Interfacial reaction growth approach to preparing patterned nanomaterials and beyond

Abstract

In conventional template-assisted synthesis, a template is used as a physical confiner to control the morphologies of the prepared materials. However, the template can also initiate interfacial reactions to form various patterned materials at its exposed surfaces. This concept can evolve into a general interfacial reaction growth approach in nanofabrications. In this tutorial review, we describe its fundamental issues, recent progress and achievements, and the properties and applications of the prepared nanomaterials before we summarize and give our personal perspectives.

---

Jiefeng Yu

Jiefeng Yu was born in Zunyi, Guizhou Province of China, and grew up in Shanghai and Dalian. In 2005 he earned a BS degree in Applied Chemistry from Shanghai Jiao Tong University. He is currently working towards a PhD degree in Materials Chemistry under the supervision of Prof. Kai Wu at Peking University.

Fan Wang

Fan Wang was born in Guangxi, China. He received a BS degree in Chemistry from East China University of Science and Technology in 1995 and a PhD degree with Professor Kai Wu from Peking University in 2004. He is now an associate Professor of Chemistry at Guangxi University in Nanning, China. His research interests include nano- and microfabrication, nanostructured materials, surface chemistry and functional inorganic materials.

Yu Wang

Yu Wang was born in China. He received his BE degree in Chemistry from Jinan University in 2000 and PhD degree from Peking University in 2007. During his PhD studies, he worked under the supervision of Prof. Kai Wu on the syntheses of functional nanomaterials through interfacial reaction growth and aluminothermal growth approach. Then he became a postdoctoral fellow with Prof. Yiying Wu at the Ohio State University in 2007. Currently, he is a research fellow at the Institute of Chemical and Engineering Science in Singapore. His research interests lie in the preparations and applications of functional materials.

Han Gao

Han Gao was born in Zhejiang, China. He received his PhD in physical chemistry from Zhejiang University. He was a postdoctoral fellow at Peking University with Prof. Kai Wu from 2000 through 2002. He is currently a senior research engineer in the Institute of Material Research and Engineering, Singapore.

Jianlong Li

Jianlong Li received his BS degree in Applied Physics from Dalian University of Technology, China, in 1992, and PhD degree in condensed matter physics from the Chinese Academy of Sciences in 2001. He was a postdoctoral fellow at Harvard University and the University of Texas at Austin from 2001 to 2006. He is currently an associate professor at the College of Chemistry and Molecular Engineering in Peking University.

Kai Wu

Kai Wu got his BS degree in chemistry from Zhejiang University in 1987 and PhD degree in physical chemistry from Dalian Institute of Chemical Physics in 1991 with Prof. Xiexian Guo. He worked as a researcher at the State Key Laboratory of Catalysis in that institute and became a postdoctoral fellow/associate with Prof. G. Ertl at Fritz-Haber-Institut in 1995 and with Dr J. P. Cowin at Pacific Northwest National Laboratory in 1998. In 2000, he joined Peking University in China where he is now a professor in chemistry. He is an editorial/advisory board member for several journals including Advanced Functional Materials. His research area is surface and materials science.

---

1. Introduction

Size matters. This is particularly true for nanomaterials that are essentially a bridge between bulk materials and atoms or molecules. Bulk materials possess well-defined physical properties which are insensitive to their size, but nanomaterials can display a wealth of novel properties as their size changes on the nanometric scale. The emerging novel properties of the nanomaterials originate from two main aspects: surface and quantum effects. As the material downsizes, more and more surface atoms are exposed. The surface properties previously overwhelmed by bulk materials become dominating in nanosystems. On the other hand, the physical properties may become quantized as the electrons are spatially confined in the quantum wells of the nanostructures. Both lead to the general acceptance that size variation of the nanostructures can provide one of the most powerful means to tune the electrical, optical and magnetic properties of the functional materials.

Many conventional approaches such as chemical vapor deposition (CVD), laser ablation, electrodeposition, thermal evaporation, facile solvent-phase syntheses, hydrothermal reaction and solvothermal reaction have been adopted for the fabrication of novel nanomaterials. Among these synthesis methods, template-assisted synthesis is a powerful one, which was pioneered by Martin in the early 1990s.¹ Although the templates have to be prepared in advance, template-assisted syntheses do have some advantages over the templateless ones in controlling the morphologies, dimensions and compositions of the prepared nanomaterials. Traditionally, in the template-assisted syntheses, templates are only utilized in a physical way, serving as scaffolds for shaping the morphologies of nanomaterials. Afterwards, the templates are selectively removed in order to obtain free-standing target nanostructures, though it is normally difficult. Alternatively, in the interfacial reaction growth (IRG) approach, templates not only serve as molds for materials duplication but also as a part of the starting reactants involved in the formation of products. In such a way, the whole template may be turned into the product, avoiding post-synthesis removal of the template. Obviously, this makes full use of the template.

The concept of interfacial reaction actually has roots in catalysis and surface science. In the 50-year history of modern surface science, researchers have been trying to characterize the surface and interfacial reactions in order to explore the mechanism of catalysis on the molecular level.² Great advances have been achieved in characterizing the active surfaces and studying interfacial reactions.³ However, it is only in the past decade that researchers introduced the concept of interfacial reaction into materials synthesis, and significant progress has been achieved in preparing patterned nanomaterials on either two-dimensional (2D) or one-dimensional (1D) templates. In this tutorial review, we describe some recent achievements of the IRG approach, discuss the fundamental issues, illustrate the properties of the prepared functional materials, and provide our perspectives of the IRG approach.

2. IRG approach to synthesizing nanomaterials

As schematically shown in Fig. 1, the IRG approach involves a template, normally a solid substrate. The interfacial reaction takes place at the substrate surface between the employed template and incoming precursor to yield a layer of new material, which can be either viewed as a target product or utilized as an intermediate buffer layer for subsequent fabrications of other materials. If the produced layer is the needed product, the chemical composition of the template has to be carefully chosen because in addition to its templating functionality, it also serves as a starting reactant. If the produced layer is further employed as an intermediate buffer layer or a secondary template for preparations of other materials, more parameters like lattice constant, surface free energy, and diffusivity have to be taken into account in order to obtain subsequent target products. Many 2D structures such as anodic aluminium oxide (AAO, also termed in the literature as porous anodic alumina, PAA, or simply as porous alumina, PA), anodic titanium oxide (ATO), porous silicon (PS) and individual/patterned 1D nanomaterials can well serve as the templates in the IRG approach.

Fig. 1 Schematic conceptual illustration of the IRG approach. (a) Incoming reacting species impinge and penetrate a substrate surface. (b) The incoming reacting species react with the substrate to form a solid layer (usually being polycrystalline) that may act as an intermediate buffer layer for subsequent growth of a second material. (c) If the lattice constant of the buffer layer well matches that of the subsequently prepared material, one may mimic the epitaxial growth at atmospheric pressure to grow a second layer of crystalline material. Obviously, both layers of the prepared materials should duplicate the surface morphology of the underlying substrate.

2.1 IRG on 2D porous hard templates

So far, the porous templates used mainly include AAO, ATO, PS, and some track-etched polymers. Track-etched polymers are commercially available for filtration and are limited in facile syntheses due to their poor thermal stability. Since high temperatures are routinely involved in the IRG process, AAO and ATO turn out to be ideal templates. In this review, we take AAO as a typical template used in the IRG approach for the sake of conceptual illustration. When used to prepare nanowires or nanotubes, the AAO template can be viewed as a 1D template; however, when employed to fabricate 2D nanonets (see below), it can be defined as a 2D template.

A large-area AAO template can be achieved via electrochemical anodization of a high purity aluminium sheet. The produced AAO template possesses hexagonally packed nanopores that are perpendicular to the template surface. The diameter, density, and length of the nanopores can be tuned by controlling the experimental parameters during the electrochemical anodization. To prepare well-ordered nanoporous structures in AAO, two or multiple steps of anodization are normally adopted. On such an AAO template with ordered nanopores inside, ordered nanodots, 1D nanoarrays and 2D nanonets can be prepared by various synthesis methods such as electrochemical deposition, sol–gel deposition and chemical vapor deposition (CVD). There are several excellent reviews describing the fabrication strategies using AAO as a physical template.^4,5 Here, we focus on the IRG approach using AAO as a reactive template.

The IRG approach could be used to synthesize various nanostructures of carbon,^6–8 semiconductors,^9–12 and metals.¹³ The conventional template-assisted synthesis approach only involves the interior surfaces of the nanopores in AAO.^4,5 In contrast, in the IRG approach, both the interior and exterior surfaces of the AAO template are utilized to fabricate 1D nanomaterials^6,7,9 and 2D nanonets,^9–11,13 respectively.

We first applied the IRG approach to prepare carbon nanotube (CNT) arrays on the AAO template.⁶ In combination with the conventional CVD approach, AAO served as both the confined template and a catalyst that cracked small hydrocarbons to form graphite layers covering all exposed interior and exterior surfaces of the AAO template. After the CVD growth, the exterior surface of the AAO template underwent ion milling to remove exterior carbon layers. After the ion milling and acid etching of the AAO template, open-ended CNTs grown inside the nanopores were exposed. These prepared CNTs were strictly parallel to each other and served as excellent field-emitters.⁶

Following the above procedure, Mu et al.⁷ developed a ubiquitous approach to fabricating uniform nanotube arrays by using AAO. As shown in Fig. 2, amorphous CNTs were initially grown at the interior surfaces of the nanopores in AAO by thermal cracking of ethyne, then one side of the as-prepared CNTs/AAO membrane was coated by a layer of gold as an electrode and subsequently nickel (Ni) was electrodeposited into the pores of the CNTs/AAO membrane to form nanowires inside. After heating in air, the CNTs were burnt out, and the Ni nanowires were oxidized into nickel oxide (NiO), which actually led to the formation of an annular channel inside each pore of the NiO/AAO/Au sample. In such a way, a series of metals could be electrodeposited into the annular nanochannels to form coaxial core–shell nanostructures. Finally, the NiO oxide core and AAO were chemically removed to obtain the metal nanotube arrays. This has been developed into a universal method to prepare metal nanotube arrays.

Fig. 2 Flow chart explaining the universal approach developed to produce nanotube arrays from the AAO template. The IRG approach is initially employed to grow CNTs at the interior surfaces of the nanopores inside the AAO template. Details for subsequent processes are given in the main text. ↑ means combustion of CNT in air or dissolution of AAO and NiO in acid. ↓ means vapor or electrochemical deposition of metals. [O] represents oxidation of the deposited Ni in air. See the main text for details.

On the AAO template, simple physical deposition suffers from the line-of-sight in coating the interior surfaces and normally results in a poor uniformity of the prepared nanostructures with a high aspect ratio. However, by using AAO as a reactive template in the IRG approach, highly ordered and uniform crystalline nanotube arrays can be routinely obtained. For example, when the AAO template was buried in the precursor, ZnO powder, and thermally treated by a steady H₂ flow at about 680 °C in a CVD furnace, a seamlessly integrated zinc aluminate (ZnAl₂O₄) nanotube/nanonet structure (Fig. 3a–d) was produced at both interior and exterior surfaces of the template.⁹ The proposed interfacial reactions are as below:

H₂(g) + ZnO(s) → H₂O(g) + Zn(g) (1)

Zn(g) + 1/2O₂(g) + Al₂O₃(s) → ZnAl₂O₄(s) (2)

Zn(g) + H₂O(g) + Al₂O₃(s) → ZnAl₂O₄(s) + H₂(g) (3)

In the CVD furnace, the Zn vapor generated by thermal reduction of ZnO in H₂ reacted with Al₂O₃ (AAO) to form ZnAl₂O₄ in the presence of residual oxygen or formed water. The dimensional parameters of the produced ZnAl₂O₄ nanotubes were dictated by the pore diameter and inter-pore distance of the AAO template, and the thicknesses of the ZnAl₂O₄ nanotube wall and nanonet, by the growth time. From the growth kinetics curve shown in Fig. 3e, one can see that at about 680 °C, the nanonet thickness (or nanotube wall thickness) rapidly increased with the growth time in the initial stage, then gradually augmented and finally reached its saturation value at 3000 min. This indicates a clear diffusion-limited growth mechanism for the ZnAl₂O₄ nanotubes/nanonet. As schematically shown in Fig. 3f, there establishes a concentration gradient of the incoming Zn vapor from outside to deep into the amorphous AAO template. Therefore, the stoichiometric ratio of Zn, Al, and O in the formed compounds in different regions is expected to be different. As a consequence, large-area crystalline Zn_xAl_yO_z (x = 4, 1, 1; y = 22, 2, 0; z = 37, 4, 1) nanonets, namely, Zn₄Al₂₂O₃₇, ZnAl₂O₄, and ZnO, have been successfully prepared and identified (Fig. 4).¹⁰ Note that ZnS was here used as the precursor, and the same products were obtained as with those from the ZnO precursor, further supporting that Zn vapor diffusion is indeed the limiting step in the IRG approach. Experiments showed that at the front of the diffusing Zn vapor inside the AAO template, a Zn₄Al₂₂O₃₇ nanonet (Fig. 4e) was formed because the Zn concentration was low; near the interface region produced were the normal ZnAl₂O₄ nanotube and nanonet (Fig. 4a, b and f); continuous supply of the Zn vapor led to the formation of pure ZnO nanonet (Fig. 4c, g and h) on top of the normal ZnAl₂O₄ layer because the Al₂O₃ species could not back-diffuse towards the outside of the AAO template. The sequentially grown materials are conceptually illustrated in Fig. 4d where materials grown at the interior surfaces of the nanopores in AAO are omitted for the sake of clarity.

Fig. 3 (a) Large-scale and (b) enlarged SEM images of the prepared free-standing ZnAl₂O₄ nanotube array (AAO already removed). Inset in (b) is a close-up of the nanotubes. (c) Upside and (d) upside-down SEM sideview images of the samples. The drawings and arrows in (c) and (d) show the nanotube orientations and the ZnAl₂O₄ nanonet, respectively. The precursor used is ZnO. Note that the scale bar in (a) should be 5 μm rather than the mistyped 5 nm in ref. 9. (e) Thickness of the nanotube wall or the nanonet as a function of the growth time at about 680 °C for the seamlessly jointed ZnAl₂O₄ nanotube/nanonet on AAO. (f) Schematic drawing showing the established Zn concentration gradient, d[C_Zn]/dx, from the exposed surface to deep inside bulk AAO. X is the position axis, and x = 0 represents the vapor–solid interface. The precursor used is ZnO (© The American Chemical Society, reprinted with permission).⁹

Fig. 4 TEM images of the prepared (a) Zn₄Al₂₂O₃₇ nanotube plus ZnAl₂O₄ nanotube array/nanonet, (b) ZnAl₂O₄ nanonet and (c) ZnO nanonet on top of the ZnAl₂O₄ nanonet. The precursor used is ZnS. Note that the pore diameter of the ZnO nanonet is much larger than that of the ZnAl₂O₄ nanonet because of the pore enlargement effect during the acid etching of the underlying AAO template and zinc aluminate nanotube/nanonet. (d) Experimentally tested growth model for the Zn_xAl_yO_z nanostructures on AAO. (e)–(h) HRTEM images showing the lattice structures of the prepared hcp Zn₄Al₂₂O₃₇ nanonet, fcc ZnAl₂O₄ nanonet, hcp ZnO nanonets along the [001] and [011] directions on top of the ZnAl₂O₄ nanonet, respectively. All insets in (e)–(h) are the corresponding SAED patterns. The precursor used is ZnS. Note that in all structures, there exists a lattice constant of ∼0.28 nm, indicating the good lattice matching of the crystals (©Wiley-VCH Verlag GmbH & Co. KGaA, reprinted with permission).¹⁰

It is interesting to note that a common lattice parameter, 0.28 nm, exists in the produced face-centered cubic (fcc) Zn₄Al₂₂O₃₇, hexagonally close-packed (hcp) ZnAl₂O₄ and hcp ZnO nanonets (Fig. 4e–h). The lattice mismatch is less than 2%. This suggests that the normal hcp ZnAl₂O₄ nanonet produced at the exterior surface may serve as an intermediate buffer layer so that an epitaxial growth mechanism under ambient conditions may be mimicked for the growth of the ZnO nanonet atop. This is indeed verified by experiments. By varying the growth temperature, the ZnO nanonet could be grown along either the [001] or the [011] direction (Fig. 4g and h).¹⁰ This case study presents an excellent example showing that the morphology (nanotube or nanonet), lattice structure (hcp or fcc), chemical composition (Zn₄Al₂₂O₃₇, ZnAl₂O₄ or ZnO) and growth directions ([001] or [011]) can be simultaneously controlled and tuned by simply changing the growth temperature during the IRG process.

The above approach can be used in different ways. Preparations of various materials on AAO under various experimental conditions have been previously reported.¹² Single crystalline ZnTe nanowires, core–shell structures of single crystalline Te nanowires surrounded by polycrystalline ZnAl₂O₄ shells (Te/ZnAl₂O₄), pure single crystalline Te nanowires, and polycrystalline ZnAl₂O₄ nanotubes have been achieved by using [Zn(TePh)₂(tmeda)] (tmeda = N,N,N′,N′-tetramethylethylenediamine) as the sole precursor. Under vacuum conditions, thermolysis of Zn(TePh)₂(tmeda) in the AAO pores yielded single crystalline ZnTe nanowires that subsequently reacted with residual oxygen to generate ZnO and Te nanowires, and the freshly prepared ZnO reacted with the AAO template to form polycrystalline ZnAl₂O₄ shells. After selective etching of the Te core in alkali solution or vigorous sonication removal of the ZnAl₂O₄ shell, separated ZnAl₂O₄ nanotubes and Te nanowires were obtained.

To further extend the applicability of the IRG approach, we have prepared metallic nanonets.¹³ There are two general approaches for the preparations of 2D metallic nanonets. One is the “top-down” approach that uses a focused ion beam (FIB) to drill nanoholes in a metallic sheet. This is obviously time-consuming and cost-inefficient. The other is the “bottom-up” approach to depositing a metal layer on top of a porous template, which normally encounters the problem of shrinking into small nanoparticles upon annealing due to the surface energy difference. This trick has been extensively used in the vapor–liquid–solid (VLS) approach where metal nanoparticles are needed to serve as catalysts for the growth of CNTs, 1D nanowires, etc. Our developed IRG approach may overcome the above problem to obtain 2D crystalline metallic nanonets by first depositing an oxide layer on top of the AAO template. One may then use the deposited oxide layer as an intermediate that is slowly reduced into a metal layer duplicating the exterior surface morphology of the AAO in a reducing atmosphere and at a high temperature. Typically, an oxide may well spread over another oxide surface and a slow temperature elevation may not cause a turbulent shrinkage of the top oxide layer. In such a way, large-area metal oxide nanonets can be achieved on AAO. Crystalline body-centered cubic (bcc) Mo and fcc Cu nanonets have been prepared in such a way on the AAO template.¹³

Chemically decorated AAO templates can also be utilized in the IRG approach. Chiu’s group has introduced sodium nanotubes inside the AAO template, denoted as Na@AAO, by using the thermally decomposable NaH as a reactive template to fabricate CNTs (see, for example, Fig. 5)⁸ and 1D TiO₂ nanostructures.¹⁴ Na@AAO reacts with the vapor of chlorides which contain chemical elements of the target products (e.g. hexachlorobenzene for CNTs, and TiCl₄ for TiO₂) to yield 1D nanostructures and NaCl byproduct in the AAO nanopores. NaCl can be easily washed off by an aqueous solution, or vaporized in vacuum at high temperatures. The prepared composite nanostructures can be directly turned into the product or further oxidized by O₂ to form 1D oxide nanostructures. Apparently, the IRG approach provides an inexpensive but efficient way to fabricate various nanostructures including carbon, oxide, complex oxide and metal nanotubes/nanonets. Since these nanostructures grow under the assistance of the underlying AAO template, their pore size and interpore distance can be easily tuned by changing the electrochemical parameters during the fabrication of the AAO template. Such advantages stem from the merits of the AAO template, its ubiquitous applicability and its limitations as well. A freshly prepared AAO template is amorphous in nature and possesses a high reactivity towards other reactants. If the growth temperature is higher than 800 °C, the amorphous AAO will crystallize into γ-Al₂O₃ or δ-Al₂O₃ that displays a low reactivity to form complex oxides at its exposed interfaces because crystallized structures can seriously prevent the incoming reactive species from diffusing into bulk γ-Al₂O₃ or δ-Al₂O₃ and, hence, prevent subsequent reactions from happening. Therefore, a typical growth temperature in the IRG approach should be lower than 800 °C. This temperature obviously restricts the employment of precursors with a low vapor pressure below 800 °C, and limits the applicability of the IRG approach. However, this limitation can be overcome by replacement of the AAO template with some other porous templates such as PS, ATO, mesoporous materials and many others. Similarly, other 2D templates can also be chemically decorated to deposit metal nanostructures.¹⁵ Martin’s group has prepared Au nanotubes and films at both interior and exterior surfaces of track-etched polycarbonate membranes. A catalyst was attached to all exposed surfaces of the template which was subsequently immersed into a solution containing Au(I) species and a reductant. Since the reactions took place at the surfaces covered with the catalyst, Au nanotubes and films were therefore successfully fabricated (Fig. 6a). Obviously, the thicknesses of the Au nanotube walls and film can be simply controlled by the electroless plating time (Fig. 6b).¹⁶

Fig. 5 (a) An illustrating scheme showing the reactions between the AAO template and the NaH and hexachlorobenzene precursors to form Na@AAO and C@AAO. (b) TEM image of an open-ended carbon nanotube prepared at 623 K and annealed at 1073 K. (c) Sideview SEM images of (c) the carbon nanotubes inside AAO pores and (d) the carbon nanotube bundles after removal of the AAO template (© Royal Society of Chemistry, reprinted with permission).⁸

Fig. 6 (a) Schematic drawing of the electroless plating process employed to fabricate the Au nanotube membranes. (b) The effective inner diameter of the prepared nanotube as a function of the electroless plating time (© Wiley-VCH Verlag GmbH, reprinted with permission).¹⁶

The interior surfaces of the freshly prepared AAO templates can be tethered by various anions during the anodization process and have an oxidation tendency towards some reactive species. In contrast, the exposed surfaces of the PS template show a reduction capability in the IRG process.¹⁷ Three different materials—SnSe, Sn and SnO₂—from the same precursor source, Sn(SePh)₄, have been fabricated in the pores of PS and AAO due to the different redox properties of the templates used. SnSe nanowires, nanotubes or microtubes can be obtained by thermolysis of Sn(SePh)₄ infiltrated into the pores of AAO or PS. Further annealing of the SnSe nanowires in AAO anodized in aqueous sulfuric acid yields single crystalline SnO₂ nanowires due to the oxidization tendency of the interior surfaces of the AAO template. However, single crystalline Sn microtubes are prepared from the reducible SnSe microtubes inside PS due to the reduction capability of the interior surfaces in PS. This is an example showing that the interfacial reactions between the same reactive precursor and different templates of distinct redox properties can generate different target products.

Furthermore, crystalline silica microtubes of quartz, tridymite, and cristobalite can be fabricated by thermal oxidation of the PS template coupled with Li-containing species as catalysts.¹⁸ The unknown Li-containing species was involved in the interfacial reaction of the Si surface and oxygen to control the crystallization of SiO₂.

Another example illustrating the oxidization capability of the AAO interior surface¹⁹ is the preparation of a Pt@CoAl₂O₄ peapod nanostructure. This structure consisting of the Pt nanoparticles encapsulated in CoAl₂O₄ nanotube shells was realized by pulsed potential electrodeposition of Co/Pt multilayered nanowires into AAO and subsequent annealing. Two interfacial reactions presumably took place during the annealing process: (1) partial oxidation of Co at the surface of AAO to CoO by reaction with the oxygen released from the AAO interior surface, and (2) combination reaction between CoO and Al₂O₃ at the interface of the Co-containing segments and the interior surface of AAO. The Pt@CoAl₂O₄ peapod nanostructure was formed by the combination of the Co species diffused into the AAO template. No reaction involving Pt took place during the annealing process. Such a process can also happen on the ATO template. Schmuki and co-workers have fabricated an ordered lead titanate nanocellular structure by annealing electrodeposited Pb on the ATO template.²⁰ A Ti substrate was first anodized to generate a TiO₂ nanotube layer (Fig. 7a) that was subsequently filled by Pb (Fig. 7b) to form the Pb wire/TiO₂ nanotube composite material. Afterwards, the composite material was annealed to trigger the interfacial reaction to produce a PbTiO₃ nanotube array (Fig. 7c) that could be directly peeled off the Ti substrate to yield a free-standing PbTiO₃ nanocellular layer (Fig. 7d). The yielded PbTiO₃ nanocellular layer was polycrystalline in nature (Fig. 7e).

Fig. 7 SEM images of (a) TiO₂ nanotube layer, (b) the same layer after filling with Pb; (c) the Pb-filled TiO₂ nanotube layers after annealing for 3 h at 550 °C, and (d) a piece of the PbTiO₃ nanocellular layer peeled off the Ti substrate. (e) The SAED ring pattern taken from the walls of the cells shown in (d) (© Wiley-VCH Verlag GmbH & Co. KGaA, reprinted with permission).²⁰

In addition to the hard templates such as AAO, ATO and PS, meso- or macro-porous substrates can also be employed in the IRG approach. Thomas and co-workers have synthesized metal nitride nanoparticles by using mesoporous carbon nitride as the template (Fig. 8a).^21–23 The carbon nitride template served as the supplying source of nitrogen. When different precursors were used, various binary and ternary metal nitride nanoparticles could be routinely obtained.^22,23 Even further, macroporous carbon nitride can be turned into pure TiN (Fig. 8b) or TiN/carbon (Fig. 8c) composite materials, depending on the molar ratio of the template to metal and the metal dispersion as well.²⁴ The whole process for the interfacial reactions is schematically shown by Fig. 8d.

Fig. 8 TEM images of (a) the pore walls of mesoporous C₃N₄, and the TiN nanoparticles synthesized in mesoporous C₃N₄ using (b) diluted and (c) concentrated precursor solutions. (d) Schematic drawing of the formation of the macroporous TiN/carbon composite from macroporous C₃N₄ (© Wiley-VCH Verlag GmbH & Co. KGaA and The American Chemical Society, reprinted with permission).^22,24

2.2 IRG on 1D templates

Ever since the discovery of CNTs in 1991, many 1D nanomaterials have been fabricated. CNTs and other inorganic nanowires can also serve as solid templates in the IRG approach to fabricating 1D nanomaterials such as nanowires, nanotubes and 1D core–shell nanostructures. Their diameter and length are dictated by the outer diameter and length of the original 1D template. The interfacial reactions can be conducted at solid–solid, solid–liquid, and solid–gas interfaces by means of atomic layer deposition (ALD), facile solvent-phase synthesis and CVD.

In addition to the role as a scaffold, the CNTs grown by the solid–vapor interfacial reaction were used as a starting reactant to supply carbon for subsequent products²⁵ or as a reduction agent.^26,27 Solid carbide nanorods of single crystalline TiC, polycrystalline NbC, amorphous Fe₃C, single crystalline SiC, and polycrystalline BC_x were prepared by the reaction of CNTs with corresponding volatile oxides or halides.²⁵ The similarity of the diameters of the resulting carbide nanorods and the starting CNTs implied that the CNTs acted not only as reagents in the reactions but also as templates to confine the geometric morphologies of the prepared nanomaterials. By using the same synthetic strategy and a second vaporized reactant, single crystalline GaN²⁶ and GaP²⁷ nanorods have been successfully prepared. In these reactions, the CNTs actually served as a sacrificial template reductant in the reactions and turned into byproduct-CO₂ gas which made the diameters of the prepared GaN and GaP nanorods slightly deviated from those of the CNT templates.

Using 1D nanowires as reactive templates is also a powerful method to fabricate various 1D nanostructures. Fan et al.²⁸ have developed a fabrication strategy for preparing silica nanotubes arrays on silicon nanowire array templates. Silicon nanowire arrays were produced via the VLS mechanism. Uniform and continuous silica shells were generated around the silicon nanowire cores through the interfacial oxidation reaction of the silicon nanowire surface and oxygen at high temperatures. After chemical etching of the silicon nanowire cores, the silica nanotube arrays were fabricated. Wu et al.²⁹ have prepared polycrystalline MgB₂ nanowires on amorphous boron nanowire templates. Mg vapor could easily diffuse into amorphous boron nanowires to react and transform boron nanowires into MgB₂.

In the IRG approach, the prepared nanomaterials at the solid–liquid interfaces of the 1D templates are normally polycrystalline. However, there are several examples showing that single crystalline nanowires can be prepared on the single crystalline nanowire templates. Gates et al.³⁰ reported the first example of such single crystalline nanowires by template-assisted growth in a solution phase at room temperature. Single crystalline Ag₂Se was prepared by an interfacial reaction between the single crystalline t-Se nanowires and AgNO₃ solution. The growth mechanism was proposed as follows: Ag⁺ catalyzed the disproportionation reaction of Se⁰ into Se²⁻ and SeO₃²⁻, and Se²⁻ species reacted with Ag⁺ to form Ag₂Se while SeO₃²⁻ species left the templates as a byproduct. During all these reactions, the atomic structure of the t-Se template largely remained intact. However, if the single crystalline t-Se nanowire template was replaced by a bi-crystalline Ag nanowire that further reacted with a Se colloid, a polycrystalline Ag₂Se nanotube formed.³¹ The only difference was that it took a much longer time to complete the whole process. A possible growth scenario was that the Se colloid oxidized Ag into Ag⁺ which migrated away and broke the crystal lattice of the bi-crystalline Ag nanowire template to form a polycrystalline Ag₂Se nanotube. The above results show that the single crystalline nature of the 1D template is a pre-requisite for the preparation of single crystalline nanomaterials. Again, by using the t-Se nanowire template, a Se/CdSe nanocable can be fabricated by the reaction of the t-Se nanowire and the Cd²⁺ cations, and a CdSe nanotube survives the removal of the unreacted inner core of the t-Se template.³²

Song et al.³³ have developed a strategy to prepare continuous polycrystalline noble metal nanowires of Au, Ag, and Pt by using the reducing LiMo₃Se₃ molecular wire bundles as templates. A series of interfacial reactions took place: the LiMo₃Se₃ nanowire bundles reduced aqueous noble metal ions such as AuCl₄⁻, Ag⁺ and PtCl₄⁻. The metal atoms produced by reduction of the ions deposited in situ onto the LiMo₃Se₃ nanowire bundles, while the LiMo₃Se₃ templates were oxidized into soluble Mo₃Se₃. Finally, polycrystalline noble metal nanowires were produced. Similarly, Ag nanowire templates reacted with aqueous precursors such as AuCl₄⁻, Pt²⁺ and Pd²⁺ to prepare noble metal nanotubes of Au, Pt and Pd.³⁴ The reduction reactions happened at the solid–liquid interface so that the produced noble metal atoms nucleated and grew into a thin shell around the templates. Ag⁺ and noble metal ions could penetrate the shell until the Ag templates were thoroughly consumed. Consequently, noble metal nanotubes with seamless walls could be obtained. This synthetic strategy is applicable for the preparation of shape-controlled nanoparticles by controlling the morphologies of the templates used. It is noticed that single crystalline Ag templates often lead to single crystalline shells of the noble metal products. Besides the element template, cation exchange reactions also happen with compound templates.^35,36 CdSe nanocrystals completely react with Ag⁺ ions to turn into Ag₂Se nanocrystals that can be converted back into CdSe nanocrystals again by a reverse cation exchange reaction.³⁵ With partial cation exchange, CdS–Ag₂S nanorod superlattices rather than CdS–Ag₂S core–shell nanostructures can be yielded on CdS nanorod templates.³⁶

In addition to the inorganic nanostructures mentioned above, organic nanotubes may also be synthesized via the IRG approach. Pan et al.³⁷ reported the synthesis of polyaniline (PAni) nanotubes on manganese oxide nanowire templates which served as a chemically oxidative initiator for the polymerization of aniline. Different from the chemical polymerization of PAni on physical templates such as track-etched polymers and AAO, aniline was oxidatively polymerized at the surface of the manganese oxide nanowire template, and the manganese oxide is consumptively reduced to soluble Mn²⁺ ions that could be removed after the polymerization. Based on this interfacial reaction, double-shelled nanotubes of PAni could also be obtained by using the manganese oxide nanotubes as a secondary template. Aniline was polymerized at both the exterior and interior surfaces of the manganese oxide nanotube template, leading to product that fully duplicated the surface morphology of the nanotube template. With other suitable polymerization precursors, this methodology might be routinely adopted to fabricate various nanostructures of conjugated polymers that have great potential applications in organic nanodevices.

In applications of the IRG approach involving solid–solid interfaces, the employed nanowire templates are usually coated with a layer of conformal sheath of a second material to fabricate core–shell nanocable structures by magnetron sputter deposition,³⁸ metalorganic chemical vapor deposition (MOCVD)³⁹ or ALD.^40,41 A subsequent annealing process triggers the solid–solid interfacial reactions and forms complex oxide nanowires^38,39 or nanotubes^40,41 depending on the diffusivities of the involved reacting species. If the outward-diffusion rate of the inner core material in the interfacial reaction overtakes the inward-diffusion rate of the outer shell material, the so-called Kirkendall effect will come into play, which eventually leads to hollow nanotubes. On the contrary, if the inward-diffusion rate of the outer shell material surpasses, solid nanowires will be prepared. Let us take the formation of Zn₂TiO₄ nanowires as an example.³⁸ Pre-prepared single-crystalline ZnO nanowires were coated with an inhomogeneous layer of amorphous Ti by magnetron sputter deposition. The sputtering source was adjusted to a Ti : Zn atomic ratio of 1 : 2. During the annealing process in a vacuum, Ti reacted with oxygen supplied by both the ZnO nanowires and the residual gas to form TiO_x wrapping the ZnO nanowires. The diffusion rate of Ti into ZnO overtook that of Zn into TiO_x, resulting in a net diffusion of Ti into the ZnO lattice. The ZnO nanowire templates spatially restricted the interfacial reaction and turned into Zn₂TiO₄ nanowires.

MOCVD can be applied to prepare core–shell nanocable structures. Chang and Wu have prepared Ga₂O₃/ZnO core–shell nanowires with a single crystalline Ga₂O₃ core and an MOCVD-grown nanocrystalline ZnO shell.³⁹ The thickness of the grown ZnO shell and the thermal budget of the annealing process played important roles in the production of various nanomaterials. Post-annealing initially helped complete the interfacial reaction between Ga₂O₃ and ZnO to form ZnGa₂O₄. Subsequent products included Ga₂O₃/ZnGa₂O₄ core–shell nanowires, single crystalline ZnGa₂O₄ nanowires, and single crystalline ZnGa₂O₄ nanowires inlaid with ZnO nanocrystals by variation of the ZnO shell thickness in the Ga₂O₃/ZnO core–shell nanowire precursor.

ALD is another powerful approach to fabricating core–shell nanostructures. It actually belongs to the IRG approach because the interfacial reactions between the incoming vapor molecules and the solid substrate surface take place in a cyclic manner. Since Leskelä and Ritala have written an excellent minireview about the ALD approach,⁴² here we only give a brief introduction and two fantastic examples. In a typical ALD growth cycle, two vapor precursors are inlet alternately into the reaction chamber. The strict separation of the precursors from each other establishes the self-limiting surface reactions during the film growth, which means the amount of the film material deposited in each reaction cycle is constant. The self-limiting film growth mechanism allows ALD to control film thickness at the atomic level. Fan et al.⁴⁰ have prepared hollow ZnAl₂O₄ nanotubes on 1D ZnO nanowires. The whole process (Fig. 9b) is similar to that for the preparation of the Zn_xAl_yO_z nanotubes/nanonets^9,10 mentioned above: a ZnO nanowire is coated by a thin layer of Al₂O₃via ALD to form the Al₂O₃/ZnO core–shell structure (Fig. 9a) that is further annealed to initiate the interfacial reaction to produce ZnAl₂O₄ nanotubes (Fig. 9c) with the help of the Kirkendall effect. Ideally, nanotubes with a cylindrical inner void and a single-phase outer wall would be obtained by exploiting the Kirkendall effect without excess reactants left after the solid–solid interfacial reaction, which requires a proper thickness match between the core and shell. Usually, it is difficult to obtain uniform ZnO nanowires via the VLS mechanism because of the size distribution of gold catalyst nanoparticles that can thermally aggregate. ALD can, however, assure a uniform layer of Al₂O₃ coating the prepared ZnO nanowires on the atomic scale. Formation of the uniform ZnAl₂O₄ nanotubes suggests that an optimized diameter of the ZnO nanowires could be achieved. A similar strategy has been successfully applied to prepare MgAl₂O₄ spinel nanotubes through the interfacial reaction in MgO/Al₂O₃ core–shell nanocables.⁴¹

Fig. 9 (a) TEM image of a ZnO/Al₂O₃ core–shell nanowire. (b) Schematic illustration of the transformation process from the initial ZnO nanowire to the Al₂O₃/ZnO core–shell nanowire, and to the final ZnAl₂O₄ nanotube. Interfacial reaction occurs during the annealing of the Al₂O₃/ZnO core–shell nanowire. (c) TEM image of the prepared ZnAl₂O₄ nanotubes (© Nature Publishing Group, reprinted with permission).⁴⁰

It is noteworthy to point out that patterned nanomaterials can obviously be fabricated if the employed 1D templates are pre-patterned into nanowire or nanotube arrays.

3. Morphology and crystallinity controls

In the previous sections, we have summarized some of the main achievements in materials synthesis by the IRG approach. In the following, we will deal with the morphology and crystallinity controls of the nanomaterials fabricated by the IRG approach.

3.1 On 2D AAO template

1D nanomaterials synthesized by traditional liquid-phase approaches such as sol–gel deposition and electrochemical deposition on AAO normally possess poor crystallinity. With the IRG approach, single crystalline 1D nanowires^12,17 and nanotubes⁹ and highly crystalline 2D nanonets^10,11,13 can be routinely fabricated on the amorphous AAO template. The sole factor preventing the fabrication of single crystalline 2D nanonets relates to the AAO template structure rather than the IRG approach itself. Laboratory-prepared AAO is produced by electrochemical anodization of a high purity Al sheet whose crystal domain size determines the achievable domain size of the nanopores, typically 1–3 μm in size, in AAO. The domain size of the hexagonally packed nanopores in AAO accordingly limits the crystal domain size of the prepared 2D nanonets. Needless to say, one may anticipate that single crystalline 2D nanonets could be obtained with an AAO template prepared by anodization of a single crystalline Al sheet, though the cost is unbearable.

The amorphous nature of the AAO template can facilitate the diffusion of the incoming vapor precursor to form highly crystalline nanomaterials at a high temperature. For instance, Zn vapor generated by reduction of ZnO or ZnS in H₂ can feasibly penetrate the amorphous AAO template and react with the template to form a highly crystalline ZnAl₂O₄ nanonet^9,10 and a single crystalline ZnAl₂O₄ nanotube array.⁹ In contrast, a ZnAl₂O₄ film with a sub-micron domain size could only be achieved by the interfacial reaction between the Zn-containing vapor generated by carbothermal reduction of ZnO and the c-sapphire substrate at a much higher temperature.⁴³ The reason is that the single crystalline nature of the c-sapphire substrate makes it difficult for the vapor precursor to penetrate, so only small domains of ZnAl₂O₄ are produced on the c-sapphire substrate. On amorphous AAO templates, polycrystalline¹² and nanocrystalline¹⁹ structures can be easily fabricated at the solid–solid interfaces by the IRG approach.

3.2 On 1D templates

At the solid–vapor interfaces, single crystalline nanomaterials may also be produced on amorphous templates by the IRG approach. However, 1D single crystalline nanomaterials can only be fabricated on 1D single crystalline templates at the solid–solid^38–41 and solid–liquid^30,34 interfaces at low temperatures. Generally, single crystalline products are hardly achievable at the solid–liquid interfaces in a solution environment at room temperature, except for that a minimal reconstruction of the crystal lattice is required in the template.³⁰ This is the case for the preparation of single crystalline Ag₂Se nanowires. Single crystalline Ag₂Se nanowires can be prepared on single crystalline t-Se nanowire templates (lattice constants: a = b = 0.698 nm, and c = 0.496 nm) in AgNO₃ solution at room temperature. The size-dependent crystal structure of the Ag₂Se nanowires was experimentally observed. When the t-Se nanowire diameter was smaller than ∼40 nm, tetragonal Ag₂Se nanowires (lattice constants: a = 0.436 nm, and c = 0.495 nm) were formed while orthorhombic Ag₂Se nanowires (lattice constants: a = 0.705 nm, b = 0.782 nm, and c = 0.434 nm) were produced when the t-Se nanowire diameter was larger than 40 nm. One can immediately identify that the corresponding lattice constants of the c- and a-axis for the tetragonal and orthorhombic Ag₂Se nanowires hardly change, so the single crystalline t-Se templates are highly adaptable in this case. This phenomenon may also apply for other 1D templates.

To fabricate 1D single crystalline nanowires^38,39 or nanotubes^40,41 at the solid–solid interfaces, 1D single crystalline nanowire templates have to be employed. There are many reports concerning solid–solid reactions for bulk materials that could be used as a source of references for the IRG approach. However, in contrast to the solid–solid reaction of bulk materials, the much lower diffusivity of the vapor precursor in the single crystalline solid template could greatly devalue the IRG approach.

4. Applications of the patterned nanomaterials prepared by the IRG approach

The IRG approach can produce various types of low-dimensional functional nanomaterials that may find wide applications including laser, field emitter, chemical sensor, superconductor, surface-enhanced Raman spectroscopy and so on and so forth. Nevertheless, most reports in the literature simply focus on the synthesis strategy rather than the applications of the prepared materials, although 1D nanowires and nanotubes possess more extraordinary electrical, optical, thermoelectrical and magnetic properties than their bulk counterparts. For applications depending on the collective behavior of regular structures on the nanometric scale, the IRG approach with AAO is an efficient method to fabricate 2D functional nanotube arrays and nanonets. In the following, we present three examples of the applications of low dimensional nanomaterials prepared by the IRG approach: field emitter, laser, and molecular filter and sensor.

4.1 CNT field emitter

CNTs have many fascinating properties such as a concentric planar wall structure, controllable nanometric diameter, high aspect ratio, chemical inertness, and extraordinary mechanical strength. These make them a good candidate for high-performance field-emitters consuming much less energy. But processing individual carbon nanotubes into an applicable large-area field emitter is a big challenge. To bypass the difficult post-processing, an ordered array of graphitized CNTs can be prepared by cracking of ethylene on AAO.⁶ With proper after-treatments, a large area CNT field emitter with a low turn-on field and a large and reproducible emission current density is produced, which makes use of the IRG approach to prepare large area, graphitized and well-patterned CNTs. The samples prepared by the IRG approach can easily reach an emission current density of 10 mA cm⁻², which is a basic requirement for a flat panel display. The measured turn-on field (E_to) is 2.8 V μm⁻¹ for an emission current density of 10 μA cm⁻². A current density of 24 mA cm⁻² is achieved for 30 min at an applied field of 12 V mm⁻¹, and the emission current density can be stable for more than 18 h (the current density fluctuation being less than 2% for ∼5 mA cm⁻² at 8 V μm⁻¹). In comparison with the field emission properties of previously reported CNTs/AAO field-emitters with a high threshold field and a low current density, the graphitization of the CNTs plays an important role in enhancing the field emission: on the one hand, the graphitized CNTs can bear a larger emission current with less damage and hence betters their stability. On the other hand, the graphitized CNTs and their good ohmic contact with the underlying electrode can provide efficient conducting channels to transport electrons.

4.2 ZnO nanonet laser

ZnO is thought to be a good lasing material at room temperature due to its broad band gap (3.37 eV) and high exciton binding energy (60 meV). The lasing properties of various ZnO nanomaterials such as random particles, 1D nanowires and 2D films have been explored. A major concern is how to prepare single crystalline ZnO structures with fewer defects that may disturb their light emission. The intrinsic band gap of ZnO restricts the wavelength of its emitted light unadjustably centered around 370–420 nm. To get around this restriction, people have been trying to prepare 2D ZnO photonic crystals whose optical band gaps could be tuned by controlling their structural parameters. In such a way, the wavelength of the emitted light could be adjusted. To do this, one may use the ion-drilling technique to punch periodic holes in an epitaxially grown ZnO film on sapphire.⁴⁴ Apparently, this is laborious, time-consuming and cost-expensive in preparing large area ZnO photonic crystals.

In contrast, the IRG approach can be applied to easily prepare large-area (from millimetres to centimetres) crystalline ZnO nanonets with periodic nanopores on AAO, as shown in Fig. 10.¹¹ A polycrystalline ZnAl₂O₄ nanonet initially grows via a gas–solid interfacial reaction on the AAO exterior surface, then a ZnO nanonet is “epitaxially” grown atop because of the good lattice matching of ZnAl₂O₄ and ZnO crystals, as mentioned before. As can be seen in Fig. 10a, the ZnO/ZnAl₂O₄/AAO has a unique structure that determines its lasing behavior. First, a continuous polycrystalline ZnO nanonet with periodic nanopores is formed on amorphous AAO, which can suppress defect-trapping of the electrons to a certain extent. Second, the as-prepared ZnO nanonet is about 1.5 μm in thickness, much longer than its emitted light wavelength (about 387–390 nm). Third, two sharp interfaces are naturally established: at the top is the air/ZnO interface and at the bottom the ZnO/ZnAl₂O₄ interface. Since the refractive indices for air, ZnO, and ZnAl₂O₄ are 1, 2.45, and 1.77, respectively, these sharp interfaces can therefore form a good cavity (Fig. 10b). The produced photons in the nanonet can traverse back and forth between the two interfaces to achieve substantial gain before they escape the cavity. All these factors make this nanonet a good laser candidate.

Fig. 10 (a) Scheme of the preparation procedure of the ZnO nanonet, starting from the amorphous AAO template. (b) Model of the prepared ZnO/ZnAl₂O₄/AAO composite sample used for the PL experiment, with their thicknesses marked beside each layer. The refractive indices for all species including air are given in the model, showing the formation of sharp cavity interfaces at both sides of the ZnO nanonet. (c) PL measurements of the as-prepared ZnO/ZnAl₂O₄/AAO composite sample as a function of energy density of the pump laser. Photos sitting on the traces are the streak camera pictures, showing the narrowing-down of the emitted light from the sample with the increase in the energy density of the pump laser. (d) Angular dependence of the light intensity around the surface normal direction (inset is the experimental geometry model). Lines with and without dots are the experimental and theoretically simulated data, respectively (© American Institute of Physics, reprinted with permission).¹¹

As shown by the photoluminescence (PL) measurements in the far field in Fig. 10c, only a broad emission feature at 379.8 nm is initially observed with a full width at half maximum (FWHM) of ∼14 nm (trace a in Fig. 10c), which is the spontaneous emission from the ZnO sample. As the incident pump laser energy density increases, a narrow feature at ∼386.0 nm grows up (trace b). This narrow feature dominated at a pump laser energy density higher than 400 μJ cm⁻² (trace c). Further increase in the pump laser energy density leads to the appearance of other features (traces c–e) and finally a sharp feature at 389.9 nm (trace f) shows up with an FWHM of ∼1.1 nm, ten times smaller than that of the broad feature in trace a. The streak camera pictures next to each trace in Fig. 10c graphically show the drastic spectral narrowing-down of the emission peak width with the increase in the pump laser energy density. In Fig. 10c, the emission intensity shows a non-linear increase above ∼400 μJ cm⁻². The fast spectral narrowing-down and the clear energy-density threshold indicate that stimulated emission from the ZnO nanonet occurs.

Fig. 10d displays the angular distribution of the emitted light of the ZnO nanonet. The remarkable result is that the emitted light from the ZnO nanonet is spatially confined around the surface normal direction. Its half FWHM is ∼5°, much narrower than those for the pressurized ZnO wafer and the ZnO (001) single crystal.¹¹ In the wafer sample, the small crystallites randomly orient and thus the light does not show an angular preference. In the ZnO (001) sample, the emitted light intensity gradually decays and its 45° intensity is still about one third of that at 0°. Both the pressurized ZnO wafer and ZnO (001) single crystal are in sharp contrast with the ZnO nanonet in their light emissions. Simple theoretical simulations based on Kirchhoff’s theory indicate that the spatial confinement of the light from the ZnO nanonet originates from the diffraction and interference effects of the periodical air cylindrical nanopores in the nanonet.

This example shows that a large area ZnO photonic crystal laser, though not perfect, can be feasibly prepared on the amorphous AAO template by the IRG approach.

4.3 Au nanotube molecular filter and sensor

The porous membranes such as AAO and track-etched polymers can be used as filters. In the pharmaceutical industry, if the pores were small and uniform, the membranes could separate molecules. But neither AAO nor track-etched polymers can be used directly as molecular filters due to the large diameter of their pores caused by the limitations of the fabrication process. There could be two ways to overcome these shortcomings. One is to prepare a hybrid membrane consisting of mesoporous silica in the AAO channels.⁴⁵ The other is to fabricate Au nanotubes with a smaller diameter in the pores of the track-etched polymers mentioned above.⁴⁶ The inner diameter of the Au nanotube can be continuously adjusted to reach the molecular dimensions (<1 nm)^46–48 to satisfy the goal of molecular sieving.

In addition to the molecular sieving application, the Au nanotube membranes can also be applied as sensors for the detection of molecules and ions.^16,49 There are several reviews contributed by Martin and co-workers on this topic,^50–52 so here we briefly mention the basic concept and idea. In a typical sensing process, two salt solutions are separated by the Au nanotube membrane. Then a constant potential is applied and an analyte is added to one of the salt solutions. If the size of the analyte is comparable to the inner diameter of the nanotubes, a substantial decrease in the corresponding current occurs.^16,49

5. Summary and perspectives

In this tutorial review, we have described the IRG approach and its applications in preparing various functional nanomaterials ranging from 1D nanowires, nanotubes and nanocables to 2D nanonets, depending on the type of the template used. The key in the IRG approach is that these templates serve as not only spatial confiners but also reactants for interfacial reactions at the exposed template surfaces. By carefully selecting the types of the templates and the interfacial reactions, one can control the morphologies and structural parameters of the prepared nanomaterials.

Several types of examples have been illustratively presented in this review. Ordered arrays of nanotubes and nanonets with hexagonal nanopores of semiconductors and metals can be fabricated on the 2D AAO template, one of the most popular porous hard templates in nanofabrication, by the IRG approach. The diameter, length and wall thickness of the nanotubes can be adjusted by tweaking the structural parameters of the templates and selecting the precursors. Other 2D nanoporous templates such as porous silicon, ATO and other anodic valve metal oxides can also be used in the IRG approach to yield many other low-dimensional nanostructures. Kinetic and thermodynamic controls in the IRG approach are discussed with respect to the morphology and crystallinity controls of prepared nanomaterials.

1D templates have the advantage in preparing 1D single crystalline nanomaterials because various methods to fabricate 1D single crystalline templates have been well established in nanofabrications. These 1D templates can serve as secondary templates in the IRG approach to preparing other 1D nanomaterials. Depending on the diffusivities in the templates of the incoming reacting species used in the IRG approach, hollow and full solid 1D inorganic and organic nanostructures can be routinely achieved on the 1D templates by exploiting the interfacial reactions involving either solid–solid, solid–liquid or solid–vapor interfaces.

The IRG approach can be extensively employed to fabricate many low-dimensional functional nanomaterials that register wide applications in nanoscience and nanotechnology. We select three examples to show that a large area 2D ZnO photonic crystal laser, a 2D CNT field emitter, and a gold nanotube molecular filter and sensor have been successfully prepared with the IRG approach and to illustrate that the properties of these nanodevices can be tweaked by the collective behaviors of nanonets and nanotube arrays.

The full potential of the IRG approach has yet to be explored and uncovered. Although not mentioned in detail in the main text, zero-dimensional nanoparticles can be obviously employed to fabricate various quantum dots, core–shells and shells with the IRG approach. As a matter of fact, any templates whose exposed surfaces are reactive towards the incoming precursors can be used in the IRG approach. We expect that many more functional nanomaterials will be obtained by the IRG approach and will find unexpected applications in electronics, photonics, magnetics, energy conversion devices, sensors and actuators, catalysis, drug-delivery, information storage, and even the emerging spintronics and molspintronics.

Acknowledgements

This work is jointly supported by NSFC (20573001, 50821061, 20773001, 20827002) and MOST (2006CB806102, 2007CB936202, 2009CB929403), China.

References

1. C. R. Martin, Science, 1994, 266, 1961–1966.
2. G. A. Somorjai, Introduction to Surface Chemistry and Catalysis, Wiley, 1994.
3. M. W. Roberts, Catal. Lett., 2000, 67, 41–53.
4. J. C. Hulteen and C. R. Martin, J. Mater. Chem., 1997, 7, 1075–1087.
5. H. Chik and J. Xu, Mater. Sci. Eng., R, 2004, 43, 103–138.
6. H. Gao, C. Mu, F. Wang, D. Xu, K. Wu, Y. Xie, S. Liu, E. Wang, J. Xu and D. Yu, J. Appl. Phys., 2003, 93, 5602–5605.
7. C. Mu, Y. Yu, R. Wang, K. Wu, D. Xu and G. Guo, Adv. Mater., 2004, 16, 1550–1553.
8. L.-S. Wang, C.-Y. Lee and H.-T. Chiu, Chem. Commun., 2003, 1964–1965.
9. Y. Wang and K. Wu, J. Am. Chem. Soc., 2005, 127, 9686–9687.
10. Y. Wang, Q. Liao, H. Lei, X. Ai, J. Zhang and K. Wu, Adv. Mater., 2006, 18, 943–947.
11. Q. Liao, Y. Wang, J. Li, K. Wu, X. Ai and J. Zhang, Appl. Phys. Lett., 2007, 91, 041103.
12. L. Zhao, M. Yosef, E. Pippel, H. Hofmeister, M. Steinhart, U. Gösele and S. Schlecht, Angew. Chem., Int. Ed., 2006, 45, 8042–8045.
13. F. Wang, Y. Wang, J. Yu, Y. Xie, J. Li and K. Wu, J. Phys. Chem. C, 2008, 112, 13121–13125.
14. Y.-A. Chung, C.-Y. Lee, C.-W. Peng and H.-T. Chiu, Mater. Chem. Phys., 2006, 100, 380–384.
15. V. P. Menon and C. R. Martin, Anal. Chem., 1995, 67, 1920–1928.
16. M. Wirtz, M. Parker, Y. Kobayashi and C. R. Martin, Chem.–Eur. J., 2002, 8, 3572–3578.
17. L. Zhao, M. Yosef, M. Steinhart, P. Göring, H. Hofmeister, U. Gösele and S. Schlecht, Angew. Chem., Int. Ed., 2006, 45, 311–315.
18. L. Zhao, N. Li, A. Langner, M. Steinhart, T. Y. Tan, E. Pippel, H. Hofmeister, K.-N. Tu and U. Gösele, Adv. Funct. Mater., 2007, 17, 1952–1957.
19. L. Liu, W. Lee, R. Scholz, E. Pippel and U. Gösele, Angew. Chem., Int. Ed., 2008, 47, 7004–7008.
20. J. M. Macak, C. Zollfrank, B. J. Rodriguez, H. Tsuchiya, M. Alexe, P. Greil and P. Schmuki, Adv. Mater., 2009, 21, 3121–3125.
21. F. Goettmann, A. Fischer, M. Antonietti and A. Thomas, Angew. Chem., Int. Ed., 2006, 45, 4467–4471.
22. A. Fischer, M. Antonietti and A. Thomas, Adv. Mater., 2007, 19, 264–267.
23. A. Fischer, J. O. Müller, M. Antonietti and A. Thomas, ACS Nano, 2008, 2, 2489–2496.
24. A. Fischer, Y.-S. Jun, A. Thomas and M. Antonietti, Chem. Mater., 2008, 20, 7383–7389.
25. H. Dai, E. W. Wong, Y. Z. Lu, S. Fan and C. M. Lieber, Nature, 1995, 375, 769–772.
26. W. Han, S. Fan, Q. Li and Y. Hu, Science, 1997, 277, 1287–1289.
27. C. Tang, S. Fan, M. L. de la Chapelle, H. Dang and P. Li, Adv. Mater., 2000, 12, 1346–1348.
28. R. Fan, Y. Wu, D. Li, M. Yue, A. Majumdar and P. Yang, J. Am. Chem. Soc., 2003, 125, 5254–5255.
29. Y. Wu, B. Messer and P. Yang, Adv. Mater., 2001, 13, 1487–1489.
30. B. Gates, Y. Wu, Y. Yin, P. Yang and Y. Xia, J. Am. Chem. Soc., 2001, 123, 11500–11501.
31. B. Gates, B. Mayers, Y. Wu, Y. Sun, B. Cattle, P. Yang and Y. Xia, Adv. Funct. Mater., 2002, 12, 679–686.
32. X. Jiang, B. Mayers, T. Herricks and Y. Xia, Adv. Mater., 2003, 15, 1740–1743.
33. J. H. Song, Y. Wu, B. Messer, H. Kind and P. Yang, J. Am. Chem. Soc., 2001, 123, 10397–10398.
34. Y. Sun, B. T. Mayers and Y. Xia, Nano Lett., 2002, 2, 481–485.
35. D. H. Son, S. M. Hughes, Y. Yin and A. P. Alivisatos, Science, 2004, 306, 1009–1012.
36. R. D. Robinson, B. Sadtler, D. O. Demchenko, C. K. Erdonmez, L.-W. Wang and A. P. Alivisatos, Science, 2007, 317, 355–358.
37. L. Pan, L. Pu, Y. Shi, S. Song, Z. Xu, R. Zhang and Y. Zheng, Adv. Mater., 2007, 19, 461–464.
38. Y. Yang, X. W. Sun, B. K. Tay, J. X. Wang, Z. L. Dong and H. M. Fan, Adv. Mater., 2007, 19, 1839–1844.
39. K.-W. Chang and J.-J. Wu, J. Phys. Chem. B, 2005, 109, 13572–13577.
40. H. J. Fan, M. Knez, R. Scholz, K. Nielsch, E. Pippel, D. Hesse, M. Zacharias and U. Gösele, Nat. Mater., 2006, 5, 627–631.
41. H. Fan, M. Knez, R. Scholz, K. Nielsch, E. Pippel, D. Hesse, U. Gösele and M. Zacharias, Nanotechnology, 2006, 17, 5157–5162.
42. M. Leskelä and M. Ritala, Angew. Chem., Int. Ed., 2003, 42, 5548–5554.
43. J. Grabowska, R. T. Rajendra Kumar, E. McGlynn, K. K. Nanda, S. B. Newcomb, P. J. McNally, L. O’Reilly, J.-P. Mosnier and M. O. Henry, Thin Solid Films, 2008, 516, 1725–1735.
44. X. Wu, A. Yamilov, X. Liu, S. Li, V. P. Dravid, R. P. H. Chang and H. Cao, Appl. Phys. Lett., 2004, 85, 3657–3659.
45. A. Yamaguchi, F. Uejo, T. Yoda, T. Uchida, Y. Tanamura, T. Yamashita and N. Teramae, Nat. Mater., 2004, 3, 337–341.
46. K. B. Jirage, J. C. Hulteen and C. R. Martin, Science, 1997, 278, 655–658.
47. M. Nishizawa, V. Menon and C. R. Martin, Science, 1995, 268, 700–702.
48. J. C. Hulteen, K. B. Jirage and C. R. Martin, J. Am. Chem. Soc., 1998, 120, 6603–6604.
49. Y. Kobayashi and C. R. Martin, Anal. Chem., 1999, 71, 3665–3672.
50. H. Bayley and C. R. Martin, Chem. Rev., 2000, 100, 2575–2594.
51. M. Wirtz, S. Yu and C. R. Martin, Analyst, 2002, 127, 871–879.
52. Y. Kobayashi and C. R. Martin, J. Electroanal. Chem., 1997, 431, 29–33.

---