Metal Oxide Nanoparticles

Journal

J. Am. Ceram. Soc., 95 [3] 831850 (2012) DOI: 10.1111/j.1551-2916.2011.05056.x 2012 The American Ceramic Society

Metal Oxides Mono-Dimensional Nanostructures for Gas Sensing and Light EmissionCaterina Soldano, Elisabetta Comini, Camilla Baratto, Matteo Ferroni, Guido Faglia, and Giorgio SberveglieriCNR-IDASC and Dipartimento di Chimica e Fisica per lIngegneria e i Materiali, ` Universita degli Studi di Brescia, 25131 Brescia, Italy

The present Review opens with an overview of metal oxides and the description of their material properties. The rst part mainly deals with various preparation techniques and growth mechanisms, which have been largely developed in the last few decades to fabricate dierent types of metal oxide nanostructures. The second part, on the other hand, will provide the Reader with an up-to-date summary of dierent applications in which metal oxides nanostructures have been successfully implemented and/or suggested. In particular, great attention will be dedicated to the implementation of metal oxide nanostructures for gas sensing as well as light emission applications, for which both working principles and integration, as well as the latest reported cutting edge results are herein reported. Finally, we emphasize on the current potentials of metal oxide nanostructured materials and future challenges that this eld in continuous expansion is currently facing. I . Introduction II. Preparation techniques and growth mechanisms 1. Doping of quasi-one-dimensional metal oxide nanostructures 2. Preparation of quasi-one-dimensional metal oxide heterostructures III. Applications of metal oxide nanostructures IV. Metal oxide gas sensors a. Working principle b. Nanowires integration c. Functional devices i. Conductometric chemical sensors ii. Optical chemical sensors V. Light emission a. Working principle b. Nanowires integration c. Light emitting diodes

VI. Conclusions VII. Acknowledgments VIII. References

I.

Introduction

D. J. Greencontributing editor

Manuscript No. 30088. Received July 29, 2011; approved December 07, 2011. Author to whom correspondence should be addressed. e-mail: caterina.soldano@ ing.unibs.it

oxides (MOxs) represent a vast class of materials of interest for various scientic communities, ranging from physics to chemistry, from material science to engineering.14 There exist a large variety of metal oxides compounds, mostly depending on the type of metals used along with oxygen. To the latter, most of the properties characterizing those types of materials, are related. A number of very dierent properties have been observed so far, ranging from metallic, semiconducting as well as insulating behavior, mainly depending on the specic electronic structure of the particular oxides. This broad range of materials properties has opened up the way to numerous and versatile applications, often very dierent one from each other such as microelectronic circuits,5 sensors,6 piezoelectric devices,7 fuel cells,8 anti-corrosive coatings,9 catalysts,10 and more. For decades now, metal oxides have been successfully used in various forms in the eld of gas sensing, where the conductometric properties of those materials are exploited based principally on the induced variation of the electrical resistance upon interaction (absorption, chemisorption or physisorption) of a gas molecule on the oxide surface. MOxs catalytic properties are also very well known and exploited in the chemical and petrochemical industries, where a large number of processes implicate the usage of these materials. Further, catalysts- and absorbents-oxides-based are frequently used to remove species such as CO, NOx, and SOx, common by-products of fuel-based combustion processes, and, at the same time, to monitor and control environmental pollution.11 Microelectronics and semiconductor industries represent further additional and predominant elds for metal oxides applications; in fact, most commercially available chips nowadays contain at least one metal oxide element. However, there are still a number of not yet very much investigated and unexplored elds in which those materialsETAL

M

Feature

832

Journal of the American Ceramic SocietySoldano et al.

Vol. 95, No. 3

can be successfully employed, such as lighting1214 [in particular for light emitting diodes (LEDs) and lasers], solar cells,15 eld emitters16 as well as high-capacity batteries.17 In the continuous quest for novel applications, new functionalities have been introduced by simply reducing the size of the material of interest toward the nanoscale world, as in the case of nanoparticles, nanowires, carbon nanotubes,18 fullerenes,19 and graphene.20 In the particular eld of metal oxides, quasi one-dimensional (1-D) structures21 have shown several advantages with respect to their traditional thin- and thick-lm counterpart such as very large surface-to-volume ratio, dimensions comparable to the extension of surface charge region, greater stability as a result of the high degree of crystalline order,22 relatively simple and low-cost preparation methods,10 possibility of dierent chemical surface functionalization23 with selective species, modulation of their operating temperature to select the specic reactions, catalyst deposition over the surface for promotion or inhibition of specic reactions. Moreover, there is also the possibility of eld-eect transistors (FETs) conguration, which probes the gate-modulated transport characteristics to tailor sensitivity and selectivity.24 Novel synthesis approaches, physical as well as chemical, have been developed to fabricate metal oxides nanostructures (nanoparticles, nanowires, nanotubes, whiskers, and a variety of more exotic nanoscale objects). Both top-down and bottom-up fabrication techniques are used: the former, involving the micro-fabrication techniques to reduce the lateral dimension of a bulk material, the latter based on the assembly of molecular building blocks from the bottom by various synthesis methods which envision an atom-byatom or molecule-by-molecule growth. In this Review, the Reader will be provided with stateof-the-art overview of the dierent preparation techniques currently used to grow metal oxides nanostructures and architectures, as well as various methods to dope one-dimensional nanostructures to tune their electronic properties. Among a large numbers of possible applications (which will be herein only briey summarized), most of the attention will be dedicated to the use of metal oxide nanostructures to fabricate devices for gas sensing as well as light emission. Details will be given for both elds in terms of their working principles, as well as the integration of those nanostructures in working or existing architectures, together with the most recent results reported in literature. When needed, the Reader will be referred to more specic and detailed literature.

II.

Preparation Techniques and Growth Mechanisms

One-dimensional structures can be prepared following two dierent approaches: bottom-up and top-down technologies. Top-down technology is based on standard microfabrication equipment with deposition and etching on planar structures to reduce the lateral dimensions of lms down to the nanoscale level. To selectively remove the material the

following techniques can be used: electron-beam, focused ion beam, X-ray lithography, nano-imprinting, and others. Drawbacks of top-down reside in the extremely elevated costs and long preparation times, although advantages are the great knowledge and control for the well developed technology coming from the semiconductor industry; furthermore, the device is directly prepared on planar surfaces which allows for easier subsequent contacting approaches. Top-down approaches have been so far used to fabricate highly ordered nanowire,2528 however, the 1-D nanostructures produced with these techniques are in general not single crystal. The bottom-up approach instead is based on the assembly of molecular building blocks or chemical synthesis directly in the nanosized morphology. This approach presents many advantages such as the high purity and crystallinity of the produced materials, the easy achievement of reduced dimensions, the low cost of the experimental set-ups and the possibility to easily dope and form junctions. On the other hand, the integration on planar substrates, necessary for the full exploitation of their useful properties, can be troublesome. In addition, the alignment and patterning of the nanostructures could be more challenging when using the bottom-up approach.29 The combination of the bottom-up approach (for the production of high quality nanostructures) with top-down approach (that can lead to large scale fabrication) represents at the moment the most promising strategy to fabricate highly functional devices. In literature, depending on the creativity and imagination of the authors, a lot of dierent names have been used, such as whiskers, nanowires, nanotubes, nanocastles, bers, brils, nanotubules, nanocables, etc., as shown in Fig. 1. Some of these manifold morphologies have indeed potential applications as functional devices. The preparation of lms or small particles can be obtained with an early stop of the growth process; whereas for the achievement of one-dimensional morphologies with an isotropic atomic bonding it is necessary to induce a break in the symmetry during the growth. There must be a preferential growth direction characterized by a much faster growth rate. The number of synthesis techniques is growing very rapidly; we can distinguish between catalyst-free and catalyst-assisted methods and between vapor- and solution-phase growth. Vapor-phase preparation techniques are very popular in literature, although solution-phase ones provide exible synthesis procedures with cheaper equipment. Depending on the presence of the catalyst during the growth, dierent mechanisms have been proposed, i.e., vapor-liquid-solid (VLS), solutionliquid-solid (SLS) or vapor-solid process (VS) (Fig. 2). Wagner and Ellis in 196430 discovered the controlled catalytic growth of whiskers and named the VLS mechanism for the three phases involved: the Vapor-phase precursor, the Liquid catalyst droplet, and the Solid crystalline product. Silicon whiskers could be grown by heating a Si substrate covered with Au particles in a mixture of SiCl4 and H2 .

Fig. 1. Dierent morphologies for sensing and lighting applications: (a) nanowire, (b) longitudinal heterostructure, (c) nanotube, (d) core-shell heterostructure, (e,f) nanobelts, and (g) hierarchical structure.

March 2012

One-Dimensional Metal Oxide for Gas Sensing and Light Emission

833

Fig. 2. Vapor-Liquid-Solid (VLS) growth of nanowires. The catalyst is in the liquid phase and precursors in the vapor phase can adsorb and condense to form the nanowire. In the Solution-LiquidSolid (SLS) growth of nanowires the precursors in the liquid phase react to form the nanowire.

Few but interesting research publications31,32 present a shape diagram of nanostructures as a function of selected parameters such as source and substrate temperature as well as type of catalyst; even if there is no clear explanation of the dierent results obtained, it is indeed useful for an eventual conrmation and/or confutation of subsequent theoretical studies. The presence of the catalyst particle at the top of the nanowire is the result of a VLS growth mechanism; however, this does not dene the catalyst phase and role during the growth. It has been shown that for all the vapor phase processes there is a dead time before the nanowires growth begins.33 Dierent methods can be used for the preparation of the catalytic nanoparticles, including thermal evaporation, sputtering, colloidal solutions; nevertheless, the focus has to remain on the agglomeration of the nanoparticles because the dimension of the clusters strongly inuences the diameter of the growing nanowire. Since the nanowire deposition is performed at elevated temperatures, even if the metal forms a thin layer covering the entire substrate, clusters will be formed due to the Ostwald ripening.34 When the nanowires have a uniform cross-section along their entire length, the condensation happens mainly at the catalyst surface, and then it diuses and segregates at the interface with the nanowire. In this case, a uniform catalyst dimension on the substrate should result in a uniform nanowire diameter, if nanoparticles coalescence eect is negligible.35,36 Whereas, if the catalyst is consumed or evaporated during the growth process, the diameter of the nanowire will decrease, resulting in conical shape nanowires. Growth temperature is one of the key parameters to control the critical phases during the nanowires elongation, dening the adsorption, diusion (surface and bulk), and solubility. Temperature, pressure, and gas composition must be well controlled to obtain reproducible and reliable nanowires growth. Furthermore, the selection of the catalyst is crucial since it has to lower the activation energy of nucleation at the nanowire-catalyst interface to ensure the preferential onedimensional growth.37 During a VS growth, instead, the nanowire crystallization originates from the direct condensation from the vapor phase. First, it was attributed to the presence of lattice defects, but defect-free nanowires obtained by VS are also frequently reported in literature. Very little is still known, even if a number of experimental and theoretical works have been presented. At high temperatures, the source material is evaporated and then condensed on the substrate at lower temperatures. The initially condensed molecules form seed crystals that serve as the nucleation sites; as a result of which, they facilitate directional growth to minimize the surface energy. This self-catalytic growth associated with many thermodynamic parameters is a rather complicated process that often needs quantitative modeling.38

Recently, some investigations on the nanowires growth have been presented in literature, according to which dierent growth mechanisms have to be taken into account all together with a large number of parameters involved which makes a systematic study more dicult.39 A quantitative investigation on the nucleation and growth of indium oxide nanowires on single crystal substrate, with catalytically active gold particles, was studied through thermal evaporation of pure In2O3 powders with the eect of substrate seeding with gold catalysts.40 The nanowires crystallize in the bodycentered cubic (bcc) structure and grow along the [100] vector of the cubic crystalline cell without any preferential growth direction. The measured growth rate can be satisfactorily explained taking into account a concurrent direct VS and catalyst-mediated VLS mechanism during nanowire growth, with the formation of high-index lateral faces which regulates longitudinal elongation at high temperature, according to the periodic bond chain theory. Most of the combined theoretical and experimental investigations rst perform the synthesis and afterward the structure analysis. There is lack of detailed in situ structural monitoring. Although lately some in situ transmission and scanning electron microscopy (TEM and SEM, respectively) observations of the nanowires growth were presented limited to a small number of nanowires such as silicon,41 germanium,4244 gallium arsenide,45,46 bismuth,47 CdTe48 a complete picture is still missing. The application at elevated temperatures and in a controlled gas environment for electron microscopy still remains an experimental challenge, like other structural characterization methods such as SPM (Scanning Probe Microscopy) or X-ray diraction techniques. The growth of ZnO nanobelts was monitored in situ using X-ray diraction for the rst time,49 by heating metallic zinc powder in air at 368C568C. A morphology diagram for the synthesized products was generated after systematic study of the experimental parameters. Higher temperatures and faster heating rates favor one-dimensional growth. Faster growth was observed for samples with higher growth temperatures, lower heating rates, and one-dimensional growth. Recently, direct in situ optical and photoelectron emission microscopy studies of VO2 nanostructures growth of using vapor transport of V2O5 in vacuum and in inert gas environment were reported.50 The formation, coexistence, and transformation of the intermediate oxide phases and morphologies were structurally and compositionally characterized. Both kinetic and thermodynamic factors seem to play a role in the composition, structure, and morphology during multiple phase transformations. The key factors governing the growth emerged from this study were: 1. abundance of the precursor on the surface and its temperature; 2. liquid droplets anity to the substrate; 3. competition between oxygen loss (if in vacuum or a reductive atmosphere) and heating rate. Controlling these parameters carefully, the chemical composition, morphology, and size of the nal structures may be tailored. Using a heating rate lower than the vanadia reduction rate allowed the interruption of the growth at an intermediate step obtaining V2O5, V6O13, or VO2. Lately, a lot of research has also been performed on nanowire preparation by oxidation in controlled environment of the metal composing the metal oxide, especially in the case of copper oxide.5153 The growth mechanism for the formation of nanowires by thermal oxidation is still under debate, but the most interesting thing is that the wires are single crystals.54 The models presented suggest that the nanowire formation in a vapor-solid growth process55 is related to stress relaxation at the interface56 or to competing grain boundary and lattice diusion of copper ions across the

834

Journal of the American Ceramic SocietySoldano et al.

Vol. 95, No. 3

Cu2O layer.49,57 Nevertheless, this technique oers many advantages such as an easy patterning of the nanowires using conventional patterning techniques on the metal lm, large area, and catalyst-free nanowire production, and the inexpensive equipment. Finally, another possible mechanism for the preparation of nanowires is solution-based catalyzed-growth (SLS), which is similar to the VLS: a metallic nanoparticle catalyzes the precursors decomposition and crystalline nanowire growth.58,59 There are many experimental procedures for the preparation of nanowires from the liquid phase, but in general VLS produce nanowires with better crystalline properties. Large research eort has been devoted to the template-free methods for the deposition of one-dimensional nanostructures in liquid environment exploring surfactant-assisted, sonochemical, hydrothermal, electrochemical and electrospinning deposition techniques. Surfactant-assisted deposition is a trial-and-error procedure, where the surfactant is used to promote the anisotropic crystal growth and conne it as in a micrometer-size reactor. Dierent surfactants have been proposed such as oleic acid, hexylphosphonic acid (HPA), tetradecylphosphonic acid, trioctylphosphine oxide, and trioctylphosphine60; dierent conditions and surfactants must be used for dierent materials. In addition, sonochemical method instead uses ultrasonic waves to modify the crystal growth; in fact, bubbles, formed in the aqueous solution, tend to grow and subsequently collapse leading to extreme reaction conditions needed to form nanowires.61,62 Nevertheless, both hydrothermal and electrochemical processes are well-known procedures for material synthesis. In the rst, a solution of an aqueous mixture of soluble precursors metal salt (metal and/or metal-organic) is placed in an autoclave at high temperature (between 100C and 300C) and relatively high pressure (>1 atm) conditions.63,64 Electrochemical deposition instead, is obtained using an electrolytic cell with the substrate acting as cathode and where the metal ions in the solution are moved by an electric eld to coat the substrate. Nanowires may be formed directly or by using a template.65 Electrospinning is known since 1994 and exploits an electrical charge to force the formation of mats of ne bers. A solid ber is produced as the electried jet is continuously stretched due to the electro-static repulsions between the surface charges and the evaporation of solvent.6668

(1) Doping of Quasi 1-D Metal Oxide Nanostructures The doping issue is extremely important for both electrical and optical properties of nanowires and, of course, it does also inuence the morphology as well. The formation of precipitates, the segregation, and nucleation of second phases must be carefully avoided such as in single crystals, dopant atoms may produce defects and distortions in the lattice,69 thanks to the conned dimension of nanowires much more tensile stress can be managed. Furthermore, unintentional defect formation may result in a decrease in the charge carrier mobility and cause defect emission in the optical spectra.70,71 The dopant has to change the electrical and optical properties avoiding any structural and chemical stability deterioration. Doping during the growth can be very easy, but there is still lack in control necessary for a reliable preparation. Furthermore, a quantitative study of the doping density and prole along the nanowire is challenging, due to the reduced dimensionality, and a high spatial and compositional resolution is necessary.72 Dierent methods have been proposed for nanowires doping73: the dopant can be added in the vapor- or liquid-phase during the growth process74 or introduced in the nanostructure after the growth. By modifying the composition of the precursor in the evaporation-condensation process, the dopant may be easily added in the lattice

of the growing nanowire, but a ne control in the amount of dopant remains challenging. There is of course a signicant dierence between the composition of the precursor and the composition of the obtained nanowires due to dierences in the evaporation rate and in the condensation within the nanowire lattice.75 Another interesting technique to introduce doping into nanowires is the use of low temperature electrochemical process as reviewed in reference.76 In this scenario, ZnO nanowires have been doped with Co and Ni.77 In 2008, the monolayer doping technique (MLD)78 has been proposed, which consists in the formation of self-assembled monolayers of dopant-containing molecules on the nanostructure surface followed by a thermal diusion of dopant atoms by rapid thermal annealing. The dose can be tuned by the molecular footprint of the precursor (i.e., smaller molecules allow for higher doses), whereas the annealing temperature and time, allow for the control of the junction depth. Furthermore, more conventional doping techniques such as ion implantation and diusion may be used, with attention to side eects such as modication in the structure and morphology due to ion bombardment and also amorphization at higher doses.79 For particular metal oxides, oxygen stoichiometry can modify the electrical properties, changing the carrier concentration, the mobility, and the overall electrical resistivity.80 The oxygen vacancies concentration may be controlled through variation of the oxygen content in the gas carrier during the nanowires growth, or by post-synthesis treatment in reducing or oxidizing atmospheres.75 The most studied metal oxide for doping is by far zinc oxide, especially for optical and optoelectronic applications; it has a wideband gap (~3.37 eV), it is transparent to visible light and it has a room temperature and high temperature luminesce (e-h binding energy ~60 meV). ZnO is naturally n-type doped although achieving reliable and stable p-type doping is still challenging and the key factors leading to reproducible and stable p-type doping have not yet been identied. To grow p-type ZnO, the acceptor concentration has to be higher than the unintentional donor concentration. Dierent authors have reported the eect of doping with V-group elements such as As, P, N,8185 and eect of doping also with Tm, Yb, and Eu using ion implantation and post annealing was reported in the literature.86 Recently, also Sb has been proposed as acceptor doping.87 Currently, the challenge is the formation of homojunctions exploiting the natural n-type behavior of ZnO and controlling the acceptor dopant introduction selectively during the nanowires growth process. Few reports on ZnO nanowires homojunctions have been reported in literature. In Ref. [88] arsenic atoms, used as acceptors, were introduced into ZnO lattice by diusion from the GaAs substrate using thermal annealing after ZnO nanowire array were grown by chemical vapor deposition. While in Ref. [89] As atoms were added by ion implantation along the direction perpendicular to the vertically aligned ZnO nanowires, the nanowires section was intentionally increased with their length to avoid ion implantation into the sidewalls. Dopant activation was obtained by annealing of the As-implanted ZnO nanowires in vacuum. Another acceptor dopant used for the preparation of homojunctions is phosphorus that has been introduced either during the nanowire growth90 (using a P2O5 powder as a precursor in the chemical vapor deposition process91 or adding NH4H2PO4 during the hydrothermal growth process92) or by ion implantation and successive annealing treatment to activate the acceptors, as shown in Fig. 3.93 A lot of eort has been devoted to possibly control nanowires doping and, although there is still the need of further research, interesting results have been presented and the real integration into reliable electrical and electro-optical devices will surely be soon available.

March 2012

One-Dimensional Metal Oxide for Gas Sensing and Light Emission

835

Fig. 3. (a) Top and (inset) side view of the as-grown ZnO rods. (b) Top and (inset) 45 tilted view of ZnO:P rods embedded in SOG layer. (c) Low-magnication TEM image of an annealed ZnO:P rod. (Inset) Change in lattice spacing along the a- and c-axis along the ZnO rod. [(d)(f)] High-resolution TEM images from dierent parts of the rod as indicated in (c). The arrows indicate the growth direction. (g) SAED pattern of region labeled d (Reprinted with permission from Appl. Phys. Lett., 95, 133124 (2009). Copyright 2009, American Institute of Physics).

(2) Preparation of Quasi One-Dimensional Metal Oxide Heterostructures The preparation of well controlled heterostructures has an intrinsic complexity, especially when radial or longitudinal heterostructures are concerned. The preparation of metal oxides heterostructures in particular, is more dicult with respect to IIIV semiconductors.94 Several examples have been reported in literature for radial, longitudinal, and dierent morphologies of heterostructures. They can be prepared in a single crystalline fashion with a change in precursors or vapor phase during the deposition process or with a 2-step growth of catalysts particles onto the nanowires surface. By periodically controlling the growth condition during the synthesis process, longitudinal heterojunctions can be created along the length of the nanowires. On the other hand, dierent heterostructures can be easily formed just decorating with a secondary phase the nanowires to create new metaloxide and oxideoxide interfaces. The VLS and VS growth mechanisms can be combined for heterostructures production. A 2-step VLS growth with the deposition of catalyst particles on the nanowires obtained from the rst VLS process leads to formation of dendritic structures, as previously shown in Fig. 1(f). A 2-step VLS and VS growth may lead to the formation of core-shell structures,9597 or the decoration of nanowires with small crystals.17,98 Epitaxial relationship is typically found between the crystal lattice of the nanowire acting as the backbone in

branched heterostructures or in the core-shell geometry and the second phase. Various mechanisms have been proposed for production of core-shell heterojunctions. For example, vertically aligned n-GaN/ZnO coaxial heterostructures were prepared on p-GaN substrates by using metal organic vapor phase epitaxy to grow GaN layers onto ZnO nanoneedles.99 ZnCdO/ZnO core-shell heterostructures were fabricated in the form of vertically aligned arrays by combining a chemical vapor deposition and pulsed laser deposition method.100 Using two steps, BiFeO3/ZnO coreshell heterostructures were prepared: rst a ZnO core by hydrothermal method then a BiFeO3 shell was deposited by sputtering and the end product annealed in an oxygen atmosphere.101 Moreover, the synthesis of the low density core-shell ZnO/ ZnMgO heterostructures was reported in Ref. [102] using successively two pulsed laser deposition chambers: a highpressure quartz glass pulsed laser deposition chamber for the nanowire growth and a conventional (low pressure) pulsed laser deposition chamber for the initial buer layer growth and for the shell growth. The introduction of a ZnO buer layer enables the fabrication of individual nanowires with low areal density, suppressing any shadowing eect by neighboring nanowires during subsequent growth. Both radial and longitudinal nanosized In2O3SnO2 heterostructures were prepared by sequential VLS and VS steps, as shown in Fig. 4. Radial heterostructures were obtained by two subsequent VS condensations, whereas longitudinal heterostructures were obtained with two VLS

836

Journal of the American Ceramic SocietySoldano et al.

Vol. 95, No. 3

Fig. 4. Structural characterization of longitudinal heterostructures. (Main panel) TEM image (panoramic view) of an In2O3 nanowire, with the SnO2 nanowire extending along the length. The black arrow marks the termination of the SnO2 nanowire and the catalytic Au nanoparticle, which assisted the VLS growth. (1) SAED pattern showing the cubic single crystal arrangement for the indium oxide nanowire. (2a) CBED pattern and (2b) high-resolution image from the tin oxide nanowire, demonstrating its single crystalline tetragonal (cassiterite) arrangement. (34) High magnication TEM image and corresponding SAED pattern of the heterojunction, where superimposition of both indium and tin oxides has been recorded (Reprinted with permission from Nano Lett., 7, 35538 (2007). Copyright 2007, American Chemical Society).

deposition steps, exploiting the same catalytic gold cluster during both VLS steps. In the rst one, there was an epitaxial growth of crystalline tin oxide on the In2O3 single crystal at rst, but due to the lattice mismatch between the cubic In2O3 and the tetragonal SnO2 lattices, it resulted in polycrystalline shell. In the latter heterostructure instead, both metal oxides are single crystals. The synthesis of highly ordered ultra-dense hetero-epitaxial Si/ZnO hierarchical nanostructures has been recently proposed.103 The rst step is the deposition of silicon nanowires by chemical vapor deposition, followed by deposition of ZnO seeds on the nanowires itself by atomic layer deposition (ALD); then the nanowires are dispersed on a new substrate and the ZnO nanorods growth is performed in aqueous media. Particle-nanowire heterostructures have been prepared using dierent techniques for the dispersion or growth of nanostructures over the oxide nanowires. For example, heterostructures of ZnO/Fe3O4 and ZnO/Fe2O3 were prepared by depositing magnetic nanoparticles on ZnO microrod templates by a low-temperature hydrothermal procedure.104 Quantum dots of PbS were freshly deposited by atomic layer deposition on silicon nanowires.105

Metal oxides such as ZrO2, Al2O3, TiO2, and SiO2 are further widely used as catalyst supports. Oxides-supported gold catalysts are active towards many reactions, including oxidation of CO, selective oxidation (alkenes, alcohols, and even alkanes), water-gas shift, and removal of atmosphere pollutants, such as NOx. These properties open up the way to numerous applications, especially in chemistry, where those materials can be used in reactions such as selective oxidation of alcohols, oxidation of CO, reduction of selective reduction of nitro groups.111,112

III.

Applications of Metal Oxide Nanostructures

This part of the Review is dedicated to the multiple applications of metal oxide nanostructures in dierent elds. For some of those applications, which do not represent a comprehensive collection of all possible applications, we will briey summarize herein the main aspects and characteristics, for others such as gas sensing (Section IV) and light emission (Section V) we will provide dedicated sections with further and more deep detailed description.

(1) As Catalyst Catalysts are species capable of favoring, from a thermodynamic point of view, reactions (chemical or not) while remaining unaltered at the end of the reaction itself.11,106 The eectiveness of a catalytic material is often measured in terms of its inuence and eects on the reaction kinetics. It has been shown that gold nanoparticles, traditionally known for being catalytically inactive in the bulk form, when supported by a metal oxides substrate become very active catalysts.107,108 As a matter of fact, the catalytic activity of nanostructured gold catalysts depends on the size of the gold particles (inactive for diameter larger than ~8 nm).109,110

(2) As Absorbents Metal oxides nanostructures have been recently proposed as ecient adsorbents in case of environmental contaminations, ranging from radioactive residues, by-products of nuclear ssion reaction and leakage of the nuclear reactors to air and water purication. It is indeed of vital importance to have materials capable of absorbing those contaminants in an irreversible and selective fashion while remaining stable at the same time. Due to their large high surface areas and elevated concentration of reactive edges, corners, and defect sites, MOx nanostructures represent very exible and promising candidates for a new generation of absorbent materials, which could lead to safe disposal of undesired species. Compared to currently used technology as in the case of zeolites, activated alumina or activated carbon, metal oxide nanostructures are in general more ecient and have a broader range of sensitive materials. For example, a composite material based on nanostructured SiO2 and TiO2 has been shown to be eective for the removal of elemental mercury vapor113 under UV exposure, which has clearly a great impact both from an environmental as well as health point of view. Volatile organic compounds as well as acid gases cause great concerns for the quality of the air; those compounds are commonly removed by using activated carbon, which, however, is not capable of eciently removing SO2 [Ref. 114] or destructively remove (no desorption over time) organic molecules, as for example copper oxide does.115 (3) As Field Emission Devices As for other types of one-dimensional nanostructures such as carbon nanotubes,116 characterized by sharp ends and large aspect ratio, metal oxide one-dimensional nanostructures exhibit impressive eld-induced electron emission properties,

March 2012

One-Dimensional Metal Oxide for Gas Sensing and Light Emission

837

indicating their potential as future electron sources and displays. Nanowires,16 needles shape,117 nanopyramids,118 and others have been widely investigated and studied in terms of their eld emission features; Bhise et al. have for example extensively demonstrated ecient eld emission properties for SnO2 nanowires with dierent types of doping (Sb,119 RuO2,120 Fe,121 In122), both in terms of emission from a single structure as well as from ordered arrays. Further, conductive vertically aligned TiO2 nanotube arrays123 have been shown to be ecient and stable electron eld emitters even at room temperature with rather low turnon elds ( $ 280 V per electrode distance of 100 lm).

In this specic eld, crystallinity15 plays a crucial role due to the increased diusion length in the anode. Further, the possibility of tailoring the electronic properties of nanowires allows for a consequent tuning of the absorption spectrum with the solar spectrum. Zinc oxide nanowires have been widely applied in dyesynthetized solar cells (DSSC); the nanowire nature provides a direct path for charge collection, thus reducing the recombination processes that are mainly responsible for reduction of the overall cell eciency. Innovative photoanodes have been proposed by using ZnO and SnO2 single crystalline nanowires and TiO2 nanotubes,135 as well as nanowires-based hybrid structures in conjunction with quantum dots acting as light absorbers.136

(4) As BioSensing Devices Nanostructured metal oxides have recently gathered great interest also in the eld of biosensing,124,125 due to their biocompatibility, non-toxicity, and enhanced electron-transfer kinetics and strong adsorption capability; these features provide a suitable microenvironment for the immobilization of biomolecules and result in enhanced electron transfer and in improved biosensing characteristics.126 A biosensor is an integrated miniaturized device that employs a biological element (antibody, enzyme, receptor protein, nucleic acid, entire cell or tissue section) as a sensing element coupled to a transducer for signal detection. A biosensor combines the molecule selectivity with the progress of microelectronics, hence providing a powerful analytical tool with applications in medical diagnostics and other areas. In this perspective, metal oxide nanostructures are able to selectively immobilize species (enzymes,114 cholesterol,127 antibody,128 DNA,129) hence reaching a high degree of selectivity and stability. Very recently metal oxide nanostructures materials [ZnO, Cu(I)/(II) oxides, MnO2, TiO2, CeO2, SiO2, ZrO2] have been proposed and investigated as glucose biosensors.130 We here refer the Reader to more specic literature114 and references therein for a detailed description of this topic, which is of great interest for the medicine community. Due to their specicity, portability, rapid response time and low cost, biosensors are expected to play a critical role in both clinical and non-clinical applications. (5) As Battery Owing to their reduced lateral size and large surface area, nanomaterials represent potential candidates for batteries fabrication, in particular for lithium-based ones. Single hybrid nanostructures (SnO2In2O3)131 have been shown to possibly enhance the electrochemical performances of a battery, due to a large increase of the electronic conductivity as compared to single material nanostructure (SnO2) as a result of the doping-like eect of the In2O3 lattice during the nucleation and growth of the shell structure. In addition, array or complex structures based on metal oxide nanowires are of large importance in the implementation and performance enhancement of the anode. In this scenario, Wan et al.132 have investigated the synthesis of massive SnO2 nanowires by the thermal evaporation method on button-type electrode for Li-ion batteries. Direct growth of one-dimensional SnO2 nanowires on the current collector133 [Si (100) or stainless steel foil] via a VLS method at 600C have also been suggested. (6) As Solar Cells The MOx nanowires have been recently applied in the eld of solar cells. CdSe nanorods-polymer structure134 was suggested, where tailoring the CdSe diameter allowed optimizing the overlap between the absorption spectrum of the cell and the solar emission spectrum.

IV.

Metal Oxide Gas Sensors

(1) Working Principle Deviation of stoichiometry is responsible for high gap MOxs semiconducting properties137; in SnO2 for example shallow states made up of oxygen vacancies are double n-type donors donating electrons to the conduction band.138 Metal oxides are generally operated in air in the temperature range between 500 and 800 K where conduction is electronic and oxygen vacancies are doubly ionized and xed. The gas sensing properties arise from surface chemisorption of oxygen, which acts as a surface state capturing an electron or a hole139 from the conduction band: charge transfer between bulk and surface takes place producing a non-neutral region (with a non-zero electric eld) in the semiconductor bulk, usually referred to as the surface space-charge region (SCR) and semiconductor bands are bent upward near the surface. The implementation of MOx nanowires in the eld of gas sensing has had a major impact on conductometric-based devices; in fact the high crystalline nature, the regular shape of the crystal and the large length-to-width ratio render the nanowires a model system to investigate the eect of gas adsorption on the electrical properties. The variation in photoluminescence emission of MOx nanowires upon gas exposure have been also used as working principle for gas detection and will be later discussed, in more details. On the other hand, electrochemical sensors present advantages in using nanowires whereas other architectures such as microbalance and surface-acoustic-waves devices are still in the early stage of development. The process of gas detection is intimately related to the oxidation reactions between the species to be detected and the chemisorbed surface oxygen,140 which release electrons from surface states to the conduction band and decrease the height of the surface barrier. Other target species, such as the strongly electronegative NO2, directly chemisorb over the oxide surface decreasing sensor conductance by trapping electrons. Electrons can also be injected as an eect of chemisorption, as in the case of water adsorption,119 which forms a hydroxylated surface, where the OH ion is bounded to the cation and the H+ ion to the oxide anion. As regards the overall electrical properties, by contacting the two opposite bases of a single crystal quasi 1-D NW, after placing it on a thermally oxidized highly doped Si substrate, and depositing a gate back contact, one obtains a Single Nanowire Transistor (SNT) structure, as shown in Fig. 5(b). The current ows along the cylindrical axes and parallel to the surface where electrons are trapped: when the nanowire section is thin enough, almost all available electrons are trapped and only the ones thermally activated from surface states are responsible for conduction. Besides if a gate bias is applied, the position of the electrochemical potential and the availability of electrons for surface reactions can be modulated (electroadsorbitive eect141); a majority carrier channel in accumulation mode is indeed created upon application of a positive gate bias as in n-channel Thin Film Transistors (TFT).142,143

838

Journal of the American Ceramic SocietySoldano et al.

Vol. 95, No. 3

Fig. 5. Current transport in nanowire bundles (a) and in a single nanowire transistor (b) (Reprinted with permission from Int. J. Nanotechnol., 5 [45], 45074 (2008). Copyright 2008, Inderscience).

Instead when contacting nanowire bundles the conduction mechanism is dominated by the SCR created at the intercrystalline boundaries between NWs: the carriers must overcome the energy barrier to cross from one nanowire to the neighbor as reported in Fig. 5(a). In this case, the conductance of the bundle (as for polycrystalline gas sensor) can 2 2 therefore be expressed as: G G0 eqVs =kT G0 eq Nt =2eNd kT , where Vs is the bulk to surface potential barrier, T the temperature, Nd is the donor density (oxygen vacancies), Nt the density of charged surface states and G0 is a pre-exponential term, independent in rst approximation from the surface adsorption and temperature. The metal semiconductor junction that forms at the interface between the nanowires and the contacts can play a role in gas detection: the contact resistance is more important for a single NW since it is in series to the semiconductor resistance than that for bundles where it is connected to a large number of resistances. Addition of a small amount of noble metals over the MOx surface such as Au, Pd, Pt, and Ag can speed up surface reactions and improve selectivity toward target gas species. From the energetic point of view, the eect of catalysis is to provide a more favorable reaction path. Catalyst should be dispersed as small crystallites over the surface of the oxide to be active near the grain boundaries where carrier transport takes place. As for signal transduction, the easiest measurable physical quantity is the sensor conductance in DC conditions. The sensor response toward a target gas concentration is dened as the (relative) change of conductance (resistance in presence of NO2 or other oxidizing species). Starting from the sensor response it is possible to derive the sensor response curve, which is the representation of the steady state output as a function of the input concentration.144 The sensor response curve is frequently erroneously called sensitivity, which is instead the derivative of the sensor response curve.

(2) Nanowires Integration For fabrication of devices, nanostructures should be grown or transferred on a suitable substrate and provided with electrical contacts. A proper design of the transducer including electrical contacts, bonding, and packaging should be made to assure correct device behavior. Silicon and alumina are the most used substrates, since they are commonly available and not very expensive. Sapphire can be used if transparency is required, but high cost limits its application to commercial devices. Plastic and paper

substrates are being employed recently for portable and exible devices. In this case the use of low temperature growth process is required, or a transfer technique such as drop coating or roll-transfer printing145 must be used to avoid damage of the substrate; room temperature operation is also needed in this case. For higher temperature operation, a metallic meander is used as a heater to maintain the NWs at the working temperature: this feature can be integrated into micromachined silicon substrate or deposited on the backside of the alumina/ sapphire substrate. The working temperature of the metal oxide gas sensors is maintained applying a constant voltage to the heating meander. Electrical contacts are deposited in two- or four-terminal conguration to measure the nanowire resistivity. The contact must provide as low contact resistance as possible to minimize the voltage drops along the interconnections and should be ohmic. Platinum is one of the most used metallization source since it has a good ohmic contact with most of the metal oxides, it does not oxidize at high temperatures, it has a low diusivity and it is resistive to corrosive gases. In the case of bundles of nanowires either two contacts are deposited on top of the mesh after the growth or nanowires are transferred to micromachined substrate. Then, nanowires are removed from the growth substrate, dispersed in a solvent and using drop coating. To realize single nanowire device, a conveniently placed nanowire is selected. If needed, it can be aligned between specic contacts with nanomanipulation or dielectrophoresis (DEP). Most of the manipulations are performed in association with high-resolution imaging techniques such as scanning probe microscopy (SPM and SEM). DEP technique utilizes the dielectrophoretic force acting on particles to induce spatial movement when exposed to a non-uniform electric eld in the suspension medium. DEP has been shown to be capable of aligning tin oxide nanobelts,146 zinc oxide nanobelts,147,148 and nanoparticles149 and GaN nanowires.150 After the nanowires were placed between contacts, platinum or gold stripes are deposited using Focused Ion Beam (FIB)151 or electron-beam lithography (EBL) to fabricate nanoscale contacts between pre-deposited contacts and NW or to improve existing electrical contacts.

(3) Functional Devices (A) Conductometric Chemical Sensors: In the basic layout, a sensor device is made of a sensing layer with two conducting electrodes for the two-probe measurement of the

March 2012

One-Dimensional Metal Oxide for Gas Sensing and Light Emission

839

electrical resistance, which is the principal gure of merit for the characterization of the sensing performance. Several works have been done for implementation of nanowires in a dedicated layout: from a technological point of view, bundled nanowire devices are much simpler to prepare than single nanowire devices, as it does not require the manipulation of individual nanostructures. In 1996, Yumoto et al.152 rst reported on the gas sensing properties of indium-tin-oxide 1-D structures known as whiskers grown by VLS mechanism. Comini et al. reported a signicant response to CO, ethanol and NO2 in pure SnO2 nanowire bundles prepared by thermal evaporation, and subsequently transferred.153 The device was operated at 400C. A number of publications have been reported since, addressing the enhancement of the sensor fabrication and performance. Wang et al.154 grew SnO2 nanowires directly over functional substrates. Vomiero et al.155 combined two consolidated techniques (vapor transport and condensation and wet etching of a sacricial layer) for the patterned fabrication of metal oxide nanowires directly over the functional substrate. Three-dimensional (3-D) architectures of tungsten oxide prepared by evaporation-condensation method have been used to develop conductometric gas sensors highly sensitive to NO2.156 Nanowires from tens to hundreds of nanometers in width were assembled in a three-dimensional open structure with a large number of nanowire-nanowire interfaces. The measured high sensitivity has been associated to this peculiar hierarchical three-dimensional network. The exposure to 50 ppb of NO2 increased the electrical conductance by a factor of 5. In addition, the nanowire network featured a sensitivity to gases such as ammonia or CO lower than the one observed with WO3 thin lms.157 Reducing the crystallite size allows one to increase the sensitivity.158 Accordingly, Polleux and co-workers adopted this approach to detect sub-ppm concentrations of NO2.159 In this work, a signicant temperature eect was reported for nanowires (about 1 nm in diameter) bundles prepared by softchemistry route. The response decreased by about one order of magnitude by increasing the temperature from 150C to 225C. The low temperature range (down to room temperature) was investigated by Kim et al. working with sub-stoichiometric tungsten oxide (WO2.72) nanowires, few nm in diameter.160 As shown in Fig. 6, the normalized resistance decreases upon exposure to ammonia at 200C operating temperature, although it decreases unexpectedly at lower temperature. As the nanowire diameter is comparable with the mean free path of electrons, it has been supposed that adsorbed molecules may act as scattering centers, thus increasing the overall resistance. The addition of dopants or catalysts to sensitize nanowires has been also exploited. Wan et al. reported the use of Sb doping to tune the resistivity of SnO2 nanowires trough to the formation of shallow donor levels by introduction of Sb+5 into SnO2.161 Authors also recorded an increase in sensor kinetics, attributed to Sb doping, which would favor and accelerate the absorption of oxygen molecules and the formation of oxygen ions on the surface. Kolmakov et al. report the eect of Pd nanoparticles over a single SnO2 nanowire.162 The improved sensing performance to oxygen and hydrogen was attributed to the combined eect of spillover of atomic oxygen catalytically formed and the back spillover eect in which molecular oxygen migrates to the Pd nanoparticles and is catalytically dissociated. Sysoev et al. comparatively investigated long-time stability of SnO2 nanoparticles and nanowires.163 As shown in Fig. 7, nanoparticles show higher response than nanowires during the rst few days of operation. Monitoring the variation along time, the authors concluded that the time-dependence of the response could be due to coarsening eect in nanoparticles.

Fig. 6. Normalized resistance (Rgas/Rair) at dierent operating temperatures for a tungsten oxide nanowire-based sensor (Reprinted with permission from Appl. Phys. Lett., 86, 213105 (2005). Copyright 2005, American Institute of Physics).

Single nanowire devices oer the opportunity to exploit the peculiar nanoscaled size and high crystalline quality, without eects arising from nanowire-nanowire interfaces. The high crystalline quality of nanowires immediately suggested their potentialities as suitable materials to develop eld-eect transistor.164 Zhang et al.165 investigated the possibility to use the gate voltage to modulate the performance of FET gas-sensors. In addition, considering that the high electrical conductivity owing in a nanowire connected by two electrodes, allows increasing the wire temperature, extremely low power consumption can be obtained.166,167 This approach toward selfheating devices revealed very promising even with respect to thin suspended micromachined substrates, both in terms of power consumption and thermal dynamics. (B) Optical Chemical Sensors: In MOx nanostructures the use of transduction mechanism dierent from electrical one is desirable to overcome diculties in contacting nanostructures. One possible solution is to use the bright room temperature photoluminescence (PL) signal shown by nanowires of SnO2 and ZnO to realize all-optical gas sensors for low concentrations of pollutants such as NO2. The SnO2 nanostructures typically exhibit strong visible PL emission at room temperature with a broad emission band peaked at about 2 eV,168170 commonly attributed to radiative defective states within the bandgap (oxygen vacancies).15 On the other hand, ZnO nanostructures show room temperature PL spectra composed by an UV peak due to excitonic recombination at about 3.26 eV,171 and broad emission band in the range from green to yellow depending on deep defect states lying inside the bandgap,172,173 where the position of this band is determined by growth conditions.174,175 Faglia et al.176 showed that PL in SnO2 is quenched in a reversible way by NO2 at a temperature of 120C: the gas adsorbs over the surface creating competitive non-radiative recombination paths. Lettieri et al.177 demonstrated that larger PL quenching by NO2 absorption is observed at room temperature with Continuous Wave (CW). Similarly, photoluminescence of ZnO is quenched by interaction with NO2 room temperature.178,179 Dierently from SnO2, either UV or visible peaks can be investigated showing that they are quenched in the same fashion without any peak shift. Detec-

840

Journal of the American Ceramic SocietySoldano et al.

Vol. 95, No. 3

Fig. 7. Sensor baseline of SnO2 mesoporous layer and nanowire bundle versus time (a). Experiments have been rst carried out in dry air, then (since day 26) in humidied air. (b) Open and lled circles correspond to nanoparticles and nanowires (Reprinted with permission from Sensors Actuat. B, 139, 699703 (2009). Copyright 2009, Elsevier).

tion of very low concentration of NO2 as low as 0.1 ppm has also been shown by Baratto et al.180 (as shown in Fig. 8). By trapping an electron from the conduction band, NO2 adsorbs over the metal oxide surface as NO2. At room temperature, reversibility of chemisorption is guaranteed by UV excitation that supplies the energy needed for desorption. Both for SnO2168 and ZnO,171 time-resolved photoluminescence (TRPL) measurements in presence of NO2 showed small modication of recombination rates and lifetimes upon introduction of the gas. The results support a surface static quenching model, according to which the gas molecules suppress a fraction of radiative transitions instead of simply reducing their probabilities. At higher NO2 concentration, a saturation eect has been observed by Valerini et al.,181 suggesting limitation in usage of optical gas sensor at concentrations higher than 20 ppm. The same authors have also shown the eect of the ZnO nanostructure morphology: relative response to NO2 is higher for sample constituted by pencil-like rods well separated one from each other, characterized by larger surface available for gas-surface interaction. Literature available on this subject shows that alloptical gas sensor based on nanowires is feasible for NO2 detection, even if stability and reproducibility issue were not addressed. Concerning cross interference issue, humidity and ethanol caused an increase in PL signal thus demonstrating that cross interference eect must be taken into account.171

V.

Light Emission

(1) Working Principle Light emitting diodes are dierent from traditional light sources in the way the light is produced; in fact, light emission is achieved by lament heating (incandescent lamp) or an electric arc exciting mercury atoms (uorescent lamp). An LED, in contrast, is basically a semiconductor diode, based on a p-n junction, where current ows from the p-side to the n-side, but not in the reverse direction, when an external bias is applied to the junction itself. Charges (both electrons and holes) pass then through the junction and when an electron meets a hole, it falls into a lower energy level, and releases photons (light). The emission wavelength (or equivalently the color) in this case mainly depends on the constituent material. Metal oxides are usually high band gap semiconductors not feasible to develop light emitting diodes. Some of them, such as SnO2, have shown emission in the visible region, which is indeed due to defects or impurities, which are in general dicult to control during the growth process. Zinc oxide, with a bandgap of 3.37 eV, represents an exception due its large exciton binding which lead to an eective radiative recombination at room temperature, allowing for ecient LEDs in the short-wavelength range (UV). The simplest type of LED based on ZnO is made from a p-type and n-type homojunction diode forward biased as shown in Fig. 9: when a current passes through the diode,

Fig. 8. Dynamic photoluminescence quenching of the signal coming from ZnO nanowires by sub-ppm concentration of NO2 (Reprinted with permission from Sensors Actuat. B Chem., 140 [2] 4616 (2009). Copyright 2009, Wiley).

Fig. 9. Working principle of a LED based on a p-n homojunction.

March 2012

One-Dimensional Metal Oxide for Gas Sensing and Light Emission

841

Fig. 10. Band osets of GaN, NiO, Si and CuAlO2 semiconductors respect to ZnO.

electrons in the conduction band ow across the junction from the n-doped side, and holes in the valence band ow from the p-doped side. The result is that a signicant number of electrons and holes recombine at the junction where both their densities are quite high, emitting light with energy equal to the exciton recombination energy. Due to diculties to p-type dope ZnO, a possible alternative is to prepare heterojunctions with an easily p-dopable materials. Further, to provide an ecient holes injection in the n-ZnO regions, one has to select materials with lower (or negative) valence band osets as compared to ZnO. Figure 10 shows band approximate osets of GaN,182 CuAlO2,183 Si, and NiO184 p-type semiconductors with respect to ZnO; it is clear that the most suitable candidate is p-type GaN, whereas other materials valence band osets are too large to provide eective hole injection. Thus, LED eciencies reported in literature for CuAlO2, Si, and NiO are quite low. On the contrary, GaN is a wurtzite-type material (similarly ZnO) with a lattice constant similar to that of ZnO (mismatch is about 1.9%). In accordance with the Anderson model, a simplied picture of bands at equilibrium for the pGaN/n-ZnO heterojunction, which disregards polarization charge at interfaces, is reported in Fig. 11. As long as forward bias is applied to the junction, eective minority carrier injection takes place [see Fig. 11(b)] and satisfactory luminescence can be obtained from radiative recombination in the ZnO region. Unfortunately, emission from the p-GaN region is detected as well. Therefore, this simple heterostructure can be further improved by introducing blocking (i.e., MgO) layers with the objective to conne carriers in the ZnO regions, as shown in Fig. 12, where electrons experience a much higher energy barrier than the holes

(which can easily tunnel through the barrier) and carrier densities are eective for recombination only in the ZnO region. It is possible to induce electroluminescence even in a nGaN/n-ZnO heterojunction by applying a considerably high voltage; holes are created by defect assisted tunneling or impact ionization. This eect is briey shown in Fig. 13, where indeed the emission is induced in the GaN applying a positive voltage to the ZnO side. As for commercial diodes, higher internal quantum eciency can be obtained only by preparing double heterostructures (DHs) through band gap engineering, with the introduction of ternary compounds as MgZnO, CdZnO, and BeZnO. A DH is made by an active region, a quantum well (QW), in which recombination occurs, and two connement layers cladding the active region. The two cladding or connement layers have a larger bandgap than the active region. Band conguration is benecial also from the point of view of photons reabsorption, which cannot take place in the higher bandgap claddings. The resulting radiation is better dened due to carrier connement and the optical power generated per unit volume is much larger as well. Figure 14 shows a typical DH conguration. For a more realistic view, considering the polar nature of both ZnO and GaN, the piezoelectric and spontaneous polarizations need to be incorporated into the band model, appearing as positive or negative energy steps depending on sign as in Fig. 15. Although polarization is usually detrimental because the electric eld moves away electrons and holes, in some congurations it can be exploited to improve charge connement in the active regions. Discussions so far reported in terms of ZnO bands, are generally valid both for thin lms and for nanowires, considering that the nanoscale objects are usually considerably

Fig. 11. Energy band diagram of the n-ZnO/p-GaN heterojunction at (a) equilibrium and (b) forward bias.

842

Journal of the American Ceramic SocietySoldano et al.

Vol. 95, No. 3

Fig. 12. Energy band structure of a p-GaN/n-ZnO diode with a MgO layer.

injection rate signicantly increases for nanoscale contacts in Schottky diodes,187 nanowires grown directly on p-type lm could benet from increased carrier injection eciency due to nanosized junctions. In general, it is necessary to achieve high quality interface between the heterojunctions to get high-eciency LEDs, since the interface states generally act as non-radiative centers that annihilate free electrons and holes. Vertically aligned nanowires are usually employed in these types of devices to take advantage of the waveguiding eect in nanowires, even if the emission from non-aligned mesh of nanowires is reported.188 Electroluminescence in these junctions is investigated in a typical two-terminal device, fabricated by depositing appropriate lm as the conducting electrode on top of the ZnO NW arrays. The metal used should ensure ohmic contact with ZnO: ITO was used for p-ZnO, Au for p-GaN while Ti/ Au, In, and Al were used for n-ZnO. To prevent the short circuit of connection between the top and the bottom electrodes in the two-terminal devices, an insulating layer made of a polymer matrix [i.e., poly(methyl methacrylate) (PMMA) or spin-on-glass (SOG)] is spin-coated on the NW arrays, followed by oxygen plasma treatment to remove the excess PMMA on top of NWs. A scheme of the heterojunction device with vertical nanowires is reported in Fig. 16.

larger than the ZnO Bohr radius (2.34 nm) and quantum connement eects can be neglected. Indeed nanowires growth produces grain-boundary free and a much less strained and defective interface, which are sources of nonradiative recombination, compared to thin lms. In addition, nanowires can act as direct waveguides and favor directional light extraction without use of lenses and reectors.

(2) Nanowires Integration The LED structure-based nanowires reported in literature so far mainly deals with ZnO homojunctions and heterojunctions, with some rather limited and preliminary investigations for SnO2 [Ref. 185] and In2O3. For homojunctions based on a nanowires/nanowires junction, the introduction of a doping element during the growth process can lead to the synthesis of p- and n-type segments in ZnO NWs, hence forming the junction, where the p- and npart are naturally aligned due to the continuity of the growth process itself.186 In some cases, ZnO layer is present at the base of ZnO nanowires, thus preventing use of nanojunctions eects. On the other end, heterojunction-based devices are mostly realized by growing n-type ZnO nanowires on p-type substrate, mainly thin lm. Since it was reported that the carrier

(3) Light Emitting Diodes In this scenario, metal oxides, and in particular ZnO, have been proposed as possible materials for light emission and optoelectronic applications. As previously mentioned, due to the high excitonic binding energy and the wide room temperature energy gap, zinc oxide is suitable for emission in the near-UV (%370 nm). Further, the value of the energy gap can be selectively tuned by introducing a very small amount of doping atoms (i.e., Mg,189,190 Cd191,192), which could shift the emission toward the visible region. In addition, ZnO is piezoelectric,193 biocompatible and bio-safe.194 The growth of ZnO in various nanostructured forms can be achieved using dierent high- as well as low-temperature (