Electrical Properties of Sorted Carbon Nanotube Networks and Their Application for Gas Sensors

Abstract

Carbon nanotube (CNT) is considered as a promising material for future electronics due to its excellent electrical properties such as a high current density, a high on-off ratio, a high mobility and a low subthershold swing. However, practical applications for CNT-based devices are hindered by complex and time-consuming fabrication processes. For example, the chirality of an as-grown CNT is determined randomly, so As-grown CNTs has the problem that they consist of semiconducting CNTs (s-CNTs) and metallic CNTs (m-CNTs). The electrical properties of s-CNTs and m-CNTs are opposite to each other. Thus, it is difficult to expect a good performance when a device is made using as-grown CNTs. To solve the problem, recently, a density gradient ultracentrifugation (DGU) method was developed to sort as-grown CNTs into m-CNTs and s-CNTs. However, practical applications using sorted-CNTs are still being studied because of the complexity of the fabrication process. In this dissertation, I and co-works have explored electrical properties of sorted CNTs. In addition, we will propose a new strategy for the fabrication of CNT-based gas sensors as a practical application. First, we will discuss pristine semiconducting carbon nanotube network-based devices exhibiting intrinsic characteristics. Conventional sorted CNT-based devices have disadvantages that surfactants or CNT bundles are exist in their channels. These surfactants and CNT bundles lower the performance of the performance of CNT-based devices, but it is extremely difficult to remove both surfactants and CNT bundles simultaneously. Here, the solutions of pristine s-CNT without bundles or organic impurities were prepared in 1, 2-dichlobenzene solutions via filtration and centrifugation processes. A FET device based on a pristine s-CNT network exhibited a rather large on−off ratio up to over ∼106 and a subthreshold swing as small as ∼490 mV/dec, which are comparable to those of devices based on a single s-CNT. Next, we will discuss the scanning noise microscopy method for the measurement of the noise characteristics of sorted carbon nanotube networks. Here, we measured the current maps and the electrical noise power spectral density (PSD) maps of a s-CNT network and a m-CNT networks via a modified conducting AFM system. By analysing the current maps and the PSD maps, we investigated the noise source activities of sorted CNTs. We found that the noise activities of sorted CNT networks were just depending on the diameter of each CNT and the density of CNT network. Also, in the case of a s-CNT network, we found that noise activities were high at crossed-CNT junctions. Lastly, a dye-functionalized sol-gel matrix on a s-CNT network for dual-mode sensing will be discussed. Here, the CNT-dye hybrid gas sensors were fabricated by functionalizing dye molecules on top of s-CNT networks via a sol-gel method. The CNT-dye hybrid gas sensors could selectively detected SO2, NH3 and Cl2 gases. The sensitivity of the gas sensor exhibited more than 50 % by the exposure to the gas species with its concentration even under its permissible exposure limit. Also, we could refresh used gas sensors simply by exposing it to fresh N2 gas without any heat treatment. These works could provide an important insight regarding the electrical properties of sorted CNTs and their possibility for practical applications.