Au Stabilization and Coverage of Sawtooth Facets ... - ACS Publications

Aug 19, 2008 - ... Wang , Christian Wiethoff , Tobias Nabbefeld , Frank Meyer zu Heringdorf , and Michael Horn-von Hoegen. ACS Nano 2011 5 (2), 1313-1...
0 downloads 0 Views 565KB Size
NANO LETTERS

Au Stabilization and Coverage of Sawtooth Facets on Si Nanowires Grown by Vapor-Liquid-Solid Epitaxy

2008 Vol. 8, No. 9 3065-3068

Christian Wiethoff,*,† Frances M. Ross,‡ Matthew Copel,‡ Michael Horn-von Hoegen,† and Frank-J. Meyer zu Heringdorf† Department of Physics and Center for Nanointegration, Duisburg-Essen (CeNIDE), UniVersity of Duisburg-Essen, 47057 Duisburg, Germany, and IBM T. J. Watson Research Center, Yorktown Heights, New York 10598, USA Received April 23, 2008

ABSTRACT Si nanowires grown in UHV by Au-catalyzed vapor-liquid-solid epitaxy are known to exhibit sidewalls with {112}-type orientation that show faceting. To understand the origin of the faceting, Au induced faceting on Si(112) surfaces was studied in situ by spot-profile-analyzing low-energy electron diffraction. With increasing Au coverage at 750 °C, the Si(112) surface undergoes various morphological transformations until, at a critical Au coverage of about 3.1 × 1014 atoms/cm2, a phase consisting of large (111) and (113) facets forms, similar in structure to the nanowire sidewalls. This phase is stable at larger Au coverages in equilibrium with Au droplets. We suggest that Si nanowire surfaces exhibit this structure, and we derive the Au coverage on the two types of facets.

Pressured by the need to scale down microelectronic devices, a great deal of effort has been undertaken to use a vapor-liquid-solid (VLS) process to grow Si nanowires several micrometers in length and only a few tens of nm in diameter. Au droplets are often used as the catalyst to grow nanowires.1,2 For diameters above about 10 nm, these nanowires typically grow in the [111] direction with hexagonal cross sections and {112}-type sidewalls.3,4 Au offers several advantages as a catalyst: growth occurs over a wide range of conditions, and the Au can be placed lithographically to give full control over the wire locations. However, the enormous impact that even the slightest concentrations of Au dopant atoms will have on the electronic properties of the nanowire makes it important to understand the extent to which Au diffuses throughout the wire and the substrate and remains in the wire and on its surface.5 Recently, in situ studies of Si nanowires grown by chemical vapor deposition under ultra high vacuum conditions6 demonstrated that Au from the catalyst is quite mobile at temperatures above about 600 °C. In fact, the Au readily diffuses up and down wire sidewalls and across the substrate between wires, leading to Ostwald ripening of the catalyst droplets and consequent tapering of the wires. Because the dominant transport mechanism is surface diffusion,8 this suggests that there may be a significant amount of Au on * Corresponding author: [email protected]. † University of Duisburg-Essen. ‡ IBM T.J. Watson Research Center. 10.1021/nl801146q CCC: $40.75 Published on Web 08/19/2008

 2008 American Chemical Society

the wire and substrate surfaces.6 A measurement of the Au coverage would be useful in quantifying wire properties and understanding growth morphology. In particular, the {112} sidewalls of UHV-grown Si nanowires are known to exhibit “sawtooth” faceting.9 Because Au-induced faceting is known to occur in a wide variety of substrate orientations, depending on the temperature and the specific surface,10,11 it is of interest to investigate whether the sawtooth faceting is related to Au coverage. Nanowire sidewalls are narrow and difficult to analyze directly. Therefore, in this letter, we investigate Au-induced faceting of larger areas of Si(112) wafers. This allows us to prepare sample surfaces with accurate Au coverages to examine the faceting process in detail. We show that Au can indeed cause faceting that appears like the sawtooth faceting; we identify the indices of the sawtooth facets and estimate the Au coverage on the different facet planes. The experiments are performed in a standard ultra high vacuum system, equipped with spot-profile-analyzing lowenergy electron diffraction (SPA-LEED).12-14 Clean Si(112) surfaces were prepared by degassing samples cut from Si(112) wafers at 600 °C, followed by a short flash to 1250 °C to desorb the native oxide. The samples were heated by direct current parallel to the step edges to avoid currentinduced step-bunching.15,16 Different coverages of Au were deposited from an electron beam Au evaporator at 750 °C substrate temperature. The evaporation rate was calibrated ex situ by performing

Figure 2. Angles between the facets of each surface phase. The dotted line is a guide to the eye. For comparison, the range of angles measured from TEM is denoted as a gray box. The variation of (2.5° is due to the existence of smaller facets on each larger facet, some of which can be seen close to the resolution limit of the TEM images.

Figure 1. SPA-LEED spot profiles in [111j] direction taken during Au deposition on a Si(112) surface at 750 °C, plotted vs Au coverage. We used an electron energy of 147 eV, which translates into a vertical momentum transfer of 6.6 Å-1. The observed surface phases are separated by white dotted lines and named from A-H. Each change of diffraction spots indicates a rearrangement of the surface step structure. A diffraction pattern of the initial Si(112) surface is given in the inset, where the spot profile direction is denoted with black arrows. Note that the inset is plotted in linear grayscale, for better visibility, while all other SPA-LEED data is given in logarithmic grayscale.

medium energy ion scattering17 on samples with four different Au coverages. SPA-LEED was used to characterize the surface morphology. Two types of information were collected. To follow the structural and morphological changes during Au deposition, one-dimensional (1D) SPA-LEED data was recorded during deposition using a RHEED-like geometry.13 We used an electron energy of 147 eV, which translates into a vertical momentum transfer of 6.6 Å-1 or an electron scattering phase S ) 1.1513 for the {112} step height of 1.11 Å. After deposition, SPA-LEED was used to obtain reciprocal space maps (RSMs)18 to determine the facet orientations at different Au coverages. RSMs are a representation of the surface lattice rods in reciprocal space and therefore provide information about the surface structure and morphology. They are, in particular, well suited to determine facet orientations of a surface with high accuracy. To improve the signal-to-noise ratio, all RSMs were taken at room temperature (RT). To ensure that the surface topography was unaltered during the temperature change, we verified that the diffraction spot positions remained constant while cooling the samples to RT. We compared the facet angles determined from the SPA-LEED analysis with those previously observed during nanowire growth using in situ transmission electron microscopy.9 Figure 1 shows the in situ results obtained during Au deposition. The surface step structure is reflected by the position and arrangement of the different diffraction spots. 3066

The data is taken in the [111j] direction, indicated by black arrows in the inset of Figure 1 in which a diffraction pattern of the initial (112) surface is shown. During deposition, the spots move, appear or disappear, and each change corresponds to a rearrangement of the steps on the surface. Altogether eight surface phases were observed, denoted A to H. Each of these corresponds to a distinct surface morphology, with differently stepped surfaces and facets. A thorough analysis of the detailed step structure in each of the eight phases, derived from RSMs, is beyond the scope of this letter and will be published separately.19 To compare the observed phases with TEM measurements of nanowire sidewalls, we plot in Figure 2 the angle between the observed facets of all phases as well as the angle determined on nanowire sidewalls with in situ TEM. We show the difference angle, rather than the two angles between the facets and the (112) surface, because this is the easiest parameter to measure accurately in TEM. The angle between the facets increases with Au coverage up to phase G, and this phase is the only one that agrees with the TEM measurement; phase H again has a smaller angle between facets. Because of the good agreement between the facet angles observed on the nanowires and in phase G, we now focus on phase G. Figure 3 shows the RSM of this phase. Only the (111) and (113) reciprocal lattice rods are observed. The (113) and (111) facets are tilted by R ) -10.0° and β ) +19.5°, respectively, from the (112) plane. For comparison, a TEM image of a nanowire sidewall is also given, and the similarity of the angles is clearly visible. It is also clear from the TEM image that the facets exhibit different widths. From geometric considerations, the ratio of the (113) and (111) facet widths must be 2:1 in order to preserve the macroscopic [112] orientation of the sample. This is consistent with the TEM image, where the total width of one facet type is approximately half the width of the other. Furthermore, it is interesting to note that phase G has the largest facets of any of the phases seen here. Because the width of the reciprocal lattice rods is limited by the transfer Nano Lett., Vol. 8, No. 9, 2008

Figure 3. Reciprocal space map of phase G showing that only 111 and 113 facet rods are present, with Bragg conditions indicated. A TEM image of the surface facets on a nanowire is also shown, illustrating clearly that the reciprocal lattice rods are perpendicular to the facets. The scale bar is 50 nm in length.

width of our SPA-LEED, it is impossible to determine the size of facets that are larger than 40 nm. We therefore performed ex situ AFM measurements of phases A-H. Of all the phases, phase G has the highest root-mean-square roughness, around 4 nm, and the widest facets, up to 30 nm for (111) and 60 nm for (113). This is a striking result given that nanowire sidewalls in TEM also exhibit rather large facets.9 We now consider the Au coverage of phase G. The (111) facets in phase G are (3 × 3)R30° reconstructed, as we conclude from the corresponding spots in the LEED pattern. Different structure models have been proposed for the (3 × 3)R30° reconstruction,20-22 with a Au coverage for a saturated (3 × 3)R30° surface between 2/3 and 1 ML (here 1 ML corresponds to 7.8 × 1014 atoms/cm2). In Figure 1, we showed that phase G is stable for an Au coverage in the range 3.1-3.6 × 1014 atoms/cm2. From the size ratio of the facets and the coverage of the (3 × 3)R30° reconstructed (111) areas, we can estimate a range for the Au coverage of the (113) facets. If Ahkl denotes the fractional area of each facet, and θhkl denotes its Au coverage, then A111 × θ111 + A113 × θ113 ) θtotal. This provides a range of values of θ113 from 0.5 to 2.6 × 1014 atoms/cm2. So the (113) facets, although they account for 2/3 of the total area, contain less Au than the (111) facets by at least a factor of 2. Such a decay of a faceted surface into Au enriched and Au depleted areas has already been observed on vicinal Si (001) surfaces23 using X-ray photoemission electron microscopy. Nano Lett., Vol. 8, No. 9, 2008

It is believed that the separation into Au enriched and depleted areas provides the driving force for the formation of such facets. Assuming that nanowire sidewalls have the same structure as the phase G that forms on the flat (112) surface, we would then conclude that the sidewall mean Au coverage is 3.1-3.6 × 1014 atoms/cm2, or 0.40-0.46 ML, with the majority of Au located on the smaller (111) facets. This coverage provides a fundamental limitation to the length of nanowires that can be grown (in UHV) from a droplet of a given size before the droplet is consumed by transfer to the sidewalls. In the calculations of ref 6, a surface coverage of 1.0-1.5 ML was estimated from the shortest nanowires visible without droplets. Because coarsening also plays a role in reducing the droplet size, that estimate will be an upper limit and is therefore consistent with the 0.40-0.46 ML found here for phase G. We finally consider the relative stability of phase G during nanowire formation. On the flat (112) surface, we have shown that the faceting transition does not end in phase G but evolves into phase H if further Au is deposited. Phase H is not observed on nanowires even though well over 0.48 ML of Au is present: typically, 2-3 ML Au is deposited initially and then agglomerated into droplets to catalyze wire growth. However, if we anneal phase H, say for 120 min at 825 °C, we observe (by SPA-LEED and ex situ scanning electron microscopy) a transformation to phase G plus Au droplets of size approximately 400 nm on the surface. Apparently phase H is metastable on the (112) surface and there exists a kinetic barrier to maintaining phase G and directly nucleating excess Au into droplets; it is known that smaller sized droplets are less stable than larger ones7 and, thus, the kinetic limitation to form droplets at all drives the system into phase H. During nanowire growth, this barrier may not be present because larger droplets are available from the start. The coexistence of phase G with large droplets appears to be the energetically most favorable configuration for high Au coverages. In conclusion, we have studied Au-induced faceting of Si(112) by SPA-LEED both in situ during Au deposition, after cooling, and after annealing. We find that Au can stabilize a complex series of eight morphological transitions between stepped and faceted surfaces depending on the coverage. By comparing our results to the structure seen on the sidewalls of UHV-grown nanowires, we identify one particular phase G, with an overall Au coverage between 0.40-0.46 ML Au, as the relevant phase for sawtooth faceting. The driving force for the faceting is the formation of Au rich (111) facets and Au depleted (113) facets. A nonzero Au surface coverage, i.e., the presence of Au at places other than the end of the wire, may be significant for the processing and operation of devices made from nanowires. Phase G consists of (113) and (111) facets, with the Au coverage much greater on the (111) compared to the (113) facets. On UHV-grown nanowires, these facets can have widths up to 20 nm. Thus there is presumably a quasiperiodic modulation of the coverage along the nanowire sidewalls, which may go hand in hand with a modulation of 3067

the electronic surface properties and could thus be of potential importance for future nanowire applications. References (1) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89. (2) Law, M.; Goldberger, J.; Yang, P. Annu. ReV. Mater. Res. 2004, 34, 83. (3) Westwater, J.; Gosain, D. P.; Tomiya, S.; Usui, S. J. Vac. Sci. Technol., B 1997, 15, 554. (4) Zhang, X. Y.; Zhang, L. D.; Meng, G. W.; Li, G. H.; Jin-Phillips, N. Y.; Phillips, F. AdV. Mater. 2001, 13, 1238. (5) Allen, J. E.; Hemesath, E. R.; Perea, D. E.; Lensch-Falk, J. L.; Li, Z. Y.; Yin, F.; Gass, M. H.; Wang, P.; Bleloch, A. L.; Palmer, R. E.; Lauhon, L. J. Nat. Nanotechnol. 2008, 3, 168. (6) Hannon, J. B.; Kodambaka, S.; Ross, F. M.; Tromp, R. M. Nature 2006, 440, 69. (7) Wagner, C. Z. Elektrochem. 1961, 65, 581. (8) Kodambaka, S.; Hannon, J. B.; Tromp, R. M.; Ross, F. M. Nano Lett. 2006, 6, 1292. (9) Ross, F. M.; Tersoff, J.; Reuter, M. C. Phys. ReV. Lett. 2005, 95, 146104.

3068

(10) Minoda, H.; Shimakura, T.; Yagi, K.; Meyer zu Heringdorf, F. J.; Horn-von HoegenM. Surf. Sci. 1999, 432, 69. (11) Dickinson, J. W.; Moore, J. C.; Baski, A. A. Surf. Sci. 2004, 561, 193. (12) Scheithauer, U.; Meyer, G.; Henzler, M. Surf. Sci. 1986, 178, 441. (13) Horn-von Hoegen, M. Z. Kristallogr. 1999, 214, 591. (14) Horn-von Hoegen, M. Z. Kristallogr. 1999, 214, 684. (15) Latyshev, A. V.; Aseev, A. L.; Krasilnikov, A. B.; Stenin, S. I. Surf. Sci. 1989, 213, 157. (16) Homma, Y.; McClelland, R. J.; Hibino, H. Jpn. J. Appl. Phys. 1990, 29, L2254. (17) van der Veen, J. F. Surf. Sci. Rep. 1985, 5, 199. (18) Meyer zu Heringdorf, F.-J.; Horn-von Hoegen, M. ReV. Sci. Instrum. 2005, 76, 085102. (19) Wiethoff, C.; Meyer zu Heringdorf, F.-J.; Horn-von Hoegen M. Manuscript in preparation. (20) Huang, J. H.; Williams, R. S. Phys. ReV. B 1988, 38, 4022. (21) Grozea, D.; Bengu, E.; Marks, L. Surf. Sci. 2000, 461, 23. (22) Chester, M.; Gustafsson, T. Phys. ReV. B 1990, 42, 9233. (23) Meyer zu Heringdorf, F.-J.; Schmidt, T.; Heun, S.; Hild, R.; Zahl, P.; Ressel, B.; Bauer, E.; Horn-von Hoegen, M. Phys. ReV. Lett. 2001, 86, 5088.

NL801146Q

Nano Lett., Vol. 8, No. 9, 2008