pubs.acs.org/NanoLett
Hexagonal Close-Packed Structure of Au Nanocatalysts Solidified after Ge Nanowire Vapor-Liquid-Solid Growth Ann F. Marshall,*,† Irene A. Goldthorpe,†,‡,⊥ Hemant Adhikari,†,‡,¶ Makoto Koto,†,‡,# Young-Chung Wang,§ Lianfeng Fu,§ Eva Olsson,§ and Paul C. McIntyre†,‡ †
Geballe Laboratory for Advanced Materials and ‡ Materials Science and Engineering Department, Stanford University, Stanford, California 94305, § FEI Company, Hillsboro, Oregon 97124, and | Department of Applied Physics, Chalmers University, Goteburg, Sweden ABSTRACT We report that approximately 10% of the Au catalysts that crystallize at the tips of Ge nanowires following growth have the close-packed hexagonal crystal structure rather than the equilibrium face-centered-cubic structure. Transmission electron microscopy results using aberration-corrected imaging, and diffraction and compositional analyses, confirm the hexagonal phase in these 40-50 nm particles. Reports of hexagonal close packing in Au, even in nanoparticle form, are rare, and the observations suggest metastable pathways for the crystallization process. These results bring new considerations to the stabilization of the liquid eutectic alloy at low temperatures that allows for vapor-liquid-solid growth of high quality, epitaxial Ge nanowires below the eutectic temperature. KEYWORDS Hexagonal close-packed Au, Ge nanowires, Au catalyst, subeutectic vapor-liquid-solid growth, aberration-corrected transmission electron microscopy
C
atalysts play a central role in the widely used vapor-liquid-solid (VLS) process for growing semiconductor nanowires (NWs). In the VLS process, the catalyst forms a eutectic liquid nanodroplet with the NW material that enables one-dimensional growth. The growth of Ge NWs using Au as the catalyst is a system of particular interest in understanding VLS growth because the NWs can grow at substantial undercoolings relative to the eutectic temperature at which the catalyst is expected to solidify.1-6 Such low-temperature growth is important for incorporating NWs into devices and increases their compatibility with standard Si processing. In situ growth studies7 and postgrowth transmission electron microscopy (TEM) heating studies8 have demonstrated that the catalyst can indeed remain liquid well below the eutectic temperature. We have observed that the majority of the Au nanoparticles that solidify at the Ge NW tips during postgrowth cooling have the equilibrium face-centered cubic (fcc) close-packed structure of gold and are randomly oriented.9 Analysis and modeling of the stability of the undercooled eutectic liquid have been carried out by considering kinetic barriers to the nucleation of Au due to size effects and to supersaturation of the liquid with Ge from the supplier germane gas.7,8,10,11
In this paper, we report that some of the Au catalysts remaining at the tips of the Ge nanowires following growth have the hexagonal close-packed (hcp) structure. The observation of an hcp structure in Au is rare. Unlike some metals, allotropic forms have not been observed for bulk Au. Similar fcc metals, such as Ni and Ag, have been observed to form hcp structures when synthesized as nanoparticles or nanowires12-23 and molecular dynamics simulations predict the formation of hcp Au in 2-3 nm clusters.24 However the occurrence of Au in hcp form has only been identified under other unique circumstances, notably in very thin regions at special types of boundaries,25,26 and under extremely high pressure (240 GPa).27 We present transmission electron microscopy (TEM) results identifying the presence of the hcp phase in an estimated 10% of nanowires. Diffraction analysis and aberration (Cs)-corrected high-resolution imaging confirm the hcp structure, while compositional analysis shows that the Ge content is minimal. We suggest that this unusual observation of hcp stacking in gold is related to a metastable crystallization pathway specific to Ge nanowire growth. Our results indicate multiple pathways for crystallization of the nanocatalyst and bring new considerations to the stabilization of the subeutectic Au-Ge liquid that allows for low-temperature NW growth. Standard vapor-liquid-solid (VLS) growth of Ge NWs uses GeH4 as the germanium vapor source and Au as the catalyst, forming a Au-Ge eutectic liquid droplet that remains at the end of the growing NW as Ge precipitates out to form the solid NW. The catalyst remains liquid at temperatures as low as 280 °C, well below the eutectic temper-
* To whom correspondence should be addressed. E-mail:
[email protected]. ⊥ ¶
Present address: Eastman Kodak Co., Rochester, New York 14610. Present address: GLOBALFOUNDRIES, Albany, NY.
# Present address: Corporate R&D Headquarters, Canon Inc. Tokyo, Japan. Received for review: 03/15/2010 Published on Web: 08/05/2010
© 2010 American Chemical Society
3302
DOI: 10.1021/nl100913d | Nano Lett. 2010, 10, 3302–3306
FIGURE 1. Ge NW images: (a) a low-magnification image of a group of NWs, grown epitaxially (vertically) using 40 nm Au catalysts and sheared slightly sideways by the TEM sample preparation process. The solidified Au catalysts are the dark tips at the top of the NWs. (b) An image of a single gold catalyst at the end of a NW, showing the 4.7 Å spacing. The insets show a higher-magnification image of the lattice and the corresponding diffraction pattern, both confirming the 4.7 Å spacing. A number of stacking faults are also visible.
FIGURE 2. A series of diffraction patterns confirming the hcp structure. Beginning with (a) a [11-20] orientation, a Au nanoparticle is tilted 30° about the vertical axis to (b) a [10-10] zone axis. From this orientation, the nanoparticle is tilted in the perpendicular direction, toward the vertical axis, and (c) the [10-11] zone axis is found at about 43° of tilt. This three-dimensional tilting gives patterns with the lattice spacings, geometry, and tilt distances expected for an hcp structure of Au. (d) The same diffraction orientation as in (b) but viewed at lower magnification and with the central beam in the lower left-hand corner. This view shows both the zero-order layer of reciprocal lattice reflections around the central beam, and the outer circular set of reflections from the next, upper layer of reciprocal space. Although the 4.7 Å reflection is forbidden by the hcp space group in the zero-order layer (the periodicity indicated by the large arrows), it is allowed in the upper layer, which therefore has double the spacing (small arrows) of the zero-order layer.
ature of 361 °C. The NWs grown at these low temperatures are defect-free, single-crystal epitaxial NWs with uniform diameter, in contrast to the tapered NWs observed after high-temperature growth.6 During postgrowth cool-down, the alloy catalyst liquid crystallizes. Although metastable Au-Ge alloy phases have been demonstrated in splat cooling experiments carried out on Au-Ge liquids,28-31 the majority of the crystallized catalysts that we observe in Ge NW growth are randomly oriented fcc Au.9 The occurrence of the hcp structure was first indicated by our observation in some of the Au catalysts of a lattice spacing of approximately 4.7 Å with a corresponding reflection in the diffraction pattern. An example is shown in Figure 1, along with a low-magnification image of a typical NW sample. The observed lattice spacing is larger than any possible for the Au fcc structure, which has a 4.08 Å lattice parameter. The 4.7 Å lattice is approximately twice that of the close-packed {111} planes of fcc Au, that is, it is the spacing one would expect for hexagonal close-packing of Au. Diffraction Analysis. We analyzed the crystal structure using TEM diffraction tilting experiments. A typical set of diffraction patterns from one nanoparticle is shown in Figure 2a-c, and a convergent beam pattern in panel d, which gives additional three-dimensional information for a single projected diffraction pattern. Several features of these patterns combine to identify the hcp phase. First, the [11-20] pattern and the [10-10] pattern, 30° apart when tilting about the vertical direction, are precisely the patterns and angle one would obtain while tilting perpendicular to an hcp [0001] axis. The forbidden (0001) spacing appears by double diffraction in the first pattern, but not in the second, as expected from the hcp space group. Furthermore, the double spacing corresponding to this reflection does appear in the higher order zone of [10-10], as shown in Figure 3d, making it clear that this is a real lattice dimension. Second, we tilted parallel to the [0001] axis to map the reciprocal lattice in the third dimension. The [10-11] pattern is found after tilting approximately 40-45°, as estimated from the first and second tilt axes readouts, along the (1-210) Kikuchi lines (diffracting planes) of the [10-10] pattern, that is, toward the © 2010 American Chemical Society
FIGURE 3. EDS spectra of a Au catalyst (blue) and the Ge NW (red) below it. The GeKr and AuLr peaks, which are the major peaks used for quantitative analysis, overlap. Minimal intensity of the GeKβ and GeL peaks are observed in the Au catalyst spectrum. Quantitative analysis of this spectrum gives the Ge content as 3.6 at. %.
[0001] axis. The calculated angle between [10-10] and [1011] is 43°. We note that it is not possible to tilt the entire 90° to the [0001] hexagonal axis; however these three patterns are sufficient to identify the hcp structure. Compositional Analysis Using Energy Dispersive Spectrometry (EDS). It is important to determine if any Ge remains in the Au when it crystallizes from the eutectic liquid, because a hexagonal metastable alloy phase with lattice parameters similar to those expected for hcp Au, has been reported in 3303
DOI: 10.1021/nl100913d | Nano Lett. 2010, 10, 3302-–3306
FIGURE 4. Cs-corrected HRTEM images of the hcp nanocatalyst. (a) This catalyst particle showed little faulting overall (compare with Figure 1). (b) An enlarged view of the region marked in (a). The stacking pattern symmetry of the outlined triangles in (b) is the same as for the hcp 〈11-20〉 view in (c), thus demonstrating the ABABAB stacking of the hcp structure.
FIGURE 5. Schematic Gibbs free energy vs composition curves for a deep subeutectic temperature. The tangent to the liquidus curve at composition XGe(1) represents supersaturation during growth and indicates the energy barrier to Au nucleation. As the composition shifts back toward the eutectic composition at the end of growth, the metastable phase curve may cross the tangent line (e.g., at XGe (2)), while the driving force for fcc Au nucleation remains negative, creating a transient cooling regime where nucleation of the metastable β phase is favorable and competes with nucleation of equilibrium Au.
28-31
splat cooling experiments. This so-called β phase is reported to have a composition of 16-25 at. % Ge. We used EDS analysis in STEM mode to determine the Ge content of the hcp nanoparticles. The GeKR and AuLR peaks, the major peaks generally used for quantification because they are in the low background region of the EDS spectra, overlap so that a direct visual estimation of Ge content was not possible. Quantitative analyses using standard deconvolution EDS software gave compositions ranging from 0 to 4 at. % for several hcp particles (see Methods in Supporting Information). A typical example is shown in Figure 3. We note that minimal intensity is observed for the nonoverlapping GeKβ and GeL peaks. We conclude that there is minimal Ge in the hcp Au nanoparticles, at most a few atomic percent. This is consistent with the 3 at. % solubility limit for fcc Au near the Au-Ge eutectic temperature given by the phase diagram.28 High-resolution transmission electron microscopy (HRTEM) was used to study the catalyst particles. Because of spherical aberration of the objective lens in conventional electron microscopes, the interpretable spatial resolution is limited.32,33 Moreover, the lens aberrations lead to a blurring effect, referred to as contrast delocalization, for nonperiodic structures such as interfaces or edges in nanowires and catalyst particles. Here, we used an electron microscope with a Cs-corrected imaging lens to study the nanostructures. Cscorrected electron microscopy can provide image resolution equivalent to the information limit, allowing direct interpretation of images with sub-angstrom resolution. Thus, it allows us to directly examine the ABAB stacking sequence of the hcp phase and distinguish it from the ABCABC stacking of the fcc phase. A Cs-corrected HRTEM image of a nanoparticle viewed in 〈11-20〉 is shown in Figure 4a,b. A schematic comparing fcc and hcp stacking, viewed along the fcc 〈110〉 and the hcp 〈11-20〉 zone axes, is shown in Figure 4c. Figure 4b clearly exhibits the ABAB hcp stacking sequence. Both hcp and fcc Au nanoparticles are found in our samples and occur whether the substrate is Si or Ge, and for a variety of nanowire samples within the range of our © 2010 American Chemical Society
processing parameters, where the growth temperature varies from 280 to 310 °C and the initial colloidal Au nanocatalyst size is 40 nm, resulting in 40-50 nm Au tips. We note, however, that we have not found hcp Au in nanowire samples where the catalyst has been remelted and crystallized during in situ TEM heating experiments following growth. We found that the fcc nanocatalysts exhibit a random orientation relationship with the nanowire, whereas the hcp nanoparticles exhibit noticeable orientation relationships, the predominant one having the (0001) planes oriented at 60-65° to the NW {111} growth planes. This orientation relationship corresponds to {111}Ge//{10-11}hcpAu; 〈110〉Ge // hcpAu. We do not, at present, have an explanation for the observed relationship, since the relevant planes do not appear to present a good lattice match. However, we have used this observation to find many more of the hcp Au particles by tilting to a 〈110〉 zone axis of a NW, and then looking for a hcp pattern in the Au catalyst. This has been a consistently reliable way to find the hcp structures. As a result of this searching process, we estimate roughly 1 in 10 Au nanocatalysts have the hcp structure. Our results clearly demonstrate the formation of hcp Au occurring in some of the liquid Au-Ge catalysts as they crystallize during postgrowth cooldown of Ge NWs and suggest that the formation of hcp Au depends on the nanoscale size of this solidification system. The dearth of literature reports of hcp structures in Au nanoparticles produced by other processes suggests that our results are specific to certain aspects of nanowire growth and of the Au-Ge system. We also note that where Au catalyst structures have been characterized for other nanowire sytems, the hcp structure has not been reported.34,35 Observation of a distribution of hcp and fcc Au nanocatalysts in a given 3304
DOI: 10.1021/nl100913d | Nano Lett. 2010, 10, 3302-–3306
sample suggests that kinetics plays a role in which structure forms at the end of each NW. We therefore propose that the formation of hcp Au may be preceded by transient formation of the metastable hcp β-alloy phase that was reported for splat cooling experiments, followed by compositional segregation and Ge deposition onto the NW. The β-phase is reported to be an hcp solid solution of Au and Ge with lattice parameters similar to those expected for hcp Au,28-31 as determined from the lattice parameter of fcc Au, so it is a plausible precursor to the hcp Au crystal. This proposed metastable pathway to hcp Au may be promoted by the constraints on fcc Au nucleation during undercooling in such a nanoscale system8,10,11 and by the initial Ge supersaturation of the catalyst liquid that exists as nanowire growth terminates.7,10 With regard to undercooling, we note that our quench rates are expected to be similar for both as-grown NWs and for TEM heating experiments on the order of 1 °C/s. This is in contrast to typical splat cooling rates, which can be 105 °C/s or greater.30,31 However, the smallest dimension in typical splat cooling experiments is on the order of tens of micrometers. The much smaller size of the 40-50 nm liquid droplets involved in our nanowire growth allows for a high degree of undercooling, even at relatively low cooling rates, thus increasing the possibility of forming metastable phases. Studies of Au-Ge melts show that in the Au-rich region, up to about 25-30 at. % Ge, the liquid structure has a density consistent with close-packing31,36 while splat-cooling studies of the effect of the initial melt temperature suggest that viscosity decreases significantly with temperature at and below the eutectic temperature.30 With such small volumes, it becomes relatively easy to cool to a temperature without Au nucleation at which the liquid becomes sufficiently viscous to trap the Ge solute in the metastable hcp β structure. Solid-state diffusion of Ge from this alloy nanoparticle to the Ge NW may occur subsequently, during the rather slow cooldown to room temperature, leaving minimal Ge in the hcp β structure. The observations of hcp Au in as-grown samples, but not during postgrowth TEM melting and crystallization, suggests that the proposed transient β phase formation may also depend on Ge supersaturation of the liquid Au-Ge. This is expected to occur during NW growth under GeH4/H2 pressure, but not during in situ TEM heating in vacuum. Increased Ge content resulting from supersaturation can inhibit fcc Au solidification in the Au-Ge droplet by creating a kinetic barrier to crystal nucleus formation. Figure 5 shows a schematic free energy diagram at a deep subeutectic temperature, where a two-phase mixture of Au and Ge is the equilibrium condition, but the droplet remains liquid. We have added a hypothetical curve for the metastable β-phase based on its occurrence in splat cooling experiments.28-31 As the composition shifts back toward the eutectic composition at the end of growth, for example, from XGe(1) to XGe(2), there may be a transient regime where nucleation of the β-phase becomes favorable and competes with formation © 2010 American Chemical Society
of equilibrium fcc Au. We propose that formation of the hexagonal β phase may be followed by compositional segregation, resulting in a mixture of fcc and hcp Au nanoparticles at the Ge NW tips. In conclusion, using TEM diffraction and compositional analysis, and Cs-corrected HREM imaging, we have characterized the formation of an hcp structure in Au catalyst nanoparticles during crystallization following Ge NW growth. The identification of both fcc and hcp phases indicates multiple pathways for crystallization of the nanocatalyst and suggests that stabilization of the subeutectic Au-Ge liquid involves more complex phase relationships than previously considered. We suggest that this unusual observation of hcp Au may result from the ability of the Au-Ge system to form the structurally similar metastable hcp alloy β-phase, followed by out-diffusion of Ge from the crystal onto the Ge nanowire. This metastable pathway may be followed because of the kinetic barrier to direct nucleation of equilibrium fcc Au, which results from both size effects and supersaturation of the eutectic liquid with Ge. Acknowledgment. We thank the Intel Foundation for a Graduate Fellowship (I.A.G.), and gratefully acknowledge the support of Canon Corp., the Agilent Foundation, DARPA/ SPAWAR Grant N66001-04-1-8916, and the FCRP MSD Center (Theme ID: 887.011). Support from STINT (The Swedish Foundation for International Cooperation in Research and Higher Education) and Chalmers Nano-Initiative is gratefully acknowledged. Supporting Information Available. Description of synthesis and details of EDS analysis. This material is available free of charge via the Internet at http://pubs.acs.org. REFERENCES AND NOTES (1) (2) (3) (4) (5) (6) (7) (8) (9)
(10) (11) (12) (13) (14) 3305
Bootsma, G. A.; Gassen, H. J. J. Cryst. Growth 1971, 10, 223. Miyamoto, Y.; Hirata, M. Jpn. J. Appl. Phys. 1975, 14, 1419. Kamins, T. I.; Williams, R. S.; Basile, D. P.; Hesjedal, T.; Harris, J. S. J. Appl. Phys. 2001, 89, 1008. Wang, D.; Dai, H. Angew. Chem., Int. Ed. 2002, 41, 4783. Greytak, A. B.; Lauhon, L. J.; Gudiksen, M. S.; Lieber, C. M. Appl. Phys. Lett. 2004, 84, 4176. Adhikari, H.; Marshall, A. F.; Chidsey, C. E. D.; McIntyre, P. C. Nano Lett. 2006, 6, 318. Kodambaka, S.; Tersoff, J.; Reuter, M. C.; Ross, F. M. Science 2007, 316, 729. Adhikari, H.; Marshall, A. F.; Goldthorpe, I. A.; Chidsey, C. E. D.; McIntyre, P. C. ACS Nano 2007, 1, 415. Marshall, A. F.; Adhikari, H.; Chidsey, C. E. D.; McIntyre, P. C. Microscopy and Microanalysis, Proceedings of the Microscopy and Microanalysis Meeting; Price, R. L., Ed.; Fort Lauderdale, FL, August 5-7, 2007; Cambridge University Press: New York, 2007; Vol. 13, p 718. Adhikari, H.; McIntyre, P. C.; Marshall, A. F.; Chidsey, C. E. D. J. Appl. Phys. 2007, 102, No. 094311. Schwalbach, E. J.; Voorhees, P. W. Nano Lett. 2008, 8, 3739. Tian, W.; Sun, H. P.; Pan, X. Q.; Yu, J. H.; Yeadon, M.; Boothroyd, C. B.; Feng, Y. P.; Lukaszew, R. A.; Clarke, R. Appl. Phys. Lett. 2005, 86, 131915. Taneja, P.; Banerjee, R.; Ayyub, P.; Dey, G. K. Phys. Rev. B 2001, 64, No. 033405. Liu, X.; Luo, J.; Zhu, J. Nano Lett. 2006, 6, 408. DOI: 10.1021/nl100913d | Nano Lett. 2010, 10, 3302-–3306
(15) Liang, C.; Terabe, K.; Hasegawa, T.; Aono, M. Jpn. J. Appl. Phys. 2006, 45, 6046. (16) Singh, A.; Ghosh, A. J. Phys. Chem. C 2008, 112, 3460. (17) Zhou, Y. Nanotechnology 2008, 19, 285711. (18) Mi, Y.; Yuan, D.; Liu, Y.; Zhang, J.; Xiao, Y. Mater. Chem. Phys. 2005, 89, 359. (19) Illy, S.; Tillement, O.; Machizaud, F.; Dubois, J. M.; Massicot, F.; Fort, Y.; Ghanbaja, J. Philos. Mag. A 1999, 79, 1021. (20) Chinnasamy, C. N.; Jeyadevan, B.; Shinoda, K.; Tohji, K.; Narayanaasamy, A.; Sato, K.; Hisano, S. J. Appl. Phys. 2005, 97, 10J309. (21) Tzitzios, V.; Basina, G.; Gjoka, M.; Alexandrakis, V.; Georgakilas, V.; Niarchos, D.; Boukos, N.; Petridis, D. Nanotechnology 2006, 17, 3750. (22) Pirkkalainen, K.; Vainio, U.; Kisko, K.; Elbra, T.; Kohout, T.; Kotelnikova, N. E.; Serimaa, R. J. Appl. Crystallogr. 2007, 40, s489. (23) Carturan, G.; Cocco, G.; Enzo, S.; Ganzerla, R.; Lenarda, M. Mater. Lett. 1988, 7, 47. (24) Chushak, Y.; Bartell, L. S. Eur. Phys. J. D 2001, 16, 43. (25) Radetic, T.; Lancon, F.; Dahmen, U. Phys. Rev. Lett. 2002, 89, No. 085502.
© 2010 American Chemical Society
(26) Lucadamo, G.; Medlin, D. L. Science 2003, 300, 1272. (27) Dubrovinsky, L.; Dubrovinskaia, N.; Crichton, W. A.; Mikhaylushkin, A. S.; Simak, S. I.; Abrikosov, I. A.; de Almeida, J. S.; Ahuja, R.; Luo, W.; Johansson, B. Phys. Rev. Lett. 2007, 98, No. 045503. (28) Okamoto, H.; Massalski, T. B. Bull. Alloy Phase Diagrams 1984, 5, 601. (29) Anantharaman, T. R.; Luo, H.-L.; Klement, W. Trans. Metall. Soc. AIME 1965, 233, 2014. (30) Scott, M. G. Mater. Sci. Eng. 1975, 18, 279. (31) Rivlin, V. G.; Waghorne, R. M.; Williams, G. I. Gold Bull. 1976, 9, 84. (32) Rose, H. Ultramicroscopy 2005, 103, 1. (33) Fu, L. F.; Wang, Y. C.; Jiang, B.; Shen, F.; Strauss, M.; Van Leer, B.; Senowitz, C.; Buxbaum, A. IFSTA 2008: Proceedings from the 34th International Symposium for Testing and Failure Analysis, Portland, OR, November 2-6, 2008; pp 14-20. (34) Harmand, J. C.; Patriarche, G.; Pere-Laperne, N.; Merat-Combes, M.-N.; Travers, L.; Glas, F. Appl. Phys. Lett. 2005, 87, 203101. (35) Park, H. D.; Gaillot, A.-C.; Prokes, S. M.; Cammarata, R. C. J. Cryst. Growth 2006, 296, 159. (36) Waghorne, R. M.; Rivlin, V. G.; Williams, G. I. J. Phys. F 1976, 6, 147.
3306
DOI: 10.1021/nl100913d | Nano Lett. 2010, 10, 3302-–3306
