Orientation-controlled nanotwinned copper prepared by electrodeposition

Orientation-controlled nanotwinned copper prepared by electrodeposition

Electrochimica Acta 178 (2015) 458–467 Contents lists available at ScienceDirect Electrochimica Acta journal homepage: www.elsevier.com/locate/elect...

3MB Sizes 27 Downloads 169 Views

Electrochimica Acta 178 (2015) 458–467

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Orientation-controlled nanotwinned copper prepared by electrodeposition Madoka Hasegawaa,* , Maxime Mieszalaa , Yucheng Zhanga,b , Rolf Ernib , Johann Michlera , Laetitia Philippea a Empa - Swiss Federal Laboratories for Materials Science and Technology, Laboratory for Mechanics of Materials and Nanostructures, Feuerwerkerstrasse 39, 3602 Thun, Switzerland b Empa - Swiss Federal Laboratories for Materials Science and Technology, Electron Microscopy Center, Überlandstrasse 129, 8600 Dübendorf, Switzerland

A R T I C L E I N F O

A B S T R A C T

Article history: Received 5 May 2015 Received in revised form 15 July 2015 Accepted 4 August 2015 Available online 7 August 2015

Highly-textured nanotwinned Cu (nt-Cu) films with different twin orientations were prepared by pulse electrodeposition in an acidic Cu sulfate bath containing a combination of conventional bath additives. We found that the orientation of nanotwins is dependent on the deposition potential. Copper deposits with horizontally oriented nanotwins were formed at an on-potential of -0.2 V vs. saturated calomel electrode (SCE), while deposits with vertically oriented nanotwins are obtained at -0.6 V vs. SCE. For horizontally oriented nt-Cu, the twin spacing was found to decrease with an increase of pulse-off time, accompanied by a decrease in the number of stacking faults that are vertical or inclined to the horizontal direction. The findings offer the possibility of more flexible engineering of nanotwin structure in Cu electrodeposits. ã 2015 Elsevier Ltd. All rights reserved.

Keywords: Nanotwinned Cu pulse electrodeposition twin boundary engineering

1. Introduction Mechanical properties of metals are highly dependent on their microstructure such as grain size, crystal texture, crystal imperfections, and impurities. On the one hand, the designing and engineering of the grain size has been a main strategy for a long time. Many studies of nanocrystalline metals showed that the grain boundaries which impede the glide of dislocation, leading to a strengthening of material [1–5]. Alloying of metal with solid solution elements also offers the strengthening of metals [6–9]. However, these solutions unfortunately increase the electrical resistivity. On the other hand, nanotwins, which are often observed in metals and alloys of low and medium stacking-fault-energy elements, are considered to be one of the most promising strengthening methods without losing a good electrical resistivity. It has been shown that nanoscale twin boundaries restrict the dislocation motion during plastic deformation and thereby nanotwins in metals improve the strength analogous to the grain refinement strengthening [10,11]. Importantly, unlike the grain boundaries, nanoscale twins do not significantly deteriorate the

* Corresponding author. E-mail addresses: [email protected] (M. Hasegawa), [email protected] (M. Mieszala), [email protected] (Y. Zhang), [email protected] (R. Erni), [email protected] (J. Michler), [email protected] (L. Philippe). http://dx.doi.org/10.1016/j.electacta.2015.08.022 0013-4686/ ã 2015 Elsevier Ltd. All rights reserved.

electrical conductivity because the electron scattering of twin boundary is about one order of magnitude less than that of incoherent high-angle grain boundaries, [12] which is highly desirable especially for applications in electronic devices. Nanotwinned Cu (nt-Cu) has been investigated intensively by many researchers mainly due to the interests for the application to interconnects in advanced electronics [13]. The experimental and theoretical studies [14–17] have shown that well-ordered twin boundaries with their spacing of 100 nm or less act as a perfect interface for strengthening of materials. It has been shown that the twin spacing determines the strength of nt-Cu, following a HallPetch-type relationship [11,18,19]. The anisotropy in mechanical properties of highly-oriented nanotwins has been discussed by some researchers [20–22]. The previous studies also revealed the outstanding properties of nt-Cu: not only for mechanical and electrical properties but also electromigration resistance, [23] which is extremely important for sub-micrometer scale interconnects application. Recently, highly (111) textured nt-Cu, where dense nanotwins are preferentially oriented in the horizontal direction within columnar grains, has attracted many researchers’ attention as a model material to investigate the deformation mechanism of twinned metals [11,13,18,20,24–27]. Highly-textured nt-Cu has been synthesized either by electrodeposition [13,14,27] or by sputtering [24,28,29]. By employing electrodeposition techniques, nt-Cu with various dimensions including thin films [13,20,27] and nanopillars [14,17] and nanowires, [26,30] has

M. Hasegawa et al. / Electrochimica Acta 178 (2015) 458–467

been reported so far. These nt-Cu films and structures were achieved by employing different electroplating parameters as well as plating baths. Liu and co-workers, [27] who reported the nt-Cu prepared by DC electrodeposition in a copper sulfate bath containing some additives, claimed that highly (111)-textured nanotwin formation is highly dependent on the agitation speed of the plating bath and the current density. Pulse electrodeposition is another choice to produce (111)-textured nt-Cu electrodeposits [13,14,17,26,31]. Jang and coworkers, [14] who studied the mechanical properties of nt-Cu nanopillars fabricated by pulse plating, reported that twin spacing can be adjusted by the length of on-time, although the detailed information was not provided. Chan and coworkers [30] reported that the density of nanotwins increases when pulse plating is carried out at a low current density and at a low temperature. As described above, highly textured nt-Cu films and nanostructures have been fabricated using various electrodeposition methods and conditions by many different groups. However, the very limited information is available in terms of the mechanism of the well-oriented nanotwin formation, despite the intensive studies of their mechanics. The growth twins formed in randomly oriented grains have been classically observed in many electrodeposits, especially produced by pulse electrodeposition techniques. The twinning in electrodeposits have been believed to be related to the relaxation of the internal stress, which is generated due to film growth phenomena such as nucleation, coalescence of nuclei and thickening of the film. Xu and coworkers [32,33] measured the stress evolution in Cu during pulse electrodeposition in order to explain the twinning in Cu electrodeposits. Their in-situ stress measurements revealed that the stress rapidly increases during pulse-on times, while it is relaxed during the pulse-off times. The stress relaxation is attributed to the rearrangement of Cu atoms in the film to form a more energetically favored structure. Nanotwinning is considered to occur during this relaxation step. Twinning of a highly textured nt-Cu is assumed to occur in a similar manner, and the reduction of Cu ions during the pulse-off time as well as the stress-relaxation process during the pulse-off time should play key roles in controlling both the orientation and the density of twins in columnar grains in electrodeposited films. However, as far as the authors are aware, the synthesis of nt-Cu has been carried out empirically, and the roles of pulse-on/off on nanotwinning of electrodeposits have rarely been discussed in the previous reports. Such fundamental knowledge in nanotwinning mechanism is indeed very important for the more precise engineering of microstructure of nanotwinned metals with a desired property. Therefore, in this study, we investigated the effect of pulse electrodeposition on microstructure of nt-Cu in order to obtain insight into the role of the pulse-on/off times on the twinning behavior, namely, the twin orientation and density of highly-textured nt-Cu films.

459

2. Experimental Section Cu electrodeposition was performed at room temperature on Si substrate sputter-coated with a 100 nm-thick Au layer on a 10 nmthick Cr adhesion layer. The plating bath consist of 65 g/L CuSO45H2O (Sigma-Aldrich) and 196 g/L H2SO4 (Sigma-Aldrich) as main constituents. The bath also contains some conventional bath additives: 50 mg/L chloride ions (added as HCl, SigmaAldrich), 100 mg/L polyethylene glycol (Mw 4000, Sigma-Aldrich) and 10 mg/L 3-Mercapto-1-propanesulfonic acid sodium salt (Sigma-Aldrich). A Pt plated Ti mesh was used as the counter electrode, while a saturated calomel electrode (SCE) was used as the reference electrode. The electroplating cell was connected to a computer-aided potentiostat system (Autolab PGSTAT30, Metrohm) controlled by GPES software. Prior to electrodeposition experiments, substrates were cleaned in Piranha solution (the mixture of H2O2 (30%) and H2SO4 with a mixing ratio of 1:3) for at least 10 min and then thoroughly rinsed in deionized water (18.2 MVcm). Subsequently, the conductive surfaces of the cleaned substrates were covered with a masking tape leaving a circular open area of 1 cm diameter for electrodeposition. All electrodeposition experiments were performed at room temperature. Electrodeposition was carried out by a pulse plating technique, which applies the periodical square wave potential pulses. The potential during on-time was set at -0.2 V or -0.6 V vs. SCE while that for offtime was the open circuit potential (+0.055 V vs. SCE). The on-time was fixed at 20 ms whereas the off-time was varied from 1 to 4 s. After the deposition, samples were rinsed in deionized water and dried under clean air flow. Microstructure of Cu electrodeposits was observed with a focused ion beam (FIB, VELA FIB-SEM, Tescan). Specimens were primarily cleaved and its cross-section was milled by FIB at an acceleration voltage of 30 keV and a probe current of 1000 pA. The cross-sections of samples were observed with a sample tilt angle of 45 with respect to the direction of Ga ion beam. The imaging was carried out with an acceleration voltage of 30 keV and a probe current of 10 pA. Twin spacing was measured from randomly selected around 20 different grains in FIB images of different parts of a sample for each growth condition. Transmission electron microscopy (TEM) was performed for selected samples in order to obtain further insights into the microstructure of Cu deposits. The samples for TEM were prepared by FIB-milling. The TEM samples were prepared by FIB-milling. A JEOL 2200 FX microscope operated at 200 keV was used to perform high-resolution imaging of the samples. The crystal structure of nt-Cu was analyzed by X-ray diffraction (XRD, Bruker) with a Cu Ka radiation source (40 kV, 40 mA). The texture coefficient [34,35] was defined as TC ðhklÞ ¼

IðhklÞ =I0ðhklÞ

ð1=nÞSIðhklÞ =I0ðhklÞ

ð1Þ

, where I(hkl) is the relative peak intensity from (hkl) reflection of a Cu electrodeposit, I0(hkl) is that of randomly oriented powder Cu

Fig. 1. A FIB image of nt-Cu prepared by pulse plating at an on-potential of -0.2 V vs. SCE (ton: 20 ms, toff: 3 s).

460

M. Hasegawa et al. / Electrochimica Acta 178 (2015) 458–467

3. Results and discussion

plane in Cu-(111). In addition to the grain boundary, it can be often observed that the presence of vertically oriented stacking faults interconnecting the twinning boundaries, as shown in Fig. 2c. The nature of the stacking faults is expected to be a twin boundary in the (112) plane, known as an incoherent twin boundary, which plays an important role in the de-twinning of nt-Cu during plastic deformation [15,36,37].

3.1. Horizontally oriented nt-Cu.

3.2. Effect of pulse-off time on nanotwinning.

Fig. 1 shows the cross-sectional FIB-scanning ion images of a Cu film prepared by pulse plating at an on-potential of -0.2 V vs. SCE with an on-time of 20 ms and an off-time of 3 s. The FIB image shows that the deposit consists of columnar grains with 1 to 2 mm in diameter. Channeling contrasts of horizontally aligned lamella patterns, suggesting the existence of nanotwins, were found in most of the grains. A TEM study of this sample confirmed the presence of the horizontally oriented nanotwins, as shown in Fig. 2. Fig. 2a is a low magnification image of the sample cross-section, which reveals a high density of nanotwins with an average spacing of around 100 nm, consistent with the FIB observation. A high-resolution image showing the lattice plane is presented in Fig. 2b. A twin boundary in the (111) plane is clearly seen. It stops at an intersection with a grain boundary. The Fourier transformation of the image shows the reflections of the nanotwin with the mirror

Cu electrodeposition experiments were performed with various off-times in order to obtain insights into formation of horizontally oriented nt-Cu. Fig. 3 compares the cross-sectional FIB images of nt-Cu samples deposited with the various off-times from 1 to 4 s. For these samples, the on-time as well as the on-potential was fixed at 20 ms and -0.2 V vs SCE, respectively. Fig. 3a shows the image of a Cu deposit prepared by potentiostatic electrodeposition at -0.2 V (i.e. toff = 0s). The deposit consists of small grains uniformly distributed across the deposit thickness in the asdeposited state. For such a plating condition, electrodeposits are known to undergo recrystallization even at room temperature [38– 40]. We also observed this recrystallization phenomenon. Hence, the observation of FIB image as well as XRD characterization for this condition was performed within a few hours after preparation in order to avoid the microstructure change due to the roomtemperature recrystallization. It is interesting to note that the

sample (JCPDS, Reference code: 00-004-0836), and n is the reflection number. For this study, TCs of all reflection appeared up to (420) reflection, except for (222) and (400) peaks which are higher order reflections of (111) and (200), are calculated for each sample.

Fig. 2. (a) A low-magnification image of the Cu cross-section showing the horizontally aligned nanotwins. The twin boundaries (TBs) are denoted with the red arrows. (b) A high-resolution image showing the interaction of a twin boundary (TB) with a grain boundary (GB). The inset is the Fourier transformation of the image with the indices of the resolved lattice planes. (c) and (d) show interconnection of TBs by a vertical stacking fault (SF).

M. Hasegawa et al. / Electrochimica Acta 178 (2015) 458–467

461

Fig. 3. (a-e) Cross-sectional FIB images of nt-Cu deposits prepared by pulse plating at an on-potential of -0.2 V vs. SCE (ton: 20 ms) and with different pulse-off times and (f) TChkl of each sample obtained from XRD measurements. (g-i) show the twin spacing distribution in nt-Cu prepared with different pulse-off times.

room-temperature recrystallization did not occur in any samples prepared by pulse electrodeposition, indicating that the introduction of off-times during electrodeposition resulted in a less movable or more stable microstructure in the as-deposited state, clearly showing a significant change in microstructure during pulse-off times. With an off-time of 1s, we observed randomly oriented coarse grains of a few micrometers in the deposit. Although some grains seem to contain some twins, they were oriented in highly random directions. The horizontally oriented nanotwins were observed with an off-time of 2 s and more (Fig. 3ce). With an off-time of 2 s, twins in grains formed with an irregular spacing, indicating that twinning in this condition happened randomly during deposition. Even though their orientation is essentially parallel to the surface, there were also many steps on these twin boundaries are visible in the FIB images, suggesting the presence of different types of twin boundaries and microstructural defects. Interestingly, the density of horizontally oriented twins in grains increased when employing longer off-time periods. With an off-time of 3 s (Fig. 3d), which is a same growth condition as the

sample for TEM observation (Fig. 2), the number of visible steps of horizontally oriented twin boundaries was reduced significantly, although many horizontally twin boundaries do not transverse the grain column and their ends are connected to vertically formed other boundaries, which were confirmed to be stacking faults by the TEM observation (Fig. 2c). The number of the vertical stacking faults tends to decrease with a further increase in the off-time to 4 s (Fig. 3e), and many nanotwins transverse the grain columns in this sample. It should be noted here that such stacking faults are expected to play an important role in the mechanical deformation of nt-Cu. According to the previous reports, [15,36,41] such stacking faults migrate during plastic deformation, leading to the de-twinning of coherent twins in the (111) plane. Hence the density of these stacking faults along with that of coherent nanotwins is a key factor to control the mechanical property. Corresponding to the changes in microstructure of Cu, the crystal structure of film also changed significantly. Fig. 3f shows the texture coefficients (TCs) of Cu samples plotted as a function of the pulse-off time. The plot revealed that samples prepared by

462

M. Hasegawa et al. / Electrochimica Acta 178 (2015) 458–467

pulse electrodeposition were highly (111)-textured regardless of the presence of nanotwins, while the sample prepared without potential pulses was almost polycrystalline. We also found that there is a tendency that the texture coefficient for (111) further increases with an increase in the off-time. XRD profiles for the samples obtained with pulse-off times of 3 s and 4 s essentially showed only the reflection peaks of (111) and (222), clearly demonstrating that the deposits is highly (111)-textured. The high (111)-texture corresponds well to the highly oriented horizontal nanotwins observed in these samples. The increase in (111) orientation of pulse plated samples can be explained by the difference in the surface energies of different crystal planes. As discussed by other researchers, [42] (111) surface is the lowest surface energy among other principle crystal planes. Therefore it is reasonable that introduction of off-time leads to the growth of a (111)-textured film, which is favored in terms of the surface energy. This change in texture is also considered to contribute to the welldefined orientation of nanotwins. The distribution of twin-lamella spacing estimated from crosssectional FIB images are shown in Fig. 3g-i. The graphs revealed that the major distribution of twin spacing found in the sample is 50-100 nm regardless of the off-time. However, the number of large twin lamellae with more than 300 nm was reduced significantly with an increase in the pulse-off time, showing that more frequent nanotwinning occurred at longer pulse-off times.

Consequently, the average twin spacing of Cu samples decreased from 180 nm to 106 nm with an increase of the period of off-time from 2 s to 4 s. 3.3. Vertically oriented nt-Cu. We found that nanotwin orientation can be manipulated by changing the deposition potential. At -0.6 V vs SCE (Fig. 4), the growth of large fiber-type grains was observed. Interestingly, for this sample, the channeling contrast in grains was observed as vertical stripe patterns, suggesting a significant change in nanotwin formation with a pulse-on potential. From the FIB image, twin boundaries in this sample traverse grains almost in the vertical direction to the substrate, although some of the twins were slightly slanted (< 5 ). Moreover, textured surface morphology appeared on the deposit, showing the anisotropy of grain growth of Cu along the film growth direction. The top view of the sample (Fig. 4c) shows that grains were formed without any preferred in-plane orientation. As is seen in the image, all grains contain many parallel patterns that are attributed to twin boundaries. These observations revealed that twin boundaries were formed along the growth direction in the early stage of film growth and they continued to grow upward during subsequent crystal growth. The pulse-off time was also found to play an important role for the samples deposited at -0.6 V. At this potential, a sample

Fig. 4. A FIB image of nt-Cu prepared by pulse plating at an on-potential of -0.6 V vs. SCE (ton: 20 ms, toff: 4 s): (a,b) cross-sectional and (c) top view.

M. Hasegawa et al. / Electrochimica Acta 178 (2015) 458–467

463

Fig. 5. Cross-sectional FIB images of Cu electrodeposits prepared at -0.6 V vs. SCE with a toff of (a) 0 s (without pulses) and (b) 1 s.

Fig. 6. Cross-sectional FIB images of Cu electrodeposits prepared by pulse electrodeposition at a pulse-on potential of -0.5 V (ton: 20 ms, toff: 2 s).

prepared without pulses (toff = 0 s) resulted in nodular electrodeposits due to significant H2 evolution during deposition (Fig. 5a). On the other hand, a nanotwinned continuous film was observed at a pulse-off time of 1 s (Fig. 5b), although twin direction was more random and more inclined from the vertical direction compared with the vertical nt-Cu deposit obtained at a pulse-off time of 4 s (Fig. 4). Such randomly slanted nanotwins were observed also when the samples were prepared at a less negative -0.5 V (Fig. 6). The differences in the effects of plating parameters on nanotwinning behavior of Cu deposits grown at -0.2 V and -0.6 V suggest that the growth mechanism of vertically oriented nt-Cu is significantly different from that of the horizontally oriented nt-Cu. The presence of the vertically oriented nanotwins can be confirmed in high-resolution TEM imaging, as shown in Fig. 7. A high density of the twin boundaries with the direction in parallel with the growth direction is shown in a low-magnification image in Fig. 7a. The Fourier transformation of the high-resolution image in Fig. 7b is consistent with the pattern from a nanotwin. In addition, the resolved lattice planes show mirror symmetry, as seen in Fig. 7c and d. Unlike the horizontal oriented nanotwins, the vertically oriented nanotwins propagate through the grain up to the surface without intersections by grain boundaries or stacking

faults, suggesting a significant difference in the growth mechanism between the horizontal and vertical nt-Cu. According to the TEM observation, the texture of the film should be {211}, whose reflection is missing in XRD. Accordingly, XRD profiles of vertically oriented nt-Cu samples showed only very weak reflection peaks, whose intensities were extremely smaller than those for horizontally oriented nt-Cu samples with a similar thickness. For instance, the peak intensity of (111) reflection of the vertically oriented nt-Cu was less than 2% of that of the horizontally oriented nt-Cu. Jang and coworkers [14] reported that the twin orientation in a Cu nanopillar can be inclined 18 from the horizontal direction by changing the angle between the cathode and anode. However, the change which they observed is much smaller than that shown in Fig. 4, and we consider that they are completely different findings. To our knowledge, the synthesis of such a vertical nt-Cu bulk film has never been reported so far. The significant change in the twin orientation from the horizontal to vertical direction in the samples prepared at different growth potentials suggest that twins form in the different mechanism in these samples (Fig. 8). For the horizontally oriented nt-Cu, twin lamellae are created layer by layer. The difference between the samples prepared by DC and

464

M. Hasegawa et al. / Electrochimica Acta 178 (2015) 458–467

Fig. 7. (a) A low-magnification image of the Cu cross-section showing the vertically oriented nanotwins. The twin boundaries (TBs) are denoted with the red arrows. (b) A high-resolution image showing a twin boundary (TB). The inset is the Fourier transformation of the image with the indices of the resolved lattice planes. (c) and (d) show digitally magnified images of the TB. The mirror planes of the nanotwin are denoted by the dotted lines.

pulse plating clearly show that nanotwins form during the off-time as a consequence of the stress relaxation. At this growth potential, the current density during pulse-on times is 100-150 mA/cm2. Assuming 100% current efficiency, this current density corresponds to the growth of 0.7 to 1.1 nm Cu per pulse-cycle, which is much thinner than the twin lamellae observed in the sample. Therefore, in this growth condition, we consider that new twin boundaries (TBs) nucleate and laterally propagate once the accumulated residual stress reaches at a critical point. The length of the pulse-off time up to some seconds affects the twin spacing and the frequency of the vertical or inclined stacking fault formation, may reflect the fact that the stress relaxation process is a relatively slow process. On the other hand, for vertical nt-Cu, TBs are generated in initial stages of film growth and they propagate along the growth direction, resulting in penetration of TBs vertically through the grains. It is worth notifying that there are many submicrometerscale steps along the twin lamellae on the surface of this film. This surface morphology clearly shows the anisotropic nature of the crystal growth of vertically oriented nt-Cu. According to the twinning model proposed by Ming and Sunagawa [43] for fcc crystals, there are two possible outcrops of twin lamellae formed on growth surfaces for (111) twinning in fcc crystals: {111} and {001} planes depending on fault vectors. One of the twin growth mechanism is driven by reentrant corners appear at the TB formed between two {111} planes on the growing surface. The reentrant corners are expected to be preferential growth sites because the

adatom at a reentrant corner has 4 nearest neighbors, whereas that at the smooth {111} surface and convex corners has only 3 nearest neighbors. The growth surfaces with {001} planes also should act as growth promoting sites because the {001} surface is atomically rough compared with the {111} surface and has 4 nearest neighbors which is the same number as that for the reentrant corners at TBs with two {111} growth surface. The above discussed twinning mechanism should be varied also for the horizontally oriented nt-Cu. However, in the case of the horizontally oriented nt-Cu, the twin propagation direction is normal to the general film growth direction, which may lead to the more frequent irregular staking fault formation in the horizontally oriented nt-Cu compared to the vertically oriented nt-Cu. Because nanotwins are formed in (111) plane, the texture of the horizontally oriented nt-Cu and that of the vertically oriented ntCu are (111) and (11-2), respectively. The texture development in electrodeposited metal films have often been explained in terms of the minimization of the surface energy or the strain energy, which are highly dependent on planes due to the crystal anisotropy [42,44,45]. From the consideration of the surface energy minimization, (111) texture is favored for fcc metals. Therefore, (111)textured films tend to be obtained at a low current density or a low over potential because the stress development in such deposits is generally low. On the other hand, at a high current density (or overpotential), other textures start to develop because the internal

M. Hasegawa et al. / Electrochimica Acta 178 (2015) 458–467

465

Fig. 8. Growth of horizontal and vertical nt-Cu at different deposition potentials and pulse-off times.

stress accumulates more significantly in the film and the strain energy becomes more important than the surface energy. The nt-Cu films obtained at -0.2 V in this study show strong (111) texture, which is consistent with the surface energy minimization theory. For the deposition at a pulse-on potential of -0.2 V, the peak current density during the on-time was -0.1 to -0.15 A/cm2, while it increased up to -0.25 A/cm2 for a pulse-on potential of -0.6 V. A higher current density allows more Cu adatoms to be generated and incorporated in the crystal during ontime period, which will in turn result in a more significant tensile stress development. We believe that an increase in the stress caused by the high deposition rate is partially responsible for the (11-2)-texture development in the vertically oriented nt-Cu films. According to the theoretical studies carried out by Zhang and coworkers [44], {211} texture is one of the four textures which are energetically preferred to {111} texture when the internal stress is sufficiently high for plastic deformation to occur. Provided that the sufficient stress was accumulated during the film growth at -0.6 V, the growth of (11-2) texture is therefore reasonable from the consideration of the strain energy minimization. Nevertheless, the strain energy alone does not fully explain the very strong texture developed in the vertically-oriented nt-Cu because other three textures, namely {110}, {100} and {511}, are more favored textures in terms of the strain energy, according to their results. The selective growth of (11-2) texture may be induced by the existence of surface adsorbates such as hydrogen adatoms and additives, which inhibit the growth of specific planes. Regarding these aspects, we should point out the fact that the hydrogen evolution occurs at -0.6 V, while it is negligible for the deposition at -0.2 V. Furthermore, hydrogen is possibly incorporated in films [46], which could affect the stress development. It has been reported, in

the case of Ni electrodeposition, that the incorporation of hydrogen induces an increase in the compressive stress by expanding the lattice, while its diffusion out from the deposits causes an increase of the tensile stress [47]. The bath additives used in the present study may also play roles in the texture development by changing the surface electrochemistry, although we did not investigate their effects. The incorporation of additives, however, is unlikely because the chemical analysis showed that the concentrations of impurities in Cu films prepared at different growth potentials are similar to each other; the concentrations of carbon and sulfur were both below 0.001wt%, regardless of the deposition potential. Because there are many different factors which could affect the growth of nt-Cu as discussed above, the detailed mechanism of the growth nt-Cu with different twin orientations remains unclear. However, we speculate that the synergy effect of the internal stress and the complicated surface chemistry during electrodeposition leads to the specific texture development, which results in the significant changes in twin orientation of nt-Cu at different growth potentials. Highly textured nt-Cu is known to exhibit strong anisotropy in mechanical properties [20,21], which are also dependent on twin spacing [11,18]. These findings provide insights into controlling and designing of twin boundaries in nt-Cu which will be of importance for many practical applications which requires well-defined properties. 4. Conclusion We reported the electrodeposition of nt-Cu films with a controllable twin orientation in an acidic Cu sulfate bath containing a combination of chloride ions, polyethylene glycol

466

M. Hasegawa et al. / Electrochimica Acta 178 (2015) 458–467

and mercapto-1-propanesulfonic acid as additives. The microscopic studies revealed that the twin orientation can be altered from the horizontal to vertical direction by adjusting the growth potential. The changes in the orientation of nanotwins clearly demonstrate that the significant change in the twinning mechanism which is attributed to the difference in the stress evolution during pulse-on times, leading to a difference in the stressrelaxation process during pulse-off times. For the horizontally oriented nt-Cu, we found that an increase of the pulse-off time resulted in an increase in the twin density, indicating that the longer pulse-off time allows for the enhanced twinning which is associated with the stress relaxation. These results provide important insights into the nanotwin formation mechanism which will be useful for the microstructure engineering of metals with nanotwins. Acknowledgements This research was financially supported by the Eureka-Eurostars project E!6921 MASSMICRO. MH acknowledges the MarieCurie cofound “Empa Postdocs” fellowship for the financial support. References [1] G. Mohanty, J.M. Wheeler, R. Raghavan, J. Wehrs, M. Hasegawa, S. Mischler, et al., Elevated temperature, strain rate jump microcompression of nanocrystalline nickel, Philos. Mag. (2014) 1–18, doi:http://dx.doi.org/ 10.1080/14786435.2014.951709. [2] K.S. Kumar, K. Biswas, Effect of thiourea on grain refinement and defect structure of the pulsed electrodeposited nanocrystalline copper, Surf. Coat. Technol. 214 (2013) 8–18, doi:http://dx.doi.org/10.1016/j.surfcoat.2012.10.018. [3] T.J. Rupert, J.R. Trelewicz, C.A. Schuh, Grain boundary relaxation strengthening of nanocrystalline Ni–W alloys, J. Mater. Res. 27 (2012) 1285–1294. [4] J.D. Giallonardo, U. Erb, K.T. Aust, G. Palumbo, The influence of grain size and texture on the Young's modulus of nanocrystalline nickel and nickel-iron alloys, Philos. Mag 91 (2011) 4594–4605, doi:http://dx.doi.org/10.1080/ 14786435.2011.615350. [5] L. Lu, M.L. Sui, K. Lu, Superplastic extensibility of nanocrystalline copper at room temperature, Science (80-.). 287 (2000) 1463-1466. doi:10.1126/ science.287.5457.1463. [6] T.J. Rupert, J.C. Trenkle, C.A. Schuh, Enhanced solid solution effects on the strength of nanocrystalline alloys, Acta Mater. 59 (2011) 1619–1631. doi:http:// dx.doi.org/10.1016/j.actamat.2010.11.026. [7] C.A. Schuh, T.G. Nieh, H. Iwasaki, The effect of solid solution W additions on the mechanical properties of nanocrystalline Ni, Acta Mater. 51 (2003) 431–443. doi:http://dx.doi.org/10.1016/S1359-6454(02)00427-5. [8] O. Younes-Metzler, L. Zhu, E. Gileadi, The anomalous codeposition of tungsten in the presence of nickel, Electrochim. Acta 48 (2003) 2551–2562, doi:http:// dx.doi.org/10.1016/s0013-4686(03)00297- 4. [9] O. Younes, E. Gileadi, Electroplating of Ni/W alloys - I. Ammoniacal citrate baths, J. Electrochem. Soc. 149 (2002) C100–C111, doi:http://dx.doi.org/ 10.1149/1.1433750. [10] L. Lu, R. Schwaiger, Z.W. Shan, M. Dao, K. Lu, S. Suresh, Nano- sized twins induce high rate sensitivity of flow stress in pure copper, Acta Mater. 53 (2005) 2169–2179. doi:http://dx.doi.org/10.1016/j.actamat.2005.01.031. [11] L. Lu, X. Chen, X. Huang, K. Lu, Revealing the Maximum Strength in Nanotwinned Copper, Science (80-.). 323 (2009) 607-610. doi:10.1126/ science.1167641. [12] L. Lu, Y.F. Shen, X.H. Chen, L.H. Qian, K. Lu, Ultrahigh strength and high electrical conductivity in copper, Science (80-.). 304 (2004) 422–426. doi:10.1126/science.1092905. [13] H.Y. Hsiao, C.M. Liu, H.W. Lin, T.C. Liu, C.L. Lu, Y.S. Huang, et al., Unidirectional Growth of Microbumps on (111)-Oriented and Nanotwinned Copper, Science (80-.). 336 (2012) 1007–1010. doi:10.1126/science.1216511. [14] D. Jang, X. Li, H. Gao, J.R. Greer, Deformation mechanisms in nanotwinned metal nanopillars, Nat Nano. 7 (2012) 594-601. doi:http://www.nature.com/ nnano/journal/v7/n9/abs/nnano.2012.116.html#supplementary-information. [15] Y.M. Wang, F. Sansoz, T. LaGrange, R.T. Ott, J. Marian, T.W. Barbee Jr, et al., Defective twin boundaries in nanotwinned metals, Nat Mater. 12 (2013) 697– 702. doi:10.1038/nmat3646 http://www.nature.com/nmat/journal/v12/n8/ abs/nmat3646.html#supplementary-information. [16] K.A. Afanasyev, F. Sansoz, Strengthening in Gold Nanopillars with Nanoscale Twins, Nano Lett. 7 (2007) 2056–2062, doi:http://dx.doi.org/10.1021/ nl070959l. [17] D. Jang, C. Cai, J.R. Greer, Influence of Homogeneous Interfaces on the Strength of 500nm Diameter Cu Nanopillars, Nano Lett. 11 (2011) 1743–1746, doi:http:// dx.doi.org/10.1021/nl2003076.

[18] L. Lu, M. Dao, T. Zhu, J. Li, Size dependence of rate-controlling deformation mechanisms in nanotwinned copper, Scr. Mater. 60 (2009) 1062–1066, doi: http://dx.doi.org/10.1016/j.scriptamat.2008.12.039. [19] L. Lu, T. Zhu, Y. Shen, M. Dao, K. Lu, S. Suresh, Stress relaxation and the structure size-dependence of plastic deformation in nanotwinned copper, Acta Mater. 57 (2009) 5165–5173. doi:http://dx.doi.org/10.1016/j.actamat.2009.07.018. [20] Z. You, X. Li, L. Gui, Q. Lu, T. Zhu, H. Gao, et al., Plastic anisotropy and associated deformation mechanisms in nanotwinned metals, Acta Mater. 61 (2013) 217– 227, doi:http://dx.doi.org/10.1016/j.actamat.2012.09.052. [21] J.C. Ye, Y.M. Wang, T.W. Barbee, A.V. Hamza, Orientation-dependent hardness and strain rate sensitivity in nanotwin copper, Appl. Phys. Lett. 100 (2012) 5, doi:http://dx.doi.org/10.1063/1.4731242. [22] A.M. Hodge, T.A. Furnish, C.J. Shute, Y. Liao, X. Huang, C.S. Hong, et al., Twin stability in highly nanotwinned Cu under compression, torsion and tension, Scr. Mater. 66 (2012) 872–877, doi:http://dx.doi.org/10.1016/j. scriptamat.2012.01.027. [23] K.-C. Chen, W.-W. Wu, C.-N. Liao, L.-J. Chen, K.N. Tu, Observation of Atomic Diffusion at Twin-Modified Grain Boundaries in Copper, Science (80-.). 321 (2008) 1066–1069. doi:10.1126/science.1160777. [24] J. Bezares, S. Jiao, Y. Liu, D. Bufford, L. Lu, X. Zhang, et al., Indentation of nanotwinned fcc metals: Implications for nanotwin stability, Acta Mater. 60 (2012) 4623–4635. doi:http://dx.doi.org/10.1016/j.actamat.2012.03.020. [25] J.R. Greer, Nanotwinned metals: It’s all about imperfections, Nat Mater. 12 (2013) 689–690, doi:http://dx.doi.org/10.1038/nmat3721. [26] C.-N. Liao, Y.-C. Lu, D. Xu, Modulation of Crystallographic Texture and Twinning Structure of Cu Nanowires by Electrodeposition, J. Electrochem. Soc. 160 (2013) D207–D211, doi:http://dx.doi.org/10.1149/2.046306jes. [27] T.-C. Liu, C.-M. Liu, H.-Y. Hsiao, J.-L. Lu, Y.-S. Huang, C. Chen, Fabrication and Characterization of (111)-Oriented and Nanotwinned Cu by Dc Electrodeposition, Cryst. Growth Des. 12 (2012) 5012–5016, doi:http://dx.doi. org/10.1021/cg300962v. [28] A.M. Hodge, Y.M. Wang, T.W. Barbee, Large-scale production of nano-twinned, ultrafine-grained copper, Mater. Sci. Eng. A 429 (2006) 272–276, doi:http://dx. doi.org/10.1016/j.msea.2006.05.109. [29] A.M. Hodge, Y.M. Wang, T.W. Barbee Jr, Mechanical deformation of high-purity sputter-deposited nano-twinned copper, Scr. Mater. 59 (2008) 163–166. doi: http://dx.doi.org/10.1016/j.scriptamat.2008.02.048. [30] T.-C. Chan, Y.-M. Lin, H.-W. Tsai, Z.M. Wang, C.-N. Liao, Y.-L. Chueh, Growth of large-scale nanotwinned Cu nanowire arrays from anodic aluminum oxide membrane by electrochemical deposition process: controllable nanotwin density and growth orientation with enhanced electrical endurance performance, Nanoscale 6 (2014) 7332–7338, doi:http://dx.doi.org/10.1039/ c3nr06194a. [31] C.-M. Liu, H.-W. Lin, C.-L. Lu, C. Chen, Effect of grain orientations of Cu seed layers on the growth of <111>-oriented nanotwinned Cu, Sci. Rep. 4 (2014) , doi:http://dx.doi.org/10.1038/srep06123. [32] D. Xu, V. Sriram, V. Ozolins, J.-M. Yang, K.N. Tu, G.R. Stafford, et al., In situ measurements of stress evolution for nanotwin formation during pulse electrodeposition of copper (vol 105, 023521, 2009), J. Appl. Phys. 105 (2009). doi:10.1063/1.3503174. [33] D. Xu, W.L. Kwan, K. Chen, X. Zhang, V. Ozolins, K.N. Tu, Nanotwin formation in copper thin films by stress/strain relaxation in pulse electrodeposition, Appl. Phys. Lett. 91 (2007) , doi:http://dx.doi.org/10.1063/1.2825412. [34] B. Hong, C.-H. Jiang, X.-J. Wang, Influence of complexing agents on texture formation of electrodeposited copper, Surf. Coat. Technol. 201 (2007) 7449– 7452, doi:http://dx.doi.org/10.1016/j.surfcoat.2007.02.011. [35] G.B. Harris, QUANTITATIVE MEASUREMENT OF PREFERRED ORIENTATION IN ROLLED URANIUM BARS, Philos. Mag. 43 (1952) 113–123. ://WOS: A1952UV23400010. [36] N. Lu, K. Du, L. Lu, H.Q. Ye, Motion of 1/3 < 111 > dislocations on Sigma 3 {112} twin boundaries in nanotwinned copper, J. Appl. Phys. 115 (2014) , doi:http:// dx.doi.org/10.1063/1.4861868. [37] N. Li, J. Wang, J.Y. Huang, A. Misra, X. Zhang, Influence of slip transmission on the migration of incoherent twin boundaries in epitaxial nanotwinned Cu, Scr. Mater 64 (2011) 149–152. doi:http://dx.doi.org/10.1016/j. scriptamat.2010.09.031. [38] J.M.E. Harper, C. Cabral, P.C. Andricacos, L. Gignac, I.C. Noyan, K.P. Rodbell, et al., Mechanisms for microstructure evolution in electroplated copper thin films near room temperature, J. Appl. Phys. 86 (1999) 2516–2525, doi:http://dx.doi. org/10.1063/1.371086. [39] C. Lingk, M.E. Gross, Recrystallization kinetics of electroplated Cu in damascene trenches at room temperature, J. Appl. Phys. 84 (1998) 5547–5553, doi:http://dx.doi.org/10.1063/1.368856. [40] M. Hasegawa, Y. Nonaka, Y. Negishi, Y. Okinaka, T. Osaka, Enhancement of the Ductility of Electrodeposited Copper Films by Room-Temperature Recrystallization, J. Electrochem. Soc. 153 (2006) C117. [41] J. Wang, N. Li, O. Anderoglu, X. Zhang, A. Misra, J.Y. Huang, et al., Detwinning mechanisms for growth twins in face-centered cubic metals, Acta Mater 58 (2010) 2262–2270. doi:http://dx.doi.org/10.1016/j.actamat.2009.12.013. [42] T.-C. Chan, Y.-L. Chueh, C.-N. Liao, Manipulating the Crystallographic Texture of Nanotwinned Cu Films by Electrodeposition, Cryst. Growth Des. 11 (2011) 4970–4974, doi:http://dx.doi.org/10.1021/cg200877f. [43] N. Ming, I. Sunagawa, Twin lamellae as possible self-perpetuating step sources, J. Cryst. Growth 87 (1988) 13–17, doi:http://dx.doi.org/10.1016/0022- 0248 (88)90339-9.

M. Hasegawa et al. / Electrochimica Acta 178 (2015) 458–467 [44] J.-M. Zhang, K.-W. Xu, V. Ji, Dependence of strain energy on the grain orientations in an FCC-polycrystalline film on rigid substrate, Appl. Surf. Sci. 185 (2002) 177–182, doi:http://dx.doi.org/10.1016/S0169-4332(01)00586-4. [45] K. Kremmer, O. Yezerska, G. Schreiber, M. Masimov, V. Klemm, M. Schneider, et al., Interplay between the deposition mode and microstructure in electrochemically deposited Cu thin films, Thin Solid Films 515 (2007) 6698– 6706. doi:http://dx.doi.org/10.1016/j.tsf.2007.01.043.

467

[46] S. Nakahara, Y. Okinaka, Microstructure and Mechanical Properties of Electroless Copper Deposits, Annu. Rev. Mater. Sci. 21 (1991) 93–129, doi: http://dx.doi.org/10.1146/annurev.ms.21.080191.000521. [47] S.J. Hearne, J.A. Floro, Mechanisms inducing compressive stress during electrodeposition of Ni, J. Appl. Phys. 97 (2005) 014901, doi:http://dx.doi.org/ 10.1063/1.1819972.