With the discovery of low-dimensional conductive materials advancing the miniaturization of electronic components, field-effect transistors (FET) can now incorporate single-molecule elements as a channel or gate. Among their applications, these small architectures allow single-molecule studies, for instance by observing their folding-unfolding or binding dy- namics. These studies were mainly carried out with field-effect transistors based on one- dimensional (1D) materials such as carbon nanotubes (CNT) or silicon nanowires (SiNW). Due to their reduced dimensionality, these materials offer good control on their interaction with 0D molecules, and therefore of their integration into the circuit. However, these 1D materials present reproducibility and scaling issues, due to the fact that they are difficult to grow and/or assemble in FET devices. This thesis focuses on the use of a two-dimensional carbon-based material, graphene, as an alternative for the fabrication of devices for studying the dynamics of single molecules. Graphene is a hexagonal network of carbon atoms that offers an excellent electrical conduc- tivity as well as a carbon-based chemistry for anchoring biological molecules on its surface, this makes it a prime candidate for the electrical detection of individual molecules. Above all, its dimensionality is compatible with large-scale microfabrication processes, which offer the possibility of statistical studies on a large number of devices. Thus, the detection of biological molecules using graphene-based field-effect transistors (GFET) has experienced significant development over the past decade, but several aspects remain to be resolved, including scale- up of the manufacturing, control of the functionalization chemistry, and miniaturization of the channel at the single molecule scale. In this thesis, I present contributions on these three aspects. First, I describe a method for scaling up graphene transfer in an industrialization perspective, by designing and implementing a graphene transfer setup allowing automation for increasing the yield of GFET fabrication. I then focus on the functionalization dynamics of graphene devices with an anchor molecule named PBASE (1-Pyrenebutyric acid N-hydroxysuccinimide ester) commonly used in the case of GFET-based biosensors, which reveals the adsorption and accumulation kinetics of the molecule on the graphene surface. Finally, I describe the design of a GFET architecture based on nanoconstrictions implemented in the graphene channel, designed to host a single molecule. These constrictions were obtained using electron beam lithography (EBL) and deep reactive ion etching (DRIE), which allows the modeling of high-resolution features of a few nanometers in the graphene channel. Nanowells were opened in the resin perpendicular to the constriction, promoting single-point, single-molecule chemistry. I then explore the immobilization of a single strand of DNA on nanoconstriction, and the dynamic study of its folding. This thesis therefore presents innovative results in terms of architectures and scaled implementation processes of GFET for biodetection purposes.