This paper presents a novel gas-sensing scheme using a single graphene field-effect transistor (GFET) and machine learning to achieve gas selectivity. Instead of using multiple functional materials, the GFET's gas-sensing conductivity profiles are recorded, decoupled into four physical properties (electron mobility, hole mobility, carrier concentration, and the ratio of residual carrier concentration to charged impurity concentration), and projected onto a 4D feature space. Machine learning algorithms classify these 4D vectors to differentiate target gases. The system successfully classified water, methanol, and ethanol vapors with high accuracy individually and qualitatively distinguished methanol from water in a binary mixture, suggesting its potential for real-world applications with background humidity.

Takeshi Hayasaka, Albert Lin, Vernalyn C. Copa, Lorenzo P. Lopez Jr., Regine A. Loberternos, Laureen Ida M. Ballesteros, Yoshihiro Kubota, Yumeng Liu, Amel A. Salvador, Liwei Lin