ABSTRACT
Sensors are, in general terms, devices that convert chemical or physical quantities in a signal, containing information, suitable for an operator. For example some gas sensors translate the interaction of gaseous molecules to the variation of an electrical signal. Gas sensors are employed in a large number of applications: aerospace, medicine, robotics, warfare, pollution control, security threats, safety, etc. Unfortunately, the commercially available gas sensors do not completely meet the expected specifications for most of applications especially concerning sensitivity, selectivity, compactness, versatility, lower power consumption and time response. In this context nanomaterials can allow people to obtain breakthrough performances and to constitute a real alternative to more traditional technologies (e.g., metal oxides [1]). CNTs are one-dimensional molecular structures obtained by rolling up one graphene sheet (single-wall carbon nanotube, i.e., SWCNT) or more (multi-wall carbon nanotube, i.e., MWCNT). Since the first sensing measurements performed at Stanford in 2000 [2], many scientific teams have focused their interests on CNT-based chemical and biological sensors [3-5]. Indeed, different kinds of gas/chemical sensors, based on different working principles, have been fabricated thanks to CNTs: miniaturized ionizing gas sensors (also called “micro-gun” sensors) [6], CNT thin films with variable resistance as a function of the adsorbed gas properties [7-10] and finally carbon nanotube field effect transistor (CNTFET)-based sensors. Concerning the so called “micro-gun sensors,” briefly, these are sort of miniaturized gas chromatography systems exploiting the gas ionization produced by the nanotube tips. One of the main issues about these devices is the detection limit. Indeed they are able to detect only some percent of the gas in air (which corresponds to hundreds of ppm). The other concern is that the breakdown voltages are quite high (hundreds of volts) and so power consumption too. CNT-based resistive sensors exploit the change of CNT mat after having interacted with the targeted gas molecules. In this case the main drawbacks are: the recovery time is too high (more than 30 min at least), the response time is quite long (these are larger devices than CNTFET-based sensors) and finally the physics of the phenomenon is not clearly understood; thus, it is difficult to
improve their performances (the right parameters to improve are not well known). Focusing on CNTFET-based sensors, these devices have the potential to be ultra compact, effective at room temperature, with low power consumption, with a very fast response, a short recovery time (a few seconds) and show a good versatility (they can be used for several gases). Moreover, we can achieve relatively low cost fabrication (using a CMOS compatible technology). For this reason we have decided to deeply analyze the state of the art in the domain of this kind of gas/chemical sensors and to review the different physical arguments developed to explain the gas/device interaction, trying to find a coherence between them in order to answer the different unsolved questions. A section is devoted to the different routes developed to improve selectivity. This is a very important issue considering that a high selectivity is the key factor to definitively strike huge markets such as security (e.g., detection of warfare gases such as nerve agents or of highly volatile explosive traces), healthcare (e.g., NO sensors for asthma monitoring or CO2 for emergency airway management), safety (e.g., CO or BTEX sensors) and environmental monitoring (e.g., NO2, benzene).
