Progress in Materials Science 54 (2009) 1–67
Contents lists available at ScienceDirect
Progress in Materials Science journal homepage: www.elsevier.com/locate/pmatsci
Quasi-one dimensional metal oxide semiconductors: Preparation, characterization and application as chemical sensors E. Comini, C. Baratto, G. Faglia, M. Ferroni, A. Vomiero, G. Sberveglieri* SENSOR Lab, CNR-INFM, Dipartimento di Chimica e Fisica per l’Ingegneria e per i Materiali, Brescia University, via Valotti 9, 25133 Brescia, Italy
a r t i c l e
i n f o
Article history: Accepted 16 June 2008
a b s t r a c t The continuous evolution of nanotechnology in these years led to the production of quasi-one dimensional (Q1D) structures in a variety of morphologies such as nanowires, core–shell nanowires, nanotubes, nanobelts, hierarchical structures, nanorods, nanorings. In particular, metal oxides (MOX) are attracting an increasing interest for both fundamental and applied science. MOX Q1D are crystalline structures with well-defined chemical composition, surface terminations, free from dislocation and other extended defects. In addition, nanowires may exhibit physical properties which are significantly different from their coarse-grained polycrystalline counterpart because of their nanosized dimensions. Surface effects dominate due to the increase of their specific surface, which leads to the enhancement of the surface related properties, such as catalytic activity or surface adsorption: key properties for superior chemical sensors production. High degree of crystallinity and atomic sharp terminations make nanowires very promising for the development of a new generation of gas sensors reducing instabilities, typical in polycrystalline systems, associated with grain coalescence and drift in electrical properties. These sensitive nanocrystals may be used as resistors, and in FET based or optical based gas sensors. This article presents an up-to-date review of Q1D metal oxide materials research for gas sensors application, due to the great research effort in the field it could not cover all the interesting works
* Corresponding author. Tel.: +39 030 3715771; fax: +39 030 2091271. E-mail address:
[email protected] (G. Sberveglieri). URL: http://sensor.ing.unibs.it (G. Sberveglieri). 0079-6425/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.pmatsci.2008.06.003
2
E. Comini et al. / Progress in Materials Science 54 (2009) 1–67
reported, the ones that, according to the authors, are going to contribute to this field’s further development were selected and described. Ó 2008 Elsevier Ltd. All rights reserved.
Contents 1. 2.
3.
4. 5. 6.
7.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Deposition techniques and growth mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.1. Vapor phase growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.1.1. Vapor–liquid–solid mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.1.2. Vapor–solid mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2. Solution phase growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2.1. Template-assisted synthesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.2. Template-free methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Vertical and horizontal alignment techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.1. Electric field alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.2. Nanomanipulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Doping of quasi 1D metal oxide nanostructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Preparation of quasi 1D metal oxide heterostructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Applications of metal oxide nanostructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 6.1. Metal oxide gas sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 6.1.1. Surface adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 6.1.2. Detection through surface reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 6.1.3. DC resistance transduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 6.1.4. Conductometric gas sensors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 6.1.5. Single nanowire transistor (SNT) based gas sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 6.1.6. PL based gas sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 6.2. Other application fields. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 6.2.1. Lasers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 6.2.2. Solar cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 6.2.3. Field emitters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 6.2.4. Li-ion batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 6.2.5. Single nanowire transistors for biosensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
1. Introduction The increasing concerns with pollution on health and safety stress the need of monitoring all aspects of the environment in real time, and in turn led to a tremendous effort in terms of research and funding for the development of sensors devoted to several applications [1–9]. As far as chemical sensing is concerned, it has been known, from more than five decades, that the electrical conductivity of metal oxides semiconductors varies with the composition of the surrounding gas atmosphere. The sensing properties of semiconductor metal oxides in form of thin or thick films other than SnO2, like TiO2, WO3, ZnO, Fe2O3 and In2O3, have been studied as well as the benefits from the addition of noble metals – Pd, Pt, Au, Ag – in improving selectivity and stability. In 1991 Yamazoe showed that reduction of crystallite size went along with a significant increase in sensor performance [10]. In a nanosized grain metal oxide almost all the carriers are trapped in surface states and only a few thermal activated carriers are available for conduction. In this configuration the transition from activated to strongly not activated carrier density, produced by target gases species, has a huge effect on sensor conductance. Thus, the technological challenge moved to the fabrication of materials with small crystallize size which maintained their stability over long-term operation at high temperature. A huge variety of devices have been developed mainly by an empirical approach
E. Comini et al. / Progress in Materials Science 54 (2009) 1–67
3
and a lot of basic theoretical research and spectroscopy studies have been carried out to improve the well known ‘‘3S” of a gas sensor, namely sensitivity, selectivity and stability. Nanotechnology is nowadays producing sensing materials such as quasi-1D metal oxides (MOX), carbon nanotubes, and nano porous materials. In particular, metal oxides are an attractive and heterogeneous class of active materials covering the entire range from metals to semiconductors and insulators and almost all aspects of material science and physics in areas including superconductivity and magnetism. After the first publications demonstrating the ability of metal oxide nanowires in detecting a variety of chemical species [179,188], the interest in this research area was growing exponentially in the past years as testified by literature. Significant progress has been made both in terms of our fundamental understanding of the interplay between bulk and surface properties and processes in MOX nanowires sensors together with their development as real world sensing platforms. Q1D metal oxide nanostructures have several advantages with respect to traditional thin- and thick film sensors such as very large surface-to-volume ratio, dimensions comparable to the extension of surface charge region, superior stability owing to the high crystallinity [11], relatively simple preparation methods that allow large-scale production [14], possible functionalization of their surface with a target-specific receptor species [190], modulation of their operating temperature to select the proper gas semiconductor reactions, catalyst deposition over the surface for promotion or inhibition of specific reactions and finally the possibility of field-effect transistors (FET) configuration that allows the use of gate potential controlling the sensitivity and selectivity [188]. Preparation and performances of these emerging nanosized structures have been reviewed by a number of authors [12–15], but this research field is growing so fast that there is still the need of a review focused on sensing applications. This review article is focused on the description of metal oxide single crystalline Q1D nanostructures used for gas-sensing application, specifically on the promising approaches that are going to contribute to the further development of this field. The overview will start from presenting the fabrication techniques and the growth mechanisms, focusing on their development and improvements, and pointing out the steps critical for application in real environments. Then the application as chemical sensors will be addressed. Furthermore an outlook on other possible new applications of metal oxide single crystalline nanowires will be presented.
2. Deposition techniques and growth mechanisms Nanocrystalline materials can be classified into different categories depending on the number of dimensions that are nanostructured (with dimensions lower than 100 nm); we will follow one of the possible classification: i.e. zero dimensional for clusters, mono dimensional for nanowires and two dimensional for films. There are two different approaches to the production of 1D structures: top-down and bottom up technologies. The first one is based on standard micro fabrication methods with deposition, etching and ion beam milling on planar substrates in order to reduce the lateral dimensions of the films to the nanometer size. Electron beam, focused ion beam, X-ray lithography, nano-imprinting and scanning probe microscopy techniques can be used for the selective removal processes. The advantages are the use of the well developed technology of semiconductor industry and the ability to work on planar surfaces, while disadvantages are their extremely elevated costs and preparation times. In the top-down approach highly ordered nanowires can be obtained [16–19], but at the moment this technology does not fulfil the industrial requirements for the production of low cost and large numbers of devices. Furthermore the 1D nanostructures produced with these techniques are in general not single-crystalline. The second approach, bottom-up, consists of the assembly of molecular building blocks or chemical synthesis by vapor phase transport, electrochemical deposition, solution-based techniques or template growth. Its advantages are the high purity of the nanocrystalline materials produced, their small
4
E. Comini et al. / Progress in Materials Science 54 (2009) 1–67
Fig. 1. Schematic drawing of some of the possible morphologies: (a) nanowire, (b) core–shell nanowire, (c) nanotube, (d) nanobelt, (e) hierarchical structure, (f) nanorod and (g) nanoring.
diameters, the low cost of the experimental set ups together with the possibility to easily vary the intentional doping and the possible formation of junctions. The main disadvantage regards their integration on planar substrates for the exploitation of their useful properties, for example transfer and contacting on transducers can be troublesome. The bottom-up approach allows low cost fabrication although it could be very difficult to get them well arranged and patterned [20]. Furthermore more control and insight into the growth process must be achieved for their fruitful integration in functional devices. The most promising approach to produce functional nanowires will be the combination of the two preparation technologies. This review article will be focused on the bottom-up techniques for the preparation of 1D singlecrystal nanostructures. Numerous one-dimensional oxide nanostructures with useful properties, compositions, and morphologies have recently been fabricated using bottom-up synthetic routes. Some of these structures could not have been created easily or economically using top-down technologies. A nomenclature for these peculiar structures has not been well established. In the literature a lot of different names have been used, like whiskers, fibers, fibrils, nanotubules, nanocable, etc. The definition of these 1D nanostructures is not well established. A few classes of these new nanostructures with potential as sensing devices are summarized schematically in Fig. 1. The geometrical shapes can be tubes, cages, cylindrical wires, rods, nails, cables, belts, sheets and even more complex morphologies. When developing 1D nanocrystals the most important requirements are dimensions and morphology control, uniformity and crystalline properties. In order to obtain one-dimensional structures a preferential growth direction with a faster growth rate must exists. Achieving 1D growth in systems with a isotropic atomic bonding requires a break in the symmetry during the growth and not just stopping the growth process at an early stage (0 and 2D). In the past years the number of synthesis techniques has grown exponentially. We can divide these growth mechanisms in different categories, first of all catalyst-free and catalyst assisted procedures and then we can distinguish between vapor and solution phase growth. As far as metal oxides are concerned the most used procedure is the vapor phase one. But solution phase growth techniques provide a more flexible synthesis process with even lower production costs. There are different growth mechanism depending on the presence of a catalyst, i.e. vapor–liquid– solid (VLS), solution–liquid–solid (SLS) or vapor–solid (VS) process. 2.1. Vapor phase growth The vapor phase approach was used in the early 60’ for the preparation of micrometer-size whiskers. These whiskers were prepared either by simple physical sublimation of the source material or
E. Comini et al. / Progress in Materials Science 54 (2009) 1–67
5
through reduction of a volatile metal halide. In the last years this method was used to prepare different materials in form of nanowires. The growth was performed in tubular furnace studied to obtain the proper temperature gradient. The source material once evaporated is transported by a gas carrier towards the growth site where it nucleates. The nucleation can start from particles or catalyst, following the VS, VLS mechanisms. 2.1.1. Vapor–liquid–solid mechanism The controlled catalytic growth of whiskers, and more recently nanowires, was discovered by Wagner and Ellis in 1964 [21], they found that Si whiskers could be grown by heating a Si substrate covered with Au particles in a mixture of SiCl4 and H2 and their diameters was determined by the size of Au particles. Wagner and Ellis named the VLS mechanism for the three phases involved: the vaporphase precursor, the liquid catalyst droplet, and the solid crystalline product (Fig. 5). VLS in the last decades was one of the most important methods for preparing 1D structures, it is promising as a scalable, economical and controllable growth of different materials (oxide, semiconductors,. . .). Understanding the growth dynamics is important to have a greater control in the nanowires shape, diameter and for a selective growth. In general the presence of a metal particle, of size comparable to the nanowire, at its apex leads to the conclusion that the growth mechanism followed the vapor–liquid–solid (VLS) process, but this does not determine the phase of the catalyst during growth. In most of the catalytic growths, nanowires have uniform diameters. The section can be rounded or polygonal with atomically sharp lateral terminations. The growth process takes some dead time, a starting period before the real growth begins, this was experimented also for vapor phase processes [22]. The catalytic particles can be formed by vapor phase and/or surface diffusion transport or be deposited from the evaporation of a colloidal solution or by deposition of a thin film onto the substrate. If the metal does not wet the substrate, it will form clusters as the result of Volmer–Weber growth [23] or when the substrate is kept at the high temperatures required for the growth process, the onset of Ostwald ripening [24] will lead to a distribution of cluster sizes. In some cases the catalyst clusters that initiate the NWs growth can also be formed at the initial deposition step; for example when carbothermal reduction is used to generate a volatile metal that is transported from a carrier gas and then condense on the substrate. Sometimes the catalyst may undergo other processes before becoming active for the growth of nanowires after its formation or deposition. A mixture of the growth compound and the metal might be more active for the NW formation than the pure metal catalyst, and may be required to form an alloy, a true eutectic or some solid/liquid solution. In this case, saturation of the catalytic particle with the growth material or the formation of the proper composition may explain the dead time period before growth. The incorporation of a significant amount of growth material into the catalytic particle is expected to change the volume and, in turn, the diameter of the catalyst from its initial value with a change in the NW section. Consequently Ostwald ripening and incorporation of growth material contribute in changing the size of the catalytic particles. A constant section NWs growth may correspond to a condensation on the catalyst surface and diffusion and segregation at the interface between catalyst and nanowire. When the condensation and incorporation is occurring only on the catalyst and not onto the NW sides, a constant catalyst section results in a constant nanowire section. The dimensions of the catalyst clusters can determine the NW section either by direct matching of the size or by mechanism involving the catalyst curvature in which strain and lattice matching are important. The NW section will decrease and eventually the growth process will end if the catalyst is consumed or evaporates during the growth, or when the material is no longer supplied, or if the temperature is reduced below a critical value necessary for the growth process. Temperature is a key factor in determining processes such as dissociative adsorption, surface diffusion, bulk diffusion through the catalyst, solubility and thermodynamic stability of certain phases. The catalyst cluster can offer a higher sticking coefficient, but the difference in sticking coefficients alone cannot account for the NWs growth process. Further considerations must be performed to explain the preferential incorporation at the interface between nanowire and catalyst. For example
6
E. Comini et al. / Progress in Materials Science 54 (2009) 1–67
the catalytic particle can lower the energy barrier for the incorporation of new material at the growth interface compared to the one needed for nucleation of an island on a sidewall or on the substrate. Adsorption occurs from the fluid (gaseous, liquid or supercritical) phase, it can be molecular or dissociative and may occur on nanowire, catalyst, or substrate. The catalyst can activate the growth with a sticking coefficient higher on its surface and vanishing elsewhere. After the adsorption there is the adatoms diffusion onto or into the catalyst, across the substrate, or on the NWs lateral sides. In order to have the unidirectional growth, the last two processes must be rapid and avoid secondary nucleation. The nanowires can grow from the top or the bottom of the catalyst cluster and as reported in Fig. 2, a catalyst cluster can give rise to single or multiple nanowires growth. The catalyst can be found at the bottom or top of the nanowire. In single NW growth there is a one-to-one correspondence between catalyst and nanowires. In single wire growth control over the nanowire diameter should be obtained controlling the catalyst radius. While in multiple nanowires growth the section must be related to other factors such as the curvature of the growth interface and lattice matching between the catalytic particle and the nanowire. Regardless of the phase of the catalyst, the major requirement is the mobility of the growth material that can allow reaching the growth interface with a low probability of nucleation in sites other than the nanowire–catalyst interface. The growth activation energy can be related to activated adsorption or with surface or bulk diffusion. The essential role of the catalyst appears to be lowering the activation energy of nucleation at the interface. There is a substantial barrier associated with the formation of the critical nucleation cluster at a random position on the substrate or nanowire according to classical nucleation theory. If the catalyst can lower the nucleation barrier at the particle/nanowire interface, then growth may only occur
Fig. 2. The processes that occur during catalytic growth. (a) In root growth, the particle stays at the bottom of the nanowire. (b) In float growth, the particle remains at the top of the nanowire. (c) In multiple prong growth, more than one nanowire grows from one particle and the nanowires must necessarily have a smaller radius than the particle. (d) In single-prong growth, one nanowire corresponds to one particle. One of the surest signs of this mode is that the particle and nanowire have very similar radii. Reprinted from Ref. [22]. License number 1905961408030.
E. Comini et al. / Progress in Materials Science 54 (2009) 1–67
7
on the catalyst. The most important role of the catalyst particle is to ensure that the material is preferentially incorporated at the growth interface. Understanding the dynamics of VLS nanowires growth is essential in order to relate the properties of the wire to their processing conditions. A theory for VLS growth has been presented in reference [25] incorporating the surface energy of the solid–liquid, liquid–vapor, and solid–vapor interfaces. The catalyst concentration profile in the droplet, the degree of supersaturation, and the modification to the shape of the solid–liquid interface were predicted as functions of the material properties and process parameters. The calculated growth rate found has the same dependence on diameter as the flux of growth material at the liquid–vapor interface; thus, for radius independent flux, growth rate results also radius independent. It is often found that, the growth rate should decrease with decreasing diameter [26,27]. The effects of size on the growth kinetics of nanowires by the vapor–liquid–solid mechanism were addressed from the theoretical point of view in [27]. The dependences of the growth rate and the activation energy of crystallization on size were given quantitatively. The obtained theoretical results showed that the smaller the nanowire radius, the slower the growth rate, and the activation energy of crystallization increases with decreasing radius of the nanowire. These theoretical predictions are in agreement with the experimental cases. However, this conclusion depends on the growth conditions [28] since the extent of supersaturation within the catalyst depends on the temperature and gas-phase composition. Transitions from smaller diameters having lower growth rates to smaller diameter having higher growth rates can occur as temperature and gas-phase composition are changed. Although it is commonly believed that in the VLS process, the size of the catalyst particles determines the NWs width, this is not true for all deposition conditions. Experimental studies on ZnO NWs growth on Al0.5Ga0.5N substrate confirm that this rule only applies when the catalyst particles are reasonably small (
