STUDENT PROJECTS

Introduction
Nanometer-scale structures have attracted great research interest due to their unique physical properties and potential applications. Nanotubes and nanowires in particular have novel properties differing from bulk materials owing to their one-dimensionality.

Monoclinic gallium oxide (b-Ga₂O₃) has one of the widest band gaps (4.9 eV) among semiconducting metal oxides. Gallium oxide has already been used for the emission of blue light and as a gas sensor [1]. At temperatures greater than 900 °C gallium oxide thin films have shown significant conductance response to oxygen gas concentrations [2], and at lower temperatures the sensitivity switches to reducing gases such as carbon monoxide, methane, and ammonia [3]. Low-dimensional gallium oxide nanostructures such as nanowires, nanosheets and nanoribbons have been created by physical evaporation [4] and arc-discharge methods [5]. These structures exhibit a very high surface to volume ratio, which provides a larger reaction surface using only very small amounts of oxide material. This simultaneously increases gas-sensing reaction times while reducing power requirements associated with heating the sensors. These attributes provide great advantages in gas-sensing over gallium oxide thin films and can provide a means by which gallium oxide based systems can effectively solve problems associated with conventional tin oxide sensors [6].

Experimental Procedure
My synthesis of gallium oxide nanostructures is based on chemical vapor deposition (CVD). The reaction used for the formation of gallium oxide is simply expressed:

In my experiment, the above process was completed in an ordinary alumina tube furnace. Two 0.7-mm² silicon wafer substrates were first coated with a 0.01 M solution of Ni(NO₃)₂ catalyst, after which the catalytic solution was heat-evaporated. This process was performed three times to leave a film on the substrate surfaces. A small bead of gallium 2 mm in diameter was placed at one end of each silicon wafer and the substrates were then loaded into the furnace tube in a ceramic boat. A 4-mL beaker of water was placed at the upstream end of the tube. The tube was pumped to a vacuum, then flooded with argon gas and repumped. This cleaning procedure was repeated three times. The substrate was then heated slowly under argon gas flow of 90 sccm at a constant pressure of 300 torr. The temperature was increased by about 17 °C/min to 850 °C, where it was held constant for 20 minutes and then decreased again by about 17 °C/min. When the temperature had decreased to 650 °C the gas was stopped and vacuum was again achieved. The sample was then allowed to cool naturally back to room temperature.

The sample was first analyzed using an optical microscope, which revealed gray and white deposits on the substrate of a fine structure. Cocoons of white and transparent silicon oxide structures had encapsulated the gallium beads. A scanning electron microscope (SEM) (FEI, Strata DB235 Focused Ion Beam (FIB) system) was then used to investigate the specific morphologies of the structures created.

Results and Analysis
The SEM images reveal a variety of nanostructures upon a bed of micrometer-sized gallium oxide particles (Fig. 1). Nanowires were found in almost every place where gallium oxide crystals had formed, varying only in length and diameter. The nanowire diameters ranged from over a hundred nanometers to only 30 nm. Some nanowires were seen to be many tens of micrometers long, showing that growth along the wire axis was over a thousand times faster than that in the radial direction.

Many of the nanowires in Figure 1 terminate in a crystalline nanoparticle. These particles have diameters of about 200 to 300 nm, larger than the diameters of the nanowires themselves. It is noted that not all the nanowires were found with such a particle.

The process by which gallium oxide nanostructures are formed has yet to be fully described. The catalytic vapor-liquid-solid (VLS) crystal growth mechanism [7, 8] has been used previously as a description for the synthesis of silicon nanowires [9], and has been suggested as the mechanism for gallium oxide nanowire growth, as well. By this process, evaporated gallium oxide is absorbed by a spherical catalytic alloy particle. As this particle becomes supersaturated gallium oxide is deposited in a lattice structure on the substrate, pushing the particle along in the preferential growth direction. Because this process requires a catalytic particle attached to each nanowire, electron microscope images reveal such particles at the nanowire tips. These catalytic particles have diameters comparable to, but larger than the diameters of the nanowires themselves. Synthesis of gallium oxide nanowires by the VLS process has been reported at high temperature using a gold layer as a catalyst.

The absence of the catalyst element at the nanowire tips, however, as is observed in the SEM images, implies that another process is at work. The defect growth process [4, 5, 10] offers another explanation. Defect growth describes the growth of nanostructures as arising from defects in the gallium oxide parent microparticles. The high temperature during synthesis leads to microparticles with many defects of various types. Twin defects, for example, create a crystal notch in which evaporated gallium oxide can bind to the lattice in two places. This notch thus represents a lowest-energy configuration in comparison to the other crystal faces, and thus the crystal growth moves fastest along the axis of the defect. The high surface entropy of gallium oxide crystals, as noted above, ensures that the sides of the nanowires and nanobelts are very smooth. Particles that do bind at these surfaces are likely to migrate towards the growth end. Gallium oxide nanowires aligned along twin defects in the crystal lattice have been observed in previous research.

Summary
In summary, monoclinic gallium oxide nanostructures were grown on a silicon substrate with Ni(NO₃)₂ catalyst. One-dimensional nanowires and two-dimensional nanobelts and nanosheets were observed with preferential growth along one or two axes, respectively.

Acknowledgements: This project was financially supported by a grant from the U.S. National Science Foundation Division of International Programs. The support of Shang-Fen Ren of Illinois State University, Da-Peng Yu of Peking University, and the School of Physics at Peking University is thankfully acknowledged. I am grateful to Yan-Wu Zhu and Jie Xiang, as well, for their assistance and collaboration.