Experimental imaging of silicon nanotubes

APPLIED PHYSICS LETTERS 86, 231901 共2005兲

Experimental imaging of silicon nanotubes M. De Crescenzi, P. Castrucci, and M. Scarselli Dipartimento di Fisica and Unita’ INFM, Universita’ di Roma “Tor Vergata,” 00133 Roma, Italy

M. Diociaiuti Dipartimento di Tecnologie e Salute, Istituto Superiore di Sanita’, 00161 Roma, Italy

Prajakta S. Chaudhari, C. Balasubramanian,a兲 Tejashree M. Bhave, and S. V. Bhoraskar Department of Physics, University of Pune, Pune-411 007, India

共Received 16 September 2004; accepted 6 May 2005; published online 31 May 2005兲 Transmission electron microscopy 共TEM兲, electron energy loss near edge structures 共EELNES兲 and scanning tunneling microscopy 共STM兲 were used to distinguish silicon nanotubes 共SiNT兲 among the reaction products of a gas phase condensation synthesis. TEM images exhibit the tubular nature with a well-defined wall. The EELNES spectra performed on each single nanotube show that they are constituted by nonoxidized silicon atoms. STM images show that they have diameter ranging from 2 to 35 nm, have an atomic arrangement compatible with a puckered structure and different chiralities. Moreover, the I-V curves showed that SiNT can be semiconducting as well as metallic in character. © 2005 American Institute of Physics. 关DOI: 10.1063/1.1943497兴 The discovery of carbon nanotubes1 has stimulated intense studies concerning one-dimensional nanostructures for their potential applications in nanotechnology. For silicon, the formation of nanowires and nanoclusters is greatly favoured by the sp3 diamondlike hybridization. There are reports on growth of Si nanotubes;2 however these tubes have a large diameter 共艌50 nm兲, a huge wall thickness of 4–6 nm, and are embedded in SiOx layers. Theoretical predictions, however, have suggested the stability of a single sheet of silicon 共rolled up to form a tubular structure兲 in the nanometer scale and proposed structural models involving either sp2 or sp3, or sp2-sp3 mixed hybridization.3–7 We report transmission electron microscopy 共TEM兲, electron energy loss near edge spectroscopy 共EELNES兲, and scanning tunneling microscopy 共STM兲 studies on the reaction products of a gas-phase condensation synthesis. The first two techniques evidenced nanotubes and their silicon, mostly nonoxidized, content. STM resolved their surface atomic network and determined their chirality. Comparisons between calculations and atomically resolved STM images suggested an atomic arrangement compatible with a puckered structure. Moreover, we found that these silicon tubular structures can be semiconducting as well as metallic in character. Si nanotubes have been synthesized in gram quantities by the dc-arc plasma method. Details of the experimental apparatus are described elsewhere.8 The arc current between the electrodes, one constituted by high purity silicon powder, was maintained at 75 A; the voltage was about 30 V. No gas or metallic elements have been used as catalysts. The synthesis products are off-white. TEM-EELNES measurements were performed by an Energy Filtering ZEISS 902 共80 keV兲 apparatus while nanoselected area electron diffraction 共SAED兲 has been carried out using a Philips EM430 共300 keV兲 apparatus. STM images were recorded with an OMICRON equipment operating in ultrahigh vacuum and at room temperature. The reaction products were dispersed with an isopropyl alcohol droplet on a carbon-coated 共about 40 Present address: INFN 共Istituto Nazionale di Fisica Nucleare兲, Laboratori Nazionali di Frascati, Via E. Fermi 40, 00044 Frascati, Italy.

a兲

nm-thick兲 copper grid for TEM, on a gold grid 共mesh 1000兲 for SAED and on a freshly cleaved highly oriented pyrolitic graphite 共HOPG兲 for STM studies. TEM analysis showed that the reaction products are constituted by nanoclusters of silicon and tubular structure, in the ratio of ten to one. These latter are characterized by different diameters 关Fig. 1共a兲兴 and their length amounts to hundreds of nanometers. Figure 1共b兲 shows a tubular nanostructure of about 7-nm average diameter. The inner part of these nanostructures appears homogeneously bright with no features of encapsulated particles. The application of SAED to small areas containing only one

FIG. 1. 共Color online兲 共a兲 TEM image reporting several silicon nanotubes and nanoparticles. 共b兲 TEM image reporting a silicon nanotube with diameter of about 7 nm and a typical line profile 共c兲, 共d兲 EELNES 共electron energy loss near edge structure兲 at the Si-L2,3 edge recorded in a circular area of 30-nm diam, centered on a single tube.

0003-6951/2005/86共23兲/231901/3/$22.50 86, 231901-1 © 2005 American Institute of Physics Downloaded 01 Jun 2005 to 160.80.2.4. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

231901-2

De Crescenzi et al.

FIG. 2. 共Color online兲 Top panel: STM image 共200 nm⫻ 200 nm兲 showing two silicon straight nanotubes, lying on the highly oriented pyrolic graphite surface together with some silicon nanoparticles. Itunn = 0.8 nA; Vbias = 0.7 V. Lower image: The line profile obtained along the indicated arrow.

single nanotube did not show any diffraction pattern, suggesting that the inner volume is not made of any periodic atomic arrangement. On the other hand, SAED 共same diaphragm and magnification兲 performed on the sample areas, but including spherical nanoparticles, gave rise to patterns characterized by wide rings typical of a polycrystalline material. This demonstrates that these strecthed nanostructures are empty. Moreover, TEM images clearly show the presence of a dark line, of the order of nanometer 关see the line profile in Fig. 1共c兲兴, running along the edges of these nanostructures. The occurrence of this line has been discussed by Quin and Peng9 in a paper describing the imaging by HRTEM of a single SWCNT 共single wall carbon nanotube兲. In our experimental conditions 共phase contrast and Scherzer defocus at about 32 nm兲 the image simulation leads to a dark line, in accordance with our observation. All these considerations suggest us that we are dealing with actual single wall nanotubes.10 In Fig. 1共d兲 we report the Si-L2,3 EELNES spectrum recorded on the tube shown in Fig. 1共b兲. This spectrum has been obtained from a circular area of 30-nm diam, centered on the tube itself. The presence of a signal different from the bare carbon background of the grid demonstrates that this structure is formed by silicon atoms. In particular, the spectral feature located at about 101.5 eV corresponds to a dipole transition from the Si 2p shell.11–14 Moreover, the absence of the characteristic edge resonance for silicon oxide in the 105–110 eV region and for a silicon carbon compound in the 103–105 eV range, confirms the absence of any oxygen contamination14 or silicon carbide formation.11 The former hint is confirmed by the negligible signal of elemental oxygen detected at its K edge, located at about 533 eV. SiC alloy or carbon nanotubes formation is excluded also after a careful check of the carbon content, performed by comparing

Appl. Phys. Lett. 86, 231901 共2005兲

FIG. 3. 共Color online兲 Upper panel: atomically resolved STM image 共4 nm⫻ 4 nm兲 of a part of the lateral surface of a silicon nanotube. The lower and the upper lines represent the armchair tube axis direction and the normal to the tube axis direction, respectively. Itunn = 0.8 nA; Vbias = −0.5 V. Lower panel left: I-V curves obtained by scanning tunneling spectroscopy on six individual silicon nanotubes characterised by different chiralities. Lower panel right: normalized conductances V / I共dI / dV兲.

the intensity of the carbon K edge recorded on the bare grid 共on the same electron area兲 and that detected on the tube. We stress that the EELNES analysis excludes presence of other elements such as metals, within the limits of the experimental sensitivity 共0.5–1.0 %兲. STM characterization has been performed on different surface regions of the sample and the nanotubes were found to lay on the HOPG surface as tangles, as bundles of a few tubes or alone. Their lengths amount to hundreds of nanometers and sometimes few micrometers. Their measured diameter ranges between 2 and 35 nm. In Fig. 2 we report the STM image 共200⫻ 200 nm2兲 of two straight nanotubes together with some silicon nanoparticles. Though the tubes appear to have the same height, they have a very different diameter, one being almost three times smaller than the other 共see the line profile in Fig. 2兲. This could mean that the tubes experience a strong radial compression, induced by the van der Waals interactions between the nanotubes and the substrate. For carbon nanotubes15 this happens in case of a low number of inner shells or a single wall, because the multiwalled tubes should show a diameter comparable with their height. Several noncontact atomic force microscopy measurements on these samples, confirmed this deformation effect for large nanotubes. We succeeded in achieving atomically resolved STM images for most of the tubes. It is worth noting that if an oxide sheath covers the tubes, no imaging with atomic resolution would have been possible due to its insulating nature.16 An atomically resolved STM image 共4 ⫻ 4 nm2兲 of a small area of a nanotube lateral surface is reported in Fig. 3, upper panel. This image is a magnified

Downloaded 01 Jun 2005 to 160.80.2.4. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

231901-3

Appl. Phys. Lett. 86, 231901 共2005兲

De Crescenzi et al.

version of the straight tubes reported in Fig. 2 and shows a honeycomb lattice with a feature in the center. If we consider a hexagonal net, this means that we are observing three over six atoms. Therefore, a chirality can be associated to each tube, by identifying the armchair tube axis direction relative to the sample tube axis, as for carbon nanotubes 共see the two arrows in the STM image reported in Fig. 3, the upper one is parallel to the tube diameter兲.10 Furthermore, we measured the I-V curves for several atomically resolved tubes and we obtained the normalised conductance, V / I共dI / dV兲, which has been shown to provide a good measurement of the main features in the local density of electronic states for metals and semiconductors. In Fig. 3 lower panel, we display a selection of I-V and normalized conductance curves for six individual nanotubes characterized by different chiralities.17 Our results give hints of a metallic nanotube behavior for chiral angles different from 0°, i.e., for those nanotubes that are not armchair. Although we have probed a large number of nanotubes, we do not know, if this result can be considered as a general one as discussed by Zhang et al.6 for nanometer-scaled nanotubes or, if the tube diameter can also play an important role as for carbon nanotubes or for silicon nanowires.18 For the armchair nanotube, we measured an energy gap of 1.27 eV and a diameter of about 24 nm. It is worth noting that if we are dealing with silicon nanowires, no metallic behavior would be detected.18–20 An important point of debate is the silicon atomic arrangement to form silicon nanotubes and therefore its hybridization. Three theoretical approaches have already been reported. The first one assumes a Si atomic network having sp2 hybridization, such as graphene,3 and calculates how much this is unfavorable with respect to diamond structure. The second one predicts puckered 共with a corrugated surface兲 sp3 hybridized silicides or SiH sheets as precursors of silicon nanotubes.4–6 The third one mixes sp2 and sp3 hybridization.7 The atomic averaged 共over one hundred STM images兲 distance is compatible with the Si–Si bond angle and bond values obtained by theoretical calculations for puckered and sp3 hybridized tubular silicon structures.4–6 Nonetheless, we cannot completely exclude the presence of sp2 hybridization because of the surprising inertness of these nanotubes to oxygen and carbon contaminations.

In conclusion, we report experimental evidence of having synthesized clean silicon nanotubes showing a very thin wall smaller than that of all silicon nanotubes reported up to now. We found that they are organized in a puckered lattice and they can assume several chiralities showing metallic as well as semiconducting character. This novel form of aggregation for Si is expected to be useful for the fabrication of novel nanoelectronic devices and circuitry. S. Iijima, Nature 354, 56 共1991兲. S. Y. Jeong, J. Y. Kim, H. D., Yang, B. N. Yoon, S-H Choi, H. K. Kang, C. W. Yang, and Y. H. Lee, Adv. Mater. 共Weinheim, Ger.兲 15, 1172 共2003兲. 3 S. B. Fagan, R. J. Baierle, R. Mota, A. J. R. da Silva, and A Fazzio, Phys. Rev. B 61, 9994 共2000兲. 4 G. Seifert, Th. Köhler, H. M. Urbassek, E. Hernandez, and Th. Frauenheim, Phys. Rev. B 63, 193409 共2001兲. 5 R. Q. Zhang, S. T. Lee, C. K. Law, W. K. Li, and B. K. Teo, Chem. Phys. Lett. 364, 251 共2002兲. 6 M. Zhang, Y. H. Kan, Q. J. Zhang, Z. M. Su, and R. S. Wang, Chem. Phys. Lett. 379, 81 共2003兲. 7 J. W. Kang, J. J. Seo, and H. J. Huang, J. Nanosci. Nanotechnol. 2, 687 共2002兲. 8 C. Balasubramanian, V. P. Godbole, V. K. Rohatgi, A. K. Das, and S. V. Bhoraskar, Nanotechnology 15, 370 共2004兲. 9 C. Qin, and L. M. Peng, Phys. Rev. B 65, 155431 共2002兲 10 H. Dai, Surf. Sci. 500, 218 共2002兲. 11 W. M. Skiff, R. W. Carpenter, and S. H. Lin, J. Appl. Phys. 62, 2439 共1987兲. 12 H. Ma, S. H. Lin, R. W. Carpenter, and O. F. Sankey, J. Appl. Phys. 68, 288 共1990兲. 13 H. Ma, S. H. Lin, R. W. Carpenter, and O. F. Sankey, Phys. Rev. B 44, 13393 共1991兲. 14 X. H. Sun, Y.-H. Tang, P. Zhang, S. J. Naftel, R. Sammynaiken, T. K. Sham, H. Y. Peng, Y.-F. Zhang, N. B. Wong, and S. T. Lee, J. Appl. Phys. 90, 6379 共2001兲. 15 T. Hertel, R. E. Walkup, and P. Avouris, Phys. Rev. B 58, 13870 共1998兲. 16 D. D. D. Ma, C. S. Lee, and T. S. Lee, Appl. Phys. Lett. 79, 2468 共2001兲. 17 We calibrated the x-y piezos for tip scanning by measuring the atomic image of the HOPG substrate detected in the same acquisition. Before and after the I-V measurements on a SiNT, measurements were made on HOPG substrate to ensure their reliability. The I-V curves of SiNT of various diameters were measured at constant tip height 共about 0.1 nm兲 where atomically resolved STM images were obtained. A series of I-V curves performed at various tip-to-surface distances showed that the I-V curves were relatively insensitive to the tip height. The tip convolution has been estimated by resolving a monoatomic step of the HOPG. 18 D. D. D. Ma, C. S. Lee, F. C. K. Au, S. Y. Tong, and S. T. Lee, Science 299, 1874 共2003兲. 19 A. J. Read, R. J. Needs, K. J. Nash, L. T. Canham, P. D. J. Calcott, and A. Qteish, Phys. Rev. Lett. 69, 1232 共1992兲. 20 B. Delley and E. F. Steigmeier, Appl. Phys. Lett. 67, 2370 共1995兲. 1 2

Downloaded 01 Jun 2005 to 160.80.2.4. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp