Report

Superplastic Extensibility of Nanocrystalline Copper at Room Temperature

See allHide authors and affiliations

Science  25 Feb 2000:
Vol. 287, Issue 5457, pp. 1463-1466
DOI: 10.1126/science.287.5457.1463

Abstract

A bulk nanocrystalline (nc) pure copper with high purity and high density was synthesized by electrodeposition. An extreme extensibility (elongation exceeds 5000%) without a strain hardening effect was observed when the nc copper specimen was rolled at room temperature. Microstructure analysis suggests that the superplastic extensibility of the nc copper originates from a deformation mechanism dominated by grain boundary activities rather than lattice dislocation, which is also supported by tensile creep studies at room temperature. This behavior demonstrates new possibilities for scientific and technological advancements with nc materials.

Plastic deformation in polycrystalline solids occurs by movement of lattice dislocations and/or diffusional creep. The dislocation movement mechanism predominates in most conventional materials at relatively low temperatures when atomic diffusivity is minor. A strain-hardening effect is always associated with this mechanism, which arises from the pileup of lattice dislocations blocked at grain boundaries. The strain-hardening effect often restricts the mechanical processing of metals, which must be eliminated by thermal annealing in order to perform further deformation without crack formation. Plastic deformation occurs at higher temperatures when atomic diffusivity is considerably increased (at grain boundaries or inside the lattice); in this case, diffusion creep becomes a dominant mechanism. The diffusion creep rate (ɛ˙), which is dominated by grain boundary diffusion, is related to grain size by (1) Embedded Imagewhere σ is tensile stress, Ω is atomic volume, d is average grain size, B is a numerical constant, D gb is grain boundary diffusivity, δ is grain boundary thickness, and k is Boltzmann's constant. This relation suggests that diffusional creep may be enhanced when grain size is reduced and/or grain boundary diffusivity is enhanced. Based on this mechanism, a decade ago Gleiteret al. (2, 3) predicted that a nanocrystalline (nc) material, one in which the crystallite size is in the nanometer regime, would make it amenable to high creep rates and large-scale deformation at much lower homologous temperatures, so that ductile ceramics and diffusional creep of pure metals would be possible even at room temperature (3). This possibility is of great interest because not only may it shed light on the deformation mechanism of the nanostructure, but it also may significantly facilitate the fabrication process for producing components with complex shapes. Such expectations are supported by recent computer simulation results, which indicate that plastic deformation of nc metals occurs by grain boundary sliding and motion, with minor lattice dislocation activity (4, 5). The observed deformation mode is in some ways similar to the way grain boundaries carry most of the deformation in superplasticity (6). A softening with grain size (decreasing) in nc Cu is also seen in the simulation as a result of the dominant grain boundary deformation mechanism (4).

However, recent experimental observations are disappointing: Most nc metal samples are very brittle, with little tensile ductility and an extremely small diffusional creep rate at room temperature (7–9). Although softening of nc metals was detected from hardness measurements in several metal and alloy systems, directly measured creep rates in the consolidated nc metals at room temperature are normally two to four orders of magnitude smaller than the expected values (10). Depression in superplastic temperatures recently has been observed in some nc metals and alloys (11), but the expected room-temperature superplasticity in nc metals has not been observed until now. Many of these disappointing experimental results can be attributed to artifacts, which are present in nc samples that were induced during sample processing procedures, such as imperfect bonding and/or porosity, contamination, and large microstrain. To eliminate these factors, which influence the deformation behavior and reveal the intrinsic mechanical properties of nc metals, “ideal” nc samples are urgently needed—that is, samples with pure (free of contamination) and dense (perfect bonding of crystallites and porosity-free) nc samples with a minimized defect density (microstrain).

Among the various synthesis methods of nc materials, electrodeposition is one of the effective approaches for producing porosity-free nc metals with a small microstrain (12). In this study, a nc pure Cu (purity better than 99.993 atomic % and a total oxygen content of about 24 ± 1 parts per million) was prepared by an electrodeposition technique with an electrolyte of CuSO4 and a Ti substrate. The sample density is 8.91 ± 0.03 g/cm3, or 99.4% of the theoretical density of Cu. Positron annihilation spectroscopy measurements verified the high density and indicated no vacancy-cluster–like volumes or missing crystallites (pores) existing in the as-deposited sample. The average grain size determined from x-ray diffraction (XRD) analysis in terms of diffraction line broadening of seven single Bragg reflection peaks [by the Wilson method (13)] is about 28 nm. The microstrain in the as-deposited sample determined by XRD is only 0.03%, which is about one order of magnitude lower than that in the ball-milled and consolidated nc metals. Microstructure characterization by means of high-resolution transmission electron microscopy (HRTEM) observation indicated that the as-deposited sample consists of ultrafine crystallites (or crystalline domains) with sizes ranging from a few nanometers to about 80 nm. The average grain size is about 20 nm, which is close to the XRD result. Most of the nanometer-sized crystallites are equiaxed and separated by small-angle grain boundaries that consist of dislocation arrays, in contrast to the conventional large-angle grain boundaries in consolidated and ball-milled nc metals. Figure 1 shows a typical HRTEM observation of a few nanocrystallites; between them are small-angle (1° to 10°) grain boundaries.

Figure 1

HRTEM observation of a few nanocrystallites in the as-deposited nc Cu sample. Four crystallites (A, B, C, and D) are outlined by dashed white lines. The orientation difference angle of (111) plan between A and B is about 1°, that between B and C is about 9°, and that between A and D is about 10°. Most nanocrystallites in the as-deposited sample are separated by the small-angle (1° to 10°) grain boundaries.

Plastic deformation of the as-deposited nc Cu specimen was done by cold rolling at room temperature. A piece of the nc Cu sample (16 × 4 × 1 mm) cut from the as-deposited sheet was rolled at room temperature with a twin-roller (diameter, 40 mm) apparatus. The strain rate during rolling was controlled to be around 1 × 10−3 to 1 × 10−2 s−1. Cold rolling resulted in a continuous increase in the sample length in the direction of rolling, while the sample width was kept unchanged. After repeated rolling, the nc Cu sample became a strip with no cracks forming at the sample edges. With the progress of repeated rolling, the sample became longer and longer, as shown in Fig. 2, and eventually resulted in a uniform, long thin ribbon (about 20 μm thick) with smooth surfaces and straight edges. The total degree of deformation, ε = (t 0t)/t, in our samples (where t 0 and t denote thickness of the initial and the as-rolled samples, respectively) can be as large as 5100%. Further cold rolling is still possible to extend the deformation. Such extreme extensibility of the nc Cu sample has not been observed in the conventional coarse-grained polycrystalline Cu, which typically would break after extensions of 800%.

Figure 2

The nc Cu specimens before and after cold rolling at room temperature with different amounts of deformation degree (ɛ) as indicated.

The nc Cu was annealed at 500°C for 48 hours in a vacuum, which resulted in a much larger grain size of about 100 μm. The as-annealed Cu sample was rolled under the same circumstances. It was found that, when the deformation degree reaches about 700%, the sample is remarkably hardened and cracks are formed at sample edges, as usually observed for conventional coarse-grained Cu. This observation implies that the extreme extensibility in the as-deposited Cu originates from the ultrafine-grained structures, and other effects, including the purity effect, can be ruled out.

Microhardness measurements showed a slight hardness increase in the as-deposited nc Cu induced at the initial stage of rolling, from 1.05 GPa (as-deposited) to 1.20 GPa (ɛ = 800%), as shown in Fig. 3, and no strain hardening effect was seen with further rolling (ɛ > 800%). This result contrasts with the evident strain hardening effect in conventional coarse-grained Cu from 0.6 GPa (initial state) to 1.57 GPa (ɛ = 800%, with crack formation), and in the as-annealed Cu from 0.51 GPa (initial state) to 1.45 GPa (ɛ = 650%, with crack formation), also in Fig. 3, respectively. The hardness variation tendencies differ fundamentally between the nc and the coarse-grained Cu samples, which indicates different underlying deformation mechanisms.

Figure 3

Microhardness variation with deformation degree of cold rolling for the as-deposited nc Cu specimen, the conventional coarse-grained polycrystalline Cu, and the as-annealed coarse-grained Cu as indicated. Cracks were formed at sample edges when ɛ ≈ 800% for the coarse-grained Cu sample.

HRTEM observations showed that, in the as-rolled nc Cu samples with a deformation degree of 4800%, there is no obvious grain size difference as in the as-deposited nc Cu sample, and the crystallites are still roughly equiaxed. However, the dislocation density (mainly located at grain boundaries) in the as-rolled specimens increased, which led to an evident increase in the grain boundary misorientation angles between the nanocrystallites, as shown in Fig. 4. The misorientation between neighboring Cu nanocrystallites can be estimated from the observed moiré finger modulations. We found that, for the cold-rolled nc sample, the misorientation angle of the nanocrystallites ranges from about 6° to 18°, which is evidently larger than that for the as-deposited sample (1° to 10°). That means the nanocrystallites (or domains) separated by small-angle grain boundaries in the as-deposited nc Cu may transform into grains (or domains) with increased misorientations during the cold-rolling process. Quantitative XRD measurements confirmed that the average grain size of the nc Cu remained unchanged during rolling, whereas the microstrain (which is a signature of the defect/dislocation density) increased in the initial stage (ɛ < 1000%) and tended to saturate at a value of about 0.16% when ɛ ≥ 1000% (seeFig. 5). Such variation of microstrain in the nc Cu shows a completely different behavior from that of the as-annealed coarse-grained Cu (Fig. 5).

Figure 4

HRTEM observation of the as-rolled nc Cu sample with a deformation degree of 4800%.

Figure 5

XRD measurement results of the microstrain (open circles) and the average grain size (open squares) for the nc Cu samples and the microstrain for the as-annealed coarse-grained Cu sample (solid circles) as a function of deformation degree of cold rolling.

The variation tendency of microstrain agrees very well with the hardness measurement results, indicating that when the as-deposited nc Cu is rolled, a minor dislocation activity (generation and movement) may occur in the initial stage, presumably in large grains, that corresponds to a slight increment in hardness. This phenomenon was observed in computer simulations of the deformation in a nc Cu (4). Further deformation results in saturation of the lattice dislocating density when the dislocation generation and annihilation are compensated. As soon as the large-angle grain boundary grains are formed, plastic deformation might be carried predominantly by grain boundary activities, such as grain boundary sliding or slip, as seen in other nc materials (14). The constant grain size, constant dislocation density, and constant hardness confirmed that plastic deformation of the nc Cu is dominated by the grain boundary mechanism instead of by dislocation activities.

Tensile creep measurements around room temperature [0.22 to 0.24 melting temperature (T m)] were performed with the as-deposited nc Cu (15). We find evidence of obvious creep near room temperature with threshold stress σ0 that decreases with increasing temperature. The steady state creep rate at 0.23 T m was measured to be as large as 1 × 10−6 s−1, which is much larger than that for coarse-grained Cu and the consolidated nc Cu (8 × 10−9 s−1) (10) as well. The creep rates increased linearly with an increasing applied stress:dɛ/dt = 2.51 × 10−14σ (where σ is in Pa and dɛ/dt is in s−1) at 303 K in the as-deposited nc Cu sample. Following the creep rate equation dominated by grain boundary diffusion (i.e., the Coble creep), the predicted value of the creep rate isdɛ/dt = 1.65 × 10−12 σ at 303 K when the diffusivity (D gb) value is taken from a consolidated nc Cu (16). When the diffusivity for the conventional grain boundary was applied, a smaller creep rate was obtained: dɛ/dt = 1.63 × 10−13 σ. This value agrees with our observed creep rate in the present nc Cu when the special (small-angle) grain boundary in our sample is taken into account.

However, even though the observed deformation behavior in this sample can be qualitatively or semiquantitatively interpreted by the grain boundary diffusion creep model, other possible deformation mechanisms could not be ruled out at this stage—for example, the grain boundary sliding model based on the rate-limiting step of deformations (17). The intrinsic deformation mechanism for the nc Cu sample should be clarified by more quantitative experimental results and theoretical analysis.

Our success in obtaining such superplastic extensibility at room temperature in the nc Cu sample might be attributed to the minimal artifacts in the nc sample. The high purity and full density of this nc Cu eliminated the effects of contamination and imperfect bonding (or porosity), which normally exist in consolidated nc specimens and which are effective in restricting the movement of grain boundaries and hence resisting deformation. High-angle random grain boundaries may slide more readily than the small-angle boundaries or special, such as low σ, boundaries. However, as seen in the present nc Cu, creep behavior around room temperature is evident. It indicates that the small-angle grain boundaries remain effective in carrying out large deformations at room temperature when the grain size is ultrafine, as the creep rate is proportional to d −3. It is also seen that, when the deformation degree exceeds 1000%, with which the grain boundary misorientation has been increased, the grain boundary sliding or slip became more predominant in deformation and no strain hardening effect was detected.

The traditional deformation-annealingdeformation technique that is routinely used for deforming metals may be much simplified by using nc metals as starting materials. With proper post-heat treatments to the deformed nc metals, the microstructure of the metal may be easily controlled by optimizing the grain growth kinetics so that novel properties and behaviors in the final products can be achieved. Such an innovative material processing technique by means of superplastic deformation of nc metals will be useful in modern industry in fields such as micromachining, nanotechnology, and electronics.

  • * To whom correspondence should be addressed. E-mail: kelu{at}imr.ac.cn

REFERENCES AND NOTES

View Abstract

Stay Connected to Science

Navigate This Article