Orientation of silicon nanowires grown from nickel-coated silicon wafers

Orientation of silicon nanowires grown from nickel-coated silicon wafers

Journal of Crystal Growth 404 (2014) 26–33 Contents lists available at ScienceDirect Journal of Crystal Growth journal homepage: www.elsevier.com/lo...

4MB Sizes 0 Downloads 43 Views

Journal of Crystal Growth 404 (2014) 26–33

Contents lists available at ScienceDirect

Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro

Orientation of silicon nanowires grown from nickel-coated silicon wafers Feng Ji Li a,b, Sam Zhang a,n, Jyh-Wei Lee c,d, Jun Guo e, Timothy John White f, Bo Li g, Dongliang Zhao g a

School of Mechanical and Aerospace Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore PVD Department of Plating Division, Singapore Epson Industrial Pte Ltd, Singapore 628162, Singapore c Department of Materials Engineering, Ming Chi University of Technology, Taipei 24301, Taiwan, ROC d Center for Thin Film Technologies and Applications, Ming Chi University of Technology, Taipei 24301, Taiwan, ROC e Analysis and Testing Center, Soochow University, Suzhou 215123, China f School of Materials Science and Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore g Central Iron and Steel Research institute, 76 South Xueyuanlu Road, Haidan District, Beijing 100081, China b

art ic l e i nf o

a b s t r a c t

Article history: Received 22 February 2014 Received in revised form 23 May 2014 Accepted 12 June 2014 Communicated by J.M. Redwing Available online 30 June 2014

Growth orientation of silicon (Si) nanowires is the key in tailoring the optical and electrical characteristics of semiconductor devices. To date, however, the distribution and dictator are still unclear. In this work, Si nanowires are grown via thermal annealing of nickel (Ni) coated Si wafers. The morphology, growth orientation and the relation to the seeding Ni catalyst particles are examined via high resolution transmission electron microscopy and selected area electron diffraction pattern. Statistical results show that Si nanowires prefer to be along the o112 4 orientation, followed by the ones in the o110 4, o 111 4 , o001 4, o113 4 and o1334 orientations. Besides surface energy that is commonly believed to control the nanowire's growth, this work found that the nanowire's growth follows certain structure-sensitive principle at the wire/catalyst interface to minimize the mismatch in lattice spacing and dihedral angle. & 2014 Elsevier B.V. All rights reserved.

Keywords: A1. Interfacial structure A1. Surface energy A3. Solid–liquid–solid growth A3. Growth orientation B1. Nanoparticle B1. Nanowire

1. Introduction Silicon (Si) nanowires are important nanotechnology building blocks in optical devices [1–3], field-effect transistors [4–6], lithium batteries [7,8], and power generators [9,10]. They can grow by laser ablation of Si and Fe powder mixture [11], chemical vapor deposition of SiCl4/H2 [12–14], or SiH4 gases [15], thermal degradation of diphenylsilane in a supercritical hexane fluid [16,17], or thermal evaporation of bulk Si onto metal-covered substrates under ultrahigh vacuum [18,19]. In the growth process, Si atoms come from vaporization of powder mixture [11], decomposition of gaseous molecules [12–14], degradation of diphenylsilane [16,17], or sublimation of bulk Si [18,19], and then, melt into the metal-Si eutectic droplet. Upon supersaturation of Si, Si precipitates out to seed a nanowire, followed by growth of the nanowire with continuous supply of precipitating Si atoms from the liquid droplet. If the growth takes place on a crystal substrate, e.g., Si wafer, the droplet becomes displaced from the substrate

n

Corresponding author. E-mail address: [email protected] (S. Zhang).

http://dx.doi.org/10.1016/j.jcrysgro.2014.06.033 0022-0248/& 2014 Elsevier B.V. All rights reserved.

and “rides” atop the grown nanowire [12]. Si nanowires also grow by thermal annealing of Si wafers that are covered with metallic catalyst (e.g., Ni, Fe or Au) without additional vapor or liquid Si source [20–27]. In the process, the metallic catalyst melts with the beneath Si wafer to form the metal-Si eutectic droplet. Si atoms continuously diffuse into the droplet. Upon supersaturation, Si is precipitated from the droplet to grow into nanowires. To facilitate the growth, Si from the Si wafer serves as the source and the liquid metal-Si droplet acts as the medium. The nanowire always grows out from the liquid droplet while the droplet always stays on the wafer (thus the catalyst droplet is always found at the “root” of the nanowire, i.e., at the wafer substrate). This is called solid–liquid– solid (SLS) growth. In SLS growth, Si atoms directly come from the solid Si wafer. Thus, diphenylsilane, supercritical hexane fluid or Si-containing flammable gases are not necessary. Moreover, the growth condition is simple, i.e., a motor pump, a heating unit and an Ar gas atmosphere, avoiding high energy laser, high vacuum atmosphere or special chemical reaction environment. Growth orientation is crucial in tailoring Si nanowire's properties. For instance, the band gap, electrical conductance and mechanical properties are dramatically different in nanowires with different growth orientations [28–34]. The o100 4 oriented

F.J. Li et al. / Journal of Crystal Growth 404 (2014) 26–33

Si nanowires exhibit a significantly higher exciton energy than the o110 4 counterparts [16]. The o 110 4 oriented wire has the lowest total energy, and the highest sensitivity to surface modification [29], whereas the o111 4 oriented is structurally stable to the change of diameter, and of the lowest sensitivity to surface modification [30]. The o110 4 oriented one displays a direct transition, the o111 4 oriented possess a competitive indirect– direct gap character, and the o112 4 oriented has an indirect band-gap [32]. Tetrahedral Si nanowires oriented in o111 4 are the most stable and the best suited for quantum confinement effects [33]. The o 112 4 nanowires exhibit the highest tensile strength of 12.31 GPa, and the weakest link fracture is a byproduct of orientation-dependent flaw populations [34]. To date, what really determines the growth orientation is still unclear, though lots of factors have been reported to have effect, including surface chemical treatment of substrate [35], nanowire diameter [15,36,37], composition of the initial alloy droplet [38,39], growth temperature [40–42], reaction pressure [16,43], and precursor molar ratios [44]. Most people believe the preferential orientation is controlled by the energetics at the solid–liquid interface of the nanocrystal emerging from the liquid droplet. It involves the surface energy of various nanocrystal planes [15,45,46]. As one of the metal-seeded Si nanowire growth methods, however, SLS growth involves (1) the formation of the metal-Si eutectic droplet from solid [20–23], and (2) the subsequent Si precipitation or a crystallization process. The first process requires thermal energy to melt the solid Si and Ni to liquid Si–Ni eutectic, i.e., an energyconsuming process. The subsequent crystallization process solidifies the liquid eutectic to solid Si nanowires, i.e., an energy-releasing process [47]. Therefore, energy may not be necessary in the second process. It is therefore concluded that there may be alternative determinant factors controlling the growth orientation. In the growth of Si nanowires, Au was widely used as the catalyst; however, the incorporation of Au in Si nanowires would possibly induce a deep-level trap, causing a reduction in the efficiency of the solar cells [48]. In addition, the ionization energy level of Ni within Si is located far from the mid-band gap, offering a higher threshold level (5  1015 cm  3) in impurity concentration compared with that (3  1013 cm  3) of Au [49]. With these concerns, in this work, highly crystalline Si nanowires were grown via thermal annealing of Ni-coated Si wafers. The morphology, growth orientation, and the relation to the seeding Ni catalyst particles were examined through high resolution transmission electron microscopy and selected area electron diffraction pattern. The determinant factors for the growth orientation were explored.

2. Experimental details The Ni catalyst layer is sputtered on N-type Si (100) wafers (10 mm  10 mm in area, 475 μm in thickness and 0.5 nm in room mean square surface roughness) in an E303A magnetron sputtering system (Penta-Vacuum, Singapore) [50]. Before loading the wafers into the sputtering chamber, the substrates were ultrasonically cleaned in acetone for 20 min, followed by 10 min in alcohol. Once the base pressure reached 1.0  10  1 Pa, Ar was introduced at a flow rate of 50 standard cubic centimeters per minute (sccm). High energy Ar þ ions were generated at  300 V substrate bias for substrate cleaning for 15 min to remove the native silicon oxide. Then, 5 min sputtering of Ni target (purity, 99.99%) in 0.47 Pa pressure at room temperature deposited about 30-nm thick Ni catalyst layer on the substrate. And then, about 570 nm amorphous carbon was sputtered from a graphite target (purity, 99.999%) on the Ni layer in the same work pressure for 60 min to retard the oxidation of grown Si nanowires [23,24]. During the process, about 4.16 at% Si was co-sputtered from a near

27

Si target (99.999% in purity) on the Ni layer [23]. The sputtering power densities for Ni, Si and C were respectively RF 1.85 W cm  2, RF 2.47 W cm  2 and DC 12.22 W cm  2. In this study, the top layer is labelled a-C layer because the top layer is only sputtered to retard oxidation and the amount of Si in the layer is very few. The a-C/Ni films underwent rapid thermal annealing (RTA, Jipelec Jetfirst 100 rapid thermal processor, France) in an Ar ambient at 1100 1C for 1 min. To obtain a clean growth atmosphere, the RTA chamber was purged 10 times with Ar (purity, 99.999%) at 2000sccm before ramping up to 600 1C at 58 1C s  1 , which is controlled by a thermocouple. After dwelling for 6 s at 600 1C, the chamber was further ramped up to 1100 1C at 42 1C s  1, which is controlled by a pyrometer, and then held for 1 min for the growth of Si nanowires. After annealing, the chamber was cooled to 500 1C at 43 1C s  1 , and then naturally cooled down to room temperature. During the whole annealing process (ramping, holding, and cooling), the inflow of the Ar gas was maintained at 2000 sccm to keep the chamber pressure at 1.08  105 Pa. Field emission scanning electron microscopy (FESEM, JEOL, JBM-7600F, JEOL Ltd., Japan, 5-kV operating voltage) shown that wire-like nanostructures were grown on the annealed film surface. A high resolution transmission electron microscope (HRTEM, JEOL 2010, 2100F, 200-kV operating voltage, Japan) was employed to examine the structure of the grown nanowires. The size of the wire was calculated by randomly taking the mean value of 20 wires. Electron dispersive X-ray spectroscopy (EDX, EDAX Inc., USA) and selected area electron diffraction pattern were carried out to characterize the chemical composition and the growth orientation of the grown nanowires and the seeding Ni catalyst particles. To determine the correct growth direction, the SAED pattern is indexed for a specific zone axis of Si. This requires the electron beam perpendicular to a specific plane family of the crystal, i.e., zone axis, which could generate the pattern. In the process, the SAED pattern or systematic spots were generated by careful adjusting the tilt angle along the xand y-axes of the sample holder to match the standard database that is cited from software JEMS.

3. Results and discussion 3.1. Morphology Fig.1a shows a bundle of primarily wire-like nanostructures entangling into each other with remarkably uniform diameter on the order of 80 nm, with length longer than 10 mm. Statistical calculation on 20 randomly selected coaxial Si nanowires reveals that the mean diameter of the Si core is 17.07 6.3 nm, and the mean thickness of SiOx (x E2) sheath is 19.7 73.7 nm. The nanowire diameter equals that of the Si core plus two times of the SiOx sheath thickness. The nanowires terminate at the root in dark ball-shaped nanoclusters with diameter 1.2 to 1.5 times that of the connected nanowires (cf., Fig. 1b). But no nanocluster is observed at the tip (cf., Fig. 1c). Similar morphology is reported in another paper [51]. This is suggestive of the Ni-catalyzed SLS growth similar to the vapour–liquid–solid growth, except for the opposite seeding location of the nanoclusters. Upon annealing, the Ni layer collapses into particles due to surface tension. Ni particles melt with the beneath Si wafer to form Ni Si eutectic, leading to the continuous diffusion of Si into the droplet. Upon supersaturation, Si is precipitated from the droplet to grow into nanowires. As Si is from the Si wafer and the liquid Ni  Si droplet acts as the medium, the nanowire always grows out from the liquid droplet while the droplet always stays on the wafer. Experimentally, Ni is not detected in the X-ray photoelectron spectroscopy and FESEM image of the annealed Si wafer surface [23]. Therefore, the Ni Si eutectic droplet always stays at the root of the wire.

28

F.J. Li et al. / Journal of Crystal Growth 404 (2014) 26–33

Fig. 1. Morphology of Si nanowires. (a) A bundle of Si nanowires. (b) Dark nanoclusters are located at the root of Si nanowires. (c) Nanocluster-free nanowire tip, where the inner dark Si core is surrounded by the gray SiOx outer sheath.

3.2. Growth orientation To date, three categories of techniques have been applied to determine the growth orientation: (1) HRTEM image and the corresponding two dimensional Fast Fourier Transform (FFT) pattern [11,17]; (2) TEM image and the corresponding SAED pattern [15,21]; and (3) geometric relation to the substrate surface by optical microscope, SEM or TEM [52]. Approaches (1) and

(2) are the most accurate ways since they completely eliminate the influence from the environment and instrument. Approach (3) may be the most convenient, but of the biggest error since the observed orientations are likely changed during sampling, especially for the long and curved nanowires. In this study, the growth orientation was determined by HRTEM image and the corresponding 2D FFT pattern, in conjunction with the TEM image and the corresponding SAED pattern.

F.J. Li et al. / Journal of Crystal Growth 404 (2014) 26–33

3.2.1. Growth orientation determined by HRTEM image and 2D FFT pattern As shown in Fig. 2, HRTEM images captured on three individual Si nanowires along different growth orientations provide atomic insight into the structure of the nanowires. The nanowires consist of a uniform diameter crystalline Si core surrounded by an amorphous SiOx sheath. The core and sheath are separated by an atomically sharp interface.

29

The amorphous SiOx sheath surrounding the crystalline Si core could be removed by hydrofluoric acid [11]. In Fig. 2a, the reciprocal lattice peaks obtained from the 2D FFT of the atom-resolved image of the Si core are indexed for the [102] zone axis of crystalline Si, suggesting that the nanowire grows along the [2, 1, 1̄] orientation. This orientation is further confirmed in the atom-resolved TEM image of the crystalline Si core, which clearly shows the (2, 1,̄ 1̄) atomic planes (spacing, 2.22 Å),

Fig. 2. HRTEM images of Si nanowires growing along different orientations. (a) [2, 1, 1̄ ] oriented; (b) [0, 1, 1̄ ] oriented; and (c) [1, 1, 1̄ ] oriented Si nanowire. Inset shows the respective FFT pattern of the atom-resolved Si core generated from the zone axis perpendicular to the growth axis.

30

F.J. Li et al. / Journal of Crystal Growth 404 (2014) 26–33

the (020) atomic planes (spacing, 2.72 Å), and their dihedral angle 65.91. Similarly, Fig. 2b and c respectively index the [111] and [112] zone axis of the crystalline Si from the 2D FFT pattern of the atomresolved Si core. In conjunction with the HRTEM images, the nanowires grow along [0, 1,̄ 1] and [1, 1, 1̄] orientations, respectively. Fig. 2b reveals the atomic planes of (1,̄ 1, 0 and (0, 1,̄ 1). (spacing, 3.84 Å) and their dihedral angle 120.01, where (0, 1,̄ 1) plane is perpendicular to the nanowire axis. Fig. 2b has been used in the previous work to prove the grown Si nanowires are highly crystalline [23]. In this work, this figure is cited to prove HRTEM and 2D FFT pattern can determine the different growth orientations. Fig. 2c shows the (2, 2̄ , 0) atomic planes (spacing, 1.92 Å), and the (1, 1, 1̄ ) (spacing, 3.14 Å), and their dihedral angle 90.01, where (1, 1, 1̄ ), planes are perpendicular to the growth axis. 3.2.2. Growth orientation determined by TEM image and SAED pattern As shown in Fig. 3a, the SAED pattern in the inset reveals the crystalline Si core and the amorphous SiOx sheath, i.e., c-Si/a-SiOx. A unit cell of face-centred-cubic structure of Si (5.43 Å in lattice parameter) is consistent with the diffraction pattern. It is indexed for the Si [001] zone axis, indicating the Si nanowire grows along the [110] orientation. Likewise, the insets of Fig. 3b–f respectively show the SAED pattern generated from the zone axis of Si [101], [101], [315], [323], and [116], determining the nanowires grow along the [1, 1,̄ 1], [010], [2̄ , 1, 1], [1,̄ 3, 1̄ ], and [3̄ , 3̄ , 1] orientations. Statistical results of 33 nanowires (cf., Fig. S1) exhibit an occurrence preference of 30.3% for the nanowires growing along the o1124 orientation, followed by 27.3% nanowires growing along the o1104

orientation, 24.3% ones along the o1114 orientation, and 12.1% ones in the o0014 orientation. Few nanowires grow in the o1134 and o1334 orientations. Similar orientation distribution has been reported for nanowires (diameter ranging from 10 nm to 20 nm) grown by chemical vapour deposition of SiH4 on Au-covered polyL-lysine on Si wafer substrates [15,36], and those synthesized through thermal evaporation of SiO powders in an alumina tube under H2 þAr atmosphere [46]. Both of them believe the orientation preference is determined by the surface energy of the various Si nanocrystal planes [15,46]. However, the energy argument made by Ref. [15] is under the assumption that the flat Si/Au interface is always the Si {111} plane (cf., Fig. 3 in Ref. [15]). Ref. [46] believes Si nanoparticles that are precipitated through the disproportionation of SiO, i.e., 2SiO-Si þSiO2 [46,53], act as the seeds to the growth of Si nanowires without any supply of metallic catalyst. Si nanowire nucleates from the Si {111} facets to reach the minimum surface energy [53]. However, in their experiments with SiO powder, reduction of Si2 þ to Si0 and oxidation of Si2 þ to Si4 þ (in formation of SiO2) take place only with migration of electrons thus the process does not involve physical movement of Si atoms or SiO2 molecules [54]. Recent study shows that metallic aluminum coming from the reduction of alumina tube by H2 highly possible acts as the catalyst for Si nanowire seeding and growth [54]. On the other hand, surface energy quantifies the disruption of intermolecular bonds that occur when a surface is created [55]. For a liquid, the surface energy density are identical everywhere. In the present growth process, surrounding residual oxygen may enter the surface of the droplet to form Ni SiO droplet. The growth of Si nanowires takes place as Si in the liquid Ni SiO

Fig. 3. TEM images of Si nanowires growing along different orientations. (A) [110], (B)[1, 1,̄ 1], (C)[010], (D)[2̄ , 1, 1], (E)[1,̄ 3, 1̄ ], and (F) [3̄ , 3̄ , 1]. Inset shows the respective SAED pattern generated from different zone axis.

F.J. Li et al. / Journal of Crystal Growth 404 (2014) 26–33

eutectic droplet becomes supersaturated. At this moment, therefore, the surface energy density is identical everywhere in the droplet. For a solid, however, the molecules on the surface have more energy compared with the molecules in the bulk of the material. For a solid Si nanowire that is already grown from the liquid droplet, the Si-oxide shell must have higher surface energy than the Si core because it is on the outer surface. But the growth orientation is already determined at the moment of precipitating Si from the droplet. Moreover, if the formation of Si-oxide shell and Si core takes place simultaneously, the interfacial energy between Si-oxide shell and Si core may be considered instead of the surface energy of a bare facet. The reasonably low Si/Si-oxide interfacial energy may explain why few o1134 and o1334 oriented Si nanowires with relatively high surface energy are detected. It is therefore concluded that the theory of surface energy may be not enough to explain the orientation preference in the seeding and growth of Si nanowires. There might be other determinant factors related to the metallic catalyst controlling the growth orientation. 3.3. Interfacial structure between Si nanowire and Ni catalyst Fig. 4a shows an elliptic Ni catalyst particle seeding a Si nanowire. The particle is of around 47.0 nm in semi-major axis and 39.3 nm in semi-minor axis. The particle is surrounded by a 7.2-nm-thick amorphous SiOx layer, forming the Ni  Si  O nanocluster. The EDX spectrum shown in Fig. 4b is generated from the nanocluster, confirming the chemical composition of the core in the seeding nanocluster is Ni, where the signals of C and Cu are from the TEM grid, while Si and O are from the SiOx layer surunding the Ni core. The SAED systematic rows, generated by directing the electron beam on the area of both the seeding Ni particle and the connected grown Si nanowire provide the crystalline information of both the catalyst and the nanowire (cf., Fig. 4c). These systematic rows are indexed for the zone axis of Ni [112]. The dotted white squares are respectively indexed for Ni (3̄ , 1, 1), Ni (1,̄ 1,̄ 1) and Ni (1, 1, 1̄ ). Though the information in the SAED spots is not enough to determine the growth orientation of the Si nanowire, the dotted white circle near Ni (1,̄ 1,̄ 1) still could be indexed for Si (020) based on the measured lattice spacing of around 2.70 Å (cf., Fig. 4d), and the relationship to the spot reflected by the Ni (1,̄ 1,̄ 1) plane. There is around 20.61 drift in anti-clockwise during HRTEM observation. It is noted that the Si-wire/Ni-catalyst interface exhibits a round profile, rather than the flat ones reported in Ref. [15]. The round Ni/Si interfacial shape further reveals the formation of the liquid Ni  Si  O phase does take place during the growth process since the liquid Ni  Si eutectic droplet could form at only around 964 1C [56], at least 100 1C lower than the annealing temperature. Ni (1,̄ 1,̄ 1) (spacing, 2.03 Å) and Si (020) (spacing, 2.72 Å) planes are almost parallel with a small dihedral angle of 31 (cf., Fig. 4c and d), though there is about 25.4% difference in lattice spacing in reference to Si (cf., (2.72–2.03)/2.72). It is noted that Ni (1,̄ 1,̄ 1) planes match with Si (020) planes though the nanowire has a change in morphology along the growth orientation within 5–10 nm of the wire/catalyst interface. Growth orientation is determined during precipitation of Si from the liquid Ni  Si eutectic droplet. As such, images showing lattice match during growth between the catalyst and wire of different growth orientations would provide stronger evidence. This can be a topic for future study.

31

believed that the preference of growth orientations was dictated by the surface energy of the various Si nanocrystal planes [15,46]. In this study, however, the growth of Si nanowires involves mass transfer of Si atoms from the Ni  Si  O droplet to a solid crystalline Si phase. It is an energy-releasing process. Therefore, surface energy may not be necessary in nucleation and growth of nanowires. As shown in Fig. 4, a structure-sensitive seeding principle is found at the Ni-catalyst/Si-wire spherical interface, i. e., Ni (1,̄ 1,̄ 1) is parallel to Si (020) with only 31 in dihedral angle and 25.4% mismatch in lattice spacing. It indicates that the matching degree of the lattice planes at the Si-wire/Ni-catalyst interface may also dictate the nucleation and growth of Si nanowires. As such, it is reasonable to believe that the nanowire's growth also follows certain structure-sensitive principle to achieve the most stable structure with the minimum mismatch in lattice spacing and dihedral angle at the wire/catalyst interface (in this study, Si and Ni). This structure-sensitive principle also works in other semiconductor nanostructures, such as ZnSe and ZnO [45,57]. For instance in Ref. [45], besides the dominant (111) planes with the lowest energy (cf., Fig. S2a, Fig. 28d in Ref. [45]), small fraction of (001) planes are also at the Au/ZnSe interfaces. It proves that the lowest surface energy may be not necessary in the nanowire's nucleation and growth. By performing the 2D FFT pattern on the ZnSe nanowires, it is further found that (1) (1,̄ 1, 1), (1,̄ 1,̄ 1), and (0, 2̄ , 0) planes of ZnSe and Au are parallel with each other in the [1,̄ 2, 1] oriented nanowire (cf., Fig. S2a, Fig. 28d in Ref. [45]); (2) (1, 1,̄ 1̄ ) and (1, 1, 1̄ ) planes of ZnSe and Au are parallel with each other in the [1,̄ 0, 1) oriented nanowire (cf., Fig. S2b, Fig. 28e in Ref. [45]); (3) (1, 1, 1̄ ) and (1,̄ 1, 1) planes of ZnSe and Au are almost parallel with each other in the [0, 1,̄ 0] oriented nanowire (cf., Fig. S2c, Fig. 28f in Ref. [45]). Careful calculation finds that the {111}ZnSe and {002}ZnSe planes respectively match the {111}Au and {002}Au planes with a mismatch of 28.04% in lattice spacing in reference to ZnSe (cf., (5.67–4.08)/5.67). In Ref. [57], the interface of the [0001] growing ZnO nanowire is composed of (0001)ZnO and (020)Sn planes (cf., Fig. S3a, Fig. 7a in Ref. [57]). The atoms in the ZnO (0001)ZnO plane have 6-fold symmetry. The angle between (101)Sn and (1,̄ 0, 1)Sn is 57.11. Thus, the atoms in the (020)Sn plane have quasi-6-fold symmetry. Two of the {0, 1, 1,̄ 0}ZnO planes match the {101}Sn planes with a lattice mismatch as small as 0.7% in reference to ZnO. The third one matches the {200}Sn with a 3.6% lattice mismatch. The interface of the [0, 1, 1,̄ 0] growing ZnO nanobelts is composed of (0, 1, 1,̄ 0)ZnO and (200)Sn planes (cf., Fig. S3b, Fig. 7b in Ref. [57]). The lattice mismatches between {2, 1,̄ 1,̄ 0}ZnO and (002)Sn and {0, 1, 1,̄ 0}ZnO and (101)Sn are 2.1% and 0.7%, respectively. The interface of the [2, 1,̄ 1,̄ 0] growing ZnO nanobelt is composed of (0, 1, 1,̄ 0)ZnO and (100)Sn planes (cf., Fig. S3c, Fig. 7c in Ref. [57]). The lattice mismatches between (0001)ZnO and (020)Sn, and (0, 1, 1,̄ 0)ZnO and (200)Sn planes are respectively about 3.6% and 12.0%. It is also found that the ordered planes of the Sn catalyst at the ZnOnanostructure/Sn-catalyst interface playing an important role in initiating the nucleation and growth of the ZnO nanostructures [57]. As the temperature drops to room temperature, Sn preserves the orientation as defined by the interface with the ZnO nanostructures. It is in turn inferred that the ordered planes of the metallic catalyst at the wire/catalyst interface also control the growth orientation of the nanowire Table 1.

3.4. Determinant factors of growth orientation

4. Conclusion

In this work, Si nanowires are of many growth orientations instead of certain preferred ones such as o112 4, o110 4 or o111 4 that are reported in Refs. [15,36,46]. It was commonly

Thermal annealing of a nickel coated silicon (Si) wafer gives rise to highly crystalline Si nanowires in many growth orientations. The nanowires prefer to grow in the o112 4 orientation.

32

F.J. Li et al. / Journal of Crystal Growth 404 (2014) 26–33

Fig. 4. Interfacial structure between a Si nanowire and the seeding Ni catalyst particle. (a) A Si nanowire is seeded by an elliptic Ni particle. (b) EDX spectrum generated from the nanocluster. (c) SAED systematic rows generated from the area covering both the ending nanocluster and the connected nanowire, which are indexed for the zone axis of Ni [112]. (d) High resolution interfacial structure between the seeding Ni catalyst particle and the grown Si nanowire.

F.J. Li et al. / Journal of Crystal Growth 404 (2014) 26–33

Table 1 Statistical distribution of Si nanowires along different growth orientations. Statistics

o 1124

Quantity 10 Percentage (%) 30.3

o 1104

o1114

o001 4

o 113 4

o 133 4

9 27.3

8 24.3

4 12.1

1 3.0

1 3.0

Surface energy is not the necessary in the nucleation and growth of nanowires. The nanowire's growth follows certain structuresensitive seeding principle with the metal catalyst to minimize the mismatch in lattice spacing and dihedral angle at the wire/catalyst interface. The growth orientation of nanowires is determined by the ordered planes of the metallic catalyst at the wire/catalyst interface at the onset of the growth.

Acknowledgments This work was supported by the Singapore Ministry of Education's Research Grant T208A1218 ARC4/08. This work is also supported by CISRI Grant 11020990A.

Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jcrysgro.2014.06.033. References [1] Y. Xia, P. Yang, Y. Sun, Y. Wu, B. Mayers, B. Gates, Y. Yin, F. Kim, H. Yan, Adv. Mater. 15 (2003) 353. [2] Y. Huang, X. Duan, C.M. Lieber, Small 1 (2005) 142. [3] A.P. Goodey, S.M. Eichfeld, K.K. Lew, J.M. Redwing, T.E. Mallouk, J. Am. Chem. Soc. 129 (2007) 12344. [4] Y. Cui, Q. Wei, H. Park, C.M. Lieber, Science 293 (2001) 1289. [5] M.Y. Bashouti, R.T. Tung, H. Haick, Small 5 (2009) 2761. [6] G. Zheng, X.P.A. Gao, C.M. Lieber, Nano Lett. 10 (2010) 3179. [7] C.K. Chan, H. Peng, G. Liu, K. McIlwrath, X.F. Zhang, R.A. Huggins, Y. Cui, Nat. Nanotechnol. 3 (2008) 31. [8] Q. Zhang, W. Zhang, W. Wan, Y. Cui, E. Wang, Nano Lett. 10 (2010) 3243. [9] L. He, D. Lai, H. Wang, C. Jiang, Rusli, Small 8 (2012) 1664. [10] A.I. Boukai, Y. Bunimovich, J. Tahir-Kheli, J.K. Yu, W.A. Goddard Iii, J.R. Heath, Nature 451 (2008) 168. [11] A.M. Morales, C.M. Lieber, Science 279 (1998) 208. [12] R.S. Wagner, W.C. Ellis, Appl. Phys. Lett. 4 (1964) 89. [13] E.I. Givargizov, J. Cryst. Growth 20 (1973) 217. [14] E.I. Givargizov, J. Cryst. Growth 31 (1975) 20. [15] Y. Wu, Y. Cui, L. Huynh, C.J. Barrelet, D.C. Bell, C.M. Lieber, Nano Lett. 4 (2004) 433. [16] J.D. Holmes, K.P. Johnston, R.C. Doty, B.A. Korgel, Science 287 (2000) 1471. [17] A.T. Heitsch, D.D. Fanfair, H.Y. Tuan, B.A. Korgel, J. Am. Chem. Soc. 130 (2008) 5436.

33

[18] N. Zakharov, P. Werner, L. Sokolov, U. Gösele, Physica E: Low-Dimens. Syst. Nanostruct. 37 (2007) 148. [19] L. Schubert, P. Werner, N.D. Zakharov, G. Gerth, F.M. Kolb, L. Long, U. Gösele, T.Y. Tan, Appl. Phys. Lett. 84 (2004) 4968. [20] H.F. Yan, Y.J. Xing, Q.L. Hang, D.P. Yu, Y.P. Wang, J. Xu, Z.H. Xi, S.Q. Feng, Chem. Phys. Lett. 323 (2000) 224. [21] B.T. Park, K. Yong, Nanotechnology 15 (2004) S365. [22] T.X. Nie, Z.G. Chen, Y.Q. Wu, J.L. Wang, J.Z. Zhang, Y.L. Fan, X.J. Yang, Z.M. Jiang, J. Zou, J. Phys. Chem. C 114/36 (2010) 15370. [23] F.J. Li, S. Zhang, J.H. Kong, J. Guo, X.B. Cao, B. Li, Thin Solid Films 534 (2013) 90. [24] F.J. Li, S. Zhang, J.H. Kong, W.L. Zhang, Nanosci. Nanotechnol. Lett. 3/2 (2011) 240. [25] H. Kohno, S. Takeda, e-J. Surf. Sci. Nanotechnol. 3 (2005) 131. [26] F.J. Li, S. Zhang, J.W. Lee, D. Zhao, J. Cryst. Growth 381 (2013) 87. [27] Y.J. Xing, D.P. Yu, Z.H. Xi, Z.Q. Xue, Appl. Phys. A: Mater. Sci. Process. 76 (2003) 551. [28] R. Rurali, Rev. Mod. Phys. 82 (2010) 427. [29] D.B. Migas, V.E. Borisenko, J. Appl. Phys. 105 (2009) 104316. [30] M.F. Ng, L. Zhou, S.W. Yang, L.Y. Sim, V.B.C. Tan, P. Wu, Phys. Rev. B:Condens. Matter Mater. Phys. 76 (2007) 155435. [31] M.F. Ng, L. Shen, L. Zhou, S.W. Yang, V.B.C. Tan, Nano Lett. 8 (2008) 3662. [32] D.B. Migas, V.E. Borisenko, Nanotechnology 18 (2007) 375703. [33] I. Ponomareva, M. Menon, E. Richter, A.N. Andriotis, Phys. Rev. B—Condens. Matter Mater. Phys. 74 (2006) 125311. [34] M.S. Steighner, L.P. Snedeker, B.L. Boyce, K. Gall, D.C. Miller, C.L. Muhlstein, J. Appl. Phys. 109 (2011) 033503. [35] A. Mikkelsen, J. Eriksson, E. Lundgren, J.N. Andersen, J. Weissenreider, W. Seifert, Nanotechnology 16 (2005) 2354. [36] Y. Cui, L.J. Lauhon, M.S. Gudiksen, J. Wang, C.M. Lieber, Appl. Phys. Lett. 78 (2001) 2214. [37] C.X. Wang, M. Hirano, H. Hosono, Nano Lett. 6 (2006) 1552. [38] H. Jagannathan, M. Deal, Y. Nishi, J. Woodruff, C. Chidsey, P.C. McLntyre, J. Appl. Phys. 100 (2006) 024318. [39] M.S. Song, J.H. Jung, Y. Kim, Y. Wang, J. Zou, H.J. Joyce, Q. Gao, H.H. Tan, C. Jagadish, Nanotechnology 19 (2008) 125602. [40] S.G. Ihn, J.I. Song, T.W. Kim, D.S. Leem, T. Lee, S.G. Lee, E.K. Koh, K. Song, Nano Lett. 7 (2007) 39. [41] C.X. Shan, Z. Liu, S.K. Hark, Appl. Phys. Lett. 90 (2007) 193123. [42] S.A. Fortuna, J. Wen, I.S. Chun, X. Li, Nano Lett. 8 (2008) 4421. [43] A. Lugstein, M. Steinmair, Y.J. Hyun, G. Hauer, P. Pongratz, E. Bertagnolli, Nano Lett. 8 (2008) 2310. [44] S.A. Dayeh, E.T. Yu, D. Wang, Nano Lett. 7 (2007) 2486. [45] N. Wang, Y. Cai, R.Q. Zhang, Materials Science and Engineering R: Reports 60 (2008) 1. [46] C.P. Li, C.S. Lee, X.L. Ma, N. Wang, R.Q. Zhang, S.T. Lee, Adv. Mater. 15 (2003) 607. [47] H.J. Scheel, T. Fukuda, Crystal Growth Technology, J. Wiley, Chichester, West Sussex; New York, 2003 (c2003). [48] C.H. Chiang, J.W. Ci, W.Y. Uen, S.M. Lan, S.M. Liao, T.N. Yang, J. Electrochem. Soc. 159 (2012) H112. [49] S.W. Jee, J. Kim, J.Y. Jung, H.D. Um, S.A. Moiz, B. Yoo, et al., Appl. Phys. Lett. 97 (2010) 042103. [50] S. Zhang, X.L. Bui, Y.Q. Fu, Surf. Coat. Technol. 167 (2003) 137. [51] F.J. Li, S. Zhang, J. Guo, B. Li, Nanosci. Nanotechnol. Lett. (2014), http://dx.doi. org/10.1166/nnl.2014.1817 (In press). [52] V. Schmidt, S. Senz, U. Gösele, Nano Lett. 5 (2005) 931. [53] N. Wang, Y.H. Tang, Y.F. Zhang, C.S. Lee, S.T. Lee, Phys. Rev. B—Condens. Matter Mater. Phys. 58 (1998) R16024. [54] F.J. Li, S. Zhang, J.W. Lee, Thin Solid Films 558 (2014) 75. [55] S. Hartland, Surface and Interfacial Tension: Measurement, Theory, and Applications, Marcel Dekker, New York; Basel, 2004 (c2004). [56] H. Okamoto, Desk Handbook: Phase Diagrams for Binary Alloys, ASM International, Materials Park, Ohio, 2000 (c2000). [57] Y. Ding, P.X. Gao, Z.L. Wang, J. Am. Chem. Soc. 126 (2004) 2066.