Role of Surface Polarity in Self-Catalyzed Nucleation and Evolution of

Article pubs.acs.org/crystal

Role of Surface Polarity in Self-Catalyzed Nucleation and Evolution of GaN Nanostructures Prasana Sahoo,*,† S. Dhara,*,† S. Amirthapandian,‡ M. Kamruddin,† S. Dash,† B. K. Panigrahi,‡ and A. K. Tyagi† †

Surface and Nanoscience Division and ‡Materials Physics Division, Indira Gandhi Center for Atomic Research, Kalpakkam-603102, India S Supporting Information *

ABSTRACT: Self-catalytic growth of GaN nanotips and nanoparticles, grown by chemical vapor deposition technique, are investigated. Three important parameters, comprised of incubation time, anisotropy of diffusion, and rate-limiting factors of Ga and N adatoms migration over polar and nonpolar surfaces, are found to play significant roles in determining the final morphology of these nanostructures. Nucleation of GaN nanotips takes place under Ga-rich conditions. As the reaction proceeds, the stochiometry changes occur as a result of a shift in Ga-rich to N-rich conditions on the surface. In all of these cases, the growth continues to be in vapor−solid mode. The conical shape of the nanotips is explained in terms of differential growth in the reduced surface diffusion of Ga under N-rich conditions on polar surfaces (0001) relative to nonpolar surfaces (101̅0). Nanoparticles are grown initially in N-rich conditions with significantly shorter incubation times. A mechanistic approach that expounds evolution of nanotips and nanoparticles is elucidated in details using crystallographic and electronic structural studies.

O

modes significantly influence the properties of resulting nanostructures, which include shape, size, and crystal structure. Apart from nanowires, nanorods, and nanotubes, there are considerable interests in exploring various forms of GaN nanostructures, such as nanotips14 and nanoparticles (NPs).15 The aim of this paper is to cast light on the evolutionary stages of nucleation-mediated morphology of nanotips and NPs. Wide band gap nitride-based nanotips with low work function are known to be hard and chemically stable. These qualities make it potentially useful for applications in the field emission devices and scanning near field optical microscopy.1,16 Until now, electron beam or focused-ion-beam (FIB) lithography and dry etching techniques, employing a SiO2 nanomask and subsequent reactive ion etching, are widely used for large area growth of aligned nanotips.17 The processes are limited to specific substrates only. The synthesis of wellaligned GaN nanotips in a self-catalytic process has raised sufficient scientific excitement.18 We describe here a single-step self-catalytic nucleation and growth process involving the formation of GaN nanostructures of nanotips and NPs using a CVD technique in low scale (Supporting Information, Figure S1). This process is reproducible. In this technique, growth morphology is easily controlled by varying Ga and Na flux rates during early nucleation. Structural, vibrational, and optical properties are

ne-dimensional (1D) materials have attracted the attention of researchers due to their interesting physical properties and potential applications.1 In this context, GaN is considered as a material of choice for the development of highperformance optoelectronic devices. It has recently gained much attention because of the capability to promote hydrogen production by water splitting.2 Until now, several prototype devices already have been demonstrated. However, large-scale applications of these devices are limited because of the cost and effort involved in developing high pure-aligned GaN nanostructures. The vapor−liquid−solid (VLS) process has been well established and widely exploited in demonstrating controlled growth of 1D nanostructures. Nevertheless, the vapor−solid (VS) process excludes inclusion of foreign atoms, which are ubiquitous in the catalytic VLS process.3 In this context, efforts are being made worldwide to grow uniformly distributed 1D GaN nanostructures in a self-catalyzed process via molecular beam epitaxy using a template approach along with controlled Ga and N flux rates.4−10 It is based on the fact that large anisotropy with respect to Ga and N adatom diffusion along the polar and nonpolar surfaces leads to the formation of 1D nanowires and nanorods. However, the underlying mechanistic concepts involved in uniaxial self-catalyzed growth of 1D GaN nanostructures are still in infancy.8−11 As a matter of fact, reports on self-catalyzed chemical-vapor-deposition (CVD) grown GaN nanostructures with controlled morphology are limited because of the difficulties encountered in controlling important parameters such as Ga and N flux rates and subsequent switching between the fluxes.12,13 Growth © 2012 American Chemical Society

Received: January 11, 2012 Revised: March 17, 2012 Published: March 21, 2012 2375

dx.doi.org/10.1021/cg300037q | Cryst. Growth Des. 2012, 12, 2375−2381

Crystal Growth & Design

Article

used to substantiate the nucleation process. Distinct mechanisms are elucidated to describe the evolution of nanotip and NP morphologies. Nanotips and NPs, in the present report, were synthesized using a hot-walled CVD reactor.19 For the growth of ultralong nanotips, Ga (99.999%, Aldrich) nodules (∼0.5 g) were initially annealed inside a high pure alumina boat (99.95%) at 500 °C for 10 min with a base pressure of 200 mbar in a tubular furnace. The temperature of the chamber was slowly increased to 900 °C with a ramping rate of ∼15 °C per min maintaining the same base pressure. After the chamber temperature reached 900 °C, the base pressure was slowly increased to 1 atm pressure in 15 min by bleeding high pure NH3 gas at a flow rate of 10 sccm. The reaction was terminated after a duration of 1 h. The system was allowed to cool slowly to room temperature (RT) in the same NH3 ambient. A large quantity (∼10 mg per 1 h growth) of free-standing nanotips were found on the alumina boat around the source material. The reaction time was varied from 1 to 3 h to obtain a substantial amount of nanotips while keeping all other parameters constant. GaN nanotips were further annealed at 800 °C for 1 h in continuous NH3 gas flow for improving the crystalline quality. The reaction parameters, namely, growth rate, temperature, growth time, and NH3 partial pressure, were varied independently to identify the growth mechanism of nanotips. However, the growth of nanotips was observed at an optimum growth temperature of 900 °C and NH3 flow rate of 10 sccm. GaN NPs were also prepared using the same CVD technique. Similar initial conditions were maintained for nanotip growth. In this case, Ga nodules (0.5 g) were annealed at 300 °C for 10 min at a base pressure of 200 mbar. Subsequently, the temperature of the growth chamber was increased to 900 °C with a ramp rate of 40 °C per min, which was approximately three times more than that used for the nanotip growth. Liquid Ga was vaporized above 800 °C to be deposited on the boat at 200 mbar chamber pressure. Subsequently, 10 sccm of NH3 was bled after the temperature reached 850 °C in 3 min time. The chamber pressure of 200 mbar was not changed substantially during the growth process. Then, the chamber pressure was slowly raised to atmospheric pressure in 10 min at a constant flow of NH3. The growth continued for 1 h at 900 °C. During this process, small powderlike GaN material with a light yellowish color was formed around the source material within the alumina boat. The growth temperature was varied from 880 to 950 °C to study its effect on NP morphology. Morphological features of the nanotips and NPs were examined by optical and field emission scanning electron microscopies (FESEM; model Zeiss SUPRA 40). Highresolution transmission electron microscopy (HRTEM; Libra 200 Zeiss), micro-Raman, and microphotoluminescence (micro-PL) studies were performed for structural characterization. Micro-Raman and micro-PL studies were conducted using 514.5 and 325 nm excitations with 1800 and 2400 g/mm gratings, respectively, in the backscattering configuration of a spectrometer (inVia Renishaw, United Kingdom). A thermoelectric-cooled back-thinned CCD detector was used for acquiring signal in the range of 200−900 nm (∼1.8−3.8 eV). Typical optical micrographs of as synthesized, free-standing nanotips grown over 1 and 3 h durations are shown in Figure 1a,b. Fine branches of nanotips are also observed in the case of growth carried out for 3 h (Figure 1b). The high-yield, freestanding nanotips grow over a large area around the Ga source

Figure 1. Optical image of the large-scale nanotips grown for (a) 1 h, (b) 3 h, and (c) typical FESEM image of the 3 h grown nanotips. The inset in panel c shows the transition of the Ga-rich base region to the N-rich smooth conical region and a lateral cross-section of a broken nanotip.

kept inside the alumina boat as a substrate. Figure 1c shows typical FESEM images of a 3 h grown single nanotip. The inset shows the magnified image of the base region that depicts a smooth transition from columnar to conical morphology. Most of the conical nanotips are solid and homogeneous, as shown in a typical cross-sectional image (inset of Figure 1c).The samples grown over 1 h duration possess a base diameter of ∼20 μm, tip diameter of 200 nm (Supporting Information, Figure S2), and length extending up to hundreds of micrometers. The aspect ratio (length/base diameter) is of the order of 20, which is 1 order of magnitude less than the sample grown over 3 h of duration. At the same time, the base diameter of 1 h grown nanotips is nearly half that of 3 h grown nanotips. Incidentally, the 3 h grown nanotip including the base part (Figure 1c) has a length more than 1 mm. The nanotips grown over 3 h have a higher surface gradient than that of the 1 h grown sample. Nanotips are grown with varying times to observe the evolution of tip structure (Figure 2). The formation of Ga droplet of size

Figure 2. Optical microscopy images of the formation of (a) a Ga droplet by annealing a Ga nodule in 10 min. Nucleation of GaN by self-agglomeration followed by the growth of nanotips after (b) 20, (c) 30, (d) 60, (e) 90, and (f) 180 min. The nanowire grown above 1 h possesses a sharp tip type morphology.

∼5−20 μm from the Ga nodule and further growth of nanotips are shown with increasing growth period. Magnified morphological images are also shown (Supporting Information, Figure S3) for certain growth periods. Morphological features for GaN NPs are shown in Figure 3a−c. The FESEM image (Figure 3a) of the sample grown at 900 °C and the corresponding TEM image (Figure 3b) show that the as-grown samples have a platelet type surface morphology with an average particle size of ∼100 nm and thickness less than 15 nm (indicated by arrow in Figure 3a), whereas the sample grown at 950 °C mostly exhibits elongated 2376



Article

Figure 3. (a) FESEM image of the NPs grown at 900 °C and the corresponding TEM image of NPs grown at (b) 900 and (c) 950 °C.

morphology (Figure 3c). All of the samples are crystalline in nature. The energy dispersive X-ray analysis (EDX) of a nanotip shows a decrease in the Ga/N ratio from 4.8 to 0.64 across the longitudinal direction (Supporting Information, Figure S4) in atomic percentage. It indicates that the growth that occurs in the Ga-rich condition eventually ends up with the N-rich condition. The HRTEM image along the basal plane shows that the surface is relatively smooth and consists of lattice fringes (Figure 4a). The near-perfect lattice spacing, with an

Figure 5. (a) HRTEM image of the NPs grown at 900 °C and (b) SAED pattern of single NPs obtained along the WZ ⟨12̅10⟩ zone axis.

from a NP (Figure 5b), which is indexed to ⟨12̅10⟩ zone axis of the WZ-GaN. A typical micro-Raman spectrum (Figure 6a) for the nanotip shows mode frequencies around 535, 567, and 726 cm−1.

Figure 4. HRTEM image of a single nanotip around (a) the base region, (b) the tip region, and (c) a high-resolution image of the transition region between polar (0001) and nonpolar (1010̅ ) surfaces, arround the tip region of the nanotips. The inset in panel c shows the magnified region of the transition region with large stacking faults.

interplanar spacing of 2.598 nm, corresponds to a (0002) plane of wurtzite (WZ) GaN. In the present case, the growth is dominated by the (0002) polar GaN surface. Growth occurs along the [0001] direction with a relatively smooth surface termination having a minimal amount of extended defects like dislocations and stacking faults along the surface. However, two distinct interfaces are observed around the tip region of the nanotips (Figure 4b,c). It is found that the surface layer consists of mostly (101̅0) nonpolar and few polar (101̅1) planes. These planes are presumably grown over (0002) polar surface belonging to the WZ-GaN phase. The transition from polar to nonpolar surface results in accumulation of a large amount of stacking faults at the interface. This is shown in the magnified region (inset of Figure 4c). The formation of these stacking faults may have been favored by the N-rich condition.20 The lattice images became obscured toward the core region as the thickness increases. Figure 5a shows the typical lattice images of crystalline NPs grown at 900 °C. Interplanar spacing of 2.76 nm corresponds to the (101̅0) polar surface of WZ-GaN. The corresponding selected area diffraction (SAED) pattern is taken

Figure 6. Raman spectra for (a) nanotips and (b) NPs showing WZ GaN phase.

Raman spectra for NPs show (Figure 6b) modes around 533, 569, and 732 cm−1, which correspond to E1(TO), E2(high), and A1(LO) phonon modes of the WZ-GaN phase, respectively.21 All of the spectra exhibit a strong E2(high) mode. A relatively smaller line width of 8 cm−1 of the nanotips than that of NPs (11.9 cm−1) proves the superior crystalline quality of nanotips. The broad line width of NPs may arise due to small grain size. Additional peaks at 253−265 and 421 cm−1 are primarily attributed to zone boundary phonons (ZB) arising from the finite size effect.22 LO modes are blue-shifted to 9 (line width ∼17 cm−1) and 3 cm−1 (line width ∼18 cm−1) for nanotips and NPs, respectively. In addition, the observed new peaks at 96, 609, and 702 cm−1 in NPs may be related to acoustic phonon overtones. The broad peaks around 634 and 693 cm−1 in NTs 2377



Article

Figure 7. (a) Typical Raman spectra along the diameter of a GaN nanotip at three different locations, surface, middle, and core region. Crosssectional Raman area mapping of a nanotip (b) along the diameter and (c) intensity distribution at 569 cm−1. The intensity of GaN peak drops slowly in the middle as a result of strain develop alongside.

are surface disorder-induced surface optical (SO) modes.22 Typical phonon modes along the lateral cross-section are shown in Figure 7a. Similar phonon spectra, as recorded for the surface region of nanotips (Figure 6a), are observed around the shell region. However, a blue shift of 18 cm−1 for the E2(high) mode is estimated at the core region of the nanotips. It shows the presence of a compressive strain developed along the core direction during the radial growth. The relation between the frequency shift Δω (cm−1) of the E2 (high) mode and biaxial stress σ(GPa) is estimated to be Δω = (4.2 ± 0.3)σ.23,24 Thus, the compressive stress developed inside the nanotip core nearly equals 4.29 GPa. Furthermore, the alternation of LO mode from A1 to E1 symmetry in the core region may be of geometric origin. The basal planes are predominant in the core, causing the E1 mode to be the preferred mode.25 The asymmetric behavior of optical phonons along the lateral cross-section (Figure 7b) of the nanotips is further analyzed by 3D mapping over an area of 20 μm × 20 μm. The map of the intense band of GaN phase with 568 cm−1, E2(high) mode frequency for the bulk GaN, illustrates (Figure 7c) that the strain free GaN phase is intensely distributed along the surface region with a contrast at the core region due to strain accumulation. A differential image (Figure 7d), obtained by comparing the 2D mapping of peak intensity corresponding to observed E2(high) mode at 569 cm−1 and the shifted peak at 587 cm−1 depicting the strained region in the core, is also shown as further evidence. RT micro-PL spectra of a single nanotip and NPs show strong emission in the photon energy in the range of 3.0−3.6 eV (Figure 8a,b). All of the observed emission lines possess broad and asymmetric line shapes with weak band edge emissions. The peak at 3.54 eV for the nanotip (Figure 8a) may correspond to the recombination of excitons bound to neutral donors (I2).23,26−29 The I2 line is blue-shifted by 80 meV with a broad line shape of ∼105 meV. Such a shift in this system is not expected due to quantum confinement effect. However, the PL line shift and additional broadening possibly arise due to strain in the system.23 The relation between the change in the excitonic PL, δE (meV), and biaxial stress σ(GPa) is estimated to be δE = (20 ± 3)σ.23 Thus, we have estimated a compressive

Figure 8. Typical micro-PL spectra for (a) nanotips and (b) NPs. The peaks at 3.54 and 3.3 eV corresponding to near band edge (I2) and FB emission, respectively. The inset in panel a shows the PL spectrum at the base region of nanotips with dominate emission peak at 3.1 eV. The inset in panel b shows peaks observed at 3.5 eV and corresponds to the I2 line; 3.27 and 3.12 eV correspond to neutral donor−acceptor pairs (DoAo) and their LO phonon replica (DoAo-LO), respectively; BB at 3 eV.

stress of 4 GPa in the nanotip. This is in close agreement with the previously calculated value of compressive stress using Raman shift of E2 (high) mode. The I2 line in the PL spectrum of GaN NPs shows a broad line shape (∼124 meV) with a blue shift of 30 meV. Because the NPs are grown in the vapor phase without using any substrate, the effect of stress is ruled out. The ambiguity of the observed blue shift, broad line width, and the asymmetric line shape of the I2 line of the NPs may be attributed to small-sized crystallites. However, unlike I2 emission, intense broad bands centered at 3.29 and 3.27 eV for the nanotip and NPs, respectively, are observed with similar line widths of ∼134 meV. The peak at 3.29 eV of the nanotip has been assigned to a zero phonon line and its phonon replica of the free-to-bound (FB) emission line. The FB transition is mediated by the free to a deep acceptor (DA) state. This possibly stems from Ga vacancy (VGa) formed under N-rich conditions and will be discussed later. However, additional peaks appearing in the spectra of NPs (Figure 8b) belong to various electronic origins. The peaks centered at 3.27 and 3.128 eV are due to recombination of neutral donor−acceptor pairs 2378



Article

(D0A0) and their LO phonon replica (D0A0-LO), respectively.30 An additional peak at 3 eV may be assigned to a blue band (BB).29 Schematic band diagrams (Figure 9a,b) describe

Ga might have helped the nucleation process. In the present study, different ramping rates of temperature of the reactor have been used to control incubation time periods for adatom migration times. These are detrimental in controlling the size for forming nanosized Ga droplets. These nanodroplets of Ga are likely to be used as nucleating centers for the formation of nanotips and NPs.8 In the case of nanotips, the slow ramping rate of temperature of the reactor (∼15 °C/min) leads to the formation of large Ga droplets of size ∼1−20 μm as a result of a self-agglomeration process. Energy stored in the defects and dangling bonds on the surface of Ga droplets might act as possible nucleating centers for the growth of the GaN phase. The slow ramping rate (higher incubation time) of the reaction temperature leads to the formation submicrometer-sized droplets, which take part in the nucleation process for the growth of GaN nanotips. The formation of oxide phase is likely at a slow ramping rate, as the nitride formation temperature at ∼900 °C is delayed. During this process, the residual and the adsorbed O2 species from the alumina surface reacted simultaneously and helped in the nucleation process with a formation of complex oxy-nitride phase.33 There are several reports on the oxide-assisted nucleation process for the growth of GaN nanostructures.33,34 Reduction or conversion of Ga2O3 phase to GaN phase at moderate temperature between 900 and 950 °C are also reported during the ammoniation process.35,36 Being a low symmetric crystal oxide phase, Ga2O3 (as observed in the Supporting Information, Figure S5) accelerates the nucleation process. Upon increasing the local ambient from Garich to N2-rich conditions, the oxygen contents reduced subsequently. Alternatively, it can be also stated that the N2 overpressure may have helped in the nucleation of GaN, inhibiting further formation of the oxide phase. The submicrometer-sized GaN column grows from the GaN seeds in a diffusion-limited self-agglomeration process promoted possibly by high vapor pressure of Ga and slow ramping rate toward reaction temperature.37 Furthermore, the conically shaped nanotips grow smoothly over the GaN column up to several millimeters. The growth is favored in the longitudinal direction due to the fact that vapor pressure at the conicalshape convex surface of the nanotips is much higher than that on the top edge.37 The growth can possibly be explained by the VS mechanism, which is very sensitive to gas phase kinetics during the growth process. The observed conical shape of nanotip morphology can further be elucidated by considering the characteristic length scale of the polar surface with respect to different reaction conditions (either Ga-rich or N-rich). It is observed that the Ga-rich condition results in broad polar faces, while in N-rich conditions, morphology mainly contains rough surfaces and facets.5−10 Ga-stabilized polar surfaces play an important role during GaN growth because Ga atoms have a significantly lower diffusion barrier on this polar surface.5 The effective diffusion length of Ga reactant on the Ga-stabilized polar surface is sensitive to the chemical environment. If excess N reactant is available, as in the N-rich condition, the free surface diffusion of Ga is disrupted because of the formation of Ga−N covalent bonds, which ultimately inhibits Ga diffusion length. However, a rapid ramping rate (about three times more than that used in case of nanotip growth) of the temperature of the reactor is employed for the formation of NPs. The rapid ramping rate of the temperature on the substrate reduces the effective adatom migration period. This hinders self-agglomeration and limits the size of Ga droplets to a smaller size than

Figure 9. Schematic representation of the emission mechanism in the (a) nanotips and (b) NPs.

the energetic of emission lines for nanotip and NPs. The wellknown yellow band centered at 2.2 eV is not noticeable in both of these samples, indicating high purity and crystallinity.27 Overtones of LO phonon modes up to third-order are also observed in the right arm of the PL spectra (Figure 8a,b). Higher order phonon modes arise due to strong electron− phonon coupling mediated Frolich interaction with excitations above the band gap (325 nm, ∼3.81 eV). This supports the above argument of higher crystalline quality in the nanostructures.30 On the other hand, the PL spectrum (inset in Figure 8a) from the base region of the nanotip, which is thought to be a possible nucleation site of nanotips (Figure 1c), shows a broad band emission around 3.1 eV. However, the near-band edge (NB) emission is found to be suppressed significantly. The base region of the nanotip is Ga-rich and contains oxide phases, as confirmed by EDX (Supporting Information, Figure S4) and Raman analyses (Supporting Information, Figure S5). A significant amount of nitrogen vacancies (VN) are present in the base region, which act as shallow donors.27 The transition from a shallow donor state (VN) to a deep acceptor state is expected to occur around 3.26 eV in the case of GaN epilayers.27,28 Nevertheless, we have assigned it as donor− acceptor pair (DAP) recombination after taking into account the transition from a shallow donor state VN to a deep acceptor state, such as VGaON type defect complex present in the nucleating region. The broad DAP line is red-shifted by 100 meV, indicating the presence of a significant amount of defect in the base region. Nevertheless, oxide-related native defects might have supported the foundation of strong nucleation sites for nanotips, as the incubation time with a low N 2 concentration is found to be fairly high in the initial stage for low ramping rate. The issue will be further elaborated in the subsequent paragraph. These observations lead us to propose a growth mechanism where the Ga/N ratio in the vapor phase plays an important role in determining the resulting morphology.6−10 During the vacuum annealing of the Ga precursor and subsequent increase in temperature from 800 to 900 °C at 200 mbar of pressure, nanosized Ga liquid droplets might have formed around the source material. This is supported by the recent reports, where small uniform-sized Ga droplets have been used as effective catalysts for the growth of aligned SiO2 nanowires.31,32 Weak Ga−Ga bonds, long liquid ranges, and high vapor pressures of 2379



Article

what is encountered during nanotip growth.8 Meanwhile, NH3 is introduced before the small Ga droplet ripens into a big cluster and helps in the nucleation process at the vapor phase itself. This occurs before the onset of the incubation time period.8 Thus, small Ga droplets react with NH3 in forming a large number of GaN islands and subsequent formation of NPs in the N-rich condition. Interestingly, we can notice a unique relationship among the Ga and N flux rates, growth orientation, and the resulting morphologies of nanotips and NPs. HRTEM data reveal that the nanotips growth occurs along [0001] polar direction (Figure 4, schematic in Figure 10). The characteristic length of

system appearing due to the differential growth rate among competitive facets. To summarize our model, three important consequences can be listed. These are (1) incubation time, (2) anisotropy of diffusion, and (3) rate-limiting factors of Ga and N adatoms in polar (0001) and nonpolar (101̅0) surfaces of GaN facets (schematic in Figure 10) play significant roles in deciding the final morphology of nanotips (schematic in Figure 11a,b) and

Figure 11. Schematic showing the possible growth mechanism of (a) GaN nanotips in different Ga-rich and N-rich growth conditions, (b) growth on the polar surface, (c) GaN NPs in N-rich conditions, and (d) growth on the nonpolar surface. Figure 10. Shematic of atomistic arrangement of the polar (0001) and the nonpolar (101̅0) surface of WZ GaN.

NPs (schematic in Figure 11c,d). Molecular N flux is found to have a stronger influence on the growth rate than nucleation.39 It is well supported by the observed variation of N 2 concentration, which increases longitudinally along the nanotip axis (Supporting Information, Figure S4). It is also found that few nanotips have branches (Figure 1b and Figure 2f, Supporting Information, Figure S6) at the end of the tip during longer (3 h) growth periods. These branches may be due to the formation of facets that occur during N-rich conditions (schematic in Figure 11a). On the other hand, NPs grown initially in N-rich conditions show the appearance of nanometric seeds with a multitude of facets (Figure 11c). These facets are made up of nonpolar surfaces (1010̅ ), (1120̅ ), and (110̅ 0) of Figure 11d. Such phenomena limit isotropic growth that invariably results in platelet morphology (schematic in Figure 11c,d). In conclusion, high-yield single crystalline and self-catalyzed WZ GaN nanotips and NPs have been synthesized by vaporsolid growth process using CVD technique. The nanotips nucleate basically in a Ga-rich condition and get converted to N-rich conditions on the surface as the reaction moves forward. N2 overpressure is proposed to play an important role in the nucleation of the GaN phase inhibiting further formation of the oxide phase nucleated during long incubation times with slow ramp rates for nanotips. The conical shape of the nanotips is explained in terms of reduced surface diffusion length in the Nrich condition than that of Ga-rich core region. Unidirectional growth of the nanotips is proposed to be driven by the strain developed due to the differential growth rate among these competitive facets. On the other hand, NPs are nucleated initially in a N-rich condition, which does not favor faceting with low surface diffusion length. The ability to control the growth of nanotips and NPs in a self-catalyzed process, independently, is expected to be advantageous in the realization of complex structures achievable by CVD process.

the polar surface at the base region is ∼20 μm. However, after a typical distance from the base region (≥1 μm), the surface layers of the nanotips are by and large nonpolar. It is basically dominated by (101̅0) planes (Figure 4c). The polar surface is found to shrink gradually toward the tip region. This variation of growth orientation is strongly supported by our growth conditions in which Ga and N flux rates have been altered after a typical incubation time of ∼10 min during the growth.5−7,20 In the case of nanotip growth, the reaction condition is initially Ga-rich, which favors growth of polar surfaces. This can be explained from the fact that the interaction between adsorbate and surface is predominately realized by very weak delocalized metallic Ga−Ga bonds, which are isotropic with low diffusion barriers ≈0.4 eV along the (0001) polar surface.5−10 However, anisotropic and differential growth orientations with a shrunken polar surface are expected when the reaction condition becomes relatively N-rich upon NH3 flow. It is known that the N-rich condition increases the nucleation rate of GaN due to higher strength of Ga−N bond and shorter diffusion length of Ga than that of Ga−Ga bond. Therefore, the Ga atoms will be bonded to N atoms before they complete their migration to step edges, to promote 3D growth and introduce roughness.5−8,20,38 Apart from this, recently, first principle calculations by Lymperakis et al.38 have revealed that the N-rich condition promotes island nucleation and the Ga adatom incorporation is more probable on the (0001) polar surface than that of other nonpolar surfaces (a- and m-planes). Thus, the diffusion-induced mechanism of adatom migration on the polar surface promotes faster material transport occurring from the side facets (nonpolar surface) toward the top. Thus, the higher nucleation rate toward the top is expected to stimulate and favor the axial growth over radial growth for the nanotips.38 In other words, the observed nanotip morphology can be thought of driven by the strain, as indicated from the Raman (Figure 7) and PL (Figure 8a) peak positions shift, in the 2380



■

Article

(23) Kisielowski, C.; Kruger, J.; Ruvimov, S.; Suski, T.; Ager, J. W.; Jones, E.; Weber, Z. L.; Rubin, M.; Weber, E. R.; Bremser, M. D.; Davis, R. F. Phys. Rev. B 1996, 54, 17745. (24) Davydov, V. Y.; Averkiev, N. S.; Goncharuk, I. N.; Nelson, D. K.; Nikitina, I. P.; Polkovnikov, A. S.; Smirnov, A. N.; Jacobson, M. A.; Semchinova, O. K. J. Appl. Phys. 1997, 82, 5097. (25) Gorczyca, I.; Christensen, N. E.; Peltzer y Blanca, E. L.; Rodriguezt, C. O. Phys. Rev. B 1995, 51, 11936. (26) Bae, S. Y.; Seo, H. W.; Park, J.; Yang, H.; Kim, H.; Kim, S. Appl. Phys. Lett. 2003, 82, 4564. (27) Reshchikova, M. A.; Morkoç, H. J. Appl. Phys. 2005, 97, 061301. (28) Sahoo, P.; Oliveira, D. S.; Cotta, M. A.; Dhara, S.; Dash, S.; Tyagi, A. K.; Raj, B. J. Phys. Chem. C 2011, 115, 5863. (29) Leroux, M.; Grandjean, N.; Beaumont, B.; Nataf, G.; Semond, F.; Massies, J.; Gibart, P. J. Appl. Phys. 1999, 86, 3721. (30) Dhara, S.; Chandra, S.; Mangamma, G.; Kalavathi, S.; Shankar, P.; Nair, K. G. M.; Tyagi, A. K.; Hsu, C. W.; Kuo, C. C.; Chen, L. C.; Chen, K. H.; Sriram, K. K. Appl. Phys. Lett. 2007, 90, 213104. (31) Zheng, B.; Wu, Y.; Yang, P.; Liu, J. Adv. Mater. 2002, 14, 122. (32) Pan, Z. W.; Dai, Z. R.; Ma, C.; Wang, Z. L. J. Am. Chem. Soc. 2002, 124, 1817. (33) Xue, S.; Zhang, X.; Huang, R.; Zhuang, H.; Xue, C. Dalton Trans. 2008, 32, 4296. (34) Shi., W. S.; Zheng, Y. F.; Wang, N.; Lee, C. S.; Lee, S. T. Chem. Phys. Lett. 2001, 345, 377. (35) Li, J.; An, L.; Lu, C.; Liu, J. Nano Lett. 2006, 6, 148. (36) Lan, Z. H.; Liang, H.; Hsu, C. W.; Wu, C. T.; Lin, H. M.; Dhara, S.; Chen, K. H.; Chen, C.; Chen, C. C. Adv. Funct. Mater. 2004, 14, 233−237. (37) Choi, H. J.; Kim, D. H.; Kim, T. G.; Lee, J. C.; Sung, Y. M. J. Electroceram. 2006, 17, 221. (38) Lymperakis, L.; Neugebauer, J. Phys. Rev. B 2009, 79, 241308. (39) Bertness, K. A.; Roshko, A.; Mansfield, L. M.; Harvey, T. E.; Sanford, N. A. J. Cryst. Growth 2007, 300, 94.

ASSOCIATED CONTENT

S Supporting Information *

Additional figures. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (P.S.) or [email protected]. in (S.D.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS P.S. acknowledges the Department of Atomic Energy for financial aid. We thank Avinash Patsha for his help in experimental work. We also thank G. Amarendra and C. S. Sundar for their support in continuing this work.

■

REFERENCES

(1) Sahoo, P.; Dhara, S.; Dash, S.; Tyagi, A. K. Nanosci. Nanotechnol.Asia 2011, 1, 140. (2) Wang, D.; Pierre, A.; Kibria, M. G.; Cui, K.; Han, X.; Bevan, K. H.; Guo, H.; Paradis, S.; Hakima, A. R.; Mi, Z. Nano Lett. 2011, 11, 2353. (3) Dimakis, E.; hnemann, J. L.; Jahn, U.; Breuer, S.; Hilse, M.; Geelhaar, L.; Riechert, H. Cryst. Growth Des. 2011, 11, 4001. Fortuna, S. A.; Li, X. Semicond. Sci. Technol. 2010, 25, 024005. (4) Deb, P.; Kim, H.; Rawat, V.; Oliver, M.; Kim, S.; Marshall, M.; Stach, E.; Sands, T. Nano Lett. 2005, 5, 1847. (5) Nam, C. Y.; Tham, D.; Fischer, J. E. Appl. Phys. Lett. 2004, 85, 5677. (6) Chèze, C.; Geelhaar, L.; Jenichen, B.; Riechert, H. Appl. Phys. Lett. 2010, 97, 153105. (7) Gotschke, T.; Schumann, T.; Limbach, F.; Stoica, T.; Calarco, R. Appl. Phys. Lett. 2011, 98, 103102. (8) Consonni, V.; Trampert, A.; Geelhaar, L.; Riechert, H. Appl. Phys. Lett. 2011, 99, 033102. (9) Lefebvre, P.; Garrido, S. F.; Grandal, J.; Ristić, J.; García, M-A. S.; Calleja, E. Appl. Phys. Lett. 2011, 98, 083104. (10) Garrido, S. F.; Grandal, J.; Calleja, E.; García, M. A. S.; Romero, D. L. J. Appl. Phys. 2009, 106, 126102. (11) Schumann, T.; Gotschke, T.; Limbach, F.; Stoicaand, T.; Calarco, R. Nanotechnology 2011, 22, 095603. (12) Cai, X. M.; Djurišić, A. B.; Xie, M. H.; Chiu, C. S.; Gwo, S. Appl. Phys. Lett. 2005, 87, 183103. (13) Bae, S. Y.; Seo, H. W.; Park, J.; Yang, H.; Kim, H.; Kim, S. Appl. Phys. Lett. 2003, 82, 4564. (14) Lin, C.; Yu, G.; Wang, X.; Cao, M.; Lu, H.; Gong, H.; Qi, M.; Li, A. J. Phys. Chem. C 2008, 112, 18821. (15) Schwenzer, B.; Hu, J.; Morse, D. E. Adv. Mater. 2011, 23, 2278. (16) Chattopadhyay, S; Chen, L. C.; Chen, K. H. Crit. Rev. Solid State Mater. Sci. 2006, 31, 15. (17) Yoshida, H.; Urushido, T.; Miyake, H.; Hiramatsu, K. Jpn. J. Appl. Phys. 2001, 40, L1301. (18) Jacobs, B. W.; Crimp, M. A.; McElroy, K.; Ayres, V. M. Nano Lett. 2008, 8, 4353. (19) Sahoo, P.; Dhara, S.; Dash, S.; Raj, B.; Manna, I.; Tyagi, A. K. Appl. Phys. Lett. 2011, 98, 043103. (20) Myoung, J. M.; Gluschenkov, O.; Kim, K.; Kim, S. J. Vac. Sci. Technol., A 1999, 17, 3019. (21) Sahoo, P.; Dhara, S.; Das, C. R.; Das, S.; Tyagi, A. K.; Raj, B. Int. J. Nanotechnol. 2010, 7, 823. (22) Dhara, S.; Sahoo, P.; Tyagi, A. K.; Raj, B. Nanowires: Implementations and Applications; Intech Open Access: Croatia, 2011; pp 3−26. 2381


Role of Surface Polarity in Self-Catalyzed Nucleation and Evolution of

Recommend Documents