Grain storage is an essential part of the food production chain. Therefore, pre- venting grain deterioration is a key issue in a grain storage system. There are several causes for spoilage, all resulting in grain quality and quantity loss. One approach to detect incipient spoilage is by detecting the produced volatiles. In the past, many sensors for detecting volatiles have been developed and are used in industry. However, most of the commercial gas sensors are bulky with high power consumption, mainly limited in range of operating temperature, or require a restricted control over temperature and humidity. This thesis describes the design, fabrication and evaluation of a gas sensor capable of detecting volatiles and considers the potential use of polymer- based sensors. Conductive polymer-based sensors have been reported sensitive to a wide range of volatiles but are commonly evaluated under a controlled environment. Conventional sensor reproducibility and repeatability are also a concern due to the difficulties associated with polymer composite film preparation. In addition, current studies have not fully explored sensor properties in response to humidity, a common factor in any environment, and a variable parameter in grain storage facilities. Moreover, these sensors suffer from ambient temperature dependency as they work based on partitioning mechanism. To enhance sensor performances and eliminate the temperature dependency, a new sensor structure is proposed. The new design uses standard lithography process to fabricate a thermally isolated cantilever containing interdigitated electrodes and a micro-heater to efficiently heat and maintain a constant temperature throughout the interdigitated electrodes. This structure eliminates sensor response drifts caused by ambient temperature variations. Capacitive measurements are performed as the means of volatile detection, which simplify the use of polymers due to the absence of conductive filler and the challenges associated with it. Frequency spectroscopy provides additional information regarding the presence of volatiles compared to conventional resistive sensors, since mechanisms other than swelling are involved. Moreover, frequency and temperature modulations can be employed to further enhance sensor performance, enabling the use of a reduced number of sensors in a sensor array.