(23g) Controlled Nanostructure and Catalyst Approaches of Metal Oxide Semiconductors for Gas Sensing

Applications of chemical sensors include environmental monitoring, automotive applications, emission monitoring, and aerospace vehicle health monitoring. The superiority of nanomaterials for sensor applications is clear. High surface area and controlled structure are the hallmarks. Each aspect is particularly relevant to sensors. Surface area is critical to reactive gas adsorption and translates into high sensitivity. Relative to micron sized grains or powders or layers or films, nanoparticles offer a 10 to 100-fold increase in each parameter. Controlled structure provides the reactive sites for adsorption and their effect upon the overall conductance. Nanomaterials often possess enhanced reactivity due both to a strained surface lattice and/or increased prevalence of step, edge, corner and terrace sites. Such activity further enhances sensitivity and potentially lowers temperature operation. Operation at lower temperatures saves power, extends operating lifetime and maintains reproducibility by preventing sintering-induced grain growth. Finally, lower temperature, combined with structure control can advantageously yield selectivity.

In recent years two competing approaches have been developed for synthesizing 1-dimensional forms of metal oxide semiconductors (MOSs): electrospinning and CVD synthesis. Each method produces 1-d sensor elements that can be incorporated into next generation sensors. Each synthesis method and product has attendant advantages and limitations. Apart from device fabrication and manufacturing issues, these two methods produce elements that differ primarily in their crystallinity. CVD synthesis produces single crystalline nanowires. Electrospinning produces polycrystalline elements upon calcination of the (as-spun) sol-gel fiber. Though each form is a nanoscale material, an additional parameter, crystallinity, will have profound consequences upon the viability of each 1-d element for reactive gas sensing and devices incorporating them.

With a diameter approaching twice the nominal charge depletion layer thickness (~ 10's of nanometers, dependent upon temperature), the full conducting channel of a nanowire can be affected by adsorption of oxidizing gases, analogous to the electrical conductivity modulation in a field effect transistor. Although somewhat thicker, the polycrystallinity and cylindrical structure of a nanofiber permits deeper penetration of the depletion layer into the structure. Essentially the same narrowing of the ?effective? conduction channel occurs. Carrier depletion (or replenishment) throughout the ?bulk? nanostructure will expand the sensor dynamic range by the virtue of adsorbates leading to full charge depletion (or replenishment) with corresponding infinite or near zero resistance respectively. With the entire bulk of the material responsive to surface adsorbed species, sensitivity gains may be achieved relative to 2-d films where often only the outer surface layer is affected while the underlying portion remains unchanged (maintaining constant conductance).

In light of these idealized constructs, electron microscopy illustrates the differences between a single crystal nanorod and a polycrystalline nanofiber. In the former case a continuous depletion layer forms around the rod perimeter. If it is of sufficiently small diameter, the entire rod is volumetrically depleted of electron density. In the case of the nanofiber, the net conductivity is the summation of the myriad potential barriers between particles and grains. While it would appear that the nanorod is the limit of the summation describing the nanofiber case, in practice the nanorod diameter is generally larger than twice the depletion layer thickness. Moreover in principle the crystalline structure may not support a high surface density of defect sites or concentration of oxygen adsorbates. Alternatively in the polycrystalline nanofiber not all junctions may be accessible to adsorbates. Such spots would correspond to a ?short?. Moreover there could be a considerable variation in potential boundaries, given the random orientation of single crystal particles with respect to each other.

To further enhance the sensitivity of these materials to gases, catalyst nanoparticles were deposited upon the surface of these 1-d nanoforms. The utility of catalysts upon active oxide supports is well known in catalysis but has yet received little attention in sensing. Most approaches using catalysts of either noble metals or other heteroelements dope the material. With a homogeneous distribution, most of the catalyst (whether atomically distributed or as nanophase particles) is inaccessible to the near-surface region, the region where active sensing occurs. Secondly, an atomic dispersion will alter lattice and grain structure, but not specifically create catalyst sites upon the sensor surface. The latter is the desirable structure from a catalytic material viewpoint and is exploited for the 1-d sensing elements reported here. In contrast to traditional doping approaches where noble metals are incorporated within the lattice structure, our approach uses deposition of discrete metal (catalyst) nanoparticles upon the oxide surface. We note that this approach is frequently used in catalysis where the noble metal nanoparticles and/or the interfacial region between the particle and oxide support greatly accelerates the reaction compared to the bare oxide surface.

In the form of discrete catalyst nanoparticles, catalysts were applied to the nanorods and carbon nanotubes to adjust the free carrier charge density and band energy levels. This particle coating will permit exposure of the underlying metal oxide support and most importantly establishes an interface between the particle and support oxide. The interfacial junctions will be self-polarized (forming a Schottky junction) by virtue of charge transfer due to differences in work function and electron affinity between the metal and oxide, respectively. This interface is expected to be highly reactive for well crystallized metal nanoparticles. This reactivity is critically dependent upon the nature (composition, crystallinity, oxygen sites and their binding energy, etc.) of the oxide support material and availability of oxygen from the junction region surrounding the metal nanoparticle. By reference to single crystal catalysis studies, it is well known that steps, edges and terraces upon metal surfaces are the reactive sites compared to single crystal planes given differences in electron density and energy levels. Such differences are expected to be magnified by the polarization junction that forms between the oxide and metal.

The central goal of this investigation was to operationally test the difference between these two nanostructured materials forms, with and without catalysts. In the fabrication of the prototype devices we gained practical knowledge of each synthesis method and realized caveats for integration of each element type towards commercial device manufacture. Advantages and limitations of each method will be presented.