Real-Time Observation of Collector Droplet Oscillations during Growth

Feb 14, 2014 - In VLS-grown nanowires,(1-4) the atoms of the semiconductor ... Next, we focus on the hutlike sidewall growth because the sawtooth ...
0 downloads 0 Views 4MB Size
Letter pubs.acs.org/NanoLett

Real-Time Observation of Collector Droplet Oscillations during Growth of Straight Nanowires Miroslav Kolíbal,*,†,§ Tomás ̌ Vystavěl,‡ Peter Varga,§ and Tomás ̌ Šikola†,§ †

Institute of Physical Engineering, Brno University of Technology, Technická 2, 616 69 Brno, Czech Republic FEI Company, Podnikatelská 6, 612 00 Brno, Czech Republic § CEITEC BUT, Brno University of Technology, Technická 10, 61669 Brno, Czech Republic ‡

S Supporting Information *

ABSTRACT: A liquid droplet sitting on top of a pillar is crucially important for semiconductor nanowire growth via a vapor−liquid−solid (VLS) mechanism. For the growth of long and straight nanowires, it has been assumed so far that the droplet is pinned to the nanowire top and any instability in the droplet position leads to nanowire kinking. Here, using real-time in situ scanning electron microscopy during germanium nanowire growth, we show that the increase or decrease in the droplet wetting angle and subsequent droplet unpinning from the growth interface may also result in the growth of straight nanowires. Because our argumentation is based on terms and parameters common for VLS-grown nanowires, such as the geometry of the droplet and the growth interface, these conclusions are likely to be relevant to other nanowire systems. KEYWORDS: Germanium, nanowires, facet, contact angle, vapor−liquid−solid growth, in situ scanning electron microscopy

I

n VLS-grown nanowires,1−4 the atoms of the semiconductor material are supplied to the metal droplet either from a gas phase (chemical vapor deposition, CVD) or atomic vapor (physical vapor deposition, PVD). The supersaturation in the alloy droplet increases, and after it reaches a certain value, nucleation of the excess material is favorable at the liquid/solid interface.5 If the growth criteria are fulfilled,6 then upon completion of a monolayer of a semiconductor material the droplet slides up and another monolayer is grown, giving rise to a nanowire with the collector droplet on top. Naturally, in this growth mode, the droplet−nanowire interface is crucial for determining the final morphology of the wire. Ideally, the droplet shape would be a spherical cap because of energy minimization. However, because the semiconductor surfaces are generally faceted, the nanowire cross-section usually has the shape of a convex polygon instead of a circle. Using in situ TEM, it has been observed that the growth interface is not planar but truncated and the small side facets serve as preferential nucleation sites for oscillatory growth.7,8 If a new facet is introduced to the system, then it rapidly grows9 and the nanowire kinks in a new direction because the droplet stays pinned to the triple phase line (TPL, the interface between the three phases) and the former interface facet winds out.4,10−12 However, as more complicated nanowire morphologies are being fabricated, it has been speculated13 that under specific conditions the droplet might repeatedly pin and unpin from the TPL.14,15 Here, we demonstrate such droplet behavior using real-time observations from scanning electron microscopy (SEM), where one can easily obtain 3D information on the © 2014 American Chemical Society

sample morphology by tilting the sample and observing it from different angles. If the nanowire constituent atoms are supplied to the sample by evaporation, which is the case in our experiment, then the droplet acts as a collector2,16 of diffusing atoms along the substrate and nanowire sidewalls. Here, the experimental conditions to achieve the nanowire growth are limited with respect to CVD experiments because the amount of atoms collected by the droplet depends on an interplay between diffusion, desorption, and nucleation on the sample surface. The choice of the ratio of evaporation rate to substrate temperature is made to allow for a large diffusion length of adatoms and a small nucleation rate. For the Ge NW system seeded with Au droplets on a Ge substrate (having either (111) or (100) orientations), the optimum conditions are slightly above the eutectic temperature of Au/Ge melt using a very low evaporation rate10,17 (see Methods for experimental details). Under these growth conditions, the noncatalyzed growth on the substrate is significantly suppressed. The two major nanowire growth directions, ⟨110⟩ and ⟨111⟩, on substrates of both orientations were observed in situ10 (see Supporting Information Figures S1 and S2 for chemical analysis). However, the sidewalls of the nanowires growing in these two distinct directions show different morphologies. Although the ⟨110⟩oriented nanowires are long and have smooth and stable (111) sidewall facets,10,16 the ⟨111⟩-oriented nanowires on these two Received: November 8, 2013 Revised: February 4, 2014 Published: February 14, 2014 1756

dx.doi.org/10.1021/nl404159x | Nano Lett. 2014, 14, 1756−1761

Nano Letters

Letter

Figure 1. ⟨111⟩-oriented nanowire morphology and associated collector droplet behavior for nanowires grown on both substrates. (a) ⟨111⟩oriented nanowire grown on the Ge(111) substrate as seen by SEM along the [112̅] (left) and [01̅1] directions (middle). The inset shows a different nanowire from the top, with the top (111) terminating facet marked as T (outlined with a dashed red line) and three (111)-oriented sidewall facets (S1, S2, and S3, with facet S1 outlined in red in the right schematic). A schematic view of the ⟨111⟩-oriented nanowire is shown on the right. (b, c) Images taken during real-time in situ observation of ⟨111⟩-oriented nanowire growth on the Ge(100) substrate at 400 °C. The droplet mostly resides on the top facet, T, but, alternatively, the triple phase line (TPL) slides down and wets the top facet as well as (111)-oriented sidewalls S1, S2, or S3 simultaneously. The scale bars are 100 nm.

a diameter of ∼103 nm growing in the ⟨111⟩ direction on the Ge(100) substrate and monitored the time that the droplet spends on each facet. The results are shown in Figure 2a together with SEM images of the studied nanowire showing its evolution during growth. In the beginning of the observation, the nanowire was already 230 nm long . The droplet spends a longer time wetting only the T facet rather than both the top and one of the sidewall facets simultaneously; however, with increasing nanowire length, this behavior reverses. In Figure 2b, we show a detailed time-resolved view of the growth during one sidewall wetting cycle. Because nucleation is favorable at the liquid−solid interface, it is not surprising that when the droplet wets the top facet only the nanowire grows in the ⟨111⟩ direction. This results in the shrinking of the top T facet because of the mutually inclined sidewalls. Once the droplet unpins and wets both the top and sidewall facet simultaneously, growth is also possible at the sidewall, which is indeed observed (see the solid white line in image 7 in Figure 2b). This compensates for the shrinking of the top T facet. Moreover, the step formed at the sidewall because of the catalyzed growth serves as a nucleation site for noncatalyzed step−flow growth on the sidewall, which is the mechanism for hut growth. This is the only case that we have seen noncatalyzed sidewall growth in this particular experiment. We illustrate the evolution of the corresponding facets during each growth step depicted in Figure 2b using schematic outlines of nanowire-growth evolution. When only the terminating T facet is wetted, the nanowire cross-sectional area actually shrinks because of the inward motion of the top facet edges (edges between the (111) sidewalls and the T facet, see schematic 3 and 4 in Figure 2b). Knowledge of the nanowire growth rate and geometry allowed us to calculate an elongation of the nanowire by 1.4 nm during the corresponding step. As a consequence, the T facet becomes narrower by 0.5 nm. This

substrates are generally shorter18,19 with a complicated morphology because of the absence of stable facets parallel to the growth direction (Figure 1). The nanowires growing in the ⟨111⟩ direction show a hexagonal terminating top (111) facet (marked as T in Figure 1). Similar to previous studies of Si nanowires,15,20 three sidewalls are sawtooth-faceted ((1̅1̅2), (12̅ 1)̅ , and (211̅ )̅ ). Contrary to previously studied Si nanowires, the remaining three sidewalls are not smooth but have a hut shape (outlined in red in Figure 1a) with faceted sides and smooth (111)-oriented top and bottom parts (marked as the S1, S2, and S3 facets in Figure 1). Postgrowth analysis of ⟨111⟩oriented nanowire hexagonal cross-sections showed that the hexagon is irregular and slightly different from wire-to-wire, depending on their length. Next, we focus on the hutlike sidewall growth because the sawtooth faceting was discussed previously.15,20 Real-time in situ SEM observations show unusual droplet behavior on top of a ⟨111⟩-oriented nanowire. The droplet does not wet the terminating T facet only, as is usually the case in VLS nanowire growth. Instead, the droplet undergoes an oscillatory motion. This is shown in Figure 1b,c, where a ⟨111⟩oriented nanowire is growing at the (100) substrate. Most of the time, the droplet resides on the top facet; however, it repeatedly unpins from one of the edges and slides down one of the (111) sidewalls (denoted S1, S2, or S3) of a hut. In the next step, the droplet pins again to the edge between the sidewall and the top facet and it stays there for a while until another unpinning event occurs. This oscillatory droplet motion was observed over a wide range of ⟨111⟩-oriented nanowire diameters (from 70 to 400 nm) and evaporation rates (3−8 Å/min) on both substrate orientations, differing only in the frequency of pinning and unpinning events. The real-time in situ SEM experiment allowed us to study the droplet behavior in more detail. We chose a nanowire with 1757

dx.doi.org/10.1021/nl404159x | Nano Lett. 2014, 14, 1756−1761

Nano Letters

Letter

Figure 2. continued image is a schematic showing the evolution of the nanowire crosssection (left) and side view (right) during individual steps (not to scale). The solid line represents the nanowire outline, the dashed line marks this outline in the previous step, and the red arrows highlight the direction of facet growth.

forces the droplet to finally unpin from one of the edges and to wet a sidewall. Subsequently, the sidewall grows outward (0.9 nm in direction normal to the sidewall, assuming the same growth rate), and the corresponding edge length decreases (schematic 5 and 6) as the T facet widens again (by 0.6 nm). The droplet dewetts the sidewall and is pinned back to the T facet (schematic 6). The quantitative analysis shows that the sidewall growth compensates for T-facet shrinking, although not fully. The reason is that in the consecutive steps the droplet wets other sidewalls while the T facet remains wetted, so the top facet edge from which the droplet was unpinned moves inward again. Hence, the system dynamically adjusts the facet wetting time and the sequence of the wetted sidewalls to compensate for T-facet shrinking (graph in Figure 2a). During elongation in the ⟨111⟩ direction, we noticed that the nanowires are more often kinking toward the ⟨110⟩ direction. This is because the droplet contact angle in this regime is very close to the advancing (or receding) one and the droplet is more susceptible to variations in the local environment (adatom flux changes, etc.).9 Kinking occurs because at one moment the droplet does not dewett the sidewall, subsequently changing the growth direction to ⟨110⟩. Nevertheless, we have observed ⟨111⟩-oriented nanowires with an aspect ratio 1:10 or more. Before we analyzed the nanowire growth in the new ⟨110⟩ direction, we focused on the fact that the droplet unpins from the top facet edges and partially wets the (111)-oriented sidewalls and not the sawtooth-faceted ones. Generally, a liquid droplet pinned to an edge remains there even if its volume, and hence the contact angle, is partially increased (or decreased); the well-known Young equation is no longer valid. The unpinning of the droplet from a sharp edge was studied by Gibbs, who derived the following criterion for the maximum (advancing) contact angle, θA, that the droplet can sustain, which is referred to as the Gibbs inequality21 θR ≤ θ ≤ θA = (180° − ϕ) + θY

Here, θ is the actual contact angle, ϕ is the angle between the two surfaces (Figure 3), and θY is the equilibrium contact angle (fulfilling the Young equation22 in an ideal case). θR is the minimum (receding) contact angle; the TPL moves inward if this value is reached. It follows that (1) the actual contact angle can vary over a certain range defined by the Gibbs inequality without TPL movement and (2) for larger ϕ, the advancing contact angle for droplet unpinning becomes smaller. The sawtooth-faceted sidewalls are perpendicular to the top facet and thus the angle ϕ is 90°. Because the angle ϕ between the top and (111) sidewall facets is 109.5°, the unpinning from the edge between these two latter planes is preferred as a result of the Gibbs inequality. To provide additional evidence for this, we performed simulations of droplet shape based on minimization of the droplet surface energy using a Surface Evolver code,23 as shown in Figure 3. It is apparent both from the experiment (Figure 3a) and simulations (Figure 3b) that the opposite contact angles are unequal because of a nonregular

Figure 2. Quantification of the droplet motion on the top of a ⟨111⟩ nanowire grown on the Ge(100) substrate at 400 °C with 3 Å/min Ge deposition rate from a real-time in situ experiment. (a) Time evolution of the time period during which the droplet wets only the T facet or both the T and sidewall facets simultaneously (S1, S2, or S3). Although the intervals are not regular, the general trend of shortening the wetting time of the T facet only is clearly visible. SEM images 1 and 2 show the nanowire shape evolution during observation. The dashed white line in image 2 marks an outline of the initial nanowire from image 1. (b) Images 3 (244 min after the growth start), 4 (plus 22 s), 5 (plus 2 s), 6 (plus 14 s), and 7 (plus 2 s) show a detailed view of the nanowire top during one sidewall wetting cycle. The images were captured during the recording of a real-time movie (see Supporting Information Videos S3 and S4). The dashed white line marks the nanowire shape from the previous step, and the solid white line highlights the actual nanowire outline. In images 3 and 6, a contact angle, θ, is marked. Scale bars are 100 nm. To the right of each SEM 1758

dx.doi.org/10.1021/nl404159x | Nano Lett. 2014, 14, 1756−1761

Nano Letters

Letter

(position O in the figure), the droplet oscillations begin. Clearly, the unpinning is possible from the longer edges only because θA is reached sooner. However, it is difficult to correlate the experimental values of the contact angles obtained from the SEM images (Figure 2b) with the simulations because their changes during droplet oscillations are too small to be quantified accurately. As discussed earlier, the droplet finally stays, wetting both the top T facet and one of the (111) sidewalls (this kinking can also be achieved intentionally by changing the evaporation rate).10 The droplet−nanowire interface consists of two inclined {111} facets having a trigonal (former sidewall) and truncated trigonal (former T facet) shape. Hence, the growth should proceed in the ⟨110⟩ direction, which is confirmed in Figure 4a (different nanowire than the one observed in Figure 2). However, the nanowire morphology is different from that which is usually obtained:10,24 there is a zigzag trench visible between two adjacent (111) sidewalls. Furthermore, the nanowire cross-section is formed by two truncated triangles, resembling the dumbbell-like cross-section that so far have been studied only theoretically.25 The formation of this crosssection is rather easy to explain because the simulations show that the very sharp corners cannot be wetted by the droplet.16,26 Therefore, in the absence of noncatalyzed sidewall growth (resulting from our in situ observations), the previously trigonal facet quickly grows into an irregular hexagon, and the cross-section becomes symmetric. To explain the formation of the zigzag trench, we have followed the growth of such a nanowire (Figure 4b, ⟨110⟩-viewing direction), concluding that the shape evolution is again accompanied by cyclic droplet motion. Compared to the previously discussed NW growth in the ⟨111⟩ direction, the droplet is significantly stretched if it fully wets the top interface because the interface area is not matched to the droplet volume.4 The contact angle is therefore small, and it soon reaches the critical receding value, θR. The droplet unpins from the outer edges (image 3) and wets onehalf of the top interface only (image 4). Now only the wet facet grows (image 5), which results in its shrinking. The contact angle rapidly increases, and as it reaches the critical advancing value, θA, the TPL moves again, and the droplet starts to wet the top interface fully (image 6). However, because the two inclined facets are unequal in size and the supersaturation in the droplet is small,10 the growth rate of each facet differs (the larger one grows faster,9 which results in its shrinking). Finally, the droplet unpins again, and the full cycle is completed. The droplet motion on top of a nanowire could have been also explained by effects other than nanowire geometry. At the higher growth temperatures employed here (compared to conventional CVD experiments),7 the droplets could generally be considered to be more unstable on the nanowire top. The growth of long ⟨110⟩-oriented nanowires with smooth sidewalls on the sample at the same time (see Supporting Information Figure S1), however, indicates that the temperature-related effect is not the driving force for droplet instability. The droplets on top of these nanowires are stable and do not unpin from the growth interface. The droplet volume changes (e.g., loss of gold because of Ostwald ripening)27 could be another cause of oscillatory motion. In this case, the nanowires would exhibit a tapered geometry, which was not seen in our experiments (see Supporting Information Figure S5). Additionaly, the oscillatory motion, as explored here, would require a periodical increase and decrease of the droplet volume, which is not a plausible scenario.

Figure 3. Contact-angle asymmetry of a droplet wetting the nonregular hexagonal top (111) facet. (a) SEM images taken during the growth show a ⟨111⟩-oriented nanowire viewed from the top and side; the top facet is outlined by the dashed line. The scale bar is 100 nm. (b) Droplet shape simulated using Surface Evolver according to the experimental conditions in panel a. The important angles are defined in the side view. (c) Dependence of the droplet contact angle (assuming the droplet constant volume) on the relative T-facet area. The shrinking of the top facet is caused by its continuous growth compared to that of the (111) sidewalls, resulting in the inward motion of its edges (schematic in Figure 2), thus corresponding to the experimental observations. Experimentally deduced advancing and receding contact-angle values are marked by horizontal lines; in the shaded areas, the droplet unpins from one of the edges.

hexagonal shape of the top facet. The contact angle is much larger at the longer edge (between the top and (111) sidewall facets, marked as θ) than at the shorter one (between the top facet and the sawtooth-faceted sidewalls parallel with the growth direction, marked as β). Hence, according to the Gibbs inequality, the unpinning is favored at the longer edge. In Figure 3c, the dependence of the droplet contact angles on the T-facet area relative to that of a regular hexagon is shown, as calculated using Surface Evolver and assuming a constant droplet volume. In the very beginning, we assume a nanowire with the hexagonal cross-section (position H in the figure). Geometrical constraints cause top facet shrinking during nanowire growth, and its shape changes from regular hexagon to a nearly triangular one. The contact angle, θ, increases, and contact angle β, defined at the narrower edge (opposite side), decreases. The horizontal lines mark the experimentally obtained values for the advancing contact angle θA = 125° (Figure 2b, image 4) and receding contact angle θR = 35° (Figure 2b, image 6). Once the critical contact angle is reached 1759

dx.doi.org/10.1021/nl404159x | Nano Lett. 2014, 14, 1756−1761

Nano Letters

Letter

Figure 4. Growth of a ⟨110⟩-oriented nanowire. (a) Top left: SEM image showing a nanowire growing in the ⟨111⟩ direction. The droplet wets the T facet only. The top right image shows the nanowire after the droplet jumps to the sidewall and the growth continued for a while with the droplet wetting both the sidewall and T facet. The growth direction was changed to ⟨110⟩. Bottom: detailed view of the nanowire after a prolonged growth period. The zigzag trench between the (111) sidewalls is clearly visible, with turning points marked by numbers according to the images in panel b. The inset shows the morphology of the top facet (postgrowth image of the facet without the droplet). The scale bars are 100 nm. (b) Series of in situ SEM images (⟨110⟩ observation direction) outlining the cyclic droplet movement on top of such a kind of nanowire during growth (different nanowire than in panel a). The white arrows mark the direction in which the TPL will move in the next step. The schematics to the right illustrate the droplet shape in each step, with black and red arrows showing the catalyzed growth direction of the facet and the force applied to the TPL by the droplet, respectively.

evaporator crucible temperature and was calibrated before the experiment using the crystal quartz thickness monitor. We did not observe any effect of electron-beam irradiation on nanowire growth even if the beam current was increased above the usually used 50 pA. The reported nanowire shapes were observed on the whole sample surface.

In conclusion, the droplet motion on top of a growing nanowire presented here is driven by changes in the contact angle caused by geometrical constraints as the growth interface shrinks or widens. In an example shown here, the geometry of the wire is dictated by the preference of the (111)-oriented sidewalls. On the basis of real-time microscopic observations, we have excluded other effects that could possibly explain the oscillatory motion of the collector droplet on nonplanar growth inteface. The revealed mechanism allows us to grow straight nanowires despite droplet instability. It is reasonable to expect that this droplet behavior can explain the plethora of nanowire shapes observed in other nanowire systems where a geometrical constraint exists, and its understanding will allow us to design and fabricate a broader variety of nanowire morphologies than are available today.



ASSOCIATED CONTENT

S Supporting Information *

Postgrowth analysis of Ge nanowires deposited on a Ge(111) sample at 400 °C; details of the Ge 2p, Ge 3d, and O 1s peaks measured by X-ray photoelectron spectroscopy on the Ge(100) substrate covered with both ⟨111⟩- and ⟨110⟩-oriented Ge NWs; real-time SEM movie showing the oscillatory motion of the eutectic Au−Ge droplet on top of a growing nanowire; realtime SEM movie showing a droplet wetting a sidewall facet of a Ge wire; postgrowth SEM image of a ⟨111⟩-oriented nanowire grown on Ge(111); and real-time evolution of the ⟨111⟩oriented nanowire growth on Ge(111). This material is available free of charge via the Internet at http://pubs.acs.org.



METHODS We equipped a FEI Quanta 3D FEG with a high-temperature heating stage and germanium evaporator10 to allow for in situ experiments with nanowire growth. In the usual geometry, the incidence angle of germanium atoms (evaporated from a solid source) is 70°, and the electron beam is tilted by 52° to the sample normal. The manipulator is capable of tilting and rotating the sample in almost any direction, thus allowing the observation angle to be changed. The pressure in the microscope chamber during the experiment was between 7 × 10−5 and 9.5 × 10−5 Pa. Germanium substrate with either (100) or (111) orientations was dipped in a solution of 40 nm colloidal gold for 30 min and immediately transferred to the SEM. The temperature was raised to 400 °C, and the evaporation was done using a flux of 3−8 Å/min. The germanium flux is changed by changing the



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]; Phone +420 54114 2813. Author Contributions

M.K. conducted the simulations and designed and performed the experiments together with T.V. P.V. and T.S. supervised the project and co-wrote the paper. All authors discussed the results and reviewed the manuscript. Notes

The authors declare no competing financial interest. 1760

dx.doi.org/10.1021/nl404159x | Nano Lett. 2014, 14, 1756−1761

Nano Letters



Letter

(20) Xu, T.; Nys, J. P.; Addad, A.; Lebedev, O. I.; Urbieta, A.; Salhi, B.; Berthe, M.; Grandidier, B.; Stiévenard, D. Faceted sidewalls of silicon nanowires: Au-induced structural reconstructions and electronic properties. Phys. Rev. B 2010, 81, 115403. (21) Oliver, J. F.; Huh, C.; Mason, S. G. Resistance to spreading of liquids by sharp edges. J. Colloid Interface Sci. 1977, 59, 568. (22) Marmur, A. Solid-surface characterization by wetting. Annu. Rev. Mater. Res. 2009, 39, 473. (23) Brakke, K. The surface evolver. Exp. Math. 1992, 1, 141. (24) Wu, Y.; Cui, Y.; Huynh, L.; Barrelet, C. J.; Bell, D. C.; Lieber, C. M. Controlled growth and structures of molecular-scale silicon nanowires. Nano Lett. 2004, 4, 433. (25) Kyogoku, S.; Iwata, J.-I.; Oshiyama, A. Relation between nanomorphology and energy bands of Si nanowires. Phys. Rev. B 2013, 87, 165418. (26) Crawford, S.; Lim, S. K.; Gradečak, S. Fundamental insights into nanowire diameter modulation and the liquid/solid interface. Nano Lett. 2013, 13, 226. (27) Hannon, J. B.; Kodambaka, S.; Ross, F. M.; Tromp, R. M. The influence of the surface migration of gold on the growth of silicon nanowires. Nature 2006, 440, 69.

ACKNOWLEDGMENTS We acknowledge J. Č echal for critically reading the manuscript. This work was supported by the Grant Agency of the Czech Republic (P108/12/P699) and by the European Regional Development Fund (CEITEC, CZ.1.05/1.1.00/02.0068). M.K. acknowledges support from FEI Company.



REFERENCES

(1) Wagner, R. S.; Ellis, W. C. Vapor-liquid-solid mechanism of single crystal growth. Appl. Phys. Lett. 1964, 4, 89. (2) Wacaser, B. A.; Dick, K. A.; Johansson, J.; Borgström, M. T.; Deppert, K.; Samuelson, L. Preferential interface nucleation: An expansion of the VLS growth mechanism for nanowires. Adv. Mater. 2009, 21, 153. (3) Sutter, E.; Sutter, P. Phase diagram of nanoscale alloy particles used for vapor-liquid-solid growth of semiconductor nanowires. Nano Lett. 2008, 8, 411. (4) Hillerich, K.; Dick, K. A.; Wen, C.-Y.; Reuter, M. C.; Kodambaka, S.; Ross, F. M. Strategies to control morphology in hybrid group III− V/group IV heterostructure nanowires. Nano Lett. 2013, 13, 903. (5) Givargizov, E. I. Fundamental aspects of VLS growth. J. Cryst. Growth 1975, 31, 20. (6) Nebol’sin, V. A.; Shchechitin, A. A. Critical parameters of the vapor−liquid−solid growth of silicon whiskers. Inorg. Mater. 2003, 39, 899. (7) Gamalski, A. D.; Ducati, C.; Hoffmann, S. Cyclic supersaturation and triple phase boundary dynamics in germanium nanowire growth. J. Phys. Chem. C 2011, 115, 4413. (8) Wen, C.-Y.; Tersoff, J.; Hillerich, K.; Reuter, M. C.; Park, J. H.; Kodambaka, S.; Stach, E. A.; Ross, F. M. Periodically changing morphology of the growth interface in Si, Ge, and GaP nanowires. Phys. Rev. Lett. 2011, 107, 025503. (9) Schwarz, K. W.; Tersoff, J. Elementary processes in nanowire growth. Nano Lett. 2011, 11, 316. (10) Kolíbal, M.; Vystavěl, T.; Novák, L.; Mach, J.; Šikola, T. In-situ observation of ⟨110⟩ oriented Ge nanowire growth and associated collector droplet behavior. Appl. Phys. Lett. 2011, 99, 143113. (11) Hocevar, M.; Immink, G.; Verheijen, M.; Akopian, N.; Zwiller, V.; Kouwenhoven, L.; Bakkers, E. Growth and optical properties of axial hybrid III−V/silicon nanowires. Nat. Commun. 2012, 3, 1266. (12) Dayeh, S. A; Wang, J.; Li, N.; Huang, J. Y.; Gin, A. V.; Picraux, S. T. Growth, defect formation and morphology control of germaniumsilicon semiconductor nanowire heterostructures. Nano Lett. 2011, 11, 4200. (13) Oliveira, D. S.; Tizei, L. H. G.; Ugarte, D.; Cotta, M. A. Spontaneous periodic diameter oscillations in InP nanowires: The role of interface instabilities. Nano Lett. 2013, 13, 9. (14) Wen, C.-Y.; Reuter, M. C.; Tersoff, J.; Stach, E. A.; Ross, F. M. Structure, growth kinetics, and ledge flow during vapor-solid-solid growth of copper-catalyzed silicon nanowires. Nano Lett. 2010, 10, 514. (15) Ross, F. M.; Tersoff, J.; Reuter, M. C. Sawtooth faceting in silicon nanowires. Phys. Rev. Lett. 2005, 95, 146104. (16) Kolíbal, M.; Kalousek, R.; Vystavěl, T.; Novák, L.; Šikola, T. Controlled faceting in ⟨110⟩ germanium nanowire growth by switching between vapor-liquid-solid and vapor-solid-solid growth. Appl. Phys. Lett. 2012, 100, 203102. (17) Porret, C.; Devillers, T.; Jain, A.; Dujardin, R.; Barski, A. Importance of kinetic effects in the growth of germanium nanowires by vapour-liquid-solid molecular beam epitaxy. J. Cryst. Growth 2011, 323, 334. (18) Xu, T.; Sulerzycki, J.; Nys, J. P.; Patriarche, G.; Grandidier, B.; Stiévenard, D. Synthesis of long group IV semiconductor nanowires by molecular beam epitaxy. Nanoscale Res. Lett. 2011, 6, 113. (19) Dumbrovskii, V. G.; Xu, T.; Lambert, Y.; Nys, J. P.; Grandidier, B.; Stiévenard, D.; Chen, W.; Pareige, P. Narrowing the length distribution of Ge nanowires. Phys. Rev. Lett. 2012, 108, 105501. 1761

dx.doi.org/10.1021/nl404159x | Nano Lett. 2014, 14, 1756−1761