1865
2007, 111, 1865-1867 Published on Web 01/13/2007
Controlled Growth of Silicon Oxide Nanowires from a Patterned Reagent Feng Wang,*,†,‡ Marek Malac,†,‡ Ray F. Egerton,†,‡ Alkiviathes Meldrum,† Peng Li,† Mark R. Freeman,†,‡ and Jonathan G. C. Veinot§ Department of Physics, UniVersity of Alberta, Edmonton T6G 2G7, Canada, National Institute for Nanotechnology, 11421 Saskatchewan DriVe, Edmonton T6G 2M9, Canada, and Department of Chemistry, 11227 Saskatchewan DriVe, UniVersity of Alberta, Edmonton T6G 2G2, Canada ReceiVed: NoVember 14, 2006; In Final Form: December 18, 2006
A new process to fabricate silicon oxide nanowires (NWs) from a patterned reagent is reported. Arrays of NWs grow on patterned nanodots containing exposed hydrogen silsesquioxane (HSQ)/Fe-SiO2 nanocomposites during annealing at 900 °C. The NWs were seeded by metallic iron nanoparticles, and the resulting microstructure and morphology of the NWs is directly related to the size of the individual iron nanoparticles. The growth process could be dominated by a solid-state transformation mechanism in which iron nanoparticles, originally embedded in a SiO2 matrix, diffuse to the surface and act as nucleation sites, the exposed HSQ being the source for the growing NWs.
Silicon-based nanowires, including crystalline and amorphous silicon and silicon oxide nanowires, have shown promising applications in nanoelectronics and integrated optic devices, such as low dimensional waveguides for functional microphotonics, scanning near field optical microscopy, optical interconnects on optical microchips, biosensors, and optical transmission antennae.1-3 Synthetic methods including chemical vapor deposition,4 laser ablation,3 sol-gel,5 and thermal evaporation,6 have been used for random (i.e., nonpatterned) growth of nanowires (NWs). However, for many potential applications, it is desirable to control the position and size of the NWs so that post-growth manipulation is not required.7,8 Recently, selective growth of silica NWs via a vapor-liquid-solid mechanism9 was achieved using an ion implantation mask.10 Here we show that patterning with much higher (by several orders of magnitude) spatial resolution can be achieved by employing electron-beam lithography. The diameter of the NWs is determined by the iron nanoparticles encapsulated at the end of individual NWs, similar to other reports in Si and GaP nanowires,11,12 but in our case, the NWs grow directly from a patterned reagent, and inflow of a Si-containing gas is not required. The fabrication procedure can be divided into three stages: (i) formation of the Fe-SiO2 nanocomposites (i.e., iron nanoparticles embedded in a SiO2 matrix), (ii) patterning of the reagent, and (iii) the growth of the silicon oxide nanowires. FeSiO2 nanocomposites were fabricated by electron-beam deposition of SiO2 (purity ) 99.999%) and Fe (99.95%) to form a SiO2/Fe/SiO2 trilayer on 50 nm-thick silicon nitride (Si3N4) membranes and annealed in situ at 880 °C, in an ultrahigh vacuum (UHV) system with base pressure 10-10 Torr.13 A 6 nm-thick carbon film was pre-coated on the Si3N4 membrane * Corresponding author. E-mail:
[email protected]. † Department of Physics, University of Alberta. ‡ National Institute for Nanotechnology. § Department of Chemistry, University of Alberta.
10.1021/jp0675476 CCC: $37.00
to suppress charging during electron beam lithography. Following fabrication of the Fe/SiO2 composite, a 60 nm-thick layer of hydrogen silsesquioxane (HSQ; Fox17, Dow Corning, which is a well-known negative planarizing electron beam resist) was spin-coated onto the sample surface. Regular arrays of nanodots of desired dimensions were patterned in the HSQ layer using a Raith 150 electron-beam lithography system. Areas of the Fe/ SiO2 nanocomposite not protected by the exposed HSQ were subsequently removed by dry etching in an argon ion mill, leaving the Fe/SiO2 nanocomposite film only in areas under the exposed HSQ dots. The samples were then annealed for 1 h at 900 °C in a slight overpressure of N2/H2 (95%/5%) forming gas. Microstructural and chemical investigations were carried out in an analytical transmission electron microscope (TEM, JEOL 2010), using bright-field imaging, electron energy-loss spectroscopy (EELS) and energy-loss near-edge structures (ELNES). An array of circular nanodots of 50 nm diameter and 1 µm spacing, is shown in Figure 1a. A limited number of iron nanoparticles (typically less than 5) are present within each nanodot (see, in the inset of Figure 1a); Figure 1b shows a TEM image of an array of NWs grown from the patterned nanodots. NWs grew only from the original patterned HSQ/Fe nanodots, leaving the areas between the dots bare and completely free of NWs. The inset of Figure 1b shows NWs produced from a typical dot, with higher magnification. The length of the NWs on the nanodots is less than 1 µm in this experiment. It is determined by the supply of the stock material and this is controllable, so the length of the NWs can be further tailored in accordance with the requirements of practical applications. It is worth noting that the number of iron nanoparticles is determined by the size of each nanodot and the areal density of Fe nanoparticles. Therefore, it is expected that a suitable choice of Fe nanoparticle growth parameters and HSQ dot diameters can be used to control the number of iron © 2007 American Chemical Society
1866 J. Phys. Chem. C, Vol. 111, No. 5, 2007
Letters
Figure 3. Dependence of the diameter of the nanowires on that of the encapsulated iron particles. The solid line is a fit to the measured data: y ) 0.96x + 2.48 (y ) diameter of the nanowire; x ) diameter of the nanoparticle (exclusive of the shell); the dotted line y ) x is used for comparison.
Figure 1. Bright-field TEM images for (a) arrays of iron nanoparticles after lithographic patterning of 50 nm diameter and 1 um spacing (the inset shows one nanodot containing several nanoparticles) and (b) arrays of nanowires grown from the patterned nanodots (the inset shows nanowires on a nanodot).
Figure 2. Bright-field TEM image of a bundle of individual nanowires projecting from the substrate, with iron nanoparticles encapsulated (darker region) at the NW tips. The insets are selected-area diffraction patterns of iron nanoparticles that are consistent with the bcc-Fe phase.
nanoparticles and consequently the number of NWs grown from each dot. Although the average number of NWs on each nanodot can be controlled, the possibility to grow one individual nanowire from each dot was not demonstrated to date. Using a thick Si3N4 membrane as substrate poses difficulties for TEM investigation; however, this problem was avoided by examining NWs that were suspended between gaps in a broken Si3N4 film, allowing microstructural and chemical analysis without influence from the substrate. Figure 2 shows a TEM image of a bundle of NWs protruding into vacuum from the Si3N4 substrate. The nanowires have a smooth surface and a closed end containing a single iron particle. The nanowires have diameters of 10-30 nm, comparable to the diameter of the iron
nanoparticles encapsulated at the nanowire tips. TEM imaging and diffuse rings in electron-diffraction pattern (not shown here) collected from the body of nanowires showed that the nanowires are noncrystalline. Indexing the electron diffraction pattern collected from the particles at the end of the NWs (inset in Figure 2) indicated that bcc-Fe is the main phase; no reflections associated with iron oxides were found. The diameters of individual nanowires and encapsulated nanoparticles are correlated. A total of 98 nanowires were measured from plan-view TEM images for a quantitative comparison. Figure 3 shows the relation between nanowire diameter and the iron particle diameter. The solid line is a linear least-square fit. The results suggest that the diameter of the nanowires is predominantly determined by the iron nanoparticles, which leads to the possibility of synthesizing diametercontrolled silicon oxide nanowires. Control of Fe seed particle size has been demonstrated through a multiple-layer synthesis procedure.13 Analysis of electron energy-loss and Auger spectra showed that the nanowires are composed of silicon and oxygen. To further clarify the chemical environment of the silicon in the nanowires, the silicon L23 and oxygen K-ionization edges were measured from individual nanowires protruding from the substrate (shown in Figure 2). The Si L23 edge shows a different fine structure than that recorded from nanowires on the Si3N4 substrate (where the spectra are dominated by the substrate), and no nitrogen K-edge was discerned in the energy loss spectra. The integral intensities of oxygen K and silicon L-edges allowed determination of the [O]:[Si] ratio, using O-K and Si-L ionization cross-sections calculated from SIGMAK3 and SIGMAL3 programs,14,15 and the resulting atomic [O]:[Si] ratio was 1.7 ( 0.1. In addition, the near-edge fine structure of the silicon L23-edge of SiO2 and elemental Si differ near the threshold energy,16,17 negating the possibility that the wires contain elemental Si. The morphology of the NWs (with metal particles attached to the end) and the linear relation between the diameters of the nanowires and the encapsulated nanoparticles suggest the operation of a classical vapor-liquid-solid (VLS) growth mechanism.9 Nevertheless no vapor source such as silane (SiH4) or other gaseous silicon compound was provided in the present growth process. Furthermore, the presence of liquid iron droplets, on which gaseous silicon precursor might produce silicon for the growth of nanowires,2,9,11 is unlikely in this system because the annealing temperature used (900 °C) was much lower than both the melting point of iron and the eutectic point for iron-silicon or iron-silicon oxides,18 even when the surface
Letters
J. Phys. Chem. C, Vol. 111, No. 5, 2007 1867 regarding the nanowire growth mechanism and are subject of further investigation. In summary, a process for fabricating arrays of silicon oxide nanowires from a patterned reagent has been developed, in which the diameter of the nanowires depends on the nanoparticles attached at the end. Such an unusual growth process, being unexplainable by the VLS and SLS mechanisms, may be attributed to direct solid-state transformation.
Figure 4. Proposed growth process of nanowires: (a) as-produced Fe-SiO2 nanocomposites, with spin-coated HSQ on top; (b) migration of iron nanoparticles onto the top of the nanodots; (c) diffusion of Si atoms; (d) growth of nanowires onto the surface of nanodot.
energy of the cluster is considered.19,20 Therefore, it is unlikely that the present nanowires grew by a VLS-like mechanism. The SLS (solid-liquid-solid) mechanism can also be excluded, in view of the top-growth mode and the low processing temperature in this growth process.21 The direct solid-state transformation mechanism has been proposed for the formation of amorphous silicon oxide nanowires from silica films, in which nickel nanoparticle nucleating centers are formed on the surface from nickel atom diffusion through a reduction layer.22 This mechanism provides a possible explanation for the growth observed in our experiments. Iron nanoparticles, initially embedded in the SiO2 matrix, diffuse to the surface when the sample is heated to 900 °C. The formation of the iron nanoparticle on the surface has been observed in our previous experiments;23 the diffusion mechanism remains to be determined. Fe particles at the exposed HSQ surface act as NW nucleation sites. NWs nucleate underneath the iron nanoparticles and grow by the diffusion of the Si atoms from the underlying exposed HSQ. Our proposed mechanism is summarized in Figure 4. HSQ is a well-defined cage molecule with an empirical formula (HSiO3/2)n. It has been reported, in the absence of conclusive structural characterization, that, upon exposure to electron-beam irradiation, some of the Si-H bonds are broken and an insoluble crosslinked siloxane network is formed. 24 This ill-defined, crosslinked network is expected to collapse upon annealing at 900 °C to form a “SiOx-like” film containing regions rich in low-valent Si. The diffusion and nucleation of oxide-encapsulated, low-valent Si on metal nanoparticles has been reported.25 In this regard, it is reasonable that the present “SiOx-like” film, produced from annealing exposed HSQ, enhances the nucleation and one-dimensional growth of the silicon NWs, in a similar fashion to that proposed for “SiO” by Wang et al.25 In the present system, as-grown nanowires likely form as noncrystalline Si nanowires that subsequently oxidize rapidly to SiO2 upon exposure to the ambient atmosphere. In support of this proposed mechanism, control experiments using an Fe/SiO2 nanocomposite, with no overlaying exposed HSQ layer as well as with an Fe/SiO2 nanocomposite bearing an unexposed HSQ film, afforded no NWs when annealed under identical conditions. These experiments clearly show that a crosslinked SiOx-layer obtained from electron-beam exposure of HSQ is crucial to the formation of the present NWs. At present, the exact role of the crosslinked HSQ remains unclear. Structural analysis of beam-crosslinked HSQ films as well as an in situ study of as-grown nanowire structure and chemical composition are expected to provide valuable information
Acknowledgment. This work was supported by NRC and NSERC. The electron-beam lithographic patterning was done at Nanofab in the University of Alberta. The authors thank Dr. Ken Bosnick and Dr. Yucheng Lan for helpful discussions and Don Mullin and Greg Popowich for technical assistance. F.W. acknowledges support from the Killam Trust. References and Notes (1) Zheng, B.; Wu, Y.; Yang, P.; Liu, J. AdV. Mater. 2002, 14, 122. (2) Carter, J. D.; Qu, Y.; Porter, R.; Hoang, L.; Masiel, D. J.; Guo, T. Chem. Commun. 2005, 2274. (3) Yu, D. P.; Hang, Q. L.; Ding, Y.; Zhang, H. Z.; Bai, Z. G.; Wang, J. J.; Zou, Y. H.; Qian, W.; Xiong, G. C.; Feng, S. Q. Appl. Phys. Lett. 1998, 73, 3076. (4) Hu, J. Q.; Jiang, Y.; Meng, X. M.; Lee, C. S.; Lee, S. T. Chem. Phys. Lett. 2003, 367, 339. (5) Zhang, M.; Bando, Y.; Wada, L.; Kurashima, K. J. Mater. Sci. Lett. 1999, 18, 1911. (6) Liang, C. H.; Zhang, L. D.; Meng, G. W.; Wang, Y. W.; Chu, Z. Q. Non-Cryst. Solids 2000, 277, 63. (7) Friedman, R. S.; McAlpine, M. C.; Ricketts, D. S.; Ham, D.; Lieber, C. M. Nature 2005, 434, 1085. (8) Gu, Q.; Dan, H.; Cao, J.; Zhao, J.; Fan, S. Appl. Phys. Lett. 2000, 76, 3020. (9) Wagner, R. S.; Ellis, W. c. Appl. Phys. Lett. 1964, 4, 89. (10) Sood, D. K.; Sekhar, P. K.; Bhansali, S. Appl. Phys. Lett. 2006, 88, 143110. (11) Cui, Y.; Lauhon, L. J.; Gudiksen, M. S.; Wang, J. Appl. Phys. Lett. 2001, 78, 2214. (12) Gudiksen, M. S.; Lieber, C. M. J. Am. Chem. Soc. 2000, 122, 8801. (13) Wang, F.; Malac, M.; Egerton, R. F.; Meldrum, A.; Zhu, X.; Liu, Z.; Macdonald, N.; Li, P.; Freeman, M. R. J. Appl. Phys. (accepted). (14) Egerton, R. F. Electron Energy-Loss Spectroscopy in the Electron Microscope, 2nd ed.; Plenum Press: New York, 1996. (15) The calculations are based on the formula: ([O]/[Si]) ) (IO(β,∆)/ ISi(β,∆)) (σSi(β,∆)/σO(β,∆)), in which IO(β,∆) and ISi(β,∆) are experimentally obtained integral intensities of O K-edge and Si L-edge respectively, using window with ∆ ) 100 eV, and semi-collection angle β ) 3.0 mrad; σO(β,∆) and σSi(β,∆) are cross-sections of O K-edge and Si L-edge respectively, calculated using SIGMAK3 and SIGMAL3 programs. (16) Botton, G. A.; Gupta, J. A.; Landheer, D.; McCaffrey, J. P.; Sproule, G. I.; Graham, M. J. J. Appl. Phys. 2002, 91, 2921. (17) Batson, P. E. Microsc. Microanal. Microstruct. 1991, 2, 395. (18) The eutectic point for FeSi2 is 1207 °C, according to Binary alloy phase diagrams, 2nd ed.; Massalski, B. M. T., Okamoto, H., Subramanian, P. R., Kacprzak, L., Eds.; ASM International: Materials Park: Ohio, 1990; the eutectic point for FeO-SiO2-Fe is 1188 °C, according to FactSage thermodynamic database computing system, http://www.factsage.com. (19) Allen, G. L.; Bayles, R. A.; Gile, W. W.; Jesser, W. A. Thin Solid Films 1986, 144, 297. (20) Nanda, K. K.; Sahu, S. N.; Behera, S. N. Phys. ReV. A 2002, 66, 13208. (21) Yan, H. F.; Xing, Y. J.; Hang, Q. L.; Yu, D. P.; Wang, Y. P.; Xu, J.; Xi, Z. H.; Feng, S. Q. Chem. Phys. Lett. 2000, 323, 224. (22) Lee, K. H.; Yang, H. S.; Baik, K. H.; Bang, J.; Vanfleet, R. R.; Sigmund, W. Chem. Phys. Lett. 2004, 383, 380. (23) Wang, F.; Malac, M.; Egerton, R. F.; Schofield, M.; Zhu, Y. unpublished. (24) Namatsu, H.; Yamaguchi, I.; Nagase, M.; Yamazaki, K.; Kurihara, K. Microelectron. Eng. 1998, 41/42, 331. (25) Wang, N.; Tang, Y. H.; Zhang, Y. F.; Lee, C. S.; Lee, S. T. Phys. ReV. B 1998, 58, R16024.
