The advance of nanotechnology has opened new opportunities to develop ever more sensitive sensors. Single-walled carbon nanotube (SWNT) is one of the most exciting materials in this regard due to its high surface-to-volume ratio and unique electronic structure. Ironically the ultrahigh sensitivity of SWNTs is inevitably compromised by various unintentional contaminants from the device fabrication process as well as the ambient environment. Indeed, our studies showed that despite significant progress made in the last decade, we are still nowhere close to what a pristine SWNT-sensor can truly offer. Under in situ ultraviolet (UV) light illumination, the electrical response of pristine SWNTs became extremely sensitive to gaseous molecules such as nitric oxide (NO) with a detection limit as low as 590 parts-per-quadrillion (ppq) at room temperature, which is ~10,000 times superior to any reported sensor to date. Gas sensing on NO₂ and NH₃ further confirmed that pristine nanotubes could have sensitivities orders of magnitude better than what previously had achieved. Furthermore, we observed the electrical response to reverse its direction upon NH₃ exposure after applying in situ UV light illumination, which gave us clear indications of the origin of gas ultrasensitivity. Because of its simplicity the concept of in situ cleaning can be deployed in existing sensor architectures. The study is expected to pave the way for developing the next generation of ultrasensitive sensors.

At the core of the detection sensibility of the sensors produced at our lab was the change in conductivity produced by the exposure of the nanotubes to the molecules of NO, NO₂, and NH₃. Given that the films were a network of tubes and junctions, it was reasonable to assume that the total conductivity of the film was the combination of the intrinsic transport response of the tubes and the less defined contribution from the junctions. Our first exploration of these questions was done on the study of the gases on the intrinsic transport characteristics of the tubes. The effects of the molecules on the transport response of the tubes arises from the combined effect of doping, through the injection of carriers, and the introduction of scattering centers affecting the transport process. These are not independent effects. On the contrary, they are both the result of the mechanisms of interaction between the molecules and the tubes. As a consequence, our theoretical modeling consisted of a series of carefully planned ab-initio calculations followed by their theoretical interpretation in order to obtain general knowledge of the charge transfer and scattering mechanisms governing the transport process in the pristine tubes.

The calculation was based on the kinetic Langmuir model, with the addition of poison sites, from which no desorption occurs during the experiment. The model reproduced the available experimental data very well. It not only explains the microscopic phenomena governing adsorption, but also the model will guide us the direction to improve the sensitivity and detection limit of the sample. Further experimental work on the effect of temperature, UV light, gate voltage, and defect density can potentially further improve their performance and achieve the ultimate sensitivities of those gas sensors.