Letter pubs.acs.org/NanoLett
Phase Selection Enabled Formation of Abrupt Axial Heterojunctions in Branched Oxide Nanowires Jing Gao,† Oleg I. Lebedev,‡ Stuart Turner,§ Yong Feng Li,† Yun Hao Lu,∥,⊥ Yuan Ping Feng,∥ Philippe Boullay,‡ Wilfrid Prellier,‡ Gustaaf van Tendeloo,§ and Tom Wu*,† †
Division of Physics and Applied Physics, School of Physical and Mathematical Sciences, Nanyang Technological University, Singapore 637371, Singapore ‡ Laboratoire CRISMAT, ENSICAEN, CNRS UMR 6508, 6 Boulevard du Maréchal Juin, 14050 Caen, France § EMAT, University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerpen, Belgium ∥ Department of Physics, National University of Singapore, 2 Science Drive 3, Singapore, 117542, Singapore ⊥ Laboratory of New-Structured Materials, Department of Materials Science and Engineering, Zhejiang University, Hangzhou, 310027, People’s Republic of China S Supporting Information *
ABSTRACT: Rational synthesis of nanowires via the vapor− liquid−solid (VLS) mechanism with compositional and structural controls is vitally important for fabricating functional nanodevices from bottom up. Here, we show that branched indium tin oxide nanowires can be in situ seeded in vapor transport growth using tailored Au−Cu alloys as catalyst. Furthermore, we demonstrate that VLS synthesis gives unprecedented freedom to navigate the ternary In−Sn−O phase diagram, and a rare and bulk-unstable cubic phase can be selectively stabilized in nanowires. The stabilized cubic fluorite phase possesses an unusual almost equimolar concentration of In and Sn, forming a defect-free epitaxial interface with the conventional bixbyite phase of tin-doped indium oxide that is the most employed transparent conducting oxide. This rational methodology of selecting phases and making abrupt axial heterojunctions in nanowires presents advantages over the conventional synthesis routes, promising novel composition-modulated nanomaterials. KEYWORDS: Vapor−liquid−solid growth, indium tin oxide, branched nanowire, phase selection, heterojunction
I
applications in nanoelectronics, photonics, and energy harvesting. But the experimental realizations of such interesting VLSgrown nanostructures with epitaxial interfaces have been limited to a few material systems such as silicon, germanium, III−V compounds, and some oxides. Regarding the growth of branched nanowires, ex situ seeding is the most employed method where the main nanowire trunks were grown first and the subsequently decorated catalyst nanoparticles seed the growth of nanowire branches. As a drastically different approach, in situ seeding was reported recently for silicon nanowires.15,24 It relies on controllably destabilizing the catalyst nanoparticles at the end of primary growth so that the derived nanoparticles can migrate along the primary nanowire trunks and seed the subsequent growth of branches. The in situ seeded growth process is illustrated in Figure 1A. The onset of catalyst migration depends on various factors such as growth temperature and pressure, and what matters is the overall change of chemical potentials when the catalyst starts to migrate from the nanowire tips onto the
n the nanowire growth via the vapor−liquid−solid (VLS) process and its variants,1−3 the complex interactions between vapor, liquid and solid phases give rise to rich phenomena,4−10 often going beyond the general conception that nanowires are merely a miniature form of their bulk counterpart sharing the same crystalline phase. In VLS synthesis of nanowires, liquid-phase catalyst nanoparticles play pivotal roles by collecting vapor-phase precursors, and their dynamic supersaturation drives the precipitation of crystalline nanowires.1,11 Therefore, the nanoparticles serve as tiny nanoscale reactors, making the VLS growth depart from the conventional synthesis routes. Chemical pressures and the associated thermodynamic equilibrium in catalyst nanoparticles can be dynamically tailored to control the nucleation at the nanoparticle/nanowire interface.12 Recently, there have been notable advances in VLS synthesis propelled by the needs in constructing nanowire-based functional devices. In particular, the dynamics of catalyst nanoparticles and modulation of the vapor-phase reactants during VLS growth have been harnessed in synthesizing complex nanowire-based architectures, such as branched nanowires,4,13−16 abrupt heterojunctions,7,17−20 superlattices,6,21,22 and kinked superstructures.23 These structure and composition tailored nanostructures are promising for © 2011 American Chemical Society
Received: October 6, 2011 Revised: November 23, 2011 Published: December 5, 2011 275
dx.doi.org/10.1021/nl2035089 | Nano Lett. 2012, 12, 275−280
Nano Letters
Letter
situ seeding the growth of nanowire branches, assisting selective stabilization of different phases, and serving as a template for the formation of ultrathin epitaxial SnO2 shells. To select the appropriate catalyst nanoparticles for the in situ seeded growth, we systemically tested various metals as well as their alloys. The binary phase diagram of the Au−Cu system indicates that the melting temperature of Au−Cu alloys is lower than both Au and Cu, implying a possible high mobility of the alloy nanoparticles27,28 Furthermore, Au and Cu share the same crystal structure with similar lattice constants, so alloying is an energetically favored process. The ITO nanowires were grown using a vapor transport method in a horizontal tube furnace equipped with pumping and gas regulating apparatus as described previously.29−31 Au−Cu alloy catalyst nanoparticles were prepared via either deposition of metal bilayers or via the doping of Cu through the vapor source. Figure 1B illustrates the designed growth process aiming to achieve in situ seeded branched nanowires (see Supporting Information Section I for more details). Stages I and II are designed to grow nanowire trunks and branches, respectively. Between them is a transition period in which the growth temperature is abruptly reduced from 840 to 600 °C and the pressure drops from 30 to 0.3 mbar. These optimized growth parameters, in particular the abrupt modulations, are critical to cause the breakup and migration of the catalyst nanoparticles to seed the branch growth. Among the various factors affecting the nanowire growth, the abrupt dropping of the furnace temperature significantly increases the Sn partial pressure within the vapor flux as a result of the lower carbon-reduction temperature of SnO2 than that of In2O3 in the source. This enables the phase selection in the indium tin oxide nanowires, as we will discuss in detail. As shown in the scanning electron microscopy (SEM) image in Figure 1C, after the stage I aligned ITO nanowires epitaxially grow on (100) yttria-stabilized zirconia (YSZ) substrates. Recently, these ITO nanowires have been shown to exhibit versatile physical properties.14,32−35 In the stage II, disintegration and migration of catalyst particles accomplish the in situ seeding process and reliably lead to the growth of branched nanowires (Figure 1D, also see Supporting Information Sections II and III). This in situ seeding process hinges on the dynamic changes of the growth conditions23,36 as well as the lower melting point and probably the enhanced mobility of the AuCu alloy. The branched nanowires exhibit a 4-fold symmetry, reflecting the expected crystallinity and the epitaxial growth. Moreover, as shown in Figure 1E, the average length of the nanowire branches can reach 400 nm and be well controlled by adjusting the duration of growth stage II. We surveyed many nanowire branches, and generally the thicker ones show higher growth rates, which is consistent with the Gibbs−Thomson effect37 (see Supporting Information Figure S5C). In our experiments, the abrupt pressure change between the growth stages I and II presumably breaks up the original nanoparticles into much smaller ones (see Supporting Information Section III), and the enhanced mobility of Au− Cu alloys facilitates the migration of the generated nanoparticles that seed the subsequent growth of branches. Importantly, our synthesis strategy also brings about the stabilization of a rare cubic phase, demonstrating the advantages of our approach of using bimetallic Au−Cu alloys in the VLS synthesis, which will be the focus of following discussions.
Figure 1. Growth of branched In−Sn−O nanowires using Au/Cu alloys as catalyst. (A) Schematic illustration of the in situ seeded growth of branched nanowires. In step 1, Au alloys with Cu forming bimetallic droplets that then initiate the growth of primary nanowire trunks in step 2. In step 3, the nanoparticles are disintegrated into much smaller ones that subsequently migrate and seed the growth of nanowire branches in step 4. (B) Diagram of a typical growth process. The three stages serve to (I) grow the primary nanowire trunks, (II) seed the branches, and (III) terminate the growth. (C) SEM image of aligned ITO nanowires grown on a YSZ substrate without in situ seeding. Scale bar, 1 μm. (D) SEM image of branched ITO nanowires after the grown process shown in panel B. Both growth stages (I and II) last for 45 min. Scale bar, 1 μm. (E) Dependence of the branch length on the duration of the growth stage II.
sidewalls.25 In the in situ seeding method, the synthesis of branched nanowires can be accomplished within a single growth process without ex situ depositing nanoparticles on the nanowire trunks. This approach effectively avoids undesirable surface contamination or oxidation which often happens in the ex situ seeding process.15 However, in situ seeding is difficult to achieve in growing a wide range of functional oxide nanowires because the presence of oxygen significantly impedes the Au diffusion.26 Besides in situ seeded branched oxide nanowires, fabrication of abrupt axial heterojunctions with compositional modulation and epitaxial interfaces has not been reported so far in oxide nanowires although functional oxides with compatible crystalline structures are common in the bulk and thin film forms. In this work, we significantly advanced the conventional VLS growth by using dual-metal Au−Cu alloys as catalyst to enhance the mobility of catalyst nanoparticles, enabling the in situ seeded growth of branched indium tin oxide nanowires. Furthermore, utilizing the variation of the carbothermal reduction temperatures of tin and indium sources and their sequential inputs into the growth process, we were able to dynamically tune the chemical potentials in the catalyst nanoparticles and selectively stabilize a cubic indium tin oxide phase in the nanowires. This cubic indium tin oxide phase is rare in the bulk and thin film counterparts, forming epitaxial axial heterojunctions with the conventional tin-doped indium oxide phase within individual nanowires. Overall, instead of merely a collector of source vapors, the Au−Cu alloy nanoparticles play multiple roles during the VLS growth: in 276
dx.doi.org/10.1021/nl2035089 | Nano Lett. 2012, 12, 275−280
Nano Letters
Letter
Figure 2. Observation of cubic In−Sn−O phase. (A) Bright-field TEM image of a branched In−Sn−O nanowire. The growth directions of the trunk and the branch are indicated by arrows. Scale bar, 100 nm. (B) ADF-STEM image of the top part of an individual branch. The sharp boundary between the conventional ITO phase and a new indium tin oxide (denoted here as ISO) is highlighted. Scale bar, 20 nm. (C) HRTEM image of the ITO/ISO interface. Insets are the SAED patterns taken on the individual nanowire segments (A and B mark the positions of the electron beam). (D) Corresponding EDX data. (E) EELS spectrum of the ISO phase in comparison to the reference spectra of ITO, In2O3, and SnO2. (F) Nanobeam ED patterns recorded for the ISO phase. The strong Bragg reflection present in the reciprocal space can be indexed according to a fluorite structure (SG: Fm3̅m, a = 0.5 nm). As shown in the ⟨110⟩ and ⟨210⟩ zone axis patterns, all ED patterns except [100] exhibit highly structured diffuse scattering related to the presence of short-range order. (G) Schematic of the simplified fluorite structure of ISO. Note that in the actual oxygen deficient fluorite phase the cationic sublattice is disordered and one-eighth of the oxygen sublattice is vacant (the vacancy denoted by the red dot should be randomly distributed).
The bright-field transmission electron microscopy (TEM) image in Figure 2A shows a nanowire branch, forming a perpendicular single crystalline junction with the trunk. Interestingly, the annular dark-field scanning TEM (ADFSTEM) image in Figure 2B shows a clear boundary at the waist of the nanowire branch. The high-resolution TEM (HRTEM) image in Figure 2C suggests that the lower part of the branch adopts the conventional ITO bixbyite phase, while the upper part exhibits a different crystalline structure, giving rise to an abrupt defect-free heteroepitaxial interface. The electron diffraction (ED) patterns (insets of Figure 2C) confirm the existence of two distinct structures within the nanowire. The energy dispersive X-ray (EDX) spectrum taken at the upper part of the nanowire (Figure 2D, upper panel) indicates that the new structure we observed is indium tin oxide with a Sn/In ratio close to 1, which is much higher than that of the conventional ITO (
