Strain Mapping and Raman Spectroscopy of Bent GaP and GaAs Nanowires.

Strain engineering of nanowires (NWs) has been recognized as a powerful strategy for tuning the optical and electronic properties of nanoscale semiconductors. Therefore, the characterization of the strains with nanometer-scale spatial resolution is of great importance for various promising applications. In the present work, we synthesized single-crystalline zinc blende phase GaP and GaAs NWs using the chemical vapor transport method and visualized their bending strains (up to 3%) with high precision using the nanobeam electron diffraction technique. The strain mapping at all crystallographic axes revealed that (i) maximum strain exists along the growth direction ([111]) with the tensile and compressive strains at the outer and inner parts, respectively; (ii) the opposite strains appeared along the perpendicular direction ([2̅11]); and (iii) the tensile strain was larger than the coexisting compressive strain at all axes. The Raman spectrum collected for individual bent NWs showed the peak broadening and red shift of the transverse optical modes that were well-correlated with the strain maps. These results are consistent with the larger mechanical modulus of GaP than that of GaAs. Our work provides new insight into the bending strain of III-V semiconductors, which is of paramount importance in the performance of flexible or bendable electronics.

Introduction

One-dimensional semiconductor nanostructures [typically nanowires (NWs)] have attracted much attention over the past two decades because they are promising bottom-up building blocks of future nanodevices.[1] Recently, strain engineering of NWs has been recognized as a powerful strategy for tuning the optical and electronic properties of nanoscale semiconductors, and thus it is receiving an increasing amount of attention.[3] The prominent properties of tensile- and compressive-strained NWs are smaller and larger band gaps, respectively, as compared to those of corresponding unstrained NWs.[2−8] For bent (or buckled) NWs, a red-shifted emission has been observed, which has been explained by the piezotronic effect, in which under a piezoelectric field, the number of photons emitted from the outer surface of the NW is much larger than that from the inner surface.[9−13] The outstanding piezoelectric properties of ZnO and CdS NWs have facilitated the development of various piezotronic and piezo-phototronic devices such as nanogenerators, field-effect transistors, and light-emitting diodes, which rely on the piezoelectric potential created along the strained NWs.[3,14−16] Moreover, the enhanced carrier mobility or photocurrent of strained NWs has attracted widespread research interest owing to their great potential in a rich variety of applications.[17−24]

To expedite the fabrication of advanced flexible electronic devices, it is essential to examine the strains with nanometer-scale spatial resolution so that the modification of the properties can be better understood. Such nanoscale examination can be carried out using transmission electron microscopy (TEM) techniques such as aberration-corrected high-resolution TEM (HRTEM),[25] electron diffraction (e.g., convergent beam and nanobeam),[26−30] geometrical phase analysis of high-angle annular dark-field scanning TEM,[31] and dark-field electron holography.[32] In particular, a precession-assisted nanobeam electron diffraction (NBED) technique had been developed to visualize the strain of materials with high precision.[28−30] This technique makes use of electron beam precession together with spot diffraction pattern recognition, resulting in reliable strain maps with a spatial resolution down to 2 nm and excellent precision (0.02%) on a field emission gun TEM.[33,34] Furthermore, the NBED technique has a great advantage to provide the strain distribution at a specific crystallographic axis. Nevertheless, strain mapping using NBED for strained NWs has not been reported yet. The strain distribution at individual crystallographic axis would provide valuable information to understand their novel properties.

In the present work, GaP and GaAs NWs were synthesized using the chemical vapor transport method. The NWs exhibited a single-crystalline zinc blende (ZB) phase with a uniform growth direction ([111]). The NWs were bent by the mechanical buckling of poly(dimethylsilioxane) (PDMS), which transformed the initially straight NWs into wavy shapes via releasing of the prestrain.[35] For strain mapping of the bent NWs, we performed NBED processing using a NanoMEGAS’s “Digistar” add-on device.[33] Moreover, because Raman spectroscopy is one of the most straightforward and broadly used techniques to monitor the strains of the NWs,[6,19,36−40] we collected Raman spectra for individual GaP (or GaAs) NWs. The aim of the current study was to correlate the bending strains of the NWs with their Raman spectrum to provide insight into lattice distortion within the NWs.

Results and Discussion

We synthesized GaP and GaAs NWs on Si substrates with high yield using a typical vapor–liquid–solid growth mechanism, which makes use of Au nanoparticles as catalysts for growth. The X-ray diffraction (XRD) peaks of the GaP and GaAs NWs were assigned to the ZB phase of GaP (JCPDS Card no. 32-0397; a = 5.450 Å) and GaAs (JCPDS Card no. 80-0016; a = 5.654 Å), respectively, as shown in Figure S1 (Supporting Information).

The scanning electron microscopy (SEM) and HRTEM images show that GaP and GaAs NWs all had a straight and smooth surface, with no amorphous oxide layer shell (Figure a,b, respectively). The average diameter is 160 nm. Figure c,d shows the HRTEM and corresponding fast Fourier transform (FFT) images (zone axis = [011̅]) of the GaP and GaAs NWs, respectively, revealing their single-crystalline ZB phase and [111] growth direction. The d-spacings between neighboring (111) planes of the GaP and GaAs NWs were 3.15 and 3.26 Å, respectively, which are the same as those of the bulk: d111 = 3.1460 Å (GaP) and 3.2643 Å (GaAs).

Figure 1

(a) SEM and (b) HRTEM images of NWs, showing a general morphology. The HRTEM and corresponding FFT images (zone axis = [011̅]) of individual (c) GaP and (d) GaAs NWs.

We then examined the strain mapping of NWs subjected to bending deformation. The NWs were directly transferred from the sample substrates onto a holey carbon TEM grid by rubbing. Some of the NWs were lying bent, which provided an opportunity to evaluate their atomic structures under bending strain. Thus, bent NWs with various curvatures were examined. The present NBED technique allows the relative extent of lattice fringe distortion across the NWs to be evaluated with respect to an unstrained area within the projected lattice, which is taken as a reference. The two-dimensional mapping of the lattice expansion (positive value) or contraction (negative value) relative to a reference (d0), that is, (d – d0)/d0 × 100%, was carried out for each NBED pattern spot. The reference was taken at the center of the NW (marked by the dots in the maps), where presumably no lattice deformation occurred. The precision of the strain was <0.03%, and the spatial resolution was 4 nm.

Figure shows an HRTEM image of a GaP NW along the [011̅] zone axis, the NBED pattern for the marked region in the HRTEM image, and the corresponding strain maps and profiles. The NW diameter was 150 nm, and the local diameter of curvature was 5.6 μm (Figure a). The bending strain (ε) of the NW was estimated using the formula ε = r/(r + R) ≈ r/R, where R is the local radius of curvature and r is the radius of the NW.[9,11] The calculated ε value was 2.7%. The strain mapping was carried out for the [111], [011], [1̅11], and [2̅11] directions, as indicated in the NBED. Tensile (red) and compressive (blue) strains formed uniformly along the NW axis, with the unstrained (green) center region. Different scale bars were used to represent properly the range of strain for the selected direction. The strain profile (at the rectangular region marked in the maps) shows a change from tensile to compressive (or vice versa) strain across the diameter (see the graphs). Strain mapping was also obtained for straight NWs, which confirmed that the strain was formed by the bending (Supporting Information, Figure S2).

Figure 2

HRTEM images, NBED pattern (zone axis = [011̅]), and its corresponding strain maps of the GaP NWs with a bending strain of (a) ε = 2.7% and (b) ε = 2%, along the [111], [011], [1̅11], and [2̅11] directions. Strain maps of the lattice expansion (positive value) or contraction (negative value) relative to a reference (d0) at the center of the NW (marked by the dots), where no lattice deformation presumably occurs, that is, (d – d0)/d0, were taken on the individual direction. The scale bars represent the strain range. The strain profile is for the red line region in the strain map.

In the [111] growth direction, the tensile and compressive strains existed at the outer and inner regions, respectively, with 3.0 and −2.5%. These values are close to the bending strain (ε = 2.7%) calculated above. The tensile and compressive strains at the edge region decrease to 2.2 and −1.2% along the [011] direction, respectively, which was tilted 35.3° from the growth direction. Along the [1̅11] direction, which was tilted 70.5° from the growth direction, there was negligible strains. In the perpendicular direction ([2̅11]), the opposite strains formed; the compressive and tensile strains at the outer and inner regions are −0.45 and 0.6%, respectively.

For GaP NWs with ε = 2%, the HRTEM image, NBED pattern (zone axis = [011̅]), and the corresponding strain maps for the [111], [011], [1̅11], and [2̅11] directions are shown in Figure b. The strain profile shows a change of the strain across the diameter (see the graphs). In the [111] growth direction, the tensile and compressive strains at the outer and inner regions appeared with 2.2 and −1.6%, respectively. The tensile and compressive strains decreased to 1 and −0.5%, respectively, along the [011] direction. Along the [1̅11] direction, the strain became almost zero. For the perpendicular direction ([2̅11]), the compressive and tensile strains existed at the outer and inner regions, respectively, which were about −0.2 and 0.6%.

The strain mapping obtained for other GaP NWs consistently showed that the strain was maximized along the growth direction that it decreased at the higher tilt directions, whereas the opposite strain appeared along the perpendicular direction. We plotted the tensile/compressive strain versus the orientation angle of the axis to show the dependence of the strain on the direction of axis in the GaP NW (Figure a). The strains were obtained at the same region for many axes, including [133], [311], and [100] directions, which were tilted from the [111] growth direction by 17.6°, 27.4°, and 54.7°, respectively. The strain was almost zero at ∼70°. The tensile strain along the perpendicular direction was about one-fifth that along the growth direction. It was also found that the tensile strain was always larger than the coexisting compressive strains along all axes, which means that the bent NWs possess an excess tensile strain.

Figure 3

Tensile (positive value) and compressive (negative value) strains as a function of the orientation angle relative to the growth direction ([111]); (a) GaP and (b) GaAs NWs at various bending strains (ε). The error range for each data point is marked by the lines.

We also measured the lattice-resolved TEM images of the bent GaP NWs and confirmed that the d-spacings at the outer and inner regions of the NW were expanded and reduced, respectively, as shown in Figure S3 (Supporting Information). The lattice expansion (d111) along the growth direction is larger than the lattice contraction, which is consistent with the strain mapping we observed.

For the GaAs NWs with ε = 3 and 2%, the HRTEM images, NBED patterns at the [011̅] zone axis, and their corresponding strain maps along the growth and perpendicular directions are shown in Figure a,b, respectively. The strain profile across the diameter (at the region marked by the red line in the maps) is also shown. The results are summarized as follows: (i) the tensile and compressive strains at the outer and inner regions, respectively, were greatest along the growth direction, that is, 3% tensile and −2.5% compressive strains for ε = 3%, and 2.1% tensile and −1.9% compressive strains for ε = 2%. (ii) Along the perpendicular direction, the compressive and tensile strains formed at the outer and inner regions, respectively, were −0.8 and 1.3% for ε = 3%, and −0.3 and 1.3% for ε = 2%. (iii) The tensile strain was always larger than the coexisting compressive strain.

Figure 4

HRTEM images, NBED patterns (zone axis = [011̅]), and their corresponding strain maps of the GaAs NWs with a bending strain of (a) ε = 3% and (b) ε = 2%, along the growth direction and its perpendicular direction. Strain maps of the lattice expansion (positive value) or contraction (negative value) relative to a reference (d0) at the center of the NW (marked by the dots), where no lattice deformation presumably occurs, that is, (d – d0)/d0, were taken on the [111] and [2̅11] directions. The scale bars represent the strain range. The strain profile is for the red line region in the strain map.

The strain versus axis for the GaAs NWs showed that the strain decreased to almost zero along axes that were tilted ∼60° from the growth direction (see Figure b). Remarkably, the GaAs NWs exhibited larger strain along the perpendicular direction as compared to the GaP NWs. The tensile strain along the perpendicular direction was approximately one-half that along the growth direction. Careful examination of the strain maps along the growth direction showed that the area of the strained region is wider for the GaAs NWs as compared to that of GaP NWs. The line profiles of GaAs NWs consisted of a plateau edge region (∼20 nm) with a steep crossover slope at the center (than that of the GaP NWs) due to a narrower unstrained green region. In contrast, the line profiles of GaP NWs showed a gradual change of the strain over the cross section. It was also found that the GaAs NWs hold a larger excess tensile strain than the GaP NWs.

We now compare quantitatively the opposite strain that occurred along the perpendicular direction for the GaP and GaAs NWs. When the tensile strain at the outer region along the growth direction was 3%, the compressive strain formed perpendicularly as −0.45 and −0.8% for the GaP and GaAs NWs, respectively. The compressive strain at the inner region along the growth direction was −2.5%, whereas the tensile strain formed perpendicularly as 0.6 and 1.3% for the GaP and GaAs NWs, respectively. Thus, the GaP NWs exhibited two times less change of the lattice constant in the opposite direction toward the same bending strain. These results mean that the GaP NWs would be more rigid than the GaAs NWs. The Young’s moduli of bulk GaP and GaAs materials were experimentally determined to be 102.8 and 85.3 GPa, respectively; the respective bulk moduli were 88.8 and 75.5 GPa, respectively, showing the larger value for GaP.[41] Therefore, the higher rigidity of the GaP NWs as compared to that of the GaAs NWs is consistent with the mechanical modulus of the bulk phase.

Another important observation is that the tensile strain was larger than the coexisting compressive strain along all axes, suggesting that the optical properties of the bent NWs may be dominated by the tensile strain. It has been observed that the emission of bent (or buckled) NWs red shifts due to a piezoelectric field.[9−13] We predict that the band gap of the bent GaP and GaAs NWs would decrease owing to the dominant tensile strain. Furthermore, the larger excessive tensile strain of GaAs NWs suggests the larger band gap decrease as compared to that of GaP NWs.

We then measured the Raman spectra of individual GaP and GaAs NWs having bending strains. The procedure for obtaining in-plane buckled NWs with various bending strains (ε < 3.5%) and the use of a micro-Raman spectrometer for measuring the Raman scattering spectrum of individual NWs are explained in Figure S4 (Supporting Information). The strain of the bent NWs was decreased by stretching the PDMS substrates using a mechanical micrometer stage, and the Raman spectrum was monitored as the curvature of the NW increased. To calculate the ε value, the curvature radii and diameters of the bent NWs in the optical images were calibrated by tracing the bent NW backbone in the TEM or atomic force microscopy images. The Raman spectrum was measured at positions close to the regions with the highest curvature and consequently at the points (green spots) of highest strain of the NWs, as shown in the optical microscopy images (Figure a). Each acquired Raman spectrum represents the total average of the spectrum excited from the sampled volume with a diameter of 1 μm.

Figure 5

(a) Optical images show the strain of bent NWs (ε = 0, 1, 2, and 3%). Raman spectra of the (b) GaP and (c) GaAs NWs at various strains. The data points were resolved into bands (colored lines) using a Voigt function. The black line represents the sum of the resolved TO and LO bands. The peak shift of TO modes is illustrated by the green line. (d) Peak broadening (FWHM) of the TO mode and (e) its peak shift as a function of bending strain (ε). Raman signals were obtained using an Ar ion laser with a wavelength of 514.5 nm.

Figure b shows the Raman spectra of the GaP NW with various bending strains. The peaks were resolved using the Voigt function. The black line representing the sum of the resolved bands (represented by colored lines) shows good fit to the data points. The spectra of the straight NW (ε = 0%) consisted of a transverse optical (TO) mode peak at 366 cm–1 and a longitudinal optical (LO) mode peak at 400 cm–1.[42] There is a shoulder peak at 360 cm–1, which is red-shifted from the TO mode by 6 cm–1. Bakker explained its origin as a frequency-dependent damping oscillator.[43] As the bending strain increased, significant peak broadening was observed, with a peak shift to a lower frequency region. At ε = 0%, the full width at half-maximum (FWHM) of the TO peak was 4.5 ± 1 cm–1. As the bending strain increased to ε = 3%, the FWHM increased to 15 ± 1 cm–1. The peak shift (Δν) was approximately −1.2 cm–1 for ε = 3%. The peak shift is illustrated by the green line. In the case of the LO mode, it was difficult to analyze due to the low intensity. Therefore, we focused on the change of the TO mode peak toward the bending strain. Previous studies of the polarized Raman spectrum showed that the peak intensity of the TO mode mainly originated from the [111] growth direction.[44] We performed polarization-dependent Raman scattering experiments to confirm this model (see Supporting Information, Figure S5).

Figure c shows the Raman spectra of the GaAs NWs, under various bending strains. As the bending strain increased, significant peak broadening was consistently observed. At ε = 0%, the TO and LO mode peaks at 269 and 288 cm–1, respectively, were fitted using the Voigt function. The FWHM of the TO peak was 6 ± 1 cm–1. As the ε increased to 3%, the FWHM increased to 15 ± 1 cm–1. We also observed a red shift of approximately −1.4 cm–1 for ε = 3%. We performed Raman spectrum analysis for a large number of NWs and plotted the FWHM as a function of ε, which is shown in Figure d. The linear fit provided a peak broadening rate (slope) of 3.4 and 3.0 cm–1 for the GaP and GaAs NWs, respectively, showing presumably the similar peak broadening. We plotted the peak position versus ε, which is shown in Figure e; the figure shows a red shift upon increasing the strain. The linear fit provides peak shift rates of −0.40 and −0.47 cm–1, respectively, for the GaP and GaAs NWs. To our best knowledge, the peak shift has never been reported for the bent NWs so far.

Previous studies on bent InP NWs reported that peak broadening of the Raman mode increases with increasing bending strain.[36] Calculations showed that peak broadening of the bent InP NWs resulted from both the tensile strain-induced red shift and compressive strain-induced blue shift. The present strain mapping visualizes clearly the coexisting tensile and compressive strains that resulted in the broadening of the Raman peak. The similar peak broadening of GaP and GaAs NWs is ascribed to the same maximum tensile/compressive strains at the edge region along the growth direction. The red shift of the Raman peaks is probably related more to the larger tensile strain than the coexisting compressive strain that we observed from the strain mapping. A larger red shift for the GaAs NWs may be due to the larger excess tensile strain than that of the GaP NWs.

The Grüneisen parameter of the TO mode, which describes the shift in the frequency of a phonon due to a variation of the lattice’s volume as a result of pressure change, was measured to be 0.88 and 1.3, respectively, for bulk GaP and GaAs (also NWs).[40,45] The value of the Grüneisen parameter predicts a larger Raman peak shift for GaAs than that for GaP. Because the Young’s moduli of bulk GaP and GaAs are 102.8 and 85.3 GPa, respectively, the red shift rates were calculated to be −3.9 and −5.5 cm–1/TPa, for the GaP and GaAs NWs, respectively. The GaAs NWs show 1.4 times larger peak shift than the GaP NWs, which is consistent with the ratio of the Grüneisen parameters (1.5). Therefore, the Raman spectrum of the bent NWs is correlated with the mechanical modulus. We suggest that the peak broadening and red shift of the Raman TO modes can be used to estimate precisely the bending strain of both GaP and GaAs NWs.

Conclusions

Single-crystalline ZB phase GaP and GaAs NWs (avg. diameter = 160 nm) were synthesized with the [111] growth direction using the chemical vapor transport method. The NBED technique was used to provide strain maps of the bent NWs (ε = 1.5–3%) at the individual crystallographic axis. The bent NWs exhibited the largest strain along the growth direction with the tensile and compressive strains at the outer and inner regions, respectively, and the value of the strain is consistent with the bending strain (ε). The strain mapping revealed two more important results: (i) the opposite strains existed along the perpendicular direction ([2̅11]) and (ii) the tensile strain was larger than the compressive strain at all axes. The opposite strain was less for the GaP NWs, which could be explained by their higher rigidity as compared to that of the GaAs NWs.

The NWs were bent by the mechanical buckling of PDMS, which transformed the straight NWs into wavy shapes via the releasing of the prestrain. The Raman spectra of individual NWs showed peak broadening and red shifts of the TO mode under bending strains (ε = 0–3.5%). The peak broadening rate was approximately 3 cm–1 for both GaP and GaAs NWs. Strain maps confirmed that the peak broadening resulted from the tensile/compressive strain along the growth direction. The red shift could be determined by the dominant tensile strain that was revealed by the strain mapping. The Raman peak change was also consistent with the larger mechanical modulus of GaP than that of GaAs.

Experimental Details: Materials and Methods

GaAs (99.999%, Alfa Aesar) or GaP (99.999%, Alfa Aesar) powders were placed inside a quartz tube reactor. A silicon (Si) substrate, on which a 3 nm thick Au film was deposited, was positioned 10 cm from the powder source. Ar gas was continuously flowed at a rate of 200 sccm during the whole growth process. H2 gas flow was added during the NW growth, with a rate of 10 sccm. The temperature of the powder sources was set to 1000 °C and that of the Si substrate to 800 °C.

The structure and the composition of the products were analyzed by SEM (Hitachi S-4700), field emission TEM (FEI Tecnai G2, 200 kV), high-voltage TEM (JEOL JEM ARM 1300S, 1.25 MV), and energy-dispersive X-ray fluorescence spectroscopy. FFT images were generated by the inversion of the TEM images using Digital Micrograph GMS1.4 software (Gatan Inc.). A tilt holder (Dual Orientation Tomography Holder 927, Gatan Co.) was used for the TEM measurements. High-resolution XRD patterns were obtained using the 9B and three-dimensional beam lines of the Pohang Light Source (PLS) with monochromatic radiation.

The strain mappings were acquired using TEM (FEI TECNAI G2 200 kV) operated at 200 keV. Both scanning and precession were enabled through a NanoMEGAS’s Digistar system hardwired into the microscope scan control boards. The system was controlled through the NanoMEGAS TOPSPIN software package using a Stingray fast capture charge-coupled device camera to capture the diffraction patterns as seen on the small viewing screen of the microscope. The precession diffraction measurements were based on the principle of NBED, and scanning TEM coils were used for the precession of the electron beam. As a consequence of using NBED, a script written in DigitalMicrograph was required to scan the beam across the specimens and individually save the acquired patterns. This was done using the software TOPSPIN so that the deformation maps could be acquired using many thousands of diffraction patterns. To measure the lattice deformation in the NW, an array of 100 × 50 diffraction patterns was acquired using a step size of only 1.75 nm. A precession angle of 0.25° was used to provide a probe of less than 2 nm in diameter.

For the formation of bent NWs, elastomeric substrates of PDMS (SYLGARD 184, Dow Corning) were prepared by mixing a base resin and a curing agent in a ratio of 10:1. The bubbles in the unset PDMS film were removed in a vacuum chamber, and the films were cured at 70 °C for 12 h. The PDMS films were then cut into 40 × 10 mm pieces. The PDMS substrates were uniaxially stretched 30–100% using a custom-made strain stage. A contact printing method applied to the NWs as follows. The as-grown NWs on the Si substrate were rubbed onto a receiver Si substrate (with a 50 nm thick oxide layer on the top) under an applied load. The aligned NWs were then transferred to the prestretched PDMS substrates by applying a uniaxial tensile force along the axis of the NWs. Releasing the prestretched substrate induced the compressive strain that generated the curvature of the NWs (see Supporting Information, Figure S4). To estimate the diameter and curvature of the bent NWs, the topology measurements were carried out using an atomic force microscopy system (XE-100 AFM, Park Systems).

The Raman spectra of individual NWs were measured with a home-made micro-Raman system at room temperature. The spectral resolution is about 1 cm–1. Raman scattering signals were obtained in a back-scattering configuration using a 100× objective (NA 0.9) and an Ar ion laser with a wavelength of 514.5 nm. The laser spot size was approximately 1 μm, which, in combination with the imaging capabilities of the microscope, allowed for routine single-wire identification. A laser power below 0.5 mW was used to avoid heating effects. We used a mechanical stage to stretch the NW-embedded PDMS substrates and release the strains of the wavy NWs in a controlled way using a micrometer.