material, samples were furnace annealed over a range of temperatures from 50 to 950 1C for 1 h (+10 min for heatup), as well as isothermally for various times at 150 and 450 1C. For the isothermal tests, the specimen temperature was continuously monitored by a thermocouple. Following heat treatment, each specimen was quenched in water. To preserve the microstructures developed during each thermomechanical treatment, the cryorolled as well as annealed samples were stored in a freezer at 20 1C prior to examination. To obtain insight into the three-dimensional nature of microstructure development, metallurgical observations for the asrolled and the rolled-and-annealed samples were made in the plane containing the RD and ND (i.e., the longitudinal plane) as well as in the plane containing the RD and TD (i.e., the rolling plane). (Per the typical flat-rolling convention, RD denotes the rolling direction, TD the transverse direction, and ND the normal direction of the rolled sheet.) Specifically, microstructure and texture were determined using the electron backscatter diffraction (EBSD) technique. For this purpose, samples were prepared using conventional metallographic techniques followed by electropolishing in a solution of 70 pct. orthophosphoric acid in water at ambient temperature with an applied potential of 5 V. In the longitudinal plane, all observations were made at the midthickness of the rolled sheet. High-resolution EBSD analysis was conducted using a Hitachi S-4300SE field-emission gun, scanning-electron microscope equipped with a TSL OIM™ EBSD system. Depending on the particular microstructure, the EBSD scan step size ranged from 0.1 to 5 μm. To improve the reliability of the EBSD data, small grains comprising three or fewer pixels were automatically
Fig. 1. Effect of annealing temperature on microhardness. Error bars show standard deviation of the measurements.
removed from the maps using the grain-dilation option in the TSL software in the TSL software. Furthermore, to eliminate spurious boundaries caused by orientation noise, a lower limit boundary-misorientation cutoff of 21 was used. A 151 criterion was used to differentiate low-angle boundaries (LABs) and high-angle boundaries (HABs). Because the microstructures developed during large deformation are frequently characterized by a complex mixture of LABs and HABs, there is often confusion regarding the definition of grains. To avoid ambiguity, the term “grain” in the present work refers to a crystallite bordered by a continuous boundary having a misorientation of greater than 151. For the cryo-rolled and partiallyrecrystallized microstructures at temperatures of 150 1C and below, the grain size was determined by the linear intercept method along the ND (i.e., “grain thickness”). At higher annealing temperatures, essentially equiaxed grain structures had developed, and the grain size in these cases was quantified by measurement of the grain area (ignoring annealing twin boundaries) and calculation of the circle-equivalent diameter (i.e., the so-called grain-reconstruction method [15]). To obtain a broader view of underlying microstructure changes, the Vickers microhardness was also measured on each sample using a load of 50 g for 10 s. At least 10 measurements were made in each case to obtain an average value.
3. Results 3.1. General trends of microstructure evolution The broad aspects of the annealing behavior in terms of the evolution of microhardness, microstructure, and texture during isochronal annealing in the range of 50–950 1C are summarized in this section. 3.1.1. Microhardness The effect of annealing temperature on microhardness is illustrated in Fig. 1. Four different temperature regimes were noted as follows: (i) At temperatures below 150 1C, the microhardness decreased rapidly with increasing temperature. (ii) In the temperature range of 150–300 1C, the hardness tended to saturate.
180
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Fig. 2. Selected portions of grain-boundary EBSD maps showing the grain structure which developed (a) after cryo-rolling and subsequent annealing at (b) 50 1C, (c) 150 1C, (d) 300 1C, (e) 400 1C, or (f) 800 1C. In the maps, red, black, and gray lines depict LABs, HABs, and Σ3 twin boundaries (within a 51 tolerance), respectively; RD, ND and TD are the rolling, normal, and transverse directions, respectively. Note the different magnifications. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
(iii) At 350 1C, an abrupt softening took place followed by a very gradual decrease of microhardness until a temperature of 700 1C; (iv) In the temperature range of 750–950 1C, the microhardness showed a slight increase. The observed microhardness behavior suggested that microstructural changes during annealing were complex. 3.1.2. Grain structure A number of details of microstructure evolution were revealed by EBSD maps for the cryo-rolled and annealed specimens; typical examples are shown in Fig. 2. In these maps, LABs, HABs, and Σ3 twin boundaries (within a 51 tolerance) were depicted by red,1 black, and gray lines, respectively. The overall statistics from the grain-size and misorientation measurements are given in Fig. 3. The microstructure of the as-cryo-rolled material (Fig. 2a) was dominated by pancake-shaped grains which were highly elongated along the RD. The grains contained a dense sub-boundary network; the LABs comprised 51% of the total grain-boundary area. Accounting for boundaries with a misorientation below 21, the actual LAB proportion was likely even higher. In addition to the LAB network, a number of fine, equiaxed grains decorating the boundaries of the coarse elongated grains were noted; an example is indicated by the arrow in Fig. 2a. As found in previous 1
color.
The reader is referred to on-line version of the paper to see the figures in
work [16], the fine grains originated from local grain boundary bulging occurring during warming of the cryo-rolled material to ambient temperature after deformation. After annealing at 50 1C, the grain structure (Fig. 2b) was noticeably different relative to the as-rolled condition. The principal features consisted of a number of abnormal, coarse grains within the originally fine-grained matrix; i.e., the microstructure had become essentially bimodal. The very large difference between the grain sizes suggested that the material had undergone the initial stage of discontinuous recrystallization (with a few rapidly growing nuclei). The abnormal, coarse grains were typically free of LABs, but contained annealing twins and even sporadic, fine unconsumed grains. The fraction of fine, equiaxed grains also increased, thus resulting in a reduction of the mean grain size (Fig. 3a), and the fraction of LABs decreased slightly (Fig. 3b). The latter effect was obviously associated with LAB elimination by abnormally-growing grains. The material annealed at 150 1C exhibited a nearly-completelyrecrystallized grain structure (Fig. 2c). However, the microstructure was rather inhomogeneous, comprising both relatively-small ( 1 μm) and large ( 410 μm) grains. Generally, recrystallization enlarged the mean grain size significantly (Fig. 3a) and reduced the LAB fraction to 10 pct (Fig. 3b). As expected, recrystallization was accompanied by the formation of a large number of annealing twins, which substantially increased the proportions of Σ3 and Σ9 boundaries (Fig. 3b). In the temperature range of 150–300 1C, the grain structure was found to be relatively stable showing no significant changes in morphology (Fig. 2c, d), mean grain size (Fig. 3a), and misorientation distribution (Fig. 3b).
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
181
Fig. 3. Effect of annealing temperature on (a) mean grain size and (b) area fractions of LABs, Σ3, and Σ9 twin boundaries. In (a), the mean grain size was measured ignoring twin boundaries. In (b), the fractions of Σ3 and Σ9 twin boundaries are given within Brandon′s range. Note: Tm is melting point.
For an annealing temperature of 350 1C and higher, the grain structure began to coarsen noticeably (Figs. 2e and 3a). More importantly, the microstructure was markedly bimodal (Fig. 2e) thus suggesting that grain growth following the completion of recrystallization was abnormal in nature. Moreover, the LAB fraction tended to increase unexpectedly with annealing temperature (Fig. 3b); this effect is discussed further in Section 3.1.3. In contrast, the proportion of twin boundaries was noticeably decreased, and Σ9 (secondary twin) boundaries almost disappeared from the microstructure (Fig. 3b). At very high annealing temperatures (4750 1C), the mean grain size increased substantially (e.g., approximately doubling from 750 1C to 800 1C) (Fig. 3a). Some grains even achieved a millimeter scale (Fig. 2f). This enhanced grain growth was accompanied by a substantial increase in the fraction of Σ3 annealingtwin boundaries (Fig. 3b).
3.1.3. Texture To evaluate texture evolution, orientation distribution functions (ODFs) were derived from large EBSD maps typically containing 10,000–60,000 grains (including twins). In the case of the material annealed in the temperature range of 750–950 1C, however, the total statistics included only 3600–8000 grains because of the coarse-grain nature of the microstructures (Figs. 2f and 3a). For comparison purposes, several important ideal rolling and recrystallization texture components for face-centered cubic metals are listed in Table 1. ODFs illustrating the change in texture with annealing temperature are summarized in Fig. 4; several important ideal orientations are also indicated in the ODFs. The evolution of the peak ODF intensity is given in Fig. 5a. The development of selected important orientations (within 151 tolerance) as a function of temperature is presented in Fig. 5b, c. The spatial distributions of these textural components in the various microstructures are summarized in Fig. 6. The cryo-rolled material was characterized by a relativelystrong texture (Fig. 5a) which was dominated by the {011}〈211〉 Brass component (Figs. 4a and 5b). The formation of the Brass texture during cryo-rolling of copper has been reported previously [6,17] and has been attributed to the suppression of cross slip [6]. In the present work, the Brass texture was characterized by a large orientation spread giving rise to noticeable fractions of {011}〈100〉 Goss and {013}〈100〉 Cube-RD orientations (Figs. 4a and 5b). All three preferential orientations were arranged as heavily elongated grains in the deformed microstructure (Fig. 6a). The material also contained a small amount of {001}〈100〉 Cube texture (Fig. 6a).
Table 1 Several ideal rolling and recrystallization textures for face-centered cubic metals. Orientation
Euler angles (deg)
Miller indices
Φ
φ2
Rolling plane
Rolling direction
45 45 0 27
0 90 0 45
{011} {011} {001} {4;4;11}
〈211〉 〈100〉 〈100〉 〈11;11;8〉
Recrystallization textures Cube-RD 0 22 Brass-R 80 31
0 35
{013} {236}
〈100〉 〈385〉
φ1 Rolling textures Brass 35 Goss 0 Cube 0 Dillamore 90
Annealing at 50 1C significantly weakened the strength of the overall texture (Fig. 5a) as well as the intensity of the Brass and Goss components (Fig. 5b). On the other hand, the Cube and CubeRD components were strengthened (Fig. 5b). In addition, it was found that the appearance of components with orientations of (55;30;0) and (55;60;0) (arrows in Fig. 4b) evolved and were associated with the development of the abnormal, coarse grains (Fig. 6b). Interestingly, the Cube-RD and (55;30/60;0) orientations were often found to be nearly twin-related (circled in Fig. 6b). After annealing at 150 1C, the texture was substantially altered relative to that developed at the lower temperature (Fig. 4c). In particular, it became very weak (Fig. 5a), and the original Brass and Goss components had almost completely disappeared (Fig. 5b). Instead, the texture comprised mainly (55;30/60;0) and Cube-RD components as well as newly-developed {236}〈385〉 Brass-R and {4;4;11}〈11;11;8〉 Dillamore orientations (Figs. 4c and 5b, c). All of these components were distributed nearly randomly in the microstructure (Fig. 6c). In the temperature range of 150–300 1C, the texture was found to be stable and showed no significant changes in strength (Fig. 5a) or volume fractions of the principal components (Fig. 5b, c). At 350 1C and higher, the texture began to strengthen (Fig. 5a). This observation might be related to the observed increase in LAB fraction in the microstructure (Fig. 3b). Indeed, the stronger texture can be rationalized on the basis of a larger fraction of grains with close orientations and thus the higher probability that the misorientation between neighboring grains is low-angle in nature. From a quantitative standpoint, the texture strengthening effect originated mainly from the rapid increase in the fractions of Brass and Goss components (Figs. 4d and 5b). In the microstructure, these orientations were
182
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Fig. 4. Orientation distribution functions (ODFs) showing the textures which developed after (a) cryo-rolling, and then subsequent annealing at (b) 50 1C, (c) 150 1C, (d) 400 1C, or (e) 950 1C. In the ODFs, the Brass, Goss, Cube, Cube-RD, Brass-R, and Dillamore orientations are depicted as B, G, C, C-RD, BR, and D, respectively.
Fig. 5. Effect of annealing temperature on (a) the peak intensity in the ODFs, as well as the volume fraction of the main texture components of (b) the as-cryo-rolled texture and (c) the recrystallization texture. Note: Tm is melting point.
arranged as relatively coarse grains (Fig. 6e). It appeared that the formation of the bimodal microstructure in this temperature range (Fig. 2e) was essentially a result of the preferential growth of the Brass and Goss orientations. The texture component (90;45;0) also developed at 400 1C (arrow in Fig. 4d). This finding can be attributed to the formation of annealing twins within grains with the Brass and Goss orientations. At temperatures exceeding 750 1C, the volume fractions of (55;30/ 60;0), Brass-R, and Dillamore orientations rapidly decreased (Fig. 5c), and the texture was dominated by the Brass and Goss grains (Figs. 4e and 6f).
3.1.4. Summary Based on the experimental observations summarized in Section 3.1, two principal temperature regimes in the annealing behavior of the cryo-rolled copper may be defined, viz., below and above 350 1C ( 0.5Tm, where Tm is the melting point).
At the lower temperatures, the annealing process may be interpreted in the terms of recrystallization. In the temperature range of 50 to 150 1C, grains having orientations close to (50;30/ 60;0), Brass-R, Dillamore and Cube-RD grew rapidly consuming the deformed microstructure (Fig. 6b, c). This process led to significant softening (Fig. 1), a coarsened grain structure (Fig. 3a), and reduced LAB fraction. It also stimulated the formation of annealing twins (Fig. 3b) and changed the texture noticeably (Figs. 4c and 5). Equally, if not more, important, the texture which evolved was drastically different from the conventional Cube texture which is usually observed in recrystallized copper [18]. The recrystallization process occurred via a nucleation2-andgrowth mechanism, and thus it may be considered to be discontinuous in nature. At temperatures of 150–300 1C, recrystallization
2 The process of nucleation of the recrystallized is considered in detail in Section 3.2.
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
183
Fig. 6. Distributions of the main texture components within the EBSD maps: (a) as-cryo-rolled material, or material which was rolled and then annealed at (b) 50 1C, (c) 150 1C, (d) 300 1C, (e) 400 1C, or (f) 950 1C. In the maps, black and gray lines depict HABs and Σ3 twin boundaries (within 51 tolerance), respectively; RD, ND and TD are the rolling, normal and transverse directions, respectively. In (f), three isolated EBSD maps ware merged together to improve the statistics. Note the different magnifications. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
was nearly complete during the 1-h annealing period and therefore the material exhibited limited change in strength (Fig. 1), microstructure morphology (Fig. 2c, d), grain size (Fig. 3a), misorientation distribution (Fig. 3b), and texture (Fig. 5a–c). At temperatures of 350 1C and higher, microstructure evolution was dominated by grain growth. The initiation of this process at 350 1C can be surmised to be responsible for the abrupt softening found at this temperature (Fig. 1). Progressive grain growth with temperature led to a drastic coarsening of the grain structure (Fig. 2e, f) and a fundamental alteration of the texture (Figs. 4d, e and 5). The latter effect was mainly related to the abnormal growth of the Brass and Goss grains (Fig. 6e, f). Their preferential development provided considerable texture strengthening (Fig. 5a) and a concomitant increase in the proportion of LABs (Fig. 3b). On the other hand, the muted material strengthening effect observed above 700 1C (Fig. 1) is not clear. According to Garcia et al. [19], it may be associated with the precipitation of Cu2O particles, but this hypothesis requires experimental verification. 3.2. Recrystallization More detailed insight into the recrystallization process was obtained from cryogenically-rolled material which was isothermally annealed at 150 1C for various times ranging from 1 min to 10 h, as well as an additional specimen which had been quenched immediately upon reaching 150 1C (i.e., a heat-up time of approximately
2 min). The evolution of microhardness, microstructure, and texture during these isothermal experiments is summarized in this section.
3.2.1. Broad aspects of recrystallization The effect of heat treatment at 150 1C on microhardness is shown in Fig. 7a. One of the most striking features was the drastic softening from 185 to 120 Hv due solely to the heat-up to 150 1C, thus indicating that the material experienced substantial microstructural changes even prior to any of the isothermal hold times. During the hold at temperature, the microhardness rapidly decreased further to 70–80 Hv tending to saturate after 4 min at temperature. EBSD measurements confirmed the marked instability of the cryo-rolled material. The recrystallized fraction in material simply heated to 150 1C was determined to be 77 pct. (Figs. 8a and 7b). Therefore, the incubation period for recrystallization was very short, if any. The recrystallized fraction grew rapidly with time; it exceeded 95 pct. after 8 min at temperature (Figs. 8a–d and 7b). On the other hand, isolated “islands” on unrecrystallized material survived even after 10 h of annealing; an example is circled in Fig. 8f. Interestingly, the coarse recrystallized grains were often elongated in the RD (Fig. 8). As expected, the size of the recrystallized grains did not change significantly during the isothermal-hold period, which represented a very late stage of recrystallization as noted above (Fig. 9a). Moreover,
184
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Fig. 7. Effect of heat treatment time at 150 1C on (a) microhardness and (b) recrystallized volume fraction.
Fig. 8. Selected portions of low-resolution grain-boundary EBSD maps showing the grain structure which developed (a) after heating to 150 1C, and subsequent annealing at this temperature for times of (b) 1 min, (c) 2 min, (d) 8 min, (e) 32 min, and (f) 10 h. In the maps, red and black lines depict LABs and HABs, respectively. For simplicity, Σ3 twin boundaries (within 51 tolerance) were removed from the maps; RD, ND and TD are rolling, normal and transverse directions, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
the grain-size distribution (GSD) of the recrystallized material was relatively broad with grain sizes between 1 and 40 μm (Fig. 9a). The recrystallized texture was also stable (Fig. 9b). In agreement with Figs. 5b, c, it consisted primarily of (50; 30/60;0), BrassR, Dillamore and Cube-RD orientations. The fractions of Brass and
Goss grains were small (Fig. 9b). On the other hand, the crystallographic orientation of the unrecrystallized areas was dominated by the Brass component (not shown). These observations suggested that recrystallization was very sluggish in the Brass and Goss grains.
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
185
Fig. 9. Effect of annealing time at 150 1C on the (a) grain-size distributions and (b) volume fractions of texture components of recrystallized material.
Fig. 10. High-resolution EBSD analysis of apparently-unrecrystallized material revealing the microstructure morphology at (a) low and (b) high magnifications as well as (c) the evolution of misorientation-angle distribution relative to the as-cryo-rolled state. In (a) and (b), red, black and gray lines depict LABs, HABs and Σ3 boundaries (within 51 tolerance), respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
3.2.2. Nucleation of recrystallization 3.2.2.1. Microstructure fragmentation. As shown in Section 3.2.1, the cryo-rolled material underwent substantial recrystallization even during the heating stage, and thus the details of the onset of recrystallization were not available for analysis. To help overcome this difficulty (at least partially), unrecrystallized areas (indicated in red in Fig. 8) were examined. In particular, the specimen exposed 1 min at 150 1C was used for this purpose. A high-resolution EBSD map of the unrecrystallized material is shown in Fig. 10a; a section of the map at higher magnification is shown in Fig. 10b. These maps revealed that a nearly-equiaxed grain structure predominated in the material. It seems therefore that the pancake-shaped cryo-rolled grains started to fragment prior to isothermal-recrystallization heat treatment. It was also evident that the fragmented grains were often free of LABs and typically contained annealing twins. In other words, they were
essentially recrystallized already. If so, the recrystallized fraction shown in Fig. 7b was underestimated. In principle, the fragmentation of pancake-shaped grains may have occurred via the transformation of transverse LABs into HABs or by thermally-induced impingement of mobile longitudinal grain-boundary bulges (i.e. so-called geometrical recrystallization [18]). An example of geometrical recrystallization is circled in Fig. 10b. To examine these two possibilities further, the evolution of the misorientation-angle distribution (Fig. 10c) was further examined. If the LAB-to-HAB transformation mechanism had predominated, a substantial increase in the area of grain boundaries with moderate misorientation ( 10–201) would have been expected. This was not observed experimentally, however; i.e., the area of boundaries with low and moderate misorientation was actually significantly reduced (Fig. 10c). Hence, the observed effect may be explained by the elimination of the LABs by migrating
186
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Table 2 Evaluation of stored energy in cryo-rolled material and in the apparently unrecrystallized material annealed at 150 1C for 1 min. Texture component Number of measured grains
Brass Goss Brass-R Dillamore (55;30/60;0) Cube-RD Cube
Mean grain orientation spread (deg)
Cryo-rolled
Annealed
Cryo-rolled
Annealed
1391 750 202 235 204 438 97
1536 657 326 212 145 551 195
2.1 2.1 1.6 1.7 1.5 1.6 1.5
2.3 2.1 1.2 0.9 0.9 1.8 1.6
Table 3 Grain-structure characteristics for different texture components in apparently unrecrystallized material annealed at 150 1C for 1 min. Texture component Length of 401 〈111〉 boundaries Equivalent grain diameter excluding per unit of grain area within twins (μm) a 10-deg tolerance (μm 1)
Brass Goss Brass-R Dillamore (55;30/60;0) Cube-RD Cube
0.319 0.420 0.595 0.589 0.367 0.521 0.513
Modal
Maximum
1.4 1.3 1.1 1.4 1.3 1.3 1.8
6.4 3.1 2.5 2.7 3.9 3.9 6.1
Note: The main recrystallization textures are highlighted in gray. Note: The main recrystallization textures are highlighted in gray.
grain boundaries, and thus it appears that the fragmentation process was driven mainly by the geometrical-recrystallization mechanism. This mechanism is believed to be very sensitive to local variations of dislocation density as well as grain-boundary mobility and thus the recrystallization process should develop highly inhomogeneously. This may be one of the possible reasons for the relatively broad GSD in the recrystallized material seen in Fig. 9a. 3.2.2.2. Recrystallization nuclei. According to the “oriented nucleation” theory, recrystallization nuclei are characterized by relatively-low stored energy [18]. To assess the applicability of this theory, it was necessary to estimate the distribution of stored energy in the deformed material (and hence potential nuclei) as well as its variation during annealing. For this purpose, the grain orientation spread (GOS) within different texture components in the cryo-rolled as well as fragmented materials was measured. The GOS approach quantifies the geometrically-necessary component of a dislocation array and thus may provide a broad indication of the stored energy. To measure the grain orientation spread, an average misorientation between each pixel of a grain and the average orientation of the grain was calculated. In the case of the fragmented material, the grain orientation spread was measured only for grains larger than 1 μm. In smaller grains, the surface energy likely exceeds the energy of the eliminated dislocations, and thus these grains are probably unable to serve as recrystallization nuclei [18]. The measurements (Table 2) revealed that the Brass and Goss texture components exhibited by the largest GOS values in both the as-cryo-rolled and the annealed conditions. It may thus be hypothesized that these two orientations were characterized by relatively large stored energy, thus in line with the experimentally-observed disappearance of these textural components during recrystallization (Figs. 5b and 9b). However, its fundamental reason is not clear and requires additional work. Because the recrystallization texture was dominated by the (55;30/60;0), Brass-R, and Dillamore orientations (Figs. 5c, 9b), these textural components were of particular interest. In the ascryo-rolled state, these grains were characterized by relatively-low values of GOS, and thus likely served as the recrystallization nuclei. Moreover, the GOS values were reduced substantially during annealing (Table 2). Assuming that pancake-grain fragmentation occurred via geometrical recrystallization, these trends may be concluded to be related to the elimination of dislocations by migrating grain boundaries. However, it is not evident from Table 2 why the (55;30/60;0) grains predominated over the Brass-R and Dillamore orientations in the recrystallization texture (Figs. 5c and 9b). This effect may be related to the specific spatial distribution of the (55;30/60;0) grains. Specifically, this
Fig. 11. Effect of annealing time at 450 1C on microhardness.
orientation is relatively close to Brass (Fig. 4b). Thus, it is possible that the (55;30/60;0) grains originated from orientation spread of the Brass component. If so, the (55;30/60;0) grains would have been frequently surrounded by high-energy Brass grains, thereby promoting their growth The Cube-RD and Cube grains were also characterized by relatively low values of GOS in the as-cryo-rolled state (Table 2), whose magnitude, however, was not reduced in the fragmented microstructure. Thus, the rate of recovery of these orientations appears to have been relatively slow. This observation was also in good agreement with the very modest contribution of these two orientations to the recrystallization texture (Figs. 5b and 9b), but the physical basis for this behavior is also unclear. Furthermore, the slight increase observed in the Cube-RD component (Fig. 9b) is thought to be related to the formation of annealing twins in the (55;30/60;0) grains, examples of which are circled in Fig. 6b. According to the “oriented growth” theory, the selective growth of the recrystallization nuclei may also be related to highly mobile 401〈111〉 boundaries [18]. To examine this possibility, the total length of these boundaries (per unit area of the EBSD map) of the different texture components was measured (Table 3). The dominant recrystallization-texture component (55;30/60;0) was characterized by one of the lowest specific lengths of 401〈111〉 boundaries. It is therefore unlikely that oriented growth played a significant role in the recrystallization process in the present work. Additionally, the preferential development of the nuclei did not appear to be related with their relatively large size (Table 3).
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
187
Fig. 12. Selected portions of grain-boundary EBSD maps showing the grain structure which developed (a) after heating to 450 1C, and subsequent annealing at this temperature for (b) 1 min, (c) 4 min, (d) 8 min, (e) 32 min, and (f) 10 h. In the maps, red and black lines depict LABs and HABs, respectively. For simplicity, Σ3 twin boundaries (within 51 tolerance) were removed from the maps; RD, ND and TD denote the rolling, normal and transverse directions, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
It seems therefore that the recrystallization was mainly driven by the oriented-nucleation effect.
3.2.3. Summary Grain boundaries in the cryo-deformed copper seem to be very mobile, being prone to migrate even at relatively low temperatures. Due to the pinning effect associated with low-mobility LABs, however, grain-boundary migration degenerated into local grain boundary bulging which fragmented the elongated rolled grains into lower-aspect-ratio grain segments. An important consequence of this process was a reduction in stored energy within the grains. This process was deduced to occur most rapidly in grains having orientations close to (55;30/60;0), Brass-R, and Dillamore thus providing them a growth advantage. Due to the small fraction of these components in the deformed material, their preferential development led to discontinuous recrystallization. Recrystallization occurred very rapidly. The cryo-deformed material was found to recrystallize to a large extent during heat-up to 150 1C. On the other hand, Brass-oriented grains were found to be very resistant to recrystallization, and thus isolated islands of the Brass grains survived even after long-term heat treatment at 150 1C. Finally, the grain-size distribution of the recrystallized grains was relatively broad with sizes ranging from 1 to 40 μm.
3.3. Grain growth More detailed insight into the grain-growth process was obtained from the cryogenically-rolled material which was isothermally annealed at 450 1C for times ranging from 1 min to 10 h. Again, one specimen was immediately quenched after reaching the soak temperature; the heating time in this case was 30 s.3 3.3.1. Microstructure The effect of heat treatment time at 450 1C on microhardness and microstructure morphology is shown in Figs. 11 and 12, respectively. It is clear that the cryo-rolled material was completely recrystallized during the heat-up to the annealing temperature (Figs. 11 and 12a). With further time at 450 1C, the grain structure coarsened substantially (Fig. 12), thereby leading to a further, moderate degree of softening (Fig. 11). However, softening tended to saturate after 4 min at 450 1C (Fig. 11). The grain structures developed during heat treatment at 450 1C were noticeably bimodal (Fig. 12b–d) thereby suggesting that grain growth was abnormal in nature. Nevertheless, the GSDs appeared to have a single peak (Fig. 13a). Therefore, to quantify further the 3 The different heating time during isothermal annealing at 150 and 450 1C (i.e. 2 min and 30 s) was caused by different furnaces used for these experiments.
188
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Fig. 13. Effect of annealing time at 450 1C on grain-size frequency distributions as a function of (a) circle-equivalent grain diameter or (b) the logarithm of D/Davg. (Davg. ¼ average grain diameter).
Fig. 15. Effect of annealing time at 450 1C on volume fractions of the main texture components. Fig. 14. Effect of annealing time at 450 1C on grain boundary length for different types of boundaries (measured in an EBSD map) normalized by the area of the EBSD maps.
nature of grain growth, normalized GSDs were calculated (Fig. 13b). It was found that the normalized distributions changed significantly with annealing time thus confirming the abnormal character of growth. Moreover, all of the GSDs were distinctly different from log-normal ones, being skewed toward larger grain sizes. Due to the large spread in grain size, the evolution of the mean grain size with annealing time was concluded not to be a good indicator of grain-growth kinetics. Instead, grain-boundary length per unit area of the EBSD map was determined (Fig. 14). In such a plot, the normalized length of HABs (excluding annealing twin boundaries) was reduced by a factor of 3.5 during the beginning of annealing; this indicated rapid grain growth. However, the grain boundary length tended to stabilize after 10 min, suggesting a saturation of the grain-growth process. On the other hand, the area of LABs and Σ3 boundaries was found to change relatively little during annealing (Fig. 14).
3.3.2. Texture The effect of annealing time on the volume fractions of the main texture components is shown in Fig. 15. Similar to the findings for the recrystallization process, the principal graingrowth texture components were (55;30/60;0), Brass-R, and Dillamore, whereas the fraction of Cube-RD and Cube components was relatively low (compare Figs. 9a and 15). A distinctive feature
of grain growth, however, was a substantial strengthening of the Brass and Goss components (Fig. 15). As shown in the previous section, the cryo-rolled material was completely recrystallized during the heat-up stage (Figs. 11 and 12a), and the LAB fraction was very low (Fig. 14). According to the Humphrey′s cellular model, abnormal grain growth and its associated significant texture change are likely to be associated with a large spread in grain size [18]. In this regard, GSDs were measured for each of the principal texture components; the results are summarized in Table 4. As expected, the maximum grain size for the Cube and Cube-RD orientations was substantially smaller than that of the (55;30/ 60;0), Brass-R, and Dillamore components. As shown in Section 3.2, the latter three orientations had a growth advantage during recrystallization and thus could achieve relatively large sizes. Because the grain-size spread in the as-heated material was also very large (Fig. 13), the tendency for abnormal grain growth would have been exacerbated. On the other hand, the predominant (55;30/60;0) grains would often impinge upon each other during growth thus giving rise to LABs. This “orientation pinning” effect would gradually suppress the development of these orientations [9], as was observed (Fig. 15). The abnormal growth of Brass and Goss grains was striking (Fig. 15). As shown in Section 3.2, these orientations were fairly resistant to recrystallization, being stable even after a 10-h anneal at 150 1C. It may be hypothesized that some portion of these long as-rolled grains survived (but recovered) during the heat-up stage and thus underwent subsequent abnormal growth. Per the results in Table 4, the as-heated material did indeed contain relatively coarse Brass grains. On the other hand, the Goss grains were
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Table 4 Grain-structure characteristics for different texture components developed after heating cryo-rolled material to 450 1C. Texture component
Cube-RD Cube Brass Goss Dillamore Brass-R (55;30/60;0)
Number of measured grains
Grain size Mean Maximum
Length of 401 〈111〉 boundaries per unit of grain area within a 10-deg tolerance (μm 1)
2737 840 1212 380 3323 2164 3355
4.2 4.0 3.7 3.8 5.5 4.5 4.8
0.063 0.078 0.073 0.074 0.053 0.055 0.065
17.8 14.2 27.5 17.8 37.5 22.3 33.6
Note: The main grain-growth textures are highlighted in gray.
relatively small, and thus the reason for the abnormal growth of this orientation was not clear. The specific grain-boundary area of highly-mobile 401〈111〉 boundaries was also measured in the as-heated condition at 450 1C (Table 4). This area did not vary significantly among the various texture components, and thus abnormal grain growth was not likely associated with an oriented-growth effect. 4. Summary and conclusions The annealing behavior of cryogenically-rolled copper was quantified. To this end, the material was rolled to a 90-pct. thickness reduction at liquid-nitrogen temperature and then isochronally annealed at temperatures between 50 and 950 1C for 1 h as well as isothermally for various times at 150 or 450 1C. Grain-structure and texture changes were quantified using an EBSD technique. The main conclusions from this work are as follows: (1) The material had very poor microstructure stability. Gross evidence of recrystallization was observed after annealing at 50 1C for 1 h whereas the recrystallization process was almost complete after annealing at 150 1C for 8 min. (2) Microstructure evolution at temperatures below 350 1C ( 0.5Tm) may be interpreted in the terms of discontinuous static recrystallization. Grains having crystallographic orientations close to (55;30/60;0), Brass-R, and Dillamore were deduced to recover most rapidly, thus providing them with a subsequent growth
189
advantage. On the other hand, the Brass and Goss orientations were found to be resistant to recrystallization. The grain-size distribution of the recrystallized grains was relatively broad, spanning 1 to 40 μm. (3) At temperatures above 300 1C, the annealing behavior was found to be driven primarily by abnormal grain growth. The abnormal nature of this process was related to the relatively large grain-size spread produced during the preceding recrystallization.
Acknowledgments Financial support from the Russian Fund of Fundamental Research (Project no.12-08-97008) is gratefully acknowledged. The authors would like to thank Professor G.A. Salishchev for suggesting this research. They are also very grateful to Dr. R.M. Galeyev and Dr. O.R. Valiakhmetov for graciously providing the material used in this work, P. Klassman and T.I. Nazarova for technical assistance during cryogenic rolling, and Dr. R. R. Kabirov for help in performing the annealing experiments.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]
Y.S. Li, N.R. Tao, K. Lu, Acta Mater. 56 (2008) 230–241. Y. Zhang, N.R. Tao, K. Lu, Acta Mater. 56 (2008) 2429–2440. Y. Huang, P.B. Prangnell, Acta Mater. 56 (2008) 1619–1632. S.K. Panigrahi, R. Jayaganathan, Mater. Sci. Eng. A. 480 (2008) 299–305. S.K. Panigrahi, R. Jayaganathan, V. Chawla, Mater. Lett. 62 (2008) 2626–2629. T. Konkova, S. Mironov, A. Korznikov, S.L. Semiatin, Acta Mater. 58 (2010) 5262–5273. V. Subramanya Sarma, J. Wang, W.W. Jian, A. Kauffmann, H. Conrad, J. Freudenberger, Y.T. Zhu, Mater. Sci. Eng. A 527 (2010) 7624–7630. G.H. Zahid, Y. Huang, P.B. Prangnell, Acta Mater. 57 (2009) 3509–3521. D. Juul Jensen, Acta Mater. 43 (1995) 4117–4129. H.D. Meingelberg, M. Meixner, K. Lucke, Acta Metal. 13 (1965) 835–844. Y.M. Wang, E. Ma, Acta Mater. 52 (2004) 1699–1709. T. Konkova, S. Mironov, A. Korznikov, S.L. Semiatin, Scr. Mater. 63 (2010) 921–924. V. Subramanya Sarma, K. Sivaprasad, D. Sturm, M. Heilmaier, Mater. Sci. Eng. A 489 (2008) 253–258. S.V. Dobatkin, G.A. Salischev, A.A. Kuznetsov, T.N. Konkova, Mater. Sci. Forum 558–559 (2007) 189–194. F.J. Humphreys, J. Microsc. 195 (1999) 170–185. T. Konkova, S. Mironov, A. Korznikov, S.L. Semiatin, Mater. Sci. Eng. A 528 (2011) 7432–7443. J. Pospiech, K. Lucke, Acta Mater. 23 (1975) 997–1007. F.J. Humphreys, M. Hatherly, Recrystallization and Related Phenomena, second ed., Elsevier, Oxford, 2004. V.G. Garcia, J.M. Cabrera, J.M. Prado, Mater. Sci. Eng. A 488 (2008) 92–101.
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Annealing behavior of cryogenically-rolled copper T. Konkova a, S. Mironov b,n, A. Korznikov a, M.M. Myshlyaev c, S.L. Semiatin d a
Institute for Metals Superplasticity Problems, Russian Academy of Science, 39 Khalturin Str., Ufa 450001, Russia Department of Materials Processing, Graduate School of Engineering, Tohoku University, 6-6-02 Aramaki-aza-Aoba, Sendai 980-8579, Japan Baikov Institute of Metallurgy and Material Science, Russian Academy of Science, 49 Lenin-av., Moscow 119991, Russia d Air Force Research Laboratory, Materials and Manufacturing Directorate, AFRL/RXCM, Wright-Patterson AFB, OH 45433-7817, USA b c
art ic l e i nf o
a b s t r a c t
Article history: Received 8 May 2013 Received in revised form 30 June 2013 Accepted 4 July 2013 Available online 31 July 2013
The static annealing behavior of cryogenically-rolled copper over a wide temperature range (50–950 1C) was established. At temperatures below 350 1C ( 0.5Tm), microstructure and texture evolution were interpreted in terms of discontinuous recrystallization. Grains having orientations close to (55;30/60;0), {236}〈385〉 (Brass-R), and {4;4;11}〈11;11;8〉 (Dillamore) were shown to recover rapidly and thus exhibited preferential growth during subsequent static recrystallization. At temperatures of 350 1C and higher, annealing behavior was dominated by abnormal grain growth. The abnormal character of this process was attributed to the relatively large spread in grain sizes produced during preceding recrystallization. & 2013 Elsevier B.V. All rights reserved.
Keywords: EBSD Nanostructured materials Thermomechanical processing Recrystallization Grain growth
1. Introduction There is significant commercial interest in the development of materials with ultrafine grain structures for structural applications. This interest is mainly driven by a substantial improvement in strength and ductility as well as a good balance of these properties. In addition, the control of the mechanical properties by processing may be an attractive alternative to expensive alloying. This could result in the use of fewer and simpler industrial alloys and would lead to economic benefits as well as improved recyclability. Techniques for the production of fine-grain alloys are thus of considerable commercial interest. Of particular importance are cost-effective methods that can be used to obtain large quantities of such materials. In this regard, an approach involving large deformation at cryogenic temperatures has recently attracted significant attention. It is believed that low temperatures may suppress dynamic recovery and stimulate mechanical twinning (e.g., [1, 2]) thereby enhancing the grain-refinement effect. This may decrease the level of strain necessary to achieve an ultra-fine microstructure and thus enable the application of conventional working processes such as rolling to produce such materials.
n
Corresponding author. Tel.: +81 22 795 7353; fax: +81 22 795 7352. E-mail addresses: [email protected] (T. Konkova), smironov@material. tohoku.ac.jp, [email protected] (S. Mironov), [email protected] (A. Korznikov), [email protected] (M.M. Myshlyaev), [email protected] (S.L. Semiatin). 0921-5093/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.msea.2013.07.042
To date, the majority of research in the field of cryogenic working has focused on aluminum and copper alloys [e.g., 1–7], most likely because of the superior ductility of these materials. It has been established that the key mechanism governing grainstructure evolution in both materials at cryogenic temperatures is the geometrical effect of strain per se. In other words, grains change their shape in proportion to the imposed strain, and noticeable grain subdivision and mechanical twinning are not observed [3,6]. By this means, a reasonably homogeneous grain structure, dominated by heavily elongated grains aligned with the direction of macroscopic material flow, is developed. Such grain structures typically contain a significant proportion of low-angle boundaries [3,6] and, in the case of copper, a high density of free dislocations [6]. The limited formation of deformation-induced boundaries during cryogenic deformation is believed to be partially associated with suppression of cross-slip at low temperatures [6]. This effect is also responsible for the strengthening of the {110}〈112〉 Brass texture in cryo-rolled materials [3,6]. Despite these characteristics, cryogenic deformation has been shown to be able to produce ultra-fine grain structures in both materials in some cases [3,6]. Not surprisingly, one important requirement to obtain and maintain a fine-grain microstructure is a high degree of microstructural stability within the expected range of use temperature and time. Hence, the annealing behavior of such materials is of interest. For cryo-deformed aluminum, this issue has been studied systematically by Zahid et al. [8]. For example, it was found that the fraction of low-angle boundaries increased progressively with
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
179
annealing temperature, leading eventually to discontinuous recrystallization. At intermediate temperatures, such behavior resulted in a bimodal structure comprising bands of coarse grains and fine subgrains. This unusual behavior has also been related to a strong texture in deformed material giving rise to the so-called “orientation pinning” effect [9] during subsequent grain growth. In contrast to aluminum, relatively little attention has been paid to the annealing behavior of cryo-deformed copper. It has been demonstrated that this material is often unstable at low and ambient temperatures; i.e., it is prone to discontinuous recrystallization or abnormal grain growth [6,10–13]. However, a comprehensive understanding of the effect of annealing temperature on microstructure and texture evolution of cryo-deformed copper is still lacking. The objective of the present work was to fill this gap in knowledge.
2. Material and experimental procedures The material used in the present work consisted of 99.9 wt pct. pure copper supplied as a hot-rolled bar. The as-received material was preconditioned by severe “abc” deformation [14] and then cryogenically rolled to 90 pct. overall thickness reduction (true strain¼ 2.3). The total thickness reduction was achieved using multiple passes of 10 pct. each. In order to provide cryogenicdeformation conditions, the rolling perform and work rolls were soaked in liquid nitrogen prior to each pass and held for 20 min. Immediately after the pass, the workpiece was re-inserted into liquid nitrogen. The total time of each pass (i.e., the exposure time of the specimen under ambient conditions) was only a few seconds. Heat transfer calculations revealed that the warming of the rolls and copper specimens prior to rolling due to free convection in air was small, resulting in temperature increases of the order only 1–4 1C. Additional details of the cryo-rolling process are described elsewhere [6]. To investigate the subsequent annealing behavior of the cryorolled material, samples were furnace annealed over a range of temperatures from 50 to 950 1C for 1 h (+10 min for heatup), as well as isothermally for various times at 150 and 450 1C. For the isothermal tests, the specimen temperature was continuously monitored by a thermocouple. Following heat treatment, each specimen was quenched in water. To preserve the microstructures developed during each thermomechanical treatment, the cryorolled as well as annealed samples were stored in a freezer at 20 1C prior to examination. To obtain insight into the three-dimensional nature of microstructure development, metallurgical observations for the asrolled and the rolled-and-annealed samples were made in the plane containing the RD and ND (i.e., the longitudinal plane) as well as in the plane containing the RD and TD (i.e., the rolling plane). (Per the typical flat-rolling convention, RD denotes the rolling direction, TD the transverse direction, and ND the normal direction of the rolled sheet.) Specifically, microstructure and texture were determined using the electron backscatter diffraction (EBSD) technique. For this purpose, samples were prepared using conventional metallographic techniques followed by electropolishing in a solution of 70 pct. orthophosphoric acid in water at ambient temperature with an applied potential of 5 V. In the longitudinal plane, all observations were made at the midthickness of the rolled sheet. High-resolution EBSD analysis was conducted using a Hitachi S-4300SE field-emission gun, scanning-electron microscope equipped with a TSL OIM™ EBSD system. Depending on the particular microstructure, the EBSD scan step size ranged from 0.1 to 5 μm. To improve the reliability of the EBSD data, small grains comprising three or fewer pixels were automatically
Fig. 1. Effect of annealing temperature on microhardness. Error bars show standard deviation of the measurements.
removed from the maps using the grain-dilation option in the TSL software in the TSL software. Furthermore, to eliminate spurious boundaries caused by orientation noise, a lower limit boundary-misorientation cutoff of 21 was used. A 151 criterion was used to differentiate low-angle boundaries (LABs) and high-angle boundaries (HABs). Because the microstructures developed during large deformation are frequently characterized by a complex mixture of LABs and HABs, there is often confusion regarding the definition of grains. To avoid ambiguity, the term “grain” in the present work refers to a crystallite bordered by a continuous boundary having a misorientation of greater than 151. For the cryo-rolled and partiallyrecrystallized microstructures at temperatures of 150 1C and below, the grain size was determined by the linear intercept method along the ND (i.e., “grain thickness”). At higher annealing temperatures, essentially equiaxed grain structures had developed, and the grain size in these cases was quantified by measurement of the grain area (ignoring annealing twin boundaries) and calculation of the circle-equivalent diameter (i.e., the so-called grain-reconstruction method [15]). To obtain a broader view of underlying microstructure changes, the Vickers microhardness was also measured on each sample using a load of 50 g for 10 s. At least 10 measurements were made in each case to obtain an average value.
3. Results 3.1. General trends of microstructure evolution The broad aspects of the annealing behavior in terms of the evolution of microhardness, microstructure, and texture during isochronal annealing in the range of 50–950 1C are summarized in this section. 3.1.1. Microhardness The effect of annealing temperature on microhardness is illustrated in Fig. 1. Four different temperature regimes were noted as follows: (i) At temperatures below 150 1C, the microhardness decreased rapidly with increasing temperature. (ii) In the temperature range of 150–300 1C, the hardness tended to saturate.
180
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Fig. 2. Selected portions of grain-boundary EBSD maps showing the grain structure which developed (a) after cryo-rolling and subsequent annealing at (b) 50 1C, (c) 150 1C, (d) 300 1C, (e) 400 1C, or (f) 800 1C. In the maps, red, black, and gray lines depict LABs, HABs, and Σ3 twin boundaries (within a 51 tolerance), respectively; RD, ND and TD are the rolling, normal, and transverse directions, respectively. Note the different magnifications. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
(iii) At 350 1C, an abrupt softening took place followed by a very gradual decrease of microhardness until a temperature of 700 1C; (iv) In the temperature range of 750–950 1C, the microhardness showed a slight increase. The observed microhardness behavior suggested that microstructural changes during annealing were complex. 3.1.2. Grain structure A number of details of microstructure evolution were revealed by EBSD maps for the cryo-rolled and annealed specimens; typical examples are shown in Fig. 2. In these maps, LABs, HABs, and Σ3 twin boundaries (within a 51 tolerance) were depicted by red,1 black, and gray lines, respectively. The overall statistics from the grain-size and misorientation measurements are given in Fig. 3. The microstructure of the as-cryo-rolled material (Fig. 2a) was dominated by pancake-shaped grains which were highly elongated along the RD. The grains contained a dense sub-boundary network; the LABs comprised 51% of the total grain-boundary area. Accounting for boundaries with a misorientation below 21, the actual LAB proportion was likely even higher. In addition to the LAB network, a number of fine, equiaxed grains decorating the boundaries of the coarse elongated grains were noted; an example is indicated by the arrow in Fig. 2a. As found in previous 1
color.
The reader is referred to on-line version of the paper to see the figures in
work [16], the fine grains originated from local grain boundary bulging occurring during warming of the cryo-rolled material to ambient temperature after deformation. After annealing at 50 1C, the grain structure (Fig. 2b) was noticeably different relative to the as-rolled condition. The principal features consisted of a number of abnormal, coarse grains within the originally fine-grained matrix; i.e., the microstructure had become essentially bimodal. The very large difference between the grain sizes suggested that the material had undergone the initial stage of discontinuous recrystallization (with a few rapidly growing nuclei). The abnormal, coarse grains were typically free of LABs, but contained annealing twins and even sporadic, fine unconsumed grains. The fraction of fine, equiaxed grains also increased, thus resulting in a reduction of the mean grain size (Fig. 3a), and the fraction of LABs decreased slightly (Fig. 3b). The latter effect was obviously associated with LAB elimination by abnormally-growing grains. The material annealed at 150 1C exhibited a nearly-completelyrecrystallized grain structure (Fig. 2c). However, the microstructure was rather inhomogeneous, comprising both relatively-small ( 1 μm) and large ( 410 μm) grains. Generally, recrystallization enlarged the mean grain size significantly (Fig. 3a) and reduced the LAB fraction to 10 pct (Fig. 3b). As expected, recrystallization was accompanied by the formation of a large number of annealing twins, which substantially increased the proportions of Σ3 and Σ9 boundaries (Fig. 3b). In the temperature range of 150–300 1C, the grain structure was found to be relatively stable showing no significant changes in morphology (Fig. 2c, d), mean grain size (Fig. 3a), and misorientation distribution (Fig. 3b).
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
181
Fig. 3. Effect of annealing temperature on (a) mean grain size and (b) area fractions of LABs, Σ3, and Σ9 twin boundaries. In (a), the mean grain size was measured ignoring twin boundaries. In (b), the fractions of Σ3 and Σ9 twin boundaries are given within Brandon′s range. Note: Tm is melting point.
For an annealing temperature of 350 1C and higher, the grain structure began to coarsen noticeably (Figs. 2e and 3a). More importantly, the microstructure was markedly bimodal (Fig. 2e) thus suggesting that grain growth following the completion of recrystallization was abnormal in nature. Moreover, the LAB fraction tended to increase unexpectedly with annealing temperature (Fig. 3b); this effect is discussed further in Section 3.1.3. In contrast, the proportion of twin boundaries was noticeably decreased, and Σ9 (secondary twin) boundaries almost disappeared from the microstructure (Fig. 3b). At very high annealing temperatures (4750 1C), the mean grain size increased substantially (e.g., approximately doubling from 750 1C to 800 1C) (Fig. 3a). Some grains even achieved a millimeter scale (Fig. 2f). This enhanced grain growth was accompanied by a substantial increase in the fraction of Σ3 annealingtwin boundaries (Fig. 3b).
3.1.3. Texture To evaluate texture evolution, orientation distribution functions (ODFs) were derived from large EBSD maps typically containing 10,000–60,000 grains (including twins). In the case of the material annealed in the temperature range of 750–950 1C, however, the total statistics included only 3600–8000 grains because of the coarse-grain nature of the microstructures (Figs. 2f and 3a). For comparison purposes, several important ideal rolling and recrystallization texture components for face-centered cubic metals are listed in Table 1. ODFs illustrating the change in texture with annealing temperature are summarized in Fig. 4; several important ideal orientations are also indicated in the ODFs. The evolution of the peak ODF intensity is given in Fig. 5a. The development of selected important orientations (within 151 tolerance) as a function of temperature is presented in Fig. 5b, c. The spatial distributions of these textural components in the various microstructures are summarized in Fig. 6. The cryo-rolled material was characterized by a relativelystrong texture (Fig. 5a) which was dominated by the {011}〈211〉 Brass component (Figs. 4a and 5b). The formation of the Brass texture during cryo-rolling of copper has been reported previously [6,17] and has been attributed to the suppression of cross slip [6]. In the present work, the Brass texture was characterized by a large orientation spread giving rise to noticeable fractions of {011}〈100〉 Goss and {013}〈100〉 Cube-RD orientations (Figs. 4a and 5b). All three preferential orientations were arranged as heavily elongated grains in the deformed microstructure (Fig. 6a). The material also contained a small amount of {001}〈100〉 Cube texture (Fig. 6a).
Table 1 Several ideal rolling and recrystallization textures for face-centered cubic metals. Orientation
Euler angles (deg)
Miller indices
Φ
φ2
Rolling plane
Rolling direction
45 45 0 27
0 90 0 45
{011} {011} {001} {4;4;11}
〈211〉 〈100〉 〈100〉 〈11;11;8〉
Recrystallization textures Cube-RD 0 22 Brass-R 80 31
0 35
{013} {236}
〈100〉 〈385〉
φ1 Rolling textures Brass 35 Goss 0 Cube 0 Dillamore 90
Annealing at 50 1C significantly weakened the strength of the overall texture (Fig. 5a) as well as the intensity of the Brass and Goss components (Fig. 5b). On the other hand, the Cube and CubeRD components were strengthened (Fig. 5b). In addition, it was found that the appearance of components with orientations of (55;30;0) and (55;60;0) (arrows in Fig. 4b) evolved and were associated with the development of the abnormal, coarse grains (Fig. 6b). Interestingly, the Cube-RD and (55;30/60;0) orientations were often found to be nearly twin-related (circled in Fig. 6b). After annealing at 150 1C, the texture was substantially altered relative to that developed at the lower temperature (Fig. 4c). In particular, it became very weak (Fig. 5a), and the original Brass and Goss components had almost completely disappeared (Fig. 5b). Instead, the texture comprised mainly (55;30/60;0) and Cube-RD components as well as newly-developed {236}〈385〉 Brass-R and {4;4;11}〈11;11;8〉 Dillamore orientations (Figs. 4c and 5b, c). All of these components were distributed nearly randomly in the microstructure (Fig. 6c). In the temperature range of 150–300 1C, the texture was found to be stable and showed no significant changes in strength (Fig. 5a) or volume fractions of the principal components (Fig. 5b, c). At 350 1C and higher, the texture began to strengthen (Fig. 5a). This observation might be related to the observed increase in LAB fraction in the microstructure (Fig. 3b). Indeed, the stronger texture can be rationalized on the basis of a larger fraction of grains with close orientations and thus the higher probability that the misorientation between neighboring grains is low-angle in nature. From a quantitative standpoint, the texture strengthening effect originated mainly from the rapid increase in the fractions of Brass and Goss components (Figs. 4d and 5b). In the microstructure, these orientations were
182
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Fig. 4. Orientation distribution functions (ODFs) showing the textures which developed after (a) cryo-rolling, and then subsequent annealing at (b) 50 1C, (c) 150 1C, (d) 400 1C, or (e) 950 1C. In the ODFs, the Brass, Goss, Cube, Cube-RD, Brass-R, and Dillamore orientations are depicted as B, G, C, C-RD, BR, and D, respectively.
Fig. 5. Effect of annealing temperature on (a) the peak intensity in the ODFs, as well as the volume fraction of the main texture components of (b) the as-cryo-rolled texture and (c) the recrystallization texture. Note: Tm is melting point.
arranged as relatively coarse grains (Fig. 6e). It appeared that the formation of the bimodal microstructure in this temperature range (Fig. 2e) was essentially a result of the preferential growth of the Brass and Goss orientations. The texture component (90;45;0) also developed at 400 1C (arrow in Fig. 4d). This finding can be attributed to the formation of annealing twins within grains with the Brass and Goss orientations. At temperatures exceeding 750 1C, the volume fractions of (55;30/ 60;0), Brass-R, and Dillamore orientations rapidly decreased (Fig. 5c), and the texture was dominated by the Brass and Goss grains (Figs. 4e and 6f).
3.1.4. Summary Based on the experimental observations summarized in Section 3.1, two principal temperature regimes in the annealing behavior of the cryo-rolled copper may be defined, viz., below and above 350 1C ( 0.5Tm, where Tm is the melting point).
At the lower temperatures, the annealing process may be interpreted in the terms of recrystallization. In the temperature range of 50 to 150 1C, grains having orientations close to (50;30/ 60;0), Brass-R, Dillamore and Cube-RD grew rapidly consuming the deformed microstructure (Fig. 6b, c). This process led to significant softening (Fig. 1), a coarsened grain structure (Fig. 3a), and reduced LAB fraction. It also stimulated the formation of annealing twins (Fig. 3b) and changed the texture noticeably (Figs. 4c and 5). Equally, if not more, important, the texture which evolved was drastically different from the conventional Cube texture which is usually observed in recrystallized copper [18]. The recrystallization process occurred via a nucleation2-andgrowth mechanism, and thus it may be considered to be discontinuous in nature. At temperatures of 150–300 1C, recrystallization
2 The process of nucleation of the recrystallized is considered in detail in Section 3.2.
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
183
Fig. 6. Distributions of the main texture components within the EBSD maps: (a) as-cryo-rolled material, or material which was rolled and then annealed at (b) 50 1C, (c) 150 1C, (d) 300 1C, (e) 400 1C, or (f) 950 1C. In the maps, black and gray lines depict HABs and Σ3 twin boundaries (within 51 tolerance), respectively; RD, ND and TD are the rolling, normal and transverse directions, respectively. In (f), three isolated EBSD maps ware merged together to improve the statistics. Note the different magnifications. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
was nearly complete during the 1-h annealing period and therefore the material exhibited limited change in strength (Fig. 1), microstructure morphology (Fig. 2c, d), grain size (Fig. 3a), misorientation distribution (Fig. 3b), and texture (Fig. 5a–c). At temperatures of 350 1C and higher, microstructure evolution was dominated by grain growth. The initiation of this process at 350 1C can be surmised to be responsible for the abrupt softening found at this temperature (Fig. 1). Progressive grain growth with temperature led to a drastic coarsening of the grain structure (Fig. 2e, f) and a fundamental alteration of the texture (Figs. 4d, e and 5). The latter effect was mainly related to the abnormal growth of the Brass and Goss grains (Fig. 6e, f). Their preferential development provided considerable texture strengthening (Fig. 5a) and a concomitant increase in the proportion of LABs (Fig. 3b). On the other hand, the muted material strengthening effect observed above 700 1C (Fig. 1) is not clear. According to Garcia et al. [19], it may be associated with the precipitation of Cu2O particles, but this hypothesis requires experimental verification. 3.2. Recrystallization More detailed insight into the recrystallization process was obtained from cryogenically-rolled material which was isothermally annealed at 150 1C for various times ranging from 1 min to 10 h, as well as an additional specimen which had been quenched immediately upon reaching 150 1C (i.e., a heat-up time of approximately
2 min). The evolution of microhardness, microstructure, and texture during these isothermal experiments is summarized in this section.
3.2.1. Broad aspects of recrystallization The effect of heat treatment at 150 1C on microhardness is shown in Fig. 7a. One of the most striking features was the drastic softening from 185 to 120 Hv due solely to the heat-up to 150 1C, thus indicating that the material experienced substantial microstructural changes even prior to any of the isothermal hold times. During the hold at temperature, the microhardness rapidly decreased further to 70–80 Hv tending to saturate after 4 min at temperature. EBSD measurements confirmed the marked instability of the cryo-rolled material. The recrystallized fraction in material simply heated to 150 1C was determined to be 77 pct. (Figs. 8a and 7b). Therefore, the incubation period for recrystallization was very short, if any. The recrystallized fraction grew rapidly with time; it exceeded 95 pct. after 8 min at temperature (Figs. 8a–d and 7b). On the other hand, isolated “islands” on unrecrystallized material survived even after 10 h of annealing; an example is circled in Fig. 8f. Interestingly, the coarse recrystallized grains were often elongated in the RD (Fig. 8). As expected, the size of the recrystallized grains did not change significantly during the isothermal-hold period, which represented a very late stage of recrystallization as noted above (Fig. 9a). Moreover,
184
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Fig. 7. Effect of heat treatment time at 150 1C on (a) microhardness and (b) recrystallized volume fraction.
Fig. 8. Selected portions of low-resolution grain-boundary EBSD maps showing the grain structure which developed (a) after heating to 150 1C, and subsequent annealing at this temperature for times of (b) 1 min, (c) 2 min, (d) 8 min, (e) 32 min, and (f) 10 h. In the maps, red and black lines depict LABs and HABs, respectively. For simplicity, Σ3 twin boundaries (within 51 tolerance) were removed from the maps; RD, ND and TD are rolling, normal and transverse directions, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
the grain-size distribution (GSD) of the recrystallized material was relatively broad with grain sizes between 1 and 40 μm (Fig. 9a). The recrystallized texture was also stable (Fig. 9b). In agreement with Figs. 5b, c, it consisted primarily of (50; 30/60;0), BrassR, Dillamore and Cube-RD orientations. The fractions of Brass and
Goss grains were small (Fig. 9b). On the other hand, the crystallographic orientation of the unrecrystallized areas was dominated by the Brass component (not shown). These observations suggested that recrystallization was very sluggish in the Brass and Goss grains.
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
185
Fig. 9. Effect of annealing time at 150 1C on the (a) grain-size distributions and (b) volume fractions of texture components of recrystallized material.
Fig. 10. High-resolution EBSD analysis of apparently-unrecrystallized material revealing the microstructure morphology at (a) low and (b) high magnifications as well as (c) the evolution of misorientation-angle distribution relative to the as-cryo-rolled state. In (a) and (b), red, black and gray lines depict LABs, HABs and Σ3 boundaries (within 51 tolerance), respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
3.2.2. Nucleation of recrystallization 3.2.2.1. Microstructure fragmentation. As shown in Section 3.2.1, the cryo-rolled material underwent substantial recrystallization even during the heating stage, and thus the details of the onset of recrystallization were not available for analysis. To help overcome this difficulty (at least partially), unrecrystallized areas (indicated in red in Fig. 8) were examined. In particular, the specimen exposed 1 min at 150 1C was used for this purpose. A high-resolution EBSD map of the unrecrystallized material is shown in Fig. 10a; a section of the map at higher magnification is shown in Fig. 10b. These maps revealed that a nearly-equiaxed grain structure predominated in the material. It seems therefore that the pancake-shaped cryo-rolled grains started to fragment prior to isothermal-recrystallization heat treatment. It was also evident that the fragmented grains were often free of LABs and typically contained annealing twins. In other words, they were
essentially recrystallized already. If so, the recrystallized fraction shown in Fig. 7b was underestimated. In principle, the fragmentation of pancake-shaped grains may have occurred via the transformation of transverse LABs into HABs or by thermally-induced impingement of mobile longitudinal grain-boundary bulges (i.e. so-called geometrical recrystallization [18]). An example of geometrical recrystallization is circled in Fig. 10b. To examine these two possibilities further, the evolution of the misorientation-angle distribution (Fig. 10c) was further examined. If the LAB-to-HAB transformation mechanism had predominated, a substantial increase in the area of grain boundaries with moderate misorientation ( 10–201) would have been expected. This was not observed experimentally, however; i.e., the area of boundaries with low and moderate misorientation was actually significantly reduced (Fig. 10c). Hence, the observed effect may be explained by the elimination of the LABs by migrating
186
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Table 2 Evaluation of stored energy in cryo-rolled material and in the apparently unrecrystallized material annealed at 150 1C for 1 min. Texture component Number of measured grains
Brass Goss Brass-R Dillamore (55;30/60;0) Cube-RD Cube
Mean grain orientation spread (deg)
Cryo-rolled
Annealed
Cryo-rolled
Annealed
1391 750 202 235 204 438 97
1536 657 326 212 145 551 195
2.1 2.1 1.6 1.7 1.5 1.6 1.5
2.3 2.1 1.2 0.9 0.9 1.8 1.6
Table 3 Grain-structure characteristics for different texture components in apparently unrecrystallized material annealed at 150 1C for 1 min. Texture component Length of 401 〈111〉 boundaries Equivalent grain diameter excluding per unit of grain area within twins (μm) a 10-deg tolerance (μm 1)
Brass Goss Brass-R Dillamore (55;30/60;0) Cube-RD Cube
0.319 0.420 0.595 0.589 0.367 0.521 0.513
Modal
Maximum
1.4 1.3 1.1 1.4 1.3 1.3 1.8
6.4 3.1 2.5 2.7 3.9 3.9 6.1
Note: The main recrystallization textures are highlighted in gray. Note: The main recrystallization textures are highlighted in gray.
grain boundaries, and thus it appears that the fragmentation process was driven mainly by the geometrical-recrystallization mechanism. This mechanism is believed to be very sensitive to local variations of dislocation density as well as grain-boundary mobility and thus the recrystallization process should develop highly inhomogeneously. This may be one of the possible reasons for the relatively broad GSD in the recrystallized material seen in Fig. 9a. 3.2.2.2. Recrystallization nuclei. According to the “oriented nucleation” theory, recrystallization nuclei are characterized by relatively-low stored energy [18]. To assess the applicability of this theory, it was necessary to estimate the distribution of stored energy in the deformed material (and hence potential nuclei) as well as its variation during annealing. For this purpose, the grain orientation spread (GOS) within different texture components in the cryo-rolled as well as fragmented materials was measured. The GOS approach quantifies the geometrically-necessary component of a dislocation array and thus may provide a broad indication of the stored energy. To measure the grain orientation spread, an average misorientation between each pixel of a grain and the average orientation of the grain was calculated. In the case of the fragmented material, the grain orientation spread was measured only for grains larger than 1 μm. In smaller grains, the surface energy likely exceeds the energy of the eliminated dislocations, and thus these grains are probably unable to serve as recrystallization nuclei [18]. The measurements (Table 2) revealed that the Brass and Goss texture components exhibited by the largest GOS values in both the as-cryo-rolled and the annealed conditions. It may thus be hypothesized that these two orientations were characterized by relatively large stored energy, thus in line with the experimentally-observed disappearance of these textural components during recrystallization (Figs. 5b and 9b). However, its fundamental reason is not clear and requires additional work. Because the recrystallization texture was dominated by the (55;30/60;0), Brass-R, and Dillamore orientations (Figs. 5c, 9b), these textural components were of particular interest. In the ascryo-rolled state, these grains were characterized by relatively-low values of GOS, and thus likely served as the recrystallization nuclei. Moreover, the GOS values were reduced substantially during annealing (Table 2). Assuming that pancake-grain fragmentation occurred via geometrical recrystallization, these trends may be concluded to be related to the elimination of dislocations by migrating grain boundaries. However, it is not evident from Table 2 why the (55;30/60;0) grains predominated over the Brass-R and Dillamore orientations in the recrystallization texture (Figs. 5c and 9b). This effect may be related to the specific spatial distribution of the (55;30/60;0) grains. Specifically, this
Fig. 11. Effect of annealing time at 450 1C on microhardness.
orientation is relatively close to Brass (Fig. 4b). Thus, it is possible that the (55;30/60;0) grains originated from orientation spread of the Brass component. If so, the (55;30/60;0) grains would have been frequently surrounded by high-energy Brass grains, thereby promoting their growth The Cube-RD and Cube grains were also characterized by relatively low values of GOS in the as-cryo-rolled state (Table 2), whose magnitude, however, was not reduced in the fragmented microstructure. Thus, the rate of recovery of these orientations appears to have been relatively slow. This observation was also in good agreement with the very modest contribution of these two orientations to the recrystallization texture (Figs. 5b and 9b), but the physical basis for this behavior is also unclear. Furthermore, the slight increase observed in the Cube-RD component (Fig. 9b) is thought to be related to the formation of annealing twins in the (55;30/60;0) grains, examples of which are circled in Fig. 6b. According to the “oriented growth” theory, the selective growth of the recrystallization nuclei may also be related to highly mobile 401〈111〉 boundaries [18]. To examine this possibility, the total length of these boundaries (per unit area of the EBSD map) of the different texture components was measured (Table 3). The dominant recrystallization-texture component (55;30/60;0) was characterized by one of the lowest specific lengths of 401〈111〉 boundaries. It is therefore unlikely that oriented growth played a significant role in the recrystallization process in the present work. Additionally, the preferential development of the nuclei did not appear to be related with their relatively large size (Table 3).
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
187
Fig. 12. Selected portions of grain-boundary EBSD maps showing the grain structure which developed (a) after heating to 450 1C, and subsequent annealing at this temperature for (b) 1 min, (c) 4 min, (d) 8 min, (e) 32 min, and (f) 10 h. In the maps, red and black lines depict LABs and HABs, respectively. For simplicity, Σ3 twin boundaries (within 51 tolerance) were removed from the maps; RD, ND and TD denote the rolling, normal and transverse directions, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
It seems therefore that the recrystallization was mainly driven by the oriented-nucleation effect.
3.2.3. Summary Grain boundaries in the cryo-deformed copper seem to be very mobile, being prone to migrate even at relatively low temperatures. Due to the pinning effect associated with low-mobility LABs, however, grain-boundary migration degenerated into local grain boundary bulging which fragmented the elongated rolled grains into lower-aspect-ratio grain segments. An important consequence of this process was a reduction in stored energy within the grains. This process was deduced to occur most rapidly in grains having orientations close to (55;30/60;0), Brass-R, and Dillamore thus providing them a growth advantage. Due to the small fraction of these components in the deformed material, their preferential development led to discontinuous recrystallization. Recrystallization occurred very rapidly. The cryo-deformed material was found to recrystallize to a large extent during heat-up to 150 1C. On the other hand, Brass-oriented grains were found to be very resistant to recrystallization, and thus isolated islands of the Brass grains survived even after long-term heat treatment at 150 1C. Finally, the grain-size distribution of the recrystallized grains was relatively broad with sizes ranging from 1 to 40 μm.
3.3. Grain growth More detailed insight into the grain-growth process was obtained from the cryogenically-rolled material which was isothermally annealed at 450 1C for times ranging from 1 min to 10 h. Again, one specimen was immediately quenched after reaching the soak temperature; the heating time in this case was 30 s.3 3.3.1. Microstructure The effect of heat treatment time at 450 1C on microhardness and microstructure morphology is shown in Figs. 11 and 12, respectively. It is clear that the cryo-rolled material was completely recrystallized during the heat-up to the annealing temperature (Figs. 11 and 12a). With further time at 450 1C, the grain structure coarsened substantially (Fig. 12), thereby leading to a further, moderate degree of softening (Fig. 11). However, softening tended to saturate after 4 min at 450 1C (Fig. 11). The grain structures developed during heat treatment at 450 1C were noticeably bimodal (Fig. 12b–d) thereby suggesting that grain growth was abnormal in nature. Nevertheless, the GSDs appeared to have a single peak (Fig. 13a). Therefore, to quantify further the 3 The different heating time during isothermal annealing at 150 and 450 1C (i.e. 2 min and 30 s) was caused by different furnaces used for these experiments.
188
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Fig. 13. Effect of annealing time at 450 1C on grain-size frequency distributions as a function of (a) circle-equivalent grain diameter or (b) the logarithm of D/Davg. (Davg. ¼ average grain diameter).
Fig. 15. Effect of annealing time at 450 1C on volume fractions of the main texture components. Fig. 14. Effect of annealing time at 450 1C on grain boundary length for different types of boundaries (measured in an EBSD map) normalized by the area of the EBSD maps.
nature of grain growth, normalized GSDs were calculated (Fig. 13b). It was found that the normalized distributions changed significantly with annealing time thus confirming the abnormal character of growth. Moreover, all of the GSDs were distinctly different from log-normal ones, being skewed toward larger grain sizes. Due to the large spread in grain size, the evolution of the mean grain size with annealing time was concluded not to be a good indicator of grain-growth kinetics. Instead, grain-boundary length per unit area of the EBSD map was determined (Fig. 14). In such a plot, the normalized length of HABs (excluding annealing twin boundaries) was reduced by a factor of 3.5 during the beginning of annealing; this indicated rapid grain growth. However, the grain boundary length tended to stabilize after 10 min, suggesting a saturation of the grain-growth process. On the other hand, the area of LABs and Σ3 boundaries was found to change relatively little during annealing (Fig. 14).
3.3.2. Texture The effect of annealing time on the volume fractions of the main texture components is shown in Fig. 15. Similar to the findings for the recrystallization process, the principal graingrowth texture components were (55;30/60;0), Brass-R, and Dillamore, whereas the fraction of Cube-RD and Cube components was relatively low (compare Figs. 9a and 15). A distinctive feature
of grain growth, however, was a substantial strengthening of the Brass and Goss components (Fig. 15). As shown in the previous section, the cryo-rolled material was completely recrystallized during the heat-up stage (Figs. 11 and 12a), and the LAB fraction was very low (Fig. 14). According to the Humphrey′s cellular model, abnormal grain growth and its associated significant texture change are likely to be associated with a large spread in grain size [18]. In this regard, GSDs were measured for each of the principal texture components; the results are summarized in Table 4. As expected, the maximum grain size for the Cube and Cube-RD orientations was substantially smaller than that of the (55;30/ 60;0), Brass-R, and Dillamore components. As shown in Section 3.2, the latter three orientations had a growth advantage during recrystallization and thus could achieve relatively large sizes. Because the grain-size spread in the as-heated material was also very large (Fig. 13), the tendency for abnormal grain growth would have been exacerbated. On the other hand, the predominant (55;30/60;0) grains would often impinge upon each other during growth thus giving rise to LABs. This “orientation pinning” effect would gradually suppress the development of these orientations [9], as was observed (Fig. 15). The abnormal growth of Brass and Goss grains was striking (Fig. 15). As shown in Section 3.2, these orientations were fairly resistant to recrystallization, being stable even after a 10-h anneal at 150 1C. It may be hypothesized that some portion of these long as-rolled grains survived (but recovered) during the heat-up stage and thus underwent subsequent abnormal growth. Per the results in Table 4, the as-heated material did indeed contain relatively coarse Brass grains. On the other hand, the Goss grains were
T. Konkova et al. / Materials Science & Engineering A 585 (2013) 178–189
Table 4 Grain-structure characteristics for different texture components developed after heating cryo-rolled material to 450 1C. Texture component
Cube-RD Cube Brass Goss Dillamore Brass-R (55;30/60;0)
Number of measured grains
Grain size Mean Maximum
Length of 401 〈111〉 boundaries per unit of grain area within a 10-deg tolerance (μm 1)
2737 840 1212 380 3323 2164 3355
4.2 4.0 3.7 3.8 5.5 4.5 4.8
0.063 0.078 0.073 0.074 0.053 0.055 0.065
17.8 14.2 27.5 17.8 37.5 22.3 33.6
Note: The main grain-growth textures are highlighted in gray.
relatively small, and thus the reason for the abnormal growth of this orientation was not clear. The specific grain-boundary area of highly-mobile 401〈111〉 boundaries was also measured in the as-heated condition at 450 1C (Table 4). This area did not vary significantly among the various texture components, and thus abnormal grain growth was not likely associated with an oriented-growth effect. 4. Summary and conclusions The annealing behavior of cryogenically-rolled copper was quantified. To this end, the material was rolled to a 90-pct. thickness reduction at liquid-nitrogen temperature and then isochronally annealed at temperatures between 50 and 950 1C for 1 h as well as isothermally for various times at 150 or 450 1C. Grain-structure and texture changes were quantified using an EBSD technique. The main conclusions from this work are as follows: (1) The material had very poor microstructure stability. Gross evidence of recrystallization was observed after annealing at 50 1C for 1 h whereas the recrystallization process was almost complete after annealing at 150 1C for 8 min. (2) Microstructure evolution at temperatures below 350 1C ( 0.5Tm) may be interpreted in the terms of discontinuous static recrystallization. Grains having crystallographic orientations close to (55;30/60;0), Brass-R, and Dillamore were deduced to recover most rapidly, thus providing them with a subsequent growth
189
advantage. On the other hand, the Brass and Goss orientations were found to be resistant to recrystallization. The grain-size distribution of the recrystallized grains was relatively broad, spanning 1 to 40 μm. (3) At temperatures above 300 1C, the annealing behavior was found to be driven primarily by abnormal grain growth. The abnormal nature of this process was related to the relatively large grain-size spread produced during the preceding recrystallization.
Acknowledgments Financial support from the Russian Fund of Fundamental Research (Project no.12-08-97008) is gratefully acknowledged. The authors would like to thank Professor G.A. Salishchev for suggesting this research. They are also very grateful to Dr. R.M. Galeyev and Dr. O.R. Valiakhmetov for graciously providing the material used in this work, P. Klassman and T.I. Nazarova for technical assistance during cryogenic rolling, and Dr. R. R. Kabirov for help in performing the annealing experiments.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]
Y.S. Li, N.R. Tao, K. Lu, Acta Mater. 56 (2008) 230–241. Y. Zhang, N.R. Tao, K. Lu, Acta Mater. 56 (2008) 2429–2440. Y. Huang, P.B. Prangnell, Acta Mater. 56 (2008) 1619–1632. S.K. Panigrahi, R. Jayaganathan, Mater. Sci. Eng. A. 480 (2008) 299–305. S.K. Panigrahi, R. Jayaganathan, V. Chawla, Mater. Lett. 62 (2008) 2626–2629. T. Konkova, S. Mironov, A. Korznikov, S.L. Semiatin, Acta Mater. 58 (2010) 5262–5273. V. Subramanya Sarma, J. Wang, W.W. Jian, A. Kauffmann, H. Conrad, J. Freudenberger, Y.T. Zhu, Mater. Sci. Eng. A 527 (2010) 7624–7630. G.H. Zahid, Y. Huang, P.B. Prangnell, Acta Mater. 57 (2009) 3509–3521. D. Juul Jensen, Acta Mater. 43 (1995) 4117–4129. H.D. Meingelberg, M. Meixner, K. Lucke, Acta Metal. 13 (1965) 835–844. Y.M. Wang, E. Ma, Acta Mater. 52 (2004) 1699–1709. T. Konkova, S. Mironov, A. Korznikov, S.L. Semiatin, Scr. Mater. 63 (2010) 921–924. V. Subramanya Sarma, K. Sivaprasad, D. Sturm, M. Heilmaier, Mater. Sci. Eng. A 489 (2008) 253–258. S.V. Dobatkin, G.A. Salischev, A.A. Kuznetsov, T.N. Konkova, Mater. Sci. Forum 558–559 (2007) 189–194. F.J. Humphreys, J. Microsc. 195 (1999) 170–185. T. Konkova, S. Mironov, A. Korznikov, S.L. Semiatin, Mater. Sci. Eng. A 528 (2011) 7432–7443. J. Pospiech, K. Lucke, Acta Mater. 23 (1975) 997–1007. F.J. Humphreys, M. Hatherly, Recrystallization and Related Phenomena, second ed., Elsevier, Oxford, 2004. V.G. Garcia, J.M. Cabrera, J.M. Prado, Mater. Sci. Eng. A 488 (2008) 92–101.