(321a) Invited Talk: Advancing Biosensing with Hybrid Nanomaterials and Machine Learning

Single-walled carbon nanotubes (SWCNTs) and more recently graphene derivatives have attracted considerable interest for the development of chemical and biological sensors. These carbon nanostructures are just one atom thick, and their electronic properties are extremely sensitive to adsorption of chemical species on their surface. When decorated with metal or metal oxide nanoparticles, these nanostructures exhibit a large and selective electronic response toward many analytes with potential applications in medical diagnostics. However, achieving selectivity through chemical functionalization necessitates destroying the sp²structure of SWCNTs, leading to a partial or total loss of electrical properties. To circumvent this issue, holey graphene may be used to accomplish covalent chemistry via coupling with graphene edges through carboxylic acid moieties. While pristine graphene is a zero-bandgap semiconductor, holey graphene mimics graphene nanoribbons, producing a bandgap at neck widths less than 10 nm. The resulting structure is analogous to an interconnected network of SWCNTs on a device surface and holds promise for chemical and biological sensing.

Sensor array approaches involving multivariate data analysis and ultimately machine learning techniques to classify different analytes have many emerging applications. For example, sensing specific biomarkers for cells is very challenging. The specific biosensing methods often require foreknowledge of the biomarker in question to successfully identify its presence. For biosensors, this necessitates individually designed devices for each specific target of interest. Carbon nanotube-based field-effect transistors (NTFET) are highly sensitive to changes in the local environment and are an ideal platform for the fabrication of label-free biosensors. Characteristic FET curves contain a wealth of information and have proved useful for analyzing the mechanism of interaction with target analytes. This rich information provides an interesting target for deeper analysis towards more general biosensing applications.