Dislocation-Guided Self-Running Droplets - Crystal Growth & Design

Dec 12, 2014 - The guiding effects are explained by surface stress which drives the mobile ... Win Eiwwongcharoen , Nitas Nakareseisoon , Supachok Tha...
0 downloads 0 Views 4MB Size
Communication pubs.acs.org/crystal

Dislocation-Guided Self-Running Droplets Songphol Kanjanachuchai*,† and Pat Photongkam‡ †

Semiconductor Device Research Laboratory, Department of Electrical Engineering, Faculty of Engineering, Chulalongkorn University, Bangkok 10330, Thailand ‡ Synchrotron Light Research Institute, 111 University Avenue, Muang District, Nakhon Ratchasima 30000, Thailand S Supporting Information *

ABSTRACT: Langmuir evaporation of stressed InSb and InAs (111)B crystals is studied using low-energy electron microscopy. The stress induces plastic relaxation by dislocation generation, whereas the evaporation results in self-running In droplets. The coexistence of in situ generated dislocations and self-running droplets allows droplet-dislocation interaction to be studied with minimal perturbation from other effects. The droplets run toward a preferred direction but are guided along dislocation lines before setting themself free once they accumulate mass and sufficient momentum to climb the energetic barrier posed by surface steps. The guiding effects are explained by surface stress which drives the mobile liquid along the dislocation line. This mechanism provides a means to control the direction of the self-running or self-propelled droplets, adding another degree of freedom to the growth and design of droplet-assisted micro- and nanostructures, for example, via interfacial dislocation network formed during heteroepitaxy.

L

tools; it requires bottom-up guiding of the recently discovered running droplet mechanism. The self-running metallic droplet phenomenon was first reported for Ga droplets on GaAs (001) in 2009,12 and for a limited number of metallic droplet/semiconductor substrates: Au/Si,13 Ga/GaAs,14,15 Ga/GaP,16 In/InAs,17 and In/InP18 with (001) and (111) the principal substrate orientations. These droplets are widely employed to form nanostructures: Au for NWs;1,8 Ga and In for QDs and NWs,3,4,7,8 particularly self-catalyzed III−V NWs.17 If the in-plane direction of running droplets can be arbitrarily controlled, the derived structures can form kinks, joints, and branches, and with simultaneous control of semiconducting host and dopant materials, a complex, droplet-derived optoelectronic devices and circuits compatible with planar processes could be realized. In addition, if electronic devices such as transistorswhich QDs and NWs arecan be guided laterally, it would help solve lithographic errorsa key issue affecting all in the semiconductor industry.19 Guiding liquid droplets is a fundamental fluid dynamic problem, made complicated by the reactive nature of metallic droplet/semiconductor interfaces20 where the only mode of control reported to date is in the initial positions of droplets.21 In this communication, we report an accidental discovery of a mechanism that can guide self-running droplets on III−V surfaces and may thus impact droplet-based micro- and

iquid, metallic droplets enable the realization of several key nanostructures and nanotechnologies. For example, they serve as catalysts at the heart of the vapor−liquid−solid (VLS) mechanism responsible for the growth of many semiconducting nanowires (NWs) such as Si, GaAs, and ZnO;1,2 they are also employed in epitaxy to crystallize quantum dots (QDs),3 in etching to form complex nanostructures4,5 and in plasmonics as the active medium.6 These droplet-based structures are often requiredfor ease of integration or for functionto be ordered or periodic, and since they are literally rooted from the nucleating droplets, their spatial distributions are dictated by those of the starting liquid droplets and, in many cases, by subsequent droplet motion which occurs either in plane, as in the crawling-mode growth of lateral NWs,7 or out of plane, as in the VLS growth of vertical NWs.8 To vertical NWs, droplet motion affects not only spatial distribution, but also other characteristics spanning the individual to the ensemble levels; for example, Schwarz and Tersoff showed that for individual NWs out-of-plane droplet motion affects the NW shape through liquid−solid contact angle during growth,9 and Dalacu et al. showed that for an ensemble of NWs out-of-plane droplet motion can be directed such that NWs interconnect.10 With proper process control, vertical NWs can thus be shaped and interconnected during growth, offering a great degree of freedom for wire- and circuit-level designs. But system-level integration and fabrication rely almost exclusively on planar processes where lateral NWs have an intrinsic advantage and are thus preferred.11 Integrating lateral NWs, however, is not as simple as connecting transistors using well-developed top-down © 2014 American Chemical Society

Received: September 11, 2014 Revised: November 22, 2014 Published: December 12, 2014 14

dx.doi.org/10.1021/cg5013704 | Cryst. Growth Des. 2015, 15, 14−19

Crystal Growth & Design

Communication

Figure 1. Self-running In droplets on InAs (111)B: (a) time series MEM images and (b) DIC image. Self-running In droplets on InSb (111)B: (c) time series MEM images and (d, e) DIC images. The times indicated at the bottom of the MEM images in (a) and (c) are in min:s from the leftmost frame. The droplets appear yellowish in the DIC images, and as dark circles surrounded by bright rings in MEM images.

microscopy (MEM) after which the filament current is slowly ramped down to zero. During cooling, the metallic nature of the droplets is confirmed by ultraviolet photoemission electron microscopy. After sample removal, the metallic droplets are confirmed to constitute mainly of indium by energy-dispersive X-ray spectroscopy in a scanning electron microscope (JEOL JSM-6400).18 The running In droplets are reactive (etch the surface), leaving shallow etch trails that are vaguely visible when viewed using an optical microscope (Nikon’s Eclipse ME600) in the standard, reflective geometry but are clearly visible with differential interference contrast (DIC) enhancement using a Nomarski prism. Surface morphologies are also probed by white light interferometry (WLI, Polytec’s MSA-400) in the phase-shift mode with topographic resolution better than 0.2 nm. The sublimation of a III−V surface in a vacuum at temperatures higher than congruent temperature TC generally results in a greater loss of group V element and a nonstoichiometric, group III-rich condition that yields randomly distributed III droplets on the surface.27 We observe the nucleation of In droplets on sublimating InAs and InSb (111)B at approximately the same TC of 360 °C. All running events are observed at temperatures between TC and TC + 30 °C. Figure 1 shows the morphologies of InAs and InSb (111)B surfaces during and after the evaporation. The droplets and trails are evident on both surfaces. For InAs (111)B, the droplets nucleate on a flat surface and run toward a preferred direction as shown in the MEM image series in Figure 1a, and the DIC image in Figure 1b. Droplet direction is unknown during in situ MEM imaging; it is inferred from wafer flats during ex situ DIC imaging that all droplets run toward the [1̅21̅] direction. The MEM image series in Figure 1a illustrates the main characteristics of the In droplets dynamics on InAs (111)B. The first-generation or mother droplets (marked 1 in Figure 1a) run toward the [1̅21̅] direction, leaving behind etch trails whose rough edges serve as preferential nucleation centers for second-generation or daughter droplets (2 and 3 in Figure 1a). Subsequently, the daughter droplets run and try to catch up with but avoid the trails of the mother

nanostructures by providing a method to control droplet position in situ. We demonstrate the directional, guided motion of running In droplets on sublimating InAs and InSb (111)B or (11̅ 1̅ )̅ surfaces. InAs and InSb are low-bandgap, zincblende semiconductors with long-wavelength and mid-infrared optoelectronic applications, respectively. 22 And the (111)B orientation is widely employed in the growth of III−V NWs.7,17 The running droplets are guided by buried dislocations. Dislocations are generally undesirable because their presence is often detrimental to a material’s electrical and optical properties,23 but they enable preferential nucleation of the otherwise randomly distributed nanostructures. This is particularly useful when nonlithographic but ordered structures are required. Examples of dislocation-guided nanostructures include self-assembled QDs grown by molecular beam epitaxy24,25 and NWs grown electrochemically.26 Through surface strain fields, dislocations affect adatoms motion during deposition. This communication establishes the equivalent effects on the reverse process: dislocation strain fields affect surface atoms motion during sublimation into vacuum or Langmuir evaporation, too. The In droplets formed from noncongruent evaporation of InAs and InSb (111)B here will be shown to run in specific directions and be guided by nanoscale surface steps arising from buried line dislocations. It is found that droplet size and step size are the main factors that determine the effectiveness and limits of the guiding mechanism. These insights can ultimately lead to complex and interconnected micro- and nanostructures that are grown, not fabricated. The sublimation of epi-ready InAs and InSb (111)B surfaces are carried out under ultrahigh vacuum (UHV) conditions in Elmitec’s low-energy electron microscope (LEEM III). The system’s base pressure is 2 × 10−10 Torr, and the maximum pressure during sublimation is 6 × 10−9 Torr. After in situ surface preparation, each sample is heated by a backside filament until the native oxide is desorbed and noncongruent evaporation is reached following our previous procedures.18 The nucleation of droplets and the subsequent dynamics are observed and recorded in real time using mirror electron 15

dx.doi.org/10.1021/cg5013704 | Cryst. Growth Des. 2015, 15, 14−19

Crystal Growth & Design

Communication

Figure 2. (a) DIC image of a sublimated InAs (111)B surface. Selected dislocation lines are labeled L1, L2, L3, and L4. The boxed regions i and ii are shown magnified in (b) and (c), respectively. In (b) the triangular schematic traces out the etch pit where the droplet emerges. In (c) the droplet D1 is being guided and D2 has been guided by the dislocation line L1. The topography of the boxed region iii in (a) is measured by white light interferometry and shown in (d) with a height contrast of 40 nm. DIC images of other sublimated samples: (e) InAs (111)B, (g) InSb (111)B. (f) Schematic drawing of the trajectories of droplets F1, F2, and F3 as guided by line dislocations in the [1̅10], [01̅1], and [101̅] directions, respectively. The directions and planes in (f) apply to all images as all samples are oriented such that the [01̅1] direction points up.

droplet density in the vicinity. If sufficiently steep, the temperature gradient trumps the chemical potential gradient and causes a large contact angle difference between the hot and cold regions.29 The degree of temperature gradient is unknown, but the cause is likely due to molten In globules on the backside, facing the hot filament thus partially reflecting the radiation. Figure 1d,e thus represents droplets driven principally by chemical potential and temperature gradients, respectively. But regardless of the dominant driving force, i.e., the direction, running droplets always leave triangular etch trails, a remnant of a stick−slip motion previously observed for running Ga droplets on GaAs (111)A.15 Returning to InAs (111)B, we found an unfortunate sample with a corner broken during sublimation. The majority of the droplets near the corner run toward the [1̅21̅] direction, but surprisingly a small minority are diverted by lines of uncertain origin cutting across their paths. While removing the sample from the holder, however, we noticed that the screw on the broken corner was much more tightened than the other three corners and realized that this must have inadvertently introduced stress during sublimation. Since stressing crystals at high temperatures is known to induce dislocations,30 these droplets-guiding lines was suspectedand later confirmedto have resulted from dislocations. By repeating the sublimation on different samples and overtightening the screws on various corners to varying degrees, we found that the lines created always occur in the ⟨110⟩ directionsthe principal dislocation directions in zincblende semiconductors.30 Figure 2a shows a DIC image of a sublimated InAs (111)B surface close to a broken corner;

droplets: Droplet 2 avoids the trail of droplet 1, while droplet 3 avoids the trails of droplets 1 and 2. The majority of droplets seen in Figure 1b are mother droplets. These droplets run under competing forces that derive from chemical potential μ,12 atomic scale restructring,16 and surface reconstruction.28 Self-running In droplets on InAs (111)B have previously been reported by Mandl et al., who observed running droplets with no preferential direction.17 In contrast, our results in Figure 1a,b unambiguously show a directional motion. The difference most likely stems from sample preparation, particularly the heating rate: the directional In droplets here are obtained from very slow annealing, whereas the nondirectional In droplets reported by Mandl et al. are formed by rapid thermal annealinga violent processand the droplets probably catapult rather than crawl out of the nucleating sites, resulting in random running directions. Apart from InAs (111)B, self-running In droplets have also been observed on InP (001), (111)A, and (111)B, but the running characteristics are multidirectional as a result of In droplets breakup,18 a feature not observed on InAs (111)B here. For InSb (111)B, most In droplets nucleate in triangular thermal etch pits, grow with time, and crawl out of the pits as seen in the MEM image series in Figure 1c. Once released from an etch pit, and depending on the nucleation site, an In droplet may move forward, sideway, or backward before turning toward the [1̅21̅] direction as seen in Figure 1d. Most of the 1 cm2 sample area exhibits similar characteristics, except for a small area where droplets instead run toward the equivalent [21̅1̅] direction as seen in Figure 1e. The latter is an anomaly caused by a temperature gradient as evident from the variation in 16

dx.doi.org/10.1021/cg5013704 | Cryst. Growth Des. 2015, 15, 14−19

Crystal Growth & Design

Communication

Figure 3. (a) Schematic drawing of a droplet running toward a line dislocation (⊥) buried at a distance h from the surface. The symbol ⊥ is drawn rotated to reflect the 70.5° angle between the {111} slip planes and the (111) surface. In addition to the global force Fμ, the droplet experiences location-specific surface stresses normal σyy and parallel σzz to the line dislocation. Simulated cross-sectional stress distributions (b) σyy and (c) σzz in the x−y plane for a semi-infinite dislocation at y = 0 and x = −1 μm, and (d) simulated surface stress distributions σyy0 (solid lines) and σzz0 (dashed lines) above a dislocation buried at h = 0.5 μm (lower lines) and 1 μm (upper) from the surface. The upper and lower lines in (d) are offset for clarity. The stress values in (b−d) are in units of 10−3G where G is the shear modulus of the host material.

the boxed regions i and ii are shown magnified in Figure 2b,c to illustrate a triangular pit and dislocation-guided droplets, respectively. Though most droplets nucleate on flat areas, some nucleate in triangular pits. The droplet in the boxed region i in Figure 2b, for example, originates from a triangular pit whose sides are bounded by the {111} glide planes in the ⟨110⟩ directions,31 consistent with the dislocation etch pits on the (111) surfaces of many zincblende and cubic crystals.30 The pit thus serves as an internal direction reference. The white triangular schematic outlining the pit in Figure 2b perfectly matches the dislocation lines intersection L1−L2 in Figure 2a and, see later, Figure 2d,e. Not all dislocation lines are created equal. Lines L1 and L2 are highly effective in guiding droplets. Guiding effects can be total (throughout a droplet’s entire existence) or partial (only momentarily). Droplet D1 in Figure 2c is totally guided by L1 since D1 nucleates near L1 and runs under the influence of L1 right from the beginning; its trail indicates that it runs in the [101]̅ direction, as opposed to the majority of droplets which run in the [1̅21̅] direction. Droplet D2 is partially guided by L1 since D2 nucleates further away and thus initially runs toward the majority direction before coming under the influence of L1, i.e., be guided by L1 toward the [101]̅ direction. The trail of droplet D2 in Figure 2c indicates that D2 is guided by L1 but later frees itself, continuing its journey along the majority direction. This change of direction is possible because reactive droplets etch the surface and grow as they run.12 After a critical mass or momentum is reached, the droplet can climb the energetic barrier (see later) and cross over the dislocation to the other side. The strong guiding effects of L1 are an exception, rather than the rule. Most other dislocation lines in Figure 2a are elusive and cannot guide droplets. This is evident from the topographic image in Figure 2d, obtained from WLI of the boxed region iii of Figure 2a. For example, droplet D3 runs through L3 and L4 almost unaffected. Both L3 and L4 are not clearly visible in the DIC image in Figure 2a and are only slightly discernible in the WLI image in Figure 2d. This is because the surface steps associated with the dislocation lines L3 and L4 are not sufficiently tall for the Nomarski prism to resolve. The step heights of L1, L2, L3, and L4 can however be measured from the WLI image and are found to be 7.5, 6.0, 3.1, and 4.5 nm, respectively. Since L1 and L2 do guide the droplets while L3

and L4 do not, there seems to exist a critical step height over which dislocations may guide the droplets. This critical step height lies between 4.5 and 6.0 nm, or between 7 and 10 times the InAs lattice parameter. Provided the related step heights exceed the critical step height, dislocation lines are similarly capable of guiding droplets regardless of line direction. The InAs (111)B sample used to acquire Figure 2a,d is dominated by [101̅] dislocation lines, e.g., L1 and L3, with only a few [01̅1] dislocation lines, e.g., L2. Detailed droplet−dislocation interactions can thus only be observed for [101]̅ dislocation lines. In another InAs (111)B sample, overtightened to the holder differently but otherwise subject to the same annealing treatments, [01̅1] and [1̅10] dislocation lines dominate as shown in Figure 2e. This and the results in Figure 2a−d allow basic droplet−dislocation interactions for the three line dislocation types to be concluded as schematized in Figure 2f. Basically, droplets running in the majority, [1̅21̅] direction are guided along dislocation lines before ultimately freeing themselves as shown by droplets F1, F2, and F3 upon chance meetings with [1̅10], [01̅1], and [101̅] dislocations, respectively. At dislocation intersections, droplet movements are more complicated but can readily be explained by these three basic motion patterns (see Supporting Information). Similar attempts were made to form dislocation lines on InSb (111)B, but the created lines are less well-defined, i.e., not limited to ⟨110⟩ directions, possibly due to competition among several slip systems32 during oxide desorption which occurs close to the InSb melting point.33 Nevertheless, running In droplets on InSb (111)B are similarly guided by line dislocations as shown in Figure 2g, indicating that the forces guiding the droplets on the InAs and InSb (111)B surfaces share the same origins. The origins and limits of the dislocation-guided self-running droplet mechanism can be qualitatively explained by two forces acting on a droplet running toward a dislocation as schematically shown in Figure 3a. First, the droplet is driven toward the [1̅21̅] direction by a global force Fμ due principally to chemical potential and temperature gradients.12,29 This force depends linearly on diameter12 and thus increases as droplets run and gain mass. Second, the droplet experiences a local force Fσ as a result of surface stress σ due to buried dislocations.34 Fσ is only appreciable if the droplet is close to a dislocation as the stress components perpendicular σyy and parallel σzz to a dislocation 17

dx.doi.org/10.1021/cg5013704 | Cryst. Growth Des. 2015, 15, 14−19

Crystal Growth & Design

Communication

line decreases rapidly with distance. Fσ gives rise to surface steps and, in combination with Fμ, dictates the degree of dislocation−droplet interaction and the running behaviors. The presence of dislocationhence, Fσgives rise to surface steps due to the asymmetric nature of surface stress fields. To illustrate this, the stress components σyy and σzz of a hypothetical, semi-infinite line dislocation (symbol ⊥) at y = 0 and at a depth h from the surface (x = −h = −1 μm) as shown in the cross-sectional schematic in Figure 3a are computed using a modified Eshelby technique following Andrews et al.34 and shown in Figure 3, panels b and c, respectively (see Supporting Information). The numerical values are in normalized units: 10−3G where G is the shear modulus. The areas under compression are in blue, tension in red. The surface (x = 0) above a dislocation line is thus always under compression (σ < 0) on one side (y > 0) and tension (σ > 0) on the other (y < 0) as shown in the surface stress plots in Figure 3d. Surface stresses drive surface mass transport.35 During sublimation and deposition, surface In atoms are thus driven away from the dislocation on one side and toward it on the other, creating at y = 0 a chemical potential difference Δμ and a topographic step Δx. The degree of dislocation−droplet interaction depends on the surface step Δx which in turns depends on the depth of the buried dislocation h. Generally, the closer a dislocation is to the surface, the greater the step, and the stronger the interaction. Figure 3d compares and contrasts the surface stresses arising from two hypothetical, semi-infinite line dislocations buried at 1 and 0.5 μm from the surface. In both cases, the stresses are greater around the immediate vicinity of the dislocations (around y = 0) but are smaller for areas further away. The shallow dislocation (h = 0.5 μm) results in a greater maximum stress and has a shorter range than the deep dislocation (h = 1 μm). Reducing h thus increases surface stresses σyy and σzz (and the corresponding forces), but at the same time decreases the range of influence. Shallow dislocation lines (e.g., L1 and L2 in Figure 2) are thus more clearly visible and interact with droplets more strongly than deep dislocation lines (L3 and L4). All observed running droplet behaviors can be qualitatively explained by the combining effects of Fμ and Fσ. For areas far away from dislocations, only Fμ is present and the droplets run toward the majority direction. Close to dislocations (|y| < ∼2h3h), Fσ is significant, and the droplets run under the vector sum of Fμ and Fσ. Since Fμ depends on droplet size and Fσ on dislocation depth h, droplet−dislocation interactions exhibit four types of behaviors: (1) Small droplets (low Fμ) running toward shallow dislocations (small h, high Δμ) are fully guided along the dislocations by virtue of σzz since Fμ is low and σyy is insufficient to push droplets cross the steep potential barrier Δμ. (2) Small droplets running toward deep dislocations (large h, low Δμ) may be diverted slightly or cross over the dislocation since Δx and Δμ are low. (3) Large droplets running toward deep dislocations will cross over, almost unimpeded, since Fμ ≫ Fσ. And (4) Large droplets running toward shallow dislocations maybe guided as in case 1 or cross over as in case 3. For case 1, droplets are initially guided but will ultimately be freed, unless they are diverted or blocked by the etch trails and/or dislocation, particularly at intersections (see Supporting Information). If allowed to run unhindered, all guided droplets can ultimately overcome the potential barrier Δμ and cross over. This poses a limit for which the guiding mechanism is useful if employed as a spatial control mechanism. A just-released droplet can however be guided

again and again by surface steps with successively greater energetic barriers, in reverse analogy to a cascade waterfall where water droplets stream down multiple rock steps. The dislocation-guided running droplet mechanism may exist, but have been overlooked, in other material systems. It is certainly not unique to In droplets: Ga droplets running on GaAs (111)A,15 for example, have been suspected to be partially guided by dislocation linesalong the same dislocation-prone, ⟨110⟩ directions reported here. There is one critical difference, however. The guided Ga droplets on GaAs (111)A do not cross over the dislocation lines, nor are they blocked by dislocation intersections: They run along straight segments and simply take sharp turns at dislocation intersections. This strong dislocation-guided running behavior is consistent with the description in Figure 3a under the condition Fσ ≫ Fμ. We believe the dislocation-guided droplet mechanism, and the strictly qualitative model in Figure 3, is general and should apply to other droplets systems, such as Au on Si,13 or self-propelled Leidenfrost droplets on asymmetric topography.36 The observed droplet−dislocation interaction can be interpreted as a macroscopic manifestation of the droplet-step interaction reported37 for submicron Pt−Si droplets migrating on Si (100) surfaces with atomic steps and step bunches. Sutter et al.37 found that the interaction between Pt−Si alloy droplets and atomic steps depends on (i) droplet size r: small droplets (r < rc where rc is a critical radius) are pinned or immobilized, intermediate droplets (r ≈ rc) are guided, and large droplets (r > rc) run through perpendicular steps; (ii) angle: even large droplets can be guided at shallow angles as a result of step drags canceling out the perpendicular velocity components; and (iii) step size: N-step bunches can guide larger droplets as a result of greater pinning force which is proportional to N. We speculate the droplet−dislocation interaction mechanism to be useful in two main applications. First, as a contrast enhancement technique in dislocation studies. The trails of the running droplets decorate the dislocation lines, making them visible under an optical microscope and thus providing a much quicker and broader information than transmission electron microscopy. Dislocation decoration is an important field of studies in its own right, giving insights to the growth and related processes of crystalline solids.38 Second, as a process tool to shape, direct, and interconnect droplet-assisted microand nanostructures, for example, in combination with latticemismatched heteroepitaxy and with techniques such as chemical- and ion-beam (such as Ga+)39,40 depositions. The second application is however more complicated as certain characteristics of the mechanism remain unresolved. For example, dislocation lines are dictated principally by crystallography, making arbitrary directional control difficult, or even impossible. Also, and even in the absence of dislocations, it is not yet clear what dictates the different running directions hitherto reported for the various III droplets on III−V surfaces,12−17 or why the same type of droplets exhibits starkly different dynamics on similar surfaces. In droplets, for example, break up before they run on InP,18 but run without breakup on InAs (111)B as observed here, or do not run at all on InAs (001). Despite incomplete understanding of the dislocationguided self-running droplet mechanism, and of reactive wetting in general,20 the rich dislocation-droplet dynamics uncovered here offer a simple method of in situ droplet motion control which can extend the complexity and functionality of the many droplet-based micro- and nanoscale structures and systems. 18

dx.doi.org/10.1021/cg5013704 | Cryst. Growth Des. 2015, 15, 14−19

Crystal Growth & Design

Communication

(10) Dalacu, D.; Kam, A.; Austing, D. G.; Poole, P. J. Nano Lett. 2013, 13, 2676−2681. (11) Hobbs, R. G.; Petkov, N.; Holmes, J. D. Chem. Mater. 2012, 24, 1975−1991. (12) Tersoff, J.; Jesson, D. E.; Tang, W. X. Science 2009, 324, 236− 238. (13) Shao, Y. M.; Nie, T. X.; Jiang, Z. M.; Zou, J. Appl. Phys. Lett. 2012, 101, 053104. (14) Wu, J.; Wang, Z. M.; Li, A. Z.; Benamara, M.; Salamo, G. J. ACS Appl. Mater. Interfaces 2011, 3, 1817−1820. (15) Kanjanachuchai, S.; Euaruksakul, C. ACS Appl. Mater. Interfaces 2013, 5, 7709−7713. (16) Hilner, E.; Zakharov, A. A.; Schulte, K.; Kratzer, P.; Andersen, J. N.; Lundgren, E.; Mikkelsen, A. Nano Lett. 2009, 9, 2710−2714. (17) Mandl, B.; Stangl, J.; Hilner, E.; Zakharov, A. A.; Hillerich, K.; Dey, A. W.; Samuelson, L.; Bauer, G. N.; Deppert, K.; Mikkelsen, A. Nano Lett. 2010, 10, 4443−4449. (18) Kanjanachuchai, S.; Euaruksakul, C. Cryst. Growth Des. 2014, 14, 830−834. (19) The International Technology Roadmap for Semiconductors (ITRS), Lithography, 2013, http://www.itrs.net/. (20) Kumar, G.; Prabhu, K. N. Adv. Colloid Interface Sci. 2007, 133, 61−89. (21) Tersoff, J.; Jesson, D. E.; Tang, W. X. Phys. Rev. Lett. 2010, 105, 035702. (22) Vurgaftman, I.; Meyer, J. R.; Ram-Mohan, L. R. J. Appl. Phys. 2001, 89, 5815−5875. (23) Holt, D. B.; Yacobi, B. G. Extended Defects in Semiconductors: Electronic Properties, Device Effects and Structures; Cambridge University Press: Cambridge, 2007; Chapter 5. (24) Shiryaev, S. Y.; Jensen, F.; Hansen, J. L.; Petersen, J. W.; Larsen, A. N. Phys. Rev. Lett. 1997, 78, 503−506. (25) Xie, Y. H.; Samavedam, S. B.; Bulsara, M.; Langdo, T. A.; Fitzgerald, E. A. Appl. Phys. Lett. 1997, 71, 3567−3568. (26) Huang, X.; Chumlyakov, Y. I.; Ramirez, A. G. Nanotechnology 2012, 23, 125601. (27) Chatillon, C.; Chatain, D. J. Cryst. Growth 1995, 151, 91−101. (28) Datta, S. S. J. Appl. Phys. 2010, 108, 024307. (29) Brochard, F. Langmuir 1989, 5, 432−438. (30) Abrahams, M. S.; Buiocchi, C. J. J. Appl. Phys. 1965, 36, 2855− 2863. (31) Hulme, K. F.; Mullin, J. B. Solid-State Electron. 1962, 5, 211− 247. (32) Peissker, E.; Haasen, P.; Alexander, H. Philos. Mag. 1962, 7, 1279−1303. (33) Tereshchenko, O. E. Appl. Surf. Sci. 2006, 252, 7684−7690. (34) Andrews, A. M.; LeSar, R.; Kerner, M. A.; Speck, J. S.; Romanov, A. E.; Kolesnikova, A. L.; Bobeth, M.; Pompe, W. J. Appl. Phys. 2004, 95, 6032−6047. (35) Freund, L. B.; Suresh, S. Thin Film Materials: Stress, Defect Formation and Surface Evolution; Cambridge University Press: Cambridge, 2003; Chapter 9. (36) Linke, H.; Alemán, B. J.; Melling, L. D.; Taormina, M. J.; Francis, M. J.; Dow-Hygelund, C. C.; Narayanan, V.; Taylor, R. P.; Stout, A. Phys. Rev. Lett. 2006, 96, 154502. (37) Sutter, P.; Bennett, P. A.; Flege, J. I.; Sutter, E. Phys. Rev. Lett. 2007, 99, 125504. (38) Amelinckx, S. In The Direct Observation of Dislocations; Seitz, F., Turnbull, D., Eds.; Solid State Physics: Advances in Research and Applications, Supplement 6; Academic Press: New York, 1964. (39) Kang, M.; Wu, J. H.; Sofferman, D. L.; Beskin, I.; Chen, H. Y.; Thornton, K.; Goldman, R. S. Appl. Phys. Lett. 2013, 103, 072115. (40) Xu, X.; Wu, J.; Wang, X.; Li, H.; Zhou, Z.; Wang, Z. M. Appl. Phys. Lett. 2014, 104, 133104. (41) Russell, P. E.; Stark, T. J.; Griffis, D. P.; Phillips, J. R.; Jarausch, K. F. J. Vac. Sci. Technol. B 1998, 16, 2494−2498. (42) Bei, H.; George, E. P.; Hay, J. L.; Pharr, G. M. Phys. Rev. Lett. 2005, 95, 045501.

In summary, the sublimation of InAs and InSb (111)B surfaces under UHV conditions are studied in situ. Self-running In droplets are observed in both cases, and so are line dislocations created during sublimation as a result of mechanical stress. Some dislocations interact strongly with droplets, some weakly, and some negligibly. The origin of the interaction can be explained by dislocation-related surface stress, while the degree of interaction depends mainly on droplet size, the size of dislocation-induced surface steps, and the size of stress-driven forces relative to the forces of other origins, particularly those derived from chemical potential and temperature gradients. The rich droplet−dislocation dynamics revealed in this work provides an important foundation for further exploitation of droplet-assisted micro- and nanostructures via strain engineering, through the use of templates24,25 or direct writing. For the latter, one can envisage a nanoindenter with a desired tip geometry to precisely fashion a stress distribution41,42 on a growing or sublimating surface in order to deterministically induce or craft dislocation lines.



ASSOCIATED CONTENT

S Supporting Information *

Droplets at dislocation intersections. Stress fields of a buried dislocation. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Tel: +66-81-646-3496. E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work is supported by Thailand’s National Research University Project, Office of the Higher Education Commission (WCU-036-EN-57). We are grateful for access to Polytec’s micro system analyzer (MSA-400) at Nectec’s Nanoelectronics and MEMS Laboratory, and related training by A. Wisitsoraat. We also acknowledge T. Chokamnuai, W. Eiwwongcharoen, N. Jearanaikoon, N. Nakareseisoon, and B. A. Trisna for assistance during MEM imaging.



ABBREVIATIONS DIC, differential interference contrast; LEEM, low-energy electron microscopy; MEM, mirror electron microscopy; NW, nanowire; QD, quantum dots; UHV, ultrahigh vacuum; VLS, vapor−liquid−solid; WLI, white light interferometry



REFERENCES

(1) Duan, X.; Lieber, C. M. Adv. Mater. 2000, 12, 298−302. (2) Dick, K. A. Prog. Cryst. Growth Charact. Mater. 2008, 54, 138− 173. (3) Koguchi, N.; Takahashi, S.; Chikyow, T. J. Cryst. Growth 1991, 111, 688−692. (4) Heyn, Ch. Phys. Rev. B 2011, 83, 165302. (5) Li, X.; Wu, J.; Wang, Z. M.; Liang, B.; Lee, J.; Kim, E.-S.; Salamo, G. J. Nanoscale 2014, 6, 2675−2681. (6) Blaber, M. G.; Engel, C. J.; Vivekchand, S. R. C.; Lubin, S. M.; Odom, T. W.; Schatz, G. C. Nano Lett. 2012, 12, 5275−5280. (7) Fortuna, S. A.; Li, X. Semicond. Sci. Technol. 2010, 25, 024005. (8) Fan, H. J.; Werner, P.; Zacharias, M. Small 2006, 2, 700−717. (9) Schwarz, K. W.; Tersoff, J. Phys. Rev. Lett. 2009, 102, 206101. 19

dx.doi.org/10.1021/cg5013704 | Cryst. Growth Des. 2015, 15, 14−19