Semiconductor Nanowires: Rational Synthesis Part 1 (Nanotechnology)

INTRODUCTION

Rational design and synthesis of nanoscale materials is critical to work directed towards understanding fundamental properties, creating nanostructured materials, and developing nanotechnologies. One-dimensional (1D) nanostructures [such as nanowire (NW) and nanotubes] have been the focus of considerable interest because they have the potential to answer fundamental questions about role of dimensionality in physical properties and are expected to play a central role in applications ranging from molecular electronics to scanning probe microscopy probes. To explore the diverse and exciting opportunities in 1D system requires materials for which the chemical composition, diameter, length, electronic, and optical properties can be controlled and systematically varied. To meet these requirements, we have focused our efforts on developing a general and predictive approach for the synthesis of 1D structures, much as molecular beam epitaxy has served as an all-purpose method for the growth of two-dimensional (2D) structures. Specifically, it is important to achieve the ability to design and synthesize rationally NWs with predictable control over the key structural, chemical and physical properties, since such control would greatly facilitate studies designed to understand the intrinsic behavior of 1D structures and to explore them as building blocks for nanoscale electronics. Here, in this article, we review recent advances in rational synthesis of semiconductor NWs. We will first address the key requirement for 1D growth and give a brief overview of various methods towards 1D materials. Subsequently, we will focus our discussion on growth of a broad range of semiconductor NWs via a metal-nanocluster mediated catalytic growth method based on vapor-liquid-solid (VLS) growth mechanism. Next, we further describe growth of NW materials with controlled physical size including diameter and length. Lastly, we discuss growth of NW heterostructures and superlattices with composition/doping modulation along the axial or radial direction.

SYMMETRY BREAKING: A KEY CONCEPT FOR 1D GROWTH

In general, the growth of 1D nanostructures requires that two dimensions are restricted to the nanometer regime, while the third dimension extends to macroscopic dimensions.[1-3] This overall requirement is considerably more difficult to achieve than the corresponding constraints needed for successful growth of 0D and 2D structures.[4,5] For example, many important semiconductor materials adopt a cubic zinc-blende (ZB) structure, and thus when growth is stopped at an early stage, the resulting nanoscale structures are nanocrystals with various polyhedron shapes and not 1D NWs (Fig. 1). To achieve 1D growth in systems where atomic bonding is relatively isotropic requires that the growth symmetry be broken rather than simply arresting growth at an early stage.

Over the past decade, considerable effort has been placed on the bulk synthesis of NWs, and various strategies have been developed to break the growth symmetry either "physically” or "chemically.” A common theme in many of these studies has been the use of linear templates, including the edges of surface steps,[6] nanofibers,[7,8] and porous membranes,[9] to direct chemical reactions and material growth in 1D (Fig. 2). This strategy is conceptually simple and has been used to prepare a wide range of NW materials. Despite this simplicity, template-mediated growth is also limited in that the resulting NWs are usually polycrystalline, which could limit their potential for many applications.

Another general strategy that has received increasing focus over that past several years involves exploiting a "catalyst” to confine growth in 1D. Depending on the phases involved in the reaction, this approach is typically defined as VLS,[10,11] solution-liquid-solid (SLS),[12,13] or vapor-solid (VS) [14,15] growth. In VLS growth, a vapor-phase reactant is solubilized by a liquid catalyst particle to form solid wire/rod-like structures. Solution-liquid-solid adopts the same idea with the only change that reactant comes from a solution rather than vapor phase; in VS, a vapor-phase reactant reacts directly on the solid particle to form a solid wire/tube. Vapor-liquid-solid growth has been used to synthesize a variety of whisker-like structures but with typical diameter much larger than 200 nm.[10,11] Solution-liquid-solid has been used to grow NWs of III-V materials but seems to be limited by the availability of low-melting point metal catalyst.[12'13] Vapor-solid growth has been used to grow carbon nanotubes, and a number of different oxide nanorods,[14'15] although it seemed to be limited to the growth of only some particular types of materials and the growth mechanism is yet to be understood.

Fig. 1 Schematic view of key concept for 1D growth. To achieve 1D growth in systems where atomic bonding is relatively isotropic requires that the growth symmetry be broken rather than simply arresting growth at an early stage.

Fig. 2 Schematic view of growth of NWs by physically confined chemical reaction. Growth of NWs by (A) Step edge deposition; (B) Pore filling; and (C) Deposition on preexisting 1D fibers such as carbon nanotubes.

CATALYTIC GROWTH: CONCEPTS AND SYNTHETIC DESIGN

Catalytic growth, where the catalyst is used to direct 1D growth of single crystal materials via a VLS mechanism (Fig. 3a), is a powerful concept for NW synthesis.[1-3] Here, the catalyst is envisioned as a nanocluster or nanodroplet that defines the diameter of and serves as the site that directs preferentially the addition of reactant to the end of a growing NW much like a living polymerization catalyst directs the addition of monomers to a growing polymer chain.

This synthetic concept is especially important since it readily provides the intellectual underpinning needed for the specification of the catalyst and growth conditions required for predictable NW growth. First, equilibrium phase diagrams (Fig. 3b) are used to determine catalyst materials that form a liquid alloy with the NW material of interest. The phase diagram is then used to choose a specific composition and growth temperature such that there is a coexistence of a liquid alloy and solid NW phases. The liquid catalyst alloy cluster serves as the preferential site for absorption of reactant since the sticking coefficient is much higher on liquid vs. solid surfaces and, when supersaturated, the nuclea-tion site for crystallization. Preferential 1D growth occurs in the presence of reactant as long as the catalyst nanodroplet remains in the liquid state. Within this framework, it is straightforward to synthesize NWs with different diameters and composition, if appropriate nanometer scale diameter catalyst clusters are available. Several methods that exploit this general approach are described later with an emphasis on laser-assisted catalytic growth (LCG) and metal-catalyzed chemical vapor deposition (CVD).

Fig. 3 Catalytic growth of NWs. (a) Schematics illustrating the underlying concept for catalytic growth of NWs. Liquid catalytic clusters act as the energetically favored site for localized chemical reaction, absorption of vapor-phase reactant, and crystallization of crystalline NWs. (b) Binary A-B phase diagram used as guide for choosing a catalyst for NW growth. The vertical arrow represents a specific composition of catalyst (A) to NW (B) with end point corresponding to the growth temperature. The horizontal arrow defines the composition of the catalyst liquid (L) catalyst-NW (A-B) nanodroplet and shows that pure solid NW (B) is the only solid phase at this temperature.

LASER-ASSISTED CATALYTIC GROWTH

A straightforward and general approach for producing the nanometer scale clusters required to nucleate and direct the growth of NWs is laser ablation and conden-sation.[16] In this context, NW growth can be readily achieved by laser ablation of a composite target containing the catalyst and NW material in a heated flow tube (Fig. 4). The background pressure within the flow reactor is used to control condensation of the ablated material and the cluster size, while the temperature is varied to maintain the catalyst cluster in the liquid state. When the laser-generated clusters become supersaturated with the desired NW material, a nucleation event occurs producing a (NW) solid-liquid (NW-catalyst alloy) interface. To minimize the interfacial free energy, subsequent solid growth/ crystallization occurs at this initial interface, which thus imposes a highly anisotropic growth constraint that lead to 1D nanoscale wires. We have termed this method "laser-assisted catalytic growth.”[3] This approach has proven to be very general for the synthesis of semiconductor NW materials, including group IV elemental,[17] group IV alloys, and group III-V, II-VI, and IV-VI compound semiconductor NW materials.[3,18-20] Later we demonstrate the basic principles underlying this approach with the rational and predictable growth of gallium arsenide (GaAs) NWs.

Fig. 4 Schematic of a laser-based NW growth apparatus.

Laser-Assisted Catalytic Growth of GaAs NWs

A key feature of our catalytic growth approach for NW synthesis is that equilibrium phase diagrams can be used to predict the catalyst material, chemical composition, and growth conditions, thus enabling rational and predictable synthesis of new NW materials. For example, binary phase diagrams have been used to predict the composition and growth temperatures for the synthesis of elemental semiconductor NW materials such as silicon and germanium NWs.[17] Predictable catalytic growth of compound NWs is more challenging than that of Si and Ge NWs due to the complexity of ternary and higher order phase diagrams. However, such complexity can be greatly simplified by considering pseudobinary phase diagrams for the catalyst and compound semiconductor of interest. For example, the pseudobinary phase diagram of Au-GaAs (Fig. 5A) shows that Au-Ga-As liquid and GaAs solid are the principle phases above 630° C in the GaAs-rich region of the phase diagram.[21] This information implies that Au can serve as a catalyst to grow GaAs NWs by the LCG method, if the target composition and growth temperature are chosen to be within this region of the phase diagram.

Indeed, LCG growth of GaAs NWs using Au as the catalyst produces samples consisting primarily of wirelike structure with diameters on the order of 10 nm, and lengths extending up to tens of micrometers (Fig. 5B). X-ray diffraction (XRD) data from bulk GaAs NW samples can be indexed to the ZB structure with a lattice constant consistent with bulk GaAs, and also show that the material is pure GaAs at least to 1% level. In addition, we note that high yields of GaAs NWs were also obtained using Ag and Cu catalysts. These latter data are consistent with the fact that Ag and Cu exhibit M-Ga-As (M = Ag or Cu) liquid and GaAs solid phases in the GaAs-rich region of the corresponding pseudobinary phase diagrams.[21] Taken together, these results demonstrate clearly the power of the LCG approach for predictable NW growth.

Detailed transmission electron microscope (TEM) studies along with electron diffraction (ED) and energy dispersive X-ray (EDX) analysis also revealed important features of the structure and chemical compositions of the produced GaAs NWs. First, diffraction contrast images of individual NWs (Fig. 5C) indicate that NWs are single crystal (uniform contrast) and uniform in diameter. Second, the Ga: As composition determined by EDX is consistent with stoichiometric GaAs within limits of instrument sensitivity. Third, the ED pattern recorded perpendicular to the long axis of this NW (Fig. 5C, inset) can be indexed for the (112) zone axis of the ZB GaAs structure, and thus shows that growth occurs along the [1 1 1] direction. Extensive studies of individual GaAs NWs show that growth occurs along the (1 1 1) directions in all cases. This direction and the single crystal structure have been further confirmed by lattice-resolved TEM images (e.g., Fig. 5D) that show clearly the (1 1 1) lattice planes perpendicular to the wire axis. Lastly, the TEM studies also revealed that most NWs terminate at one end with a nanoparticle of higher contrast (Fig. 5E). Energy dispersive X-ray analysis indicates that the nanoparticles are composed largely of Au. The presence of Au nanoparticles at the ends of the NWs is consistent with the pseudobinary phase diagram, and represents strong evidence for a VLS growth mechanism proposed for LCG.

Fig. 5 Laser-assisted catalytic growth of GaAs NWs. (A) Pseudobinary phase diagram of the GaAs-Au system. (B) Scanning electron microscope (SEM) image of GaAs NWs produced by LCG. (C) Diffraction contrast TEM image of a GaAs NW indicating single crystal structure. The inset shows a convergent beam ED pattern recorded along (112) zone axis, and demonstrates wire axis (and growth) is along the [1 1 1] direction. (D) High-resolution TEM image of the NW. The (1 1 1) lattice planes (lattice spacing 0.32 ± 0.01 nm) are clearly seen and perpendicular to the wire axis, further confirming the [1 1 1] growth direction. (E) High-resolution TEM image of an NW end showing a catalyst nanocluster. Energy dispersive X-ray analysis indicates that the nanoparticle is composed primarily of Au with trace amounts of Ga and As.

General Synthesis of Semiconductor Nanowires

The successful synthesis of binary GaAs NWs by LCG is representative of the broad range of binary and more complex NW materials prepared by this method.[3,18-20] In a number of cases, phase diagrams could be used to define clearly catalyst, composition and growth conditions required for successful NW growth, although appropriate phase diagrams (e.g., pseudo-binary and more complex) are not readily available for many compound semiconductor materials and catalytic metals of interest. To extend our synthetic approach to the broadest range of materials, we recognized that catalysts for LCG can be chosen in the absence of detailed phase diagram data by identifying metals in which the NW component elements are soluble in the liquid phase but that do not form solid compounds more stable than the desired NW phase; that is, the ideal metal catalyst should be ”physically active” but "chemically stable.” This basic guiding principle suggests that the noble metal Au should represent a good starting point for many semiconductor materials. Indeed, Table 1 summarizes the broad range of NW materials, with different structure types and chemical properties, synthesized as high-quality single crystals using Au as the catalyst. These results demonstrate unambiguously the generality of our approach.

The Case of GaN Nanowires

Gold nanoclusters are unable to catalyze GaN NW growth due to the low solubility of nitrogen.[19] These results are consistent with the basic requirement for a catalyst: physically active and chemically stable. In this context, Au is not physically active. A good catalyst should form a miscible liquid phase with GaN but not lead to a more stable solid phase under the NW growth conditions. This guiding principle suggests that Fe and Ni, which dissolve both gallium and nitrogen, and do not form a more stable compound than GaN, will be good catalysts for GaN NW growth.

Significantly, we found that LCG using either a GaN/Fe or GaN/Ni targets produces a high yield of nanometer diameter wire-like structures.[19] Fig. 6A shows a diffraction contrast TEM image of a GaN NW synthesized via an iron catalyst. The uniform contrast of the NW suggests that it has a single crystalline structure. Electron diffraction confirmed that the NWs are single crystals with a wurtzite structure and a [1 0 0] growth direction. Images of the NW ends also exhibited nanoclusters consisting primarily of iron, consistent with a VLS growth mechanism. Lastly, high-resolution TEM images (Fig. 6B,C) confirmed the single crystal structure and growth direction, and moreover, demonstrated that the surfaces of the GaN NWs terminate sharply with only 1-2 atomic layers of amorphous oxide.

Diameter-, Length-, and Doping-Controlled Synthesis

In the LCG method, laser ablation is used to simultaneously generate nanoscale metal catalyst clusters and semiconductor reactant that produce NWs via a VLS growth mechanism. A major advantage of this approach is its flexibility and generality since laser ablation can be used to produce nanoclusters of virtually any material, which has enabled the growth of a very wide range of NWs. To fully exploit the potential of these NW materials also requires control of physical dimensions (i.e., diameter and length) and electronic properties (e.g., doping).

Implicit to our catalytic approach to NW growth is the idea that the size of the metal catalysts determine NW diameter, and thus NWs with a narrow size distribution should be obtained from approximately monodisperse nanocluster catalysts. This important idea was verified recently through the demonstration of diameter-controlled growth of GaP,[22] InP,[23] and Si NWs[24] from monodisperse Au nanocluster catalysts. Significantly, the widths of the NW diameter distributions were essentially the same as those of the starting Au nanoclusters, thus demonstrating that NW diameter can be controlled predictably. The use of well-defined catalyst nanoclusters was also used to show that NWs with a given length can be prepared by controlling the growth time.[23] Lastly, studies have shown that controlled p- and n-type doping of these semiconductor NWs, which is critical for most nanoelectronic and many photonic applications, can be readily achieved by simply introducing dopant elements into the reactant.[25,26]

Table 1 NW materials grown by the LCG approach. All of the NWs were synthesized using gold as the catalyst except for the case of GaN (see text)

Group IV-IV	Binary	Group III-V Ternary	Group II-VI (Binary)	Group IV-VI (Binary)
Si	GaN	Ga(As(i_x)Px)	ZnS	PbSe
Ge	GaP	In(As(i_x)Px)	ZnSe	PbTe
Si(i-x)Gex	GaAs	(Ga(i_x)Inx)P	CdS
\	InP	(Ga(i_x)Inx)As	CdSe
	InAs	(Ga(i_x)Inx)
		(AS(i_x)Px)

Fig. 6 Laser-assisted catalytic growth of GaN NWs. (A) Diffraction contrast TEM image of a GaN NW that terminates in a faceted nanoparticle of higher (darker) contrast. (Inset) convergent beam ED (CBED) pattern recorded along (0 0 1) zone axis. (B), (C) Lattice resolved TEM images of GaN NW with diameters of ca. 30 and 10 nm, respectively. The image was taken along (0 0 1) zone axis. 

CATALYTIC CHEMICAL VAPOR DEPOSITION GROWTH OF NANOWIRES

An alternative to the LCG implementation of catalytic NW growth is nanocluster-catalyzed CVD, in which the reactants and dopants are well-defined gas sources.[2] The catalytic CVD method enables the NW size, composition, and doping level to be controlled in a very precise manner.[24]

A clear illustration of the potential of this method is the controlled growth of Si NWs using distinct mono-disperse diameter Au nanoclusters as the catalyst and silane as the gaseous reactant source (Fig. 7).[24] Field emission scanning electron microscope (FE-SEM) images of the Si NWs grown from Au nanoclusters dispersed on SiO2 planar surfaces showed a comparable NW density to the starting nanocluster density (Fig. 7B). Qualitatively, these images also show that the Si NWs grown from Au nanoclusters are nearly monodisperse with diameters determined by the nano-clusters. This latter point was confirmed from TEM images (Fig. 7B, inset), which show directly that the Au particles at the NW ends are similar to the NW diameters. Histograms summarizing extensive TEM studies of the NW diameters obtained from different diameter nanoclusters also show that Si NWs grown from 5 (4.9 ± 1.0), 10 (9.7 ± 1.5), 20 (19.8 ± 2.0), and 30 nm (30.0 ± 3.0 nm) Au nanoclusters have average NW diameters of 6.4 ± 1.2, 12.3 ± 2.5, 20 ± 2.3, and 31.1 ± 2.7 nm, respectively. Significantly, the dispersion of the Si NW diameters mirrors that of Au catalysts, suggesting that the NW dispersity is limited only by the dispersity of Au nanocluster catalysts.

These studies also showed that the NW diameters were on average 1-2 nm larger than the catalyst sizes. This observation is consistent with the formation of an (Si, Au) alloy prior to the nucleation of NW growth. In addition, postgrowth oxidation of the Si NWs upon exposure to air could increase observed wire diameters. Overall, these results demonstrate the power of our size-controlled growth approach, and also show that it is possible to produce nearly monodisperse Si NWs with specific core diameters ranging from only 2 to more than 30 nm. In addition, the electronic properties of the Si NWs can be precisely controlled by introducing dopant gases during growth. For example, addition of different ratios of diborane or phosphane to silane reactant during growth produces p- or n-type Si NWs, respectively, with dopant concentration varied in a well-controlled way.[25]

Fig. 7 Catalytic CVD growth of Si NWs. (A) Schematic of size-controlled NW synthesis using monodisperse Au nanocluster catalysts. (B) Atomic force microscope (AFM) image of Au nanoclusters deposited on a silicon substrate, and the Si NWs grown by CVD. (C) Distributions of NW diameters obtained from 5, 10, 20, and 30 nm diameter nanoclusters.