Volatile Organic Compounds (VOCs) are an important class of air pollutant because many of them are harmful to human health. VOCs are typically present at very low concentrations – parts-per-billion (ppb) and lower – which makes their detection a significant challenge. Standard measurement techniques, such as Gas Chromatography (GC), are typically sensitive and selective, but are limited by their large size, high cost and complexity to operate. Such factors restrict deployment for practical applications, including measurement across air quality networks. In contrast, gas sensors are typically small, inexpensive and easily deployed, but are limited by poor selectivity.

This work aims to establish the extent to which gas sensors can be used to achieve sensitive and selective detection of VOCs. Based on the processes of adsorption and desorption, it examines how temperature control of functionalised silica adsorbents can be used to produce discriminable signals of VOCs. In this context, the VOCs of interest were benzene, toluene, ethyl benzene, para-xylene (BTEX), due to their proven toxicity and prevalence in human environments. With the aim of producing a potentially deployable device, a novel sensing platform was designed and constructed. The main features of this Adsorption Device were an aluminium channel, a peltier module heating unit, flow path control and photoionisation detector (PID). In addition to unmodified silica, this thesis presents six modified silica adsorbents with amino, chloro, (n8) alkyl, fluoroalkyl, phenyl and chlorophenyl functionality.

Analysis of BTEX vapours with the seven silica adsorbents indicated adsorption was physical (physisorption) and desorption was readily reversible between 25 and 100 °C. Adsorption was influenced by the strength of adsorbent-vapour interaction, which could be increased by introducing delocalised electron density (phenyl and chlorophenyl silica), but modification could not compensate for any significant loss of surface area and pore volume that occurred. PID responses during adsorption and desorption were found to be discriminable from each other. Vapour desorption was examined with different heating profiles, which were found to initiate distinct response patterns. Principle Component Analysis (PCA) of Adsorption Device data indicated that the responses were sufficiently discriminable that they could be offer a means of vapour selectivity. Tests of the Adsorption Device indicate that selective detection of individual and dual component BTEX vapours is achievable in the ppb concentration range and with a cycle time of 10 minutes. Classification algorithms based on the Adsorption Device output were found to be at least as accurate as previously published research. This work presents significant progress towards to the development of a selective and practical sensor for air quality applications.