Self-assembly of Si nanostructures

26 May 2000

Chemical Physics Letters 322 Ž2000. 312–320 www.elsevier.nlrlocatercplett

Self-assembly of Si nanostructures Yan Qiu Zhu a , Wen Kuang Hsu a , Nicole Grobert a , Mauricio Terrones a , Humberto Terrones a , Harold W. Kroto a , David R.M. Walton a,) , Bing Qing Wei a

b

School of Chemistry, Physics and EnÕironmental Science, UniÕersity of Sussex, Brighton, BN1 9QJ, UK b Department of Mechanical Engineering, Tsinghua UniÕersity, Beijing 100084, China Received 27 January 2000; accepted 23 March 2000

Abstract Flower-like Si nanostructures are formed in high yield, by heating an SiO 2 plate at ca. 1600 8C ŽTa heater. under Ar Ž100 Torr.. The product consists of metal-free cubic phase Si nanowires surmounted by bulbous Si tips. HRTEM observations show that the nanowires contain kink and twinning defects, whereas the tips are generally well-crystallized and covered with a thin layer of amorphous SiO x Ž x s 1–2.. A growth model is proposed to account for these observations. q 2000 Elsevier Science B.V. All rights reserved.

1. Introduction Because of their valuable semiconducting, mechanical and optical properties, as well as their potential applications in nano-electronic and -technology, Si-based nanoscale materials have attracted much attention. They are, for example, considered as candidates for 1-dimensional quantum transistors, composites, and light-emitting devices w1,2x. To date, Si, SiC, Si 3 N4 and SiO x nanowires, nanorods and nanocables have been produced by various methods, e.g. photolithographic etching w3x, vapour–liquid– solid ŽVLS. growth w4–7x, laser ablation w8–10x, chemical vapour deposition ŽCVD. w11x, thermal evaporation w12x and template growth w13,14x. The properties of Si nanowires are influenced directly by

) Corresponding author. Fax: q44-1273-677196; e-mail: [email protected]

size, crystal defects and the presence of impurities, so the production of pure nanowires with minimal structural defects and of uniform size is an important goal. However, for CVD and VLS methods, metal particles Že.g. Au or Fe. are generally employed as catalysts w5,7x, so that some impurities are inevitable. Laser ablation routes may not require metal catalysts w15x, unfortunately they result in significant defect problems so that scale-up may be difficult. Previously, we successfully generated SiO x Ž x s 1–2. nanoflowers and SiC nanowires by heating SiC powder mixed with Co or Fe as a catalyst and using a carbon foil heater, under Ar or CO Žca. 100–200 Torr. w6,7,16x. In this Letter, we describe a new, high yield, route to ca. 10–100 nm diameter cubic phase Si nanowires formed by heating a SiO 2 plate supported on a Ta metal strip heater, at ca. 1600 8C under Ar Žca. 100 Torr.. SEM, XRD, HRTEM, ED and EDX analytical techniques were used to characterise the products which consist of bulbous Si nanoparticles attached to the ends of the nanowires.

0009-2614r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 9 - 2 6 1 4 Ž 0 0 . 0 0 4 4 0 - 1

Y.Q. Zhu et al.r Chemical Physics Letters 322 (2000) 312–320

313

Fig. 1. SEM images of the Si nanostructures Ža–e.. Ža. Massive amounts of nanowires grown on the upper surface of the SiO 2 plate Žcentre., corresponding to zone A; Žb. high magnification of Ža.; Žc. nanostructures on the edges, corresponding to zone B; Žd. high magnification of Žc., revealing the large tips on the top; Že. severely bent microfibres on the very edge, corresponding to zone C; Žf. diagrammatic of the structures.

Y.Q. Zhu et al.r Chemical Physics Letters 322 (2000) 312–320

314

2. Experimental The equipment was similar to that used previously by us to generate SiO x nanoflowers and SiC nanowires w6,16x. Ta foil Žca. 50 = 5 = 0.025 mm; Goodfellow, 99.85% pure. was positioned under a SiO 2 plate Žca. 10 = 5 = 1 mm; Goodfellow, 99.9 pure. in an Ar atmosphere Ž100 Torr., and a ca. 40 A d.c. was applied. As soon as the foil reached ca. 16008C Žmeasured by optical pyrometry., the temperature was maintained at this level for 30 min. Passage of the current was then discontinued. During the experiments, we observed minute flames along the edges of the SiO 2 plate, presumably due to oxidation of the Ta foil. When this occurred the current was reduced immediately so as to maintain the stipulated temperature. After each experiment, we found that the SiO 2 plate surface was covered evenly with a brown sponge-like substance surrounded, at the edge of the plate, by brown fibre-like bundles. The brown material was removed from the SiO 2 substrate, with the aid of a scalpel, and analysed

directly by XRD. Samples were dispersed ultrasonically in acetone for 10 min, then transferred to a Cu grid coated with a holey carbon film, for TEM examination. Another sample was coated with Au for 4 min, and then monitored in situ by SEM. The XRD examination was carried out on a Siemens D-5000 instrument ŽCu-k a radiation ., with a 0.02 degreermin scanning rate Žoperating at 40 mAr40 kV.. The following equipment was used as appropriate: SEM ŽLeo 5420, 10–20 keV., TEM ŽH-7100, 120 keV; CM200, 200 keV; JEM-4000, 400 keV.; and energy dispersive X-ray ŽEDX. using a Noran Instruments detector attached to a CM-200, element 0 B ..

3. Results SEM examination of the sample ŽFig. 1a. revealed the presence of numerous microscopic sponge-like domains, each consisting of fine curled

Fig. 2. XRD profiles of the products. The upper profile reveals the b-Si structure; the lower shows splits at high 2-theta angles.

Y.Q. Zhu et al.r Chemical Physics Letters 322 (2000) 312–320

315

Fig. 3. TEM images Ža and b. showing the octopus-like feature of the products, and Žc. exhibiting the polygonal tip; Žd. EDX pattern revealing the dominance of Si; Že. very thin nanowires Žca. 10 nm diam.., consisting of b-Si Žinsert ED pattern..

316

Y.Q. Zhu et al.r Chemical Physics Letters 322 (2000) 312–320

nanowires. Close inspection showed the nanowires to be of fairly uniform diameter Žca. 10–100 nm., evenly-distributed over the surface of the plate ŽFig. 1b.. At low resolution, the edges of the sample ŽFig. 1c. resembled the central material. However, at high resolution ŽFig. 1d., different features were identified: e.g. groups of vertical nanowires growing from a spherical tip Žca. 50–500 nm diam... At the very edge, large and severely bent fibres Žca. 1 mm diam. and up to 100 mm long., with rough surfaces, were detected ŽFig. 1e.. Fig. 1f illustrates the different areas Žclassified as A, B and C, respectively. associated with the results described above. XRD results revealed the sample to be dominated by the cubic b-Si phase. A broad peak accompanied by smaller peaks indicated the presence of traces of SiO x and Ta 2 O5 ŽFig. 2a.. The Si peaks were strong and sharp at low 2-theta angles, but were weaker and broadened at higher angles, revealing shoulders ŽFig. 2b, arrowed.. TEM showed the presence of octopus-like structures as a characteristic feature of the Si nanowires ŽFig. 3a, area B.. The nanowire surfaces were smooth and their diameters gradually decreased as the distance from the tips increased. Under TEM, the contrast between the tips and nanowires did not change significantly, indicating their phase uniformity. These structures differ from the SiO x nanoflowers previously reported, which exhibited numerous arms sprouting radially from the central sphere. In this case, only a few arms ŽFig. 3a. sprout from adjacent sites situated in an area localised on one side of the spherical tip. Considering the SEM results ŽFig. 1d., we conclude that the tips are supported by the nanowires. In some cases, a nanowire was observed to have several short branches surmounted by spherical tips ŽFig. 3b.. Occasionally, the tips exhibited a polyhedral structure and a typically hexagonal cross-section, under TEM, as shown in Fig. 3c. Examination of the nanowires and their tips by EDX, indicated that they were of identical composition ŽFig. 3d.. Clearly, the intensity of the Si peak is much stronger than that of O. Despite careful searches, no Ta signals were detected. This result indicates that the nanowires are composed predominantly of Si, with O as a minor Ž- 5%. contaminant. Relatively fine nanowires Žca. 6–10 nm diam.., from area A, were also observed by HRTEM ŽFig.

3e.. Their ED pattern Žinsert. can be indexed as b-Si, consistent with the XRD result.

4. Discussion Based upon our previous studies w16x, we conclude that the Ta of the heater reacts with the SiO 2 plate, according to Eqs. Ž1. and Ž2.:

™ 2Ta O q 5Si 2Ta q 5SiO ™ Ta O q 5SiO 4Ta q 5SiO 2

2

2

2

5

5

Ž 1. Ž 2.

If Eq. Ž1. predominates, Si as opposed to SiO nanowires will form. In practice, XRD analysis showed that material deposited on the top inner wall of the chamber, consisted of Ta 2 O5 and SiO, possibly due to their evaporation at the reaction temperature Žca. 16008C.. SiO vapour may also have condensed on the surfaces of the grown Si nanowires or tips. An important result derived from our experiment is that EDX examination indicated no Ta within the Si tips or the nanowires ŽFig. 3d., despite that fact that Ta must be intimately involved. These nanowires are different from those prepared by the VLS and related methods, such as laser ablation of a mixed Fe and Si target w8,12x. The EDX results do not conflict with the XRD analysis ŽFig. 2. which indicated the presence of small amounts of Ta 2 O5 and SiO phases, produced according to reaction Ž2.. Due to the steep temperature gradient within the SiO 2 plate, covered with Ta foil, only part of the resulting Ta 2 O5 Ž m.p. 18728C. sublimed Žcorresponding to the higher temperature zone.. The Ta 2 O5 residue on the SiO 2 substrate may accumulate with the Si nanowires, however they do not appear to mix. By contrast, residual SiO Ž m.p. 17028C. may easily combine with the growing nanowires, leading to disproportionation of the Si nanowires Žinner core. and SiO 2 Žouter shell.. This transformation resembles that in thermal decomposition and laser ablation approaches to creating Si nanowires ŽEq. Ž3.. w12,15x: 2SiO

™ Si q SiO

2

Ž 3.

The very small O peak, compared to the Si peak ŽFig. 3d., indicates the presence of a minute amount of SiO x in the nanowires and tips.

Y.Q. Zhu et al.r Chemical Physics Letters 322 (2000) 312–320

Considering the various structures found in different areas ŽFig. 1., we surmise that the temperature gradient occurring at the outset may be responsible for such features, by influencing the nanowire growth rate. The largest temperature gradient occurs in zone C, and the highest nanowire growth rate is achieved, thereby leading to large fibres Žca. 1 mm diam... In this context, the marked growth rate seems to result in highly defective nanowires ŽFig. 1e.. By contrast, the smallest temperature gradient is found in zone A, resulting in the most regular nanowires. In zone B Žintermediate temperature gradient. a moderate growth rate results in larger smooth-surface nanowires. HRTEM measuments showed that the nanowire tips consist of well-crystallised Si, covered with a thin amorphous SiO x layer Žca. 1–2 nm, x s 1–2, Fig. 4a., according to the EDX results. However, no obvious amorphous phase could be identified within the nanowires. By contrast with the tips, the nanowires were generally rather poorly crystallised, with twinning and kink defects ŽFig. 4b.. A noteworthy feature of the nanowires is their more-or-less constant tiprbody diameter ratio Žca. 2., although the tip sizes vary considerably ŽFig. 3b and c.. This ratio is fairly consistent with that found in the VLSproduced nanowires Žtypically ca. 1.5–2. w5,6,16x. As regards the Fe-catalysed Si nanowire growth Žgenerally involving a VLS step., the tip size is mainly determined by the initial size of the Fe particle, and the nanowire diameter is mainly controlled by the tip size w8x. In our case, we believe that the function of the in situ formed Si tip is, to some extent, analogous to the metal catalysts ŽAu, Fe or Co.. The tips may also have experienced Si Žand SiO. accumulation andror extrusion stages during nanowire growth, however their size is probably determined by the temperature. For example, when the temperature exceeds 13508C, the Si tip front will form a liquid phase by absorbing SiO or O – according to the Si–O phase diagram. Thus the higher melting points, associated with the Si and SiO 2 phases, may cause extrusion from the Si tips, leading to solid-phase nanowire growth. Because Ta has not been involved in Si nanowire growth Žas verified by EDX., it seems that the localised temperatures play an important role in the experiments. It is obvious that the temperature might affect the degree of Si

317

andror SiO vaporisation and migration. Furthermore, the localised temperature also determines the sizes of the initial Si tips Žformed from liquid phases at temperature ) 14588C.. The latter may further alter some of the initial tip sizes by affecting the nanowire growth rates. In some cases, a low growth rate was achieved Žas in zone B., indicating a high Si Žand SiO. accumulation rate and a low extrusion rate, thus resulting in large tips. However, in zone C, where the nanowire growth rate is highest Ži.e. highest extrusion rate., large nanowires were formed and very few tips were observed. In zone A Žwhere the temperature gradient is lowest., because of its position, accumulation is expected to be difficult Žreaction occurred only on the underside of the SiO 2 plate.. Thus, low accumulation and extrusion results, leading to fine uniform tipless nanowires Žarrowed in Fig. 4b.. Another interesting feature of these nanowires is that they exhibit directional crystal growth ŽFig. 4c, but not the ²111: growth observed when the material is catalysed by Au w17x.. As discussed above, the function of the Si tips is comparable with that of metal catalyst tips during the Si extrusion stage. Therefore, the growth axis of the nanowires is believed to be associated with the crystal orientation of the Si tips. TEM analyses confirmed the crystalline structure of the tips. In the XRD analyses, the highangle diffraction peaks are broad and weak, with obviously shoulders. Because the tips are crystalline, the shoulders appear to arise mainly from the nanowires. Not only are the nanowires relatively poorly crystalline, but they also occur in two typical diameter distributions, i.e. ; 10 nm Žzone A. and ) 100 nm Žzone C.. Poorly crystalline structures may reduce the peak intensities and smaller nanowires widen the peak profiles. Furthermore, since some nanowires were formed from the SiO phase ŽEq. Ž3.., in these Si crystals, lattice deformation may be retained if the transformation is not complete ŽO traces within the Si structures.. This may be responsible for the shoulders on the peaks, which occur at high diffraction angles, due to the high-angle planes associated with long-distance crystalline structures. This explanation is supported by the HRTEM observations ŽFig. 4b.. Because the tips and wires are of similar composition, the continuous accumulation of small amounts

318

Y.Q. Zhu et al.r Chemical Physics Letters 322 (2000) 312–320

Fig. 4. HRTEM images. Ža. Massive Si tip covered with a very thin SiO x layer Žindicated.. The bottom right insert Žmagnified from the upper white square. shows the well-crystallised feature. The lattice fringes are separated by ca. 0.31 nm, consistent with b-Si; Žb. kink defective structure; Žc. different growth directions of the nanowires.

Y.Q. Zhu et al.r Chemical Physics Letters 322 (2000) 312–320

319

attaches to another particle, thus generating linked structures of the type shown in Fig. 3b.

5. Conclusions

Fig. 5. A schematic model for Si nanowire growth. Ža. Nano- or micro-Si tips formed. Žb. Initial nanowire formation. Žc. Adjacent site growth, or alternatively, Žd. Coalescence.

of SiO during growth may facilitate the joining of one nanowire to another nanowire tip, leading to structures shown in Fig. 3b. Meanwhile, different extrusion sites Žnucleation sites for the nanowires. may occur at the Si tip surface Ždue to the polyhedral structure., octopus-like structures result ŽFig. 3a.. These processes are comparable with Co coalescence or bifurcation, which provides the main driving force, for the formation of SiO x nanoflowers w6,16x. The primary driving force for Si nanowire growth appears to lie in the temperature gradients, which result in products, rather different from the 3-D nanoflowers. In view of the SEM results ŽFig. 1d., we conclude that the tips are supported by the nanowires. It is apparent that only a few arms sprout from adjacent sites situated in a localised area of the spherical tip ŽFig. 3a., and that some nanowires have several short branches surmounted by larger tips ŽFig. 3b.. Based on the above observations, we propose a model ŽFig. 5. for Si nanowire growth. Ža. Si ŽEq. Ž1.. accumulates to form nano- or micro- particles on the lower temperature upper surface of the SiO 2 plate. Žb. The resulting SiO ŽEq. Ž2.. accumulates on the Si particle surface. Žc. Si extrudes from this particle, initiating nanowire growth. Žd. In several adjacent sites extrusion occurs leading to octopus-like structures ŽFig. 3a., or alternatively, Že. a nanowire

Octopus-like structures consisting of metal-free cubic Si nanowires and attached Si nanoparticles have been generated in good yield, by heating a SiO 2 plate on Ta at ca. 16008C under Ar. HRTEM observations reveal that the Si nanowires contain kinks and twinning defects, whereas their tips are generally well-crystallised and covered with a distinct amorphous SiO x Ž x s 1–2. thin layer. The effect of the temperature gradient is believed to be the primary reason for nanowire and nanostructure growth. A non-metal-catalysed mechanism has been proposed to account for the facts.

Acknowledgements We thank the Royal Society, the JFCC, ConacytMexico and DGAPA-UNAM IN 107-296 ŽHT., EP´ SRC, and Max-Planck Society Fellowship ŽBQW. for financial support. We are grateful to J. Thorpe and D. Randall ŽSussex. for assistance with TEM and SEM facilities.

References w1x A.P. Alivisatos, Science 271 Ž1996. 933. w2x D.P. Yu, Q.L. Huang, Y. Ding, H.Z. Zhang, Z.G. Bai, J.J. Wang, Y.H. Zou, W. Qian, G.C. Xiong, S.Q. Feng, Appl. Phys. Lett. 73 Ž1998. 3076. w3x H.I. Liu, N.I. Maluf, R.F.W. Pease, J. Vac. Sci. Tech. B 10 Ž1992. 2846. w4x R.S. Wagner, W.C. Ellis, Appl. Phys. Lett. 4 Ž1964. 89. w5x J. Westwater, D.P. Gosain, S. Usui, Phys. Status Solid A-Appl. Res. 165 Ž1998. 37. w6x Y.Q. Zhu, W.K. Hsu, M. Terrones, N. Grobert, H. Terrones, J.P. Hare, H.W. Kroto, D.R.M. Walton, J. Mater. Chem. 8 Ž1998. 1859. w7x Y.Q. Zhu, W.B. Hu, W.K. Hsu, M. Terrones, N. Grobert, T. Karali, H. Terrones, J.P. Hare, P.D. Townsend, H.W. Kroto, D.R.M. Walton, Adv. Mater. 11 Ž1999. 844. w8x A.M. Morales, C.M. Lieber, Science 279 Ž1998. 208.

320

Y.Q. Zhu et al.r Chemical Physics Letters 322 (2000) 312–320

w9x G.W. Zhou, Z. Zhang, D.P. Yu, J. Cryst. Growth 197 Ž1999. 129. w10x Y. Zhang, K. Suenaga, C. Colliex, S. Iijima, Science 281 Ž1998. 973. w11x G.W. Meng, L.D. Zhang, Y. Qin, F. Phillipp, S.R. Qiao, H.M. Guo, S. Y Zhang, Chin. Phys. Lett. 15 Ž1998. 689. w12x N. Wang, Y.H. Tang, Y.F. Zhang, C.S. Lee, I. Bello, S.T. Lee, Chem. Phys. Lett. 299 Ž1999. 237. w13x H. Dai, E.W. Wong, Y.Z. Lu, S.S. Fan, C.M. Lieber, Nature 375 Ž1995. 769.

w14x W.Q. Han, S.S. Fan, Q.Q. Li, Y.D. Hu, Science 277 Ž1997. 1287. w15x N. Wang, Y.F. Zhang, Y.H. Tang, C.S. Lee, S.T. Lee, Appl. Phys. Lett. 73 Ž1998. 3902. w16x Y.Q. Zhu, W.K. Hsu, M. Terrones, N. Grobert, W.B. Hu, J.P. Hare, H.W. Kroto, D.R.M. Walton, H. Terrones, Chem. Mater. 11 Ž1999. 2709. w17x J. Westwater, D.P. Gosain, S. Tomiya, S. Usui, J. Vac. Sci. Tech. B 15 Ž1997. 554.