The Synergic Effect of Atomic Hydrogen Adsorption and Catalyst

Jul 26, 2016 - Here, we used in situ scanning electron microscopy and spatially resolved Auger spectroscopy to reveal the essential role of atomic hyd...
1 downloads 6 Views 4MB Size
Letter pubs.acs.org/NanoLett

The Synergic Effect of Atomic Hydrogen Adsorption and Catalyst Spreading on Ge Nanowire Growth Orientation and Kinking Miroslav Kolíbal,*,†,‡ Tomás ̌ Pejchal,‡ Tomás ̌ Vystavěl,§ and Tomás ̌ Šikola†,‡ †

Institute of Physical Engineering, Brno University of Technology, Technická 2, 616 69 Brno, Czech Republic CEITEC BUT, Brno University of Technology, Purkyňova 123, 616 69 Brno, Czech Republic § FEI Company, Vlastimila Pecha 1282/12, 627 00 Brno, Czech Republic ‡

S Supporting Information *

ABSTRACT: Hydride precursors are commonly used for semiconductor nanowire growth from the vapor phase and hydrogen is quite often used as a carrier gas. Here, we used in situ scanning electron microscopy and spatially resolved Auger spectroscopy to reveal the essential role of atomic hydrogen in determining the growth direction of Ge nanowires with an Au catalyst. With hydrogen passivating nanowire sidewalls the formation of inclined facets is suppressed, which stabilizes the growth in the ⟨111⟩ direction. By contrast, without hydrogen gold diffuses out of the catalyst and decorates the nanowire sidewalls, which strongly affects the surface free energy of the system and results in the ⟨110⟩ oriented growth. The experiments with intentional nanowire kinking reveal the existence of an energetic barrier, which originates from the kinetic force needed to drive the droplet out of its optimum configuration on top of a nanowire. Our results stress the role of the catalyst material and surface chemistry in determining the nanowire growth direction and provide additional insights into a kinking mechanism, thus allowing to inhibit or to intentionally initiate spontaneous kinking. KEYWORDS: Nanowire, germanium, vapor−liquid−solid growth, faceting, hydrogen, nanowire kinking wires (usually less than 20 nm) prefer to grow in the ⟨110⟩ direction, while the larger diameter ones grow in the ⟨111⟩ direction.14 This behavior has been observed in many material systems, including Si,7 III-Vs,15 II-VIs,16 and also Ge nanowires grown from molecular hydride precursors in chemical vapor deposition (CVD).17 However, the model does not seem to reproduce well the experiments on Ge NWs grown from atomic vapor (by evaporation from a solid source in Molecular Beam Epitaxy, MBE), which tend to grow in the ⟨110⟩ growth direction for a wide range of diameters. Just recently, Sivaram et al.18 demonstrated that the nanowire sidewalls are covered by adsorbed hydrogen during growth from hydride precursors and Gamalski et al.19 speculate that the byproducts of digermane decomposition passivate the surface during growth, significantly influencing Ge nanowire morphology. These studies emphasize the importance of adsorbed species, hydrogen in particular, during the growth. Because the partial pressure of atomic hydrogen is quite low in experiments where atomic vapor is utilized for NW growth (≪10−5 mbar), it is most probably the key element making the difference between these two growth techniques, that is, CVD and MBE. Here, by utilizing the growth of Ge NWs by evaporation from a solid Ge source with assistance of atomic hydrogen, we are able

I

n nanowire-based three-dimensional device architecture (e.g., gate-all-around nanowire field effect transistors), it is required to intentionally grow nanowires at a precisely determined wafer location and, additionally, perpendicular to the substrate.1 The former requirement was successfully met utilizing the so-called vapor−liquid−solid (VLS) mechanism,2 where the eutectic catalyst droplet is used to localize semiconductor material nucleation and growth to the dropletsubstrate (liquid−solid) interface. The growth direction of VLS-grown semiconductor nanowires and its control is, however, a complex task attracting a lot of attention.3 It has been demonstrated that group IV nanowires preferentially grow in the ⟨111⟩ direction, irrespective of the substrate crystallographic orientation.4−6 Schmidt et al.7 analyzed the nanowire growth considering surface free energies of the growth interface and nanowire sidewalls. They realized that the nanowire growth direction is not resulting from energy minimization of the whole nanowire/catalyst system but originates from the energy balance between the catalyst/nanowire growth interface and a small nanowire segment below the catalyst where atom ordering during solidification takes place.8 This concept based solely on thermodynamics introduced two possibilities to control the nanowire growth direction: either by catalyst engineering (thus potentially changing the solid−liquid interface energy)9,10 or by passivating nanowire sidewalls with different adsorbates (solid−vapor interface energy).11−13 Another success of Schmidt’s model is that it explained an intriguing experimental observation that small diameter nano© XXXX American Chemical Society

Received: March 31, 2016 Revised: July 22, 2016

A

DOI: 10.1021/acs.nanolett.6b01352 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 1. SEM images of germanium nanowires grown in high vacuum (a,b) and with atomic hydrogen (c,d) on Ge(111) (a,c) and Ge(110) (b,d) substrates. The growth temperatures were 425 °C in (a,b) and 385 °C in (c,d). Viewing directions are close to (a) [1̅10], (b) [1̅12], (c) [11̅0], and (d) [11̅ 0] (the samples are tilted by 85° to the surface normal and by 5° to the designated viewing direction). The insets in (a,d) are top-view SEM images of the sample and the schematics illustrate the possible degenerate nanowire growth orientations (gray, ⟨110⟩; white, ⟨111⟩). Scale bars, 400 nm.

limited from the lower end by surface diffusion of germanium atoms across the substrate and NW sidewalls, and by their desorption from the upper end.20,21 Diameter values of the nanowires grown in this study, and generally in MBE, are at the larger end of the size range studied in the field (40−200 nm). All the nanowires grow in the ⟨110⟩ direction, irrespective of the growth temperature (see Supporting Information, Figure S1a), substrate crystallographic orientation, and nanowire diameter, as seen in Figure 1a,b. The assignment of the growth direction and sidewalls orientation was made by assuming the diamond cubic lattice structure and determining an angle between the growth axis and the substrate (see Figure S2) and, additionally, verified by a detailed TEM analysis (Figure S3). It is interesting to note that the ⟨110⟩ oriented NWs on the (110) substrate grow only perpendicularly to the substrate and not in the other four degenerate (crystallographically equivalent) inclined ⟨110⟩ directions (Figure 1b). This is different from the (111) substrate, where all three equivalent ⟨110⟩ growth directions are observed (Figure 1a). The nanowires grown under (ultra)high vacuum have mostly a truncated rhombohedral cross section with four major {111} and two truncating {001} sidewalls (Figure 2a). The latter are often overgrown and the cross section changes to rhombohedral.22,23 The growth interface is formed by two inclined {111} plains, which are mutually tilted. This morphology is observed irrespective of the growth temperature within the growth window (385−455 °C), see Supporting Information, Figure S1a. The presence of atomic hydrogen in the chamber during growth changes the growth direction of NWs to ⟨111⟩ (Figure 1c,d) as well as nanowire morphology (Figure 2b). However, this change does not apply to the whole temperature growth window. The nanowires shown in Figures 1c,d and 2b were grown at 385 °C. At higher temperatures, we observe increased roughening of the sidewall facets and at temperatures above 435 °C the ⟨110⟩-oriented nanowires with clearly defined {111} sidewalls grow despite the presence of atomic hydrogen in the chamber (see Supporting Information, Figure S1b). In the following text, we will primarily focus on the nanowire growth with atomic hydrogen at low temperatures (385−400

to change the growth direction of Ge nanowires and thus to mimic nanowire growth from hydride precursors in terms of nanowire morphology and growth direction. We provide an in situ spectroscopic evidence of gold presence on the nanowire sidewalls and demonstrate its role in determining the nanowire growth direction and, potentially, kinking to other crystallographic directions. Ensembles of Ge nanowires were synthesized in an ultrahigh vacuum chamber (UHV, pbase < 3 × 10−9 mbar) coupled to a scanning auger microscope system (SAM, Omicron). Ge(111) and Ge(110) substrates were cleaned by ultrasonic treatment in acetone and isopropanol and subsequently covered with gold nanoparticles (40 nm in diameter) by dipping the sample in a 450:1 mixture of a colloidal gold solution with HCl (5%). After reaching the growth temperature in the UHV chamber the growth was carried out by deposition from an effusion cell (Ge flux oriented 30° to the sample surface) at a typical operation pressure of 5 × 10−9 mbar in case of no hydrogen gas delivery (i.e., at ultrahigh vacuum). The atomic hydrogen was generated by cracking molecular hydrogen (introduced to the chamber by a leak valve, pH2 = 2 × 10−5 mbar) on a heated tungsten wire, placed in the vicinity of the sample (∼2−3 cm). To examine the initial stage of nanowire growth and the kinking mechanism, Ge NWs were also grown inside a scanning electron microscope (SEM, FEI Quanta) with the base pressure pbase < 2 × 10−6 mbar using the same deposition method as in the UHV chamber (operation pressure 2 × 10−6 mbar, high vacuum) including the extension for cracking hydrogen molecules via a heated tungsten wire (pH2 = 2 × 10−5 mbar). Further experimental details can be found in Supporting Information. It is worth noting that growth under ultrahigh and high vacuum conditions without atomic hydrogen has delivered us identical results with respect to the studied subject. Therefore, in the paper we do not make a distinction in growth of nanowires under these two different vacuum conditions. The substrate temperature window for Ge NW growth by evaporation under high (or ultrahigh) vacuum conditions (i.e., without atomic hydrogen) is quite narrow (385−455 °C), B

DOI: 10.1021/acs.nanolett.6b01352 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

temperatures with atomic hydrogen have planar (111) growth interface (possible truncation as reported previously24 is not visible due to droplet presence) and hexagonal cross-section (Figure 2b). The {211}-oriented sidewalls exhibit different roughness on opposite facets (strongly resembling the sawtooth faceting of Si NWs grown by CVD).25 The sawtooth faceting of Ge nanowires has not been discussed in literature properly yet, although it has been already observed, especially above 300 °C.18,19,26,27 In the case of gold-catalyzed Si nanowires, the partially Au-decorated sawtooth facets stabilize the growth in the ⟨111⟩ direction.28 To confirm or exclude the presence of Au on Ge nanowire sidewalls, we have analyzed the nanowires by spatially resolved Auger spectroscopy. The spectra taken on the nanowire sidewalls clearly indicate the presence of a significant amount of Au on nanowires grown under high vacuum conditions (Figure 2a) and in contrast very small Auassigned peaks detected on sidewalls of ⟨111⟩-oriented nanowires grown at low temperatures with atomic hydrogen (Figure 2b). We have detected slight differences in Au signal between the two types of {211} sidewalls of the ⟨111⟩ nanowires, but the signal was too weak to be conclusive. Experiments with other gases (molecular hydrogen, O2, Ar, and CH4) introduced into the chamber during growth have proven that the effect of changing the growth direction is related specifically to atomic hydrogen, as the other gases do not induce this effect. The results presented so far thus implicate that (i) atomic hydrogen prevents gold from diffusing out of the catalyst across the sidewalls and (ii) gold promotes the formation of {111}oriented sidewall facets. The presence of hydrogen adsorbed on germanium nanowire sidewalls up to 330 °C in the case of nanowires grown from hydride precursors has been confirmed recently by Sivaram et al.18 Here, we generate atomic hydrogen more efficiently by thermal cracking of molecular hydrogen on a heated filament at high partial pressures (pH2 = 2 × 10−5 mbar) that results in a very high adsorption rate rads (dependent on the incoming flux). The nanowire sidewalls are thus flooded with atomic hydrogen that efficiently replenishes adsorbed hydrogen despite the high desorption rates at enhanced temperatures used in our growth experiments. The surface diffusion of gold atoms is critically dependent on other adsorbates.29 Our results show that the synergic effect of adsorbed hydrogen and gold spreading is critical for determining the nanowire sidewall orientation and, consequently, the growth direction. In the following text, we focus on the solid−vapor interface (H adsorption on sidewalls) and neglect liquid−vapor interface changes (H adsorption on eutectic droplet), being justified by desorption studies of H from gold surfaces (desorption temperatures vary between 100 and 216 K).30,31 The latest experimental findings also indicate negligible adsorption of H onto an eutectic AuGe droplet at relevant temperatures (>250 °C).18 The effect of gold on germanium faceting is demonstrated in Figure 3. As the surface energies of bare Ge facets of different crystallographic orientations are similar, the equilibrium crystal shapes are complex, exposing many different facets ({311} being the most frequently observed ones).32 A stress-free homoepitaxy of Ge on Ge(111) and Ge(110) surfaces results in germanium island shapes strongly resembling those obtained using the Wulff construction (Figure 3a,c). Experimental studies of flat Ge surfaces suggest that low-index planes are stabilized after exposure to atomic hydrogen due to (1 × 1)-H reconstruction.33−35 Thus, at low temperatures the nanowire

Figure 2. Morphology of nanowires grown (a) in high vacuum on (110) substrate and (b) with atomic hydrogen on (111) substrate as observed ex situ by SEM. Sketched top views show crystallographic orientations of the sidewalls; in (b) the ones with pronounced subfaceting are highlighted by zigzag lines. A detailed view of ⟨111⟩ nanowire sidewalls is shown in (b) with the viewing direction marked by the green arrow. The colored marks indicate locations where the Auger spectra were taken. The positions of the main Au MNN peaks (2015 and 2101 eV) are designated by the arrows. (a) Scale bar, 200 nm. (b) Scale bar, 100 nm.

°C) in which case the NWs adopt ⟨111⟩ growth direction. We observed that not all the degenerate ⟨111⟩ growth directions on the (111) substrate are equal, as the one perpendicular to the substrate is observed solely (Figure 1c). In contrast to NWs grown under high vacuum, all the NWs grown in the presence of atomic hydrogen still have catalytic gold droplets on their top (the gold droplets on NWs top are visible in Figure 1 due to different contrast). The ⟨111⟩ NWs grown at low C

DOI: 10.1021/acs.nanolett.6b01352 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

due to increasing importance of sidewall energies for low diameter nanowires. The surface free energy gain due to Au decoration is a key property of the Ge/Au system. Our real-time microscopic observation of initial stages of NW growth, shown in Figure 4, demonstrates how it affects the mechanism of growth interface formation and possible subsequent nanowire kinking. Focusing on the Ge(111) substrate first, the contact angle β between the catalytic droplet and the substrate is small and the droplet is initially nearly flat (the first image in the sequences in Figure 4a,c). This is a consequence of germanium diffusion from the substrate into the gold droplet and formation of an eutectic AuGe melt, accompanied by droplet sinking into the substrate and simultaneous formation of the (111) interface (see schematics in Figure 4a,c).38 Opening the shutter of the Ge evaporator (always the second image in the sequences) results in increasing the droplet curvature (i.e., increasing the contact angle) as the nanowire starts to grow, which is documented by the enhanced contrast in secondary electron images. This process is inevitably accompanied by an inward movement of the contact line (see white arrows in Figure 4a,c), as the growth interface shrinks to fit the droplet volume and optimum contact angle on top of a nanowire.39,40 With atomic hydrogen passivating the sidewalls (Figure 4c), nanowires continue to grow in the ⟨111⟩ direction as the formation of inclined {111} facets (which may result in kinking to another growth direction) is not energetically favorable. The situation is different in high vacuum (Figure 4a), where {111} sidewalls are energetically preferred due to Au decoration and the growth follows a scenario proposed in a model by Schwarz and Tersoff.41 During the initial stage of growth, the contact line is pinned at the edge between the growth interface and the preferred {111} sidewalls, therefore the droplet cannot reduce its diameter gradually to reach an optimal contact angle42 (as is the case with atomic hydrogen, see schematics in Figure 4c). Instead, the (111) growth interface is split into two {111} facets by introducing a step and the new facet quickly grows. This is how the V-shaped interface consisting of two {111} facets is formed. The growth direction changes to ⟨110⟩ and is

Figure 3. Homoepitaxial growth of Ge on bare (600 °C) (a,c) and gold covered (400 °C) (b,d) Ge(110) (a,b) and Ge(111) (c,d) surfaces at high vacuum (further experimental details can be found in Supporting Information). In (a,c) equilibrium crystal shapes are depicted (after Stekolnikov et al.).32 The SEM images in (b,d) are tilted by 52°, (a,c) and insets in (b,d) are top-views. All the scale bars are 200 nm, except in (c) (500 nm).

sidewalls are Au-free, passivated with atomic hydrogen.18 Nanowires then exhibit nearly cylindrical4 or hexagonal cross sections as the growth is not constrained by any preferential sidewalls orientation. However, the presence of Au on the sidewalls under conditions promoting Au diffusion (e.g., high vacuum or high temperature) dramatically changes the surface free energies in favor of {111} facets, as inferred from the shape of Ge crystallites which exhibit {111} facets only (Figure 3b,d). We speculate on formation of (√3 × √3)-Au reconstruction on the {111} facets,36 which coincides with the smooth appearance of the {111} sidewalls of ⟨110⟩-oriented NWs. Given that even the nanowires with the smallest diameters reported in the literature (e.g., 3.8 nm in ref 37) are faceted, the catalyst spreading and sidewall chemistry is also likely to affect faceting of nanowires with smaller diameters than reported here

Figure 4. Real-time SEM observation of the initial stage of NW growth on Ge(111) (a,c) and Ge(110) (b,d) in high vacuum (a,b) and with atomic hydrogen (c,d). The growth temperatures were 425 °C (a,b), 385 °C (c), and 390 °C (d) and the evaporation rate was kept the same in all experiments (1 nm/min). The first image in the sequences is always taken just before the Ge evaporation starts (marked with the full circle on a schematic time scale). The white arrows in the first image in sequences (a,c) highlight the size of the catalytic droplet (which is poorly visible due to low topographic contrast). The arrows are also shown in the second images for comparison. The white arrow in (d) points to the nanowire pedestal. All the images are tilted by 52° to the surface normal and the crystallographic orientation of the sample with respect to the viewing direction is the same as in Figure 3b,d. The images of individual nanowires taken ex situ are tilted by 85°. Scale bars, 200 nm. D

DOI: 10.1021/acs.nanolett.6b01352 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 5. Real-time SEM observation of NW response to changes of growth conditions. The growth temperature and partial pressure course of hydrogen are schematically depicted by red and blue lines, respectively (dashed blue line marks the base pressure of the system). The first image in the sequence is always taken after growth of a sufficiently long nanowire (a) with atomic hydrogen or (b) in high vacuum. The first two images (and schematics below) in (b) show a close-up view of the droplet before and after the temperature was lowered. The image in (c) is an SEM side-view of a nanowire resulting from an experiment performed with the same temperature step and duration as (b) but without introduction of atomic hydrogen. The circles on a schematic time scale depict the Ge evaporation status (full circle, no evaporation; open circle, evaporation proceeds) and the time indicates a period since the evaporation was restarted. The evaporation rate was kept the same in all experiments (1 nm/min). The substrate is Ge(111) and all the images (except c)) are tilted by 52° to the surface normal. The crystallographic orientation of the sample (except c)) with respect to the viewing direction is the same as in Figure 3d and 4a,c. Scale bars are (a) 100 nm and (b,c) 200 nm.

oriented (i.e., Au-decorated) nanowires (Figures 1a,b, 2a, and 4a,b). However, a significant portion of nanowires is shorter and without the catalytic droplet on top (Figure 1a). The realtime experiments in high vacuum indicate that the nanowire growth can be stopped at any time due to rapid droplet dissolution (see Supporting Information Figure S4), which is an additional source of gold decorating the sidewalls of other (still growing) nanowires. The ability to switch the growth direction by altering the vacuum conditions during the growth is attractive from the technological point of view, as it could enable formation of intentionally kinked nanowires.48 We have performed several attempts to kink the nanowires by either pumping out atomic hydrogen or introducing it in during growth under isothermal conditions but with a little success. Such an experiment is shown in Figure 5a. First, a ⟨111⟩-oriented segment was grown with atomic hydrogen in the SEM chamber, evaporation and hydrogen flow were stopped, and the SEM was pumped down to the base pressure while keeping the sample at the same temperature. However, further growth after the evaporation was restarted in high vacuum (no hydrogen delivery) did not result in a growth direction change. Although new inclined {111} facets were formed, they were not wet by the droplet and the nanowire continued to grow in the ⟨111⟩ growth direction. Similarly, we were not able to kink the ⟨110⟩-oriented NWs into a ⟨111⟩ direction simply by introducing atomic hydrogen. On the basis of these experiments, we stress that thermodynamic considerations are not sufficient to explain or even control kinking. Hence, there exists an energetic barrier to overcome, so that kinking has to be enforced kinetically by changing pressure48,49 or evaporation rate,44 temperature,19 or combining both.11,17 As these alterations in growth conditions are often accompanied by changes in the catalytic droplet volume (and shape), we hypothesize that this energetic barrier originates from the kinetic force needed to drive the droplet out of its optimum configuration on top of the nanowire (established during the initial growth stage) and thus to adopt a new growth interface. To support this hypothesis in Figure 5b we demonstrate that kinking can be controllably induced by a counterintuitive process, that is, lowering the growth temperature. Upon decreasing the temperature, the eutectic droplet volume decreases and, subsequently, it dewetts one of the {111} growth facets. Subsequent growth under the assistance of atomic hydrogen results in the ⟨111⟩-oriented

stabilized by Au-covered {111}-oriented sidewalls (Figure 4a). Note that there are three equivalent ⟨110⟩ nanowire directions as there are three possible inclined {111} facets (see the inset in Figure 1a and the crystallite in Figure 3d, viewed in the same direction as the image sequence in Figure 4a). A closer inspection of the contact angles during the initial stages of growth (Figure 4a,c) yields unexpected observation that the contact angle is reduced in the presence of atomic hydrogen compared to the growth in high vacuum. Assuming a simple planar growth interface and sidewalls parallel to the growth direction, the droplet contact angle should be larger in the presence of adsorbed H atoms compared to the growth in high vacuum due to lower solid−vapor interface energy. However, the contact line pinning to the edges of faceted growth interface of ⟨110⟩-oriented nanowires affects the contact angle as well. Hence, the contact angle changes do not reflect the adsorption of hydrogen only but result from a complex interplay of processes involved in V-shaped interface formation.41 The nanowire growth on the (110) substrate follows the same principles (Figure 4b,d). In high vacuum (Figure 4b), the V-shaped growth interface is formed immediately after the Ge deposition starts. The NWs viewed at their early growth stage have similar shape to the crystallites grown on a relevant substrate (Figure 3b). The V-shaped interface is made of two {111} facets which are inclined by 35.3° to the substrate and the {111}-oriented NW sidewalls are parallel to the growth direction. The growth in other inclined ⟨110⟩ directions (schematic in Figure 1b) is possible only if the initially formed V-shaped interface includes one of the {111} facets perpendicular to the substrate, which is difficult on a planar (110) substrate. Note that such inclined ⟨110⟩ nanowires were observed under growth conditions similar to ours on rough and stepped surface,43 which is consistent with our explanation. Growth on the (110) substrate with atomic hydrogen is accompanied by formation of a faceted pedestal (exposing different facets, similar to Figure 3a), a remnant of the initial stage of (111) growth interface formation (Figure 4d, white arrow). This feature has been commonly observed ex situ on inclined nanowires in different material systems,12,17,44−47 as well as in theoretical modeling.41 The loss of catalyst material is usually accompanied by shrinking of the nanowire diameter and eventually growth cease.29 Surprisingly, we have not seen any tapering of ⟨110⟩E

DOI: 10.1021/acs.nanolett.6b01352 Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters



growth. Note that the growth direction change is not primarily caused by the temperature step, as the subsequent growth under the same vacuum conditions (no atomic hydrogen) yields ⟨110⟩-oriented NW segments (Figure 5c), albeit with a smaller diameter. The temperature step is introduced to allow the droplet to unpin from the original V-shaped interface and to adopt a new growth interface and geometry, corresponding to a smaller volume and different composition of the droplet. Now the situation is similar to the initial stages of the growth, where the orientation of nanowires is controlled by sidewall chemistry, leading to nanowire kinking to the ⟨111⟩ direction (when atomic hydrogen is introduced, Figure 5b) or resuming ⟨110⟩-oriented growth (no atomic hydrogen, Figure 5c). Similar to temperature changes, variations of the Ge flux may cause droplet instability, because the supersaturation in the droplet changes. This is not the case here (Figure 5) where small growth rates are utilized (additionally, control experiments to those presented in Figure 5 with a continuous Ge flux yielded the same results). However, the probability of kinking events after a Ge flux change grows with the increasing NW growth rate, where the nanowires are more susceptible to kinking events17,41 as the system is driven even further from equilibrium. In summary, we have shown that the decoration of nanowire sidewalls with different species affects the preferential nanowire growth direction. For the Ge nanowire/Au catalyst system, our spectroscopic data indicate that the adsorbed atomic hydrogen suppresses Au diffusion out of the catalyst droplet, which readily occurs during growth in high vacuum. The presence of gold on nanowire sidewalls (i) results in preferential formation of {111}-oriented facets due to energy minimization and (ii) changes catalyst wetting of the initial pedestal, which plays a crucial role in determining the growth direction. The initial formation of the growth interface is inevitably accompanied by changes in the droplet wetting angle and liquid−solid interface area as the whole system is reaching (meta)stable state. These findings, based on real-time microscopic experiments, are promising with respect to their utilization in absolute control of the growth direction. A smart choice of the catalyst material and sidewall passivation can enable the nanowire growth in a specific direction, despite the existence of multiple crystallographically equivalent directions. Additionally, although spontaneous kinking is frequently observed in nanowire growth experiments, our data suggest that in certain systems (e.g., where significant differences in morphology exist between nanowires growing in different growth directions) an energy barrier has to be overcome to initiate a nanowire growth direction change.



Letter

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This research has been financially supported by the Ministry of Education, Youth and Sports of the Czech Republic under the projects CEITEC 2020 (LQ1601)) and CEITEC Nano (LM2015041), Grant Agency of the Czech Republic (1616423Y) and Brno University of Technology (FSI/STI-J-152823). M.K. and T.P. acknowledge the support from FEI Company. The authors would like to thank J. Michalička and L. Kachtı ́k for TEM measurements and J. Č echal and P. Varga for critical reading of the manuscript.



REFERENCES

(1) Thelander, C.; Agarwal, P.; Brongersma, S.; Eymery, J.; Feiner, L. F.; Forchel, A.; Scheffler, M.; Riess, W.; Ohlsson, B. J.; Gösele, U.; Samuelson, L. Mater. Today 2006, 9, 28. (2) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89. (3) Fortuna, S. A.; Li, X. Semicond. Sci. Technol. 2010, 25, 024005. (4) Kamins, T. I.; Williams, R. S. Nano Lett. 2004, 4, 503. (5) Adhikari, H.; Marshall, A. F.; Chidsey, C. E. D.; McIntyre, P. C. Nano Lett. 2006, 6, 318. (6) Jagannathan, H.; Deal, M.; Nishi, Y.; Woodruff, J.; Chidsey, C.; McIntyre, P. C. J. Appl. Phys. 2006, 100, 024318. (7) Schmidt, V.; Senz, S.; Gösele, U. Nano Lett. 2005, 5, 931. (8) Howe, J. M.; Saka, H. MRS Bull. 2004, 29, 951. (9) Wang, J.; Plissard, S. R.; Verheijen, M. A.; Feiner, L. − F.; Cavalli, A.; Bakkers, E. P. A. M. Nano Lett. 2013, 13, 3802. (10) Kuykendall, T. R.; Altoe, M. V. P.; Ogletree, D. F.; Aloni, S. Nano Lett. 2014, 14, 6767. (11) Musin, I. R.; Filler, M. A. Nano Lett. 2012, 12, 3363. (12) Kodambaka, S.; Hannon, J. B.; Tromp, R. M.; Ross, F. M. Nano Lett. 2006, 6, 1292. (13) Shin, N.; Filler, M. A. Nano Lett. 2012, 12, 2865. (14) Wu, Y.; Cui, Y.; Huynh, L.; Barrelet, C. J.; Bell, D. C.; Lieber, C. M. Nano Lett. 2004, 4, 433. (15) Yang, Z.-X.; Yip, S.; Li, D.; Han, N.; Dong, G.; Liang, X.; Shu, L.; Hung, T. F.; Mo, X.; Ho, J. C. ACS Nano 2015, 9, 9268. (16) Cai, Y.; Chan, K. S.; Sou, I. K.; Chan, Y. F.; Su, D. S.; Wang, N. Adv. Mater. 2006, 18, 109. (17) Madras, P.; Dailey, E.; Drucker, J. Nano Lett. 2009, 9, 3826. (18) Sivaram, S. V.; Shin, N.; Chou, L. − W.; Filler, M. A. J. Am. Chem. Soc. 2015, 137, 9861. (19) Gamalski, A. D.; Tersoff, J.; Kodambaka, S.; Zakharov, D. N.; Ross, F. M.; Stach, E. A. Nano Lett. 2015, 15, 8211. (20) Porret, C.; Devillers, T.; Jain, A.; Dujardin, R.; Barski, A. J. Cryst. Growth 2011, 323, 334−339. (21) Dubrovskii, V. G.; Sibirev, N. V.; Suris, R. A.; Cirlin, G. E.; Harmand, J. C.; Ustinov, V. M. Surf. Sci. 2007, 601, 4395. (22) Kolibal, M.; Kalousek, R.; Vystavěl, T.; Novák, L.; Šikola, T. Appl. Phys. Lett. 2012, 100, 203102. (23) Hanrath, T.; Korgel, B. A. Small 2005, 1, 717. (24) Gamalski, A. D.; Ducati, C.; Hofmann, S. J. Phys. Chem. C 2011, 115, 4413. (25) Ross, F. M.; Tersoff, J.; Reuter, M. C. Phys. Rev. Lett. 2005, 95, 146104. (26) Ryu, S.-G.; Kim, E.; Yoo, J.-H.; Hwang, D. J.; Xiang, B.; Dubon, O. D.; Minor, A. M.; Grigoropoulos, C. P. ACS Nano 2013, 7, 2090. (27) Cho, H. S.; Kamins, T. I. J. Cryst. Growth 2010, 312, 2494. (28) Wiethoff, C.; Ross, F. M.; Copel, M.; Horn von Hoegen, M.; Meyer zu Heringdorf, F.-J. Nano Lett. 2008, 8, 3065. (29) Hannon, J. B.; Kodambaka, S.; Ross, F. M.; Tromp, R. M. Nature 2006, 440, 69.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.6b01352. Additional experimental details, morphology and growth direction of NWs grown within the temperature growth window (Figure S1), determination of nanowire growth direction and sidewall facets crystallographic orientation in SEM (Figure S2) and TEM (Figure S3) and SEM image sequence documenting a rapid loss of a catalyst during growth in high vacuum (Figure S4) (PDF) F

DOI: 10.1021/acs.nanolett.6b01352 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters (30) Sault, A. G.; Madix, R. J.; Campbell, C. T. Surf. Sci. 1986, 169, 347. (31) Pan, M.; Flaherty, D. W.; Mullins, C. B. J. Phys. Chem. Lett. 2011, 2, 1363. (32) Stekolnikov, A. A.; Bechstedt, F. Phys. Rev. B: Condens. Matter Mater. Phys. 2005, 72, 125326. (33) Su, C.; Tsai, C. − S.; Lin, C. − E.; Chen, K. − H.; Wang, J.-K.; Lin, J. − C. Surf. Sci. 2000, 445, 139. (34) Akazawa, H. J. Appl. Phys. 2006, 99, 103505. (35) Scholz, S. M.; Schreiner, J.; Jacobi, K. Surf. Sci. 1995, 331−333, 402. (36) Göthelid, M.; Hammar, M.; Björkqvist, M.; Karlsson, U. O.; Flodström, S. A. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 4470. (37) Wu, Y.; Cui, Y.; Huynh, L.; Barrelet, C. J.; Bell, D. C.; Lieber, C. M. Nano Lett. 2004, 4, 433. (38) Hajjar, S.; Garreau, G.; Josien, L.; Bubendorff, J. L.; Berling, D.; Mehdaoui, A.; Pirri, C.; Maroutian, T.; Renard, C.; Bouchier, D.; Petit, M.; Spiesser, A.; Dau, M. T.; Michez, L.; Le Thanh, V.; Mentes, T. O.; Nino, M. A.; Locatelli, A. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 84, 125325. (39) Schmidt, V.; Senz, S.; Gösele, U. Appl. Phys. A: Mater. Sci. Process. 2005, 80, 445. (40) Schwarz, K. W.; Tersoff, J. Phys. Rev. Lett. 2009, 102, 206101. (41) Schwarz, K. W.; Tersoff, J. Nano Lett. 2011, 11, 316. (42) Schwarz, K. W.; Tersoff, J.; Kodambaka, S.; Chou, Y.-C.; Ross, F. M. Phys. Rev. Lett. 2011, 107, 265502. (43) Schmidtbauer, J.; Bansen, R.; Heimburger, R.; Teubner, T.; Boeck, T. J. Cryst. Growth 2014, 406, 36. (44) Kolíbal, M.; Vystavěl, T.; Novák, L.; Mach, J.; Šikola, T. Appl. Phys. Lett. 2011, 99, 143113. (45) Cai, Y.; Chan, K. S.; Sou, I. K.; Chan, Y. F.; Su, D. S.; Wang, N. Small 2007, 3, 111. (46) Mikkelsen, A.; Skö ld, N.; Ouattara, L.; Lundgren, E. Nanotechnology 2006, 17, S362. (47) Fonseka, H. A.; Caroff, P.; Wong-Leung, J.; Ameruddin, A. S.; Tan, H. H.; Jagadish, C. ACS Nano 2014, 8, 6945. (48) Tian, B.; Xie, P.; Kempa, T. J.; Bell, D. C.; Lieber, C. M. Nat. Nanotechnol. 2009, 4, 824. (49) Lugstein, A.; Steinmair, M.; Hyun, Y. J.; Hauer, G.; Pongratz, P.; Bertagnolli, E. Nano Lett. 2008, 8, 2310.

G

DOI: 10.1021/acs.nanolett.6b01352 Nano Lett. XXXX, XXX, XXX−XXX