Mechanical, Electrical, and Crystallographic Property Dynamics of

6 days ago - ... Queensland University of Technology (QUT) , 2nd George Street, ... The core–shell Ge/Si nanowires were bent and strained in tension...
0 downloads 0 Views 4MB Size
Letter pubs.acs.org/NanoLett

Cite This: Nano Lett. XXXX, XXX, XXX−XXX

Mechanical, Electrical, and Crystallographic Property Dynamics of Bent and Strained Ge/Si Core−Shell Nanowires As Revealed by in situ Transmission Electron Microscopy Chao Zhang,*,† Dmitry G. Kvashnin,‡,⊥ Laure Bourgeois,§ Joseph F. S. Fernando,† Konstantin Firestein,† Pavel B. Sorokin,‡,⊥ Naoki Fukata,∥ and Dmitri Golberg*,†,∥

Downloaded via UNIV OF SUNDERLAND on October 25, 2018 at 08:44:30 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



School of Chemistry, Physics and Mechanical Engineering, Science and Engineering Faculty, Queensland University of Technology (QUT), 2nd George Street, Brisbane, Queensland 4000, Australia ‡ Inorganic Nanomaterials Laboratory, National University of Science and Technology MISIS, Leninsky prospect 4, Moscow 119049, Russian Federation ⊥ Emanuel Institute of Biochemical Physics RAS, Kosigina 4, Moscow 119334, Russian Federation § Monash Centre for Electron Microscopy and Department of Materials Science and Engineering, Monash University, Victoria 3800, Australia ∥ International Center for Materials Nanoarchitectonics (MANA), National Institute for Materials Science (NIMS), Namiki 1-1, Tsukuba, Ibaraki 3050044, Japan S Supporting Information *

ABSTRACT: Research on electromechanical properties of semiconducting nanowires, including plastic behavior of Si nanowires and superb carrier mobility of Ge and Ge/Si core− shell nanowires, has attracted increasing attention. However, to date, there have been no direct experimental studies on crystallography dynamics and its relation to electrical and mechanical properties of Ge/Si core−shell nanowires. In this Letter, we in parallel investigated the crystallography changes and electrical and mechanical behaviors of Ge/Si core−shell nanowires under their deformation in a transmission electron microscope (TEM). The core−shell Ge/Si nanowires were bent and strained in tension to high limits. The nanowire Young’s moduli were measured to be up to ∼191 GPa, and tensile strength was in a range of 3−8 GPa. Using high-resolution imaging, we confirmed that under large bending strains, Si shells had irregularly changed to the polycrystalline/amorphous state, whereas Ge cores kept single crystal status with the local lattice strains on the compressed side. The nanowires revealed cyclically changed electronic properties and had decent mechanical robustness. Electron diffraction patterns obtained from in situ TEM, paired with theoretical simulations, implied that nonequilibrium phases of polycrystalline/amorphous Si and β-Sn Ge appearing during the deformations may explain the regarded mechanical robustness and varying conductivities under straining. Finally, atomistic simulations of Ge/Si nanowires showed the pronounced changes in their electronic structure during bending and the appearance of a conductive channel in compressed regions which might also be responsible for the increased conductivity seen in bent nanowires. KEYWORDS: In situ TEM, flexible electronics, core−shell nanowire

C

response and higher carrier motilities.1 In addition, singlecrystalline one-dimensional Si and Ge nanowires can bear much higher strains than their bulky counterparts. It has also been theoretically predicted and experimentally confirmed that such nanowires are direct band gap materials, which indicates their great potential for optoelectronics.

onventional silicon-based semiconducting technology has been well developed. However, it still has some limitations, e.g. difficulties in achieving quantum tunneling effects, the need for a rigid substrate, and indirect band gaps restricting possibilities with regard to future flexible electronics and optoelectronics. Semiconducting nanotubes and nanowires are attractive candidates for next-generation applications because of their extraordinary electrical, mechanical, and optoelectronic properties. As compared to macroscale Si or Ge transistors processed by conventional photolithography, nanowires made of these elements demonstrate a faster © XXXX American Chemical Society

Received: August 22, 2018 Revised: October 8, 2018 Published: October 22, 2018 A

DOI: 10.1021/acs.nanolett.8b03398 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Results and Discussion. The two designed setups for electrical and mechanical tests are displayed in Figure 1. In situ

Generally, cubic Si and Ge structures are similar in many respects but have specific features with regard to mechanical strength and carrier mobility. Young’s modulus of bulk cubic Si1−xGex materials is presented using the following relation: Y110 = (169 − 31x) GPa along the [110] direction.2 Young’s modulus of nanowires also falls into this range. In terms of electronic properties, Ge possesses higher electron/hole mobility than Si.3 Si also has a big advantage with respect to creation of a perfect semiconducting/insulating Si/SiO2 structure, especially important in electronics. A Ge/Si core− shell nanowire has been proven to be a very promising material for high-frequency electronics and stable high-mobility programmable logic circuits.4,5 In addition, Ge and Ge/Si junctions are essential for infrared optics and optoelectronic communication applications. Thus, Ge/Si core−shell nanowires should stand out for their high flexibility and tunability of optoelectronic properties. There have been several in situ TEM studies on Si nanowires. These show that thin Si nanowires can take a much higher bending strain than bulk silicon.6−9 In the paper by Tang et al.10 an ∼400 nm long, 25.3 nm thin Si nanowire withstood a strain of up to ∼20%, while revealing a crystallineto-amorphous phase transition. Zheng et al. reported a colloidal thin film bending technique and observed a stressinduced elastic-plastic transition.11 These authors also studied dislocations and Lomer locks in bent Si nanowires under strains from 1.3% to 12.8%.8 In terms of technological applications, Si nanowire arrays have been considered as materials for high capacity flexible lithium-ion batteries,12 flexible surface enhanced Raman spectroscopy,13,14 field-effect transistors,15,16 flexible nanogenerators,17 and thermoelectric generators.18 However, direct in situ and parallel mechanical and electrical probing of Ge/Si core−shell nanowires has not been performed under detailed in-tandem crystallography observations. It is known that Si nanowires have comparable Young’s modulus with bulk silicon.10 Decent plasticity of such nanowires up to high strains has also been demonstrated.7 This phenomenon attracted our attention, and we initiated research on Ge/Si core−shell nanowires in light of future flexible and foldable electronics. Therefore, in this work, we performed in-tandem nanoscale manipulations, crystallographic analyses, and mechanical and electrical tests on Ge/Si core−shell nanowires using in situ TEM probing techniques. We found that not only elasticplastic mode transitions under bending had taken place but also nonequilibrium crystallographic changes, and, highly likely, diffusionless phase transformations had occurred. Under bending to a 7% strain, nanowire conductivity increased. However, after further applying a pressing technique and achieving a higher strain of ∼15−20%, conductivity of Ge/ Si nanowires significantly dropped. High-resolution TEM images of deformed Ge/Si nanowires were analyzed, and it was found that local strains had induced dynamic, nonpermanent changes in nanowire crystallography. The heavily deformed Ge/Si nanowires and their cores were also studied by aberration-corrected Scanning TEM (ACSTEM), and these studies confirmed that Ge/Si nanowires could keep their integrity even after ∼30−40% strains without cracking. Etching-off Si shells out of plastically deformed Ge/Si nanowires revealed that Ge cores possessed almost defectfree atomic structures with minor lattice strains on compressed nanowire sides.

Figure 1. Two designed setups for in situ TEM bending of core−shell Ge−Si nanowires during (a) electrical or (b) mechanical measurements.

experiments were conducted with two high-resolution TEMs. One TEM was a JEOL 2010 transmission electron microscope with a LaB6 filament and Zues in situ and Heracles in situ specimen holders (Zeptools, P.R. China). The other TEM was a JEOL 3100F transmission electron microscope with a field emission gun, an Omega filter, and a scanning tunnelling microscope (STM)-TEM specimen holder (NanoFactory Instruments, Inc.). The electrical probing experimental setup is shown in Figure 1a, and more details were described by us previously.19,20 The in situ mechanical setup is depicted in Figure 1b. Nanowires were probed by the atomic force microscope (AFM) cantilever, which was fixed to the holder frame with bridged electrical circuits; a nanowire sample was attached to the piezo-driven gold probe. The bright-field STEM and HAADF images of Ge/Si core− shell nanowires, the deformed Ge cores, and spatially resolved elemental maps were obtained on a double-aberrationcorrected FEI Titan3 80-300 FEGTEM. It was demonstrated that 200 kV or 300 kV electron beams with a low electron flux dose, of around 5 × 1019 e cm−2 s−1, had not affected Si or Ge crystallography.7 However, still, during acquiring electrical or mechanical data, the electron beam was blocked and/or tilted away from the samples in order to entirely avoid data artifacts due to the electron beam effects. Calculations of electronic properties (charge density, band structure, DOS) of Ge/Si core−shell nanowires were carried out using density functional theory (DFT) with the general gradient approximation (DFT-PBE) implemented in the SIESTA package with LCAO basis.21 Due to the large number of atoms in the considered supercell, the energy cutoff was set to 125 Ry. The full-scale simulations of bending and relaxation processes under room temperature of hybrid Ge/Si nanowires were carried out using classical molecular dynamics (MD) implemented in the LAMMPS simulation package.22 Interatomic interactions between Si and Ge were described with a well-developed theoretical approach of the Tersoff many-body B

DOI: 10.1021/acs.nanolett.8b03398 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters potential.23 Considered models of Si/Ge nanowires included up to 10000 atoms within the core and shell and had a length of 20 nm and a width of 3 nm. The nanowire length and diameter and the sample displacements during manipulations were measured by the “Gatan Microscopy Suite”(GMS) software. The strains (based on the measured bending radius and the wire diameter) and cross-sectional areas were determined by using the “Digimizer” software, and the details are shown in Figure S1. The bending strain was denoted as ε = d/(d+2R), where d and R are the nanowire diameter and its radius of curvature after bending, respectively. Considering that TEM provides 2D imaging, while the deformation may occur in the 3D space, the sample height was always carefully adjusted to match all stages of the experiments not only by using a TEM wobbler function but also by controlling the sliding conditions of the contact. If the sample had slightly shifted in the z direction during contact, its height was readjusted. However, during bending, the nanowire buckling direction might not be perfectly aligned to the 2D imaging plane; and, therefore, in some cases, the deformation conditions could not be precisely evaluated. All Ge/Si core−shell nanowires were grown by a chemical vapor deposition method, using GeH4, SiH4 gases, and an ntype (111) Si wafer. The details can be found in our previous publications.24,25 The as-grown Ge/Si nanowires were first characterized by STEM, and the high-resolution results are demonstrated in Figure 2a. The images were taken along the Si

from the selected areas of the nanowire are illustrated in Figure S2. The nanowire is ∼20 nm thick. The Ge core signal can easily be detected in this case. Elemental EDX maps of Si and Ge are displayed in the insets. The FFT pattern is shown in Figure 2b. The two sets of visible diffraction spots are indexed. In Figure S3, the interface mismatches are also marked. After applying a spot mask on Si 202/-20-2 and Ge 220/-2-20 spots by GMS software, the inverse FFT filtered image was created, which clearly reveals the interface mismatches.28 In Figure S3a, a representative “3Ge-2Si” mismatch is marked by red dashed lines. Figure S3b is the corresponding STEM image of the Ge/Si interface. The mismatch here is located within the area of the weaker contrast as compared to Figure 2a. By applying a spot mask and inverse FFT, the filtered image makes mismatches much clearer. The process is shown in Figure S3c. In Figure 2c, a filtered image at the lower magnification shows nine mismatches within an ∼35 nm range. It is noted that, as compared to the perfect interface, shown in Figure 2, on one side the mismatches only appear at the interface from the other side of the core−shell nanowire. This can be attributed to the nanowire growth conditions in a horizontal tube furnace. On the other side however, no mismatches were found. Among the mismatches, 5 have “2Ge3Si” structures, and 4 are of the “3Ge-2Si” type. Even though mismatches do exist, the lattice interplanar distances of Si and Ge are still identical and are 2.00 ± 0.01 nm. Sometimes, it was difficult to distinguish Si diffraction spots from those of Ge, not only due to their close bulk lattice spacings and dispersions but also due to in parallel crystal growth of the two segments, as displayed in Figure S3. In many cases, Ge and Si diffraction spots are overlapped because the segments grew along the same direction. In this case, it was difficult to extract the clear crystallographic information, that was obtained in Figure 2 and Figure S2, in which we were able to clearly determine Si and Ge lattices along the [10-1] and [1-11] zone axes. Mechanical properties of the nanowires were then tested by in situ TEM probing using an AFM holder. One representative set of strain−stress data is displayed in Figures 3a and 3b. Before the test, the nanowire was firmly fixed to the cantilever by electron beam-induced deposition (EBID).10 During the first tensile test, under moving the sample probe backward from the AFM cantilever, a linear stress−strain curve was recorded prior to the nanowire final failure. The broken end of the nanowire is shown in the inset of Figure 3a. The cross section of the broken area was calculated given the structure diameter of 41 nm. Then, the nanowire was repeatedly and firmly welded to the probe using the same EBID method, and the second tensile test was conducted. The width of the broken end, given as the nanowire diameter at this point, was 47 nm. By considering that the nanowire’s cross section can be hexagonal and even irregular,29 we set the cross-sectional range from (π/4)d2 to d2. Then we introduce an irregular shape variant p, ranging from 0.785 to 1. The maximum forces during the first and second tensile tests were measured as 7148 nN and 11526 nN, respectively. The fracture strength was F , calculated using the following formula: σ = 2 max

Figure 2. (a) HAADF-STEM image of a Ge/Si core−shell nanowire interface. (b) FFT of the nanowire. (c) Si 202/-20-2 and Ge 220/-220 filtered image at a lower magnification shows nine mismatches within an ∼35 nm range.

[10-1] and Ge [1-11] axes. The growth directions of Si and Ge sections were [202] and [220], respectively. Effectively, both Ge core and Si shell were grown along the ⟨110⟩ directions. The STEM image clearly displays the interface between Ge and Si domains. Lattice spacings for {220} in two portions are identical and equal to 1.99 ± 0.01 nm. As a reference, the lattice distances of bulk Si and Ge are 0.1920 and 0.1998 nm at normal temperature and pressure, respectively.26,27 Low magnification images and Fourier transform (FFT) patterns

pd ·cos φ cos θ

where θ is the angle between the nanowire orientation and the displacement axis of the AFM cantilever on its projection to the TEM image plane. In the experiment, the 2D angle between the nanowire projection and the AFM displacement axis was measured to be θ = 13.54. The exact angle between the nanowire and electron beam axis is actually unknown in C

DOI: 10.1021/acs.nanolett.8b03398 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 3. Tensile stress−strain curves recorded on an individual Ge/Si core−shell nanowire before (a) and after (b) its first breakage. The corresponding insets illustrate TEM images of the broken sections. Scale bars: 10 nm. (c) TEM images of a strained nanowire, showing local contrast under a 3.2% strain. Diffraction pattern of the nanowire and enlarged selected reflection are presented in the insets (from the red square area) and denote the polycrystalline Si segment and a crystalline Ge core. Scale bars: 100 nm, 20 nm, 5 1/nm; right panel: MD simulation of the Ge/Si core−shell hybrid nanowire at room temperature before and after bending. Red area denotes formed polycrystalline/amorphous Si region onto the pure crystalline Ge core. (d) Force values recorded under bending to a 5% strain followed by nanowire reloading. Red and blue dots correspond to the experimental values, whereas green circles represent the theoretically obtained values for hybrid nanowires with a 13.41 L/D ratio (analogous with the experimental values) at different bending strains.

GPa, whereas at the second breakage point it was calculated as 206 GPa. These values are close to that of a 18 nm diameter Si nanowire in the [111] direction; the latter delivered the modulus of 201 GPa.10 Damon et al. reported Young’s modulus of Ge nanowires (with a diameter from 50 to 140 nm) to be around 106 GPa; one nanowire in their work (diameter 55 nm) exhibited Young’s modulus of 190 GPa.30 Those authors also found a trend for Young’s modulus to increase when the diameter of a Ge nanowire decreases. With regards to Young’s modulus of Ge/Si core−shell nanowires, only theoretical calculations have been performed until now. For example, Liu et al. calculated that a Ge/Si core−shell nanowire should have a higher Young’s modulus than the isolated Si or Ge constituents.29 The authors inferred that the core and shell are competing with each other; this fact results in the composition-dependent Young’s modulus. An interesting point of our experiment is that the measured ultimate tensile strength and Young’s modulus values at the second breakage are higher than those of the first breakage. In fact, the tensile strength was about 2 times higher on the second breaking point, while Young’s modulus was only about 16% higher (we note though that the latter number is still within the error range). Generally, it is believed that Young’s modulus

TEM. We may take this into account and introduce an error of up to φ = ±30. Therefore, the averaged factor will become cos φ =

3 π

π /6

∫−π /6 cos φdφ. As described, pA is the cross-

sectional area of the fractured end with the irregular shape variant. Then the error was determined by p cos φ, given a considerable number of 32%. The first fracture site was close to the probe, 35 nm distance from it (original length 520 nm), whereas the second fracture was 90 nm from the probe (original length 483 nm). The ultimate tensile strength after the first breakage was then calculated to be ∼3.85 GPa at a 2.2% strain, while this value after the second break was determined to be ∼8.34 GPa at a 3.8% strain. The two values are fairly similar to the numbers peculiar to Si nanowires.10 The strains were calculated using a value of Δl/l0, where l0 was the original length of a certain part of the nanowire. The measured error of Δl was estimated to be ∼5 nm. The stress was calculated using a formula F/pd2. Finally, Young’s modulus was determined via the following equation: E =

F / pd 2 , Δl / l 0

using

the linear fit slope from Figures 3a and 3b. The linear fits are displayed in Figure S4. Young’s modulus of the Ge/Si core− shell nanowire at the first breakage event was calculated as 177 D

DOI: 10.1021/acs.nanolett.8b03398 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

straightening the wire a polycrystalline/amorphous Si region formed. Despite the regarded changes within the Si shell structure, the Ge core kept its single crystalline status. Figure 3d shows a bending strain-force diagram presented as a series of numbers of force recorded during deformation. The red square dots correspond to loading, i.e. the AFM cantilever pushes the nanowire toward the sample probe. The chart shows the positive correlation of the bending strain and the force that the nanowire takes. The nanowire bears 425 nN at a strain of 3.2%. During the unloading process, the nanowire takes less force than than during the loading process. The experimental data corresponds well with the theoretical results. We calculated a force response during the hybrid nanowire bending at different strains; see the green circles in Figure 3d. Note that one of the important parameters during the force measurements (as well as in the simulations) is the ratio between nanowire length and its diameter. Thus, we considered hybrid nanowires with different L/D ratios (from 6.42 to 29.28, see Figure S6). It was found that even thin hybrid nanowires (with a diameter of about 4 nm) with the L/ D ratio similar to that in the experiments display mechanical properties in excellent agreement with the experimental measurements (green circles compared with the red ones in Figure 3d). In Figure 4, a series of TEM images displays a set of manipulations with the wire, including its bending in one direction, bending in the other direction, and pressing straight along the nanowire axis. Figure 4a(1) shows the original state

should not vary in different morphologies of the same material, and it is also reported that sizes of nanowire have no certain relationship with Young’s modulus.30 Therefore, in our case, Young’s modulus of the nanowire can be evaluated to be the same or to be of a minor difference and is averaged at ∼192 GPa. However, the tensile strength difference at two breaking points is 217%, which is much higher than the calculated error range of 32%. The previous research results showed that wider nanowires had possessed a lower strength.10 So, highly likely, the bigger nanowires have more possibilities to have structural defects as stress concentrators and demonstrate a lower strength. It should be noted that the second measurement in Figure 3b demonstrated a turning point at a strain of ∼2% and a stress of ∼3.2 GPa, which is close to the first breakage point in Figure 3a. It can be related to the cracks and potential defects initiated during the first tensile test. Therefore, the first break and the turning point can be linked to the dislocation activity which happens in Si at a strain of ∼2%, while the second break taking place at a strain of 3.8% can be related to a Lomer-Cottrell lock, which is in accordance with the statistical analysis by Wang et al.8 An in situ bending experiment was then performed as shown in Figure 3c. The low magnification TEM image demonstrates the overall loading setup. The AFM cantilever is in the right corner. The moving probe is located on the left-hand side of the figure. The movement of the probe is made parallel to the original nanowire’s orientation. A selected-area diffraction pattern of the bent nanowire shows the related crystallography. A higher magnification image reveals uneven contrast within the nanowire at a 3.42% strain (from the area marked in the low magnification TEM image). A magnified selected area diffraction pattern (SADP) from the -1-1-1 diffraction spot shows clear Ge and Si reflections. In particular, the characteristic halo-like ring of Si {111} is solid evidence that amorphization (the process and trend of being polycrystalline/ amorphous) takes place within the Si shell, which has also been noticed previously.7,10,31 Amorphization of the Si shell during bending can also be confirmed under imaging, as shown in Figure S5. It is noted that the regarded amorphization is not general and takes place only in some local Si parts, at the contact end and at the segment in the bent region. By contrast, it is known that amorphization does not easily happen in Ge.32 This is also confirmed by simulations, for which we directly observed the Si shell distortion in the compressed region, see Figure 3c. In addition, our MD simulations showed the formation of area with a crystalline structure that differed from the standard Ge/Si lattices (being caused by a high deformation ratio) in the inner portion of the bent region, see Figure 3c. The employed algorithm is able to analyze the local environment of each atom (up to the second neighboring shell) to determine the local structural type.33 The same results were obtained for nanowires with a larger diameter (up to 6 nm). Moreover, after detailed analysis of curved nanowires with various diameters it was found that amorphization took place after the bending strain of ∼40%, for all considered sizes. This fact agrees well with the experimental data. Therefore, we suggest that the high value of a critical strain can lead to the partial amorphization of the nanowire structure in the bent region (the obtained higher critical strain value than in the experiments can be explained by a smaller diameter of the designed nanowire and its perfect structure). Removal of the load from the wire led to the fast relaxation of its original geometry (during 50 ps) at room temperature. Thus, after

Figure 4. (a) I−V curves taken from a Ge/Si core−shell nanowire at its different deformation statuses in correspondence to the consecutive TEM images shown as insets. Insets a(1−7) illustrate the wire bent in two different directions. Insets a(8−12) present TEM images recorded under reloading of the wire followed by its deformation to a large strain. The values of measured bending strains are marked on the images. (b−d) Local strains illustrated by a set of TEM images of the deformed Ge/Si nanowire. Yellow dashed lines show the areas of interest. E

DOI: 10.1021/acs.nanolett.8b03398 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

in Figure S7b. In addition, from our experimental experience, the lattice strain can be traced by referring to the movement of contrast, which is not only affected by the orientation changes within the crystal but also by structure amorphization at a high strain. A typical example of such a moving strain is illustrated in Figure S8. The electrical performance of the bent nanowires was also examined. For example, by taking the I-t curve of a nanowire during the bending process, the conductance changes were confirmed under bending-release cycles, as depicted in Figure S9. We comparatively in situ tested an individual pure Si nanowire, a pure Ge-nanowire core after Si-shell etching, and a core−shell Ge/Si nanowire. It appears that the intrinsic Si nanowire has a very low conductivity, while the Ge core possesses a relatively higher conductivity. However, during bending, the Ge core breaks easily, and therefore, the conductance dropped to zero. By contrast, the Ge/Si nanowire demonstrates the decent mechanical stability. The reasons behind the conductance changes were then analyzed by DFT calculations. We simulated a bent Ge/Si core−shell nanowire with a mean diameter of ∼3 nm with the total amount of atoms of ∼1000 (Figure 5a). We also calculated the

of a nonbent Ge/Si nanowire. The nanowire was bent in one direction, as shown in Figure 4a(2−4) at the strain values of 1.5%, 2.5%, and 3.4%. Then the nanowire was positioned in such way as to bend it in the other direction, as displayed in Figure 4a(5−7) at strain values of 5.7%, 6.3%, and 7.0%. The probe was then moved backward to recover the nanowire straight status, as shown in Figure 4a(8). By applying an adequate EBID technique, the nanowire was firmly attached to the probe. Consequently, the nanowire was fully retracted from the sample probe, and this helped to measure its full length of 153 nm in Figure 4a(9). The nanowire was then moved to contact the sample probe at the position shown in Figure 4a(10), and this action makes it possible to apply an axial push. By moving the probe forward, the nanowire was again bent. The kink point moved from the place shown in Figure 4a(11) to the left side of the wire, as illustrated in Figure 4a(12). During all the manipulations, bias-current curves were constantly recorded. As displayed in Figure 4a, all I−V curves show the perfect Ohmic behavior, with a trend that the nanowire in bent conditions possesses a significantly higher conductance. In the cases of plots (2), (3), (4), (5), and (7), the bending strains are correlated with the conductance. However, the plots (11) and (12) do not follow this trend. The conductance decreases due to the inelastic bending, which leads to the irreversible crystallography changes. It should be noted that the nanowire shows elastic behavior at bending strains of up to 7.0%, which is identical to the results of Wang et al., i.e. Si nanowires show plastic deformation via partial dislocations’ movements at a critical strain of ∼6.9%.8 Also, the experimental elastic strain values for Si nanowires are 11.5%, 16%, and 6.5%; whereas these values of Ge nanowires are 17% and 7.5%.34 Thus, our experimental value at 7.0% for the present Ge/Si nanowire is close to some of the reported Si and Ge values. However, the critical points for elastic-plastic transformations should be detailed in further studies on nanowires with different geometric and morphological parameters. Figure S6 illustrates a series of standard high-resolution TEM images, with corresponding filtered images; additional images are presented in Figure S7. These images were consecutively taken with the intervals of 3−15 s during the process of gentle bending of the selected segment of a Ge/Si nanowire. Figure S6a corresponds to the initial stage at 0 s. The alignment conditions were not changed over the whole process. The probe was manually manipulated at a speed of about 1 nm/min, which is about 3 fine steps in a minute. Figure 4b-d presents the raw images, without any filtering or processing, but it is still quite obvious that the crystallography of the nanowire significantly changes, even during subnanometer probe displacements. The yellow dashed lines from the same place represent deformation-induced changes on lattice planes. Figure 4c shows a clear contrast in the middle structure with defects in its core, while the shell becomes more disordered than it was in Figure 4b. Figure 4d shows clear crystallography at the upper part of the image (with defects), whereas the lower part of the image recovers to a regular lattice structure. During the process, a series of dynamic images were taken. This allows us to conclude that the crystal structure is unstable under strain. The blue to yellow color alternations indicate the contrast change on the filtered image. Blue areas indicate the atomic structure, which is close to the standard intrinsic lattice, while the red color represents vacuum or nearly amorphous regions. More supporting images are given

Figure 5. (a) Deformation charge density of a bent Ge/Si nanowire. Color bar denotes the electron density in electron per Å3 units. Density of electronic states and band structure for the undeformed (b) and uniaxially compressed (c) Ge/Si nanowire. Black, green, and red curves denote total DOS and partial DOS of Si and Ge atoms, respectively. The Fermi level is vanished and marked by a horizontal dashed line.

deformation charge density (difference between the valence charge density and the sum of atomic valence charge densities) for the considered nanowire, see Figure 5b. The obtained distribution contains the regions of generalized electronic density in the compressed area caused by shortening of bonds between atoms. Electronic orbital overlapping in the compression region plays a major role in the formation of a conductive channel within the nanowire. We additionally supported this assumption by calculation of density of electronic states, as well as electronic bands of uniaxially compressed nanowire with a strain compared with the compression strain in the bent area. Originally, the energy band gap of the Ge/Si structure was 0.6 eV (Figure 5b). Figure F

DOI: 10.1021/acs.nanolett.8b03398 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

electronics, this would require fundamental mechanical research on emerging semiconducting materials. Herein, we demonstrated that Ge/Si core−shell nanowires have a Young’s modulus of ∼192 GPa and a tensile strength of up to 8.34 GPa. The electrical conductance generally increases under elastic bending; however, while the strain increases, its irregular distribution over the structures results in the amorphization of Si shells and fast phase transformations within the Ge cores, and these factors significantly decrease the conductance. Amorphization of the Si shell can be reversible at a low strain (2%) under slow deformations, whereas the phase transformation of the Ge core is irreversible and only happens at a very high strain (34%). The present analysis gives one timely clues which are valuable for the development of future flexible and foldable electronic and optoelectronic components. In addition, by means of first-principal calculations, electronic properties of bent Ge/Si core−shell nanowires were estimated. It was found that within a compressed nanowire the overlapping of electronic orbitals of Si and Ge atoms caused the formation of additional energy levels near the Fermi energy. Also using MD simulations, we confirmed the possibility of amorphous domain formation within the inner parts of bent nanowire Si shells.

5c displays the changes of the bands after the compression. The pronounced closing of the band gap occurred which confirms the observed trend of conductance increase, once the Ge/Si nanowire bears a bending strain. Nevertheless, the prediction of Si-shell amorphization allows us to suggest the conductivity dropping at the ultimately larger strain. This can explain the nonmonotonic behavior of the experimentally measured conductance in Figure 4a. Under very slow bending, in a few cases, the Ge/Si core− shell nanowire can survive even at a very high strain, as shown in Figure S10 and videos in the Supporting Information. As seen in Video 1, a nanowire remains intact at a probe moving rate of ∼0.4 nm/s, while a nanowire was cracked at a probe moving rate of ∼1 nm/s as displayed in Video 2. Surprisingly, the nanowire’s crystallography has not changed under such a process, most likely due to the amorphization of Si and diffusionless phase transformation of Ge, which happens at a speed of sound.35,36 The highly strained Ge core was occasionally obtained under etching of the Ge/Si structure by a hydrofluoric acid. Such a strained Ge nanowire was then characterized by STEM, as shown in Figure S11. These figures reveal that the Ge core generally maintains its intrinsic crystallography with a slight lattice strain on the compressed nanowires side. The core−shell nanowire, however, shows an irregular and locally random amorphization within the Si segment, as shown in Video 3. We analyzed the diffraction patterns, as shown in Figure S12, taken from the original and heavily bent nanowire, at a 34% strain. The original diffraction pattern shows diffraction spots of Si or Si and Ge together. In Figure S12b, a few more 111 diffraction spots appear for a bent structure; the inner spots represent Si reflections.37 The clear reflection splitting is related to the appearance of on-axis Ge crystallographic changes under bending. It should be noted that, unlike the diffraction ring in Figure 3c, the diffraction ring in Figure S12b appears to be more distant from the origin than characteristic 111 rings peculiar to the standard Si or Ge lattices, and a lattice spacing of 0.24 nm was obtained from such ring. This is likely related to a Sn-structure type of Ge (β phase), with the (200) lattice distance at 0.248 nm and the 101 distance at 0.240 nm.27 The semiconducting phases of Ge transform to a metallic β-Sn phase at ∼10−11 GPa.27,38 It has been proved that the β-Sn phase possesses excellent electronic properties.39 These results are in accordance with our in situ TEM experiments. Yet, more investigations are needed to explore this phase transformation in greater detail for isolated Ge and core−shell Ge/Si nanowires to shed light on the atomistic dynamics of this transition. In conclusion, even though handling of nanoscale building blocks is currently not easy and straighforward, it has been predicted that piezo-driven precise nanomanipulations, which can be performed using artificial intelligence, may be generally applied to future large scale bottom-up miniaturized device fabrications. Using such bottom-up technology, one may first prepare high quality building blocks, for instance, a Ge/Si core−shell structure, to have the shell material of high strength and the core material of high carrier mobility and then employ the delicate subnanometer resolution positioning of these building blocks onto predesigned locations. Ge/Si core−shell nanowires have been reported since 2002,37 and the high carrier mobilities and special physical properties have been evidenced in them. Nowadays, since there is a high public demand for flexible, bendable, and stretchable semiconducting



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.8b03398. Supporting images with descriptions (PDF) Video of slow bending to a high strain (AVI) Video of bending to a high strain with crack (AVI) Video of bending-induced Si amorphization (AVI)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected] (D.G.). *E-mail: [email protected] (C.Z.). ORCID

Chao Zhang: 0000-0001-5309-8484 Dmitry G. Kvashnin: 0000-0003-3320-6657 Joseph F. S. Fernando: 0000-0003-0129-359X Pavel B. Sorokin: 0000-0001-5248-1799 Naoki Fukata: 0000-0002-0986-8485 Dmitri Golberg: 0000-0003-2298-6539 Author Contributions

C.Z. and D.G.K. contributed equally to this work. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest. Additional videos for this article are available upon request from the corresponding authors.



ACKNOWLEDGMENTS This work was supported by the Australian Research Council (ARC) Laureate Fellowship FL160100089, Queensland University of Technology (QUT) Project 322170-0355/51, and the National Institute for Materials Science (NIMS), Grant PE2030. C.Z. acknowledges an Ian Potter Foundation G

DOI: 10.1021/acs.nanolett.8b03398 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Assisted Chemical Etching for Surface Enhanced Raman Spectroscopy. RSC Adv. 2016, 6 (96), 93649−93659. (14) Cui, H.; Li, S.; Deng, S.; Chen, H.; Wang, C. Flexible, Transparent, and Free-Standing Silicon Nanowire SERS Platform for in Situ Food Inspection. ACS Sensors 2017, 2 (3), 386−393. (15) Shin, J. Y.; Pi, K.; Jung, S.; Cho, D. I. D. Silicon-Nanowire Field-Effect Transistors on a Flexible Substrate Using Top-down MEMS Processes. In Proceedings of the IEEE International Conference on Micro Electro Mechanical Systems (MEMS); 2016; Vol. 2016−Feb, pp 469−472, DOI: 10.1109/MEMSYS.2016.7421663. (16) Kim, D. H.; Lee, S. J.; Lee, S. H.; Myoung, J.-M. Electrical Properties of Flexible Multi-Channel Si Nanowire Field-Effect Transistors Depending on the Number of Si Nanowires. Chem. Commun. 2016, 52 (42), 6938−6941. (17) Liu, L.; Lu, K.; Wang, T.; Liao, F.; Peng, M.; Shao, M. Flexible Piezoelectric Nanogenerators Based on Silicon Nanowire/α-Quartz Composites for Mechanical Energy Harvesting. Mater. Lett. 2015, 160, 222−226. (18) Choi, J.; Cho, K.; Kim, S. Flexible Thermoelectric Generators Composed of N-and p-Type Silicon Nanowires Fabricated by TopDown Method. Adv. Energy Mater. 2017, 7 (7), 1602138. (19) Zhang, C.; Cretu, O.; Kvashnin, D. G.; Kawamoto, N.; Mitome, M.; Wang, X.; Bando, Y.; Sorokin, P. B.; Golberg, D. Statistically Analyzed Photoresponse of Elastically Bent CdS Nanowires Probed by Light-Compatible In Situ High-Resolution TEM. Nano Lett. 2016, 16 (10), 6008−6013. (20) Zhang, C.; Xu, Z.; Kvashnin, D. G.; Tang, D. M.; Xue, Y. M.; Bando, Y.; Sorokin, P. B.; Golberg, D. Opto-Mechano-Electrical Tripling in ZnO Nanowires Probed by Photocurrent Spectroscopy in a High-Resolution Transmission Electron Microscope. Appl. Phys. Lett. 2015, 107 (9), 091103. (21) Soler, J. M.; Artacho, E.; Gale, J. D.; García, A.; Junquera, J.; Ordejón, P.; Sánchez-Portal, D. The SIESTA Method for Ab Initio Order-N Materials Simulation. J. Phys.: Condens. Matter 2002, 14 (11), 2745−2779. (22) Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 1995, 117 (1), 1−19. (23) Tersoff, J. New Empirical Approach for the Structure and Energy of Covalent Systems. Phys. Rev. B: Condens. Matter Mater. Phys. 1988, 37 (12), 6991−7000. (24) Fukata, N.; Mitome, M.; Sekiguchi, T.; Bando, Y.; Kirkham, M.; Hong, J. Il; Wang, Z. L.; Snyder, R. L. Characterization of Impurity Doping and Stress in Si/Ge and Ge/Si Core-Shell Nanowires. ACS Nano 2012, 6 (10), 8887−8895. (25) Fukata, N.; Yu, M.; Jevasuwan, W.; Takei, T.; Bando, Y.; Wu, W.; Wang, Z. L. Clear Experimental Demonstration of Hole Gas Accumulation in Ge/Si Core - Shell Nanowires. ACS Nano 2015, 9 (12), 12182−12188. (26) Okada, Y.; Tokumaru, Y. Precise Determination of Lattice Parameter and Thermal Expansion Coefficient of Silicon between 300 and 1500 K. J. Appl. Phys. 1984, 56 (2), 314−320. (27) Qadri, S. B.; Skelton, E. F.; Webb, A. W. High Pressure Studies of Ge Using Synchrotron Radiation. J. Appl. Phys. 1983, 54 (6), 3609−3611. (28) Liu, Q. L.; Zhao, C. W.; Xing, Y. M.; Su, S. J.; Cheng, B. W. Quantitative Strain Analysis of Misfit Dislocations in a Ge/Si Heterostructure Interface by Geometric Phase Analysis. Opt. Lasers Eng. 2012, 50 (5), 796−799. (29) Liu, X. W.; Hu, J.; Pan, B. C. The Composition-Dependent Mechanical Properties of Ge/Si Core-Shell Nanowires. Phys. E 2008, 40 (10), 3042−3048. (30) Smith, D. A.; Holmberg, V. C.; Lee, D. C.; Korgel, B. A. Young’s Modulus and Size-Dependent Mechanical Quality Factor of Nanoelectromechanical Germanium Nanowire Resonators. J. Phys. Chem. C 2008, 112 (29), 10725−10729. (31) Wang, Y.-C.; Zhang, W.; Wang, L.-Y.; Zhuang, Z.; Ma, E.; Li, J.; Shan, Z.-W. In Situ TEM Study of Deformation-Induced Crystallineto-Amorphous Transition in Silicon. NPG Asia Mater. 2016, 8 (7), e291.

grant (RMP2018000109). C.Z. and D.G. thank the Central Analytical Research Facility (CARF) of QUT and the Monash Centre for Electron Microscopy (MCEM) at Monash University (ARC funding LE0454166) for the experimental support. P.B.S., D.G.K., and D.G. acknowledge the financial support from the Ministry of Education and Science of the Russian Federation in the framework of Increase Competitiveness Program of NUST “MISiS” (No. K2-2017-082). P.B.S. is also grateful for the financial support of the RFBR, in the frame of the research project No. 16-32-60138 mol_a_dk. D.G.K. acknowledges the Grant of President of Russian Federation for government support of young Ph.D. scientists (MK3326.2017.2).



ABBREVIATIONS TEM, transmission electron microscopy/microscope; STEM, scanning transmission electron microscopy; HAADF, annular dark-field imaging; DFT, density functional theory; DOS, density of states; MD, molecular dynamics



REFERENCES

(1) Xiang, J.; Lu, W.; Hu, Y.; Wu, Y.; Yan, H.; Lieber, C. M. Ge/Si Nanowire Heterostructures as High-Performance Field-Effect Transistors. Nature 2006, 441 (7092), 489−493. (2) Wortman, J. J.; Evans, R. A. Young’s Modulus, Shear Modulus, and Poisson’s Ratio in Silicon and Germanium. J. Appl. Phys. 1965, 36 (1), 153−156. (3) Greytak, A. B.; Lauhon, L. J.; Gudiksen, M. S.; Lieber, C. M. Growth and Transport Properties of Complementary Germanium Nanowire Field-Effect Transistors. Appl. Phys. Lett. 2004, 84 (21), 4176−4178. (4) Shim, W.; Yao, J.; Lieber, C. M. Programmable Resistive-Switch Nanowire Transistor Logic Circuits. Nano Lett. 2014, 14 (9), 5430− 5436. (5) Yan, H.; Choe, H. S.; Nam, S.; Hu, Y.; Das, S.; Klemic, J. F.; Ellenbogen, J. C.; Lieber, C. M. Programmable Nanowire Circuits for Nanoprocessors. Nature 2011, 470 (7333), 240−244. (6) Zheng, K.; Han, X.; Wang, L.; Zhang, Y.; Yue, Y.; Qin, Y.; Zhang, X.; Zhang, Z. Atomic Mechanisms Governing the Elastic Limit and the Incipient Plasticity of Bending Si Nanowires. Nano Lett. 2009, 9 (6), 2471−2476. (7) Han, X. D.; Zheng, K.; Zhang, Y. F.; Zhang, X. N.; Zhang, Z.; Wang, Z. L. Low-Temperature In Situ Large-Strain Plasticity of Silicon Nanowires. Adv. Mater. 2007, 19 (16), 2112−2118. (8) Wang, L.; Kong, D.; Xin, T.; Shu, X.; Zheng, K.; Xiao, L.; Sha, X.; Lu, Y.; Zhang, Z.; Han, X.; et al. Deformation Mechanisms of Bent Si Nanowires Governed by the Sign and Magnitude of Strain. Appl. Phys. Lett. 2016, 108 (15), 151903. (9) Wang, L.; Zheng, K.; Zhang, Z.; Han, X. Direct Atomic-Scale Imaging about the Mechanisms of Ultralarge Bent Straining in Si Nanowires. Nano Lett. 2011, 11 (6), 2382−2385. (10) Tang, D.-M.; Ren, C.-L.; Wang, M.-S.; Wei, X.; Kawamoto, N.; Liu, C.; Bando, Y.; Mitome, M.; Fukata, N.; Golberg, D. Mechanical Properties of Si Nanowires as Revealed by in Situ Transmission Electron Microscopy and Molecular Dynamics Simulations. Nano Lett. 2012, 12 (4), 1898−1904. (11) Yue, Y.; Liu, P.; Zhang, Z.; Han, X.; Ma, E. Approaching the Theoretical Elastic Strain Limit in Copper Nanowires. Nano Lett. 2011, 11 (8), 3151−3155. (12) Wang, J.; Wang, H.; Zhang, B.; Wang, Y.; Lu, S.; Zhang, X. A Stable Flexible Silicon Nanowire Array as Anode for HighPerformance Lithium-Ion Batteries. Electrochim. Acta 2015, 176, 321−326. (13) Kara, S. A.; Keffous, A.; Giovannozzi, A. M.; Rossi, A. M.; Cara, E.; D’Ortenzi, L.; Sparnacci, K.; Boarino, L.; Gabouze, N.; Soukane, S. Fabrication of Flexible Silicon Nanowires by Self-Assembled Metal H

DOI: 10.1021/acs.nanolett.8b03398 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters (32) Millet, P.; Calka, A. Amorphization of Silicon and Germanium by Mechanical Grinding : Effect of Atmosphere and Milling Conditions. Mater. Sci. Eng., A 1994, 181-182, 1222−1226. (33) Maras, E.; Trushin, O.; Stukowski, A.; Ala-Nissila, T.; Jónsson, H. Global Transition Path Search for Dislocation Formation in Ge on Si(001). Comput. Phys. Commun. 2016, 205, 13−21. (34) Wang, S.; Shan, Z.; Huang, H. The Mechanical Properties of Nanowires. Adv. Sci. 2017, 4 (4), 1600332. (35) Ovshinsky, S. R. Optically Induced Phase Changes in Amorphous Materials. J. Non-Cryst. Solids 1992, 141 (1−3), 200− 203. (36) Liu, Y.; Wang, H.; Zhang, X. In Situ TEM Nanoindentation Studies on Stress-Induced Phase Transformations in Metallic Materials. JOM 2016, 68 (1), 226−234. (37) Lauhon, L. J.; Gudiksen, M. S.; Wang, D.; Lieber, C. M. Epitaxial Core−shell and Core−multishell Nanowire Heterostructures. Nature 2002, 420 (6911), 57−61. (38) Werner, A.; Sanjurjo, J. A.; Cardona, M. X-Rays Investigation of the Α→β Phase Transition in the GexSi1-x Solid Solutions at High Pressure. Solid State Commun. 1982, 44 (2), 155−158. (39) Fan, Q.; Chai, C.; Wei, Q.; Zhou, P.; Yang, Y. Two Novel Ge Phases and Their Si−Ge Alloys with Excellent Electronic and Optical Properties. Mater. Des. 2017, 132, 539−551.

I

DOI: 10.1021/acs.nanolett.8b03398 Nano Lett. XXXX, XXX, XXX−XXX