StranskiKrastanow Growth of Germanium on Silicon Nanowires

Floyd et al.33 measured the compositions of. Ge/Si(100) islands .... (7) Zhang, Y.; Floyd, M.; Driver, K. P.; Drucker, J.; Crozier, P. A.; Smith,. D. ...
2 downloads 0 Views 812KB Size
NANO LETTERS

Stranski−Krastanow Growth of Germanium on Silicon Nanowires

2005 Vol. 5, No. 6 1081-1085

Ling Pan,† Kok-Keong Lew,†,‡ Joan M. Redwing,†,‡ and Elizabeth C. Dickey*,†,‡ Department of Materials Science and Engineering and Materials Research Institute, The PennsylVania State UniVersity, UniVersity Park, PennsylVania 16802 Received March 30, 2005; Revised Manuscript Received April 28, 2005

ABSTRACT There have been extensive studies of germanium (Ge) grown on planar silicon (Si) substrates by the Stranski−Krastanow (S−K) mechanism. In this study, we present S−K growth of Ge on Si nanowires. The Si nanowires were grown at 500 °C by a vapor−liquid−solid (VLS) method, using silane (SiH4) as the gaseous precursor. By switching the gas source from SiH4 to germane (GeH4) during the growth and maintaining the growth conditions, epitaxial Ge islands deposited on the outer surface of the initially formed Si nanowires. Transmission electron microscopy (TEM), scanning TEM, and energy-dispersive X-ray spectroscopy techniques were utilized to identify the thin wetting layer and the threedimensional Ge islands formed around the Si core nanowires. Cross-sectional TEM verified the surface faceting of the Si core nanowires as well as the Ge islands.

It is well-known that germanium (Ge) thin films grown on planar silicon (Si) substrates follow a Stranski-Krastanow (S-K) mechanism due to the 4.2% lattice mismatch between Ge and Si.1,2 In the S-K growth mode, deposition first proceeds by a two-dimensional (2D) layer-by-layer mechanism to form a wetting thin film, and as the strain energy due to the lattice mismatch increases as the thin film gets thick, three-dimensional (3D) islands form.3 There have been extensive studies on the size, morphology, structure, and chemistry of Ge or SiGe islands grown on Si(001) substrates because of the great potential of Ge/Si self-assembled quantum dots for possible device applications.4-9 It was found that the Ge or SiGe islands were dislocation-free before reaching a critical thickness.1,10 The island morphology was dependent upon growth method, growth conditions, and the SiGe composition. In general, both smaller pyramids with {105} facets and larger domes with {102} and {113} facets have been observed for Ge islands grown on Si(001) by chemical vapor deposition (CVD),4 physical vapor deposition (PVD),5 and molecular beam epitaxy (MBE)6,7 methods. Semiconductor nanowires have attracted enormous attention in recent years due to their great potential as building blocks for functional nanoscale electronic and optical devices. The fabrication of heterostructured semiconductor nanowires is of great interest both for fundamental studies of carrier confinement effects in these materials as well as nanoscale device development.11-13 The vapor-liquid-solid (VLS) mechanism, which was proposed in the 1960s for large whisker growth,14,15 has become a common approach for * Corresponding author. E-mail: [email protected]. † Materials Research Institute. ‡ Department of Materials Science and Engineering. 10.1021/nl050605z CCC: $30.25 Published on Web 05/24/2005

© 2005 American Chemical Society

fabricating Si and Ge nanowires.16-21 The essentially identical eutectic temperatures of Au-Si and Au-Ge 22 suggest that similar growth conditions could be used to form SiGe alloy nanowires and Si-Ge heterostructures. Our previous research successfully synthesized SiGe alloy nanowires with a high degree of homogeneity by the VLS mechanism using silane (SiH4) and germane (GeH4) as gaseous sources.23 However, the growth regimes for Si and Ge are very different because of the different thermal stabilities of the gaseous precursors, where the decomposition rate of GeH4 at 800 K is much greater than that of SiH4.23 Givargizov et al.24 demonstrated the VLS growth of micrometer-sized Ge whiskers on top of Si whiskers using silicon tetrachloride (SiCl4) and germanium tetrachloride (GeCl4) as gaseous sources over the temperature range of 750-850 °C. More recently, Y. Wu et al.25 synthesized axial Si/SiGe heterostructured nanowires, where Si and SiGe portions grew alternatively along the nanowire axis, with SiCl4 as the Si precursor and a Ge wafer as the pulsed laser ablation target to generate Ge vapor at high temperature. Lauhon et al.26 produced Si-Ge core-shell and coremultishell heterostructures using SiH4 and GeH4 source gases by VLS growth to form the Si or Ge core and CVD afterward to coat the nanowires with Ge or Si thin film, respectively. Theoretically, Gutkin et al.27 and Liang et al.28 have carried out research based on stress analyses, modeling the criterion for the generation of misfit dislocations to be energetically favorable in epitaxial film/substrate composites of wire form. However, they have considered only the generation of misfit dislocations as the means of strain relaxation. The possibility of island formation to relieve the misfit strain between the core and the shell has not yet been studied

theoretically or experimentally. In this study, we demonstrate the growth of 3D Ge islands around Si nanowires by the S-K mechanism. The research could open up a new opportunity for integrating quantum dots with semiconductor nanowires, and may also provide information on the ability to form strained Si and Si1-xGex core-shell nanowires for nanoscale electronics.29,30 The nanowires were grown out of two types of membranes, one is commercially available nanoporous alumina membranes (Whatman Scientific) with a nominal pore diameter of 200 nm, and the other is anodized alumina membranes prepared by anodization of aluminum (Al) foil in oxalic acid (C2H2O4) with a nominal pore diameter of 80 nm.31 The membranes served to confine the catalysts, and thus nanowire diameters. The detailed synthesis method was presented in a previous paper.18 Electrodeposited Au segments in the pores near the top of the membrane served as the catalyst. Gas sources used for nanowire growth were SiH4 (10% in H2) and GeH4 (1% in H2). The total gas flow rate and the total pressure were held constant at 100 sccm and 12 Torr, respectively. SiH4 was passed through the reaction chamber at 50 sccm for 15 min first so that the nanowires grew out of the top surface of the membrane. Taking advantage of the known growth regimes, the temperature was maintained at 500 °C and the inlet gas was switched to GeH4 for one minute at the same flow rate. Under these conditions, Ge nanowires do not grow by the VLS mechanism as evidenced by our and others studies of Ge nanowire growth.21 This results from the relatively low thermal stability of GeH4 so that at 500 °C gas-phase decomposition dominates relative to the catalyzed decomposition at the Au tip. We therefore anticipated that the Ge would deposit around the initially grown Si nanowire by CVD deposition. In some growth experiments the gas was switched back to the SiH4 for another minute after these two stages to determine if axial growth of Si could be reinitiated. Nanowires grown out of the top of the membrane were released from the surface by sonication and suspended in isopropyl alcohol (IPA). The nanowire suspension was then dropped onto a lacey carbon coated copper grid for transmission electron microscopy (TEM) analyses. Structural and chemical characterizations of the nanowires were carried out on a JEOL 2010F field emission TEM/scanning TEM (STEM) operated at 200 kV. As a result of the much higher deposition rate of GeH4 at 500 °C than SiH4, it is expected that Ge should deposit preferentially on the surface of the Si nanowires. This is exactly what we find: Ge deposits radially and epitaxially around the initially grown Si nanowires. What is interesting is that instead of coating the Si nanowires as uniform shells they form 3D island structures (denoted as Ge/Si) around the Si nanowire substrate. When the inlet gas is returned back to SiH4, the Si regains axial growth. Figure 1a shows an example of such Si-Ge/Si nanowires with a diameter about 200 nm. The dark round feature at the tip is the Au catalyst. The smooth, approximately 1 µm long segment next to the Au tip corresponds to the third growth segment, where pure SiH4 was passed in the chamber. The length of this final segment matches with the determined growth rate of 1082

Figure 1. (a) Bright-field TEM image of a Si-Ge/Si nanowire with the inset of the corresponding SADP. (b) ADF STEM image of a Ge/Si nanowire and the inset is the EDS line scan profile along the line across the nanowire. The arrow in the inset marks the Si signal in a protruding island, indicating Si diffusion into the Ge island.

Si nanowires under the same growth conditions. Energydispersive X-ray spectroscopy (EDS) confirms that the final segment is pure Si. The islanded segment next to the Si segment results from the first and second growth stages. The inset selected-area diffraction pattern (SADP) shows that the Ge is epitaxial with respect to the Si core. An annular darkfield (ADF) STEM image of a Ge/Si nanowire is shown in Figure 1b. The inset EDS line profile confirms that Ge deposits on the outer surface of the initially formed Si nanowire. It is also noticed that the islands are not grown randomly around the Si core, but they appear to be aligned along particular surfaces of the Si nanowire. Furthermore, the edges of the aligned islands are faceted along particular surfaces (to be discussed below). Three types of nanowire Nano Lett., Vol. 5, No. 6, 2005

Figure 2. (a) Bright-field TEM and (b) ADF STEM images of the Ge islands deposited on Si nanowires with Moire´ fringes clearly seen. The inset in (b) shows a thin Ge-rich layer at the surface of Si nanowire. (c) The corresponding SADP with extra spots consistent with the Moire´ fringes in (a) and (b). (d) Higher magnification TEM image of an island on the same nanowire with the arrows pointing to the misfit dislocations. (e) HRTEM image of an area in (d) with stacking faults and twin boundary (marked by arrows) clearly seen.

growth directions are observed: 〈110〉 , 〈111〉, or 〈112〉, the same as the SiGe alloy nanowires grown in a similar setup.23 Figure 2 shows an example of the Ge/Si island structure of a ∼100 nm diameter nanowire with growth direction of [111], where Figure 2a is a low magnification bright-field TEM image, Figure 2b is an ADF STEM image, and Figure 2c is the corresponding SADP. Figure 2d is a higher magnification TEM image of the same nanowire and Figure 2e is the high-resolution image of an area in Figure 2d. The same 3D island structure is present. The ADF STEM image shows a bright line at the right edge of the nanowire, which can be clearly seen from the enlarged view at the inset of Figure 2b. Since there are no Ge islands protruding from the right side, this bright line along the edge indicates the presence of a Ge-rich thin layer surrounding the nanowire. Most of the Si nanowires exhibit defect-free microstructures, while a small fraction of nanowires show growth defects such as the (111h) stacking fault as indicated by the small arrow in Figure 2a. 2D Moire´ fringes are clearly seen on the islands in Figure 2a and 2b, indicating the slight difference in the lattice constants between the Si core and the Ge islands. Correspondingly, double spots are present at primary diffraction spots in Figure 2c. In Figure 2d several (111) and Nano Lett., Vol. 5, No. 6, 2005

(111h) stacking faults and {111} type twins are present in the island, as viewed in Figure 2e, extending from the core/ island interface to the edge of the island. These stacking faults are likely bounded by Shockley partial dislocations at the core/island interface that act as misfit dislocations, as pointed by the arrows in Figure 2d. Similar misfit dislocation related stacking faults have also been observed in Ge islands grown on planar Si(001).4 The spacings between the misfit dislocations at the interface are not periodic within the length of the island, likely related to the various facet surfaces of the island and thus the contact angle between the island side facet and the substrate. Although the planes marked on Figure 2d are 2D projections of the facets of the 3D island, different contact angles between the side facets and the nanowire core are obvious. For studies of Ge islands grown on Si(001), LeGoues et al.32 have suggested that increasing the contact angle reduces the barrier for dislocation nucleation at the island edge; Liao et al.8 have proposed that the shallow facet is a stable facet for low strain status where strain has been relaxed through the formation of dislocations. EDS analyses on the islands revealed X-ray signals of Si besides those of Ge in the islands (see the arrow in the inset 1083

of Figure 1b). Floyd et al.33 measured the compositions of Ge/Si(100) islands grown by MBE over temperature range from 400 to 700 °C, and similar Si diffusion into the Ge islands was also seen. Although they did not perform growth at 500 °C, by examining their results of the Ge concentration in the islands as a function of growth temperature, a Ge concentration of approximately 70-82% would be expected for growth at 500 °C. The spacings of Moire´ fringes were measured from the Fourier transform of the real-space image. The values are 15-35% larger than the calculated spacings assuming the core and the islands exhibiting the bulk Si and Ge lattices, respectively, indicating the smaller misfit between the core and the islands than the 4.2% misfit between Si and Ge lattices. Assuming the Si core has the lattice parameter of bulk Si, and the islands are Si1-xGex whose lattice parameter follows Vegard’s law, the Ge concentration x can then be calculated from the measured Moire´ fringe spacings (dM), which is dM )

dSi[(1 - x)dSi + xdGe] [(1 - x)dSi + xdGe] - dSi

where dSi and dGe are the corresponding spacings of the same set of planes in Si and Ge lattices, respectively. The resulting Ge concentration in the islands was calculated to be 7585%. This calculation takes chemical composition as the only factor that determines the lattice spacings and ignores the strain effect. Considering the existence of residual misfit strain in the islands, the actual Ge concentration in the islands should be higher. To view the island structure more clearly, cross-sectional TEM samples were prepared by microtoming the nanowires that were embedded in epoxy. Figure 3 shows a bright-field TEM and ADF STEM images of the cross-section of a Ge/ Si nanowire with beam direction of [112h], i.e., the nanowire axis or the growth direction. Similar to what is shown in Figure 2b, a thin Ge-rich layer, 1-2 nanometers thick, is seen clearly surrounding the core, which is consistent with S-K type growth. Combined with the 3D islands observed around the nanowire surface, this is the first experimental evidence of Ge growing radially on Si nanowires via the S-K mechanism. Due to the cylindrical substrate surface, it was difficult to obtain the overall shape and the size distribution of the islands, and thus their relationship with the diameter of the substrate nanowires. However, faceting is clearly seen not only on the outer surface of the Si core but also on the surfaces of the Ge islands. It is known that Si nanowires tend to facet on low-index crystallographic planes in order to reduce the surface energy.34,35 As an example, the Si nanowire shown in Figure 3 with [112h] axis has facets at (111), (11h0), and {113} planes, forming an octagonal outer surface. Similar to what was observed for Ge islands grown on Si(001), the islands in Figure 3 exhibit {113} side facets, which are inclined by 58.5° to the (111) Si facet. Furthermore, the Ge islands appear to grow preferentially on certain Si planes, for example, (111) is more favored compared with 1084

Figure 3. (a) Bright-field TEM and (b) ADF STEM images of the cross-section of a Ge/Si nanowire, where faceting is clearly seen on the outer surfaces of the Si core and the Ge islands.

(11h0) and {113}, probably because (111) has the lowest surface energy. Since Ge deposits around the Si nanowire cores, uncatalyzed CVD instead of catalyzed VLS mechanism dominates during the second growth stage. Compared to the growth method of the Si-Ge core-shell nanowires achieved by Lauhon et al.,26 our growth was carried out at 500 °C all along, whereas they grew Si nanowire cores at 450 °C and Ge shells at 380 °C. After growing the Si core, they intentionally moved the growth substrate downstream in the furnace to favor uncatalyzed surface growth, whereas it is not necessary for our case because the growth temperature we applied, 500 °C, is much higher than the optimum growth temperature for Ge nanowires via VLS mechanism in our setting, 275-325 °C.23 As to why the S-K growth occurs in our experiments while Lauhon et al.26 obtained relatively uniform Ge shell, there could be multiple factors, an important one of which is the size of the nanowires. The diameters of the Si nanowire cores that they produced were 20-30 nm, much smaller than what we obtained in our growth experiments. According to research on thin film growth, as the substrate dimension is Nano Lett., Vol. 5, No. 6, 2005

Figure 4. ADF STEM image of a Si core-Ge shell nanowire.

reduced, approaching that of the thin film, the thickness of the strained epitaxial film will increase since the strain energy is distributed more evenly between the epilayer and the substrate.36 Similar predictions have also been made on systems where small-diameter nanowires serve as the substrate.28 More experiments were performed in order to clarify the relationship between the nanowire diameter and the epilayer thickness. Preliminary results have produced SiGe core-shell nanowires with the diameters of the Si core in the range of 10-50 nm, as shown in Figure 4. However, this has been proven to be a complex problem involving other factors such as growth direction, surface faceting, partial pressure of the inlet GeH4 gas, temperature, etc. Systematic experiments and theoretical modeling are in progress to try to understand the underlying factors that control the epitaxial growth on nanowires. In summary, this study demonstrated the S-K growth of 3D epitaxial Ge islands around Si nanowires. By varying the inlet gas species from SiH4 to GeH4 during the growth at 500 °C, uncatalyzed surface deposition of Ge was initiated around the outer surface of the Si core nanowires and a coherent wetting layer was formed followed by 3D islands. Cross-sectional TEM verified the surface faceting of the Si core nanowires as well as the Ge islands. Acknowledgment. This work was supported by the National Science Foundation under grant number DMR0103068 and The Pennsylvania State University Materials Research Science and Engineering Center (MRSEC) for Nanoscale Science (DMR-0213623). TEM work was performed in the electron microscopy facility of the Materials Characterization Laboratory at The Pennsylvania State University. References (1) Eaglesham, D. J.; Cerullo, M. Phys. ReV. Lett. 1990, 64, 1943. (2) Mo, Y.-W.; Savage, D. E.; Swartzentruber, B. S.; Lagally, M. G. Phys. ReV. Lett. 1990, 65, 1020. (3) Jain, S. J.; Willander, M. Silicon-Germanium Strained Layers and Heterostructures; Academic Press: San Diego, 2003. (4) Kamins, T. I.; Carr, E. C.; Williams, R. S.; Rosner, S. J. J. Appl. Phys. 1997, 81, 211.

Nano Lett., Vol. 5, No. 6, 2005

(5) Medeiros-Ribeiro, G.; Bratkovski, A. M.; Kamins, T. I.; Ohlberg, D. A. A.; Williams, R. S. Science 1998, 279, 353. (6) Chaparro, S. A.; Zhang, Y.; Drucker, J.; Chandrasekhar, D.; Smith, D. J. J. Appl. Phys. 2000, 87, 2245. (7) Zhang, Y.; Floyd, M.; Driver, K. P.; Drucker, J.; Crozier, P. A.; Smith, D. J. Appl. Phys. Lett. 2002, 80, 3623. (8) Liao, X. Z.; Zou, J.; Cockayne, D. J. H.; Jiang, Z. M.; Wang, X.; Leon, R. Appl. Phys. Lett. 2000, 77, 1304. (9) Hanke, M.; Schmidbauer, M.; Kohler, R.; Syrowatka, F.; Gerlitzke, A.-K.; Boeck, T. Appl. Phys. Lett. 2004, 84, 5228. (10) Daruka, I.; Baraba´si, A.-L. Phys. ReV. Lett. 1997, 79, 3708. (11) Cui, Y.; Lieber, C. M. Science 2001, 291, 851. (12) Gudiksen, M. S.; Lauhon, L. J.; Wang, J.; Smith, D. C.; Lieber, C. M. Nature 2002, 415, 617. (13) Wang, D.; Lieber, C. M. Nature Mater. 2003, 2, 355. (14) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89. (15) Wagner, R. S.; Doherty, C. J. J. Electrochem. Soc. 1966, 113, 1300. (16) Westwater, J.; Gosain, D. P.; Tomiya, S.; Usui, S.; Ruda, H. J. Vac. Sci. Technol. B 1997, 15, 554. (17) Morales, A. M.; Lieber, C. M. Science 1998, 279, 208. (18) Lew, K.-K.; Reuther, C.; Carim, A. H.; Redwing, J. M.; Martin, B. R. J. Vac. Sci. Technol. B 2002, 20, 389. (19) Wu, Y.; Yang, P. J. Am. Chem. Soc. 2001, 123, 3165. (20) Cui, Y.; Lauhon, L. J.; Gudiksen, M. S.; Wang, J.; Lieber, C. M. Appl. Phys. Lett. 2001, 78, 2214. (21) Kamins, T. I.; Li, X.; Williams, R. S.; Liu, X. Nano Lett. 2004, 4, 503. (22) Okamoto, H.; Massalski, T. B. ASM International-Binary Alloy Phase Diagrams; 2nd ed.; ASM International: Materials Park, OH, 1990; Vol. 1. (23) Lew, K.-K.; Pan, L.; Dickey, E. C.; Redwing, J. M. AdV. Mater. 2003, 15, 2073. (24) Givargizov, E. I.; Sheftal, N. N. J. Cryst. Growth 1971, 9, 326. (25) Wu, Y.; Fan, R.; Yang, P. Nano Lett. 2002, 2, 83. (26) Lauhon, L. J.; Gudiksen, M. S.; Wang, D.; Lieber, C. M. Nature 2002, 420, 57. (27) Gutkin, M. Y.; Ovid’ko, I. A.; Sheinerman, A. G. J. Phys.: Condens. Matter 2000, 12, 5391. (28) Liang, Y.; Nix, W. D.; Griffin, P. B.; Plummer, J. D. J. Appl. Phys. 2005, 97, 043519. (29) Thompson, S. E.; Armstrong, M.; Auth, C.; Cea, S.; Chau, R.; Glass, G.; Hoffman, T.; Klaus, J.; Ma, Z.; Mcintyre, B.; Murthy, A.; Obradovic, B.; Shifren, L.; Sivakumar, S.; Tyagi, S.; Ghani, T.; Mistry, K.; Bohr, M.; El-Mansy, Y. IEEE Electron DeV. Lett. 2004, 25, 191. (30) Kasper, E. In Heterostructures on silicon: one step further with silicon; Rosencher, Y. I. N. a. E., Ed.; Kluwer Academic: Dordrecht, The Netherlands, 1989; p 101. (31) Bogart, T. E.; Dey, S.; Lew, K.-K.; Mohney, S. E.; Redwing, J. M. AdV. Mater. 2005, 17, 114. (32) LeGoues, F. K.; Tersoff, J.; Reuter, M. C.; Hammar, M.; Tromp, R. Appl. Phys. Lett. 1995, 67, 2317. (33) Floyd, M.; Zhang, Y.; Driver, K. P.; Drucker, J.; Crozier, P. A.; Smith, D. J. Appl. Phys. Lett. 2003, 82, 1473. (34) Wu, Y.; Cui, Y.; Huynh, L.; Barrelet, C. J.; Bell, D. C.; Lieber, C. M. Nano Lett. 2004, 4, 433. (35) Li, C.-P.; Lee, C.-S.; Ma, X.-L.; Wang, N.; Zhang, R.-Q.; Lee, S.-T. AdV. Mater. 2003, 15, 607. (36) Lo, Y. H. Appl. Phys. Lett. 1991, 59, 2311.

NL050605Z

1085