Temperature-dependent catalyst-free growth of ZnO nanostructures on ...

Journal of the Korean Physical Society, Vol. 60, No. 11, June 2012, pp. 1877∼1885

Temperature-dependent Catalyst-free Growth of ZnO Nanostructures on Si and SiO2 /Si Substrates via Thermal Evaporation Hyo Jin Kim, Sang Han Park, Woo-Jung Lee, Jung Min Bae, Ji Min Chae and Mann-Ho Cho∗ Institute of Physics and Applied Physics, Yonsei University, Seoul 120-749, Korea (Received 23 November 2011, in final form 24 April 2012) The catalyst-free growth of ZnO nanostructures on Si and SiO2 /Si substrates as a function of substrate temperature was carried out using a thermal evaporation method. We observed that the shapes and the morphologies of the ZnO nanostructures could be controlled by using the substrate temperature and the presence of an oxide layer on the surface of the substrate. The shape of the ZnO nanostructure was changed from an embossed nanocantilever to a nanowire as the growth temperature was decreased from 500 ◦ C to 430 ◦ C. At 360 ◦ C, a winding stem-like nanostructure with thin and short branch nanowires on the facet of the nanostructure was produced. In particular, at a growth temperature of 430 ◦ C, a ZnO buffer layer was formed during the initial growth when an Si substrate was used. However, no ZnO buffer layer was observed when a SiO2 /Si substrate was used. The formation of a buffer layer significantly affected the crystalline structure. Defects were observed in the embossed nanocantilevers and the nanowires grown on SiO2 /Si but not in those grown on Si. PACS numbers: 61.46.-w, 68.37.Lp, 68.55.Ac, 68.65.-k Keywords: ZnO, Nanostructure, Surface energy, Substrate temperature, Initial growth DOI: 10.3938/jkps.60.1877

I. INTRODUCTION Nanostructured ZnO materials have attracted significant interest in recent years due to their excellent performance in electronics, optics and photonics compared to conventional bulk materials [1–3]. Novel electrical, mechanical, chemical and optical properties are achieved by reducing the sizes of the nanostructures as a result of the high surface-to-volume ratio and quantum confinement effects. Zinc oxide has a wide bandgap energy of 3.37 eV and a large exciton binding energy at room temperature that make it a promising material for UV light detection [4,5], UV nanolasers [6], sensors [7,8], light emitting diodes [9,10], and solar cells [11,12]. The lack of a center of symmetry in the wurtzite structure of ZnO makes it suitable for applications in piezoelectric devices [13]. Differently shaped ZnO nanostructures have been synthesized by using various methods, including chemical vapor deposition [14,15], molecular-beam epitaxy [16], hydrothermal methods [17], pulsed laser deposition [18], and thermal evaporation [19,20]. Among those available methods, the thermal evaporation and the vapor-phase transport method have been extensively used because well-formed nanostructures with high crystalline quality can be grown in a simple, cost-effective and scalable way [21–23]. Metal catalysts have been used to assist ∗ E-mail:

[email protected]

nanostructure growth by capturing gas phase precursors. However, a number of studies report that the metal catalysts are segregated in the nanowires, critically affecting the formation of nanostructures. To avoid the negative effects resulting from metal catalysts, ZnO nanostructures have been synthesized by using catalyst-free methods [24–26]. The self-catalyzed vapor-liquid-solid (VLS) process with Zn or ZnOx liquid droplets is generally used for the catalyst-free method. A two-step process has occasionally been used to synthesize well-aligned ZnO nanostructures on substrates such as silicon that have a large mismatch with ZnO crystal structures [19,27]. Sapphire, which has the same crystal structure and a good lattice match with ZnO, is a suitable substrate to synthesize high-quality ZnO nanostructures. However, this substrate is expensive, which prevents its widespread use in relatively inexpensive devices. On the other hand, a Si substrate is relatively inexpensive and appropriate for use in the Si-based industry. Although silicon has a large mismatch with the ZnO crystal structure and there are difficulties in the synthesis of epitaxial ZnO nanostructures, silicon is commonly used to synthesize ZnO nanostructures. On the other hand, the synthesis of ZnO nanostructures on SiO2 is rarely conducted due to the amorphous structure of the substrate, which is known to deteriorate the crystalline quality of the ZnO structure [28]. In this work, both Si and SiO2 /Si substrates were used

-1877-

-1878-

Journal of the Korean Physical Society, Vol. 60, No. 11, June 2012

Fig. 1. (Color online) Schematic diagram of the experimental setup for the growth of ZnO nanostructures in different temperature regions on Si and SiO2 /Si substrates.

for a direct comparison, which has not been done before, of the growth kinetics of ZnO nanostructures at the initial growth stage due to the influence of different surface states for various growth temperatures. At 500 ◦ C, embossed nanocantilevers were grown on both Si and SiO2 /Si substrates. However, nanowires were predominantly grown on Si while embossed nanocantilevers were grown on SiO2 /Si. At 430 ◦ C, the nanostructures of ZnO were significantly changed; i.e., only nanowires were observed on both substrates. A ZnO buffer layer was formed during the initial growth on the Si substrate while the nanowires were grown directly on the SiO2 /Si interface without a ZnO buffer layer. Since the surface energy of the clean Si surface is high, the formation of the buffer layer can lower the interfacial energy between ZnO and Si. Finally, the buffer layer prompted the growth of high-quality ZnO nanowires. At 360 ◦ C, winding stemlike nanostructures with thin and short branch nanowires on facets were grown on both substrates. A buffer layer was also formed at this temperature, but only on the Si substrate. Defects were only observed in the embossed nanocantilevers and the nanowires grown on SiO2 /Si. From these results, we concluded that the ZnO nanostructures could be controlled by using the growth conditions. In particular, the formation of a buffer layer affects the crystalline quality of the grown nanostructures.

II. EXPERIMENTAL DETAILS The catalyst-free synthesis of ZnO nanostructures by using thermal evaporation was conducted in a horizontal tube furnace system, as shown in Fig. 1. A mixture of ZnO (0.16 g, 99.999%, Sigma-Aldrich) and graphite (0.04 g, 99.5%, CERAC) powders was loaded in an alumina boat as the source material and was positioned in the heating zone of the furnace. Three pieces each of Si(100) and SiO2 /Si wafers were cleaned in organic solvent and rinsed with deionized water. A surface-oxidefree Si substrate was prepared by cleaning with HF acid. The SiO2 /Si substrate was prepared by using the following procedure: The Si substrate was cleaned by using HF acid to remove the native oxide layer, and SiO2 with a 200-nm-thick oxide layer was grown on the Si substrate by using thermal oxidation. After they had been cleaned,

the pieces of each wafer were arranged in parallel on the downstream side at distances of about 10 cm, 11.5 cm, and 13 cm from the boat containing the source material. At these positions, three different growth temperatures could be used: relatively high: 500 ◦ C, intermediate: 430 ◦ C, near the melting point of Zn (419 ◦ C), and low: 360 ◦ C, below the melting point of Zn. The chamber was evacuated to 360 mTorr by using a rotary vacuum pump. High-purity Ar (99.999%) gas was introduced into the reactor at a constant flow of 14 sccm during the entire process. The furnace was heated for 40 minutes, and the temperature of heating zone reached 1030 ◦ C. While the furnace was being maintained at this temperature for 1 hour, a mixing gas of O2 (5%) and Ar (95%) was introduced at 6 sccm. The temperatures of each substrate were measured using a thermocouple. The measured temperatures of the substrates were 500 ◦ C ± 10 ◦ C (A1, B1), 430 ◦ C ± 10 ◦ C (A2, B2), and 360 ◦ C ± 10 ◦ C (A3, B3), as shown in Fig. 1. After the growth process, the furnace was cooled to room temperature. The morphologies and the compositions of the products were examined by using field emission scanning electron microscopy (FESEM, JEOL JSM-6500F) with energy dispersive spectrometry (EDS). The crystalline structures of the nanostructures were investigated by using transmission electron microscopy (TEM, TECNAI F20) at 200 kV. The crystal structures of the products were identified by using X-ray diffraction (XRD) with Cu Kα radiation.

III. DATA AND RESULTS Figure 2 shows typical FESEM images of as-grown samples at different temperatures (500 ◦ C ± 10 ◦ C, 430 ◦ C ± 10 ◦ C, 360 ◦ C ± 10 ◦ C) on Si (left side) and SiO2 /Si (right side). The density of the ZnO nanowires grown on Si at 500 ◦ C was very low, as shown in Fig. 2(a). The diameters of the ZnO nanowires were normally 18 nm – 50 nm, and their lengths ranged from hundreds of nanometers to 1.5 µm. These nanowires had non-homogeneous diameters along their lengths. The diameters at the upper regions were slightly thinner than those of the bottom regions. Nanowires with primarily random orientations were produced while some embossed nanocantilevers were locally grown on the Si substrate. The inset of Fig. 2(a) reveals the grown embossed nanocantilevers which were tens of micrometers long and were not of uniform width. The morphology of the product grown on the SiO2 /Si substrate is significantly different from that of the product grown on a Si substrate even at the same growth temperature. In this case, ZnO embossed nanocantilevers were grown with high density. Figure 2(b) shows the characteristic shapes of the ZnO embossed nanocantilevers grown on SiO2 /Si at 500 ◦ C. An embossed nanocantilever consists of a continuous connection of thick bulging parts and thin

Temperature-dependent Catalyst-free Growth of ZnO Nanostructures · · · – Hyo Jin Kim et al.

-1879-

Fig. 3. (Color online) XRD data of the ZnO nanostructures grown under different conditions. (a), (c), and (e) are the XRD data of the ZnO nanostructures grown on Si at 500 ◦ C, 430 ◦ C, and 360 ◦ C, respectively. (b), (d), and (f) correspond to the XRD data of the ZnO nanostructures grown on SiO2/Si at 500 ◦ C, 430 ◦ C, and 360 ◦ C, respectively.

Fig. 2. FESEM images of the grown ZnO nanostructures in various growth conditions: (a) and (b) are the nanostructures grown at 500 ◦ C on Si (A1) and SiO2 /Si (B1), respectively; (a) and (b) show the characteristic shapes of the nanowires and the embossed nanocantilevers. The inset of (a) shows the embossed nanocantilevers grown on Si in places, which are the minority nanostructures of A1. (c) and (d) are the nanowires grown at 430 ◦ C on Si (A2) and SiO2 (B2), respectively; (c) shows the nanowires grown on the ZnO buffer layer formed during the initial growth stage, and (d) shows the nanowires grown on the ZnO seeds without the formation of the ZnO buffer layer. The insets of (c) and (d) show the top view images of the nanowires grown on A2 and B2, respectively. (e) and (f) are the winding stem-like nanostructures with thin and short branch nanowires on the facets grown at 360 ◦ C on Si (A3) and SiO2 /Si (B3), respectively. The insets of (e) and (f) are magnified FESEM images showing the existence of the buffer layer at the interface between the winding stem-like nanostructures and the substrates, Si and SiO2 /Si, respectively.

flat parts. The widths of the embossed nanocantilevers extended to several micrometers, and their lengths were greater than 10 µm. As the growth temperature was decreased to 430 ◦ C, which is nearly the same as the melting point of zinc (∼419 ◦ C), a significant change was observed at the interface between the substrate and the ZnO nanowires, as shown in Figs. 2(c) and 2(d). A notable observation is that a buffer layer was produced in the sample grown on the Si substrate but not in the sample grown on the SiO2 /Si substrate. The buffer layer with a thick-

ness of about 200 nm consisted of zinc and oxygen elements only (EDX data not presented). The lengths of the ZnO nanowires grown on the ZnO buffer layer were normally 500 nm – 1 µm, and their diameters were 20 – 30 nm. The inset of Fig. 2(c) shows the top image of the ZnO nanowires with high density and a hexagonally shaped cross section. Vertically-aligned nanowires with a high density were grown on the Si substrate as opposed to those grown on the SiO2 /Si substrate. The inset of Fig. 2(d) shows that ZnO nanowires with a random orientation and a low density were grown on the SiO2 /Si substrate. Instead of the ZnO buffer layer, only ZnO seeds were shown at the bottom of the ZnO nanowires grown on the SiO2 /Si substrate. The diameters and the lengths of the nanowires grown on the SiO2 /Si substrate were comparable to those of the nanowires grown on the Si substrate. As the growth temperature was decreased further to 360 ◦ C, winding stem-like nanostructures with thin and short branch nanowires on the facets were obtained as shown in Figs. 2(e) and 2(f). The XRD examination revealed that these nanostructures had hexagonal (wurtzite) structures, as shown in Fig. 3. The branch nanowires had lengths of 100 – 200 nm and were randomly oriented on the facets of the nanostems. Although both products, as shown in Figs. 2(e) and 2(f), have similar appearances, the stem-like nanostructures grown on the SiO2 /Si substrate had denser, thicker, and longer branch ZnO nanowires on the facets of the nanostems than those grown on the Si substrate did. Moreover, the surfaces of the branch nanowires grown on the SiO2 /Si were rougher than those of the branch nanowires on the Si substrate. From the vertical FESEM images of both

-1880-


Fig. 4. TEM and HRTEM images of the ZnO nanowire grown on A1. (a) – (e) HRTEM images showing the nanowire from the base to the tip. (f) HRTEM image of the rounded tip, revealing (03-37) and (0-112) planes. (The SAED pattern of the tip is represented in the inset of (f)). (g) and (h) are magnified HRTEM images of the protrusive structures on the facets of the nanowires shown in (c) and (d), respectively.

products (insets of Figs. 2(e) and 2(f)), buffer-like structures covering the substrate were observed on the Si substrate, but not on the SiO2 /Si substrate. To investigate the crystalline structures of the grown ZnO nanostructures in detail, we performed HRTEM analyses. As shown in Fig. 2(a), ZnO nanowires grown on the Si substrate at 500 ◦ C had non-uniform diameters along their lengths. The magnified TEM images of Figs. 4(a) – (e) show the change in diameter of a nanowire from base (∼34 nm) to tip (∼20 nm). When the flowing source vapors arrived at the heated Si substrate at a temperature of 500 ◦ C, most of the source atoms diffusing along the surface of the substrate were reflected from the substrate due to their sufficient thermal energies. Therefore, the density of the grown nanowires was very low. When the source atoms diffusing along the surface of the substrate came together at active sites on the surface of substrate, they formed droplets, which were preferential sites for the adsorption of Zn or ZnOx vapors and for their reaction with O2 gas. The continuous supply of the source gases to the droplets induced the growth of nanowires. When the source atoms diffusing along the surface of the substrate were supplied to the nanowires during growth, some of them adhered to the side surfaces of the nanowires. The growth process on the sides of the nanowires causes a difference in the diameters in the growth direction; i.e., the total number of impinging vapors on the side surfaces during the growth process gradually decreased along the length of the growing nanowire because the time of exposure to the source vapors decreased along the length of the growing nanowire, resulting in a larger bottom diameter and a smaller top diameter. The HRTEM image (Fig. 4(f)) and the diffraction pattern (inset of Fig. 4(f)) indicate a well-grown singlecrystalline nanowire with a growth direction of [0001] and facets enclosed with six {01-10} planes. The tip of

the nanowire is not totally flat and has rounded edges revealing (03-37) and (0-112) planes as shown in Fig. 4(f). The typical structure of the ZnO nanowire has six side surfaces with {01-10} or {2-1-10} planes and a (0001) surface on the top due to the low surface energy of the {01-10} and the {2-1-10} planes and the fast growth rate of the (0001) plane [29]. Stable crystal facets are known to have a low growth rates and can easily generate large surface areas while unstable crystal facets are known to have fast growth rates, with minimal their surface areas [30]. The high-order indexed planes are the unstable planes of ZnO crystals [30]. However, unstable planes of ZnO nanowires with high surface energy can be readily formed at a high growth temperature of 500 ◦ C [30]. Rounded and protrusive shapes were also observed on the side surfaces of most nanowires grown at the high substrate temperature of 500 ◦ C, as shown in Figs. 4(c) and 4(d). The magnified HRTEM images of the protrusive shapes (Figs. 4(g) and 4(h)) indicate that the protrusive parts were connected to the body of a nanowire. Moreover, the crystal structure of the protrusive part was identical to that of the body of the nanowire without any defects. Thus, we can conclude that the CVD process locally affected a facet of the nanowire at high substrate temperature, resulting in the formation of the tiny ZnO droplet. A 30-µm-long embossed nanocantilever grown on the Si substrate at the temperature of 500 ◦ C was investigated in detail, as shown in Fig. 5. Because we mechanically scratched the nanostructures grown on the substrate for TEM sampling, only a portion of the original nanocantilever was observed. The nanocantilever shown in Fig. 5(a) has a variable width along its length. The bottom side, which was grown earlier, was wider (2 – 3 µm), while the top side, which was grown later, was narrower (several hundreds of nanometers). The surfaces of the nanocantilevers were not flat, and embossed teethlike structures appeared repeatedly along the lengths. The distances between the embossed teeth-like structures were not uniform. The parts grown earlier had more embossed, protruded, and thicker shapes while the parts grown later had relatively flat, weakly embossed and thinner shapes. The magnified TEM image of the circled region in Fig. 5(a) shows that the teeth-like structures were connected by thinner sheets located among the teeth-like structures, as shown in Fig. 5(b). The regions indicated in Fig. 5(b) were magnified for specific examinations, as shown in the HRTEM images of Figs. 5(c) – 5(f). The positions designated “c” and “d” in Fig. 5(b) correspond to the edges in the [000-1] direction while “e” and “f” match the edges in the [0001] direction. The HRTEM image at position “c”, a boundary region between the teeth-like structure and the neighboring thinner sheet, is shown in Fig. 5(c). The fast-Fourier transform (FFT) patterns and enlarged HRTEM images (insets of Fig. 5(c)) confirm that both the thinner sheets and the teeth-like structures had the same growth directions and identical


Fig. 5. (Color online) TEM and HRTEM images of the embossed nanocantilever grown on A1. (a) TEM image showing the shape of the embossed nanocantilever and (b) magnified TEM image of the circled region in (a). The positions indicated in (b) are investigated in detail using the HRTEM image: (c) – (e) are the HRTEM images of positions “c”, “d”, and “e” in (b), respectively. (f-L), (f-M), and (f-R) are the HRTEM images of positions “f-L”, “f-M”, and “f-R” in (b), respectively. (c) and (e) are the HRTEM images of the boundary region between the teeth-like structure and the thinner sheet on the edges of [000-1] and [0001], respectively. (Insets of (c): FFT diffraction patterns and enlarged HRTEM images of the thinner sheet (upper) and the teeth-like structures (lower) at position “c”. Insets of (e): FFT diffraction patterns of the thinner sheet (upper) and the teeth-like structure (lower) at position “e”). (d) is the HRTEM image showing the edge of [000-1] of the embossed teeth-like structure. (fL), (f-R), and (f-M) are the HRTEM images showing both sides and the middle on the edge of [0001] of the embossed teeth-like structure. (Inset of (f-M): SAED pattern of position “f-M”).

hexagonal (wurtzite) structures with a well-grown single crystal. The embossed nanocantilevers were grown in the [01-10] direction, and the growth direction of the width followed the [0001] direction. The large exposed surface (through the TEM image) was the (2-1-10) plane. ZnO nanocomb structures are known to grow asymmetrically along the Zn-[0001] direction due to differences in the chemical activity between the two polar facets of ZnO, ±(0001) [31]. The positively-charged Zn-(0001) surface is chemically active while the negatively-charged O-(0001) surface is relatively inert. Such anisotropic growth is a common characteristic for the wurtzite family. As shown in Figs. 5(a) – 5(d), which contain images at the edge in the [000-1] direction, no crystalline features were formed in the [000-1] direction. On the other hand, the embossed teeth-like structures and thinner sheets

-1881-

grew in the [0001] direction. Boundaries between the teeth-like structures and the neighboring thinner sheets (region e) were also observed, as shown in Fig. 5(e). Both the teeth-like structure and the thinner sheet at position “e” have the same crystal structure and growth direction, which is very similar to the results at position “c”. To examine the teeth-like structure, we present four HRTEM images (Figs. 5(d), (f-L), (f-M), and (f-R)). Since the portion “f-M” was thicker than “d”, as shown in the contrast difference between position “f-M” and “d”, the HRTEM image at position “f-M” was not obtained, as shown in the black image of Fig. 5(f-M). That is, the teeth-like structure on the [0001] edge was much thicker than that on the [000-1] edge, indicating that the growth of the teeth-like structure along the [0001] direction proceeded continuously, different from that along the [000-1] direction. The HRTEM images of both ends of the teeth-like structures along the [01-10] and the [0110] direction were obtained, indicating that the teethlike structure was thinner at the ends, as shown in the regions “f-L” and “f-R”. The rounded shapes in the regions “f-L” and “f-R” in Fig. 5 show that specific planes were revealed at the end of the structure. The inset of Fig. 5(f-M) shows the diffraction pattern of the region in the middle of the edge of the teethlike structure. The diffraction pattern appeared at the same lattice fringe as the FFT patterns obtained in positions “c” and “e”. Thus, the embossed nanocantilevers grown on the Si substrate at 500 ◦ C had single crystals with hexagonal (wurtzite) structures and with the same growth direction in all regions. The changes in the thickness and the width of the teeth-like structure imply that an embossed nanocantilever can be formed by using a two-step process: the fast growth of the stem-like nanostructure along the [01-10] direction and the slow growth of the teeth-like structure along the [0001] direction from the stem by a self-catalyzed process. The locally-grown nanostructure on the SiO2 /Si substrate at 500 ◦ C had a similar appearance to that grown on the Si substrate at the same growth temperature, as shown in Figs. 6(a) and 6(b); i.e., the embossed nanocantilevers consisted of alternately repeating thicker teethlike structures and thinner sheets. Detailed FESEM images of the nanocantilever cross sections are shown in Fig. 6(c); the teeth-like structures of the nanocantilevers form hexagonally shaped cross sections with a thick middle region that gets thinner at the edge. The hexagonally shaped cross section is also thought to be a reasonable shape for the cross section of the teeth-like structures of the nanocantilever grown on Si at 500 ◦ C. The nanocantilevers grew along the [01-10] direction and their widths grew along the [0001] direction, which was the same as that on the Si substrate. A remarkable difference between the nanocantilevers grown on the two different substrates was the existence of a protrusive nanostructure grown out of the (000-1) plane. This protrusive nanostructure was observed in the nanocantilevers grown on the SiO2 /Si substrate, but not in those grown on the

-1882-


Fig. 6. (Color online) TEM, HRTEM, and FESEM images of the embossed nanocantilevers grown on B1. (a) and (b) TEM images showing the whole shape. (c) FESEM images showing the hexagonal-shaped cross-sections of the embossed teeth-like structures. (d) TEM image of the protrusive nanostructure grown out of (000-1) marked in (b). (e) HRTEM image showing the reconstruction of the side surface of the protrusive nanostructure. (f) HRTEM image showing the planar defect beneath the protrusive nanostructure. (Insets of (f): FFT diffraction pattern of the protrusive structure (left) and SAED pattern of the embossed nanocantilever (right)).

Si substrate, as shown in the circled area of Fig. 6(b). Figures 6(d) – 6(f) show HRTEM images of the protrusive nanostructure grown out of the (000-1) plane of the nanocantilever. A planar defect parallel to the (0001) plane was observed just beneath the protrusive nanostructure (Fig. 6(f)). The observed planar defect was found to be a twin based on the FFT diffraction pattern. The generation of planar defects stabilizing surfaces with a higher energy state resulted from the anisotropic growth of the nanostructure being fastest in a specific direction [32]. Since the ±(0001) surfaces of ZnO are polar surfaces which have higher energies, planar defects parallel to the (0001) plane were introduced to stabilize the polar surfaces. Based on the formation of a protrusive nanostructure grown from the (000-1) plane with an underlying planar defect, we can conclude that the creation of the twin parallel to the [0001] direction enabled the growth of the nanostructure out of the (0001) plane, even though the (000-1) plane was a chemically inactive surface. Figure 6(e) shows the reconstruction of the side surface of the protrusive nanostructure grown out of (000-1) plane with a series of {03-3-4} and {01-10} facets. Since a surface state covered by dangling bonds is very active, the surfaces of nanomaterials with large surface-tovolume ratios were reconstructed to reduce the surface energy [32]. The HRTEM image (Fig. 6(f)) and the diffraction pattern (insets of Fig. 6(f)) revealed that a single-crystalline protrusive nanostructure was grown with the same crystal structure as that on the nanocantilever. The char-

Fig. 7. TEM and HRTEM images of the nanowires grown on A2. (a) TEM image showing several nanowires. (b) TEM images showing two sets of triangle-shaped sharp tips and roots. (c) HRTEM image of the triangle-shaped sharp tip. (Insets of (c): magnified HRTEM image (upper) and SAED pattern (lower) of the nanowire in (c)). (d) HRTEM image showing the poly-crystalline ZnO buffer layer formed at the interface between the nanowires and the substrate.

acteristics shapes and the crystalline structures of the nanocantilevers grown on both Si and SiO2 /Si at a high temperature of 500 ◦ C have not been reported yet. However, the similar shapes and crystalline characteristics of the nanocantilevers grown on both Si and SiO2 /Si indicate that the growth process on Si is almost the same as that on SiO2 /Si at a high temperature of 500 ◦ C. The TEM image of the ZnO nanowires grown on Si at 430 ◦ C is shown in Fig. 7(a). Two sets of images, those at the tips and those at the bases, are shown in Fig. 7(b). Unlike the nanowires grown on Si at 500 ◦ C, the diameters of the nanowires grown on Si at 430 ◦ C were almost completely uniform or slightly thicker in diameter at the tip, indicating that the surface reaction on the nanowire did not actively occur at the relatively lower growth temperature of 430 ◦ C. Because of the suppressed surface reaction process, no nanocantilever structures were generated. Figure 7(c) shows the HRTEM image of the triangle-shaped sharp tip. The magnified partial HRTEM image (upper inset) and the electron diffraction pattern (lower inset) are also attached. These insets reveal that the single-crystalline nanowire grew along the [0001] direction with {01-10} facets. In particular, the (01-12) and the (0-112) facets were generated at the tips of the nanowires. The shapes of the tips and the generated facets heavily depended on the growth temperature. The nanowires mentioned earlier that were grown on Si


at 500 ◦ C had tips with a higher-order indexed plane, (03-37), and a much larger area of the (0001) plane than the tips grown on Si at 430 ◦ C due to the thermal energy being sufficiently large. The sufficiently-large thermal energy resulted in the formation of higher-order indexed planes and an increased area of the (0001) plane with high surface energy while an insufficiently-large thermal energy caused the formation of relatively lower-order indexed planes and decreased the area of the (0001) plane. Therefore, the nanowires grown at 500 ◦ C had flatter tips while the nanowires grown at 430 ◦ C had sharper tips. The directions of the two facets were symmetric to the [0001] plane and made an angle of 85.6◦ . Figure 7(d) presents the HRTEM image showing the cross section of the buffer layer at the interface between the Si substrate and the ZnO nanowires. The EDX analyses of the poly-crystalline buffer layer showed that the buffer layer contained only zinc and oxygen without any impurities. When the source vapor arrived at the surface of the Si substrate at a temperature of 430 ◦ C, the source atoms did not have sufficient energy to diffuse along the surface of the substrate. Because 430 ◦ C is very close to the melting points of Zn and ZnOx , source atoms rarely diffused and easily adhered to the substrate. The incoming atoms collided with one other and combined to form ZnO clusters with random orientations and no relationship to the substrate. The continuous supply of vapor sources enabled the additional formation of ZnO clusters near the existing clusters and local growth of 1D ZnO nanowires along the [0001] plane on the clusters. The neighboring clusters and less-aligned ZnO nanowires grown on the neighboring clusters interacted, resulting in the formation of a buffer layer. Moreover, the formation of additional ZnO clusters near the existing clusters and the lateral growth of those clusters produced a polycrystalline ZnO buffer layer, as shown in Fig. 7(d). On the other hand, nanowires vertically aligned to the substrate grew well with no formation of a buffer layer. Through the growth process, a retarded growth of less-aligned nanowires and an enhanced growth of well-aligned nanowires were achieved. As a result, relatively well-aligned nanowires with vertical orientations were grown on the buffer layer on the Si substrate. The ZnO buffer layer played an important role in the synthesis of high-quality ZnO nanowires by preventing the growth of strained nanowires due to the difference in lattice constants between the ZnO and the substrate. Therefore, defects were not observed in the nanowires grown on the ZnO buffer layer. Because the whole area of the ZnO buffer layer acted as a favorable nucleation site for nanowire growth, the ZnO nanowires grown on the ZnO buffer layer had a relatively high density compared to the nanowires grown on the SiO2 /Si, as shown in Figs. 2(c) and 2(d). We investigated the structural characteristics of nanowires grown on SiO2 /Si substrates at 430 ◦ C. The most important finding is that a ZnO buffer layer was not created in the case of the sample grown on the SiO2 /Si

-1883-

Fig. 8. TEM and HRTEM images of the ZnO nanowires grown on B2. (a) TEM image showing several nanowires. (b) Magnified TEM image of the nanowire with dew-like shapes on the facets. (c) HRTEM images showing the polycrystalline dew-like shapes on the facets. (Inset of (c): FFT diffraction pattern of the dew-like structure.) (d) HRTEM image showing many dislocations in the nanowire. (e) – (g) Magnified HRTEM images showing the polycrystals in (c). (h) Magnified HRTEM image showing the dislocations in (d).

substrate, i.e., the nanowires were grown directly on the substrate, as shown in the SEM image of Fig. 2(d). The TEM images show that the shape of the nanowire was very different than that for the Si substrate. In this case, dew-like shapes grew on the facets of the nanowires, as shown in Figs. 8(a) and (b). Moreover, the nanowires had rough surfaces and non-uniform diameters along lengths. The magnified HRTEM image (Fig. 8(c)) and the FFT pattern (inset of Fig. 8(c)) reveal that the dewlike shapes consisted of ZnO polycrystals. Figures 8(e) – (g) show magnified HRTEM images of the ZnO polycrystals in the region of Fig. 8(c) and contain various lattice images with different orientations. Compared to the structural shape with the protrusive structure on the facet of the nanowire grown on the Si substrate at 500 ◦ C, the dew-like structure is somewhat similar to the protrusive structures; i.e., the two structures were locally formed on the facets of the nanowires. In particular, polycrystalline structures with different orientations and grain boundaries were generated in the dew-like

-1884-


shape formed at 430 ◦ C while a single-crystalline structure with the same crystalline structure as that of the body of nanowire was grown on the facet of the nanowire at 500 ◦ C. The difference in the crystalline structures on the facets of the nanowires was caused by the difference in thermal energies. Since the growth temperature of 430 ◦ C is only slightly higher than the melting point of Zn or ZnOx , neither Zn or ZnOx has sufficient thermal energy to diffuse on the facet of the nanowire. This insufficient energy resulted in the formation of polycrystalline structures on SiO2 . On the other hand, at a growth temperature of 500 ◦ C, the Zn or ZnOx vapors were in a liquid state on the facets of the nanowires, and they have sufficient thermal energy to align the crystalline structure. The surface roughness of the nanowire also yields information on the growth kinetics. At a growth temperature of 430 ◦ C, the surface roughnesses of the nanowires grown on the two different substrates, SiO2 /Si and Si, were very different. Thus, the origin of the rough surface of the nanowire was not the growth temperature. We noted that the density of the nanowires grown on B2 was much lower than that of the nanowires grown on A2. When using the same quantity of supplied source material, the impinging rate of the source material to the nanowire was relatively high in the nanowire region of low density while the impinging rate was relatively low on the region of high nanowire density. Finally, the nanowires grown with a high impinging rate (A1, B2) had facets which were not smooth, with protrusive and dew-like structures, while those grown with a low impinging rate (A2) had smooth facets without protrusive structures. Another remarkable observation in the nanowires grown on the SiO2 /Si substrate at 430 ◦ C was the generation of many defects such as dislocations while perfect single crystalline nanowires without any dislocation lines were grown on the Si substrate. Many dislocations were observed continuously along the lengths of the nanowires, as shown in Fig. 8(d). A magnified HRTEM image of the dislocations is presented in Fig. 8(h). The formation of the dislocations could be due to growth without a buffer layer. Because the surface energy of SiO2 is not as high as that of Si, a ZnO buffer layer was not formed on the SiO2 /Si substrate, unlike the case of the nanowires grown on the Si substrate. Instead of a buffer layer, ZnO seeds were formed on the SiO2 /Si substrate, as observed at the bottom of the nanowires in Fig. 2(d). The ZnO seeds on the SiO2 /Si substrate did not have a specific epitaxial arrangement due to the amorphous structure of the SiO2 /Si substrate, which resulted in randomly oriented nuclei. The nanowires grown on the randomly oriented nuclei had the same orientation as the nuclei [33]. Therefore, less-aligned nanowires with poor crystalline structure were grown on the SiO2 /Si substrate. Moreover, the density of the ZnO nanowires grown on the SiO2 /Si substrate was relatively low compared to that of the nanowires grown on the ZnO buffer layer because

there were few nucleation sites due to the absence of the ZnO buffer layer. At a growth temperature of 360 ◦ C, a ZnO buffer layer was observed in the sample grown on the Si substrate while the buffer layer was not observed on the SiO2 /Si substrate. The formation of a buffer layer that depends on the type of substrate is similar to the case for a growth temperature of 430 ◦ C. The magnified FESEM images at the interface between the grown nanostructures and the substrate are shown in the insets of Figs. 2(e) and 2(f). As in the nanowires grown at 430 ◦ C, high surface energy due to dangling bonds also affected the formation of the buffer layer. The difference in the sticking coefficients between the Si substrate and the SiO2 substrate was very high and may be another important factor in the formation of the buffer layer. At the higher temperature of 500 ◦ C, the buffer layer was not generated because the source atoms had sufficient thermal energy to diffuse along the substrate. Since the growth temperature 360 ◦ C is much lower than the melting point of Zn or ZnOx , the source materials quickly condensed on the substrate without diffusing and were simultaneously oxidized by the supplied oxygen gas. The continuously supplied source materials combined with the crystallized ZnO surface in random orientations due to insufficient time and thermal energy. Therefore, the growth orientations changed continuously, forming the winding and stem-like nanostructures. The surfaces of the winding shapes geometrically provided preferential nucleation sites because many active defect states were distributed on their surfaces. When the source materials arrived at the nucleation sites, the branch nanowires grew on the side surfaces of the stem-like nanostructures.

IV. CONCLUSION The syntheses of ZnO nanostructures at different growth temperatures on Si and SiO2 /Si substrates were conducted by using a thermal evaporation method without a metal catalyst to investigate the growth kinetics of ZnO nanostructures on Si substrates as a function of growth temperature. The characteristics of the grown ZnO nanostructures depended on the growth temperatures and the surface state of the Si substrate. At a temperature higher than the melting point (500 ◦ C) of Zn, the nanowires grew on the Si substrate with low density, and embossed nanocantilevers grew on both Si and SiO2 /Si. Defect-free ZnO nanostructures were obtained on the Si substrate while planar defects beneath a protrusive nanostructure were found in the embossed nanocantilevers grown on the SiO2 /Si substrate. At 430 ◦ C, a temperature near the melting point of Zn, the initial growth mechanisms on both substrates were quite different due to the difference in the surface energies of the Si and the SiO2 /Si substrates. A ZnO buffer layer formed at the interface between the ZnO nanowires and


the Si substrate due to the high surface energy of the Si substrate while ZnO seeds were generated between the nanowires and the SiO2 /Si substrate due to the low surface energy. The density of the nanowires also depended on the existence of the ZnO buffer layer. Because the ZnO buffer layer produced numerous preferential nucleation sites for the growth of nanowires, high-density nanowires grew on the Si substrate. Finally, we concluded that the ZnO buffer layer played an important role in the growth of high-quality ZnO nanowires by preventing the growth of strained nanowires due to the difference in lattice constants between the ZnO and the Si substrate.

ACKNOWLEDGMENTS This work was supported by the New and Renewable Energy R&D program (Grant No.: 2009T100100614) under the Ministry of Knowledge Economy.

REFERENCES [1] Y. Wu, H. Yan and P. Yang, Chem. Eur. J. 8, 1260 (2002). [2] J. Hu, T. W. Odom and C. M. Lieber, Acc. Chem. Res. 32, 435 (1999). [3] G. Y. Adachi and N. Imanaka, Chem. Rev. 98, 1479 (1998). [4] C. Soci, A. Zhang, B. Xiang, S. A. Dayeh, D. P. R. Aplin, J. Park, X. Y. Bao, Y. H. Lo and D. Wang, Nano Lett. 7, 1003 (2007). [5] J. D. Prades, R. Jimenez-Diaz, F. Hernandez-Ramirez, L. Fernandez-Romero, T. Andreu, A. Cirera, A. Romano-Rodriguez, A. Cornet, J. R. Morante, S. Barth and S. Mathur, J. Phys. Chem. C 112, 14639 (2008). [6] H. Huang, S. Mao, H. Feick, H. Yan, Y. Wu, H. Kind, E. Weber, R. Russo and P. Yang, Science 292, 1897 (2001). [7] J. X. Wang, X. W. Sun, Y. Yang, H. Huang, Y. C. Lee, O. K. Tan and L. Vayssieres, Nanotechnology 17, 4995 (2006). [8] J. X. Park, D. E. Song and S. S. Kim, Nanotechnology 19, 105503 (2008). [9] A. Nadarajah, R. C. Word, J. Meiss and R. Konenkamp, Nano Lett. 8, 534 (2008). [10] X. W. Sun, J. Z. Huang, J. X. Wang and Z. Xu, Nano Lett. 8, 1219 (2008).

-1885-

[11] M. Law, E. Greene, J. C. Johnson, R. Saykally and P. Yang, Nat. Mater. 4, 455 (2005). [12] J. B. Baxter and E. S. Aydila, Appl. Phys. Lett. 86, 053114 (2005). [13] M-Y. Choi, D. H. Choi, M-J. Jin, I. S. Kim, S-H. Kim, J-Y. Choi, S. Y. Lee, J. M. Kim and S-W. Kim, Adv. Mater. 21, 2185 (2009). [14] B. P. Zhang, N. T. Binh, Y. Segawa, K. Wakatsuki and N. Usami, Appl. Phys. Lett. 83, 1635 (2003). [15] W. I. Park, D. H. Kim, S-W. Jung and G-C. Yi, Appl. Phys. Lett. 80, 4232 (2002). [16] H. T. Wang, B. S. Kang, F. Ren, L. C. Tien, P. W. Sadik, D. P. Norton, S. J. Pearton and J. Lin, Appl. Phys. Lett. 86, 243503 (2005). [17] H. Q. Yang, Y. Z. Song, L. Li, J. H. Ma, D. C. Chen, S. L. Mai and H. Zhao, Cryst. Growth Des. 8, 1039 (2008). [18] Y. Sun, G. M. Fuge and M. N. R. Ashfold, Chem. Phys. Lett. 396, 21 (2004). [19] S. Li, X. Zhang, B. Yan and T. Yu, Nanotechnology 20, 495604 (2009). [20] B. D. Yao, Y. F. Chan and N. Wang, Appl. Phys. Lett. 81, 757 (2002). [21] A. Fontcuberta-i-Morral, J. Arbiol, J. D. Prades, A. Cirera and J. R. Morante, Appl. Phys. Lett. 19, 1347 (2007). [22] J. D. Pardes, J. Arbiol, A. Cirera, J. R. Morante and A. Fontcuberta-i-Morral, Appl. Phys. Lett. 91, 123107 (2007). [23] M. H. Huang, Y. Wu, H. Feick, N. Tran, E. Weber and P. Yang, Adv. Mater. 13, 113 (2001). [24] C. Y. Geng, Y. Jiang, Y. Yao, X. M. Meng, J. A. Zapien, C. S. Lee, Y. Lifshitz and S. T. Lee, Adv. Funct. Mater. 14, 589 (2004). [25] L. S. Wang, X. Z. Zhang, S. Q. Zhao, G. T. Zhou, Y. L. Zhou and J. J. Qi, Appl. Phys. Lett. 86, 024108 (2005). [26] J. Jie, G. Wang, Y. Chen, X. Han, Q. Wang, B. Xu and J. G. Hou, Appl. Phys. lett. 86, 031909 (2005). [27] Y. Sun, Q. Zhao, J. Gao, R. Zhu, X. Wang, J. Xu, L. Chen, J. Zhang and D. Yu, Cryst. Eng. Comm. 13, 606 (2011). [28] S. Muthukumar, C. R. Gorla, N. W. Emanetoglu, S. Liang and Y. Lu, J. Cryst. Growth 225, 197 (2001). [29] Z. L. Wang, J. Phys. Condens. Matter 16, R829 (2004). [30] J. S. Jeong, J. Y. Lee, J. H. Cho, C. J. Lee, S-J. An, G-C. Yi and R. Gronsky, Nanotechnology 16, 2455 (2005). [31] Z. L. Wang, X. Y. Kong and J. M. Zuo, Phys. Rev. Lett. 91, 185502 (2003). [32] X. D. Wang, Y. Ding, C. J. Summers and Z. L. Wang, J. Phys. Chem. B 108, 8773 (2004). [33] J. S. Jeong and J. Y. Lee, Nanotechnology 21, 475603 (2010).